Citation
Apple harvesting and non-immigrant alien workers in West Virginia

Material Information

Title:
Apple harvesting and non-immigrant alien workers in West Virginia
Creator:
Adu-Nyako, Kofi, 1952-
Publication Date:
Language:
English

Subjects

Subjects / Keywords:
Agriculture ( jstor )
Apples ( jstor )
Domestic workers ( jstor )
Farm workers ( jstor )
Immigration ( jstor )
Labor ( jstor )
Labor markets ( jstor )
Labor supply ( jstor )
Mathematical variables ( jstor )
Wages ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
25103548 ( ALEPH )
20240774 ( OCLC )

Downloads

This item has the following downloads:


Full Text

























APPLE HARVESTING AND NON-IMMIGRANT ALIEN WORKERS IN
WEST VIRGINIA: AN ECONOMIC ANALYSIS








KOFI ADU-NYAKO










A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY



UNIVERSITY OF FLORIDA 1988






















To my country, Ghana, and my parents, Yaw Nyako and Yaa Asabea.
















ACKNOWLEDGEMENTS

I wish to express my appreciation and thanks to Dr. Robert Emerson, chairman of my committee, for all the help in bringing this study to fruition. I also acknowledge the diverse contributions the other members of my committee Dr. Carlton Davis, Dr. Larry Kenny and Dr. Max Langham made. Dr. Tim Taylor assisted me with various aspects of the production model and I am thankful.

The government of Ghana has supported me financially, at various

stages of my education for which I am very grateful. I also appreciate the financial support from the Food and Resource Economics Department and from USDA Cooperative Agreement 58-3J23-4-00189 with the University of Florida.

To my friend Abena, for your inspiration and encouragement I am

very grateful. My appreciation also goes to James, Liz, Kwasi, Asamoah and Joe for their friendship in Gainesville.

My thanks go to Berenda, Gwen and Dottie for all the secretarial help they gave me and also to Hope.















iii
















TABLE OF CONTENTS

ACKNOWLEDGEMENTS . . . . . . . . . .... . iii

ABSTRACT . . . . . . . . . . . . ... . vi

CHAPTERS

1 INTRODUCTION . . . . . . . . . . . . 1
Problem Statement . . . . . . . . .. . . 1
Statement of Objectives ... ............. . . 4
Historical Background of U.S. Temporary Alien Worker Program 5
World War II Period: 1941-1947 ........... 6
1948-1951 . . . . . . . . . . . 7
1951-1965 . . . . . . ... . . . 8
1965-present . . . . ... . . . . 11

2 REVIEW OF THE LITERATURE ............ . . 15

3 THEORETICAL FRAMEWORK . . . .. .. . . . . . 25
The Theoretical Model ............. . . . 25
Labor and Product Market Equilibrium in the Domestic and
Foreign Economies . . . . . . . . . 34

4 EMPIRICAL MODEL . . . . . . . . . . . 37
Production Function . . . . . . . . . . 37
The Labor Supply Relations . . . .. . . . . 46
The Adverse Effect Wage Rate . . .. . . . . 49
Data: Sources and Construction of Variables . . . . 52
Employment Data . . . . . . . . . . 52
Wages . . . . . . . . . . . . 53
Average Wage Rate for All Hired Farmworkers . . . 53 Non-farm Wage Rate . . . . . . . . . 54
Index of Congressmen Tenure and Seniority: . . . 54 Other Inputs . . . . . . . . . 56

5 EMPIRICAL RESULTS . . . . . . . . . . 57
Estimation... ... . . . . . . . ...57
Results of Estimation . . . . . . . . . . 58
Supply of Domestic Labor: . . . . . . . 58
Supply of Foreign Labor. . . . . . . . . 64
The Adverse Effect Wage . . . . . . . . 69
Share Equations . . . . . . . . . . 75
Separability . . . . . . . . . . . 82
Elasticities . . . . . . . . . . . 84


iv










6 SIMULATION . . . . . . . . . . . 90
Historical Simulation .................. 90
Alternative Scenarios .................. 97

7 SUMMARY, CONCLUSIONS AND LIMITATIONS OF THE STUDY .... . 100
Summary and Conclusions .. ................ 100
Limitations of The Study and Suggestions for Further
Research. .................. ..... 104

APPENDICES

A DATA FOR CONGRESSIONAL TENURE INDEX ............ 107

B DATA FOR THE MODEL .......... . . ...... 110

REFERENCES . . . . . . . . . . . . . 116

BIOGRAPHICAL SKETCH ................... .. . 121











































v
















Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


APPLE HARVESTING AND NON-IMMIGRANT ALIEN WORKERS
IN WEST VIRGINIA: AN ECONOMIC ANALYSIS By

KOFI ADU-NYAKO

APRIL 1988


Chairman: Robert D. Emerson
Major Department: Food and Resource Economics


This study is an analysis of the interaction of foreign migrant

H-2 workers with domestic workers in the apple harvest labor market in West Virginia. At issue is whether growers require supplemental labor that is not obtainable from domestic sources and what impact such imported labor exerts on the domestic economy.

A labor market model including the supply of domestic and foreign labor as well as the demand component represented by a translog harvesting function was constructed to gain insights into the controversy. Results of the analysis indicate that the supply of domestic labor is inelastic, lending support to the growers' argument that domestic workers are difficult to attain. The tight domestic labor market for harvest labor is aggravated by it being interlinked with the non-farm sector labor market. The evidence from the study is that domestic labor is highly responsive to wages in the manufacturing vi









sector thereby attracting labor from the agricultural sector. The supply of foreign labor was found to be very responsive to farm wages in the U.S., an attestation to the attractiveness of the large wage disparity between the U.S. and the Caribbean that serves as an incentive to immigrate.

The issue of possible job displacement by foreign workers was

investigated by estimating the Allen elasticity of substitution between domestic and foreign workers in the harvesting operation. It was found that the two categories of labor are highly substitutable in production; the computed elasticity of substitution is 83. This is evidence that barring any restrictions on the importation of foreign labor, its use could completely substitute for domestic labor in harvesting. This high substitution elasticity is accompanied by a high demand elasticity for both domestic and foreign labor.




























vii


















CHAPTER 1
INTRODUCTION

Problem Statement

The flow of foreign workers into a nation's economy has aroused a feeling of xenophobia among various groups of nationals. This xenophobia is due in part to the fear harbored by natives that aliens may cause them economic displacement. Whereas the concern by natives for the effects of influx of alien workers is a persistent phenomenon, xenophobic tendencies reach a crescendo during bust periods of the economic cycle and subside or become latent in the upswing periods. Accompanying the heightened concern of citizens are calls for action by the legislative body to stem the tide of the influx of alien workers.

In the U.S. the issue of immigrants both legal and illegal has

been a continuing problem. While certain sections of the economy look upon the presence of alien labor as necessary and desirable, other members of the economy see immigrants as potentially destabilizing both in the economic and social spheres. The latter contend that aliens place undue burden on the provision of social services for which they may not contribute. On a more serious front, temporary alien workers are seen to cause postponement of necessary economic adjustment in the system (Martin, 1979; Briggs, 1978). Advocates of liberalized immigration, however, assert that there exists an economic need for



1









2

alien labor in certain occupations for which local supply of labor is in short supply or for one reason or another is unavailable.

In agriculture, alien farm workers have been used since the 19th century when farm labor was recruited from China, Japan and the Philippines. Mexican immigrants have been crossing the border for work in the U.S. since before World War I (Morgan and Gardner,1982). A more structured program, the Bracero program, was instituted during World War II and nonimmigrant Mexicans were employed to meet the increased demand in both farm and non-farm occupations for the war effort. The program was terminated in 1964.

The H-2 program, which was initiated in 1952 and currently exists, is another example of a foreign agricultural work program that utilizes alien workers in certain agricultural enterprises, particularly in vegetable and fruit and nuts production. Most of the labor employed in this program is seasonal in nature, filling the excess labor demand during the harvest season.

Undoubtedly there exists an excess demand for labor in certain agricultural jobs and at certain times of the production season. The question then arises whether this excess demand exists only because of the presence of friction in the domestic labor market that prevents the domestic supply from responding to the demand. Are there institutional structures and / or characteristics of the domestic labor market, the removal of which would result in the domestic labor market being in equilibrium? On the other hand, if the shortage of labor is "real" and there is a need for supplementing domestic labor supply, then what









3

would be the consequence on the domestic market of the importation of foreign labor?

These questions lie at the heart of the whole immigration debate. Answers to these questions will enable informed policy decisions to be made. Recent efforts by the U.S. Congress seek to streamline U.S. immigration policy. The Immigration Reform and Control Act enacted in November 1986 makes provision for a Special Agricultural Workers Program. Under this provision seasonal agricultural workers who can show proof that they worked at least 90 days in the year prior to May 1986 would be granted temporary alien status pending permanent status. This provision also allows for replenishment of temporary residents in case of labor shortage so long as they work in agriculture for at least 90 days. Such persons are also eligible for future adjustment to permanent status. It is envisaged that for the next three years, the duration of the above provision, not more than 350,000 persons would be eligible. For an agricultural labor market consisting of 2.6 million persons (USDA, 1983)1 the proposed figure is a very significant proportion, and will most likely have a significant impact on the labor market. The impact is likely to be more pronounced if it is considered that these workers are earmarked for only a section of the entire agricultural labor market, i.e. the seasonal harvesting labor market. Ex ante, one cannot assess the direction and magnitude of the impact of the program. However, the H-2 program provides an avenue to





'It is debatable whether this figure includes illegal aliens.
Most people would argue that a substantial number of aliens are not included.









4

explore some of the positive and normative issues likely to be associated with implementation of a foreign worker program.

Recent works in evaluating the effects of foreign worker programs on the domestic agricultural economy include Morgan and Gardner (1982), Mehra (1984) and Emerson and Mehra (1985). Morgan and Gardner's work evaluates the welfare impact of the Bracero Program. This research assumes away a key question about the substitutability between foreign and domestic agricultural workers. The extent that foreign labor is substitutable for domestic workers in production is fundamental to determining the rationality of domestic producer's preference for foreign workers in certain agricultural jobs, e.g., sugarcane harvesting, apple harvesting and tobacco production.

Emerson and Mehra study the effects of the H-2 foreign worker

program in the sugar industry in Florida. The substitutability issue could not be addressed in this study because virtually no domestic labor is used in sugarcane harvesting. The present study intends to include the above neglected dimensions into an economic analysis of the impact of the H-2 foreign worker program. The analysis will be conducted within the context of the apple industry in the State of West Virginia.

Statement of Objectives

The general objective of this study is to evaluate the impact of the H-2 foreign worker program in the apple industry on producers and workers. In order to achieve the above objective an economic model of the apple harvest labor market will be constructed embodying the demand and supply relationships that pertain and incorporating existing









5

institutional arrangements that impact on this market, e.g., Adverse Effect Wage Rate (AEWR).

Secondly, from an aggregate production function of apple

harvesting, the short run elasticity of substitution between domestic and alien workers will be estimated. In modelling the harvest labor market, analysts have assumed implicitly that decisions on harvest labor factor use are taken separately -from other inputs used in the harvesting operation. This separability assumption will be tested explicitly in the models. Finally, analysis of the policy implications arising from the results of the empirical investigations above will be considered.

Historical Background of U.S. Temporary Alien Worker Programs

Although the first large scale temporary alien labor program was initiated in 1917, the stream of alien workers into the U.S. was an on going phenomenon at the turn of the 20th century. This request for supplemental labor was in response to the increased demand for labor for the World War I effort and the ensuing shortage of labor in agriculture and railroad construction. This first program of temporary alien workers comprised mainly Mexicans estimated at about 80,000 persons and a few Bahamians and Canadians. While the railroad component of the program was officially terminated at the end of the war in 1918, the agricultural employment part was extended to 1921.

The legislative authority under which the program was authorized is the ninth proviso of section 3 of the Immigration Act of 1917. The Labor Department set rigid regulations for the employer in order to obtain alien workers; however, enforcement of these regulations was









6

rather lax and this resulted in large numbers of Mexicans deserting their original agricultural employers for more lucrative pursuits in industry. In the 1920s Mexican immigration quickly expanded and within six years had doubled from 486,000 to almost 900,000 in 1926 (U.S. Senate, 1980). This growth occurred in spite of the official termination of the alien worker program in 1921. Kiser (1972) attributes the growth to the large influence that employers of Mexican labor had on policy making, albeit informally. The advent of the 1930s and the accompanying economic depression precluded any need for supplemental labor. Mexicans were repatriated, some under very unpleasant conditions. This process of repatriation resulted in a lot of resentment from the Mexicans.

World War II Period: 1941-1947

Prior to the U.S. entry into the war in December 1941 there

existed a labor surplus situation. However, the war caused an exodus of rural farm workers to higher wage industries producing for the war effort which resulted in labor shortages in agriculture. As a consequence supplemental labor was requested from Mexico. The Mexican government was reluctant and the subsequent detailed agreement wrangled out of the U.S. government was due in part to their sensitivity to the humiliating treatment meted out to their nationals after the termination of the post- World War I program. Also, the U.S. wanted an agreement which prevented adverse effects on domestic workers due to the importation of foreign labor.

Like the first program, the ninth proviso of the 1917 Immigration Act was the authority under which Mexicans were admitted contingent on









7

certification from the U.S. Employment Service that local workers were not available. Admission of other foreign nationals from the Bahamas and Jamaica was similarly under the ninth proviso. The agreements for the admission of these other foreign nationals unlike the Mexican Bracero Program involved merely a memorandum of bilateral agreements between the governments. These agreements formed the original basis of the British West Indies (B.W.I.) Program. Public Law 45, enacted in 1943, and the various intergovernmental agreements were the basis of foreign labor supply until December 1947. During this period, the U.S. government paid all transportation and recruitment expenses and was also the labor contractor. The Department of Agriculture administered the foreign labor supply program. Braceros were employed in cotton, sugar beets, fruits and vegetables and were concentrated mainly in the West and Midwest. B.W.I. workers were used mainly in the Eastern seaboard states and were extensively involved in the agricultural program of the War Food Administration. 1948-1951

The emergency war time legislation ended in December 1947.

Importation of Mexican Braceros proceeded under frequently revised international agreements. B.W.I. and Bahamian workers were contracted under waiver provided by the ninth proviso. The administration of the foreign worker program reverted to the U.S. Employment Service. While the B.W.I. Program was generally assessed favorably, the Mexican segment was criticized as causing wage depression in areas where Braceros were employed (see U.S. President's Commission on Migratory









8

Labor, Migratory Labor in American Agriculture, Washington, 1951 p.51).

1951-1965

The Bracero Program

As a response to the dissatisfaction with the Mexican Program, Public Law 78 was enacted in 1951 and provided the legal framework under which the Bracero Program operated until its termination in December 1964. Basically this law required the U.S. government to set up recruiting centers at or near the border, provide transportation and medical care from Mexican recruiting centers to the U.S. centers, and provide guarantees of performance by employers under terms of the law. Employer responsibilities under the law included provision of transportation to and from government recruitment centers, free housing and meals at reasonable cost. They were also required to pay prevailing wages in the area and guarantee employment of workers during 75% of the contract period. Craig (1971) maintains that most of the guarantees of the Bracero Program were empty guarantees since their implementation was contingent on domestic workers enjoying same and such were nonexistent.

Between 1952 and 1956 the Bracero Program expanded rapidly with the number of workers peaking at 445,197 in 1956. Part of the reason for this increased use of Braceros was due to a crackdown on illegal aliens in Operation Wetback that resulted in an increase in the apprehension of illegal aliens and made such aliens less available for use by growers who wanted to avoid the higher cost of using legals.









9

From 1956 the Department of Labor required stricter housing

regulations and issued prevailing wage rates for certain areas. These regulations, effectively increased the cost of using Braceros and curtailed the demand for them. During the latter part of the fifties the program came under increasing opposition from labor, welfare groups, and the Secretary of Labor. The main criticisms were that Braceros were adversely affecting wages, as evidenced by wages remaining stagnant for ten years in areas of Bracero use. In contrast, national wage levels had increased about 46% during the same ten year period. Also the availability of Mexican workers and the preference for them by some growers had caused job displacement of domestic workers.

In the early 1960s opposition to the program quickly increased.

The Kennedy administration tightened regulations that sought to protect domestic workers. An Adverse Effect Wage Rate was instituted for all states that used foreign workers. This is a mandatory minimum wage that must be paid by all employers to both domestic and foreign workers alike and was generally higher than prevailing wages. It became increasingly difficult for the Bracero Program supporters to gain congressional support for its renewal. The program was finally terminated in December of 1964.

The B.W.I. Program

Whereas PL78 was the legal basis for the Bracero Program until its demise in 1964, all other temporary labor programs including the B.W.I. Program derived their legislative authority from PL 414, the Immigration and Nationality Act of 1952. This act grants the Attorney General "sufficient authority to admit temporarily certain alien









10

workers, industrial, agricultural, or otherwise, for the purpose of alleviating labor shortages as they exist or may develop in certain branches of American productive enterprises, particularly in periods of intensified production" (U.S. Congress, House of Representatives, 1952). Whereas final legal authority as to who will be admitted and who will be rejected reposed in the Justice Department's Immigration and Naturalization Service (INS), the Department of Labor(D.O.L.) had an advisory role as to certification of the need for supplemental labor. During the fifties and early sixties the expanded Bracero Program overshadowed the B.W.I. Program in importance. While the former program attracted much adverse attention, the B.W.I. Program was looked upon favorably. For example, in 1953 Senator Holland of Florida contrasting the B.W.I. Program to the Bracero Program commented that the B.W.I. Program has entailed minimal government expense and supervision and was both simple and acceptable to both the employers and employees. Again in 1957 a House Subcommittee report found the importation of agricultural workers from the British West Indies beneficial for agricultural interests in the U.S. While the B.W.I. Program was small in size compared to the Bracero program, B.W.I. workers were concentrated in small areas along the Atlantic Seaboard and contributed immensely in agricultural harvesting activities ranging from citrus in Florida, to apples in the Virginias and potatoes and cranberries in the northeast. There was a steady increase in the B.W.I. contract workers from the latter part of the fifties and the early sixties.











1965 to Present

At the termination of the Bracero Program in December 1964 it was anticipated that the B.W.I. Program would be expanded to accommodate a larger foreign worker program. Contrary to expectations, the Department of Labor issued new regulations which sought to curtail even further importation of foreign labor under the H-2 program. These included higher wages than previously for domestic workers, more housing and transportation benefits and limiting the certification period to 120 days. Attempts in the courts to weaken or overrule the D.O.L. regulations did not succeed. Secondly, amendments proposed in Congress in 1965 to remove certification authority from D.O.L. to the more sympathetic Secretary of Agriculture failed. The 1965 amendment to the Immigration and Nationality Act has served as the basis for importation of temporary foreign labor to the present. Major revisions were made and published in March 1978. Some of the major developments and controversies relating to the H-2 program and the apple industry in particular will be highlighted in the following paragraphs.

In the sixties the criticism leveled against the D.O.L. was a perceived laxity in the administration of the temporary alien worker program. Rising unemployment during the seventies caused the H-2 program to be under persistent attack from labor groups and congress. In 1970 a complaint of exploitation of domestic migrant farm workers was leveled against the Rural Manpower Service of the D.O.L. A 1972 review by D.O.L. of the program found that the impact of foreign workers in apple producing areas was insignificant. Wage determination and whether the prevailing piece rates measure up to the mandated









12

adverse effect wage levels have been controversial issues. In 1976 the General Accounting Office raised the issue relating to wage calculations for the 1974 apple harvest in New Hampshire. The certification process by the D.O.L. has also been criticized on several occasions. At a House oversight hearing in 1975, Congressman Ford, Democrat of Michigan, wondered why there was so much unemployment in the Washington, D.C., area and yet the D.O.L. had certified unavailability of domestic workers for the apple harvest in Maryland, West Virginia and Virginia (U.S. Congress, Senate, 1980). To the question of the H-2 workers causing depressed wages, the president of the Apple Institute testifying to the Senate Judiciary subcommittee in 1976 maintained that on the contrary the B.W.I. workers through hard work, were able to raise the effective wage above the required minimum (U.S. Congress, Senate, 1978). During the year 1977, the H-2 program was at a crossroads as a result of a series of events including (a) the proposed D.O.L. regulations attacked by both grower and worker groups alike; (b) the president's undocumented alien program which sought to regularize the status of certain illegal aliens accompanied by sanctions against employers of illegals and a review of the H-2 program; (c) a major court decision in Virginia that ordered the D.O.L. to certify 5,000 foreign workers for apple growers in 11 eastern states, the initial refusal of the D.O.L. to comply on the grounds that Puerto Rican workers were available and the subsequent capitulation;

(d) the Senate Select Committee on Small Business hearings on Agricultural Labor certification that documents the great frustration of growers with D.O.L. intransigence even in the face of monumental









13

effort and the attendant high cost of growers' attempts to obtain domestic workers.

The controversy over certification continued into the following

year, the highlight of which was the great fiasco of the D.O.L. attempt to recruit Puerto Rican workers. After expending a lot of effort and spending $275,000, only 19 persons out of the 554 referred remained to the end of the harvest in Virginia and West Virginia. The situation in four other states was even worse.

In the 1980s, the H-2 program has continued to be controversial. Issues relating to certification, working conditions and wages have aroused both debate and legal battles. While the growers have continually challenged the D.O.L.'s denial of applications for certification of foreign workers and have won court ordered approval in some cases, farm workers have increasingly contested the D.O.L. methodology of arriving at Adverse Effect Wage Rates for the various states. The battle has been going back and forth with the courts initially affirming the D.O.L.'s methodology only to have an earlier decision overturned upon an appeal.

Rising unemployment in the early 1980s coupled with the recession heightened concern about the influx of illegal aliens and the illegal alien problem was once again thrust onto public policy debate. The concern with illegal aliens resulted in the enactment of Public Law 99-603 (1986) to reform immigration and naturalization laws in the country. The Simpson-Mazzolli Bill of 1982 which was the precursor of the PL 99-603, originally envisaged a guest worker program to replace the H-2 program. The 1986 act, however, only amended certain









14

requirements of the H-2 program including cutting the required period for requesting H-2 workers to 60 days instead of 80. A seasonal agricultural program was also authorized which permits admitting up to 350,000 workers in the event of labor shortages.


















CHAPTER 2
REVIEW OF THE LITERATURE


The central issue in the debate on immigration and influx of

foreign workers is the economic and social consequence on the domestic economy. Reubens (1983) categorized models of international migration into five categories spanning the extremes of "open door" and "closed door" policies. The former subscribes to "humanitarian-egalitarian" principles and is totally committed to free market norms. In the opposite extreme "closed door" models are predicated on the complete exclusion of immigrants on the rationale that since total domestic income will have to be shared among a larger number of people, such immigrants lower real per capita income. Within this spectrum, the model that describes best the U.S. situation is what he terms the screen door and pressure-group economics. In this model "immigration is determined by strategic games played by small pressure groups rather than consensually decided by political or market interaction of our whole electorate or labor force"(Reubens 1983, p. 179). In the agricultural sector the pressure groups can be identified as the producers or producer groups who use immigrant labor in production. This view is in consonance with that held by Reder (1963), Martin and North (1984) and others who see foreign immigrant worker programs as constituting a form of tariff or protection resulting in some kind of


15









16

producer surplus to the employers. Quoting Reder(1963, p. 229) "regularly to grant special entry permits to certain groups of workers, whatever the reason, is to subsidize their employers and (perhaps) the consumers of their output, at the expense of the native labor market competitors." The detrimental effect on native labor of admitting foreign workers includes wage depression and labor displacement in jobs that otherwise would have been occupied by domestic workers. Martin and North (1984) further argue that the current H-2 program encourages labor "wastage." They assert that the piece rate pay system lessens supervisory costs and this enables growers to maintain an excess labor pool from which they practice occupational discrimination. The H-2 program, it is claimed, causes labor market segmentation and with the increased divergence between wages in H-2 impacted markets and other markets, the increased availability of labor allows sugar and apple producers to continue in traditional growing areas, areas labor autarky certainly would have dictated otherwise.

Foreign worker programs are also held to discourage adoption of, or to delay the adoption of new technologies. Martin and North maintain that the H-2 program and the availability of cheaper harvest labor does not encourage efforts at developing dwarf varieties of apple trees that would allow women and retired persons to be used as harvest workers or renew mechanization efforts for harvesting processing apples and sugar cane. Corroborating empirical evidence on this subject is provided from the case of the Bracero Program and adoption of mechanical cotton harvesters. Although, by 1958, mechanical cotton strippers were available, only thirty-five were being used by growers.









17

Following the decline of the Bracero Program after 1956 and the increased cost of Bracero labor, there ensued a dramatic shift toward mechanical harvesters, and at the time of termination of the program in January 1965, cotton harvesting was virtually completely mechanized (Morgan and Gardner, 1982).

Whereas the foregoing discussion has focused on the seemingly detrimental effect of foreign immigrant workers, researchers and proponents of liberalized immigration policies point to the beneficial side of the story. Ethier (1985) used a trade theoretic approach to develop a model to explain the purported beneficial effect of foreign migrant labor on the domestic economy in the presence of industrial fluctuations and import competing industries. He showed that host countries smooth out native labor employment fluctuations by a combination of commodity dumping and migrant dumping. The extent to which this balancing is successful is, among other factors, critically dependent on the elasticity of substitution between domestic labor and foreign labor in production. Even those opposed to allowing foreign workers into the economy recognize that certain advantages accrue to the receiving country in general. Reder (1963) recognized that, overall, immigration raised per capita income in the domestic economy through increasing total output and by stimulating further capital investments. As pointed out earlier, Reder's concerns were with the distributional impact of immigration which he finds detrimental. The view that immigration favors the distribution of income towards owners of nonlabor inputs (capital owners) and against labor is supported by Rivera-Batiz in his 1981 and 1983 studies. In a two-sector (a modern









18

sector categorized by production of an exportable good, a high administered wage causing unemployment and a traditional sector producing an import competing good, and characterized by flexible wages and use of migrant labor) model, Rivera-Batiz (1981) showed that immigration caused domestic labor displacement and unemployment, favors income distribution to non-labor inputs and had an ambiguous but possibly negative effect on natives. This latter negative effect, contradicts the generally held view of a beneficial overall effect of immigration on the domestic economy. However, within the particular framework in which his model is cast where employment in the modern sector is fixed, this conclusion is consistent. Rivera-Batiz also claimed that his model negated the view held by Cornelius (1978) and shared by Reubens (1979) that since unemployment rates are not generally high in parts of the country which traditionally have high concentration of foreign migrant labor, introduction of migrants does not lead to a decrease in welfare of domestic residents. He argued that immigration affects unemployment positively through intersectoral movements of the displaced workers from the traditional to the modern sector.

Further insights into the sectorial impact of the introduction of alien workers into the domestic economy is gleaned from the studies conducted by Johnson (1980), Chiswick (1982) and Grossman (1984). Using a general equilibrium framework and disaggregating labor into skilled, domestic unskilled and foreign unskilled, Johnson theoretically examined the probable effects of illegal immigration on the U.S. labor market. He showed that domestic low skilled workers









19

will lose as a result of immigration and this loss will be mainly through declining wages rather than through a loss of jobs. The earnings of domestic skilled workers will improve, as will the returns to capital. He further demonstrated the conditions under which a system of progressive taxes designed for an income transfer program, (an increasingly low unskilled labor wage may force some domestic unskilled labor to the welfare rolls) may hurt both highly skilled workers and owners of capital. This condition may exist only if there is a high degree of labor market inflexibility in the domestic unskilled labor market. Chiswick's (1982) analysis corroborated Johnson's arguments and advances the theme further. He argued that depending on the skewness of capital ownership structure in the receiving country, immigration may enhance income inequality in the domestic economy. At the same time the decline in the supply of labor in the sending country raises the marginal product of labor. In this case labor gains and capital loses, thus emigration may result in raising the general income level of the sending country and cause a narrowing of income inequality. The extent of the latter effect is again contingent upon the ownership structure of capital. A more highly concentrated capital ownership will be associated with a greater decline in income inequality as a result of emigration (Chiswick, p. 293).

Chiswick (1982), Krauss and Baumol (1979) agree on the subject of income transfers. Their conclusions from these analyses are that domestic workers whose incomes decline as a result of immigration can be compensated for their loss in welfare, since immigration will result









20

in an increase in the average income of the general economy. However, if immigrants are treated in the same way as domestic workers in the income transfer program, then the attempt to raise the welfare of the impacted domestic workers will be futile. It must be noted in passing, on this issue that, most studies on illegal aliens in the United States seem to indicate that such aliens do not make much use of the income transfer and social services systems, e.g., North and Houstoun (1976). In the case of legal foreign worker programs such as the Bracero and H-2 programs, workers are barred by law from participating in the welfare system.

While most of the foregoing survey of the literature on the

effects of immigration has been conducted on the theoretical level, empirical contributions to the debate may be found in the works of Wise (1974), Morgan and Gardner (1982) and Mehra (1984).

After the termination of the Barcero Program on January 1, 1965, a controversy ensued between agricultural growers and the then Secretary of Labor, W. W. Wirtz. The growers argued that they would not be able to harvest their crops without the Braceros and also that they could not find domestic workers willing and able to perform the arduous task involved in harvesting (stoop labor hypothesis). The Secretary of Labor responded that domestic labor could be found provided the growers would offer a decent wage. Wise's study tested the "stoop labor" hypothesis using data for two California crops, winter melons and strawberries. The estimated supply elasticities for domestic labor of

2.71 and 3.35 for the two industries, respectively, reinforce the Secretary of Labor's contention that a sufficiently high wage rate









21

would elicit adequate supplies of domestic labor. His analysis showed that in the absence of immigrants, growers made adjustments in melon production by reducing acreage planted by 26%, total production fell by 23%, also domestic workers gained a 262% increase in employment and wages increased 67%. Produce price increased 6% so consumers were less off.

Morgan and Gardner (1982) also evaluated on a more aggregate level the effects of the Bracero Program in order to provide policy insights for a prospective guest-worker program. Using data for the seven states that utilized most Braceros, they construct an econometric model to estimate the demand and supply relations in the labor market and the distributional effects of the program. Their analysis showed high demand (-0.8 to -3.2) and supply (0.2 to 1.0) elasticities for the domestic farm labor market. They estimated a wage depressing effect of about 8.8%, and a corresponding expansion in the numbers of farm workers demanded of 26.0%. The wage decline also caused an estimated 8% reduction in the quantity supplied of domestic farm workers. Employers gained by employing more workers at reduced wages. The total gain for the seven state area was estimated to be $185 million; the loss sustained by domestic workers was estimated to be $135 million-a net gain of $45 million to U.S. residents. Note that there seems to have been a beneficial effect on consumers, since farm prices fell by about 5%. The results of the analysis seem to indicate a more benign effect of the Bracero Program in contrast with that of Wise. Morgan and Gardner concluded from the magnitudes of the demand and supply elasticities that the economic impact of the proposed guest-worker









22

program would be sustainable. This portends good for the agricultural sectors which currently use foreign labor since the alternative to foreign labor importation would be importation of those agricultural products to replace domestic production.

Mehra (1984) and Emerson and Mehra (1985) contribution to the

"stoop labor hypothesis" and the broader immigration issue looks at why producers are able to maintain programs that allow foreign workers to be imported even though empirical evidence suggests that it is possible to obtain enough domestic workers for the tasks performed by the alien workers. The model used explicitly endogenized the bureaucratic policy implementation process that governs the H-2 worker program with respect to job certification and the determination of the Adverse Effect Wage Rate (AEWR). Their simulations showed that contrary to expectations, as the level of the AEWR increased, grower producer surplus increased. The conclusion derived from the analysis is that increased producer rents "provide incentive for producer group political influence seeking to maintain the regulatory environment that allows non-immigrant worker program" (Mehra 1984, p. xi).

The preceding discourse on the effect of the foreign migrant labor on the domestic worker has indicated that immigrants can adversely affect domestic workers by taking jobs away from them. Most of the theoretical studies have assumed that foreign workers substitute for domestic workers on a one-to-one basis. The empirical studies have arrived at their conclusion on the basis of simulations based on implicit or explicit assumptions about the degree of labor substitutability, but the crux of the whole issue of jobs displacement









23

by alien workers hinges on the degree of substitutability of immigrants in production compared with domestic workers. Although the literature is rich in studies that have examined the substitutability of labor under various forms of disaggregationl (e.g. blue collar versus white collar, educational status, age and unionionized versus nonunionionized) there exists a dearth of studies that have examined the substitution possibilities between domestic and foreign workers. The few studies that have a bearing on the immigration issue include Grossman (1982) and Borjas (1983).

Grossman's study has a direct bearing on the degree of

substitutability of immigrants and domestic workers. Disaggregating labor into natives with native parents, natives with foreign-born parents (second generation native workers) and foreign-born persons (immigrants) he estimated the Hicks elasticity of complementarity among the factors of production. He found that both second generation and foreign born workers were substitutes for native workers, though second generation workers were to a higher degree. Also immigrants substituted better for second generation workers than for native workers. This latter result was in consonance with Chiswick's (1978) result that immigrants become economically more similar to natives the longer they stay. The magnitudes of the estimated short run employment elasticities (with respect to quantities of immigrants) were -0.08 and

-0.04 for natives and second generation natives, respectively and indicated that the employment displacement impact of immigrants on




1 See Hamermesh and Grant (1979) for a critical survey of such studies.









24

natives was negligible. The study also showed a wage depressing effect accompanied this unemployment.

A methodological question that was addressed in this study was the issue of separability of capital and labor. The empirical finding was that under certain disaggregation of labor, labor is jointly separable from capital. Another study that has incidental bearing on the issue of the degree of substitutability between immigrants and domestic workers was that by Borjas. This analysis investigated the degree of substitutability between blacks, whites, and Hispanics. To the extent that the Hispanic community is of more recent origin and also for the fact that the immigrant population has been traditionally and predominantly of Hispanic origin, this study provided insights into the pertinent debate. The result of the study indicated no evidence of substitutability among the three groups. The implication was that the wave of Hispanic immigrants has not hurt blacks nor whites and that it may have rather helped these groups owing to the complementarity in production.
















CHAPTER 3
THEORETICAL FRAMEWORK

The Theoretical Model

The theoretical foundations of the international migration issue are rooted in the international trade theory and factor mobility literature. Mundell (1957) has shown that in the presence of international factor mobility, factors of production and commodity trade are substitutes. Based on this principle, it can be shown how international labor migration can occur. And, the consequences of this migration on both the source country as well as in the receiving country can be deduced.

The model that best describes the problem confronting us in the H-2 program is that developed by Krauss (1976) and extended by RiveraBatiz (1983). They developed a simple 2x2x2 (two factors, two goods, two countries) general equilibrium model to elucidate the guest-worker problem.

Consider two countries, the United States which shall be called the recipient country and the Caribbean the source country for H-2 workers. Let us also assume that the only factors of production in these countries are capital and labor, also the countries produce two goods; one tradable and the other non-tradable and labor intensive in production. The U.S. has a relatively larger capital endowment while the Caribbean has a relatively larger endowment of labor.



25











1011
Traldab les R

H V





H'
N




H' U E E.














12
1 II' 'rldables



Figure 3.1: Gains From International Labor Migration









27

Figure 3.1 illustrates the welfare effects of the guest-worker

program on the participant countries. The OH and OF origins represent, respectively, the source and recipient countries origin. HH and FF curves indicate their initial transformation curves. With H-2 workers moving out into the U.S. to augment the agricultural labor force, the labor exporting country's transformation curve moves inward towards the OH origin. Thus the H'H' curve represents the post-immigration production possibilities of the non-immigrant population in the sending country. Similarly, from the OF origin the F'F' curve represents the production transformation curve of the natives and immigrant labor after immigration has occurred. The initial welfare equilibrium position of the sending country is attained at the tangency of HH and a social indifference curve UlH at the point E, corresponding to the price ratio
H
PN
T
This price ratio is represented by a negatively sloped curve VV. Similarly, the initial equilibrium of the closed economy of the recipient country is point E'. The relative positions of points E and E' indicate that initially the price of the tradable good in the source country is higher than in the recipient country. In the preimmigration stage, the non-immigrant population in the sending country produces at point N and consumes at point D. At this level of production and consumption they have an excess demand for non-tradables and exchange their excess supply of tradables internally with the potential migrant population for the latter's excess supply of nontradable. They attain U1H level of satisfaction.









28

With international trade, the higher price of the good in the

source country would dictate that PJ of the non-tradable is exchanged for JC of the tradable from the recipient country in order to attain the consumption point C. Equilibrium point C would have resulted in the welfare gain for both countries, U2H U1H for the sending country and U2F U1F for the receiving country. However, because of nontradability of one of the commodities, this equilibrium position is non-viable. This welfare position can, however, be attained through international labor mobility, since factor mobility serves as a perfect substitute for commodity mobility. Labor is imported from the source country into the receiving country until the production point P' on the H'H' transformation curve is reached. This new equilibrium point is given by the tangency of F'F', the production possibility curve of the receiving country with its labor force augmented by migrant labor, and H'H' curve with the international terms of trade line T'T'. At this equilibrium point, it is obvious that non-migrants' economic welfare in the sending country declines from U1N to U2N; thus, gross domestic product declines in the sending country and rises by the same amount TT' in the receiving country. However, to the extent that the incomes of the guest-workers are repatriated to their country of origin, the national income of the sending country improves. The receiving country's economic welfare also increases by U2F UlF (at C, the slope of T'T' and TT are equal). Therefore, through labor export, the resultant trade equilibrium attained is Pareto optimal and from a









29

cosmopolitan view, economic welfare of the world has increased through labor factor migration.

While this general equilibrium framework of the theoretical

underpinnings of international migration affords a more complete view of the guest-worker problem, empirically this model would be very difficult to make operable. At the same time the magnitude of the H-2 foreign worker program is quite small in relation to the entire labor market. As a consequence, a partial equilibrium approach is proposed.

The labor market for migrant labor for harvesting in the apple industry is seasonal in nature. At harvest time, a need for extra labor arises to expedite harvesting of orchards. This demand is satisfied partially from domestic sources and the remainder comes from foreign migrant H-2 workers. Thus, two forces seem to be at work, there exists an excess demand for labor in the domestic economy and this is satisfied by a flow of excess labor from the sending countries. The wage differential between the receiving country (U.S.) and the sending country (the Caribbean) is the incentive that stimulates the foreign workers to offer their services to U.S. apple growers. If immigration were unrestricted, then labor would flow from the lower wage Caribbean to the higher wage U.S. until wages were equalized in the two areas though mobility costs and different tastes and preferences might result in equilibrium without equalization of wages. This equilibration comes about because as more labor flows into the domestic economy, wages will fall, because the marginal product of the additional workers falls. On the other side of the coin, wages in
















Wages Wages SSc

Sc




_* ____Si
Wc












0 Labor 0 Ls L s Ld LO L'd Labor

3.2A: The Caribbean 3.2B: United States
.Figure 3.2: International Labor Migration









31

the sending economy rise because the marginal product of the remaining workers rises.

The preceding analysis follows Sjaastad (1962) and Morgan and

Gardner (1982) and is depicted graphically in Figure 3.2. Figure 3.2A represents the demand (Dc) and supply (Sc) relationships for the Caribbean. At the prevailing wage Wc, an excess supply of labor exists in the labor exporting country. In Figure 3.2B, Wi is the preimmigration wage rate in the U.S. At this wage rate the supply of domestic labor falls short of that required for apple harvesting by Lod Los. The higher wage in the U.S. attracts migrant labor from the Caribbean. The augmented supply of labor shifts the labor-importing country supply curve SSOU to SS'u. Accompanying this expanded labor supply is a decline in the wage rate from Wi to W*. This scenario assumes that labor immigration is not restricted by the receiving country. However, the situation in the U.S. imposes quantitative restrictions on the number of immigrants who can enter the U.S. for harvest jobs through the job certification process. Assuming a fixed quota of immigrant workers such that total supply of labor is fixed at L'd, then the domestic wage will fall only to Wi** and OL's domestic labor will be supplied, the short fall in demand L'sL'd will be supplied through international labor migration.

The distributional consequence of this movement is illustrated in Figures 3.3A and 3.3B and is based on Chiswick (1982). In this model, capital and labor, the only factors of production, are assumed homogeneous but are not perfectly substitutable in production. Figures 3.3A and 3.3B represent the situation in the labor importing and labor














Wages
ages Sic Soc
Sou Slu


D E T

Wo F
oi A


lu C B Ic Q APL

APL

4 -- -- t
oc

N L
MPL


0 L L Labor C L L Labor
Uou lu C Ic oc

3.3A: U.S. 3.3B: The Caribbean Figure 3.3: Effect of International Migration on the Level and Distribution of Income









33

exporting countries, respectively. In the pre-immigration stage the supply of labor in the receiving country is Sou and it is Soc in the sending country, the wage rate as determined by the intersection of the marginal product curves and the supply curves are Wou and Woc respectively. Aggregate income during this stage in the U.S. as given by the area under the marginal product curve is OuDALou, of which the share of labor is OuWouALou, and the return to owners of capital contributes the remainder, i.e. WouDA. Similarly, the returns to labor in the sending country during the pre-immigration period are OcWocPLoc and capital share WocTP makes up the rest of the total national income OcTQPLoc.

With immigration, aggregate income increases in the receiving country; however, with the increasing number of workers both marginal product and average products of labor decline (assuming capital stock remains constant). The functional distribution of this increased aggregate income favors capital owners against native workers. (As shown, capital share increases from WouDA to W1uDB and although total labor share increases from OWouALou to OW1uBLlu, domestic labor share declines to OuWluCLou from OWouALou). On the other hand the outmigration of agricultural labor in the sending country results in an increase in the marginal product and average product of the remaining workers. Thus, labor gains and returns to capital owners decline. Note also that H-2 migrant labor also gains LouCBLlu of the increased aggregate income in the receiving country. If the income gained by the migrant workers is remitted back home as is likely to be the case with the H-2 program (the workers stay in the U.S. no more than three









34

months) then national income of the sending country also increases and the results of the partial equilibrium analysis accord with the general equilibrium results.

Labor and Product Market Equilibrium in the Domestic and Foreign Economies

A competitive labor market structure will be assumed for the

market for harvest labor. This assumption is justified because unlike the market for H-2 workers for the sugar industry where only one or a few grower associations contract labor for all the growers (a monopsonistic market structure), many growers participate in the recruitment of temporary alien workers for harvest jobs in the apple industry. The AEWR serves as a benchmark wage for recruiting alien workers. Figure 3.4 depicts the interaction between the factor markets (both domestic and foreign labor markets) and product market (domestic apple industry). The wage differential Wus Wc between the U.S. and the Caribbean translates the stock of excess labor into a factor flow that is made available to the apple harvest labor market in the U.S. This exogenous supply of labor causes a shift of the total domestic labor supply (residents and foreign workers) curve to the right from Si to S'i. The regulatory environment in the U.S. represented by the stipulation of an adverse effect wage rate serves to dampen the potential impact of the original wage differential. Thus the supply of foreign labor forthcoming is constrained to LAL'A. Total labor available for the apple harvest expands to OL'A. Thus augmented labor supply is purchased at a lower total cost than would otherwise have been the case in the absence of international labor migration. Consequently a producer surplus AEWRWousKJ accrues to producers. This





35

w D S



S us
AEWR 5 W1 LIS
Wc usD D
c D



0 Labor 0 L L L L** Labor
(A) (B)


P DP





W1
p A



D D


SQ 0 LA L Labor

(C)
(E)

Aggregate
Q**
Production
Q function








0 L* L' L Labor us Aus Labor
(D)
(A) The Caribbean Labor Market (B) U.S. Preharvest Labor Market
(C) U.S. Harvest Labor Market (D) U.S. Apple Industry (E) Product Market Figure 3.4: Labor and Product Markets Interaction









36

increased labor supply applied to the apple harvest results in a higher level of output. Figure 3.4D shows an expansion in output of Q-Q In Figure 3.4E the increased total product is depicted as a shift in the product supply curve to the right from Sp to S'p. The representation of the expanded output as a shift of the supply curve instead of a movement along the original supply curve is based on the observation that the causal factor is due to an exogenous increase in a factor of production, labor. The increased product supply causes a decline in product price. The gain in consumer welfare which arises from the price fall is represented by the area PP*usES.
















CHAPTER 4
EMPIRICAL MODEL

The structure of the labor market specified is comprised of the supply of labor and the demand for labor components. Equilibrium in this market is attained when the two sides are equalized. The supply side of the market is represented by a set of supply of labor equations that reflect the domestic and foreign labor components of the total labor supply of the apple industry. A production function approach was used to model the demand side of the labor market. This approach permits one to obtain information embodied in apple harvesting and also estimate parameters necessary for computing labor demand parameters.

Production Function

Various functional forms exist that have been used to model a production activity. The popular forms that have been traditionally employed in empirical work include the Cobb-Douglas (CD) and Constant Elasticity of Substitution (CES) forms. These two functions however exhibit constant substitution elasticities--in the case of the CD restricted to unity and for the CES restricted to be constant and equal between all inputs. Here a functional form that imposes few a priori restrictions on harvesting technology was employed, in order to compute the elasticity of substitution between inputs employed in apple harvesting.





37









38

Flexible functional forms permit one to model production

technologies with fewer maintained hypotheses. Among the common flexible functional forms that have been employed in empirical work are the generalized Leontief and quadratic forms which can be classified as linear flexible forms while the translog form belongs to the class of non-linear flexible forms (Lopez, 1985). This classification is based on the form the left hand side variable takes after a transformation. The choice between alternative forms for empirical analysis has been arbitrary. Caves and Christensen (1980) based this choice on numerical methods for determining the regular regions of different forms. Other methods employed for selection have been based on Monte Carlo studies (Guilkey, Lovell, and Sickles, 1983), Bayesian methods (Rossi, 1983), and parametric statistical methods (Berndt and Khaled, 1979; Applebaum, 1978). Lopez (1985) has shown that between the two classes of flexible functional forms the nonlinear form does not impose quasi homotheticity and separability restrictions a priori. For this study the translog form was selected. The translog functional form may be represented mathematically as

n
n "i n (Ej-=lfijlnXj )
(4.1) y = f(Xl,X2...--Xn) a Hi-1l Xi Hi-l Xi Explicitly, for the three input production function,

3 3 3
a1 22 a3 E i I.nXj E if2 nXj ;1 .nXj (4.2) y 0XI X2 X3 X1 X2 33



Log transformation of equation (4.2) gives the popular form in which this function is written, viz.,









39


3 3 3
(4.3) In y In a + E iailnX+ nXnX




The above algebraic formulation can be interpreted in two ways: namely as an exact production function, or as a second-order Taylor series approximation to a general but unknown production function. For empirical analysis the second interpretation of the translog formulation as a second-order Taylor series approximation to an unknown production function has been widely adopted. A major drawback of the translog formulation, and for that matter, most flexible functional forms, is that they do not globally satisfy certain desirable regularity conditions for a well-behaved production function, i.e., monotonicity in output, and convexity of isoquants. Berndt and Christensen (1973 p.85) maintain that "there are regions in input space where these conditions are satisfied. These well-behaved regions may be large enough so that the translog function can provide a good representation of relevant production possibilities." Thus an alternative is to empirically estimate the translog production function, and then, with the set of parameters obtained, check for satisfaction of the regularity conditions at each input level.

Monotonicity in inputs requires that for all inputs, the first derivative of the production function should be positive; i.e.,


fi > 0 Vi


The isoquants of the production function are strictly quasi-convex if









40

the bordered Hessian matrix is negative definite. For the translog function, taking the derivatives of the equation gives


n
(4.4) fi 8Y/aXi (y/Xi)(ai + z oijlnXj) j-i


a2y y n n
(4.5) fii - ( )[ii+(ai + Z pij1nXj)2 -(ai + Z PijlnXj)]
aX2i X2i j-1 j-1


82y y n n
(4.6) fij --- (- )[(ai + E ijlnXj) (aj + E ij lnXi)]
axiaxj XiXj j1 i-1 The bordered Hessian matrix is,


0 fl f2...... fn
fl fll f12 fin
F f2 f21 f22 f2n


fn fnl fn2...... fnn



F is negative definite if the successive principal minors Fi alternate in sign, starting with FI < 0.

The translog production function is non-homogeneous. The scale elasticity for this function is equal to the sum of the production elasticities. Taking the first derivative of equation (4.3) with respect to InXi provides us with the production elasticity


n
(4.7) ei ai + Z _ijlnXj j=-1









41

A sufficient condition for homogeneity is that the coefficients on the quadratic terms sum to zero in column or row directions:


n n n n
(4.8) Z -ij Z 1ij Z Z1ij 0
i=l j-1 i=l j=l


The homogeneous translog function is homogeneous of degree K Zai, For linear homogeneity this implies that the sum of ai equals unity.

As indicated earlier, the translog function does not assume

separability in inputs a priori. Thus one can test the hypothesis of input separability by imposing this restriction on the function. Berndt and Christensen (1973) have shown that weak separability implies the Allen Elasticity of substitution (4.9) aik ajk(i, j, e N, k a Ns) For the translog function, functional separability of inputs i and j from a third input k demands that the first and second derivatives of y satisfy the condition



(4.10) fifjk fjfik 0



This implies that



(4.11) eifkj ejfik 0



Global separability requires the condition above as well as



(4.12) Pim jk ik#jm = 0 (m 1....n)









42



Equation (4.12) reduces to


ei Pik Bim
(4.13) --
ej Pjk Pjm



if fjk and Pjm are not equal to zero.

Although the three input translog function can be estimated in the form



(4.14) In y aO + allnXl + a2lnX2 + a31nX3 + l11(lInXl)2 + :f22(lnX2)2

+ ;33(lnX3)2 + 012lnXllnX2 + 131lnX1lnX3 + 23l1nX2lnX3



by single equation methods, this is fraught with problems. For one, the number of parameters to be estimated increases rapidly as the number of inputs increase, and the additional quadratic terms present a potential source for multicollinearity problems. One may also be limited by the degrees of freedom if data are limiting. Empirical implementation of the translog functional form has in practice been via indirect methods by assuming profit maximization and competitive markets. The necessary condition for profit maximization in competitive markets is for the producer to employ the factor to the point where the value of the product at the margin equals the cost of the input; i.e.,



(4.15) fi ay/8Xi = W/Py









43

where W factor price (wage)

Py price of output y.



Equating (4.4) to (4.15) gives


n
(4.16) fi (ai + Z PijlnXj) y/Xi W/Py j =1

rearranging (4.16) and using (4.7)



WXi n
(4.17) ei - ai + Z OijlnXj
Pyy j-1



Thus, ei, the production elasticity, is also equal to the value of factor i relative to the value of output or the cost share Si. Making the assumption that any deviations from the optimal factor share are attributable to errors in trying to minimize cost, an additive error term ui is appended to each share equation. This renders the share equations stochastic.



(4.18) S1 al + lllnX1 + P121nX2 + 8131nX3 + ul (4.19) S2 a2 + 021lnXl + P221nX2 + P231nX3 + u2 (4.20) S3 a3 + 8311nXl + 8321nX2 + 6331nX3 + u3



For a continuous production function, the Hessian function with respect to input quantities is symmetric; i.e.,



ij fiji i,j 1,2,3.









44

This symmetry condition will not in general be satisfied and will have to be imposed on the estimating functions. Under the assumption of symmetry and linear homogeneity, the factor shares Si sum to unity and only eight of the twelve parameters in the system of three share equations are free. Any two of the equations will fully identify the parameters of the production function.

Selecting the first two equations for estimation purposes,

symmetry implies the following set of cross equation restrictions:



(4.21) P21 P12; 613 -(ll + i12); P23 -(022 + P21)



Using these restrictions the equations to be estimated are, therefore,



(4.22) Sl al + 611(lnX1 InX3) + 812(inX2 inX3) + ul (4.23) S2 a2 + 12(lnXl lnX3) + 822(lnX2 InX3) + u2



The foregoing discourse on the production function translates for the apple harvesting function the particular output function specification below:



(4.24) y f(DL,FL,HKT)

where DL Domestic hired picking labor in man months

FL Foreign, H-2 labor in man months

HKT Expenditure on all other inputs in the harvesting operation

except labor









45

In terms of equations (4.22) and (4.23), the equations to be estimated are



(4.25) SDL al + 11(1nDL InFL) + P12(lnFL lnHKT) + ul (4.26) SFL a2 + 012(lnDL 1nHKT) + 022(1nFL lnHKT) + u2 SDL and SFL represent domestic and foreign labor shares, respectively.

The parameters of the estimated share equations (also those of the underlying translog production function) enable one to compute such measures as the Allen Elasticity of Substitution (AES) and the demand elasticities between factors of production. The Allen partial elasticity of substitution between any two inputs in a multiple input production system is defined as

the effect on the relative factor quantities
of a change in relative factor prices, holding
output and other input prices constant. (Sato and
Koizumi, 1973, p.47)



This is given in general as
n
(E Xj f )
j-1
(4.27) aij (jFij / IFI)




where oij The Allen partial elasticity of substitution between

inputs i and j.

IFI The determinant of the Hessian in equation (4.26)

IFij I The i,jth co-factor of IFI



Under the assumption of a linear homogeneous production function,









46


n
z X fj y
j-1l
Employing the parameters of the translog production function aij is given as


IHijI
(4.28) aij
IHI

where for the three input case IHI is the determinant


0 H1 H2 H3

H1 H11 H12 H13

H = H2 H21 H22 H23

H3 H31 H32 H33 and

Hi Si i,j 1,2,3.

Hii 6ii + S12 Si Hij =" ij + SiSj

The constant-output factor demand elasticity Eij is given by


a1nXi
(4.29) Eij - Sjaij
alnWj



The Labor Supply Relations

The total amount of labor supplied to the economy is based on both short run and long run considerations. For any particular sector or industry at a particular point in time, the supply of labor to the sector is devoid of long run considerations like participation rates and investment in human capital. The determinant of total labor supplied in such a situation is dependent on the labor supplier's









47

decisions as to whether to provide his or her services to the particular industry or to an alternative industry, and then to what extent he or she is willing to provide those services, i.e. the supply of effort decision. While the occupational choice decision may be dependent among other things on the pecuniary rewards as well as geographical mobility, the intensity of work is a tradeoff between income and leisure.

In the agricultural industry where the demand for particular types of labor is seasonal, the ease or otherwise of labor supply hinges to a large extent on workers' mobility and preferences for employment in agriculture. The total supply of labor for apple picking in West Virginia is comprised of hired domestic workers and foreign workers from the Caribbean. The domestic component is made up of locals-mainly women, school children on vacation, and interstate migrant workers. Foreign workers are predominantly from Jamaica and are on contract under the H-2 program. The specification of the supply of labor function in the domestic market follows Tyrchniewicz and Schuh (1969), and is modified to reflect the particular situation of the apple harvest. The aggregate supply of labor for the harvesting market is postulated to be a function of the wage rate of farm labor, the alternative employment opportunities outside apple harvesting as indicated by returns to labor in the non-farm sector, the size of the civilian labor force, the unemployment rate in the economy and the level of unemployment insurance benefits. Algebraically, the supply of domestic hired labor DL, is given as: (4.30) DL f(FWAG,NFW,DLF,RUMP,UI)









48

where DL quantity of domestic labor services in man months

FWAG wage of hired labor in dollars per hour

NFW non-farm wage in dollars per hour

DLF civilian labor force

RUMP unemployment rate

UI unemployment insurance benefits

The supply of labor to the domestic agricultural sector draws from the same pool which provides unskilled labor inputs to other non-farm activities in the manufacturing, construction and service sectors. The inclusion of the non-farm sector wage in the supply function takes into account the alternative employment opportunities available to the hired laborer. A source of the domestic seasonal harvest labor supply would come from unskilled workers who have been temporarily laid off. The decision of such persons to be part of the labor pool available for harvesting activities would depend on the level of the unemployment insurance benefits they receive and also their willingness to jeopardize this source of income. Inclusion of the unemployment rate in the specification of the supply function is justified on the premise that whether one supplies his labor services to the agricultural sector involves explicit consideration of one's opportunity cost in alternative employment, with high levels of unemployment in the economy this opportunity cost may be approaching zero.

The supply of immigrant workers from the Caribbean is hypothesized to be a function of returns to labor, alternative earning opportunities in the Caribbean, the exchange rate between the U.S. dollar and Jamaican dollar and the unemployment rate in Jamaica. Jamaica is used









49

as the representative country for economic conditions in the Caribbean because about 80% of all the H-2 workers originate here. Prior to arriving in the U.S. their earnings potential is measured by the adverse effect wage which is the mandatory minimum wage that any employer could pay. In the U.S., the foreign workers are paid the same piece rate as domestic workers. Inclusion of the exchange rate in the specification is based on the justification that a sizable portion of the immigrants' earnings is remitted back to the home country. Under such circumstances a favorable exchange rate regime will be an inducement to supply more of their services. As an indication of the alternative market conditions in the foreign labor market, the minimum wage in Jamaica is employed in our specification. Alternatively, the per capita income on a national basis may be used. Formally the foreign workers supply function was specified as follows



(4.31) FL f(FWJ,MWJ,EXJ,JUMP)

where FL Quantity of foreign labor in man months.

FWJ Hourly earnings of foreign workers.

MWJ Minimum wage in Jamaican dollars per hour.

EXJ The exchange rate between the U.S. dollar and Jamaican

dollar.

JUMP Rate of unemployment in Jamaica.



The Adverse Effect Wage Rate

The adverse effect wage is a minimum wage, the compliance of which must be met by all employers who use both foreign and domestic workers.









50

The intent of instituting the AEWR in such markets is to ensure that wages offered in such markets do not cause a depression of wages of U.S. workers similarly employed in other occupations.

The adverse wage is generally announced prior to the production season by the Department of Labor. Various methods and criteria have been used to arrive at the annual hourly wage for each state. The stipulated wage is, however, supposed to reflect interests of various parties that are affected by or are the targets of the instrument. The interested parties include the government, represented by the D.O.L., producer groups, politicians and worker groups, and it is believed that the final wage decided upon is a compromise achieved through a political-bureaucratic bargaining process. Thus, the adverse effect wage rate is hypothesized to be a function of the previous year's level, AEWRt-1, a measure of the expectation of the wages of other similarly employed workers FWWVt-l, forecast crop size PRDt-1, expected crop price PRCt-1, and an index of producer group political influence. This specification follows Mehra (1984); however, we go further in testing the hypothesis that various political factions constituting the political interest in setting the AEWR have differing agendas and may affect the level of the AEWR differently. Equation (4.38) represents the adverse wage determination process.



(4.32) AEWR (AEWRt_.,PRDt-1,FWWVtl,PRCt-1,ICT1,ICT2)



The lagged AEWR reflects the behavior of bureaucrats' decision making. It is asserted that an element of stability in the wage levels is









51

introduced by bureaucrats in basing part of their decision on past levels of the AEWR. The other side of government's interest in the level of the AEWR is its regulatory objective of preventing a wage decline of other workers as a result of introduction of foreign workers. This objective is represented in the specification by the average wage rate for all hired farm workers lagged one season. The lagged output price PRC is also included to reflect the influence of producers' crop price expectations on the adverse wage. Producers are also likely to base their decision to influence the adverse wage in one direction or the other depending on their expectations of the crop size. Thus, the forecast crop size variable PRDtl is included as an explanatory variable to take account of this producer behavior. The variables ICT1 and ICT2 are indexes of congressional tenure, for congressmen representing the state and selected congressional committee chairmen, respectively. While it is expected that through political influence seeking, producers would be able through their congressmen to limit increases in the adverse effect wage, the actions of relevant committee chairs in the wage setting process are unpredictable.

Equations (4.33) to (4.46) summarize the specified labor market model.

(4.33) SDL al + 11(lnDL InFL) + 012(lnFL InHKT) + ul (4.34) SFL a2 + 012(lnDL lnHKT) + f22(lnFL InHKT) + u2 (4.35) InDL a3 + f31lnFWAG + P32lnNFW + P33lnDLF + 034lnRUMP

+ 6351nUI + u3

(4.36) InFL a4 + 841lnFWJ + 042lnMWJ + 8431nEXJ + 644lnJUMP + u4









52

(4.37) InAEWR a5 + 651lnAEWRt-1 + P52lnPRDt-1 + P531nPRCt-1 + 054lnFWWVt-1 + 055lnICT1 + 561nlICT2 + u5 (4.38) SDL EDL/TC (FWAG DL)/TC (4.39) SFL EFL/TC (FWJ FL)/TC (4.40) TC EDL + EFL + HKT (FWAG DL) + (FWJ FL) + HKT where EDL, EFL and HKT are respectively the expenditures on domestic labor, foreign labor and other inputs and TC is total expenditures on all inputs.

Data: Sources and Construction of Variables

Data for this study were obtained from secondary sources.

However, while some series were already published, the basic data for labor and wages were in "semi-processed" form and needed to be cleaned and organized. There was also the need to convert some of the variables by various transformations into forms that would be operational in the empirical model. This section explains the nature of the data and how various variables were handled. Employment Data

The basic source of data for hired labor used in the apple

harvest operation is the In-Season Farm Labor Reports otherwise known as ES-223 report of the U.S. Department of Labor, Employment and Training Administration.

The reporting period is for the preceding two weeks; data are collected on the numbers of persons hired in the apple picking operation. The data are disaggregated on the basis of whether workers are of domestic or foreign origin. Domestic workers are further broken down to local, intrastate migratory and interstate migratory, the last









53

group also identifies Puerto Rican contract workers separately. The data were converted to man months of labor by aggregating numbers of persons for two biweekly periods and finding the average. The manmonths of work performed for each season is the sum of monthly aggregates in each season. Typically there were approximately two months in each harvesting season.

Wages

Wages for labor employed in the apple harvest were obtained from the ES-223 In-Season Farm Labor Report, and from ES-232 Domestic Agricultural In-Season Wage Reports. Only domestic workers are covered in the wage survey. Wages are quoted on a piece rate basis of cents per bushel or per box. Efforts have been made to convert the piece rate to an hourly basis by weighting the total wage bill with the total number of hours worked in that piece wage rate category. The average hourly earnings resulting from this procedure are subject to error because the hours worked figure is only a rough tally and the production figures are less than precise. For the purpose of this study, the average hourly wage is computed by summing the wage bill in each piece rate category, and dividing by the total number of hours worked.

Average Wage Rate for All Hired Farmworkers

The wage rate of all hired workers by state is published by the U.S.D.A., National Agricultural Statistic Service in Farm Labor. The annual average wage rates are weighted averages of the four quarterly estimated wage rates weighted by the number of hours worked. A break in the publication of this series occurred from 1981 onwards. Data for









54

1981 to 1984 were estimated by using the following prediction equation based on data from 1968 to 1979.



A
(4.41) FWWV 0.519 + 0.174(NFW) + 0.103(TREND) (0.098) (0.013)

R2 0.96



Non-farm Wage Rate

The appropriate non-farm wage should reflect the wages that labor employed in the apple harvest would receive if they had offered their services in the non-farm sector. Hired labor in the harvesting operation is generally unskilled labor that can easily fill a variety of unskilled jobs in the manufacturing sector and in construction. Wages in the Food and Kindred Industry are used to represent alternative earnings. The series is published by the Department of Labor in the Handbook of Labor Statistics. Index of Congressmen Tenure and Seniority

This variable is included in the AEWR determination equation to take account of the political influence that apple producer groups exert on congressmen who are involved in setting the adverse effect wage rate. Since the wage is set by negotiation between the administration, represented by the Department of Labor, and other parties including congressional committees and other interested parties, it is hypothesized that producer groups would try to influence decisions in their favor by lobbying congressmen representing their districts. Pressure may be brought to bear on these congressmen in the form of campaign contributions to their re-election effort.









55

Congressmen in turn will intercede on the behalf of producer group interests with the bureaucratic agencies. Since congressmen control agencies' budgets, the bureaucrats are likely to be responsive to their demands in anticipation of "quid pro quo" when their budgets come up for review. The extent of the clout wielded by the congressmen in influencing decisions in their committees depends to a large degree on seniority in congress, and on the particular committee. The longevity of a politician in office is measured by the number of re-elections and depends on the incumbent's ability to marshall, among other things, political contributions for the re-election effort.

In constructing the index it is hypothesized that it is likely

that congressmen from the state or district in which production occurs may influence decisions differently than congressmen from other states and serving on committees that have a bearing on decisions concerning the setting of the adverse wage. Thus two different indexes were constructed: ICTI for state congressmen and ICT2 for congressmen from other states who chair relevant committees. The index was constructed by scoring for each congressman the cumulative number of years that he has been in office and an extra score for each year of re-election. For the state congressmen, an additional point is given for each year of being on a relevant committee. The committees that are considered important in setting the adverse effect wage are for the House, Agriculture, Judiciary, and Education and Labor; in the Senate, the committees are Agriculture, Appropriation, Finance, Labor and Public Welfare, and Judiciary.









56

Other Inputs

This variable is constructed as a residual, taking account of other miscellaneous inputs other than labor that are employed in the apple harvesting operation. It is envisaged that component inputs would include capital input services provided by expenses on fuels for operating trucks and tractors, depreciation on trucks, tractors, ladders and other equipment and other overhead, including interest payments.

Data on such inputs do not exist for West Virginia. However, cost and returns data have been collected for apple production for New York State, mainly Western New York, and are published in the Cornell University Agricultural Economics Research series. The relevant part of this series is used for this study based on the assumption that similar conditions exist for the two locations. Apple production in Western New York is predominantly for the processing market; the same applies to West Virginia as opposed to Eastern New York where a substantial part of production is for the fresh market. In essence, it is being maintained that with respect to "other inputs" the expenditures are similar for the two production areas. The cost data are on a per acre basis. Since total acreage data were not available for West Virginia the original data were converted to a per bushel basis and adjusted for total annual output for West Virginia.
















CHAPTER 5
EMPIRICAL RESULTS

Estimation

Equations (4.33) and (4.34) form the set of equations that were estimated to provide parameter estimates for the translog production model. In this aggregate model, labor input quantities and prices should be considered endogenous and their values jointly and simultaneously determined in the system. Thus, the simultaneity problem arises. Also errors of the individual share equations are not likely to be independent of each other but (contemporaneously) correlated. The two stage least squares approach will take care of the simultaneity problem and is employed to estimate the unrestricted model. It is noted however that the share equations are nonlinear in the endogenous variables and applying standard 2SLS procedures to a nonlinear system will not be appropriate.

The estimation procedure employed is based on a method developed by Amemiya (1974). Amemiya shows that the optimization problem is one of estimating values of the parameter vector 0 such that the objective function equation 5.1 below is minimized



(5.1) (y f)' D (D'D)-1 D' (y f)



where

Yt ft(zt,8) + ut

57









58

zt is a [(m + k) x 1] vector of m endogenous variables and k exogenous variables. ft is a nonlinear function in both zt and 0 having continuous first and second derivatives with respect to 0. D is a (t x K) unspecified matrix of certain constants with rank K. The choice of the elements in D may be based on any set of fixed variables or functions thereof. The nonlinear two stage least squares estimator is consistent and asymptotically normal. Amemiya's method is also applicable in the case where the ft is nonlinear in the variables but linear in the parameters.

As noted earlier, the share equations can be considered logarithmic marginal productivity relations of a well-defined production function if and only if their partial derivatives are symmetric in inputs. Symmetry of /ij, however, may not be attained in empirical estimation unless imposed a priori in the estimation. Cross equation restrictions in the share equations render individual share equations interrelated with other equations and their disturbances become correlated. Thus, imposition of symmetry restrictions requires the share equations to be estimated jointly.

Results of Estimation

Supply of Domestic Labor:

The original specification of the supply of domestic labor has as arguments the hourly farm wage paid apple pickers (FWAG), the non-farm wage (NFW), domestic civilian labor force (DLF), rate of unemployment (RUMP), and the level of unemployment insurance benefits (U.I.). For the initial estimation, however, the unemployment insurance variable was dropped from the specification because on examination of the series









59

it was observed that the level of unemployment insurance benefits virtually did not vary over the observation set. In effect, one would have been including another constant term as an explanatory variable. The specification of the supply of domestic labor is log-linear, thus the coefficient estimates are elasticities. Specifically the equation to be estimated is;

In DL 00 + ~lln FWAG + 621n NFW + 931n DLF + 04RUMP + e



where DL is domestic labor in man months, and e is an error term with zero mean and constant variance a2.

The results of the two stage least squares estimates, using as

instruments all predetermined variables in the system is shown in Table

5.1. Apart from the non-farm wage variable which showed a highly statistically significant t-ratio and the correct a priori sign, all the other explanatory variables exhibited high standard errors, indicating their coefficients are not significantly different from zero. At the same time, the F statistic (with 4 and 12 degrees of freedom) is 8.3, indicating that one should reject the hypothesis that all coefficients are zero. The occurrence of large standard errors of the coefficients coupled with a high F statistic is symptomatic of the presence of a multicollinearity problem.

Whether multicollinearity in the sample is harmful depends on the degree or severity of the collinearity problem. Various approaches have been adopted in the literature to measure the degree of multicollinearity. This includes using the determinant of the moment









60


Table 5.1: Parameter Estimates for the Domestic Labor Supply
Equation


Explanatory 2SLS Variables Coefficients Coefficients


Intercept -14.3670 17.4210 (9.1935)a (2.1285) FWAG 0.0202 0.6501 (0.0568) (0.2799) NFW -6.3590 (1.5752)

DLF 0.5054 (0.9045)

RUMP 0.0386 (0.2542)

NFWUN -4.9977 (0.9651)

a Standard errors in parentheses









61

matrix (X'X), a low value indicating a high degree of multicollinearity. Alternatively, since a small value of the determinant of (X'X) implies that at least one of the characteristic roots of the moment matrix is small, Besley, Kuh, and Welsch (1980) have suggested using the ratio of the largest to the smallest square roots of the characteristic roots of (X'X), the singular values of X, as a measure of ill-conditioning of the X matrix. This ratio is called the condition number and a value exceeding 30 is considered an indication of problematic multicollinearity. When the condition number is calculated for the model a value of 763.23, far in excess of the critical level was obtained.

Which method to choose to correct for an ill-conditioned data matrix depends on the specific problem at hand. Suggested solutions include, obtaining additional information by augmenting the sample size or using extraneous information such as extraneous estimates of elasticities, or constraining some parameters to take certain values a priori. A related approach is imposing additional restrictions on the parameters of the model. Other approaches include principal components and ridge regression.

The approach selected was to impose additional restrictions as follows. Since the demographic variables, labor force and rate of unemployment move closely together, they may be considered as one measure of employment by subtracting the rate of unemployment from one and multiplying by the domestic labor force. Alternatively, one can construct an indicator of non-farm opportunities by deflating the wage in the Food and Kindred Industry by the C.P.I. and adjusting for the









62

employment rate by multiplying the deflated non-farm wage by one minus the unemployment rate. The former method was adopted by Mehra (1984), while the latter method was employed by Schuh (1962) and Emerson (1976). The latter specification represented as NFWUN in the model was adopted here.

The supply of domestic labor function was estimated employing as regressors FWAG, NFWUN and DLF. Only the coefficient on NFWUN showed statistical significance, the FWAG and DLF estimated coefficients exhibited large standard errors. The model was reestimated, omitting from the regressors the domestic civilian labor force variable since its coefficient showed the largest standard error in the previous estimation. While this approach is likely to ameliorate the effect of a poor design matrix one must take cognizance of the possible specification bias that might be introduced into the estimation process.

The results of the domestic labor supply equation with these modifications are shown in Table 5.1. The standard error on the picking wage coefficient has been reduced and the coefficient is significantly different from zero at the 5% level. The sign of the coefficient also conforms to a priori expectation. This coefficient translates as the supply elasticity of domestic hired workers in apple harvesting. The magnitude of the supply elasticity implies that the supply of domestic hired labor is inelastic. This is consistent with Tyrchniewicz and Schuh's (1969) short run elasticity estimate of .621 and is on the lower side of both Emerson et al.(1976) and Morgan and Gardner (1982) who estimated an elastic supply of domestic labor. An









63

inelastic supply of hired labor has implications for the controversy between growers and D.O.L. and seems to reinforce the farmers' argument that domestic workers are difficult to attain; considerably higher wage incentives are required to elicit an adequate supply of domestic workers.

The adjusted non-farm wage variable (i.e. wages in the Food and Kindred Industry adjusted for unemployment) NFWUN has an estimated coefficient which conforms to a priori expectations with respect to sign. The value of the t-statistic shows that this variable is highly significant in explaining variations in the supply of domestic labor. The magnitude of the estimated coefficient implies that a 10% increase in the non-farm wage will induce a decline of 50%, in the hired labor services available to the apple harvest operation. That the elasticity of supply with respect to the non-farm wage adjusted for unemployment is highly elastic is collaborated by evidence from Tyrchniewicz and Schuh (1969), Glover (1971), Emerson et al (1976) and Morgan and Gardner (1982).

Two effects are at play in influencing labor supply, the non-farm wage and the employment rate. If unemployment in the economy as a whole is low, less labor will be available in agriculture for harvesting. High employment in the economy coupled with high wages in the nonagricultural sector will result in fewer workers being available for seasonal work in agriculture. This is true, for most of the hired workers are migrants or part time workers in agriculture. Full employment in other sectors of the economy would mean fewer lay-offs and therefore a reduced source of temporary domestic workers.









64

The linkage between the agricultural labor market and the

industrial labor market is facilitated by the relatively low embodied human capital required for labor transferring from manufacturing to harvesting or picking functions. The interaction of the two markets can have another effect apart from employment effects. In the absence of importation of foreign labor, a high non-farm wage elasticity of supply would mean that high wages in the non-farm sector would attract some labor away from agricultural activities like harvesting. This would imply a shift of the domestic labor supply curve to the left. If the demand for labor stays the same then the additional supply of labor would be forthcoming only at a higher wage. Thus, the events taking place in the macro economy can impact both wages and employment in the agricultural sector.

Supply of Foreign Labor

The H-2 migrant labor supply equation, like the domestic labor supply model, is in log-linear form. Initially, labor supply was postulated as a function of the real adverse effect wage rate for the current season (AEWR), the exchange rate between the Jamaican and U.S. dollar (EXJ), and the unemployment rate in Jamaica (JUMP). Alternative earning opportunities were measured by the per capita gross domestic product (GDP) since data on minimum wages in Jamaica were not adequate, and for the years that such data were available, the minimum wage remained virtually constant. Initial estimates of the parameters consistently gave negative values for the estimated coefficient of the adverse wage. This and other explanatory variables also exhibited large standard errors of estimates. Since the numbers of workers









65

employed in apple picking have been converted to a service or effort basis, it is plausible that the negative and statistically insignificant parameter estimate was indicating that the adverse effect wage rate variable was an inappropriate wage variable for the harvesting period.

Although during the harvest period both domestic and foreign workers are paid the same piece rate, it is also known that the two groups of workers have different productivities. Thus, the earnings per hour should differ for domestic workers as compared to foreign workers. In this study, however, productivity data do not exist for foreign workers to convert the piece rate to an hourly rate. It must be noted that the source of hourly earnings for domestic workers namely the In-Season Wage Survey of the Department of Labor does not include foreign workers in the survey. The best one could do under such circumstances was to assume the same hourly earnings for domestic and foreign workers with the caveat that one would be estimating an errorin-variable, model a derivation of which is made below.

Let the foreign labor supply equation be represented as

Y1 Y171 + X1 I + el

where Y1 is the true foreign wage and is related to the observed foreign wage as Y1 = Y1* el where el is the measurement error.

Yl = (Y1* el)71 + X1P + el

In the first stage of 2SLS an estimate of Y1 is given as
A A A
Y* = Y ul ul is the estimated reduced form error.

But the expectation of Y1* is the same as that of Y1, XH1, where Hl is
A
the vector of reduced form parameters. The use of YI in the second









66

stage therefore purges the measurement error problem. Since the measurement error occurs in FWAG which is an endogenous variable, the simultaneous equation estimation of the foreign labor supply equation takes care of the bias introduced. The measurement error in the foreign worker earnings proxy variable FWAG is expected to cause an underestimation of the foreign worker earnings. The results of estimating the foreign labor supply equation with the following set of variables FWAG, GDP, EXJ and JUMP as explanatory variables are presented in Table 5.2, column 2. The proxy for foreign labor earnings FWAG has an estimated coefficient which is statistically different from zero. It also has a correct positive sign. The magnitude of the estimated coefficient indicates that the supply of labor is very responsive to wages. The signs of the coefficients on the remaining explanatory variables are plausible; however, none of the coefficients are statistically significant at the 5% significance level. The standard errors on these parameter estimates are large. Inspection of the correlation between the explanatory variables indicates high intercorrelations among the variables. The condition number (192.2) again suggests an ill-conditioned data matrix. Restrictions were imposed related to the three variables per capita gross domestic product, exchange rate and the unemployment rate, transforming them into a single measure of alternative employment opportunities in Jamaica. First the per capita income was converted to U.S. dollars by dividing G.D.P. by the exchange rate. The G.D.P. in U.S. dollars was then adjusted for unemployment by multiplying it by the Jamaican unemployment rate.









67

Table 5.2: Parameter Estimates for the Foreign Labor Supply Equation.

Explanatory 2SLS Variables Coefficients Coefficients Set (A) Set (B)


Intercept 12.776 14.6510 (6.8157)a (3.5420) FWAG 2.1209 2.0910 (0.9900) (0.9352) EXJ 0.2178 (0.4414)

GDP -0.1443 (0.3537)

JUMP 0.2704 (1.1858)

JEMX -- -0.2014 (0.1020)
a Standard errors are in parentheses.









68

Table 5.2 column 3 shows the results of the estimate of the

modified foreign labor supply equation. Both explanatory variables do show statistical significance at the 5% level. The supply elasticity with respect to earnings is elastic. This result is not surprising at all. The differential between wages in the U.S. and Jamaica is very large and therefore Jamaican laborers will exhibit a high response to any demand for their services. The wage elasticity being elastic implies that economic incentives alone would be adequate to induce Jamaican laborers to pick fruit in the U.S. The constraint on exporting all their services is policy induced by the U.S. certification process.

Mehra (1984) estimated the supply elasticity for H-2 workers

employed in sugar cane harvest to be 2.4; this is about the same as our estimate of 2.1. Although Morgan and Gardner's (1982) estimate far exceeds the aforementioned estimates, one arrives at the same conclusion that supply of foreign labor is very responsive to the farm wage in the U.S.

The variable JEMX, as indicated previously, is a composite of the per capita gross domestic product, the exchange rate, and the rate of unemployment in Jamaica. The estimated coefficient is significant and bears a negative sign. This inverse relationship between the measure of alternative employment opportunities and the supply of labor suggests that as the adjusted per capita income in the economy rises there will be more opportunities in the economy for labor with expanded economic activities. Jamaicans would therefore supply less labor for work outside their country. The magnitude of the parameter estimate,









69

however, indicates that this relationship is inelastic. Thus, a ten percent change in JEMX would bring only about a 2% decline in labor supply to the U.S. This observation is consistent with the implications deriving from the wage supply elasticity. The strong positive impact of the destination country wage is expected to override any improvements in employment opportunities in the sending country. The Adverse Effect Wage

The adverse effect wage rate equation is specified as


AEWR aAEWR 11 PRD21 PRC1 FWWVt 1 ICTl15 ICT26 where AEWR is the deflated hourly adverse effect wage for West Virginia, AEWRt-1, PRDt_l, PRCt_l and FWWVt_l are the lagged values of the adverse effect wage, output, price of output and statewide hired farm wage. ICTI and ICT2 are measures of congressional tenure and stability for state congressmen and congressional committee heads, respectively. For purposes of estimation the above equation is cast into a log-linear form by taking logs of both sides. The log-linear form

Ln AEWR In aO + lln AEWRt.l + 021n PRDt_l + P31n PRCt.l +

P41n FWWVt_ + P51n ICT1 + P61n ICT2 +e

was estimated by ordinary least squares method. The disturbance term e has zero expected mean and variance a2. The results of the estimation were disappointing with most of the coefficients exhibiting large standard errors. It was suspected that multicollinearity was a problem and the variable with the largest standard error, lagged output, was dropped from the specification. On reestimation of the model the results in Table 5.3 column 2 were obtained. Although the standard









70


Table 5.3: Parameter Estimates of the Adverse Effect Wage Equation.


Predetermined O.L.S. Hatanaka's Estimator Variables Coefficient Coefficient


Intercept 0.4176 1.5275 (1.0990)a (0.4679) AEWRtl 0.6224 0.9572 (0.2730) (0.1401) FWWVt-1 0.2518 0.1959 (0.1914) (0.06781) PRCt-_ -0.0002 (0.0489)

ICT1 -0.1233 -0.1279 (0.0702) (0.0217) ICT2 -0.05217 -0.0953 (0.0493) (0.0212) R2 .40 .91 D.W. 2.52 2.31 Durbin-h n.a. a Standard errors are in parentheses. n.a. Not applicable because n(Var(3l)) is greater than one.









71

errors of the estimated coefficients declined somewhat and the t-ratios improved, only the lagged adverse effect wage variable showed statistical significance. The magnitude of the estimated coefficient of lagged apple price is very close to zero and still exhibited a large standard error. Since a lagged dependent variable is among the explanatory variables, serial correlation of the error term is suspected of being a problem. As the model stands now, one suspects a multicollinearity problem probably due to misspecification of the model and autoregressive residuals. In the presence of the two problems any test of the latter must be robust to the misspecification problem.

Godfrey (1987) has proposed a testing strategy based on a sequence of ordered hypotheses representing increasingly restrictive assumptions about the true data process, starting from a hypothesis about misspecification through autocorrelation, to a situation of no problem. For this model, it is assumed any misspecification may be due to inclusion of the lagged output price variable and therefore omitting that variable from the estimation will take care of the collinearity problem.

Since the data are annual, the form of autocorrelation in the residuals is likely to be of the first order. In the presence of lagged endogenous variables in the model, the conventional DurbinWatson (D.W.) test for first order autocorrelation will be inappropriate since this test was derived under the assumption of a nonstochastic X matrix. (In the presence of lagged endogenous variables appearing among the explanatory variables the D.W. test tends to take a value around 2.) The alternative Durbin h-test (equation









72

5.2) although derived as a large sample test, is employed to test for first order autocorrelation in the regression.



(5.2) n
h r
l-n(Var(pl))

where

r the estimate of serial correlation coefficient from

the O.L.S. regression.

Var(1) the estimated sampling variance on the coefficient of

the lagged dependent variable in the O.L.S. regression.

n sample size (17)



This test, however, is not applicable in this case because the product of n and the variance of the coefficient of the lagged adverse effect wage variable is greater than unity, i.e. n Var(,1) 1.242 > 1. Maintaining the assumption of first order serial autocorrelation, ordinary least squares estimation of the model is no longer appropriate. The existence of contemporaneous correlation between the lagged dependent variable and the error term renders ordinary least squares estimates inconsistent and asymptotically inefficient. Consistent estimates of the parameters of this model can be obtained by applying an instrumental variable method or maximum likelihood estimator. Hatanaka's (1974) two step 'residual adjusted' estimator is an instrumental variable approach and is asymptotically efficient. This method is employed in estimating the model and the results are presented in the second column of Table 5.3.









73

There is a marked improvement in the remaining coefficient estimates and all are statistically different from zero at the 5% level. The coefficient estimate for the lagged adverse effect variable has increased dramatically in value from 0.6224 to 0.9572. This indicates that the current value of the adverse effect wage depends highly on the previous year's level: over 95% of the current period level of the adverse effect wage is accounted for by the previous year's level. The magnitude of the coefficient of the lagged dependent variable is a pointer to the decision behavior of the parties involved in determining the annual level of the AEWR. With considerable bureaucratic input in the wage determination process, one would expect a high element of stability in decision making based on past experiences and bereft of any radical changes. Mehra(1984) found similar large influences of the lagged AEWR on the current level. The lagged level of the statewide farm wage seemed to also influence the adverse wage determination process, the coefficient being significant at the 5% level. In view of the legislative directive that AEWR should be set at a rate comparable to other similarly employed workers, one would expect this variable to have assumed a strong impact. As much as the variable is an average for all hired farm workers and the term "similarly employed" is debatable, the variable used in the regression may not be accurately representing hired labor which is paid by the piece rate. Regardless, the results give an indication that the directive is being taken into consideration in the wage determination process.









74

The low standard errors on the proxy variables ICT1 and ICT2

serving as a measure of the political influence of producer groups in influencing the adverse effect wage determination process, indicate that these producer groups exert a measure of significant influence on the level of the wage. As expected, producer's group influence through representatives and senators would be directed to keeping the wage level low so as to cut down on their cost of production. The negative signs of the coefficients on the indexes of congressional tenure bear out this expectation. The magnitude of the parameter estimates seems to imply that congressmen representing producers directly from the state of interest do exert stronger influence than committee chairpersons who do not necessarily come from that state. To confirm the above observation a test of the equality of the two parameters was conducted. The F-test with 2 and 10 degrees of freedom was 449.15. Thus the null hypothesis that the two parameters are equal was rejected. The question of committee chairpersons' interests being probably different from native congressmen is not borne out by the signs on the coefficients; rather the two groups seem to reinforce each other's actions. A combined measure of the two groups' impact on the wage level also shows that they both work to influence the wage level in the downward direction. Similar negative influence of the producer group's action through congressmen on the level of the adverse effect wage rate was found by Mehra (1984).









75

Share Equations

The domestic and foreign share equations were estimated by 2SLS estimation procedures. These equations in the unrestricted form are given as SDL 1 + 11 in DD + P12 In FF + P13 In HKT

SFL P2 + P21 in DD + 022 ln FF + 823 ln HKT

The results of the estimation of the unrestricted share equations are presented in Table 5.4, column 2. As mentioned earlier, because the first and second derivatives vary with input levels, there is no guarantee that the isoquants of the translog function are globally convex. There is the need to check the estimated production function to ascertain whether or not it is well behaved. To assess the behavior of the estimated function, one has to test whether the monotonicity and convexity conditions are satisfied. Monotonicity was tested by examining the predicted factor shares using the parameter estimates and checking whether they assume positive signs. The predicted shares are shown in Table 5.5, and are all positive at all observations. The convexity assumption is tested by examining the production function for negative definiteness of the bordered Hessian matrix. For the three input production function this implies that the sign of the principal minors should alternate starting with a negative. Table 5.6 indicates that this condition is not satisfied at all data points. Thus, the convexity assumption is violated for certain points in the sample.

Generally, without the prior imposition of the symmetry condition on the parameters the translog function will not be well behaved. Symmetry is therefore imposed on the parameters of the two share









76


Table 5.4: Parameter Estimates of the Translog Function.

Constrained Coefficients Unconstrained Set(A) Set(B) Parameters Coefficients


Constant -0.0592 0.3810 0.3788
(0.2599)a (0.0633) (0.0641)

611 0.1796 0.1459 0.1456
(0.216) (0.0129) (0.0103)

812 -0.0637 -0.0796 -0.0801
(0.0274) (0.0102) (0.0104)

P13 -0.0577 -0.0662 -0.0655
(0.0219) (0.0223) (0.0226) Constant 0.0300 0.3913 0.3887
(0.2020) (0.0547) (0.0555)

821 -0.0515 -0.0796 -0.0801
(0.0168) (0.0102) (0.0104)

822 0.1420 0.1276 0.12715
(0.0213) (0.0106) (0.0107)

823 -0.0422 -0.0479 -0.0471
(0.0170) (0.0190) (0.0192) Constant 0.9707 0.2277 0.2325
( ) (0.1143) (0.1159)

P31 -0.1281 -0.0662 -0.0655
( ) (0.0223) (0.0226)

832 -0.0783 -0.0479 -0.0471
( ) (0.0190) (0.0192)

#33 0.0999 -0.1142 -0.1125
( ) (0.0401) (0.0406) a Standard errors are in parentheses.









77

Table 5.5: Predicted Shares from the Estimated Share Equations.


SDL SFL HKT


0.35567532 0.08974245 0.55461504 0.29210814 0.19621723 0.51171298 0.28284351 0.17961676 0.53757577 0.24044682 0.18060144 0.57898792 0.24450752 0.19993850 0.55558944 0.20278174 0.21368209 0.58357100 0.22923910 0.18861845 0.58217618 0.25536896 0.20137467 0.54329194 0.27144955 0.12895683 0.59962490 0.20670563 0.21888332 0.57444574 0.16211921 0.26584640 0.57207165 0.20036129 0.25196657 0.54771027 0.23807835 0.24467282 0.51728639 0.26076667 0.19737376 0.54189502 0.29418603 0.16139580 0.54445240 0.29116775 0.20968977 0.49917837 0.32139564 0.17271184 0.50592729









78

Table 5.6: Principal Minors of the Bordered Hessian Matrix


-0.12650493 -0.00815221 0.00612132 -0.08532716 0.00235576 -0.00135337 -0.08000045 0.00048975 -0.00030033 -0.05781467 -0.00076977 0.00057130 -0.05978393 0.00047739 -0.00031209 -0.04112043 -0.00076888 0.00069253 -0.05255057 -0.00071787 0.00055141 -0.06521331 0.00105718 -0.00068066 -0.07368486 -0.00337978 0.00280593 -0.04272722 -0.00039735 0.00040643 -0.02628264 -0.00208709 0.00202933 -0.04014465 0.00040432 -0.00000229 -0.05668130 0.00287041 -0.00159009 -0.06799926 0.00102001 -0.00065754 -0.08654542 -0.00076914 0.00055391 -0.08477866 0.00362140 -0.00198459 -0.10329516 0.00106109 -0.00053762









79

equations and the constrained model is estimated by 2SLS procedure. Table 5.4 column 3 presents the result of the constrained estimation. The estimated function satisfies the monotonicity requirement, however, like the unconstrained model, convexity is not satisfied at all observations in the sample. The chi-square test (one degree of freedom) of the symmetry restriction assumes a value of 0.0014. This implies that the imposition of the symmetry of the 6ij's on the model is justified. It must be noted that the symmetry test also implies a weak test of linear homogeneity. Thus, the model also satisfies linear homogeneity.

Several problems can cause a production function not to be well behaved. These include first, errors in measurement of variables, secondly misspecification of the function (i.e. the translog model may be a poor representation of the harvesting function). For the present estimation one is inclined to ascribe the first problem as critically affecting the production function. Errors of measurement of various variables loom large in the model. In obtaining the aggregate measure used in the estimation, there were biweekly labor input data points for both domestic and foreign labor for which there were no data. Missing values were obtained by taking the means of adjoining data points. There was also discretionary judgement in determining the starting and ending points to obtain the aggregate labor quantity from the biweekly data.

The method of calculating the factor shares, assuming a constant

adjustment factor for number of hours worked for each year, definitely introduces an element of error in these variables. For the foreign









80

labor share variable domestic wage was used as a proxy for foreign wage. This results in an understatement of the foreign labor share. Since it is generally held that foreign workers' output per hour is generally higher than domestic workers', converting the same piece rate to an hourly rate should result in a higher foreign wage than domestic wage.

The preceding discussion indicates that measurement errors exist for some of the variables. However, the variables so.far discussed are endogenous; therefore one has a measure of comfort in the fact that in the simultaneous equations model, the errors will be absorbed into the disturbance terms. The same thing cannot be said of the "other inputs" (capital input) variable however. This variable is considered exogenous in the system of share equations. The existence of an exogenous variable measured with error in a system of simultaneous equations presents further problems for identification and estimation (see Geraci 1983, Hsiao 1976). Generally, measurement errors in the independent variables result in biased and inconsistent parameter estimates. A derivation of the error model is shown below.

Let the system of structural relations be

r y X + u -----(1) where

y and X are the g x g and g x k matrices of endogenous and

exogenous variables, respectively. r and P are the g x g and g x k coefficient matrices.

E(u) = 0, E(uu') Zuu positive definite E(X) 0, E(XX') Zxx positive definite









81

E(uX') 0.

X is not observable but X* is observed where

X* X + e ----(2) e is a k x 1 vector of measurement errors on X.

E(e) O0, E(ee') Zee

E(eX') 0, E(eu') 0

Substituting (2) into (1) results in

ry p(x* e) + u ry fx* + (u Pe)

since the disturbance term (u pe) is not independent of X*,

E(rylX*) pX*



The situation is aggravated in nonlinear relations (Griliches and Ringstad 1970). Although various procedures have been suggested in the literature to handle such problems, the paucity of empirical applications of these can be explained by the complexity of the suggested methods.

An instrumental variable approach was employed to reestimate the share equations dropping the "other inputs" variable from the set of instruments. This approach is based on the rationale that one is able to find somewhere in the system a predetermined variable that is measured correctly and is correlated with the variable that is affected by measurement errors, but uncorrelated with the disturbance term. The estimated constrained coefficients (set B) are similar to that from the previous estimation that included "other inputs" variable as instrument. For further discussion this set of parameters will be









82

adopted. The computed t-ratios on the estimated parameters indicate that all the parameters are statistically highly significant at the 1% level.

Separability

The estimated share equations with symmetry and linear homogeneity imposed were tested for separability of inputs. For the three input production function, four types of separability are identified. These are domestic labor and foreign labor are separable from "other inputs"; domestic labor and "other inputs" from foreign labor; foreign labor and "other inputs" from domestic labor and complete separability. The functional separability of labor in harvesting from other inputs is of more interest to us in this study. If labor is functionally separable from "other inputs" then one can omit "other inputs" in the functional specifications. Since this variable is measured with considerable errors, its omission would help the estimation. As discussed previously, if the parameters 013 and 023 are non-zero, functional separability between domestic labor and foreign labor and "other inputs" imply the following restrictions

"l 023 "2 P13 0



f11 i23 021 013 0



In terms of the estimated parameters these imply

aI 612 "l P22 + 2 611 + "2 P12 0

2
P12 Pll 22 0









83

Table 5.7: Results of Parameter Analysis of Input Separability. Parameter Estimate Standard Error T-Statistic


1 623 2 P13 0 0.0272 0.0029 9.2020 i11 23 21 613 0 -0.0123 0.0043 2.8264




Wald Test for the hypothesis that the derived set of parameters are jointly zero:

chi-square 608.72 with 2 degrees of freedom









84

A Wald test was performed to ascertain whether the above conditions were satisfied, in which case one would accept that the labor inputs are functionally separable from other inputs and drop the latter from the explanatory variables of the harvesting function. The result of the test is shown in Table 5.7. The null hypothesis of each of the restrictions summing to zero should be rejected. The Wald test that the derived parameters are jointly zero is also rejected. Thus functional separability of labor from other inputs is rejected. Elasticities

Estimates of the Allen partial elasticity of substitution (A.E.S.) are presented in Table 5.8. Column 2 provides estimates of the elasticity of substitution between the two categories of labor; domestic and foreign. One notices that the signs of the estimated parameters span both negative and positive values. Since the question as to whether two inputs i and j are substitutes or complements depends on the sign of the A.E.S., one is unable unequivocally to make such inferences. Further examination of the A.E.S., however, indicates that all the positive estimates correspond consistently with data points in the sample that satisfy the regularity conditions of convexity of isoquants. By extension it can be inferred that at observations where the translog function is well behaved the sign of the A.E.S. indicates that domestic labor and foreign labor are highly substitutable in picking apples. As already noted, the observed sample data of the translog model, unlike the C.E.S. functions whose elasticities are solely a function of the regression parameters, do not provide repeated









85

Table 5.8: Allen Partial Elasticities of Substitution 1968-1984.

Year aDL FL aDL HKT aFL HKT

1968 -4.2801 2.0158 0.0241 1969 22.3433 1.5566 2.2220 1970 104.0769 -0.1109 5.0555 1971 -58.9911 1.4710 2.3610 1972 104.8139 4.1947 -0.8792 1973 -9.0196 -1.7393 4.6827 1974 -61.6184 0.5360 3.3668 1975 47.1528 2.6029 0.8894 1976 -11.8034 2.3011 0.7570 1977 -82.4311 -4.5055 6.8206 1978 -15.6158 -1.6180 3.1998 1979 1382.9586 192.5790 -144.9039 1980 19.4465 3.4692 0.0441 1981 48.6052 2.2609 1.2719 1982 -55.8812 3.7562 -2.1626 1983 15.0490 1.8241 1.8110 1984 53.8661 -0.4616 6.6404









86

observations on a single set of elasticities. For statistical inferences, one needs to know the distribution of the estimated parameters.

Anderson and Thursby (1986) have demonstrated that only elasticity estimates based on the means of the actual shares are likely to follow either the normal or ratio-of-normals distribution function. The estimates of the elasticity of substitution based on the means of the input shares and using the parameter estimates (Set B) from the restricted translog function are presented in Table 5.9. The calculated value of over 83 for the elasticity of substitution between domestic labor and foreign labor conforms to the observation made earlier that the two categories of labor are highly substitutable in harvesting of apples. The magnitude of the A.E.S. also confirms the notion held by most observers of the harvest labor market that domestic labor is perfectly substitutable with foreign labor. If foreign labor is highly substitutable for domestic labor as the data seem to indicate, then the fears of domestic workers of being displaced from employment are well founded. In the absence of any restrictions on the importation of foreign labor and given the seemingly attractiveness of H-2 foreign workers to growers, one would expect that growers would compose their labor with a high proportion of foreign workers, thereby forfeiting job opportunities for domestic workers. Note this result derives from a micro perspective; from the general equilibrium view it might be that displaced workers would be absorbed more efficiently in the system. Thus, importation of foreign workers could be welfare increasing.









87

Table 5.9: Allen Partial Elasticities of Substitution and Price Elasticities of Demand Computed at Mean of Input Shares.

Domestic Foreign Other Labor Labor Inputs



Domestic .-68.80 83.31 2.59 Allen Labor Partial
Elasticities Foreign -112.2 0.86 of Labor Substitution Other Inputs -1.51


Domestic -17.59 Price Labor Elasticities
of Foreign -21.78 Demand Labor

Other -0.83 Inputs









88

The sign of the computed Allen elasticity of substitution between domestic labor and the variable "other inputs" indicates that the two inputs are substitutable in harvesting. A similar relationship exists for foreign labor and "other inputs." The exhibited relationship is rather surprising, since the "other input" category is comprised of expenditures on fuel, truck and equipment that complement the labor inputs in the harvesting operation. Table 5.9 also presents price elasticities of demand computed from the relation Eij aijSj, where Eij is the demand elasticity. The results indicate that both domestic and foreign labor demand are highly elastic. This is contrary to the generally held view that during the harvest season the demand for labor is generally inelastic. The magnitudes of the estimated demand elasticities are, however, consistent with the high substitution elasticities computed between domestic and foreign labor. The prevailing view that the demand for labor is inelastic is correct in so far as the total labor demand is concerned, i.e. in a situation when there are no viable substitutes. The results of this analysis show that as a result of the high substitutability between the two labor categories, the demand for each component of labor is highly elastic.

During the harvesting period for apples the weather conditions tend to fluctuate considerably with the possibilities for heavy rains and snow disrupting planned activities. To the extent that such uncertainties affect crop price expectations and thus the growers' decision environment, growers would be expected to make adjustments to the use of labor in harvesting. Although they might not have much room for maneuverability in relation to numbers of persons employed, they









89

are able to spread the time of arrival of H-2 workers and then the intensity of effort by workers can be adjusted also. The piece rate system of payment definitely facilitates growers' ability to adjust the intensity of labor use. The elastic demand result of the estimates may thus be reflecting this ability of growers.















CHAPTER 6
SIMULATION

In this chapter the estimated labor market model of the apple harvest in West Virginia is simulated to a) evaluate how well the model can track the actual endogenous variables--Historical Simulation;

b) evaluate the behavior of the model under alternative scenarios assuming alternative sets of values for policy variables.

Historical Simulation.

The harvest labor market model estimated in the preceding

chapter was "validated" by testing how well the computed values of the endogenous variables approximate the actual values. For simulation purposes it was assumed that the quantity of foreign labor employed is exogenously determined instead of being jointly determined in the model as previously assumed in the estimation. The justification for doing this is based on the nature of the policy environment in which the H-2 program operates where the D.O.L has the authority to limit the final number of foreign workers allowed into the country irrespective of the demand by producer groups. The set of equations employed in the simulation model were the domestic and foreign labor share equations and the supply of domestic labor equation. The adverse effect wage determination equation was not included since it does not form part of the simultaneous system.



90









91

The solution of the simultaneous equation model was determined using Newton's method. Simulation is static in that actual values of the exogenous variables are used in the solution process. The criteria selected for assessing the goodness of fit were (1) the mean percent simulation error:


(1/n) E=l1 (Yt Yt) 100 / Yt

where

Yt actual value of the endogenous variable at time t
A
Yt predicted value of the endogenous variable at time t

t 1, 2, . n

n 17

and (2) the percentage of instances in which the direction of change of the observed variable coincided with the predicted variable.

The results of the historical simulations are presented in Table

6.1 with the actual variables. It is evident that the predicted values of the endogenous variables are remarkably close to the observed variables. The results of the validation measures show that none of the predicted variables had a mean percent simulation error over 1%, specifically the mean percent simulation errors for DL, SDL, SFL and DFWAG were respectively -0.35, -0.26, 0.17 and 0.087. Also, apart from the DFWAG for which 90% of the directional changes were correctly predicted, all the other variables had 100% correctly predicted directional changes.

The actual and predicted values of the endogenous variables were plotted against time to graphically depict their movements over time









92

Table 6.1: Actual and Simulated Values of the Endogenous Variables.


YEAR DL EDL SDL ESDL

1968 2293 2218.51270 0.36423 0.35698 1969 2500 2469.15649 0.33080 0.32749 1970 1929 1927.05957 0.30860 0.30810 1971 1926 1912.17432 0.23776 0.23705 1972 1477 1482.71167 0.25967 0.25729 1973 1176 1228.63281 0.20260 0.20902 1974 1199 1275.09497 0.22041 0.23176 1975 1493 1489.42944 0.27371 0.27049 1976 1277 1281.07495 0.23653 0.23814 1977 1107 1111.07263 0.18946 0.19130 1978 1168 1164.42285 0.17176 0.17099 1979 1523 1466.85327 0.20736 0.20083 1980 1484 1479.52930 0.25347 0.25218 1981 1510 1518.03259 0.23195 0.23363 1982 1646 1653.78943 0.26484 0.26649 1983 1500 1555.31982 0.28247 0.29040 1984 1667 1654.46008 0.31335 0.31344



YEAR SFL ESFL DFWAG EDFWAG

1968 0.10526 0.10344 0.05284 0.05294 1969 0.21789 0.21804 0.05438 0.05418 1970 0.19261 0.19252 0.05648 0.05637 1971 0.17595 0.17712 0.05743 0.05768 1972 0.21004 0.20776 0.05846 0.05715 1973 0.20766 0.20532 0.05807 0.05771 1974 0.17794 0.17588 0.05535 0.05559 1975 0.21388 0.21211 0.05636 0.05526 1976 0.11731 0.11743 0.05268 0.05303 1977 0.19671 0.19808 0.04837 0.04890 1978 0.27353 0.27351 0.05682 0.05662 1979 0.26396 0.26511 0.05278 0.05257 1980 0.26782 0.26678 0.05030 0.05000 1981 0.17368 0.17400 0.03995 0.04016 1982 0.14331 0.14352 0.04091' 0.04109 1983 0.20087 0.20064 0.03747 0.03760 1984 0.16216 0.16431 0.03807 0.03846

Simulated values are prefixed with E.






Domestic Labor
2.

2.3 2.2 2.1

2.0

1.9


1.8
0
1.7

1.6

1.5

,I 1.4 1.3

1.2

1. 1- -----111968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 Y enr Figure 6.1 Domestic Labor: Actual vs. Simulated

]Actual Domestic Labor tSimulated Domestic Labor




Full Text

PAGE 1

APPLE HARVESTING AND NONIMMIGRANT ALIEN WORKERS IN WEST VIRGINIA: AN ECONOMIC ANALYSIS KOFI ADU-NYAKO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1988

PAGE 2

my country, Ghana, and my parents, Yaw Nyako and Yaa Asab

PAGE 3

ACKNOWLEDGEMENTS I wish to express my appreciation and thanks to Dr. Robert Emerson, chairman of my committee, for all the help in bringing this study to fruition. I also acknowledge the diverse contributions the other members of my committee Dr. Carlton Davis, Dr. Larry Kenny and Dr. Max Langham made. Dr. Tim Taylor assisted me with various aspects of the production model and I am thankful. The government of Ghana has supported me financially, at various stages of my education for which I am very grateful. I also appreciate the financial support from the Food and Resource Economics Department and from USDA Cooperative Agreement 58-3J23-4-00189 with the University of Florida. To my friend Abena, for your inspiration and encouragement I am very grateful. My appreciation also goes to James, Liz, Kwasi, Asamoah and Joe for their friendship in Gainesville. My thanks go to Berenda, Gwen and Dottie for all the secretarial help they gave me and also to Hope. iii

PAGE 4

TABLE OF CONTENTS ACKNOWLEDGEMENTS iii ABSTRACT vi CHAPTERS 1 INTRODUCTION 1 Problem Statement 1 Statement of Objectives 4 Historical Background of U.S. Temporary Alien Worker Program 5 World War II Period: 1941-1947 6 1948-1951 7 1951-1965 8 1965-present 11 2 REVIEW OF THE LITERATURE 15 3 THEORETICAL FRAMEWORK 25 The Theoretical Model 25 Labor and Product Market Equilibrium in the Domestic and Foreign Economies 34 4 EMPIRICAL MODEL 37 Production Function 37 The Labor Supply Relations 46 The Adverse Effect Wage Rate 49 Data: Sources and Construction of Variables 52 Employment Data 52 Wages 53 Average Wage Rate for All Hired Farmworkers 53 Non-farm Wage Rate 54 Index of Congressmen Tenure and Seniority: 54 Other Inputs 55 5 EMPIRICAL RESULTS 57 Estimation 57 Results of Estimation 58 Supply of Domestic Labor: 58 Supply of Foreign Labor 64 The Adverse Effect Wage 69 Share Equations 75 Separability 82 Elasticities 84 iv

PAGE 5

6 SIMULATION 90 Historical Simulation 90 Alternative Scenarios 97 7 SUMMARY, CONCLUSIONS AND LIMITATIONS OF THE STUDY 100 Summary and Conclusions 100 Limitations of The Study and Suggestions for Further Research 104 APPENDICES A DATA FOR CONGRESSIONAL TENURE INDEX 107 B DATA FOR THE MODEL 110 REFERENCES 116 BIOGRAPHICAL SKETCH 121 v

PAGE 6

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy APPLE HARVESTING AND NONIMMIGRANT ALIEN WORKERS IN WEST VIRGINIA: AN ECONOMIC ANALYSIS By KOFI ADU-NYAKO APRIL 1988 Chairman: Robert D. Emerson Major Department: Food and Resource Economics This study is an analysis of the interaction of foreign migrant H-2 workers with domestic workers in the apple harvest labor market in West Virginia. At issue is whether growers require supplemental labor that is not obtainable from domestic sources and what impact such imported labor exerts on the domestic economy. A labor market model including the supply of domestic and foreign labor as well as the demand component represented by a translog harvesting function was constructed to gain insights into the controversy. Results of the analysis indicate that the supply of domestic labor is inelastic, lending support to the growers' argument that domestic workers are difficult to attain. The tight domestic labor market for harvest labor is aggravated by it being interlinked with the non-farm sector labor market. The evidence from the study is that domestic labor is highly responsive to wages in the manufacturing vi

PAGE 7

sector thereby attracting labor from the agricultural sector. The supply of foreign labor was found to be very responsive to farm wages in the U.S., an attestation to the attractiveness of the large wage disparity between the U.S. and the Caribbean that serves as an incentive to immigrate. The issue of possible job displacement by foreign workers was investigated by estimating the Allen elasticity of substitution between domestic and foreign workers in the harvesting operation. It was found that the two categories of labor are highly substitutable in production; the computed elasticity of substitution is 83. This is evidence that barring any restrictions on the importation of foreign labor, its use could completely substitute for domestic labor in harvesting. This high substitution elasticity is accompanied by a high demand elasticity for both domestic and foreign labor. vii

PAGE 8

CHAPTER 1 INTRODUCTION Problem Statement The flow of foreign workers into a nation's economy has aroused a feeling of xenophobia among various groups of nationals. This xenophobia is due in part to the fear harbored by natives that aliens may cause them economic displacement, whereas the concern by natives for the effects of influx of alien workers is a persistent phenomenon, xenophobic tendencies reach a crescendo during bust periods of the economic cycle and subside or become latent in the upswing periods. Accompanying the heightened concern of citizens are calls for action by the legislative body to stem the tide of the influx of alien workers. In the U.S. the issue of immigrants both legal and illegal has been a continuing problem. While certain sections of the economy look upon the presence of alien labor as necessary and desirable, other members of the economy see immigrants as potentially destabilizing both in the economic and social spheres. The latter contend that aliens place undue burden on the provision of social services for which they may not contribute. On a more serious front, temporary alien workers are seen to cause postponement of necessary economic adjustment in the system (Martin, 1979; Briggs 1978). Advocates of liberalized immigration, however, assert that there exists an economic need for 1

PAGE 9

2 alien labor in certain occupations for which local supply of labor is in short supply or for one reason or another is unavailable In agriculture, alien farm workers have been used since the 19th century when farm labor was recruited from China, Japan and the Philippines. Mexican immigrants have been crossing the border for work in the U.S. since before World War I (Morgan and Gardner 1982) A more structured program, the Bracero program, was instituted during World War II and nonimmigrant Mexicans were employed to meet the increased demand in both farm and non-farm occupations for the war effort. The program was terminated in 1964. The H-2 program, which was initiated in 1952 and currently exists, is another example of a foreign agricultural work program that utilizes alien workers in certain agricultural enterprises, particularly in vegetable and fruit and nuts production. Most of the labor employed in this program is seasonal in nature, filling the excess labor demand during the harvest season. Undoubtedly there exists an excess demand for labor in certain agricultural jobs and at certain times of the production season. The question then arises whether this excess demand exists only because of the presence of friction in the domestic labor market that prevents the domestic supply from responding to the demand. Are there institutional structures and / or characteristics of the domestic labor market, the removal of which would result in the domestic labor market being in equilibrium? On the other hand, if the shortage of labor is "real" and there is a need for supplementing domestic labor supply, then what

PAGE 10

3 would be the consequence on the domestic market of the importation of foreign labor? These questions lie at the heart of the whole immigration debate. Answers to these questions will enable informed policy decisions to be made. Recent efforts by the U.S. Congress seek to streamline U.S. immigration policy. The Immigration Reform and Control Act enacted in November 1986 makes provision for a Special Agricultural Workers Program. Under this provision seasonal agricultural workers who can show proof that they worked at least 90 days in the year prior to May 1986 would be granted temporary alien status pending permanent status. This provision also allows for replenishment of temporary residents in case of labor shortage so long as they work in agriculture for at least 90 days. Such persons are also eligible for future adjustment to permanent status. It is envisaged that for the next three years, the duration of the above provision, not more than 350,000 persons would be eligible. For an agricultural labor market consisting of 2.6 million persons (USDA, 1983) 1 the proposed figure is a very significant proportion, and will most likely have a significant impact on the labor market. The impact is likely to be more pronounced if it is considered that these workers are earmarked for only a section of the entire agricultural labor market, i.e. the seasonal harvesting labor market. Ex ante, one cannot assess the direction and magnitude of the impact of the program. However, the H-2 program provides an avenue to It is debatable whether this figure includes illegal aliens, st people would argue that a substantial number of aliens are not included.

PAGE 11

4 explore some of the positive and normative issues likely to be associated with implementation of a foreign worker program. Recent works in evaluating the effects of foreign worker programs on the domestic agricultural economy include Morgan and Gardner (1982) Mehra (1984) and Emerson and Mehra (1985). Morgan and Gardner's work evaluates the welfare impact of the Bracero Program. This research assumes away a key question about the substitutability between foreign and domestic agricultural workers. The extent that foreign labor is substitutable for domestic workers in production is fundamental to determining the rationality of domestic producer's preference for foreign workers in certain agricultural jobs, e.g., sugarcane harvesting, apple harvesting and tobacco production. Emerson and Mehra study the effects of the H-2 foreign worker program in the sugar industry in Florida. The substitutability issue could not be addressed in this study because virtually no domestic labor is used in sugarcane harvesting. The present study intends to include the above neglected dimensions into an economic analysis of the impact of the H-2 foreign worker program. The analysis will be conducted within the context of the apple industry in the State of West Virginia. Statement of Objectives The general objective of this study is to evaluate the impact of the H-2 foreign worker program in the apple industry on producers and workers. In order to achieve the above objective an economic model of the apple harvest labor market will be constructed embodying the demand and supply relationships that pertain and incorporating existing

PAGE 12

5 institutional arrangements that impact on this market, e.g., Adverse Effect Wage Rate (AEWR) Secondly, from an aggregate production function of apple harvesting, the short run elasticity of substitution between domestic and alien workers will be estimated. In modelling the harvest labor market, analysts have assumed implicitly that decisions on harvest labor factor use are taken separately 'from other inputs used in the harvesting operation. This separability assumption will be tested explicitly in the models. Finally, analysis of the policy implications arising from the results of the empirical investigations above will be considered. Historical Background of U.S. Temporary Alien Worker Programs Although the first large scale temporary alien labor program was initiated in 1917, the stream of alien workers into the U.S. was an on going phenomenon at the turn of the 20th century. This request for supplemental labor was in response to the increased demand for labor for the World War I effort and the ensuing shortage of labor in agriculture and railroad construction. This first program of temporary alien workers comprised mainly Mexicans estimated at about 80,000 persons and a few Bahamians and Canadians. While the railroad component of the program was officially terminated at the end of the war in 1918, the agricultural employment part was extended to 1921. The legislative authority under which the program was authorized is the ninth proviso of section 3 of the Immigration Act of 1917. The Labor Department set rigid regulations for the employer in order to obtain alien workers; however, enforcement of these regulations was

PAGE 13

rather lax and this resulted in large numbers of Mexicans deserting their original agricultural employers for more lucrative pursuits in industry. In the 1920s Mexican immigration quickly expanded and within six years had doubled from 486,000 to almost 900,000 in 1926 (U.S. Senate, 1980). This growth occurred in spite of the official termination of the alien worker program in 1921. Kiser (1972) attributes the growth to the large influence that employers of Mexican labor had on policy making, albeit informally. The advent of the 1930s and the accompanying economic depression precluded any need for supplemental labor. Mexicans were repatriated, some under very unpleasant conditions. This process of repatriation resulted in a lot of resentment from the Mexicans. World War II Period: 1941-1947 Prior to the U.S. entry into the war in December 1941 there existed a labor surplus situation. However, the war caused an exodus of rural farm workers to higher wage industries producing for the war effort which resulted in labor shortages in agriculture. As a consequence supplemental labor was requested from Mexico. The Mexican government was reluctant and the subsequent detailed agreement wrangled out of the U.S. government was due in part to their sensitivity to the humiliating treatment meted out to their nationals after the termination of the postWorld War I program. Also, the U.S. wanted an agreement which prevented adverse effects on domestic workers due to the importation of foreign labor. Like the first program, the ninth proviso of the 1917 Immigration Act was the authority under which Mexicans were admitted contingent on

PAGE 14

7 certification from the U.S. Employment Service that local workers were not available. Admission of other foreign nationals from the Bahamas and Jamaica was similarly under the ninth proviso. The agreements for the admission of these other foreign nationals unlike the Mexican Bracero Program involved merely a memorandum of bilateral agreements between the governments. These agreements formed the original basis of the British West Indies (B.W.I.) Program. Public Law 45, enacted in 1943, and the various intergovernmental agreements were the basis of foreign labor supply until December 1947. During this period, the U.S. government paid all transportation and recruitment expenses and was also the labor contractor. The Department of Agriculture administered the foreign labor supply program. Braceros were employed in cotton, sugar beets, fruits and vegetables and were concentrated mainly in the West and Midwest. B.W.I, workers were used mainly in the Eastern seaboard states and were extensively involved in the agricultural program of the War Food Administration. 1948-1951 The emergency war time legislation ended in December 1947. Importation of Mexican Braceros proceeded under frequently revised international agreements. B.W.I, and Bahamian workers were contracted under waiver provided by the ninth proviso. The administration of the foreign worker program reverted to the U.S. Employment Service. While the B.W.I. Program was generally assessed favorably, the Mexican segment was criticized as causing wage depression in areas where Braceros were employed (see U.S. President's Commission on Migratory

PAGE 15

8 Labor, Migratory Labor in American Agriculture, Washington, 1951 p. 51). 1951-1965 The Bracero Program As a response to the dissatisfaction with the Mexican Program, Public Law 78 was enacted in 1951 and provided the legal framework under which the Bracero Program operated until its termination in December 1964. Basically this law required the U.S. government to set up recruiting centers at or near the border, provide transportation and medical care from Mexican recruiting centers to the U.S. centers, and provide guarantees of performance by employers under terms of the law. Employer responsibilities under the law included provision of transportation to and from government recruitment centers, free housing and meals at reasonable cost. They were also required to pay prevailing wages in the area and guarantee employment of workers during 75% of the contract period. Craig (1971) maintains that most of the guarantees of the Bracero Program were empty guarantees since their implementation was contingent on domestic workers enjoying same and such were nonexistent. Between 1952 and 1956 the Bracero Program expanded rapidly with the number of workers peaking at 445,197 in 1956. Part of the reason for this increased use of Braceros was due to a crackdown on illegal aliens in Operation Wetback that resulted in an increase in the apprehension of illegal aliens and made such aliens less available for use by growers who wanted to avoid the higher cost of using legals

PAGE 16

9 From 1956 the Department of Labor required stricter housing regulations and issued prevailing wage rates for certain areas. These regulations, effectively increased the cost of using Braceros and curtailed the demand for them. During the latter part of the fifties the program came under increasing opposition from labor, welfare groups, and the Secretary of Labor. The main criticisms were that Braceros were adversely affecting wages, as evidenced by wages remaining stagnant for ten years in areas of Bracero use. In contrast, national wage levels had increased about 46% during the same ten year period. Also the availability of Mexican workers and the preference for them by some growers had caused job displacement of domestic workers. In the early 1960s opposition to the program quickly increased. The Kennedy administration tightened regulations that sought to protect domestic workers. An Adverse Effect Wage Rate was instituted for all states that used foreign workers. This is a mandatory minimum wage that must be paid by all employers to both domestic and foreign workers alike and was generally higher than prevailing wages. It became increasingly difficult for the Bracero Program supporters to gain congressional support for its renewal. The program was finally terminated in December of 1964. The B.W.I. Program Whereas PL78 was the legal basis for the Bracero Program until its demise in 1964, all other temporary labor programs including the B.W.I. Program derived their legislative authority from PL 414, the Immigration and Nationality Act of 1952. This act grants the Attorney General "sufficient authority to admit temporarily certain alien

PAGE 17

10 workers, industrial, agricultural, or otherwise, for the purpose of alleviating labor shortages as they exist or may develop in certain branches of American productive enterprises, particularly in periods of intensified production" (U.S. Congress, House of Representatives, 1952). Whereas final legal authority as to who will be admitted and who will be rejected reposed in the Justice Department's Immigration and Naturalization Service (INS), the Department of Labor (D 0 L. ) had an advisory role as to certification of the need for supplemental labor. During the fifties and early sixties the expanded Bracero Program overshadowed the B.W.I. Program in importance. While the former program attracted much adverse attention, the B.W.I. Program was looked upon favorably. For example, in 1953 Senator Holland of Florida contrasting the B.W.I. Program to the Bracero Program commented that the B.W.I. Program has entailed minimal government expense and supervision and was both simple and acceptable to both the employers and employees. Again in 1957 a House Subcommittee report found the importation of agricultural workers from the British West Indies beneficial for agricultural interests in the U.S. While the B.W.I. Program was small in size compared to the Bracero program, B.W.I, workers were concentrated in small areas along the Atlantic Seaboard and contributed immensely in agricultural harvesting activities ranging from citrus in Florida, to apples in the Virginias and potatoes and cranberries in the northeast. There was a steady increase in the B.W.I, contract workers from the latter part of the fifties and the early sixties.

PAGE 18

11 1965 to Present At the termination of the Bracero Program in December 1964 it was anticipated that the B.W.I. Program would be expanded to accommodate a larger foreign worker program. Contrary to expectations, the Department of Labor issued new regulations which sought to curtail even further importation of foreign labor under the H-2 program. These included higher wages than previously for domestic workers, more housing and transportation benefits and limiting the certification period to 120 days. Attempts in the courts to weaken or overrule the D.O.L. regulations did not succeed. Secondly, amendments proposed in Congress in 1965 to remove certification authority from D.O.L. to the more sympathetic Secretary of Agriculture failed. The 1965 amendment to the Immigration and Nationality Act has served as the basis for importation of temporary foreign labor to the present. Major revisions were made and published in March 1978. Some of the major developments and controversies relating to the H-2 program and the apple industry in particular will be highlighted in the following paragraphs. In the sixties the criticism leveled against the D.O.L. was a perceived laxity in the administration of the temporary alien worker program. Rising unemployment during the seventies caused the H-2 program to be under persistent attack from labor groups and congress. In 1970 a complaint of exploitation of domestic migrant farm workers was leveled against the Rural Manpower Service of the D.O.L. A 1972 review by D.O.L. of the program found that the impact of foreign workers in apple producing areas was insignificant. Wage determination and whether the prevailing piece rates measure up to the mandated

PAGE 19

12 adverse effect wage levels have been controversial issues. In 1976 the General Accounting Office raised the issue relating to wage calculations for the 1974 apple harvest in New Hampshire. The certification process by the D.O.L. has also been criticized on several occasions. At a House oversight hearing in 1975, Congressman Ford, Democrat of Michigan, wondered why there was so much unemployment in the Washington, D.C., area and yet the D.O.L. had certified unavailability of domestic workers for the apple harvest in Maryland, West Virginia and Virginia (U.S. Congress, Senate, 1980). To the question of the H-2 workers causing depressed wages, the president of the Apple Institute testifying to the Senate Judiciary subcommittee in 1976 maintained that on the contrary the B.W.I, workers through hard work, were able to raise the effective wage above the required minimum (U.S. Congress, Senate, 1978). During the year 1977, the H-2 program was at a crossroads as a result of a series of events including (a) the proposed D.O.L. regulations attacked by both grower and worker groups alike; (b) the president's undocumented alien program which sought to regularize the status of certain illegal aliens accompanied by sanctions against employers of illegals and a review of the H-2 program; (c) a major court decision in Virginia that ordered the D.O.L. to certify 5,000 foreign workers for apple growers in 11 eastern states, the initial refusal of the D.O.L. to comply on the grounds that Puerto Rican workers were available and the subsequent capitulation; (d) the Senate Select Committee on Small Business hearings on Agricultural Labor certification that documents the great frustration of growers with D.O.L. intransigence even in the face of monumental

PAGE 20

13 effort and the attendant high cost of growers' attempts to obtain domestic workers. The controversy over certification continued into the following year, the highlight of which was the great fiasco of the D.O.L. attempt to recruit Puerto Rican workers. After expending a lot of effort and spending $275,000, only 19 persons out of the 554 referred remained to the end of the harvest in Virginia and West Virginia. The situation in four other states was even worse. In the 1980s, the H-2 program has continued to be controversial. Issues relating to certification, working conditions and wages have aroused both debate and legal battles. While the growers have continually challenged the D.O.L. 's denial of applications for certification of foreign workers and have won court ordered approval in some cases, farm workers have increasingly contested the D.O.L. methodology of arriving at Adverse Effect Wage Rates for the various states. The battle has been going back and forth with the courts initially affirming the D.O.L. 's methodology only to have an earlier decision overturned upon an appeal. Rising unemployment in the early 1980s coupled with the recession heightened concern about the influx of illegal aliens and the illegal alien problem was once again thrust onto public policy debate. The concern with illegal aliens resulted in the enactment of Public Law 99-603 (1986) to reform immigration and naturalization laws in the country. The Simpson-Mazzolli Bill of 1982 which was the precursor of the PL 99-603, originally envisaged a guest worker program to replace the H-2 program. The 1986 act, however, only amended certain

PAGE 21

14 requirements of the H-2 program including cutting the required period for requesting H-2 workers to 60 days instead of 80. A seasonal agricultural program was also authorized which permits admitting up to 350,000 workers in the event of labor shortages.

PAGE 22

CHAPTER 2 REVIEW OF THE LITERATURE The central issue in the debate on immigration and influx of foreign workers is the economic and social consequence on the domestic economy. Reubens (1983) categorized models of international migration into five categories spanning the extremes of "open door" and "closed door" policies. The former subscribes to "humanitarianegalitarian" principles and is totally committed to free market norms. In the opposite extreme "closed door" models are predicated on the complete exclusion of immigrants on the rationale that since total domestic income will have to be shared among a larger number of people, such immigrants lower real per capita income. Within this spectrum, the model that describes best the U.S. situation is what he terms the screen door and pressure -group economics. In this model "immigration is determined by strategic games played by small pressure groups rather than consensually decided by political or market interaction of our whole electorate or labor force" (Reubens 1983, p. 179). In the agricultural sector the pressure groups can be identified as the producers or producer groups who use immigrant labor in production. This view is in consonance with that held by Reder (1963), Martin and North (1984) and others who see foreign immigrant worker programs as constituting a form of tariff or protection resulting in some kind of 15

PAGE 23

16 producer surplus to the employers. Quoting Reder(1963, p. 229) "regularly to grant special entry permits to certain groups of workers, whatever the reason, is to subsidize their employers and (perhaps) the consumers of their output, at the expense of the native labor market competitors." The detrimental effect on native labor of admitting foreign workers includes wage depression and labor displacement in jobs that otherwise would have been occupied by domestic workers. Martin and North (1984) further argue that the current H-2 program encourages labor "wastage." They assert that the piece rate pay system lessens supervisory costs and this enables growers to maintain an excess labor pool from which they practice occupational discrimination. The H-2 program, it is claimed, causes labor market segmentation and with the increased divergence between wages in H-2 impacted markets and other markets, the increased availability of labor allows sugar and apple producers to continue in traditional growing areas areas labor autarky certainly would have dictated otherwise. Foreign worker programs are also held to discourage adoption of, or to delay the adoption of new technologies. Martin and North maintain that the H-2 program and the availability of cheaper harvest labor does not encourage efforts at developing dwarf varieties of apple trees that would allow women and retired persons to be used as harvest workers or renew mechanization efforts for harvesting processing apples and sugar cane. Corroborating empirical evidence on this subject is provided from the case of the Bracero Program and adoption of mechanical cotton harvesters. Although, by 1958, mechanical cotton strippers were available, only thirtyfive were being used by growers.

PAGE 24

17 Following the decline of the Bracero Program after 1956 and the increased cost of Bracero labor, there ensued a dramatic shift toward mechanical harvesters, and at the time of termination of the program in January 1965, cotton harvesting was virtually completely mechanized (Morgan and Gardner, 1982). whereas the foregoing discussion has focused on the seemingly detrimental effect of foreign immigrant workers, researchers and proponents of liberalized immigration policies point to the beneficial side of the story. Ethier (1985) used a trade theoretic approach to develop a model to explain the purported beneficial effect of foreign migrant labor on the domestic economy in the presence of industrial fluctuations and import competing industries. He showed that host countries smooth out native labor employment fluctuations by a combination of commodity dumping and migrant dumping. The extent to which this balancing is successful is, among other factors, critically dependent on the elasticity of substitution between domestic labor and foreign labor in production. Even those opposed to allowing foreign workers into the economy recognize that certain advantages accrue to the receiving country in general. Reder (1963) recognized that, overall, immigration raised per capita income in the domestic economy through increasing total output and by stimulating further capital investments. As pointed out earlier, Reder 's concerns were with the distributional impact of immigration which he finds detrimental. The view that immigration favors the distribution of income towards owners of nonlabor inputs (capital owners) and against labor is supported by Rivera-Batiz in his 1981 and 1983 studies. In a two-sector (a modern

PAGE 25

18 sector categorized by production of an exportable good, a high administered wage causing unemployment and a traditional sector producing an import competing good, and characterized by flexible wages and use of migrant labor) model, Rivera-Batiz (1981) showed that immigration caused domestic labor displacement and unemployment, favors income distribution to nonlabor inputs and had an ambiguous but possibly negative effect on natives. This latter negative effect, contradicts the generally held view of a beneficial overall effect of immigration on the domestic economy. However, within the particular framework in which his model is cast where employment in the modern sector is fixed, this conclusion is consistent. Rivera-Batiz also claimed that his model negated the view held by Cornelius (1978) and shared by Reubens (1979) that since unemployment rates are not generally high in parts of the country which traditionally have high concentration of foreign migrant labor, introduction of migrants does not lead to a decrease in welfare of domestic residents. He argued that immigration affects unemployment positively through intersectoral movements of the displaced workers from the traditional to the modern sector. Further insights into the sectorial impact of the introduction of alien workers into the domestic economy is gleaned from the studies conducted by Johnson (1980), Chiswick (1982) and Grossman (1984). Using a general equilibrium framework and disaggregating labor into skilled, domestic unskilled and foreign unskilled, Johnson theoretically examined the probable effects of illegal immigration on the U.S. labor market. He showed that domestic low skilled workers

PAGE 26

19 will lose as a result of immigration and this loss will be mainly through declining wages rather than through a loss of jobs. The earnings of domestic skilled workers will improve, as will the returns to capital. He further demonstrated the conditions under which a system of progressive taxes designed for an income transfer program, (an increasingly low unskilled labor wage may force some domestic unskilled labor to the welfare rolls) may hurt both highly skilled workers and owners of capital. This condition may exist only if there is a high degree of labor market inflexibility in the domestic unskilled labor market. Chiswick's (1982) analysis corroborated Johnson's arguments and advances the theme further. He argued that depending on the skewness of capital ownership structure in the receiving country, immigration may enhance income inequality in the domestic economy. At the same time the decline in the supply of labor in the sending country raises the marginal product of labor. In this case labor gains and capital loses, thus emigration may result in raising the general income level of the sending country and cause a narrowing of income inequality. The extent of the latter effect is again contingent upon the ownership structure of capital. A more highly concentrated capital ownership will be associated with a greater decline in income inequality as a result of emigration (Chiswick, p. 293). Chiswick (1982), Krauss and Baumol (1979) agree on the subject of income transfers. Their conclusions from these analyses are that domestic workers whose incomes decline as a result of immigration can be compensated for their loss in welfare, since immigration will result

PAGE 27

20 in an increase in the average income of the general economy. However, if immigrants are treated in the same way as domestic workers in the income transfer program, then the attempt to raise the welfare of the impacted domestic workers will be futile. It must be noted in passing, on this issue that, most studies on illegal aliens in the United States seem to indicate that such aliens do not make much use of the income transfer and social services systems, e.g., North and Houstoun (1976). In the case of legal foreign worker programs such as the Bracero and H-2 programs, workers are barred by law from participating in the welfare system. While most of the foregoing survey of the literature on the effects of immigration has been conducted on the theoretical level, empirical contributions to the debate may be found in the works of Wise (1974), Morgan and Gardner (1982) and Mehra (1984). After the termination of the Barcero Program on January 1, 1965, a controversy ensued between agricultural growers and the then Secretary of Labor, W. W. Wirtz The growers argued that they would not be able to harvest their crops without the Braceros and also that they could not find domestic workers willing and able to perform the arduous task involved in harvesting (stoop labor hypothesis). The Secretary of Labor responded that domestic labor could be found provided the growers would offer a decent wage. Wise's study tested the "stoop labor" hypothesis using data for two California crops, winter melons and strawberries. The estimated supply elasticities for domestic labor of 2.71 and 3.35 for the two industries, respectively, reinforce the Secretary of Labor's contention that a sufficiently high wage rate

PAGE 28

21 would elicit adequate supplies of domestic labor. His analysis showed that in the absence of immigrants, growers made adjustments in melon production by reducing acreage planted by 26%, total production fell by 23%, also domestic workers gained a 262% increase in employment and wages increased 67%. Produce price increased 6% so consumers were less off. Morgan and Gardner (1982) also evaluated on a more aggregate level the effects of the Bracero Program in order to provide policy insights for a prospective guest-worker program. Using data for the seven states that utilized most Braceros they construct an econometric model to estimate the demand and supply relations in the labor market and the distributional effects of the program. Their analysis showed high demand (-0.8 to -3.2) and supply (0.2 to 1.0) elasticities for the domestic farm labor market. They estimated a wage depressing effect of about 8.8%, and a corresponding expansion in the numbers of farm workers demanded of 26.0%. The wage decline also caused an estimated 8% reduction in the quantity supplied of domestic farm workers. Employers gained by employing more workers at reduced wages. The total gain for the seven state area was estimated to be $185 million; the loss sustained by domestic workers was estimated to be $135 milliona net gain of $45 million to U.S. residents. Note that there seems to have been a beneficial effect on consumers, since farm prices fell by about 5%. The results of the analysis seem to indicate a more benign effect of the Bracero Program in contrast with that of Wise. Morgan and Gardner concluded from the magnitudes of the demand and supply elasticities that the economic impact of the proposed guest-worker

PAGE 29

22 program would be sustainable. This portends good for the agricultural sectors which currently use foreign labor since the alternative to foreign labor importation would be importation of those agricultural products to replace domestic production. Mehra (1984) and Emerson and Mehra (1985) contribution to the "stoop labor hypothesis" and the broader immigration issue looks at why producers are able to maintain programs that allow foreign workers to be imported even though empirical evidence suggests that it is possible to obtain enough domestic workers for the tasks performed by the alien workers. The model used explicitly endogenized the bureaucratic policy implementation process that governs the H-2 worker program with respect to job certification and the determination of the Adverse Effect Wage Rate (AEWR) Their simulations showed that contrary to expectations, as the level of the AEWR increased, grower producer surplus increased. The conclusion derived from the analysis is that increased producer rents "provide incentive for producer group political influence seeking to maintain the regulatory environment that allows nonimmigrant worker program" (Mehra 1984, p. xi) The preceding discourse on the effect of the foreign migrant labor on the domestic worker has indicated that immigrants can adversely affect domestic workers by taking jobs away from them. Most of the theoretical studies have assumed that foreign workers substitute for domestic workers on a one-to-one basis. The empirical studies have arrived at their conclusion on the basis of simulations based on implicit or explicit assumptions about the degree of labor substitutability, but the crux of the whole issue of jobs displacement

PAGE 30

23 by alien workers hinges on the degree of substitutability of immigrants in production compared with domestic workers. Although the literature is rich in studies that have examined the substitutability of labor under various forms of disaggregation-*(e.g. blue collar versus white collar, educational status, age and unionionized versus nonunionionized) there exists a dearth of studies that have examined the substitution possibilities between domestic and foreign workers. The few studies that have a bearing on the immigration issue include Grossman (1982) and Borjas (1983). Grossman's study has a direct bearing on the degree of substitutability of immigrants and domestic workers. Disaggregating labor into natives with native parents, natives with foreign-born parents (second generation native workers) and foreign-born persons (immigrants) he estimated the Hicks elasticity of complementarity among the factors of production. He found that both second generation and foreign born workers were substitutes for native workers, though second generation workers were to a higher degree. Also immigrants substituted better for second generation workers than for native workers. This latter result was in consonance with Chiswick's (1978) result that immigrants become economically more similar to natives the longer they stay. The magnitudes of the estimated short run employment elasticities (with respect to quantities of immigrants) were -0.08 and -0.04 for natives and second generation natives, respectively and indicated that the employment displacement impact of immigrants on See Hamermesh and Grant (1979) for a critical survey of such studies.

PAGE 31

24 natives was negligible. The study also showed a wage depressing effect accompanied this unemployment. A methodological question that was addressed in this study was the issue of separability of capital and labor. The empirical finding was that under certain disaggregation of labor, labor is jointly separable from capital. Another study that has incidental bearing on the issue of the degree of substitutability between immigrants and domestic workers was that by Borjas. This analysis investigated the degree of substitutability between blacks, whites, and Hispanics. To the extent that the Hispanic community is of more recent origin and also for the fact that the immigrant population has been traditionally and predominantly of Hispanic origin, this study provided insights into the pertinent debate. The result of the study indicated no evidence of substitutability among the three groups. The implication was that the wave of Hispanic immigrants has not hurt blacks nor whites and that it may have rather helped these groups owing to the complementarity in production.

PAGE 32

CHAPTER 3 THEORETICAL FRAMEWORK The Theoretical Model The theoretical foundations of the international migration issue are rooted in the international trade theory and factor mobility literature. Mundell (1957) has shown that in the presence of international factor mobility, factors of production and commodity trade are substitutes. Based on this principle, it can be shown how international labor migration can occur. And, the consequences of this migration on both the source country as well as in the receiving country can be deduced. The model that best describes the problem confronting us in the H-2 program is that developed by Krauss (1976) and extended by RiveraBatiz (1983). They developed a simple 2x2x2 (two factors, two goods, two countries) general equilibrium model to elucidate the guest-worker problem. Consider two countries, the United States which shall be called the recipient country and the Caribbean the source country for H-2 workers. Let us also assume that the only factors of production in these countries are capital and labor, also the countries produce two goods; one tradable and the other nontradable and labor intensive in production. The U.S. has a relatively larger capital endowment while the Caribbean has a relatively larger endowment of labor. 25

PAGE 33

26

PAGE 34

27 Figure 3.1 illustrates the welfare effects of the guest-worker program on the participant countries. The 0^ and Op origins represent, respectively, the source and recipient countries origin. HH and FF curves indicate their initial transformation curves. With H-2 workers moving out into the U.S. to augment the agricultural labor force, the labor exporting country's transformation curve moves inward towards the origin. Thus the H'H' curve represents the postimmigration production possibilities of the nonimmigrant population in the sending country. Similarly, from the Op origin the F'F' curve represents the production transformation curve of the natives and immigrant labor after immigration has occurred. The initial welfare equilibrium position of the sending country is attained at the tangency of HH and a social indifference curve U^ H at the point E, corresponding to the price ratio This price ratio is represented by a negatively sloped curve W. Similarly, the initial equilibrium of the closed economy of the recipient country is point E' The relative positions of points E and E' indicate that initially the price of the tradable good in the source country is higher than in the recipient country. In the preimmigration stage, the nonimmigrant population in the sending country produces at point N and consumes at point D. At this level of production and consumption they have an excess demand for nontradables and exchange their excess supply of tradables internally with the potential migrant population for the latter 's excess supply of nontradable. They attain U X H level of satisfaction.

PAGE 35

28 With international trade, the higher price of the good in the source country would dictate that PJ of the nontradable is exchanged for JC of the tradable from the recipient country in order to attain the consumption point C. Equilibrium point C would have resulted in the welfare gain for both countries, for the sending country and U2 U]_ r for the receiving country. However, because of nontradability of one of the commodities, this equilibrium position is non-viable. This welfare position can, however, be attained through international labor mobility, since factor mobility serves as a perfect substitute for commodity mobility. Labor is imported from the source country into the receiving country until the production point P' on the H'H' transformation curve is reached. This new equilibrium point is given by the tangency of F'F', the production possibility curve of the receiving country with its labor force augmented by migrant labor, and H'H' curve with the international terms of trade line T'T'. At this equilibrium point, it is obvious that non-migrants' economic welfare in the sending country declines from Ui N to u"2 N ; thus, gross domestic product declines in the sending country and rises by the same amount TT' in the receiving country. However, to the extent that the incomes of the guest-workers are repatriated to their country of origin, the national income of the sending country improves. The receiving country's economic welfare also increases by u*2 F Ui F (at C, the slope of T'T' and TT are equal). Therefore, through labor export, the resultant trade equilibrium attained is Pareto optimal and from a

PAGE 36

29 cosmopolitan view, economic welfare of the world has increased through labor factor migration. While this general equilibrium framework of the theoretical underpinnings of international migration affords a more complete view of the guest-worker problem, empirically this model would be very difficult to make operable. At the same time the magnitude of the H-2 foreign worker program is quite small in relation to the entire labor market. As a consequence, a partial equilibrium approach is proposed. The labor market for migrant labor for harvesting in the apple industry is seasonal in nature. At harvest time, a need for extra labor arises to expedite harvesting of orchards. This demand is satisfied partially from domestic sources and the remainder comes from foreign migrant H-2 workers. Thus, two forces seem to be at work, there exists an excess demand for labor in the domestic economy and this is satisfied by a flow of excess labor from the sending countries. The wage differential between the receiving country (U.S.) and the sending country (the Caribbean) is the incentive that stimulates the foreign workers to offer their services to U.S. apple growers. If immigration were unrestricted, then labor would flow from the lower wage Caribbean to the higher wage U.S. until wages were equalized in the two areas though mobility costs and different tastes and preferences might result in equilibrium without equalization of wages. This equilibration comes about because as more labor flows into the domestic economy, wages will fall, because the marginal product of the additional workers falls. On the other side of the coin, wages in

PAGE 37

30

PAGE 38

31 the sending economy rise because the marginal product of the remaining workers rises. The preceding analysis follows Sjaastad (1962) and Morgan and Gardner (1982) and is depicted graphically in Figure 3.2. Figure 3 2A represents the demand (D c ) and supply (S c ) relationships for the Caribbean. At the prevailing wage W c an excess supply of labor exists in the labor exporting country. In Figure 3.2B, is the preimmigration wage rate in the U.S. At this wage rate the supply of domestic labor falls short of that required for apple harvesting by Ld L s The higher wage in the U.S. attracts migrant labor from the Caribbean. The augmented supply of labor shifts the laborimporting country supply curve SS U to SS' U Accompanying this expanded labor supply is a decline in the wage rate from to W*. This scenario assumes that labor immigration is not restricted by the receiving country. However, the situation in the U.S. imposes quantitative restrictions on the number of immigrants who can enter the U.S. for harvest jobs through the job certification process. Assuming a fixed quota of immigrant workers such that total supply of labor is fixed at L' d then the domestic wage will fall only to Wi** and 0L' S domestic labor will be supplied, the short fall in demand L' s L' d will be supplied through international labor migration. The distributional consequence of this movement is illustrated in Figures 3 3A and 3 3B and is based on Chiswick (1982). In this model, capital and labor, the only factors of production, are assumed homogeneous but are not perfectly substitutable in production. Figures 3.3A and 3 3B represent the situation in the labor importing and labor

PAGE 39

32

PAGE 40

33 exporting countries, respectively. In the preimmigration stage the supply of labor in the receiving country is S ou and it is S oc in the sending country, the wage rate as determined by the intersection of the marginal product curves and the supply curves are W ou and W oc respectively. Aggregate income during this stage in the U.S. as given by the area under the marginal product curve is O u DAL ou of which the share of labor is O u W ou AL ou and the return to owners of capital contributes the remainder, i.e. W ou DA. Similarly, the returns to labor in the sending country during the pre -immigration period are O c W oc PL oc and capital share W oc TP makes up the rest of the total national income O c TQPL oc With immigration, aggregate income increases in the receiving country; however, with the increasing number of workers both marginal product and average products of labor decline (assuming capital stock remains constant) The functional distribution of this increased a gg re g a te income favors capital owners against native workers. (As shown, capital share increases from W ou DA to W lu DB and although total labor share increases from OW OU AL OU to OW lu BL lu domestic labor share declines to O u W lu CL ou from O u W ou AL ou ) On the other hand the outmigration of agricultural labor in the sending country results in an increase in the marginal product and average product of the remaining workers. Thus, labor gains and returns to capital owners decline. Note also that H-2 migrant labor also gains L ou CBL lu of the increased aggregate income in the receiving country. If the income gained by the migrant workers is remitted back home as is likely to be the case with the H-2 program (the workers stay in the U.S. no more than three

PAGE 41

34 months) then national income of the sending country also increases and the results of the partial equilibrium analysis accord with the general equilibrium results. Labor and Product Market Equilibrium in the Domestic and Foreign Economies A competitive labor market structure will be assumed for the market for harvest labor. This assumption is justified because unlike the market for H-2 workers for the sugar industry where only one or a few grower associations contract labor for all the growers (a monopsonistic market structure) many growers participate in the recruitment of temporary alien workers for harvest jobs in the apple industry. The AEWR serves as a benchmark wage for recruiting alien workers. Figure 3.4 depicts the interaction between the factor markets (both domestic and foreign labor markets) and product market (domestic apple industry). The wage differential W us W c between the U.S. and the Caribbean translates the stock of excess labor into a factor flow that is made available to the apple harvest labor market in the U.S. This exogenous supply of labor causes a shift of the total domestic labor supply (residents and foreign workers) curve to the right from S^ to S'i. The regulatory environment in the U.S. represented by the stipulation of an adverse effect wage rate serves to dampen the potential impact of the original wage differential. Thus the supply of foreign labor forthcoming is constrained to L A L' A Total labor available for the apple harvest expands to 0L' A Thus augmented labor supply is purchased at a lower total cost than would otherwise have been the case in the absence of international labor migration. Consequently a producer surplus AEWRW US KJ accrues to producers. This

PAGE 42

35 W usiLabor (A) 7 / / / Q Q Q 1 (E) i 1 1 t ** us A us Labor (D) (A) The Caribbean Labor Market (B) U.S. Preharvest Labor Market (C) U.S. Harvest Labor Market (D) U.S. Apple Industry (E) Product Market Figure 3.4: Labor and Product Markets Interaction

PAGE 43

36 increased labor supply applied to the apple harvest results in a higher level of output. Figure 3.4D shows an expansion in output of Q-Q In Figure 3.4E the increased total product is depicted as a shift in the product supply curve to the right from Sp to S'p. The representation of the expanded output as a shift of the supply curve instead of a movement along the original supply curve is based on the observation that the causal factor is due to an exogenous increase in a factor of production, labor. The increased product supply causes a decline in product price. The gain in consumer welfare which arises from the price fall is represented by the area PP* US ES.

PAGE 44

CHAPTER 4 EMPIRICAL MODEL The structure of the labor market specified is comprised of the supply of labor and the demand for labor components Equilibrium in this market is attained when the two sides are equalized. The supply side of the market is represented by a set of supply of labor equations that reflect the domestic and foreign labor components of the total labor supply of the apple industry. A production function approach was used to model the demand side of the labor market. This approach permits one to obtain information embodied in apple harvesting and also estimate parameters necessary for computing labor demand parameters. Production Function Various functional forms exist that have been used to model a production activity. The popular forms that have been traditionally employed in empirical work include the Cobb-Douglas (CD) and Constant Elasticity of Substitution (CES) forms. These two functions however exhibit constant substitution elasticities-in the case of the CD restricted to unity and for the CES restricted to be constant and equal between all inputs. Here a functional form that imposes few a priori restrictions on harvesting technology was employed, in order to compute the elasticity of substitution between inputs employed in apple harvesting. 37

PAGE 45

38 Flexible functional forms permit one to model production technologies with fewer maintained hypotheses. Among the common flexible functional forms that have been employed in empirical work are the generalized Leontief and quadratic forms which can be classified as linear flexible forms while the translog form belongs to the class of non-linear flexible forms (Lopez, 1985). This classification is based on the form the left hand side variable takes after a transformation. The choice between alternative forms for empirical analysis has been arbitrary. Caves and Christensen (1980) based this choice on numerical methods for determining the regular regions of different forms Other methods employed for selection have been based on Monte Carlo studies (Guilkey, Lovell, and Sickles, 1983), Bayesi an methods (Rossi, 1983), and parametric statistical methods (Berndt and Khaled, 1979; Applebaum, 1978). Lopez (1985) has shown that between the two classes of flexible functional forms the nonlinear form does not impose quasi homotheticity and separability restrictions a priori. For this study the translog form was selected. The translog functional form may be represented mathematically as n n aj_ n ^Vimlfiu lnX i > (4.1) y= f(X 1 ,X 2 ,X n ) = a 0 n i=1 X L Xi Explicitly, for the three input production function, Q l a 2 "3 42 Li i lnX j ^Ll^9i lnX i ^ i^ lnX i (4.2) y = a Q X l X 2 X 3 X l X 2 J J X 3 J J Log transformation of equation (4.2) gives the popular form in which this function is written, viz.,

PAGE 46

39 (4.3) In y In a n + S 3 ,a.lnX.+ hZ 3 S 3 .0. .lnX.lnX. v J 0 1=1 l l 1=1 J=l ij i J The above algebraic formulation can be interpreted in two ways: namely as an exact production function, or as a second-order Taylor series approximation to a general but unknown production function. For empirical analysis the second interpretation of the translog formulation as a secondorder Taylor series approximation to an unknown production function has been widely adopted. A major drawback of the translog formulation, and for that matter, most flexible functional forms, is that they do not globally satisfy certain desirable regularity conditions for a well-behaved production function, i.e., monotonicity in output, and convexity of isoquants. Berndt and Christensen (1973 p. 85) maintain that "there are regions in input space where these conditions are satisfied. These well-behaved regions may be large enough so that the translog function can provide a good representation of relevant production possibilities." Thus an alternative is to empirically estimate the translog production function, and then, with the set of parameters obtained, check for satisfaction of the regularity conditions at each input level. Monotonicity in inputs requires that for all inputs the first derivative of the production function should be positive; i.e., fi > 0 V i The isoquants of the production function are strictly quasi-convex if

PAGE 47

40 the bordered Hessian matrix is negative definite. For the translog function, taking the derivatives of the equation gives n (4.4) fi Sy/aXi = (y/XiXai + 2 ftanXO J-l (4.5) f u n n ax^i x z i (— — )[)9ii+(ai + S )9ijlnXj) z +2 ^lnX^)] j-l J-l (4.6) fjj = a 2 y n ( )[(! + 2 jSijlnXj) (04 + S ^i^nXi)] axiaxj XiXj j-i i-i The bordered Hessian matrix is, 0 fl f2 fn f l f ll f 12 f ln f 2 f2i f22 f 2n f n f nl f n2 L nn F is negative definite if the successive principal minors alternate in sign, starting with F]_ < 0 The translog production function is non-homogeneous. The scale elasticity for this function is equal to the sum of the production elasticities. Taking the first derivative of equation (4.3) with respect to lnX^ provides us with the production elasticity n (4.7) ej tti + S y3 ii lnX i j-l

PAGE 48

41 A sufficient condition for homogeneity is that the coefficients on the quadratic terms sum to zero in column or row directions: n n n n (4.8) S 0 t1 S ^3 ii = S 2 0^ = 0 1-1 j-1 1=1 j-1 The homogeneous translog function is homogeneous of degree K = Sa^. For linear homogeneity this implies that the sum of equals unity. As indicated earlier, the translog function does not assume separability in inputs a priori. Thus one can test the hypothesis of input separability by imposing this restriction on the function. Berndt and Christensen (1973) have shown that weak separability implies the Allen Elasticity of substitution (4.9) a ik = a jk (i, j € N, k € N s ) For the translog function, functional separability of inputs i and j from a third input k demands that the first and second derivatives of y satisfy the condition (4.10) fif jk = fjf ik = 0 This implies that (4.11) e^ ej/J ik = 0 Global separability requires the condition above as well as (4.12) /3 im /J jk /3 ik /3 jni = 0 (m = l,...,n)

PAGE 49

42 Equation (4.12) reduces to e i 0ik (4.13) if ySjjj and fij m are not equal to zero. Although the three input translog function can be estimated in the form (4.14) In y a 0 + a^lnX! + a 2 lnX 2 + a 3 lnX 3 + h/3n(lnX]_) 2 + h£ 22 (lnX 2 ) 2 + H£ 33 (lnX 3 ) 2 + /3 12 lnX 1 lnX 2 + /9 13 lnX 1 lnX 3 + y3 23 lnX 2 lnX 3 by single equation methods, this is fraught with problems. For one, the number of parameters to be estimated increases rapidly as the number of inputs increase, and the additional quadratic terms present a potential source for multicollinearity problems. One may also be limited by the degrees of freedom if data are limiting. Empirical implementation of the translog functional form has in practice been via indirect methods by assuming profit maximization and competitive markets. The necessary condition for profit maximization in competitive markets is for the producer to employ the factor to the point where the value of the product at the margin equals the cost of the input ; i.e., (4.15) fi = dy/dXi = W/P y

PAGE 50

43 where W — factor price (wage) Py = price of output y. Equating (4.4) to (4.15) gives n (4.16) q = (oi + S 01* 1IIX4 ) y/Xi j-1 rearranging (4.16) and using (4.7) WXi n (4.17) e t Qi + 2 /SijlnX Thus, ej_, the production elasticity, is also equal to the value of factor i relative to the value of output or the cost share Sj_. Making the assumption that any deviations from the optimal factor share are attributable to errors in trying to minimize cost, an additive error term u^ is appended to each share equation. This renders the share equations stochastic. (4.18) S;l = Q1 + PulnXi + y9i2lnX 2 + ,Si 3 lnX 3 + ui (4.19) S 2 = a2 + /92ilnX 1 + yS 2 2 lnX 2 + 023 lnX 3 + u 2 (4.20) S 3 = q 3 + fi 3 ilvXi + 032lnX 2 + £ 33 lnX 3 + u 3 For a continuous production function, the Hessian function with respect to input quantities is symmetric; i.e., w/p.

PAGE 51

44 This symmetry condition will not in general be satisfied and will have to be imposed on the estimating functions. Under the assumption of symmetry and linear homogeneity, the factor shares Sj[ sum to unity and only eight of the twelve parameters in the system of three share equations are free. Any two of the equations will fully identify the parameters of the production function. Selecting the first two equations for estimation purposes, symmetry implies the following set of cross equation restrictions: (4.21) 021 = £l2! 013 ~ "(011 + 012): ^23 -(022 + 021> Using these restrictions the equations to be estimated are, therefore, (4.22) Si = i + y3 11 (lnX 1 lnX 3 ) + £ 12 (lnX 2 lnX 3 ) + u x (4.23) S 2 = 2 + 0l2dnXl lnX 3 ) + /3 22 (lnX 2 lnX 3 ) + u 2 The foregoing discourse on the production function translates for the apple harvesting function the particular output function specification below: (4.24) y f (DL, FL.HKT) where DL = Domestic hired picking labor in man months FL = Foreign, H-2 labor in man months HKT Expenditure on all other inputs in the harvesting operation except labor

PAGE 52

45 In terms of equations (4.22) and (4.23), the equations to be estimated are (4.25) SDL <*i + /9u(lnDL InFL) + £ 12 (lnFL InHKT) + u x (4.26) SFL = a 2 + ^12( lnDL InHKT) + y9 2 2( lnFL InHKT) + u 2 SDL and SFL represent domestic and foreign labor shares, respectively. The parameters of the estimated share equations (also those of the underlying translog production function) enable one to compute such measures as the Allen Elasticity of Substitution (AES) and the demand elasticities between factors of production. The Allen partial elasticity of substitution between any two inputs in a multiple input production system is defined as the effect on the relative factor quantities of a change in relative factor prices holding output and other input prices constant. (Sato and Koizumi, 1973, p. 47) This is given in general as n (2 XAf*) j-1 (4.27) aij (| Fij | / |F|) XiXj where ajj = The Allen partial elasticity of substitution between inputs i and j |F| = The determinant of the Hessian in equation (4.26) |Fij| The i,j th co-factor of |F| Under the assumption of a linear homogeneous production function,

PAGE 53

46 n Xjfj y j-1 Employing the parameters of the translog production function a^j is given as (4.28) ay I Hi J I |H| where for the three input case |H| is the determinant and H = 0 Hi H 2 H3 Hi Si Hi H 2 H 3 H U H 12 H 13 H 21 H 2 2 H 2 3 H 31 H 32 H33 Hii = p i + Sj 2 l.J 1,2,3. H ij 0ij + SjSj The constant -output factor demand elasticity Ejj is given by (4.29) Eij = ainXi ainWi The Labor Supply Relations The total amount of labor supplied to the economy is based on bo short run and long run considerations. For any particular sector or industry at a particular point in time, the supply of labor to the sector is devoid of long run considerations like participation rates and investment in human capital. The determinant of total labor supplied in such a situation is dependent on the labor supplier's

PAGE 54

47 decisions as to whether to provide his or her services to the particular industry or to an alternative industry, and then to what extent he or she is willing to provide those services, i.e. the supply of effort decision. While the occupational choice decision may be dependent among other things on the pecuniary rewards as well as geographical mobility, the intensity of work is a tradeoff between income and leisure. In the agricultural industry where the demand for particular types of labor is seasonal, the ease or otherwise of labor supply hinges to a large extent on workers' mobility and preferences for employment in agriculture. The total supply of labor for apple picking in West Virginia is comprised of hired domestic workers and foreign workers from the Caribbean. The domestic component is made up of localsmainly women, school children on vacation, and interstate migrant workers. Foreign workers are predominantly from Jamaica and are on contract under the H-2 program. The specification of the supply of labor function in the domestic market follows Tyrchniewicz and Schuh (1969), and is modified to reflect the particular situation of the apple harvest. The aggregate supply of labor for the harvesting market is postulated to be a function of the wage rate of farm labor, the alternative employment opportunities outside apple harvesting as indicated by returns to labor in the non-farm sector, the size of the civilian labor force, the unemployment rate in the economy and the level of unemployment insurance benefits. Algebraically, the supply of domestic hired labor DL, is given as: (4.30) DL = f(FWAG,NFW,DLF,RUMP,UI)

PAGE 55

48 where DL quantity of domestic labor services in man months FWAG = wage of hired labor in dollars per hour NFW = non-farm wage in dollars per hour DLF = civilian labor force RUMP = unemployment rate UI — unemployment insurance benefits The supply of labor to the domestic agricultural sector draws from the same pool which provides unskilled labor inputs to other non-farm activities in the manufacturing, construction and service sectors. The inclusion of the non-farm sector wage in the supply function takes into account the alternative employment opportunities available to the hired laborer. A source of the domestic seasonal harvest labor supply would come from unskilled workers who have been temporarily laid off. The decision of such persons to be part of the labor pool available for harvesting activities would depend on the level of the unemployment insurance benefits they receive and also their willingness to jeopardize this source of income. Inclusion of the unemployment rate in the specification of the supply function is justified on the premise that whether one supplies his labor services to the agricultural sector involves explicit consideration of one's opportunity cost in alternative employment, with high levels of unemployment in the economy this opportunity cost may be approaching zero. The supply of immigrant workers from the Caribbean is hypothesized to be a function of returns to labor, alternative earning opportunities in the Caribbean, the exchange rate between the U.S. dollar and Jamaican dollar and the unemployment rate in Jamaica. Jamaica is used

PAGE 56

49 as the representative country for economic conditions in the Caribbean because about 80% of all the H-2 workers originate here. Prior to arriving in the U.S. their earnings potential is measured by the adverse effect wage which is the mandatory minimum wage that any employer could pay. In the U.S., the foreign workers are paid the same piece rate as domestic workers. Inclusion of the exchange rate in the specification is based on the justification that a sizable portion of the immigrants' earnings is remitted back to the home country. Under such circumstances a favorable exchange rate regime will be an inducement to supply more of their services. As an indication of the alternative market conditions in the foreign labor market, the minimum wage in Jamaica is employed in our specification. Alternatively, the per capita income on a national basis may be used. Formally the foreign workers supply function was specified as follows (4.31) FL = f(FWJ,MWJ,EXJ,JUMP) where FL Quantity of foreign labor in man months. FWJ = Hourly earnings of foreign workers. MWJ = Minimum wage in Jamaican dollars per hour. EXJ = The exchange rate between the U.S. dollar and Jamaican dollar. JUMP Rate of unemployment in Jamaica. The Adverse Effect Wage Rate The adverse effect wage is a minimum wage, the compliance of which must be met by all employers who use both foreign and domestic workers.

PAGE 57

50 The intent of instituting the AEWR in such markets is to ensure that wages offered in such markets do not cause a depression of wages of U.S. workers similarly employed in other occupations. The adverse wage is generally announced prior to the production season by the Department of Labor. Various methods and criteria have been used to arrive at the annual hourly wage for each state. The stipulated wage is, however, supposed to reflect interests of various parties that are affected by or are the targets of the instrument. The interested parties include the government, represented by the D.O.L. producer groups, politicians and worker groups, and it is believed that the final wage decided upon is a compromise achieved through a political-bureaucratic bargaining process. Thus, the adverse effect wage rate is hypothesized to be a function of the previous year's level, AEWR t i a measure of the expectation of the wages of other similarly employed workers FWJV t _i, forecast crop size PRD t _i expected crop price PRC t _i, and an index of producer group political influence. This specification follows Mehra (1984); however, we go further in testing the hypothesis that various political factions constituting the political interest in setting the AEWR have differing agendas and may affect the level of the AEWR differently. Equation (4.38) represents the adverse wage determination process (4.32) AEWR (AEWRt.!, PRD t 1 ,FWWV t 1 ,PRC t 1 ,ICTl,ICT2) The lagged AEWR reflects the behavior of bureaucrats' decision making. It is asserted that an element of stability in the wage levels is

PAGE 58

51 introduced by bureaucrats in basing part of their decision on past levels of the AEWR. The other side of government's interest in the level of the AEWR is its regulatory objective of preventing a wage decline of other workers as a result of introduction of foreign workers. This objective is represented in the specification by the average wage rate for all hired farm workers lagged one season. The lagged output price PRC is also included to reflect the influence of producers' crop price expectations on the adverse wage. Producers are also likely to base their decision to influence the adverse wage in one direction or the other depending on their expectations of the crop size. Thus, the forecast crop size variable PRD^-.^ is included as an explanatory variable to take account of this producer behavior. The variables ICT1 and ICT2 are indexes of congressional tenure, for congressmen representing the state and selected congressional committee chairmen, respectively. While it is expected that through political influence seeking, producers would be able through their congressmen to limit increases in the adverse effect wage, the actions of relevant committee chairs in the wage setting process are unpredictable Equations (4.33) to (4.46) summarize the specified labor market model (4.33) SDL ox + £ n (lnDL InFL) + ^(lnFL InHKT) + U1 (4.34) SFL = q 2 + /3 12 (lnDL InHKT) + £ 22 (lnFL InHKT) + u 2 (4.35) InDL = a 3 + £ 31 lnFWAG + /33 2 lnNFW + 0 33 lnDLF + 0 34 lnRUMP + /9 35 lnUI + u 3 (4.36) InFL = a 4 + £ 41 lnFWJ + £ 42 lnMWJ + /3 43 lnEXJ + £ 44 lnJUMP + u 4

PAGE 59

52 (4.37) InAEWR a 5 + ^ 51 lnAEWR t 1 + J 9 52 lnPRD t 1 + ^lnPRC,-.! + /3 5 4lnFWWV t 1 + £ 55 lnICTl + /3 56 lnICT2 + u 5 (4.38) SDL EDL/TC (FWAG DL) /TC (4.39) SFL = EFL/TC = (FWJ FL) /TC (4.40) TC EDL + EFL + HKT (FWAG DL) + (FWJ FL) + HKT where EDL, EFL and HKT are respectively the expenditures on domestic labor, foreign labor and other inputs and TC is total expenditures on all inputs Data: Sources and Construction of Variables Data for this study were obtained from secondary sources. However, while some series were already published, the basic data for labor and wages were in "semi-processed" form and needed to be cleaned and organized. There was also the need to convert some of the variables by various transformations into forms that would be operational in the empirical model. This section explains the nature of the data and how various variables were handled. Employment Data The basic source of data for hired labor used in the apple harvest operation is the InSeason Farm Labor Reports otherwise known as ES-223 report of the U.S. Department of Labor, Employment and Training Administration. The reporting period is for the preceding two weeks; data are collected on the numbers of persons hired in the apple picking operation. The data are disaggregated on the basis of whether workers are of domestic or foreign origin. Domestic workers are further broken down to local, intrastate migratory and interstate migratory, the last

PAGE 60

53 group also identifies Puerto Rican contract workers separately. The data were converted to man months of labor by aggregating numbers of persons for two biweekly periods and finding the average. The manmonths of work performed for each season is the sum of monthly aggregates in each season. Typically there were approximately two months in each harvesting season. Wages Wages for labor employed in the apple harvest were obtained from the ES-223 InSeason Farm Labor Report, and from ES-232 Domestic Agricultural In-Season Wage Reports. Only domestic workers are covered in the wage survey. Wages are quoted on a piece rate basis of cents per bushel or per box. Efforts have been made to convert the piece rate to an hourly basis by weighting the total wage bill with the total number of hours worked in that piece wage rate category. The average hourly earnings resulting from this procedure are subject to error because the hours worked figure is only a rough tally and the production figures are less than precise. For the purpose of this study, the average hourly wage is computed by summing the wage bill in each piece rate category, and dividing by the total number of hours worked. Average Wage Rate for All Hired Farmworkers The wage rate of all hired workers by state is published by the U.S.D.A., National Agricultural Statistic Service in Farm Labor The annual average wage rates are weighted averages of the four quarterly estimated wage rates weighted by the number of hours worked. A break in the publication of this series occurred from 1981 onwards. Data for

PAGE 61

54 1981 to 1984 were estimated by using the following prediction equation based on data from 1968 to 1979. A (4.41) FWWV 0.519 + 0.174(NFW) + 0.103(TREND) (0.098) (0.013) R 2 0.96 Non-farm Wage Rate The appropriate non-farm wage should reflect the wages that labor employed in the apple harvest would receive if they had offered their services in the non-farm sector. Hired labor in the harvesting operation is generally unskilled labor that can easily fill a variety of unskilled jobs in the manufacturing sector and in construction. Wages in the Food and Kindred Industry are used to represent alternative earnings. The series is published by the Department of Labor in the Handbook of Labor Statistics. Index of Congressmen Tenure and Seniority This variable is included in the AEWR determination equation to take account of the political influence that apple producer groups exert on congressmen who are involved in setting the adverse effect wage rate. Since the wage is set by negotiation between the administration, represented by the Department of Labor, and other parties including congressional committees and other interested parties, it is hypothesized that producer groups would try to influence decisions in their favor by lobbying congressmen representing their districts. Pressure may be brought to bear on these congressmen in the form of campaign contributions to their re-election effort.

PAGE 62

55 Congressmen in turn will intercede on the behalf of producer group interests with the bureaucratic agencies. Since congressmen control agencies' budgets, the bureaucrats are likely to be responsive to their demands in anticipation of "quid pro quo" when their budgets come up for review. The extent of the clout wielded by the congressmen in influencing decisions in their committees depends to a large degree on seniority in congress, and on the particular committee. The longevity of a politician in office is measured by the number of re-elections and depends on the incumbent's ability to marshall among other things, political contributions for the re-election effort. In constructing the index it is hypothesized that it is likely that congressmen from the state or district in which production occurs may influence decisions differently than congressmen from other states and serving on committees that have a bearing on decisions concerning the setting of the adverse wage. Thus two different indexes were constructed: ICT1 for state congressmen and ICT2 for congressmen from other states who chair relevant committees The index was constructed by scoring for each congressman the cumulative number of years that he has been in office and an extra score for each year of re-election. For the state congressmen, an additional point is given for each year of being on a relevant committee. The committees that are considered important in setting the adverse effect wage are for the House, Agriculture, Judiciary, and Education and Labor; in the Senate, the committees are Agriculture, Appropriation, Finance, Labor and Public Welfare, and Judiciary.

PAGE 63

56 Other Inputs This variable is constructed as a residual, taking account of other miscellaneous inputs other than labor that are employed in the apple harvesting operation. It is envisaged that component inputs would include capital input services provided by expenses on fuels for operating trucks and tractors, depreciation on trucks, tractors, ladders and other equipment and other overhead, including interest payments Data on such inputs do not exist for West Virginia. However, cost and returns data have been collected for apple production for New York State, mainly Western New York, and are published in the Cornell University Agricultural Economics Research series. The relevant part of this series is used for this study based on the assumption that similar conditions exist for the two locations. Apple production in Western New York is predominantly for the processing market; the same applies to West Virginia as opposed to Eastern New York where a substantial part of production is for the fresh market. In essence, it is being maintained that with respect to "other inputs" the expenditures are similar for the two production areas. The cost data are on a per acre basis. Since total acreage data were not available for West Virginia the original data were converted to a per bushel basis and adjusted for total annual output for West Virginia.

PAGE 64

CHAPTER 5 EMPIRICAL RESULTS Estimation Equations (4.33) and (4.34) form the set of equations that were estimated to provide parameter estimates for the translog production model. In this aggregate model, labor input quantities and prices should be considered endogenous and their values jointly and simultaneously determined in the system. Thus, the simultaneity problem arises. Also errors of the individual share equations are not likely to be independent of each other but (contemporaneously) correlated. The two stage least squares approach will take care of the simultaneity problem and is employed to estimate the unrestricted model. It is noted however that the share equations are nonlinear in the endogenous variables and applying standard 2SLS procedures to a nonlinear system will not be appropriate. The estimation procedure employed is based on a method developed by Amemiya (1974) Amemiya shows that the optimization problem is one of estimating values of the parameter vector 6 such that the objective function equation 5.1 below is minimized (5.1) (y f)' D (D-D)" 1 D' (y f) where Yt ~ f t< z t,0) + u t 57

PAGE 65

58 z t is a [ (m + k) x 1] vector of m endogenous variables and k exogenous variables. f t is a nonlinear function in both z t and 6 having continuous first and second derivatives with respect to 9 D is a (t x K) unspecified matrix of certain constants with rank K. The choice of the elements in D may be based on any set of fixed variables or functions thereof. The nonlinear two stage least squares estimator is consistent and asymptotically normal. Amemiya's method is also applicable in the case where the f t is nonlinear in the variables but linear in the parameters. As noted earlier, the share equations can be considered logarithmic marginal productivity relations of a well-defined production function if and only if their partial derivatives are symmetric in inputs. Symmetry of Pij however, may not be attained in empirical estimation unless imposed a priori in the estimation. Cross equation restrictions in the share equations render individual share equations interrelated with other equations and their disturbances become correlated. Thus, imposition of symmetry restrictions requires the share equations to be estimated jointly. Results of Estimation Supply of Domestic Labor: The original specification of the supply of domestic labor has as arguments the hourly farm wage paid apple pickers (FWAG) the non-farm wage (NFW) domestic civilian labor force (DLF) rate of unemployment (RUMP), and the level of unemployment insurance benefits (U.I.)For the initial estimation, however, the unemployment insurance variable was dropped from the specification because on examination of the series

PAGE 66

59 it was observed that the level of unemployment insurance benefits virtually did not vary over the observation set. In effect, one would have been including another constant term as an explanatory variable. The specification of the supply of domestic labor is loglinear, thus the coefficient estimates are elasticities. Specifically the equation to be estimated is; In DL = £ 0 + FWAG + /3 2 ln NFW + £3 lr > DLF + /94RUMP + e where DL is domestic labor in man months, and e is an error term with zero mean and constant variance The results of the two stage least squares estimates, using as instruments all predetermined variables in the system is shown in Table 5.1. Apart from the non-farm wage variable which showed a highly statistically significant t-ratio and the correct a priori sign, all the other explanatory variables exhibited high standard errors, indicating their coefficients are not significantly different from zero. At the same time, the F statistic (with 4 and 12 degrees of freedom) is 8.3, indicating that one should reject the hypothesis that all coefficients are zero. The occurrence of large standard errors of the coefficients coupled with a high F statistic is symptomatic of the presence of a multicollinearity problem. whether multicollinearity in the sample is harmful depends on the degree or severity of the collinearity problem. Various approaches have been adopted in the literature to measure the degree of multicollinearity. This includes using the determinant of the moment

PAGE 67

60 Table 5.1: Parameter Estimates for the Domestic Labor Supply Equation Explanatory Variables 2SLS Coefficients Coefficients Intercept •14.3670 (9.1935) a 17.4210 (2.1285) FWAG 0.0202 (0.0568) 0.6501 (0.2799) NFW -6.3590 (1.5752) DLF 0.5054 (0.9045) RUMP 0.0386 (0.2542) NFWUN -4.9977 (0.9651) a Standard errors in parentheses

PAGE 68

61 matrix (X'X) a low value indicating a high degree of multicollinearity Alternatively, since a small value of the determinant of (X'X) implies that at least one of the characteristic roots of the moment matrix is small, Besley, Kuh, and Welsch (1980) have suggested using the ratio of the largest to the smallest square roots of the characteristic roots of (X'X), the singular values of X, as a measure of ill -conditioning of the X matrix. This ratio is called the condition number and a value exceeding 30 is considered an indication of problematic multicollinearity. When the condition number is calculated for the model a value of 763.23, far in excess of the critical level was obtained. Which method to choose to correct for an ill-conditioned data matrix depends on the specific problem at hand. Suggested solutions include, obtaining additional information by augmenting the sample size or using extraneous information such as extraneous estimates of elasticities, or constraining some parameters to take certain values a priori. A related approach is imposing additional restrictions on the parameters of the model. Other approaches include principal components and ridge regression. The approach selected was to impose additional restrictions as follows. Since the demographic variables, labor force and rate of unemployment move closely together, they may be considered as one measure of employment by subtracting the rate of unemployment from one and multiplying by the domestic labor force. Alternatively, one can construct an indicator of non-farm opportunities by deflating the wage in the Food and Kindred Industry by the C.P.I, and adjusting for the

PAGE 69

62 employment rate by multiplying the deflated non-farm wage by one minus the unemployment rate. The former method was adopted by Mehra (1984), while the latter method was employed by Schuh (1962) and Emerson (1976). The latter specification represented as NFWUN in the model was adopted here. The supply of domestic labor function was estimated employing as regressors FWAG NFWUN and DLF. Only the coefficient on NFWUN showed statistical significance, the FWAG and DLF estimated coefficients exhibited large standard errors. The model was reestimated, omitting from the regressors the domestic civilian labor force variable since its coefficient showed the largest standard error in the previous estimation. While this approach is likely to ameliorate the effect of a poor design matrix one must take cognizance of the possible specification bias that might be introduced into the estimation process The results of the domestic labor supply equation with these modifications are shown in Table 5.1. The standard error on the picking wage coefficient has been reduced and the coefficient is significantly different from zero at the 5% level. The sign of the coefficient also conforms to a priori expectation. This coefficient translates as the supply elasticity of domestic hired workers in apple harvesting. The magnitude of the supply elasticity implies that the supply of domestic hired labor is inelastic. This is consistent with Tyrchniewicz and Schuh' s (1969) short run elasticity estimate of .621 and is on the lower side of both Emerson et al.(1976) and Morgan and Gardner (1982) who estimated an elastic supply of domestic labor. An

PAGE 70

63 inelastic supply of hired labor has implications for the controversy between growers and D.O.L. and seems to reinforce the farmers' argument that domestic workers are difficult to attain; considerably higher wage incentives are required to elicit an adequate supply of domestic workers The adjusted non-farm wage variable (i.e. wages in the Food and Kindred Industry adjusted for unemployment) NFWUN has an estimated coefficient which conforms to a priori expectations with respect to sign. The value of the t-statistic shows that this variable is highly significant in explaining variations in the supply of domestic labor. The magnitude of the estimated coefficient implies that a 10% increase in the non-farm wage will induce a decline of 50%, in the hired labor services available to the apple harvest operation. That the elasticity of supply with respect to the non-farm wage adjusted for unemployment is highly elastic is collaborated by evidence from Tyrchniewicz and Schuh (1969), Glover (1971), Emerson et al (1976) and Morgan and Gardner (1982). Two effects are at play in influencing labor supply, the non-farm wage and the employment rate. If unemployment in the economy as a whole is low, less labor will be available in agriculture for harvesting. High employment in the economy coupled with high wages in the nonagricultural sector will result in fewer workers being available for seasonal work in agriculture. This is true, for most of the hired workers are migrants or part time workers in agriculture. Full employment in other sectors of the economy would mean fewer lay-offs and therefore a reduced source of temporary domestic workers.

PAGE 71

64 The linkage between the agricultural labor market and the industrial labor market is facilitated by the relatively low embodied human capital required for labor transferring from manufacturing to harvesting or picking functions. The interaction of the two markets can have another effect apart from employment effects. In the absence of importation of foreign labor, a high non-farm wage elasticity of supply would mean that high wages in the non-farm sector would attract some labor away from agricultural activities like harvesting. This would imply a shift of the domestic labor supply curve to the left. If the demand for labor stays the same then the additional supply of labor would be forthcoming only at a higher wage Thus the events taking place in the macro economy can impact both wages and employment in the agricultural sector. Supply of Foreign Labor The H-2 migrant labor supply equation, like the domestic labor supply model, is in log-linear form. Initially, labor supply was postulated as a function of the real adverse effect wage rate for the current season (AEWR) the exchange rate between the Jamaican and U.S. dollar (EXJ), and the unemployment rate in Jamaica (JUMP). Alternative earning opportunities were measured by the per capita gross domestic product (GDP) since data on minimum wages in Jamaica were not adequate, and for the years that such data were available, the minimum wage remained virtually constant. Initial estimates of the parameters consistently gave negative values for the estimated coefficient of the adverse wage. This and other explanatory variables also exhibited large standard errors of estimates. Since the numbers of workers

PAGE 72

65 employed in apple picking have been converted to a service or effort basis, it is plausible that the negative and statistically insignificant parameter estimate was indicating that the adverse effect wage rate variable was an inappropriate wage variable for the harvesting period. Although during the harvest period both domestic and foreign workers are paid the same piece rate, it is also known that the two groups of workers have different productivities. Thus, the earnings per hour should differ for domestic workers as compared to foreign workers. In this study, however, productivity data do not exist for foreign workers to convert the piece rate to an hourly rate. It must be noted that the source of hourly earnings for domestic workers namely the InSeason Wage Survey of the Department of Labor does not include foreign workers in the survey. The best one could do under such circumstances was to assume the same hourly earnings for domestic and foreign workers with the caveat that one would be estimating an errorin-variable, model a derivation of which is made below. Let the foreign labor supply equation be represented as yi Y m + xi£ + n where is the true foreign wage and is related to the observed foreign wage as Y]_ = Y^ e^ where e^ is the measurement error. 71 (Yl* ei)71 + Xx0 + e 1 In the first stage of 2SLS an estimate of Y]* is given as A A A Yl Yi ui ui is the estimated reduced form error. A But the expectation of Y]* is the same as that of Y^ Xn lt where 11^ is A the vector of reduced form parameters. The use of Y]_* in the second

PAGE 73

66 stage therefore purges the measurement error problem. Since the measurement error occurs in FWAG which is an endogenous variable, the simultaneous equation estimation of the foreign labor supply equation takes care of the bias introduced. The measurement error in the foreign worker earnings proxy variable FWAG is expected to cause an underestimation of the foreign worker earnings. The results of estimating the foreign labor supply equation with the following set of variables FWAG, GDP, EXJ and JUMP as explanatory variables are presented in Table 5.2, column 2. The proxy for foreign labor earnings FWAG has an estimated coefficient which is statistically different from zero. It also has a correct positive sign. The magnitude of the estimated coefficient indicates that the supply of labor is very responsive to wages. The signs of the coefficients on the remaining explanatory variables are plausible; however, none of the coefficients are statistically significant at the 5% significance level. The standard errors on these parameter estimates are large. Inspection of the correlation between the explanatory variables indicates high intercorrelations among the variables. The condition number (192.2) again suggests an ill-conditioned data matrix. Restrictions were imposed related to the three variables per capita gross domestic product, exchange rate and the unemployment rate, transforming them into a single measure of alternative employment opportunities in Jamaica. First the per capita income was converted to U.S. dollars by dividing G.D.P. by the exchange rate. The G.D.P. in U.S. dollars was then adjusted for unemployment by multiplying it by the Jamaican unemployment rate.

PAGE 74

67 Table 5.2: Parameter Estimates for the Foreign Labor Supply Equation. Explanatory 2SLS Variables Coefficients Coefficients Set (A) Set (B) Intercept 12 776 14 6510 (6. 8157) a (3, ,5420) FWAG 2 1209 2 ,0910 (0, 9900) (0, ,9352) EXJ 0, 2178 (0, ,4414) GDP -0, ,1443 (0, ,3537) JUMP 0 ,2704 (1. 1858) JEMX -0, ,2014 (0, ,1020) a Standard errors are in parentheses

PAGE 75

68 Table 5.2 column 3 shows the results of the estimate of the modified foreign labor supply equation. Both explanatory variables do show statistical significance at the 5% level. The supply elasticity with respect to earnings is elastic. This result is not surprising at all. The differential between wages in the U.S. and Jamaica is very large and therefore Jamaican laborers will exhibit a high response to any demand for their services. The wage elasticity being elastic implies that economic incentives alone would be adequate to induce Jamaican laborers to pick fruit in the U.S. The constraint on exporting all their services is policy induced by the U.S. certification process. Mehra (1984) estimated the supply elasticity for H-2 workers employed in sugar cane harvest to be 2.4; this is about the same as our estimate of 2.1. Although Morgan and Gardner's (1982) estimate far exceeds the aforementioned estimates, one arrives at the same conclusion that supply of foreign labor is very responsive to the farm wage in the U.S. The variable JEMX, as indicated previously, is a composite of the per capita gross domestic product, the exchange rate, and the rate of unemployment in Jamaica. The estimated coefficient is significant and bears a negative sign. This inverse relationship between the measure of alternative employment opportunities and the supply of labor suggests that as the adjusted per capita income in the economy rises there will be more opportunities in the economy for labor with expanded economic activities. Jamaicans would therefore supply less labor for work outside their country. The magnitude of the parameter estimate,

PAGE 76

69 however, indicates that this relationship is inelastic. Thus, a ten percent change in JEMX would bring only about a 2% decline in labor supply to the U.S. This observation is consistent with the implications deriving from the wage supply elasticity. The strong positive impact of the destination country wage is expected to override any improvements in employment opportunities in the sending country. The Adverse Effect Wage The adverse effect wage rate equation is specified as AEWR = aoAEWR^i PRD^?^ FRC^i FWWV^ ICTl^ 5 ICT2^ 6 where AEWR is the deflated hourly adverse effect wage for West Virginia, AEWR t .]^ PRD t _i PRC t _i and FWWV t _;L are the lagged values of the adverse effect wage, output, price of output and statewide hired farm wage. ICT1 and ICT2 are measures of congressional tenure and stability for state congressmen and congressional committee heads, respectively. For purposes of estimation the above equation is cast into a log-linear form by taking logs of both sides. The loglinear form Ln AEWR In aO + /9].ln AEWR t ^ + ^ln PRD t .x + 0 3 ln PRC t i + 0 4 ln FWWVt.! + /3 5 ln ICT1 + 0 6 ln ICT2 +e was estimated by ordinary least squares method. The disturbance term e has zero expected mean and variance a 2 The results of the estimation were disappointing with most of the coefficients exhibiting large standard errors. It was suspected that multicollinearity was a problem and the variable with the largest standard error, lagged output, was dropped from the specification. On reestimation of the model the results in Table 5.3 column 2 were obtained. Although the standard

PAGE 77

70 Table 5.3: Parameter Estimates of the Adverse Effect Wage Equation. Predetermined Variables O.L.S. Coefficient Hatanaka's Estimator Coefficient Intercept AEWR t-1 FWWV t _x PRC,-.! ICT1 ICT2 R 2 D.W. Durbin-h 0.4176 (1.0990) a 0.6224 (0.2730) 0.2518 (0.1914) -0.0002 (0.0489) -0.1233 (0.0702) -0.05217 (0.0493) .40 2.52 n. a. 1.5275 (0.4679) 0.9572 (0.1401) 0.1959 (0.06781) -0.1279 (0.0217) -0.0953 (0.0212) .91 2.31 a Standard errors are in parentheses. n.a. Not applicable because n(Var(/3i)) is greater than one,

PAGE 78

71 errors of the estimated coefficients declined somewhat and the t-ratios improved, only the lagged adverse effect wage variable showed statistical significance. The magnitude of the estimated coefficient of lagged apple price is very close to zero and still exhibited a large standard error. Since a lagged dependent variable is among the explanatory variables, serial correlation of the error term is suspected of being a problem. As the model stands now, one suspects a multicollinearity problem probably due to misspecif ication of the model and autoregressive residuals. In the presence of the two problems any test of the latter must be robust to the misspecif ication problem. Godfrey (1987) has proposed a testing strategy based on a sequence of ordered hypotheses representing increasingly restrictive assumptions about the true data process, starting from a hypothesis about misspecif ication through autocorrelation, to a situation of no problem. For this model, it is assumed any misspecif ication may be due to inclusion of the lagged output price variable and therefore omitting that variable from the estimation will take care of the collinearity problem. Since the data are annual, the form of autocorrelation in the residuals is likely to be of the first order. In the presence of lagged endogenous variables in the model, the conventional DurbinWatson (D.W.) test for first order autocorrelation will be inappropriate since this test was derived under the assumption of a nonstochastic X matrix. (In the presence of lagged endogenous variables appearing among the explanatory variables the D.W. test tends to take a value around 2.) The alternative Durbin h-test (equation

PAGE 79

72 5.2) although derived as a large sample test, is employed to test for first order autocorrelation in the regression. (5.2) / n l-n(Var(^)) where r = the estimate of serial correlation coefficient from the O.L.S. regression. Var(@i) the estimated sampling variance on the coefficient of the lagged dependent variable in the O.L.S. regression, n = sample size (17) This test, however, is not applicable in this case because the product of n and the variance of the coefficient of the lagged adverse effect wage variable is greater than unity, i.e. n Var(/3i) ™ 1-242 > 1. Maintaining the assumption of first order serial autocorrelation, ordinary least squares estimation of the model is no longer appropriate. The existence of contemporaneous correlation between the lagged dependent variable and the error term renders ordinary least squares estimates inconsistent and asymptotically inefficient. Consistent estimates of the parameters of this model can be obtained by applying an instrumental variable method or maximum likelihood estimator. Hatanaka's (1974) two step 'residual adjusted' estimator is an instrumental variable approach and is asymptotically efficient. This method is employed in estimating the model and the results are presented in the second column of Table 5.3.

PAGE 80

73 There is a marked improvement in the remaining coefficient estimates and all are statistically different from zero at the 5% level. The coefficient estimate for the lagged adverse effect variable has increased dramatically in value from 0.6224 to 0.9572. This indicates that the current value of the adverse effect wage depends highly on the previous year's level: over 95% of the current period level of the adverse effect wage is accounted for by the previous year's level. The magnitude of the coefficient of the lagged dependent variable is a pointer to the decision behavior of the parties involved in determining the annual level of the AEWR. With considerable bureaucratic input in the wage determination process one would expect a high element of stability in decision making based on past experiences and bereft of any radical changes. Mehra(1984) found similar large influences of the lagged AEWR on the current level. The lagged level of the statewide farm wage seemed to also influence the adverse wage determination process, the coefficient being significant at the 5% level. In view of the legislative directive that AEWR should be set at a rate comparable to other similarly employed workers, one would expect this variable to have assumed a strong impact. As much as the variable is an average for all hired farm workers and the term "similarly employed" is debatable, the variable used in the regression may not be accurately representing hired labor which is paid by the piece rate. Regardless, the results give an indication that the directive is being taken into consideration in the wage determination process

PAGE 81

74 The low standard errors on the proxy variables ICT1 and ICT2 serving as a measure of the political influence of producer groups in influencing the adverse effect wage determination process, indicate that these producer groups exert a measure of significant influence on the level of the wage. As expected, producer's group influence through representatives and senators would be directed to keeping the wage level low so as to cut down on their cost of production. The negative signs of the coefficients on the indexes of congressional tenure bear out this expectation. The magnitude of the parameter estimates seems to imply that congressmen representing producers directly from the state of interest do exert stronger influence than committee chairpersons who do not necessarily come from that state. To confirm the above observation a test of the equality of the two parameters was conducted. The F-test with 2 and 10 degrees of freedom was 449.15. Thus the null hypothesis that the two parameters are equal was rejected. The question of committee chairpersons' interests being probably different from native congressmen is not borne out by the signs on the coefficients; rather the two groups seem to reinforce each other's actions. A combined measure of the two groups' impact on the wage level also shows that they both work to influence the wage level in the downward direction. Similar negative influence of the producer group's action through congressmen on the level of the adverse effect wage rate was found by Mehra (1984).

PAGE 82

75 Share Equations The domestic and foreign share equations were estimated by 2SLS estimation procedures. These equations in the unrestricted form are given as SDL Pi + Pn In DD + P12 In FF + £13 In HKT SFL = P2 + P21 In DD + P22 In FF + 023 ln HKT The results of the estimation of the unrestricted share equations are presented in Table 5.4, column 2. As mentioned earlier, because the first and second derivatives vary with input levels, there is no guarantee that the isoquants of the translog function are globally convex. There is the need to check the estimated production function to ascertain whether or not it is well behaved. To assess the behavior of the estimated function, one has to test whether the monotonicity and convexity conditions are satisfied. Monotonicity was tested by examining the predicted factor shares using the parameter estimates and checking whether they assume positive signs. The predicted shares are shown in Table 5.5, and are all positive at all observations. The convexity assumption is tested by examining the production function for negative definiteness of the bordered Hessian matrix. For the three input production function this implies that the sign of the principal minors should alternate starting with a negative. Table 5.6 indicates that this condition is not satisfied at all data points. Thus, the convexity assumption is violated for certain points in the sample. Generally, without the prior imposition of the symmetry condition on the parameters the translog function will not be well behaved. Symmetry is therefore imposed on the parameters of the two share

PAGE 83

76 Table 5.4: Parameter Estimates of the Translog Function. Constrained Coefficients Unconstrained Set(A) Set(B) Parameters Coefficients Con c; t~ Ant" -0 0592 o 3810 0 3788 (0 ,2599) a (0 .0633) (0.0641) ^11 o 1796 o 1459 0 1456 (0 .216) (0 .0129) (0.0103) 012 -0 .0637 -0 0796 -0.0801 (0 .0274) (0 .0102) (0.0104) £13 -0, 0577 -0 066? i \J \J U £, -0 0655 co! ,0219) (0 ,0223) (0.0226) Constant 0 ,0300 o 3913 0 3887 (0, 2020) (0. ,0547) (0.0555) 021 -0. 0515 -0, ,0796 -0.0801 (0. 0168) (0, ,0102) (0.0104) 022 0. 1420 0. ,1276 0. 12715 (0. 0213) (0. 0106) (0.0107) 823 -0. 0422 -0. 0479 -0.0471 (0. 0170) (0, 0190) (0.0192) Constant 0. 9707 0, 2277 0.2325 ( ) (0. 1143) (0.1159) 031 -0. 1281 -0. 0662 -0.0655 ( ) (0. 0223) (0.0226) 832 -0. 0783 -0. 0479 -0.0471 ( ) (0. 0190) (0.0192) 833 0. 0999 -0. 1142 -0.1125 ( ) (0. 0401) (0.0406) a Standard errors are in parentheses.

PAGE 84

77 Table 5.5: Predicted Shares from the Estimated Share Equations. SDL SFL HKT 0 .35567532 0 .08974245 0 .55461504 0 .29210814 0 .19621723 0 .51171298 0 .28284351 0 .17961676 0 .53757577 0 .24044682 0 ,18060144 0 .57898792 0 .24450752 0 ,19993850 0 .55558944 0 .20278174 0 ,21368209 0 .58357100 0 .22923910 0 ,18861845 0 .58217618 0 .25536896 0 ,20137467 0 .54329194 0 .27144955 0. 12895683 0 ,59962490 0 ,20670563 0, 21888332 0 57444574 0 ,16211921 0. 26584640 0 ,57207165 0 ,20036129 0, 25196657 0, ,54771027 0. 23807835 0, 24467282 0, 51728639 0. 26076667 0. 19737376 0, 54189502 0. 29418603 0. 16139580 0. 54445240 0. 29116775 0. 20968977 0. 49917837 0. 32139564 0. 17271184 0. 50592729

PAGE 85

78 Table 5.6: Principal Minors of the Bordered Hessian Matrix -0. 12650493 -0. 00815221 0, 00612132 -0. 08532716 0 ,00235576 -0. 00135337 -0, 08000045 0 ,00048975 -0, 00030033 -0, 05781467 -0, ,00076977 0 ,00057130 -0, ,05978393 0 ,00047739 -0, ,00031209 -0, ,04112043 -0, ,00076888 0 ,00069253 -0. ,05255057 -0 ,00071787 0 ,00055141 -0 ,06521331 0 ,00105718 -0, ,00068066 -0 .07368486 -0 ,00337978 0 .00280593 -0 .04272722 -0 .00039735 0 .00040643 -0 .02628264 -0 .00208709 0 .00202933 -0 .04014465 0 .00040432 -0 .00000229 -0 .05668130 0 .00287041 -0 .00159009 -0 .06799926 0 .00102001 -0 .00065754 -0 .08654542 -0 .00076914 0 .00055391 -0 .08477866 0 .00362140 -0 .00198459 -0 .10329516 0 .00106109 -0 .00053762

PAGE 86

79 equations and the constrained model is estimated by 2SLS procedure. Table 5.4 column 3 presents the result of the constrained estimation. The estimated function satisfies the monotonicity requirement, however, like the unconstrained model, convexity is not satisfied at all observations in the sample. The chi-square test (one degree of freedom) of the symmetry restriction assumes a value of 0.0014. This implies that the imposition of the symmetry of the Piy s on the model is justified. It must be noted that the symmetry test also implies a weak test of linear homogeneity. Thus, the model also satisfies linear homogeneity Several problems can cause a production function not to be well behaved. These include first, errors in measurement of variables, secondly misspecif ication of the function (i.e. the translog model may be a poor representation of the harvesting function) For the present estimation one is inclined to ascribe the first problem as critically affecting the production function. Errors of measurement of various variables loom large in the model. In obtaining the aggregate measure used in the estimation, there were biweekly labor input data points for both domestic and foreign labor for which there were no data. Missing values were obtained by taking the means of adjoining data points. There was also discretionary judgement in determining the starting and ending points to obtain the aggregate labor quantity from the biweekly data. The method of calculating the factor shares, assuming a constant adjustment factor for number of hours worked for each year, definitely introduces an element of error in these variables. For the foreign

PAGE 87

80 labor share variable domestic wage was used as a proxy for foreign wage. This results in an understatement of the foreign labor share. Since it is generally held that foreign workers' output per hour is generally higher than domestic workers', converting the same piece rate to an hourly rate should result in a higher foreign wage than domestic wage The preceding discussion indicates that measurement errors exist for some of the variables. However, the variables so far discussed are endogenous; therefore one has a measure of comfort in the fact that in the simultaneous equations model, the errors will be absorbed into the disturbance terms. The same thing cannot be said of the "other inputs" (capital input) variable however. This variable is considered exogenous in the system of share equations. The existence of an exogenous variable measured with error in a system of simultaneous equations presents further problems for identification and estimation (see Geraci 1983, Hsiao 1976). Generally, measurement errors in the independent variables result in biased and inconsistent parameter estimates. A derivation of the error model is shown below. Let the system of structural relations be ry^X + u (1) where y and X are the g x g and g x k matrices of endogenous and exogenous variables, respectively. T and £ are the g x g and g x k coefficient matrices. E(u) = 0, E(uu') S, 'uu positive definite E(X) = 0 E(XX') X ; xx positive definite

PAGE 88

81 E(uX') 0. X is not observable but X is observed where X* = X + e (2) e is a k x 1 vector of measurement errors on X. E(e) = 0, E(ee' ) = 2 ee E(eX') 0, E(eu') = 0 Substituting (2) into (1) results in Ty = 0(X* e) + u Ty £X* + (u 0e) since the disturbance term (u /3e) is not independent of X E(ry|X*) # /9X* The situation is aggravated in nonlinear relations (Griliches and Ringstad 1970) Although various procedures have been suggested in the literature to handle such problems, the paucity of empirical applications of these can be explained by the complexity of the suggested methods. An instrumental variable approach was employed to reestimate the share equations dropping the "other inputs" variable from the set of instruments. This approach is based on the rationale that one is able to find somewhere in the system a predetermined variable that is measured correctly and is correlated with the variable that is affected by measurement errors, but uncorrelated with the disturbance term. The estimated constrained coefficients (set B) are similar to that from the previous estimation that included "other inputs" variable as instrument. For further discussion this set of parameters will be

PAGE 89

82 adopted. The computed t-ratios on the estimated parameters indicate that all the parameters are statistically highly significant at the 1% level Separability The estimated share equations with symmetry and linear homogeneity imposed were tested for separability of inputs. For the three input production function, four types of separability are identified. These are domestic labor and foreign labor are separable from "other inputs"; domestic labor and "other inputs" from foreign labor; foreign labor and "other inputs" from domestic labor and complete separability. The functional separability of labor in harvesting from other inputs is of more interest to us in this study. If labor is functionally separable from "other inputs" then one can omit "other inputs" in the functional specifications. Since this variable is measured with considerable errors, its omission would help the estimation. As discussed previously, if the parameters £13 and /3 2 3 are non-zero, functional separability between domestic labor and foreign labor and "other inputs" imply the following restrictions a l 023 a 2 013 = 0 011 023 021 013 0 In terms of the estimated parameters these imply oc-l p 12 ocx 022 + a 2 011 + a 2 012 0 012 011 022 = 0

PAGE 90

83 Table 5.7: Results of Parameter Analysis of Input Separability. Parameter Estimate Standard Error TStatistic a l 023 a 2 013 ~ 0 0.0272 0.0029 9.2020 011 023 021 013 = 0 -0.0123 0.0043 2.8264 Wald Test for the hypothesis that the derived set of parameters are jointly zero: chisquare = 608.72 with 2 degrees of freedom

PAGE 91

84 A Wald test was performed to ascertain whether the above conditions were satisfied, in which case one would accept that the labor inputs are functionally separable from other inputs and drop the latter from the explanatory variables of the harvesting function. The result of the test is shown in Table 5.7. The null hypothesis of each of the restrictions summing to zero should be rejected. The Wald test that the derived parameters are jointly zero is also rejected. Thus functional separability of labor from other inputs is rejected. Elasticities Estimates of the Allen partial elasticity of substitution (A.E.S.) are presented in Table 5.8. Column 2 provides estimates of the elasticity of substitution between the two categories of labor; domestic and foreign. One notices that the signs of the estimated parameters span both negative and positive values. Since the question as to whether two inputs i and j are substitutes or complements depends on the sign of the A.E.S. one is unable unequivocally to make such inferences. Further examination of the A.E.S., however, indicates that all the positive estimates correspond consistently with data points in the sample that satisfy the regularity conditions of convexity of isoquants. By extension it can be inferred that at observations where the translog function is well behaved the sign of the A.E.S. indicates that domestic labor and foreign labor are highly substitutable in picking apples. As already noted, the observed sample data of the translog model, unlike the C.E.S. functions whose elasticities are solely a function of the regression parameters, do not provide repeated

PAGE 92

85 Table 5.8: Allen Partial Elasticities of Substitution 1968-1984. a DL FL a T>L HKT CT FL HKT 1968 -4, ,2801 2 .0158 0, ,0241 1969 22, ,3433 1 .5566 2 2220 1970 104, ,0769 -0 .1109 5 ,0555 1971 -58, ,9911 1 .4710 2 3610 1972 104, ,8139 4 .1947 -0 .8792 1973 -9, ,0196 -1 .7393 4, .6827 1974 -61, ,6184 0 .5360 3 .3668 1975 47. ,1528 2 .6029 0 ,8894 1976 -11, 8034 2 .3011 0 ,7570 1977 -82. ,4311 -4 .5055 6 ,8206 1978 -15. 6158 -1, ,6180 3 ,1998 1979 1382. 9586 192, ,5790 -144. ,9039 1980 19. 4465 3, ,4692 0, ,0441 1981 48. 6052 2 ,2609 1, ,2719 1982 -55. 8812 3, ,7562 -2. 1626 1983 15. 0490 1, 8241 1. 8110 1984 53. 8661 -0, 4616 6, 6404

PAGE 93

86 observations on a single set of elasticities. For statistical inferences, one needs to know the distribution of the estimated parameters Anderson and Thursby (1986) have demonstrated that only elasticity estimates based on the means of the actual shares are likely to follow either the normal or ratio-of -normals distribution function. The estimates of the elasticity of substitution based on the means of the input shares and using the parameter estimates (Set B) from the restricted translog function are presented in Table 5.9. The calculated value of over 83 for the elasticity of substitution between domestic labor and foreign labor conforms to the observation made earlier that the two categories of labor are highly substitutable in harvesting of apples. The magnitude of the A.E.S. also confirms the notion held by most observers of the harvest labor market that domestic labor is perfectly substitutable with foreign labor. If foreign labor is highly substitutable for domestic labor as the data seem to indicate, then the fears of domestic workers of being displaced from employment are well founded. In the absence of any restrictions on the importation of foreign labor and given the seemingly attractiveness of H-2 foreign workers to growers, one would expect that growers would compose their labor with a high proportion of foreign workers, thereby forfeiting job opportunities for domestic workers. Note this result derives from a micro perspective; from the general equilibrium view it might be that displaced workers would be absorbed more efficiently in the system. Thus, importation of foreign workers could be welfare increasing.

PAGE 94

87 Table 5.9: Allen Partial Elasticities of Substitution and Price Elasticities of Demand Computed at Mean of Input Shares. Domestic Foreign Other Labor Labor Inputs Domestic .-68.80 83.31 2.59 Allen Labor Partial Elasticities Foreign -112.2 0.86 of Labor Substitution Other Inputs -1.51 Domestic -17.59 Price Labor Elasticities of Foreign -21.78 Demand Labor Other Inputs -0.83

PAGE 95

88 The sign of the computed Allen elasticity of substitution between domestic labor and the variable "other inputs" indicates that the two inputs are substitutable in harvesting. A similar relationship exists for foreign labor and "other inputs." The exhibited relationship is rather surprising, since the "other input" category is comprised of expenditures on fuel, truck and equipment that complement the labor inputs in the harvesting operation. Table 5.9 also presents price elasticities of demand computed from the relation Ejj — a ijSj, where Ejj is the demand elasticity. The results indicate that both domestic and foreign labor demand are highly elastic. This is contrary to the generally held view that during the harvest season the demand for labor is generally inelastic. The magnitudes of the estimated demand elasticities are, however, consistent with the high substitution elasticities computed between domestic and foreign labor. The prevailing view that the demand for labor is inelastic is correct in so far as the total labor demand is concerned, i.e. in a situation when there are no viable substitutes. The results of this analysis show that as a result of the high substitutability between the two labor categories, the demand for each component of labor is highly elastic. During the harvesting period for apples the weather conditions tend to fluctuate considerably with the possibilities for heavy rains and snow disrupting planned activities. To the extent that such uncertainties affect crop price expectations and thus the growers' decision environment, growers would be expected to make adjustments to the use of labor in harvesting. Although they might not have much room for maneuverability in relation to numbers of persons employed, they

PAGE 96

89 are able to spread the time of arrival of H-2 workers and then the intensity of effort by workers can be adjusted also. The piece rate system of payment definitely facilitates growers' ability to adjust the intensity of labor use. The elastic demand result of the estimates may thus be reflecting this ability of growers.

PAGE 97

CHAPTER 6 SIMULATION In this chapter the estimated labor market model of the apple harvest in West Virginia is simulated to a) evaluate how well the model can track the actual endogenous variables-Historical Simulation; b) evaluate the behavior of the model under alternative scenarios assuming alternative sets of values for policy variables. Historical Simulation. The harvest labor market model estimated in the preceding chapter was "validated" by testing how well the computed values of the endogenous variables approximate the actual values. For simulation purposes it was assumed that the quantity of foreign labor employed is exogenously determined instead of being jointly determined in the model as previously assumed in the estimation. The justification for doing this is based on the nature of the policy environment in which the H-2 program operates where the D.O.L has the authority to limit the final number of foreign workers allowed into the country irrespective of the demand by producer groups. The set of equations employed in the simulation model were the domestic and foreign labor share equations and the supply of domestic labor equation. The adverse effect wage determination equation was not included since it does not form part of the simultaneous system. 90

PAGE 98

91 The solution of the simultaneous equation model was determined using Newton's method. Simulation is static in that actual values of the exogenous variables are used in the solution process. The criteria selected for assessing the goodness of fit were (1) the mean percent simulation error: (1/n) Et=i (Y t Y t ) 100 / Y t where Y t actual value of the endogenous variable at time t A Yt = predicted value of the endogenous variable at time t t = 1, 2, n n 17 and (2) the percentage of instances in which the direction of change of the observed variable coincided with the predicted variable. The results of the historical simulations are presented in Table 6.1 with the actual variables. It is evident that the predicted values of the endogenous variables are remarkably close to the observed variables. The results of the validation measures show that none of the predicted variables had a mean percent simulation error over 1%, specifically the mean percent simulation errors for DL, SDL, SFL and DFWAG were respectively -0.35, -0.26, 0.17 and 0.087. Also, apart from the DFWAG for which 90% of the directional changes were correctly predicted, all the other variables had 100% correctly predicted directional changes. The actual and predicted values of the endogenous variables were plotted against time to graphically depict their movements over time

PAGE 99

92 Table 6.1: Actual and Simulated Values of the Endogenous Variables YEAR DL EDL SDL ESDL 1968 2293 2218 51270 0. 36423 0 35698 1969 2500 2469 15649 0 33080 0 32749 1970 1929 1927 05957 o 30860 0 30810 1971 1926 1912 17432 0 23776 0 23705 1972 1477 1482, 71167 0. 25967 0, 25729 1973 1176 1228, ,63281 0. 20260 0, 20902 1974 1199 1275, ,09497 0. 22041 0, 23176 1975 1493 1489 ,42944 0, 27371 0, ,27049 1976 1277 1281 .07495 0, 23653 0, ,23814 1977 1107 1111 ,07263 0, 18946 0, ,19130 1978 1168 1164 ,42285 0, ,17176 0, ,17099 1979 1523 1466 ,85327 0 20736 0, ,20083 1980 1484 1479 ,52930 0, ,25347 0 ,25218 1981 1510 1518 ,03259 0, ,23195 0 ,23363 1982 1646 1653 ,78943 0, 26484 0 ,26649 1983 1500 1555 ,31982 0 28247 0 ,29040 1984 1667 1654 ,46008 0, 31335 0 ,31344 YEAR SFL ESFL DFWAG EDFWAG 1968 0 .10526 0 .10344 0 .05284 0 .05294 1969 0 .21789 0 .21804 0 .05438 0 .05418 1970 0 .19261 0 .19252 0 .05648 0 .05637 1971 0 .17595 0 .17712 0 .05743 0 .05768 1972 0 ,21004 0 .20776 0 .05846 0 .05715 1973 0 ,20766 0 ,20532 0 .05807 0 .05771 1974 0 ,17794 0 ,17588 0 .05535 0 .05559 1975 0, ,21388 0, 21211 0 .05636 0 ,05526 1976 0, 11731 0. 11743 0 .05268 0 ,05303 1977 0. 19671 0. 19808 0 .04837 0 04890 1978 0, 27353 0. 27351 0 ,05682 0, 05662 1979 0. 26396 0. 26511 0, ,05278 0. 05257 1980 0, 26782 0. 26678 0, 05030 0. 05000 1981 0. 17368 0. 17400 0, 03995 0. 04016 1982 0. 14331 0. 14352 0. 04091 0. 04109 1983 0. 20087 0. 20064 0. 03747 0. 03760 1984 0. 16216 0. 16431 0. 03807 0. 03846 Simulated values are prefixed with E.

PAGE 100

93 "" < <—> csi — O cr. co \£> m
PAGE 101

94

PAGE 102

95

PAGE 103

96

PAGE 104

97 and are presented in Figures 6.1 6.4. Again the closeness of the two plots in each case is obvious. Alternative Scenarios The model was also evaluated under alternative assumptions about the quantity of the foreign labor services that would be available and the impact on quantity of domestic labor, the wage rate and the share of domestic labor in harvesting. Within the current political environment and with the background of the Immigration Act of 1986 which seeks to penalize employers who employ illegal aliens, it is anticipated that the demand for certification of H-2 workers would expand considerably. Thus the question asked is what would be the impact on domestic employment and wages if the foreign worker variable were increased by 5% and 10% of the mean level. Also, the impact of the interaction between the level of foreign labor services and a 5% increase in the adjusted non-farm wage was investigated. The labor market impact of altering the mean level of foreign labor services available as well as the mean non-farm wage adjusted for unemployment is shown in Table 6.2. An anticipated increase in H-2 worker services causes a decline in domestic labor services and also its share in total cost. Increased use of H-2 workers also causes a depression of the farm wage. Specifically, a 5% increase in the foreign labor services resulted in a decline in the available domestic labor services by 5 8% i.e. almost the same percentage. This decline was accompanied by a 6.1% depression of the farm wage. That an increase in foreign labor services resulted in an almost equal percentage decline in domestic labor services was expected given the

PAGE 105

98 0) 2 0 in n CN rco 0) w 3 O 0 -H (U l-i C a CD C 00 cu O O "3 00 c UJ CD CD X H 14-1 o tn r-i a) < 3 .-I V4 CO CD > "O C C 35 CO a> S •3 a) •u 0 "3 CD l-( O X! o C X) t-i CO J a> l-i 0 Xi CO •X a) cn C cfl QO -l 4J CO O C U-l a) a. in a) CO 00 CD — l C ^ CD o O U Z BJ c o r-l U, o m in m3 2 i m CN GO CN rH r> CN o vo in m r-^ i-l O o O O in in iH CN r rH o co u O CN -J XJ on to a < ii

PAGE 106

99 very high elasticity of substitution between the two categories of labor previously identified. When both foreign labor services and the adjusted non-farm wage were increased by 5% each, domestic labor services available decline. However, the farm wage increased relative to only a labor increase. This result is consistent with the earlier observation that the non-farm and farm labor markets are interlinked and that higher wage incentives in the non-farm sector will result in more domestic labor moving into the non-farm sector with a concomitant improvement in the earnings of farm workers. Generally, decreasing the quantity of foreign labor services available resulted in opposite response from increasing foreign labor services In the scenario where foreign labor services were decreased but the adjusted non-farm wage was increased, (comparing column 7 with column 9) less domestic labor became available and the farm wage increased.

PAGE 107

CHAPTER 7 SUMMARY, CONCLUSIONS AND LIMITATIONS OF THE STUDY Summary and Conclusions This study sought to investigate the interaction of domestic and foreign workers in the domestic labor market within the context of the H-2 temporary nonimmigrant program in the apple industry. It was anticipated that the problem setting would be the region comprising H-2 user states like Virginia, West Virginia, and Maryland. However, lack of data for some of the states resulted in limiting the analysis to only West Virginia. The use of foreign workers in seasonal agricultural harvesting activities has long been a controversial issue. Growers have maintained that they are unable to find adequate and reliable sources of domestic workers for activities like harvesting of produce for which timeliness of operation is imperative, and the lack of it would result in devastating financial losses. They therefore, have justified their continued need and pursuit for supplementary foreign workers, who incidentally are reliable and hardworking. On the other side of the controversy are the labor groups and also the Department of Labor. This block contends that a potential source of domestic workers suitable and able to do the job exist and that low wages and poor working conditions provided by the growers are all that limit domestic harvest labor availability. Further, the knowledge that growers would be able to call on foreign sources of labor precludes any serious 100

PAGE 108

101 efforts on the part of the growers at recruiting domestic workers. The consequences of this situation being the persistence of low wages, job displacement, and subsidization of the activities of one group of people to the detriment of domestic workers. In an attempt to resolve some of the pertinent empirical issues in the controversy, a labor market model of the harvesting operation in apples was set up to characterize the supply and demand relationships that exist for both domestic and foreign labor. The complete economic model consisted of two factor share equations based on the translog technology and two supply of labor equations for domestic and foreign labor. These were estimated simultaneously by 2SLS methods. An adverse effect wage determination equation that represented the political economic environment in which the H-2 program operated was estimated by ordinary least squares. While the translog production function specification in both the unrestricted and restricted (imposition of symmetry conditions) forms satisfied the monotonicity conditions at all data points the (quasi) concavity condition was violated at certain data points. Symmetry of the parameters was not rejected and since this is also a test of homogeneity of the function, the harvesting technology was assumed to be homogeneous. Another test of the structure of harvesting technology is the issue of input separability. It was hypothesized that the nature of harvesting of fruits would lend itself to labor inputs being separable in production from all other inputs in the harvesting operation. A test that the two categories of labor inputs, domestic and foreign, were separable from other inputs was rejected.

PAGE 109

102 The parameter estimates from the restricted translog function were used to estimate the Allen elasticity of substitution and the elasticities of demand for inputs. The estimated AES between domestic and foreign labor was 83. This magnitude reinforces the general contention that domestic and foreign workers are highly substitutable in a generally low skill agricultural operation. Cornelius (1986) found a similar relation between domestic and foreign workers in citrus harvesting in Florida. He estimated the elasticity of substitution to be infinite. The magnitude of the elasticity of substitution between domestic and foreign labor implies that barring any policy controls over the number of foreign workers that are allowed into the country for harvesting it is theoretically possible to have a situation of complete displacement of domestic workers in the harvesting operation. Bearing in mind that it is relatively cheaper to use foreign workers (no payroll taxes on foreign worker use) growers would be more than willing to use an all foreign work force if they could. Both labor inputs were found also to be substitutes for "other inputs" in harvesting. This result was contrary to the complementary relation that the component parts of the "other inputs" variable play in harvesting. Demand elasticities were also computed for the various inputs. Consistent with the high own substitution elasticities, domestic and foreign labor showed highly elastic demand. The price elasticity of demand for other inputs was inelastic although approaching unitary elasticity.

PAGE 110

103 The statistical results of the original estimates of the labor supply equations were disappointing. Multicollinearity and or misspecif ication problems were suspected and various combinations of variables and formulations were experimented with. The final pretest estimators involved a completely abridged model in both supply relations. Domestic labor supply was found to be less responsive to wage changes. The estimated wage inelastic labor supply contradicts findings from similar harvest labor market studies (Wise, 1974; Emerson, et.al., 1976) and seems to bolster growers' arguments of the difficulty in finding domestic labor at prevailing wages. In contrast, the supply of foreign labor for harvesting is quite responsive to farm wages in the domestic market. Other results of the supply analysis indicate that the farm sector and non-farm sector are interconnected and that domestic labor responds highly to wage incentives in the non-farm sector to the detriment of labor availability in the farm sector. An inverse relationship between foreign labor supply to the U.S. and other economic opportunities in the home country also exists, albeit of a small magnitude. The low response to economic incentives in the home country is explainable by the overwhelmingly attractive situation in the U.S. and is evidenced by the highly el astic effect of the U.S. farm wage. The regulatory environment that impinges on the H-2 program is effected through the imposition of the adverse effect wage rate. The determination of this wage was modeled to explain the factors that go into the wage determination process and takes into consideration federal directives that seek to protect domestic workers as well as the

PAGE 111

104 political and producer group influences. The results of the OLS estimation indicate that previous levels of the AEWR have a very strong influence on current levels. Consequently there is considerable stability in the politically negotiated AEWR. The impact on the AEWR of producer group influence as they act through their political representatives is negative. The estimated parameters of the simultaneous system were used in simulations to validate the estimated model and also to simulate alternative policy scenarios. In the simulations, the level of foreign labor services was considered exogenous. The results of the historical simulations indicate a close fit between the predicted values and the observed values of the endogenous variables. Increasing the supply of foreign labor resulted in a decline in domestic labor services available. This was accompanied by a decline in the domestic wage as well. However, when the increase in the supply of foreign labor services was accompanied by an increase in the adjusted non-farm wage, the decline in the supply of domestic labor services was associated with an increase in the domestic wage. Limitation s of The Study and Suggestions for Further Research As noted earlier in this report, data problems were encountered. The twin problems of lack of data for certain variables as well as the poor quality of the data for some variables hampered the analysis. Such techniques as proxy variables and errorsinvariables methods were used in an attempt to obtain as much information out of the data as possible. In the case of the supply equations, the various modifications rendered the final model inappropriate for any meaningful

PAGE 112

105 comparative static analysis in the sense that there did not appear to be any interesting predetermined variables for such analysis. Also the final estimates did not permit any analysis to evaluate the welfare effect of the program on producers and workers. The partial equilibrium setup of the model also precluded any evaluation of H-2 program on consumer welfare. With more complete data this would be an interesting area for further study. The implicit assumption in estimating only a harvesting function is that other productive activities prior to harvesting do not affect the final harvested output. It would be interesting to empirically test this separability assumption. The H-2 program has been continued under the Immigration Reform and Control Act of 1986 with few changes from the era under which the preceding analysis was done. A new feature of the 1986 Act is the Special Agricultural Workers provision permitting agricultural workers to enter the U.S. legally. A significant difference between the legalization program and the H-2 program is that unlike the H-2 case, the workers are not tied to a particular employer. They are free to search for employment in the U.S. labor market. This presents an entirely different labor market structure than in the H-2 case, and requires an alternative specification.

PAGE 113

APPENDIX A DATA FOR CONGRESSIONAL TENURE INDEX

PAGE 114

Table A-l West Virginia: Congressmen and Committees Years in Office Committee Membership 1959Appropriations 1959Judiciary 19691959-84 Labor and Public Welfare 1959-84 Representatives (Second District) Harley D. Staggers 1949-80 Cleveland Benedict 1981-82 Harley Staggers Jr. 1983Agriculture 1983Source: Congressional Directory. Senators Robert Byrd Jennings Randolph 107

PAGE 115

108 Table A2 Chairmen of Selected U.S. Senate and House Committees and Number of Years on Committees. Year of First Years on Years as Committee Election Committee Chairperson U.S. House of Representatives Agriculture W.R. Poage T.S. Foley Judiciary Emmanuel Cellar Peter Radino 1936 1964 1922 1948 1942-76 19651930-72 19511967-74, 1976 19811949-72 1973-81 Education and Labor Carl D. Perkins U.S. Senate 1948 19491967-84 Agriculture Allen J. Ellender Herman Talmadge Jesse Helms 1937 1957 1972 1939-73 1958-80 1975pre 1960-71 1971-80 1981-84 Appropriations Carl Hayden Richard B. Russell Allen J. Ellender J. McLellan Magnus on 0. Hartfield Finance Russel B. Long Robert Dole Judiciary James 0. Eastland Edward Kennedy Strom Thurmond Labor and Public Welfare Lister Hill Ralph Yarbrough Harrison Williams Orrin Hatch 1912 1937 1943 1944 1966 1948 1969 1943 1954 1954 1938 1957 1958 1976 1928-68 1933-70 1949-78 1949-78 1953-80 19731953-81 19731945-78 196719671938-68 1959-70 1959-80 1977pre 1960-68 1969-70 1971-72 1973-78 1979-80 1981-84 1965-81 pre 1960-78 1978-80 1981-84 pre 1967-68 1969-70 1971-80 1981-84 Source: Congressional Directory

PAGE 116

APPENDIX B DATA FOR THE MODEL

PAGE 117

Table B-l Index of Congressmen Tenure, 1968-1984. CTl a CT3 C Year 1968 60 76 136 1969 67 56 123 1970 72 64 136 1971 80 52 132 1972 84 60 144 1973 92 48 140 1974 96 56 152 1975 103 57 160 1976 108 64 172 1977 116 72 188 1978 120 80 200 1979 128 59 187 1980 132 67 199 1981 106 30 136 1982 112 38 150 1983 118 46 164 1984 124 54 170 Index constructed from Congressional Directory. a Congressmen from State of West Virginia. k Chairpersons of relevant Congressional Committees. c Composite of CT1 and CT2 110

PAGE 118

Ill Table B-2 Labor Employment for the Apple Harvest, West Virginia Year Domestic Labor Foreign Labor (H-2) (man months) (man months) 1968 2,293 497 1969 2,500 1,235 1970 1,929 903 1971 1,926 1,069 1972 1,477 896 1973 1,176 904 1974 1,199 726 1975 1,493 875 1976 1,277 475 1977 1,107 862 1978 1,168 1,395 1979 1,484 1,176 1980 1,510 848 1981 1,646 668 1982 1,500 800 1983 1,667 647 1984 1,469 801 Data constructed from U.S. Dept. of Labor Employment Training Administration: E 1.S.223

PAGE 119

112 Table B-3 West Virginia Demographic Data, 1968-1984. Domestic Labor Unemployment Population Year Force (000) Rate % (000) 1968 625 000 5 60 1 763 J/ U J 1969 636 002 5 50 1 746 1970 621 099 6 10 1 7 SI J/ J Mm 1971 628 000 6 50 1776 1972 641.800 6.50 1795 1973 653.700 5.70 1788 1974 667.799 5.90 1783 1975 680.500 7.30 1799 1976 698.299 6.40 1832 1977 718.799 5.90 1853 1978 746.700 5.60 1861 1979 758.599 5.60 1939 1980 771.500 9.00 1950 1981 762.400 9.30 1946 1982 744.799 11.10 1961 1983 742.500 14.20 1962 1984 737.000 11.70 1952 Source: U.S. Dept. of Commerce.

PAGE 120

113 Table B-4 West Virginia, Output, Price and Wage Data, 1968-1984. Non Output 3 Price b AEWR C Farm CPI e Wage d 1968 220 .80 5 .81 1 .45 2 .20 42 .20 1969 260 .00 4 .55 1 ,55 2 ,33 44 ,50 1970 220 .00 4 .90 1 ,60 2 ,40 47 .10 1971 250 .00 4 ,99 1 ,81 2 .66 49 ,10 1972 215 .00 5 ,60 1 ,90 2 ,80 50 ,80 1973 225 .00 8 ,90 1. 95 3 ,00 53 ,90 1974 210 .00 9 ,40 2. 20 3 ,30 59 ,80 1975 216 .00 5 ,40 2, 48 3 .64 65 ,30 1976 200 .00 8 ,30 2, 70 3 ,90 69, ,10 1977 195 .00 9 ,80 2, 74 4 ,30 73 ,60 1978 295 .00 8 ,60 2, 81 4 ,64 79 ,20 1979 260 .00 8 ,20 3, 10 5. ,00 88 ,10 1980 245 .00 7, ,80 3, 28 5, 66 100 .00 1981 200 .00 11, ,50 3, 63 6 01 110 .40 1982 240. ,00 9, ,40 4, 24 6 ,42 117, ,10 1983 216, ,00 7, ,70 4, 24 6 ,84 120 .90 1984 225, ,00 10, 60 4. 40 7 ,09 126, ,10 a Output of apples in millions of pounds: (U.S.D.A, Agr. Stat. ) Price per pound in cents: (U.S.D.A., Agr. Stat. ) ^ Adverse Effect Wage Rate in dollars per hour: (U.S. Gen. Serv. Admn.) Food and Kindered Industry in dollars per hour: (U.S.D.L, Handbook of Labor Statistics e Consumer Price Index: (U.S.D.A., Agr. Stat. )

PAGE 121

114 Table B-5 Jamaican Demographic and Exchange Rate Data, 1968-1984. Labor Unemployment Year Force 3 Rate % Employment Population (000) (000) (000,000) 1968 726 .799 19 .90 582 .170 1 .83000 1969 798 .900 17 .60 658 .289 1 .84000 1970 799 .701 20 .30 637 .359 1 .87000 1971 800 ,000 21 .70 626 .400 1 .90000 1972 800 ,500 23 .20 614 .780 1 .93000 1973 805. ,900 21 .90 629 .409 1 .97000 1974 820. ,000 21 .20 646 .159 2 .01000 1975 857, ,700 20 ,50 681 .870 2 ,04000 1976 883. 599 22, ,40 685 .669 2 ,07000 1977 910. 000 24, ,20 689 .780 2 ,10000 1978 939. 000 24. ,50 708 ,950 2 ,12000 1979 953. 599 27. ,80 688, ,500 2 ,15000 1980 991. 200 27. 40 719, ,609 2. 17000 1981 1014. 900 25. 90 752. 039 2, 20000 1982 1043. 199 27. 40 756. 599 2, 23000 1983 974. 299 26. 30 718. 500 2. 27000 1984 978. 099 25. 20 728. 700 2. 30000 Source: Economic and Social Survey. Jamaica.

PAGE 122

115 Table B-6 Jamaican Exchange Rate, Gross Domestic Product and Consumer Price Index Data. Exchange Gross Domestic Year Rate Product CPI 1968 0 .83000 4453 .200 16 .55 1969 0 .83000 4596 .600 17 .55 1970 0 .83000 5143 .500 18 .54 1971 0 .83000 5304 .799 19 .95 1972 0 .77000 5795 .299 21 .04 1973 0 .91000 5878 .600 24 .76 1974 0 .91000 5607 .600 31. .48 1975 0 .91000 5589 .600 36. .95 1976 0 ,91000 5228 .700 40, ,57 1977 0 .91000 5104 .299 45. .11 1978 1. ,41000 5132 ,100 60. 85 1979 1. 76000 5043, ,000 78. 55 1980 1. 78000 4750. 100 100. 00 1981 1. 78000 4868, 500 112. 74 1982 1. 78000 4915. 799 120. 12 1983 1. 93000 5015. 200 134. 03 1984 3. 94000 4995. 500 171. 31 Source: International Monetary Fund, International Financial Statistics

PAGE 123

116 REFERENCES Anderson, R.G. and J.G. Thursby. "Confidence Intervals for Elasticity Estimators in Translog Models." Rev, of Econ. Stat. 68(1986) : 648655 Araemiya, T. "The Nonlinear Two-Stage Least-Squares Estimator." J of Econometrics 2 (1974) : 105 110 Applebaum, E. "Testing Neoclassical Production Theory." J of Econometrics 7 (1978) : 87 102 Berndt, E.R. and L. Christensen. "The Translog Function and the Substitution of Equipment, Structures and Labor in U.S. Manufacturing." J. of Econometrics. 1(1973) : 81-114. Berndt, E.R. and M.S. Khaled. "Parametric Productivity and Choice Among Flexible Functional Forms" J. Pol. Econ. 87 (1979) : 1220-46 Besley, D.A. E. Kuh, and R.E. Welsch. Regression Diagonistics Identify ing Influential Data and Sources of Collinearity Wiley : New York, 1980 Borjas, G.J. "The Substitutability of Black, Hispanic and White Labor." Econ. Inquiry. (1983) : 93-106 Briggs, V.M.,Jr. "Labor Market Aspects of Mexican Migration to U.S. in the 1970s," in Stanley R. Ross (ed.), Views Across the Border: The United States and Mexico. Alberquerque : Univ. of New Mexico Press, 1978. Caves, D., and L. Christensen. "Global Properties of Flexible Functional Forms." Amer. Econ. Rev. 70(1980) : 422 -432 Craig, R.B. The Brac ero Program: Interest Groups and Foreign Policy. Austin: Univ. of Texas Press, 1971. Chiswick, B.R. "The Effect of Americanization on the Earnings of Foreign Born Men." J. Pol. Econ. 86(1978) : 897-921 Chiswick, B.R. "The Impact of Immigration on the Level and Distribution of Economic Wellbeing," in Barry R. Chiswick (ed.), The Gateway. Washington, D.C.: Amer. Ent. Inst, for Pub. Pol Res 1982 p. 289-313.

PAGE 124

117 Congressional Directory Washington, D.C.: U.S. Govt. Print. Off., various issues, 1945-1984. Cornelius, J.W. Substitution Between Foreign and Domestic Labor in the Citrus Harvesting Labor Market: A Total Elasticity Analysis of the Adverse Effect Wage Rate Policy: 1959-83. Unpublished M.S. thesis, Univ. of Florida, 1986. Cornelius, W.A. Jr. Mexican Migration to the United States: Causes. Consequences, and U.S. Responses. Mig. Dev. Study Grp., Cnt. Int. Study., Massachusetts Insti. Techn. 1978. Dept. of Ag. Econ. Cornell Univ., Ag. Exp. Station A.E. Res. Fruit and Vegetable Crops. Costs and Returns From Farm Cost Accounts. Economic and So cial Survey. Jamaica. Jamaica: National Planning Agency, various issues, 1968-1984. Emerson, R.D. T.S.Walker, and C.O.Andrew. "The Market for Citrus Harvesting Labor." South. J. of Ag. Econ. 8(1976) : 149-153 Emerson, R.D., and R.Mehra. "Monopsony and the Foreign Worker Program in Agriculture." Paper presented at A. A.E. A. Meetings, Ames, Iowa 1985. Ethier, W.J. "International Trade and Labor Migration." Amer Econ Rev. 4(1985) .-691-707. Finis, W. and J.Cunningham. "Effects of Minimum Wages on the Level of Age Composition of Youth Employment." Rev. Econ. Stats 60(1978) :140-145. Geraci, V.J. "Errors in Variables and Individual Structural Equations Int'l. Econ. Rev. 24(1983)217-236. Glover, T.F. Seasonal Employment of Hired Labor in U.S. Agriculture. Ph.D thesis, Purdue Univ., 1971. Godfrey, L.G. "Discriminating between Autocorrelation and Misspecification in Regression Analysis: An Alternative Strategy Rev. Econ Stats. 69(1987) : 128-137 Grossman, J.B. "The Substitution of Natives and Immigrants in Production." Rev. Econ. Stats. 64(1982) : 596-603 Grossman, J.B. "Illegal Immigrants and Domestic Employment." Industrial and Labor Rel. Rev. 37 (1984) : 240-251 Griliches, Z. and V. Ringstad. "Errors-in-the-Variables Bias in Nonlinear Contexts." Econometrica 38(1970) : 368-370.

PAGE 125

118 Guilkey, D.K. C.A.K. Lovell, and R.C. Sickles." A Comparison of the Performance of Three Flexible Functional Forms Int'l. Econ. Rev. 24 No. 3 (1983) :591-616. Hatanaka, M. "An Efficient Two -Step Estimator for the Dynamic Adjustment Model with Autocorrelated Errors." J. of Econometrics. 2(1974) :199-220. Hamermesh, D.S., and J. Grant. "Econometric Studies of Labor Demand and Their Application to Policy Analysis." J of Human Resources 11(1979) :507-525. Hsiao, C. "Identification and Estimation of Simultaneous Equation Models with Measurement Error." Int'l. Econ. Rev. 17 (1976) • 319339. International Monetary Fund, International Financial Statistics. Washington, D.C. (1985). Johnson, G.E. "Labor Market Effects of Immigration." Industrial and Labor Rel. Rev. 33 No 3 (1980) : 331341 Kiser, G.C. "Mexican American Labor Before World War II." J of Mexican American History. 2(1972):136 Krauss, M.B., "Commodity Trade and Factor Mobility." Amer. Econ Rev 64(1974) :797-801. Krauss, M.B. "The Economics of the ""Guest Worker"" Program in Agriculture: A Neo Heckscher-Ohlin Approach." Scand. J. Econ 79(1976) :470-476. Krauss, M.B. and W.J. Baumol. "Guestworkers and Income Transfer Programs Financed by Host Governments." Kvklos 32 (1979) : 36-46 Lopez, R. "Structural Implications of Flexible Functional Forms for Profit Functions." Int'l. Econ. Rev. 26 No 3(1985) : 593-601 Martin, P.L. Cuestworker Programs: Lessons from Europe. Washington, D.C. The Brookings Institution, 1979. Martin, P.L. and D.S. North. "NonImmigrant Aliens in American Agriculture," in R.D. Emerson (ed.), Seasonal Agricultural Labor Markets in the Un ited States. Ames, Iowa: Iowa State Univ Press 1984. p. 168-199. Mehra, Rekha. International Labor Migration and Florida SugarcaneA Political-Economic Analysis, Ph.D thesis, Univ. of Florida, 1984.

PAGE 126

119 Morgan, L.C., and B.L. Gardner. "Potential for U.S. Guest Worker Program in Agriculture: Lessons from the Braceros," in Barry R. Chiswick (ed.), The Gateway. Washington. D.C.: Am. Ent. Insti. for Pub. Res., 1982. p. 361-411. Mundell, R.A. "International Trade and Factor Mobility." Amer. Econ. Rev. 47(1957) :321-325. North, D., and M.T. Houstoun. "The Characteristics and Role of Illegal Aliens in the U.S. Labor Market: An Exploratory Study." mimeo. Washington, D.C.: Linton and Co., 1976. Reder, M.W. "The Economic Consequences of Increased Immigration." Rev. Econ. Stats. 45(1963) : 221-230 Reubens, E.P. "International Migration Models and Policies." Amer Econ. Assoc. Papers and Proc 73(1983) : 178-182 Reubens, E.P. Temporary Admission of Foreign Workers: Dimensions and Policies Special Report No. 34. Washington, D.C. Special Report of the National Commission for Manpower Policy, March 1979. Rivera-Batiz, F.L. "The Effects of Immigration in a Distorted Two Sector Economy." Econ. Inquiry 19(1981) :626-639. Rivera-Batiz, F.L. "Trade Theory, Distribution of Income and Migration." Amer. Econ. Assoc. Papers and Proc. 73(1983) : 183187 Rossi, P.E. "Comparison of Alternative Functional Forms in Production." J. of Econometrics 30(1983) : 345361 Sato, R. and T.Koizumi. "On the Elasticities of Substitution and Complementarity." Oxford Econ. Papers 25(1973) :44-56 Schuh, G. "An Econometric Investigation of the Market for Hired Labor in Agriculture." J. of Farm Econ. 48(1962) : 307-321 Sjaastad, L.A. "The Costs and Returns of Human Migration." J Pol Sc 70(1962) :80-93. Tullock, G. "The Cost of Transfers," in J.M.Buchanan, R.D. Tollison, and G. Tullock (eds.). Toward a Theory of RentSeeking Society.' College Station: Texas A&M Univ. Press, 1980. Tyrchniewicz E.W. and G.E. Schuh. "Econometric Analysis of the Agricultural Labor Market." Amer. J. of. Ag. Econ. 51(1969) : 777.S. Congress, House of Representatives, House Report No. 1365 82nd Congress, 2d Sess., 1952 p. 44-55.

PAGE 127

120 U.S. Congress, Senate, Committee on the Judiciary, 95th Congress 2nd session. The West Indies (B.W.I) Temporary Alien Labor Program: 1943-77 A Report Prepared for the Subcommmitte on Immigration. Washington, D.C., Govt. Printing Office, 1978. U.S. Congress, Senate, Committee on the Judiciary, 96th Congress 2nd session. Temporary Worker Programs: Background and Issues. Report for the Select Commission on Immigration and Refugee Policy. Washington, D.C., Govt. Printing Office, 1980. U.S. Department of Agriculture, Agricultural Statistics. Washington, D.C.; various issues, 1968-1984. U.S. Department of Agriculture, Economic Research Service, The Hired Farm Working Force of 1983: A Statistical Profile. A.E.R. No. 554. U.S. Department of Agriculture, National Agricultural Statistical Service: Farm Labor Washington, D.C.; various issues, 1968-1980. U.S. Department of Commerce, Bureau of Census. Statistical Abstract of the United States. Washington, D.C.; various issues, 1968-1984. U.S. Department of Labor, Employment and Training Administration, Domestic Agricult ural InSeason Wage Report. E.S. 232. Washington, D.C.; various issues, 1968-1984. U.S. Department of Labor, Handbook of Labor Statistics. Washington, D.C.; various issues, 1968-1984. U.S. Department of Labor, Employment and Training Administration, InSeason Farm Labor Report. E.S. 223. Washington, D.C.; various issues, 1968-1984. U.S. General Services Administration, Federal Register : various issues 1968-1984. U.S. President's Commission on Migratory Labor. Migratory Labor in American Agriculture. Washingrrm D.C., 1951, p. 51. se, D.E. "The Effect of the Bracero on Agricultural Production i California." Econ. Inquiry 12(1974) : 547-558

PAGE 128

BIOGRAPHICAL SKETCH Kofi Adu-Nyako was born on November 7, 1952, in Accra, Ghana. He attended the Anglican Primary School, Larteh, proceeded to the Demonstration Middle School at Akropong from where he passed the Common Entrance examination to the Presbyterian Boys Secondary, Odumasi and Legon. He entered the University of Cape Coast in 1970 to do the science preliminary course. From 1973-1974 he worked as clerk at the Bank of Ghana, Government Securities Section. He was admitted to the University of Science and Technology (UST) Kumasi in 1974 and graduated in agriculture with first class honors in 1978. Kofi served his national service as an assistant on faculty at UST and was awarded a fellowship to Cornell University in 1980. He graduated in 1983 with a Master of Science in agricultural economics. In the fall of 1983 he commenced a Ph.D program at the Food and Resource Economics Department of the University of Florida. For the 1984-85 academic year he was an instructor with the USAID/UF project in Dschang, Cameroon. He returned to the U.S. in the fall of 1985 to complete his studies. 121

PAGE 129

I certify that I have read this study and that in my opinion it conforms to acceptable standards of schedrar/ly presentation and is fully adequate, in scope and quality, as a' dissertation f/6r the degree of Doctor of Philosophy. \ / f / \j ^^.J / h .^^ / f — j 'i Robert D. Emerson, Chair Associate Professor of Food and Resource Economics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a diss^ftaition for the degree of Doctor of Philosophy. Carlton G. Davis/ Professor of Food and Resource Economics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Max Langham Professor of Food and Resource Economics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Larry Kenny Associate Professor of Economics

PAGE 130

This dissertation was submitted to the Graduate Faculty of the College of Agriculture and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy April, 1988 Dean, Graduate School


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EE11HUFMY_RE8KT8 INGEST_TIME 2014-12-15T20:16:39Z PACKAGE AA00026503_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES