Consequences of small-farm mechanization

Material Information

Consequences of small-farm mechanization
International Rice Research Institute
Agricultural Development Council
Place of Publication:
Los Baños
International Rice Research Institute
Publication Date:
Physical Description:
vi, 183 p. : ill., maps ; 23 cm.


Subjects / Keywords:
Farm mechanization -- Government policy -- South Asia ( lcsh )
Farm mechanization -- Economic aspects -- South Asia ( lcsh )
Farms, Small -- South Asia ( lcsh )
bibliography ( marcgt )
local government publication ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references.
Electronic resources created as part of a prototype UF Institutional Repository and Faculty Papers project by the University of Florida.
Statement of Responsibility:
International Rice Research Institute and Agricultural Development Council.

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 Digital Services ( with any additional information they can provide.
Resource Identifier:
002850530 ( ALEPH )
36870013 ( OCLC )
ANY0937 ( NOTIS )
9711040824 ( ISBN )

Full Text
71. IV
AA /*
-J 7' or

C"1umSeliellces of SnilIt- 1run Aleci ha iimatiion
INTERNATIONAL RICE RESEARCH INSTITUTE Los Baios, Laguna, Philippines P.O. Box 933, Manila, Philippines

MECHANIZATION OF SMALL FARMS in South and Southeast Asia affects all production stages from land preparation through crop production, harvesting, and postharvest processing. Because adoption of agricultural machinery increases capital input, it is expected that either the level of output should increase or the level of other inputs decrease. Such changes are likely to have an impact on employment, income and income distribution, and social welfare.
Evaluation of the consequences of mechanization as well as formulation of appropriate policy guidelines is a predominantly empirical issue strongly related to local socioeconomic conditions. The most frequently used analytical approach is a detailed study of a relatively specific situation.
In 1977, the United States Agency for International Development (USAID) agreed to fund a research project The Consequences of Small Farm Mechanization on Production, Income, and Rural Employment in Selected Countries of Asia
- undertaken jointly by the International Rice Research Institute (IRRI) and the Agricultural Development Council Inc. (A/ D/C). The project has involved collaborative work with national research institutions in Indonesia and Thailand, as well as smaller projects with individual researchers in many countries.
A series of planning and monitoring workshops was held at IRRI to review project progress. The fourth workshop, in which study results from both A/D/C and IRRI were presented, was held 14-18 September 1981. A summary report of the papers and discussion is available upon request from the IRRI Agricultural Engineering Department.
A committee comprising Dr. K. Adulavidhaya, Dr. H. P. Binswanger, Mr. J. B. Duff, Mr. J. Lingard, Dr. R. Sinaga, Dr. J. P. G. Webster, and Dr. J. A. Wicks
reviewed the papers presented and selected those for inclusion in this book. The committee recommended that three papers on farm mechanization in Pakistan be consolidated into a single paper (second chapter). It also recommended that 12 papers for the Philippines, West Java, and South Sulawesi studies be summarized in the final three chapters.
The work of all who prepared papers for the workshop is gratefully acknowledged. Dr. J. A. Wicks undertook the technical editing of the papers, assisted for specific papers by Mr. J. Lingard, Dr. G. Nelson, and Dr. J. P. G. Webster. G. P. Hettel,
assisted by G. S. Argosino, edited the final papers.
This book should provide useful information on the consequences of mechanization, and we hope it will stimulate further research.
M. S. Swaminathan
Director General
International Rice Research Institute

The International Rice Research Institute (IRRI) was established in 1960 by the Ford and Rockefeller Foundations with the help and approval of the Government of the Philippines. Today IRRI is one of 13 nonprofit international research and training centers supported by the Consultative Group for International Agricultural Research (CGIAR). The CGIAR is sponsored by the Food and Agriculture Organization (FAO) of the United Nations, the International Bank for Reconstruction and Development (World Bank), and the United Nations Development Programme (UNDP). The CGIAR consists of 50 donor countries, international and regional organizations, and private foundations.
IRRI receives support, through the CGIAR, from a number of donors including:
the Asian Development Bank the European Economic Community the Ford Foundation the International Fund for Agricultural Development the OPEC Special Fund the Rockefeller Foundation the United Nations Development Programme
and the international aid agencies of the following governments:
Fed. Rep. Germany India
Netherlands New Zealand Philippines
Switzerland United Kingdom United States
The responsibility for this publication rests with the International Rice Research Institute.
ISBN 971-104-082-4

Mechanization of rice production in developing Asian countries: perspective, evidence, and issues 1
Farm mechanization in Pakistan: policy and practice 15
Innovation in the Philippine agricultural machinery industry 31
Economic analysis of the farm machinery industry and tractor contractor business in Thailand 39
Productivity growth of the agricultural machinery industry in Thailand 51
Domestic resource cost of agricultural mechanization in Thailand: a case study of small rice farms in Supanburi 61
Causes and consequences of power tiller utilization in two areas of Bangladesh 71
Economic, technical, and social aspects of tractor operation and use in South Sulawesi, Indonesia 85
Economics of pump irrigation in Eastern Nepal 95
Effect of tubewells on income and employment: a case study in three villages in Kediri, East Java, Indonesia 107
Comparative analysis of thresher adoption and use in Thailand and the Philippines 119
Labor use patterns and mechanization of rice postharvest processing in Bangladesh 139

Consequences of small rice farm mechanization in the Philippines: a summary of preliminary analyses 151
J. F. SISON, Y. TAN, and J. A. WICKS
Consequences of small rice farm mechanization in West Java: a summary of preliminary analyses 165
Consequences of small rice farm mechanization in South Sulawesi: a summary of preliminary analyses 177

R. W. Herdt
A generalized sequence of mechanization of Asian rice production is developed, based on experience in Japan, Taiwan, and Korea.
By 1978, these countries had reached a relatively high level of mechanized riceland preparation. Labor force growth rates among Asian countries are compared. A net social benefit model for evaluating the desirability of mechanizing rice production is outlined and often neglected factors are identified. Available empirical evidence on some important factors is reviewed and the question of
who benefits from mechanization is addressed.
There is considerable controversy about the desirability of agricultural mechanization in Asia. One extreme view equates mechanization with modernization so that it becomes the major indicator and requirement for development. In the late 1 970s, modernization in agriculture in the People's Republic of China was synonymous with mechanization. Those who point out the psychological benefits from riding a tractor rather than walking behind a carabao have similar ideas in mind.
A more moderate view holds that the functional relationship between power input and agricultural output is analogous to that between fertilizer and yield, so that development requires added power in the form of mechanization (Hamid 1979). Binswanger (1978) refers to this as the net contribution view.
A third view holds that the major objective is absorbing the significantly increasing numbers of laborers in the agricultural sector during the next 20 years, and that mechanization can help through intensifying land use (Southworth 1974). In this view, mechanization is the key to increased cropping intensity which will permit labor absorption at other times during the production cycle. A major benefit of mechanization then is the increased agricultural output generated from larger harvested area and higher yields resulting from deeper plowing and better cultivation practices.
The International Rice Research Institute.

A fourth view, somewhat different from the first three, opposes agricultural mechanization on the grounds that it is a straightforward substitution of capital for labor and, given the labor supply in most Asian countries, is socially undesirable (Abercrombie 1975). In some cases, this is supplemented with the idea that distortions in the price ratio of labor to capital have been a primary factor responsible for speeding mechanization, and that nonmarket forces have been largely responsible for these distortions.
The purpose of this paper is:
" to provide a perspective on the issues through an examination of historical
experience in East Asia; and
" to provide a framework for criteria tojudge the impact of machinery introduction and other technological changes.
One additional note: mechanization means different things in different situations. Our discussion deals with powered equipment, mostly two-wheel tractors (power tillers) and power threshers.
Economic, technical, and policy factors are important in determining the pattern and speed with which rice production is mechanized in a country. Alternative investment opportunities and the prices of land, labor and capital influence demand for farm machinery. The perceived social opportunity costs of these resources influence policies which restrain or encourage mechanization, and the relative abundance of resources influences their private and social costs. Technical factors which influence engineering feasibility and the relative cost of mechanization include the amount of power required for a given task, the degree of judgment needed to apply the power, and the question of whether the task requires moving through the field. Climatic or soil conditions may also influence the relative difficulty of designing successful machines.
A generalized sequence of mechanization Because every country in Asia has different technical and economic conditions, it is unlikely that identical patterns of mechanization will take place. However, the broad similarities in relative factor abundance and the tasks required for wet-rice cultivation indicate a general pattern of'rice mechanization is likely to emerge. To date, among the Asian rice economies, only Japan has fully mechanized production. Following the Japanese pattern, Taiwan and Korea are well started and a number of other Asian countries are moving in the same direction.
In East Asia, investment in land improvements and water control preceded mechanization. This was partly an accident of history. Improved water control was carried out using human and draft animal power, and was one of the few ways to improve the productivity of agricultural land in the high mainland economies. Most of Japan's rice fields were supplied with irrigation facilities by the 19th century, and the major subsequent improvement was investment in drainage, providing a high degree of water control (Ishikawa 1981). The same water control investment occurred later in Korea and Taiwan, where there were "many years of rural

infrastructure creation" prior to 1940 (Fei and Ranis 1975). Most of this took the form of gravity irrigation and drainage but, with the availability of electricity and internal combustion engines, power pumps have become one of the first machinery investments for many rice producers. In areas of South and Southeast Asia, where gravity systems are inadequate to permit efficient water control and ground water resources are available, investment in private pumps has been substantial (Patel and Patel 1971). Electric irrigation pumps replaced foot-operated pumps in Japan during the 1 920s, long before power tillers were used (Ishikawa 198 1). A 1966 study in an intensive rice double-cropping area of Taiwan noted that 1 water pump was available for every 3 farms while there was only 1 power tiller for every 18 farms (Lai 1972).
In addition to pumps, other investments in land were important prerequisites for successful mechanization in East Asia. Drainage to provide a hard enough soil base for machines, enlargement of plots and consolidation of fragments, and construction of farm roads to reach individual plots were important in some areas (Tsuchiya 1972).
The introduction of land preparation equipment and threshers followed development of a high quality land base. Small manual threshers were among the first widely adopted mechanical devices (Lee 1972).
They were later introduced into a number of other countries by the Chinese and Japanese, but never became established. Power threshers however, were widely adopted in both East Asia and some areas of other countries. In Japan, two-wheel power tillers were introduced in the early 1950s. L-and preparation was the first operation mechanized in parts of the Philippines and Thailand. Tillers may initially be very small as in the case of the 2- to 3-hp "iron cow" introduced in Taiwan (C. K. H. Wu 1972), but after some years machines in the 8- to 12-hp range seem to take over.
It seems to be more difficult to develop appropriate machines for other rice production tasks such as planting, fertilizing, cultivating, and drying. These operations present formidable technical problems. Weeding, for example, requires considerable judgment and relatively little power. Some mechanical weeders have been developed, but herbicides have proven to be cheaper and more effective at distinguishing weeds from rice. Transplanting has been mechanized, but requires special techniques for raising seedlings and is still quite labor intensive. Despite these problems, the Japanese had developed commercial machines for each major operation by the late 1970s.
There are substantial divergences from this path. Four-wheel tractors of 35-60 hp have been introduced into Thailand, Malaysia, the Philippines, and Pakistan. In some areas, these units are rented for initial land preparation, and draft cattle used for secondary land preparation. Old and new technologies coexist. The large tractors have, in some cases, preceded a high degree of water control development as in central Luzon or central Thailand where their presence in sugarcane farming may have stimulated adaption to rice. In other cases, government authorities (such as the Muda River Development Authority of Malaysia) own the tractors and rent them to farmers.
Data to measure the degree of, and forces contributing to, mechanization are

somewhat fragmentary. The FAQ publish data on wheel and crawler tractors, garden tractors, and a few other kinds of machinery, but there are many gaps in the series for Asian countries. Definitions are unclear and may be inconsistent from country to country. Time series data on the number of power tillers and farm wage rates are available for Japan, Taiwan, and Korea for a number of years, although the wage rates for Taiwan are incomplete. The data provide some clues to the importance of certain forces in the mechanization process.
A supply and demand model provides an oversimplified, but still useful,view. Rice production machinery can be supplied either by imports or from domestic production. Imports are generally under government limitation or licensing. Domestic production may occur through private initiative, but experience in Japan and Taiwan shows that concessional government credit, subsidies, tax exemptions, and government efforts have been major forces speeding the development of appropriate rice production machinery (Kudo 1972, T. C. Wu 1972).
Machinery supply from the private sector is a function of the development stage of the industrial sector and the alternative earning opportunities for industrial capacity. The latter is related to the scale of investment needed to begin production of farm machinery compared to other industrial products. Potential earnings of export industries may make governments willing to set policies encouraging or discouraging mechanization. Government investment in farm machinery research and development is an obvious encouragement, while taxation and import restrictions are barriers. Of course, there are limitations in the extent to which governments will subsidize imported machinery or invest in machinery research and development. So, the supply of agricultural machinery is determined by market forces in the industrial sector, together with the decisions of government to tax or subsidize imported machinery.
The demand for rice production machinery is determined by the degree to which it substitutes ,for labor or other inputs, the price of labor relative to substitutes, and the price of rice. Machinery well adapted to the technical requirements of a particular agricultural setting will be more productive and substitute for a greater value of alternative inputs. A high price for rice increases the demand for machinery.
The data in Table 1 are broadly consistent with the preceding static equilibrium concepts, although many other forces within each country affect the mechanization level. They show that Japan is the most highly mechanized country of the region. In the early 1970s, it had 6 power tillers for every 10 ha of cropland and 5 tractors for every 100 ha. Farm wage rates approached $10/day and the price of rice was the highest anywhere in the world. Taiwan and Korea had wage rates about four times as high as any other country except Japan and were well started toward the adoption of power tillers. Korea had the second highest rice price with Taiwan not far behind. Several countries had a small number of tractors, but these were used for road transportation, plantation and other nonrice crops (Sri Lanka and Malaysia), or on a contract basis for initial land preparation for rice (Malaysia, Philippines, Thailand). Power tillers had been introduced in small numbers to many countries, but except for Thailand and the Philippines, they were still available in strictly experimental numbers. Wage-rates and rice prices in most South and Southeast Asian countries, with the possible exception of Malaysia, were far below those in the East Asian

Table 1. Power tiller and tractor numbers, wage rates, and rice prices in Asia, 1971-75 average.a
Tillers Wheel, Farm wage rate Farm price of
(no./1000 ha crawler rough rice
cropland) tractors US$/day rice/day (US$/t)
(no./1000 ha)
South Asia b
India 1.0 0.26 2.1 125
Pakistan +c 1.6 0.39 3.3 119
Sri Lanka 0.1 6.1 0.42 2.6 161
Bangladesh + 0.3 0.68 3.3 206
Southeast Asia
Malaysia 0.4 2.3 2.53d 12.9d 195
Thailand 8.0d 1.1 0.59 7.9 75
Philippines 4.1d 1.0 0.34. 3.1 109
Indonesia + 0.5 0.71a 4.3d 167d
Burma 0.1 0.8 0.39 7.0 56
East Asia
Japan 615 48.5 8.78 15.6 563
Taiwan 38 0.6 2.80 17.1 164
Korea 20 0.1 2.07 9.3 223
China 130e 7.5 n.a. n.a. n.a.
aSources: Tiller numbers from the IRRI Agricultural Engineering Department. Tractor numbers from FAO Production Yearbooks, but tiller numbers in Thailand and China are from Ishikawa (1981). Wage rates and rice prices from World rice statistics (IRRI). World price averaged $310/t of rice equal to about $200/t of rice over the period 1971-75. Data for China from personal communication from Hua Guozhu, vice director, Chinese Academy of Agricultural Mechanization Sciences. bNo data available, but the author estimates there are less than 0.5/1,000 ha. cNo statistical estimates available; the author estimates less than 0.05/1,000 ha. dRefers to 1976. eRefers to 1978.
countries. So it appears that rice production in most of South and Southeast Asia was not poised for rapid mechanization. However, even though Japan was so far ahead of the other Asian countries, examining mechanization there may still be relevant for the rest of Asia.
Japan's experience
In 1950, agriculture in Japan wasjust entering the early stages of mechanization with many small pedal threshers, some 13,000 power tillers, and an equal number of power sprayers, but essentially no other machinery. The industrial sector, recovering from World War II, was beginning to pull increasing numbers of workers from rural areas. Thirty years later more than 4 million power tillers were being used, and rice production, from transplanting to harvesting, was essentially mechanized.
In 1950, Japan had about 2 power tillers/ 1,000 ha of cropland. Early data on threshers are not available, but in 1955 there was I thresher/ 3 ha. By 1960, there was I tiller/ 12 ha and I thresher/2.5 ha. During the 1960s, the size and capacity of Japanese tillers increased and the riding tiller was introduced. By 1970, there was 1 tiller/ 1.7 ha, apparently more than adequate. Thereafter, the number of ordinary walking tillers remained constant while there was a rapid increase in the number of riding tillers. The 1970s also saw the rapid introduction of powered rice transplant-

ing machines and combine harvesters, and more power sprayers and dusters. There were also significant numbers of power reapers and reaper-binders during the 1960s although data are fragmentary.
During the same period, there was a sharp decline in the agricultural labor force and a steady reduction in the hours of labor per hectare in rice production (Ishikawa 198 1) as machines were substituted for human labor. Some observers clearly saw the trend toward mechanization as a drive to achieve economic efficiency, *pushed by rising labor costs, rather than as a continuation of the earlier Japanese drive to increase yields. Tsuchiya (1972) claimed that "mechanization has been advanced in order to secure a certain amount of rice at the lowest possible cost, rather than to increase the yield."
Differences within East Asia
Korea and Taiwan are the only Asian rice producing countries other than Japan to have achieved any significant level of mechanization by 1980. The pattern of power tiller introduction in the three countries is in Figure 1.
In 1960, Taiwan had as many tillers per 1,000 ha as Japan had in 1950, but the number increased more slowly. After 10 years, Taiwan still had only about 2 tillers/ 100 ha. However, by 1977, 20 years after the initial introduction, there was nearly 1 tiller/ 10 ha, and by the 1980s nearly all of Taiwan's riceland was prepared by machines. Power tillers were introduced into Korea's rice sector about a decade after introduction in Taiwan, and their rapid adoption was similar to that in Japan, reaching 7 tillers/ 100 ha in about 10 years, with continued rapid increases thereafter.
Four critical points in the mechanization process are identified in Table 2: 1) the
Power tillers (na/1000ho paddy land)
1950 1956 1960 1965 1970 1975 1980
1. Power tillers in 5 Asian countries, 1950-80.

Table 2. Farm-level rice prices and wage rates during comparable periods of agricultural mechanization in Japan, Korea, and Taiwan.
Prices (US$)
-State of Tillers Real wage
mechanization (no./1000 ha) Period Rice/t Wages/ World Rea ad e
day rice (kg paddy/day)
Initial Introduction pre-1950 na na na na
Early 2.5 1950-51 311 0.70 na 3.4
Takeoff 20 1956 328 1.00 134 4.7
Full 100 1961 327 1.47 137 6.9
Initial Introduction 1961 173 0.85 154 7.2
Early 2.5 1968 216 1.36 201 9.6
Takeoff 20 1972 338 2.02 148 9.2
Full 100 1978 559 5.99 367 16.5
Initial Introduction 1955-56 86 na 134 na
Early 2.5 1961a 98 0.88 137 9.7
Takeoff 20 1970 176 1.77 143 10.1
Full 100 1978 376 7.06 367 27.9
aTaiwan passed 2.5 tillers/1,000 ha in 1958, but wage data are not available for that year. 1961 is the first year they are available. bTaiwan had 70 tillers/1,000 ha in 1978, the year for which data are shown. By 1979 or 1980, it had undoubtedly passed 100 tillers/1,000 ha.
initial stage; 2) the early stage with about 2.5 tillers/ 1,000 ha, a conspicuous number of tillers with an insignificant proportion of the land served; 3) the takeoff stage with about 20 tillers/1,000 ha, about 20% of the land served; and 4) full mechanization with about 100 tillers/ 1,000 ha, enough to (theoretically) serve the entire rice area. The data for East Asia show great differences in real agricultural wage measured in rice equivalent at mechanization takeoff. At the takeoff stage, a day of labor cost about 5 kg of rice in Japan, about 9 kg in Korea, and about 10 kg in Taiwan. These differences suggest that farm-level demand for machinery may have been very different in the three countries. However, wage rates converted into kilograms office reflect the domestic price of rice as well as of labor. All three countries, indeed all Asian countries, insulate their rice prices from the world market, thereby distorting them in one direction or the other.
In contrast, the industrial sectors of the three countries have been well aligned to opportunities in the international market. So, it can be argued that the cost of mechanization reflected the world market and the price of rice reflected policy views on how to achieve the desired pace of development. In Japan in the early 1950s, both labor and rice were so valuable as potential foreign exchange earners that there was a strong drive to mechanize. In addition, institutional factors, such as restraints on land sales and consolidation to larger units, encouraged the development of parttime farming which could only conveniently be done with machinery. Korea has followed a similar path, but with an even higher price of rice, encouraging rapid mechanization in the 1970s. Taiwan, in contrast, has had a low rice price and has maintained more labor on the farm through rural industrialization policies, and thus retarded mechanization.

The East Asian countries mechanized when labor used in rice production became sufficiently valuable in alternative industrial employment that it could no longer be used in the rice sector. The data in Table I indicate that only Malaysia, Thailand, and Burma have ratios of agricultural wages to rice prices that approach those in the East Asian countries, but that the low value of rice in Thailand and Burma makes wages appear high. It appears that Malaysia may be the next country to adopt rice mechanization.
Labor force growth rates
Knowing the growth rate of the nonagricultural labor force, the total labor force, and agriculture's current share, one may calculate the growth rate of the agricultural labor force needed to absorb the available labor in an economy (Mellor 1966). This process can be understood by examining the following relationship.
x =(I -a) Z+ aY (1)
x is growth rate in the labor force as a whole,
Z is growth rate of the agricultural labor force,
Y is growth rate of the nonagricultural labor force, and
a is proportion of the labor force in the nonagricultural sector.
This relationship states that the growth rate in the total labor force is the weighted average of the growth rates of the labor force in the two sectors of the economy. Given this, one can easily determine the growth rates, needed in the two sectors to achieve certain objectives. For example, one can calculate the growth rate in nonagricultural employment required to absorb all new entrants to the labor force. Conversely, one can determine how fast the agricultural labor force will grow, given certain initial conditions, by rewriting equation I as:
Z =(x aY)/I- a) (2)
So, for a country with 25% of the labor force in the nonagricultural sector, an 8% growth rate in the nonagricultural labor force, and a 3% growth rate in the total labor force, the relationship yields 1.33% (3 0.25(8)/ 0.75), while a country with the same proportion in the nonagricultural sector and the same overall growth in total labor force, but only 5% growth in the nonagricultural sector, yields 2.33%.
Typical values for important Asian rice-growing countries are in Table 3. The growth rate of the labor force in the nonagricultural sector is not readily available, so we use the growth in urban population. This, if anything, overstates the growth rate of the nonagricultural labor force and understates the growth rate of the agricultural labor force. Among the Asian countries, only Korea, Taiwan, and Japan have reached the stage of declining agricultural labor forces. Burma had a temporary decline in the early 1970s because of very low rice prices and earlier slow population growth. This will change over the next 2 decades with the rise in the growth rate of the total labor force. All the other countries have added to their agricultural labor force at 1%/ year or more during the 1970s and will continue to add 1-2%/ year

Table 3. Growth rates of total labor force (L. F.) and nonagricultural portion, and resulting residual rate of growth in agricultural L. F., Asian countries, 1970-2000a.
Growth rate Growth rate Resulting calculated
of labor force of nonagricl- of LF. not growth in agricultural
1970-7 197-2000 tural L.Fia, in agricultural ,F
1970.77 1977-2000 1970-75 1977 1970-75 1977-2000
Thailand 2.5 2.3 3.5 23 2.20 1.94
Pakistan 2.4 2.8 4.1 42 1.17 1.86
Philippines 2.1 2.6 3.5 49 0.75 1.74
Bangladesh 2.3 2.7 6.3 22 1.17 1.68
Burma 1.4 1.9 3.8 45 -0.56 1.46
India 1.7 1.9 3.1 27 1.18 1.46
Indonesia 2.0 1.9 3.3 40 1.13 0.97
Sri Lanka 2.1 2.2 3.7 46 0.74 0.92
Malaysia 3.6 3.0 4.8 56 2.07 0.71
Korea 2.9 1.9 5.4 55 -0.16 -2.38
Taiwan 1.9 1.6 4.4 66 -2.95 -3.84
Japan 1.3 0.8 2.4 86 -5.46 -9.03
aCalculated from data in IBRD, World Development Project Report, 1979. bGrowth rate in urban population used as proxy.
during the next 20 years. The main rice producing countries of Asia have had, and will continue to have, increasing numbers of people available for rice production.
Social benefit:cost framework
Given the increasing supply of agricultural labor, questions must be raised about the, desirability of introducing machines that replace labor. The argument that farm machinery may provide the key input necessary to overcome a labor bottleneck is persuasive, but should be reflected in economic benefits. For economies struggling with development problems, mechanization can only be justified if it generates benefits for society regardless of the psychological benefits of seeing tractors in one's field. What are the benefits and costs of mechanization and how are those costs and benefits shared within society? The net social benefit (NSB) bf mechanization, equal to agricultural output valued at shadow prices less agricultural inputs valued at shadow prices, can be stated as:
NSB = QP*r + QPn LrPl rP*k TrPK InPJ KnP*k TnP*t (3) where the subscripts r, n, l, k, and t refer to rice, other crops, labor, capital, and land, respectively; Q's refer to output quantities produced; L, K, and T refer to the input quantities of labor, capital, and land, respectively; and P*s are shadow prices.
Output can be further defined as the product of area harvested times average yield, while area harvested equals geographic area times cropping intensity. Defining Cr, the rice cropping intensity, as total area of rice harvested divided by geographic crop area and C,, the other cropping intensity, as total area of other crops harvested divided by geographic crop area gives
Qr = Cr A Yr (4)
and Q, = Cn A Y (5)

where A is the geographic crop area and Y, and Y,, are the yields per harvested hectare of rice and other crops.
Substituting equations 4 and 5 for 3 gives:
NSB = CrA YrP*r + CnA YP LrP- KrP*k
- TP* L,,P KnP* TrP* (6)
The impact of any particular change, such as that generated by mechanization, can be expressed as the total differential of equation 6:
dNSB = A YrP* + dCr + CrAP*dYr + CrA YrdP* + A YP*,dC,
+ CA P*dY, + CA YndP*n PdLr- LrdP P*kdKr KrdP*k
- P*dTr TrdP* PdLn LdP- P*k dKn KndP*k
- P*dT TndP* (7)
Some components of equation.7 are expected to change dramatically with mechanization. These are the focus of most investigations on the impact of mechanization. Other components are commonly assumed to be constant while still others are generally ignored. A brief review may be useful. P*, P*' : The prices of rice and other crops are generally assumed to be invariant
with the techniques of production although, if output effects of
changes in technique are large, they may change.
Yr, Cr, Y,, c, : The production of rice and other crops may change as a result of
changing yield and changing land use intensity.
A : The geographic crop area is generally assumed to be fixed because
there is little new land to be exploited except in a few remote areas.
Pt. Pk. P1 : The shadow prices of labor, capital, and land are generally considered
to be invariant with respect to changes in production techniques. This assumption may be valid only in the short run where the rice production sector is small relative to total use of each input.
P= = P :+i The shadow prices of inputs equal their market prices, Pi, plus adifference
which may be recognized as a distortion factor, Di.
L,, Kr, T, : The use of inputs in rice production is a major factor usually assumed
to change with technical innovation.
L,, K,, T : The use of inputs in production of other crops may also change with a
change of technology, but these are usually ignored.
Because many of the factors that may change have been ignored in the empirical work on the impact of mechanization, it is impossible to draw any firm conclusions although investigations to date provide some indications as to the size of certain impacts.
A second dimension of the impact of mechanization is the distribution of NSB among earners in the production process. Relatively little on this issue is available in the empirical literature, aside from the obvious statement that, in a market-directed economy, laborers receive the returns to labor and farmers, or capitalists, receive the

returns to capital. All consuming groups would obtain some portion of the benefits to increased output via its impact on the market price of rice.
Empirical studies have been cast in a much narrower framework than just suggested and so it is impossible to evaluate from them the NSB of any change. However, a necessary (but not sufficient) condition for society to benefit from the change is that either
" the total output value of crops produced be increased with no change in value of
inputs; or
" machinery be introduced where the social cost of labor is rising relative to the
social cost of capital.
In either case, the identified change will have a positive impact on NSB, which will be reflected in a lower social cost of rice production. This lower social cost of output, if transmitted through the market, will result in lower rice prices. If neither condition holds, then it is unlikely that there is any NSB from mechanization.
NSB, as used in this discussion, has only an efficiency implication. Equity or overall social desirability is not inferred by either its increase or decrease. A situation, leading to an increase in NSB with a decrease in equity or a decrease in NSB with an increase in equity, may be selected because society decides it is worthwhile to trade off some equity for the efficiency obtained. It seems evident, however, that society would never knowingly choose a situation that leads to a decrease in NSB and a decrease in equity. Given the existing price distortions (taxes, tariffs, subsidies, imperfect competition) in most developing economies, it is likely that use of market prices to evaluate benefits may produce an erroneous conclusion of a given change leading to increased efficiency. The importance of using shadow prices cannot be overstressed when evaluating NSB.
A general equilibrium framework
Although the formulation of NSB in the preceding section takes account of the impact on rice and other crops, it is a single-sector, partial-equilibrium model because it does not account for changes that may occur in the nonagricultural sector of the economy. That requires a whole economy or general equilibrium. approach. If rice production uses domestically manufactured machinery, rice production mechanization may generate considerable income and employment in the machinery sector. The amount will vary with the labor intensity of machinery production. The increased income generated from manufacturing machines will generate an increased demand for rice. If that increased income is concentrated in economic classes with a high income elasticity of demand for rice, the feedback impact on the rice sector will be larger than if the increased income went to classes with a low income elasticity of demand.
These interrelationships can be measured in an input-output (I-0) model of the type developed by Leontief (195 1). 1-0 models define economy as a whole in which all inputs used in the production of each commodity are identified and measured. Manipulation of these models can lead to insights on the impact the changes in one sector can have on other sectors. 1-0 analysis has the advantage of quantifying both the direct employment impact of machines substituting for labor in rice production, as well as the indirect impact of the increased labor used in producing, distributing,

and servicing these machines. Such models can be developed by building on existing 1-0 models which planning ministries have already developed for most countries.
If mechanization results in increased output, it will tend to push prices down and the benefits will be shared by rice consumers, whether they be landless agricultural workers, farmers, or urban people (Hayami and Herdt 1977). The absolute benefits to various individuals are positively related to the proportion of their incomes spent on rice consumption.
In the absence of increased output, machinery adoption may shift earnings from one group to another. That is, a machine which replaces labor will receive the wage formerly paid to the laborers. In such an event, the machine owner receives the earnings formerly paid to laborers. There is an inherent difference in the ownership pattern of capital and labor. In the absence of slavery, labor can only be owned at a rate of I unit/ person, or at most 5-10 units/ household. On the other hand, ownership of capital can be and, in most economies, is concentrated in the hands of relatively few, usually through inheritance, political power, or business acumen. Concentrated capital ownership means that income earned by capital also is concentrated. The introduction of machinery has redistributing effects which, when it also leads to increased output, will add to the welfare of low income rice consumers and technology adopters. When machinery has no output effect, it simply redistributes income.
Farmers will have an incentive to adopt machines when the machines reduce production costs. However, if reduced costs are achieved through subsidies on the purchase price or through low cost credit, then farmers are responding to artificial (policy-induced) market prices that diverge from the real, or shadow prices. If machine use results in a faster output growth rate, policyrnakers must evaluate the trade-off between more output and income redistribution from labor to capital. But if there is no output effect, the decision to promote mechanization supports a transfer of income from labor to machinery owners, without an offsetting benefit.
The extent to which landless agricultural workers depend on earnings from hired farm employment varies widely. Under many conditions, they are highly dependent on agriculture, and perhaps even more dependent on the earnings obtained during the harvesting season. Clearly, the actual impact of any reduction in the demand for labor depends on the proportion of their incomes deriving from operations which become mechanized and the alternative employment opportunities. Because that varies, careful assessment of the likely impact of mechanization is very important.

Abercrombie, M. 1975. Agricultural mechanization and employment in developing countries. In Effects of farm mechanization on production and employment. FAO, Rome. Binswanger, H. P. 1978. The economics of tractors in South Asia. Agricultural Development
Council and ICRISAT.
Fei, J. C. H. and G. Ranis. 1975. Agriculture in two types of open economies. In L. Reynolds,
ed. Agriculture in development theory. Yale University Press, New Haven.
Hamid, J. 1979. Agricultural mechanization: a case for fractional technology. In Tan Bock
Thiam and Shao-er Ong, eds. Readings in Asian farm management. University of
Singapore Press, Singapore.
Hayami, Y. and R. W. Herdt. 1977. Market price effects of technological change on income
distribution on semisubsistence agriculture. Am. J. Agric. Econ. 59(2):245-56.
Ishikawa, S. 1981. Essays on technology, employment and institutions in economic development. Econ. Res. Ser. 19. Institute of Economic Research, Hitotsubashi University,
Tokyo, Kinokuniya Company, Ltd.
Kudo, Z. 1972. Implications of farm management research for government mechanization
programs. In H. Southworth, ed. Farm mechanization in East Asia. Agricultural
Development Council, New York.
Lai, W. C. 1972. Current problems of farm management on mechanized farms. In H.
Southworth, ed. Farm mechanization in East Asia. Agricultural Development Council,
New York.
Lee, C. C. 1972. Economic and engineering aspects of mechanization of rice harvesting in
Korea. In H. Southworth, ed. Farm mechanization in East Asia Agricultural Development Council, New York.
Leontief, W. W. 1951. The structure of American economy, 1919-39. 2nd ed. Oxford
University Press, New York.
Mellor, J. W. 1966. The economics of agricultural development. Cornell U. Press, Ithaca,
New York.
Patel, S. M., and K. U. Patel. 1971. Economics of tubewell irrigation. Indian Institute of
Management, Ahmedabad.
Southworth, H. M. 1974. Some dilemmas of agricultural mechanization. In H. Southworth
and M. Barnett, ed. Experience in farm mechanization in Southeast Asia. Agricultural
Development Council, New York.
Tsuchiya, K. 1972. Mechanization and relations between farm, non-farm and government
sectors. In H. Southworth, ed. Farm mechanization in East Asia. Agricultural Development Council, New York.
Wu, C. K. H. 1972. Analysis of machinery-labor relationship in farm mechanization. In H.
Southworth, ed. Farm mechanization in East Asia. Agricultural Development Council,
New York.
Wu, T. C. 1972. Government policies promoting farm mechanization. In H. Southworth, ed.
Farm mechanization in East Asia. Agricultural Development Council, New York.

B. Lockwood, M. Munir, K. A. Hussain, and J. Gardezi
The first part of this report contains reviews of the government of Pakistan's farm mechanization policy from 1975 to the present, and of a 1975 World Bank study questioning the suitability of rapid introduction of tractors. The study concluded that although farmers with tractors had good returns on their mechanized investments, these private gains were at the expense of substantial social advances for rural society. The second part presents some results of an A/ D/ C-funded study of farm mechanization in Punjab Province conducted in 1978-79. Data on the behavioral pattern of farms with a tractor generally supported the findings of the World Bank study particularly in terms of the transfer of farmland from tenants to farmers with tractors and the loss of jobs and
earnings for the rural landless community.
Agriculture is the largest sector in the Pakistan economy and 75% of the population live in rural areas. Fifty-three percent of the work force is employed in agriculture, which produces about 30% of the GNP and accounts for about 36% of foreign exchange earnings from merchandise exports. Large parts of the industrial and service sectors depend on raw materials and customers from the agricultural sector.
The agricultural sector consists of some 5 million farms with an average area of 4 ha, a highly skewed distribution, and a high incidence of tenancy. About 700/0 of the 19.3 million ha of cultivated farmland are irrigated, mainly from a large network of canals which provide water to 74% of the irrigated area. The late 1960s to the early 1970s was a period of rapid agricultural change. There were improvements in the availability and control of water through an expansion of canal capacity and considerable farmer investment in tube wells, increasingly greater supplies of chemical fertilizer, and the rapid adoption of high yielding wheat and rice varieties. These changes led to more intensive farm operations and land use, greater farm income, and increasing demand for more and better farm power. The Agricultural Development Council, Inc., and University of Agriculture, Faisalabad.

In 1965, animals constituted the main source of farm power there were about 10.5 million work animals and only about 10,000 tractors. Between 1966 and 1970, 18,000 tractors, mainly in the 36 to 55-hp range, were imported, but there remained an increasing government concern that agriculture was being adversely affected by a power shortage, and that mechanization was the answer.
The main set of guidelines governing farm mechanization policy in Pakistan was drawn up in a major government study between 1968 and 1970 (Farm Mechanization Committee 1969, 1970). The committee estimated that tractors supplied only 14% of available farm power in 1978 and that work animals still dominated the scene with 75%. Total available farm power was estimated to be about 0.04 hp/ha of cultivated farmland; a suggested minimum requirement was 0.08 hp/ha.
The Farm Mechanization Committee proposed a tractor import program (Table 1) and the popularization of tractor-powered implements. The 1968 survey had shown that, while most tractor owners owned cultivators, very few owned moldboard and disk plows for primary tillage, or row planters, seed drills, fertilizer distributors, wheat threshers, and combines. The tractor import program was expected to increase total farm power availability from 0.04 in 1968 to 0.068 hp/ ha in 1985. The share supplied by animal power would decline from 75 to 35%, the contribution of human power would fall from I I to 6%, and the share provided by tractors would rise from 14 to 59%. In absolute terms, the committee predicted a decline in work animals from 3.65 to 3.26 million hp, a small increase in human labor from 0.53 to 0.55 million hp, and a substantial increase in tractors from 0.66 to
5.46 million hp.
This program was the crux of the committee's report and is the basis of Pakistan's farm mechanization policy to the present. The committee also made recommendations on a wide range of related aspects of farm mechanization. These included agricultural conditions, socioeconomic aspects of farm mechanization, standardization, manufacturing, spare parts, repairs and serving, prices of agricultural machinery (including duties and taxes), the system and financing of imports, credit, technical manpower requirements, facilities for education and training, and research and extension.
The committee recognized that tractors were going predominantly to large farms, and that there was a tendency for large-tractor farms to grow even larger through resumption of land previously farmed by tenants. The ex-tenants, however, were
Table 1. Farm Mechanization Committee's proposed tractor imports program.
Tractors (no.)
Time Progressive total
At beginning of Replacement Net Total import/ at the end of
plan period needs addition manufacture pa period
1969 17,100 500 4,000 4,500 21lJflO
1971-75 25,000 12,200 23,600 35,800 48,600
1976-80 48,600 24,200 32,700 56,900 81,300
1981-85 81,300 46,100 41,100 87,200 122,400

being absorbed as farm labor and the tractors on such farms were serving as a supplemental power source, not necessarily as tenant replacements. The small farmers were self-cultivators and the introduction of tractors did not create any tenant displacement problems (Farm Mechanization Committee 1970).
All machines are imported since Pakistan has not developed the capacity to manufacture tractors. Tractor introduction began slowly, and until 1957 there were no restrictions on makes and models. Consequently, there were at least 30 different makes or models operating and owners were experiencing maintenance and spare parts problems. In 1978, the government restricted future imports to seven makes, then continued to control and standardize the range of tractors available. While this policy made sense, it had a checkered history as the government depended increasingly on credit and barter trade arrangements for the supply of tractors and changed the range of acceptable makes accordingly.
Between 1965 and 1969, the World Bank, through provision of International Development Association (IDA) credits totaling $43 million, was a major financer of the farm tractor program. By 1970, the Bank had become concerned about possible adverse effects and initiated a study to examine "the major consequences of the introduction of large scale tractor technology in Pakistan." The report (Mclnerney and Donaldson 1975) was based on a survey of 202 farmers who had purchased tractors through the first IDA credit, mainly in 1967. The main findings were:
" the average size of the farms grew by a factor of 2.4,
" the average number of crops cultivated increased from 4.77 to 7.30 per farm and
cropping intensity increased from 111.5% to 119%,
" the areas sown to fodder crops decreased by 50% with land transferred mainly
to wheat and rice,
" labor use per farm increased but labor use per cultivated hectare declined by
" taking into account displacement of tenants each tractor replaced 8-12 full-time
" tractor use averaged about 1,200 hours/year,
* the private rate of return on investment was 57%,
* the social rate of return was 24%, and
* significant social costs resulted from adjustments in the pattern of resource use.
Two major policy questions arose 1) why did farmers adopt such a behavioral pattern, and 2) how can it be reversed. Mclnerney and Donaldson (1975) broadly discussed the kind of policies which could increase social benefits and reduce social costs. They also questioned the appropriateness of the basic unit of the mechanization program: a 10- to 30-hp tractor could spread the private benefits more widely among the rural population and incur lower social costs.
The land reform legislation of 1972 and 1977 put ceilings on land ownership and attempted to improve the security of tenants, but landowners were adept at getting around the legislation and the tenant position worsened. There is no evidence that the land reform legislation did anything to curb the predatory behavior of the tractor-farms, and the land accumulation activities of farmers who have acquired tractors have continued unchecked.

Government farm mechanization policy was not affected by high social costs. It continues to be based on the power shortage argument of the Farm Mechanization Committee and implementation involves little more than the importing of tractors and provision of cheap credit to tractor buyers. The policy is stated in the Fifth Five-Year Development Plan 1978-83 as 1) the "liberal import of tractors, sold at market price without subsidy;" 2) improved availability of tractors through their "assembly and progressive manufacture" in Pakistan; and 3) "allowing the importation of 2-year-old second-hand tractors and power tillers ... freely against genuine foreign exchange earnings/ savings of Pakistanis working abroad" (Planning Commission 1978).
The main question is not "whether tractors" but "how many tractors." An import program was recommended by the Farm Mechanization Committee for 1971 to 1985. In 1975, a FAO Mission was commissioned to review the program for the Fifth Development Plan and recommended that 15,000 tractors should be imported in each of 1976 and 1977 to catch up with "demand," and that during the plan period, annual imports should be between 10,000 and 11,000. When the plan came out, it specified that "about 15,000 tractors will be imported annually ... to wipe out, inter alia, the backlog demand. The net population of tractors, excluding replacement is expected to go up from about 7 1,000 in 1977-78 to 1 11,000 in 1982-83" (Planning Commission 1978). On 16 August 1981 the Federal Agriculture and Food Minister announced that the government had decided to import 20,000 tractors annually.
Average annual imports from 1965-66 to 1974-75 were 4,040 units, mainly in the 36- to 55-hp range, but between 1975-76 and 1980-81 average annual imports stood at 14,470, with a shift to the 47- to 66-hp range.
Other aspects of the mechanization policy, for example, assembly and manufacture of tractors, power tillers, and implements, have not been implemented as vigorously. The Agricultural Development Bank of Pakistan (ADBP) has played the leading role in providing cheap credit (12%1) for purchasing tractors and tubewells. It has consistently provided loans on about half the new tractors purchased each year, covering about 90% of the purchase costs. A disproportionately high percentage of these loans have gone to large (influential) farmers. To counter this, ADBP and the Rural Supply Cooperative Corporation initiated in 1979, a special scheme to lend up to 50% of the cost of a tractor to Punjab farmers with 5-10 ha. In 1979,784 tractors were financed under this scheme (13% of all tractor loans). Also in 1979, ADBP started making loans for purchasing wheat threshers. Until then, the bank had kept away from financing tractor-powered farm implements.
During 1978-79 A/ D/ C supported three projects on farm mechanization in Pakistan to investigate:
1. farmers' decision making for investment in farm machinery, with special
reference to tractors;
2. the capacity of workshops and farmers to repair and maintain farm machinery;
3. the effect of the adoption of mechanical wheat threshing on rural labor.

Study procedure
The team conducted three surveys during 1978-79, two involving farmers and one involving tractor repair/ maintenance workshops, spare parts shops, and implement manufacturers.
Data obtained in the first farm study covered factors affecting farmers' decision making, their capacity to service and maintain tractors, details of tractor use in own-farm and hire service activities, and historical information on farm mechanization. This last requirement was the main influence on the survey design. Faisalabad District was selected as the survey area and a random sample of 40 villages was drawn from the district subdivisions. The villages were stratified so that half were on main roads and half on minor roads. A presurvey visit identified 125 farmers owning 129 tractors. Villages with fewer than two tractors were excluded. Four farms were operated by managers for absentee owners and the managers did not feel free to cooperate; three were owned by nonfarmers and 19 farmers could not be contacted during the survey and two subsequent visits. The survey was conducted in 25 villages with 88 tractor-owning farmers.
The sampling method facilitated collection of historical data. It caused problems in assessing the consequences of mechanization on the size of farms, tenants, cropping patterns, and land use intensities. Because we were dealing with farmers who had owned tractors between a few months and 25 years, recall data would have been stretching the memories of some respondents. In analyzing the "before and after" data relating to the major consequences of introducing tractors (farm area, tenants), we compromised by excluding farmers who purchased tractors before 1973.
The thresher study was conducted near Multan, an area where wheat is the major winter season crop and cotton the major monsoon-summer crop. Four villages were selected two in a fairly standard farm area and two in a river-flats area of predominantly large farms. In each village, random samples of 15 farmers each were drawn from mechanical wheat thresher owners and those who threshed their 1978-79 wheat crop by traditional methods. A problem was that very few large and medium farmers were still threshing by bullock. Most were contracting mechanical threshing of their wheat. Therefore, the average wheat area for farms threshing mechanically (by farmers who owned threshers) was 19.4 ha; that for farms threshing with bullocks, only 2.8 ha. This prevented comparing farms which differed only (or even mainly) in threshing method.
Farm mechanization in Faisalabad District Most farmers in areas with usable underground water began mechanization with diesel or electric tubewells. The tubewell often preceded the purchase of a tractor and cultivator by several years. Seventeen respondents only began active farming when they bought their first tractor; before this they had leased out their farmland. Generally, tractor investment committed the farmer first to mechanized land preparation and to other operations such as threshing later. The pace of farm mechanization accelerated after 1973 (Table 2). All farmers bought their first tractor and cultivator simultaneously. Investment in other tractor attachments before 1973 was fairly uncommon but then increased rapidly.

Table 2. Farm machinery purchased by 88 sample farmers. M
Farm machinery purchased (no.)
Tubewell Tractor ana Trailer Wheat Pulley Leveler Cane Other items Tractor farms 0
cultivator thresher crusher rA
To 1965 16 5 .. 1 3 >
1966 5 2 1 --
1967 1 2 6
1968 7 3 1 2 1 1 9
1969 6 13
1970 5 3 4 1 4 1 15 ni
1971 1 5 1 2 4 2 20
1972 1 3 1 2 3 1 22 >
Subtotal 36 29 8 4 8 8 2 3
1973 3 8 2 2 3 3 29
1974 4 13 3 1 8 1 40 0
1975 2 12 8 8 9 7 2 1 50 z
1976 7 18 9 4 6 2 2 65
1977 3 19 16 11 15 12 1 2 80
1978c 5 13 13 23 18 7 2 88
Subtotal 24 83 51 49 51 44 9 7
Total 60 122a 59 53 59 44 9 10
% of farmer
owners in 1978 50 100 63 58 63 46 10 5
aulsually purchased together. Seed drill (5), rototiller (3), disk harrow (2). CTo August 1978 only. dAlthough 112 tractors had been purchased, 22 had been sold. The 90 tractors were operating on the sample farms in 1978.

Table 3. Expected benefits from tractor ownership.
Responses from 88 sample farmers
Expected benefit First Second Third Total
priority priority priority
More farm income 36 22 8 66
Timely farm operations 10 8 3 21
Self-cultivation (removing tenants) 16 4 20
Income from hire services 5 10 1 16
Easier farming 8 3 1 12
Saving bullock expenses 4 6 2 12
Overcoming labor shortage 3 5 4 12
Others 5 2 2 9
The sequence of events culminating in tractor purchase usually began with leasing tractor services for land preparation, cartin, or threshing. In our sample, 96% of 72 tractor farmers who had operated farms prior to buying a tractor had been leasing for 4 to 5 years before purchase. Expectations varied, but clearly, many anticipated increased farm income from more timely operations, capability to manage larger farms (ejection of tenants), and off-farm income from offering tractor services (Table 3). Interestingly, the sons of 30 farmers demanded a tractor before agreeing to stay on the farm.
Generally, farmers had little or no choice as they had to purchase, make and model available when their bank loans were approved, or other funds were in hand. So, ownership pattern is not a good guide to farmers' preferences. In most cases, however, farmers replaced tractors with units of higher horsepower: the trend was from 44- to 45-hp to 46- to 48-hp units and, in many cases, 55- to 65-hp units. Most farmers preferred a particular make and model and matched it and work task with considerable consistency: Ford 4000 (55 hp) for heavy duty general work, Fiat 640 (64 hp) for long periods in the summer heat driving a wheat thresher, Bylarus (55 hp) for heavy hauling of bricks, and the Massey Ferguson 135 (47 hp) for general cultivation work.
The average cost of a tractor increased from $1,800 to $5,810 between 1968 and 1978 and, while the ADBP loans rose, many farmers had to find increasingly large sums from other sources. After 1973, credit from sellers of secondhand tractors and other noninstitutional loans became important, particularly for small farm tractor buyers.
Table 4. Sources of funds for tractor purchases by 88 sample farmers.
OperationaL Percentage of purchase cost
farm size Cash Loan
in 1978 Farms
(ha) (no.) Overseas Farm Other
remittances income income Bank Private Seller Less than 10 23 24 17 14 34 1 11
19-19.5 31 11 17 7 52 9 4
20-39.5 23 2 52 2 4 3 0
40 & above 11 0 19 0 78 3 0
Total 88 10 23 6 49 6 5

Based on average cost figures (1973-78) for the different items, the package investment in a tractor and implements cost about $8,900 tractor ($5,650), cultivator ($320), thresher and pulley ($1,770), and trailer ($1,140). The total investment in farm machinery by the sample farms was about $680,000, of which 48% had come from the farmers' resources, 3 1% from bank loans, 7% from overseas remittances, and 10% from other sources. Bank loans were restricted almost completely to tractor investments, and were biased toward larger farmers (Table 4). Smaller farms relied more on overseas earnings because they bought a number of secondhand tractors on credit from sellers and noninstitutional sources. Tractor attachments were mainly financed from farmers' funds.
Consequences of tractorization
Although there were considerable differences in time and methods, between the current survey and Mclnerney and Donaldson, the results are remarkably similar.
Changes in farm size. Only the 62 farmers who bought their first tractor after 1972 were included in this assessment. Changes between the year before tractor purchase and the end of 1978 are in Table 5. The first group increased their farm size by either reducing the farmland rented out to tenants (70%1) or increasing the area rented in (30%). Forty-five tenants lost their farms, possibly more if the area rented in by the tractor-farmer was previously tenanted. Landlords, who became operating farmers only when they bought tractors, accounted for 17% of the total farm area in 1978. Thirty percent of their 1978 farm area was rented in and 70% was land previously rented out. Thirty-three tenants lost their farms as their landlords became farmers.
Our survey, made almost 10 years after Mclnerney and Donaldson's, should remove some doubts about the causal relationship between farm size expansion and tractor ownership, although not about the direction of the relationship. The resumption of tenanted land argues for the tractor as a final facilitating factor, while the increase in farmland rented in shows the tractor initiated the farm size expansion. In Table 6, classifying farmers into pretractor farm size categories shows that the "predatory behavior" of a large proportion (59%) was not restricted to those beginning as large farmers. The Farm Mechanization Committee had claimed that small farmers behaved differently. In our sample, 82% of the pretractor farmers were
Table 5. Changes in operated farm area, before and after tractor purchase, by direction of change.
Fres Area farmed Area farmed Pootoa
Change, (~n.) before tractor in 1978 Prpotang l
(no.) (ha) (ha) cag
Increased the farm 24 279 532 1.91
area operated
Did not change the 22 274 274 1.00
farm area operated
Decreased the farm 3 56 36 0.64
area operated
Began farming 13 0 171
Did not farm 1 0 0
Total or av 62 609 1,013 1.66

Table 6. Changes in average operated farm area and sources of change before and after tractor purchase, by pretractor farm size.a Pretractor Operated farm (ha) Rented in (ha) Rented out (ha) Tenants (no.) Proportional change
operated farm Farms" prtd Retd Rne
(ha) (no.) Before After Before After Before After Before After Operatedrm Rented Rented
Less than 10 16 5.6 8.8 0.5 1.6 3.0 0.3 0.8 0.2 1.58 3.00 0.11
10-19.5 24 12.2 18.1 1.1 3.2 4.6 0.4 1.1 0.1 1.48 2.96 0.09
20-39.5 8 23.2 28.1 3.2 5.2 6.6 1.9 1.9 0.4 1.21 1.64 0.29
40 and above 1 40.5 40.5 10.1 10.1 1.00 1.00
All farms 49 12.4 17.2 1.5 3.1 4.3 0.6 1.1 0.2 1.38 2.10 0.14
Nonfarmers before tractor 14 0 12.3 0.7 5.1 8.5 0.6 1.9 0.2 6.94 0.07 x
aFarmers who purchased tractors 1973-78.
Table 7. Changes in average operated farm area and sources of change before and after tractor, by year of tractor acquisitionaZ
Year of tractor Farms Operated farm (ha) Rented in (ha) Rented out (ha) Tenants (no.) Proportional change
purchase (no.) Before After Before After Before After Before After Operated Rented Rented
farm in out
1978 7 10.4 14.4 1.8 1.9 3.4 0 0.9 0 1.39 1.04
1977 11 11.2 16.4 0.4 1.6 6.2 1.9 2.0 0.5 1.47 4.33 0.31 0
1976 12 10.7 12.3 2.2 3.8 1.4 0.1 0.4 0.1 1.15 1.73 0.09 r
1975 5 15.1 25.3 2.9 7.0 8.1 0 1.4 0 1.67 2.39 <
1974 9 15.2 21.1 1.3 2.4 5.6 0.8 1.2 0.3 1.39 1.79 0.14 7>
1973 5 14.4 19.1 0 3.2 1.4 0 0.4 0 1.33
All farms 49 12.4 17.2 1.5 3.1 4.3 0.6 1.1 0.2 1.38 2.10 0.14 >
'a()n farmers who acquired their tractor 1973-78 and who had previously farmed.

"small" by the committee's reckoning (up to 20 ha), but they expanded their farms by a factor of 2.51 and expelled 33 tenants. The 14 pretractor landlords were also fairly small landowners: the average holding was 8.5 ha, and only one owned more than 20 ha. The smallest landlord (0.5 ha) expelled his single tenant and rented in another 6 ha. It is clear that expelling tenants and renting in additional land are not restricted to any one group.
We agree with the conclusion of Mclnerney and Donaldson that "the tractor is a powerful force narrowing the spread of farm sizes by pulling up the lower end of the size distribution." These practices do not appear to be declining (Table 7).
Changes in land use. Although one of the dominant arguments for mechanization is that it should lead to significant increases in cropping intensity, Mclnerney and Donaldson found an increase of only 8%. In our survey, the overall result was even more dismal an increase of 4-5%. However, there was considerable variation in cropping intensity changes when farms were categorized by initial size (Table 8). The increase was also greater on farms without tubewells by a factor of 1.20. This suggests that the tubewell farmers had already increased cropping intensity as a result of better water supply and control, and that this, more than the tractor, influenced cropping intensity. However, as in Mclnerney and Donaldson's study land for farm enlargement may have been appropriated from small farms which might have had a relatively high cropping intensity. Then, the cropping intensity of the whole area may have dropped with tractor introduction.
There was a change in the cropping pattern of tractor farms in our sample. Most farms increased the proportion of their cultivated area sown to wheat in the winter season (4%Y), rice in the monsoon-summer season (1%/), and to miscellaneous crops, such as vegetables, in both seasons (5%). At the same time, there were declines in the area sown to maize and cotton in summer (3%1) and to fodder in both seasons (5%). Apart from the decline in fodder acreage, the other changes appear to be similar to those taking place generally in the district and cannot be attributed directly to tractorization.
Effects on tenants. Eighty-eight of the original 105 tenants lost their land when the landowners bought the first tractors. The average tenant farm size declined from 4.4 to 3.4 ha (Table 9). In most cases, tenant ejection took place in the year the tractor was purchased.
Effect on livestock. Because the tractor replaced the bullock in some farm operations, we would expect a decline in draft animals on tractor farms. There were
Table 8. Cropping intensity by farm size before and after tractor purchase.
Cropping intensity
Farm category (ha) Before tractor After tractor Proportional
____________________________ (1977/78) change Less than 10 104 141 1.36
10-19.5 135 138 1.02
20-39.5 138 146 1.06
40 and above 176 155 0.88
All farms 141 146 1.04

Table 9. Tenancy before tractor purchase and in 1978.
Area leased out (ha) Tenants (no.) Av tenancy (ha)
Kind of change Farmers Percentage Beoetaor 17 Bfoercor 98
in farm size (no.) Before tractor 1978 Difference changectr 198 Bfoetrcor 17
Increased 34 278 23 255 91 59 8 4.7 2.9
"New" farmers 17 161 16 145 89 38 5 4.2 3.2
Total 51 439 39 400 91 97 13 4.5 3.0 0
No change 33 16 5 11 69 5 2 3.2 2.4 2
Decreased 4 7 13 6 72 3 2 2.4 6.3
Total 88 462 57 405 88 105 17 4.4 3.4>
a Had not farmed before owning a tractor.

about 333 bullocks on the sample farms before tractor purchase (3.8/ farm); in 1979 there remained 159 (1.8/ farm). On the average, each tractor had replaced 2.0 bullocks. In the Mclnerney and Donaldson survey, the average number of bullocks per farm declined from 5.4 to 2.9 (2.5/ tractor). However, farmers did not act uniformly: 12 sold all of their bullocks (54) after buying a tractor, 44 sold some bullocks and kept some (245 to 107), 10 kept their bullocks (30), 12 farmers had none before or after the tractor, and 10 actually increased their bullocks (4 to 22). Most farmers did not regard the tractor as a perfect substitute for bullocks in all farm operations. In particular, they preferred bullocks for chopping and carting fodder, planking, intercultural operations, cane-crushing, and line sowing. Only two farmers said that they kept bullocks as a standby when their tractors were being repaired.
There was little change in milk animal numbers. Apparently, farmers regarded draft and milk animals as distinct and a reduction in the number of draft animals did not lead to a corresponding increase in the milk herd.
Effect on labor. The tractor-farm survey in Faisalabad District did not obtain data on hired or family labor except to ask how many workers were employed on a full-time or permanent basis. Although the number for 1978 (184) was probably accurate, the pretractor number (164) spread over many years should be treated with caution. Using these numbers, there was an increase of around 12%/ in the number of permanently employed. Two factors complicate this simple arithmetic. First, only 71 of the 88 farmers were operating farms before buying their tractors and, therefore, the number of permanently hired laborers per farm fell by 8%. Second, because there was a 53% increase in the operated farm area, the number employed per cultivated hectare declined by 17%. Another complication was that .86 tenant families were expelled, substantially reducing the on-farm labor pool.
The 1979 survey of wheat threshing operations of traditional and mechanized farms near Multan obtained details of labor use during the harvesting-threshing season. Mechanization caused no change in the labor used for harvesting which was mainly contracted out and, except for carting the cut wheat to the threshing area, was not mechanized. Mechanization substantially reduced the use of labor for threshing. The differences between the traditional and mechanized sample farms farm size, average area of wheat harvested (2.87 and 19.43 ha), and average yields (1.47 and 2.29 t/ha) made direct comparisons tricky. Nonetheless, the orders of magnitude were clear. The traditional threshing method required an average of 3.26 days/ha for wheat and used 101 hours/ha of labor. The mechanical method required 0.74 day/ha and used 67 hours/ha. For traditional threshing, bullocks worked an average of 37.7 hours/ ha of wheat threshed; on farms using mechanical threshing, the tractor and thresher were used 9.3 hours/ ha.
Surprisingly, the operational costs for both methods were similar: $ 15.04/ ha for traditional methods and $ 14.20/ ha for thresher and tractor. However, the mechanical method doubled the cost of power and equipment and halved the labor cost. Therefore, mechanical wheat threshing reduced the amount of labor needed by about 36%, and reduced labor earnings by about 55%. From the farmer's point of view, the savings in labor cost were transferred mainly to fuel and machinery maintenance expenses.

Tractor use and development of a custom service market The sample tractors were operated an average of 1, 120 hours during the 12-month period: 58% on land preparation and cultivation, 20% on wheat threshing, 18% on carting and hauling work, and 4% on other jobs.
The considerable variation in tractor use was due to variation in the operated farm area (Table 10), the tractor attachments owned, and the extent to which the owner engaged in custom services. Clearly tractor use on the owners' farm was positively correlated with farm size, whereas providing custom services was negatively correlated with farm size. The type of custom service and the amount of time given to this enterprise depended on the equipment and attachments owned. Only 18 of the tractors in the survey were not used for custom jobs.
The development of the custom market in recent years is the most important change in farm mechanization. We have seen the development of a new class of entrepreneur who has invested heavily in farm machinery and has actively sought business. We were unable to measure the extent of this business or how profitable it was, but it is quite extensive and growing very rapidly.
Tractor repair and maintenance
Eighty-eight tractor-owning farmers, 47 private workshops, 11I manufacturers of farm machinery, 2 tractor dealers, 3 cooperative farm service centers, 2 general automobile workshops, and 16 spare parts dealers were interviewed in an attempt to evaluate the tractor repair and maintenance facilities in Faisalabad District. The ability of tractor-owning farmers to properly operate and maintain farm machines was very poor. The many private workshops scattered throughout the district offered a wide range of services, but most lacked appropriate tools and equipment. Skilled manpower was scarce both on and off the farm and the supply of spare parts was inadequate and costly.
Breakdowns. Fifty-nine percent of the tractors owned by the sample farmers had major breakdowns and repairs during 1978 25% of these occurred during the wheat threshing season. The average downtime was 14 days (range of 5-30 days). The frequency of breakdowns was greatest for tractors between 2 and 5 years old. Breakdowns of the hydraulic system, fuel pump, dynamo, clutch, steering mechanism, and water body were most common and often caused by misuse or poor maintenance. The average repair cost was $257 (ranging from $129 to $766), of which spare parts were the largest component.
Table 10. Tractor use by farm size.
Operated farm areaa Av hours of Tractor hours (%7)
(ha) tractor operation/year owner's farm Hire services
More than 40 1431 83 17
20-39.5 1126 80 20
10-19.5 1019 48 52
less tha 10 1059 27 73
aln 1978.

Farmers recognized that unskilled operators, often themselves, were the main cause of breakdowns. However, they also placed some blame on manufacturing defects, substandard parts, and adulterated or wrong lubricants. Farmers had poor technical knowledge about tractors. Few understood or followed manufacturers' instructions on periodic replacement of oil and filters, few owned even simple tool kits or could carry out minor repairs, and few had formal training in tractor driving and operation.
Repair facilities. In 1978, there were 146 private workshops, 3 dealers' workshops, 3 cooperative service centers, I government workshop, and a number of farm machinery manufacturers and general automobile workshops in Faisalabad District. Spare parts were available from 76 shops. Special service technicians tire vulcanizers, radiator mechanics, battery specialists, dynamo and autoelectricians and diesel-testing facilities were generally clustered with tractor workshops on the outskirt at most rural towns and Faisalabad City. Half of the workshops and spare parts shops had opened since 1975.
Ninety percent of repairs were carried out at private workshops, 5% at tractor dealers' workshops, 4% by the farmers themselves, and 1% at the government workshop in Faisalabad. None of the sample farmers patronized the cooperative service workshops. Private workshop mechanics had varied skills and experience as well as facilities to work in. Workshops were visited early in the study, assessed, and divided into four categories on the basis of the tools and equipment (Table 11). Forty-seven having 52 owner-mechanics were studied in detail.
The level of formal education of owner-mechanics was low, but most had attended short mechanics training courses such as those run by Millat Tractors. Seventy-three percent of the workshop owners had less than 2 years experience as mechanics. In the typical workshop, the owner was the only trained mechanic, although eight also employed mechanics. All workshops employed or attached "trainees," boys between the ages of 8 and 16. Of the 140 trainees attached to the sample workshops, 36% received no payment and the rest received $24$15 /month, depending on age and experience. Trainees constituted a cheap, but largely unskilled, work force while they obtained informal training and experience in tractor and other machinery repair. Employed mechanics earned $1l0-$40/ month depending on experience. There was a fairly rapid turnover with employees predominantly moving to other workshops or opening their own. The sample workshops employed
Table 11. Type of workshop, investment in tools, and equipment, and number in category, Faisalabad, Pakistan, 1978.
Workshop category Tools, equipment a Value ()Workshop (no.) in Tools Equipment District Sample
A CT +SE 560 1600 25 8
B IT +SE 300 1600 25 8
C CT 370 31 10
D IT 180 65 21
Total 146 47
a T= Complete tool kit, SE = special equipment (e.g. lathe), IT = incomplete tool kit.

211 persons, comprising 52 owner mechanics, I I family mechanics, 8 paid mechanics, and 140 trainees.
Of an estimated 2,000 repair and maintenance jobs completed in 1978, 525 were by type A workshops, 367 by type B, 778 by type C, and 330 by type D. The more complicated jobs, such as engine overhauling, went mainly to type A and B workshops. The higher job rate of type C workshops was due to their more frequent location in rural areas and willingness to provide on-farm services. Few workshops specialized in particular tractor makes or types of repairs. Seasonal peaks for tractor repair were in March-June and October-December when tractors were heavily engaged in plowing, threshing and carting. Forty percent of the workshops reported that they received as much work as they could handle. Most that could have taken more work were located in urban and rural town areas. Half of the workshops reported their ability to do many repair obs was constrained by lack of appropriate tools and equipment. Many claimed to be willing to invest more if they could obtain credit on reasonable terms.
The three cooperative farm service centers in the district were very well equipped with tools, machine, workshop areas, and other facilities. These represented a total fixed investment of about $45,000. Charges were generally lower than at private workshops and some spare parts were stocked and sold at lower prices than at private shops. However, they had very little work. One center was not covering variable costs, the second turned in a small profit, and the third had been waiting about 4 years for electricity so work could start.
Spareparts supply. The widest range of tractor spare parts, particularly imported parts, was available from tractor and parts importers and assemblers in Lahore and tractor dealers. Faisalabad City tractor dealers and subdealers also maintained reasonable stocks, although they reported delays while parts were brought from Lahore. Most rural towns had spare parts shops, but these generally maintained very limited stocks of a narrow range of fast-moving parts. The three cooperative farm service centers maintained stocks of spares, but very few of the private workshops did.
Farmers and mechanics reported that parts were not always available even in Lahore and that they were often forced to substitute "inferior" locally made parts, and to adapt parts from other makes. Most farmers were willing to pay the higher costs for "genuine' parts. Prices for all parts were considerably higher in the rural town shops than in Faisalabad and Lahore.
Since 1970, the Pakistani Government has implemented a straightforward farm mechanization policy to increase the availability of farm power through the importation and manufacture of four-wheel tractors, and the subsidization of credit for tractor purchases.
In practice, the number, makes, and models of imported tractors have been controlled and have varied from year to year according to foreign exchange allocations and special credit or trade arrangements with suppliers and supplying countries. It appears that the demand for tractors, although partly conditioned by

government allocations of credit through the Agricultural Development Bank of Pakistan, exceeds supply and that the distribution process greatly favors farmers at the top end of the traditional power structure. The recent provision of a limited amount of subsidized credit to small farmers is too small and too late to alter the overall balance. However small farmers appear to provide an expanding market for used tractors, using funds obtained through foreign remittances and sellers' credit.
Although it is government policy to increase the efficiency of tractor use and the level of farm productivity by encouraging the adoption of cultivation attachments, that policy has not been implemented as vigorously as the simpler activity of importing tractors. The recent, rapidly increasing demand for and local fabrication of tractor-driven wheat threshers is the only significant change of interest in attachments since 1970. While subsidized credit has always been provided for tractor purchases, the ADBP only began making loans for thresher purchases in 1979.
By controlling imports, the government has effectively removed competition among tractor sellers, resulting in poor service, inadequate supplies of spare parts, and a lack of interest in training in tractor use and machinery maintenance facilities. Repair and maintenance services have developed rapidly in the private sector, but facilities generally are poorly equipped with tools and technical expertise. Likewise, most tractor owners have little technical knowledge of tractor use and maintenance. However, in the present situation of relatively small tractor numbers and limited annual imports, private gains from tractor ownership are high enough to more than compensate for high maintenance costs and inefficient tractor use.
Although there is evidence of high returns to farmers on investments in tractors and attachments, there is little evidence of appreciable social benefits. Tractors do not appear to have contributed to significant increases in farm productivity, either by bringing uncultivated land into production or by increasing the intensity of cultivation and crop yields. On the other hand, the opportunity costs of foreign exchange allocated for tractor imports and fuel, social hardship and reduced earnings of the rural landless, and the reinforcement of an inequitable traditional rural power structure indicate the very high cost of the present program.
Farm Mechanization Committee. 1969. Report on the farm mechanization survey, 1968.
Government of Pakistan, Planning and Development Department, Lahore.
Farm Mechanization Committee. 1970. Farm mechanization in West Pakistan. Government
of Pakistan, Ministry of Agriculture and Works, Islamabad.
McInerney, J. P., and G. F. Donaldson. 1975. The consequences of farm tractors in Pakistan.
World Bank Staff Working Paper 210.
Pakistan Government Planning Cominission. 1978. The fifth five year plan (1978-83).

K. W. Mikkelsen and N. N. Langarn
Data from a partially completed survey were used to investigate innovative activity and product-improving technological change in the Philippine agricultural machinery industry. Most firms had one or more persons engaged in inventing new products, improving products, and improving production methods, usually on an informal basis. Nearly all firms had made significant improvements.
In most discussions of the consequences of agricultural technology, private agents in agriculture and related industries are considered to have a very minor role in generating new technology. New technology is developed outside the agricultural sector and can either be accepted at some rate and to some degree, or rejected. This view may be appropriate for basic agricultural research which, because private agents lack either the incentive or capacity, must be undertaken by public and quasi-public institutions. In contrast, a new machine, in addition to being accepted or rejected, may be modified and adapted to local conditions. This paper explores innovative activity and product-improving technological changes among agricultural machinery manufacturers in the Philippines.
The data used for this preliminary report were obtained from interviews with 47 agricultural machinery manufacturing firms. The sample was drawn from two sources. First, the survey was to cover all firms which have actively participated in
Yale University and The International Rice Research Institute.

the Industrial Extension Program developed by the Agricultural Engineering Department of the International Rice Research Institute (IRRI). Actively participating firms were defined as those having received designs of one or more machines developed at I RRI and having produced at least one commercially. Of the 37 firms initially identified, the interviews revealed only 27 were active participants. Data from the 10 other firms were retained for added information. The second source of sample firms was a list of about 75 non-IRRI agricultural machinery manufacturers compiled from several directories. After stratifying by region, a systematic sample of 25 firms was drawn. Ten of these firms have been interviewed to date. Figure I shows the geographic dispersion of the sample.
0 Cooperators producing ,,
IRRI machines =
V Cooperators not producer
a Noncooperotors
1. Location of agricultural machinery manufacturers in the Philippines.

Ninety-one percent of the sample firms manufactured more than one product. The most common agricultural machines were rice threshers and two-wheel tractors (Table 1); others were rice dryers, rice and maize mills, and maize shellers. Many firms also produced furniture and jeep bodies, or performed metalworking services. Some firms retailed machinery usually made by foreign manufacturers. Firm size showed considerable diversity (Tables 2, 3). Employment ranged from several oneman shops to three firms each employing more than 100 workers. Despite their numerical dominance, small firms accounted for only 16.3% of total employment, whereas the 8 firms employing 50 or more provided 52.3%. More than half of the firms employed fewer than 20 workers. A similar picture emerged in sales. Twelve of the 40 firms reported sales of $20,000 or less in 1980; the median value was $37,467. Ten large firms, four of which were also in the largest employment group, reported sales of more than $500,000, and together they captured 87% of 1980 sales.
Table 1. Common agricultural machinery products.
Firms Av no. Av
Product producing of units price
(no.) produced ($)
IRRI portable thresher 18 71 696
IRRI axial-flow thresher 15 58 2,548
IRRI two-wheel tractor 15 102 754
Non-IRRI two-wheel tractor 7 167 1,247
Blower 6 50 67
Non-IRRI thresher 5 47 1,002
Table 2. Distribution of firms by level of employment.
Range Firms Employees Epoesfr
(no.) (no.) Epoesfr
1-9 employees 16 87 5.4
10-19 employees 10 126 10.6
20-49 employees 13 411 31.6
50 and above 8 685 85.6
Total 47 1,309 27.9
Table 3. Distribution of firms by sales.
Range Firms Sales (thousand $)
(thousand $) (no.) Total Average
Up to 20 12 76 6.4
21-40 12 380 31.6
41-100 6 396 65.9
101-500 6 1,936 322.6
501 and above 4 3,658 914.5
Total 40 6,446 (av) 16 1. 1

The most common measure of innovative activity is expenditure on research and development (R&D). Seventeen of the sample firms reported some R&D expenditure in 1980 (Table 4). R&D activities and firm size appeared to be positively correlated and no firms with annual sales of $20,000 or less reported R&D. However, these figures probably understate the amount of innovative activity, particularly among smaller firms where R&D is frequently not explicitly organized as a separate activity within the firm and may not be identified or reported. Measure of innovative activity which includes informal innovative effort, sometimes callled "blue-collar R&D," was obtained by asking each firm how many persons performed various technical functions within the firm, and what percentage of their time was spent on each function in 1980. Three of these functions, corresponding to the central purposes of formal R&D, were inventing new products, improving products, and improving production methods (Table 5). Almost 25% of the firms made no effort to invent new products, nearly all attempted to improve existing products or processes. Only 2 of the 45 responding firms reported no innovative activity. Summing the three activities, an average of about one man-year/ firm was devoted to technology improvement. Unlike the R&D measure, which indicated that less than half of the firms participated in innovative activity, the function-oriented measure showed that even the smallest firms participated. This appears consistent with the evidence on product improvement.
Table 4. Distribution of firms performing R&D, by firm sales.
Range Firms Firms (no.)
(thousand $) (no.) reporting R&D
Up to20 11 0
21-40 12 5
41-100 6 3
101-500 6 5
501 and above 4 3
Total 39 16
Table 5. Allocation of resources to various forms of innovative activity.
Firms (no.) engaged in given innovative activity Allocation Inventing Improving Improving production
(man-3rr) new products products methods
0 10 4 3
0.0-0.25 19 28 25
0.26-0.5 7 6 9
0.51-1.0 6 4 5
1.01-2.0 0 1 0
2.01 and more 2 2 2
Total 44 45 44
Average level .3.35 .34

To identify the important resources used in innovation, firms were asked to rate sources of technology and new ideas on a scale from 1 (not important at all) to 4 (extremely important). The average rating for customers (3.5) was the highest. Customers frequently identified problems in machine performance and offered suggestions for improvements based on personal experience with the products. In some cases, features requested by the buyer of a custom-built machine proved so successful that they were incorporated into the standard design. The second ranking source was technical personnel with the firm (3.3). In most cases, innovative activity was not carried out by specialized personnel. The owner or manager of the firm was frequently involved, as were production supervisors and foremen. Formal technical training among the 208 persons reportedly performing any of the innovative functions included 4 MIS degrees, 47 BS degrees, and some training past high school for an additional 48 persons. IRRI was ranked third as a source of new technology (3.0). In addition to supplying machine designs, IRRI responds to requests from cooperating firms with advice on production management and assistance in product improvements initiated by the firm. Less important sources of technology were suppliers, subcontractors, non-IRRI consultants, workers, trade journals, and patent documents.
The success of innovative efforts in generating new technology was investigated through information collected about product improvements. Each firm was asked to list all changes made in its four main agricultural machinery products since the beginning of 1976. When possible, this list was supplemented by, and checked against, an actual examination of the current product. Forty-two of 46 firms made at least one change (Table 6).
Changes can be classified as minor or major. A typical minor change was alteration of material specifications. Other minor changes included modifying the shape of various parts or the configuration of parts on the machine, and adding parts or features to enhance the machine functions. Some major changes required considerable redesign. One indication that a firm's product changes were not trivial was the number of patents held. Patents are only granted for changes meeting certain standards of novelty, usefulness, and significance. Sixty patents had been issued to 18 firms, and applications were pending for another eight. An additional 27 changes
Table 6. Distribution of firms by number of product changes.
Nube f hngsFirms Av no. of
Numer f hanes(no.) changes/firm
0 4 0.0
1-5 22 3.0
6-10 11 7.3
11-15 7 11.7
16-20 1 17.0
21 or more 1 23.0
Total 46 5.8

were judged by both the authors and the innovating firms to be patentable, although no application had been made. These patented and patentable changes accounted for 35% of the listed changes.
The extent to which these changes indicate activity among agricultural machinery firms is also influenced by the degree of overlap or duplication of changes between firms. Several features of the industry suggested that the changes made by one firm may not be very different from those made by others. First, it was apparehit from the survey that most firm managers were quite familiar with the competing designs sold in their region. Second, firms had little reservation about copying designs. Even though many firms held patents, few considered it worthwhile to prosecute for patent infringement. Third, firms felt a great deal of competitive pressure. Seventy percent indicated that their customers as well as potential customers would have little or no difficulty finding an alternative supply source if they stopped production. Under such conditions, rapid dissemination of profitable innovations would be anticipated.
To investigate their individuality, a list was compiled of the changes made in the most common products the IRRI-designed two-wheel tractor, portable thresher, and axial-flow thresher. Individual changes were grouped by feature, or component. For example, all changes on the threshing drum were grouped together, all changes on the blower formed a separate group. The results (Table 7) showed that, for each of the three products, most changes were made by a very few firms. This high degree of individuality actually understated the situation since firms changing the same component often made very different improvements. This individuality can probably be attributed to the segmented nature of the agricultural machinery market. Most firms sell to a local, not national or international, market. Localities differ in agricultural conditions such as soil type, water patterns, and in customer preferences. Consequently, a change that enhances product performance or desirability in one area would not necessarily be effective elsewhere.
As noted earlier, firms rely heavily on their customers as a source of new ideas. One point of interaction between manufacturers and users is direct sales. All but six of the sample firms made at least some sales directly to end users, and 68% sold 50% or more of their output by this method. The industry has responded to contacts with the local market by developing a variety of differentiated products appealing to
Table 7. Distribution of component changes by firms. a Component changes (no.)
Two-wheel Portable Axial-flow
tractor thresher thresher
Firms (no.) 11 14 13
Byl1firm 11 5 11
By 2firms 3 5 2
By 3firms 3 4 3
By 4or more firms 2 4 3
Total 19 18 19
aNumbers in parentheses indicate the number of firms that made changes in the given unit.

different needs and tastes, instead of highly homogeneous products. This allows farmers and other users to choose the machine with features most suited to their conditions.
Any mechanical technology introduced to the agricultural sector is likely to be altered. In a sample of Philippine agricultural machinery manufacturers, almost all firms engaged in some technology-improving activities. Most firms made improvements in existing machine designs, and different firms tended to make different improvements. Over time, such private efforts can help to increase the level of technology and to adapt it to diverse agricultural conditions.

S. Wattanutchariya
This paper provides information on the local farm machinery industry in Thailand and looks at the economics of farm machinery hire services. The local industry, developed during the past 2 decades, is now facing strong competition within the country and from abroad. The tractor contractor has provided access to farm machinery throughout the country. Custom services are profitable but because of increases in operating costs and competition, contractors'profits are declining and farmers are paying higher custom
In most developing countries, food policy has aimed at increasing production through introduction of modem technologies. These include the development of new crop varieties, improved cultural practices, irrigation systems, and farm mechanization. Although mechanization may be blamed for increasing unemployment, it is frequently justified because it permits faster and more efficient cultivation, particularly of heavier and more difficult soils. This may lead to increases in cropping intensity and employment.
In developing countries where labor is abundant, attempts to introduce higher levels of mechanization should be preceded by investigation of both social and economic repercussions. .Many studies on these questions have been conducted in Thailand. Two specific aspects, the development of the local farm machinery industry and the economics of providing farm machinery custom services, are investigated in this paper.
According to Kitdhakorn (1971), one of the pioneers of modern farming in Thailand, the first tractor was imported during World War I for experimental use in the Kasetsart University.

rice fields near Bangkok, but was too heavy for the soil conditions. In 1920, Kitdhakorn used a tractor for cultivating upland crops in Pracheub Kirikhan, about 200 km south of Bangkok.
After World War 11, heavy mechanical equipment was needed to open up new land for cultivation of upland crops. This led to the importation of farm tractors. In 1957, the Department of Customs recorded the import of 267 tractors, mostly from England. It was not until 1970 that imported farm tractors were recorded separately from other tractors. Imports peaked at 6,161 in 1977 (Table 1). The decline thereafter was mainly due to an increase in the number of locally produced or assembled tractors. The Thailand Office of Agricultural Economics (1980) reported that there were 230,591 two-wheel tractors, 31,158 small four-wheel tractor, 33,285 full-size tractors, and 8,000 motor rollers during the 1979-80 crop year, an average of 1 tractor/ 51.4 ha.
The Engineering Division, Department of Agriculture, Ministry of Agriculture and Cooperatives, has long been involved in developing and testing farm machinery adapted to the Thai environment. Examples of machinery developed are the wellknown Tebariddhi pump in 1955 and the "Iron Buffalo" in 1958. Several models of the "Iron Buffalo" were developed during the early 1960's and sold by a private company (Taenkam 1980). Due to the small scale of operation, relatively high costs, and competition from other local and imported machines, the company ceased operation in 1967.
In addition to government agencies, various private firms have undertaken research and development of farm machinery, in particular power tiller or two-wheel tractor. The design was originally simplified from a Japanese power tiller. In the 1980 survey, several firms accredited Singkru Utsahagamn at Prapradaeng as their source of design. That firm's first two-wheel tractor was a modification, to suit local conditions, of a Mitsubishi tiller imported in 1963. These tractors were first marketed in the area in 1965 and gained popularity due to the low price compared to imported machines, and their ability to perform well under local conditions. Because this tractor design is simple and requires no special manufacturing technology, many firms have started manufacture in Prapradaeng and other provinces.
Table 1. Farm tractors imported or locally produced during l97S-79'
Item 1975 1976 1977 1978 1979
Imported farm tractors 4,231 5,257 6,161 4,298 3,559
Small four-wheel tractors 2,582 2,914 4,568 5,631 4,920
(locally produced)
Two-wheel tractors 2,6 1764,2 2215,2
(locally produced) 2,6 1764,2 2215,2
&sources: Department of Customs; estimates by Thailand Office of Agricultural Economics, Ministry of Agriculture and Cooperatives.

Even though the demand for small, locally produced tractors is increasing, there is considerable competition among producers. Many large producers, who have access to credit and modern equipment, are more capable than smaller firms of producing at low cost. In addition, the marketing and managerial abilities of the owner are important for success. The survey, found that many firms, especially in the Prapradaeng area, have ceased doing business because of competition.
The development of an alternative to the "Iron Buffalo" started a few years after the local two-wheel tractor was widely accepted. A simple four-wheel tractor was developed by a group of firms in Ayutthaya by adding two more wheels so that the driver could operate from a seat instead of walking. The design was modified later to include gear and hydraulic systems similar to those of imported tractors. Development of small four-wheel tractors is still in progress, but the two-wheel tractor design remains unchanged.
In general, a firm will produce more than one type of farm machine so as to fully utilize its resouces and to meet seasonal market demands. For example, a firm may produce small tractors between January and June, before the planting season, and rice threshing machines and water pumps during the rest of the year. According to the survey, which covered most major firms in Bangkok metropolis and the nearby provinces, more than 33% of the firms produce only two-wheel tractors, 20% produce two-wheel and four-wheel tractors, 20% produce two-wheel tractors and rice threshers, and 9% produce all three machines (Table 2). For subsequent analysis, firms were divided into three types based on their main product as defined by value:
* Type I firms (23 of 35) produced mainly two-wheel tractors.
* Type 2 firms (7) produced mainly four-wheel tractors.
* Type 3 firms (5) produced mainly threshing machines.
The former activity of the manufacturer is in Table 3. Firm size in each type ranged from a few to more than 100 workers. Table 4 shows the number of firms classified by the number of workers and the operating period during the year. The survey showed that more than half of the firms produced throughout the year, but some reduced the number of workers in the off-season, maintaining sufficient skilled labor to produce tractors essential for stock.
Table 2. Surveyed firms classified by type of machinery.
Type of machinery No.
Two-wheel tractor 12 34
Four-wheel tractor 1 3
Maize thresher 5 14
Both two-wheel and 72
four-wheel tractors 2
Both two-wheel tractor and 7 20
rice thresher
Two-wheel, four-wheel tractors 3 9
and rice thresher
Total 35 100

Table 3. Frequency of former activities by manufacturers.
Manufacturers (no.)
Occupation Type 1 Type 2 Type 3 Tot
firm firm firm
Metal worker 6 4 3 13
Machinery repairman 2 2 3 7
Farm machinery dealer 3 1 4
Metal dealer 2 1 3
Dealer in other products 6 6
Farme 2 2
Others 9 2 3 14
Total 30 10 9 49
aType 1 firm produced mainly two-wheel tractors; type 2 f'rm, four-wheel tractors; and type 3 firm, threshing machines, bRice miller, automotive repairman, bicycle repairman, water pump manufacturer, etc.
Table 4. Firms classified by number of workers and operating periods.
Firmsa (no.)
Item Type 1 Type 2 Type 3 Total
Number of workers
Less than 20 11 2 1 14
20-39 5 1 3 9
40-100 6 1 1 8
More than 100 1 3 -- 4
Total 23 7 5 35
Operating period
Throughout the year 13 6 19
By order 6 1 1 8
Seasonal 4 4 8
Total 23 7 5 35
aType 1 firm produced mainly two-wheel tractors; type 2 firm, four-wheel tractors; and type 3 firm, threshing machines.
There has never been any patent law covering farm machinery in Thailand so firms are free to copy from abroad or from each other. Table 5 shows design sources reported by the sample firms. Although several firms reported designing their own machines many admitted copying from other sources, especially from Singkru Utsahagam. In fact, many of those who claimed to be designers probably copied a machine and made some minor changes. It was also found that many firms introduced changes based on customers' recommendations (Table 6). Design changes were most frequently made to improve performance and suit a customer's needs. Some firms reported changes in design to eliminate unnecessary costs.
The main tractor components produced by the firms are body parts such as wheels, engine housing, chassis, and handle. Other parts, such as engine, chain, and bearing,

Table 5. Frequency of sources of design for machines.
Frequency (no.))
Sources of design Two-wheel tractor
Two-wheel tractor Four-wheel tractor Rice thresher Maize thresher 0
with clutch
Any brand in the market 7 1 1 1 >
Another brand in the same location 2 2 2 Z
Another brand in different location 15 2 3 2
Foreign design 1 2 3 2
Own design 1 3 6 5 o
IRRI design 1 4
Others' 3 2 1
Total 30 8 15 13 6
aDivision of Agricultural Engineering, mechanic previously employed by another firm, etc. W

Table 6. Source of changes in design and reason for changing.
Firmsa (no,)
Item Type 1 Type 2 Type 3 Total
Sources of changes
Recommended by customer 19 3 5 27
Owner 6 1 1 8
Followed other firms 6 6
Copied from imported machine 1 1 2
Own mechanic 1 2 3
Others 2 1 3
Total 34 6 9 49
Reasons for changing
Reduced cost 5 2 7
Improved performance 20 2 5 27
Suited customer's need 8 2 3 13
Better looks 4 4 8
Others 2 1 3
Total 39 11 8 58
aType 1 firm produced mainly two-wheel tractors; type 2 firm, four-wheel tractors; and type 3 firm, threshing machines.
are purchased. Machines used can be very simple. A small firm may have a welder, lathe, sprayer, and acetylene cutter, while a large firm may employ more laborsaving machines such as hydraulic press, shapers, steel roller, honing machine, electric power drive, and sprocket wheel cutter (Pinthong 1974).
Estimated costs of production of two-wheel and four-wheel tractors are shown in Table 7. Excluding the diesel engine, steel, which is the main input for two-wheel tractors, accounts for 50% of the cost. The cost of an 8-hp diesel engine is around $650. Total cost of two-wheel tractor with engine is $849 compared to $1,500 for an imported two-wheel tractor. However, imported tractors are better built, fighter, and perform more functions. A small four-wheel tractor, including the 15-hp diesel engine, costs around $2,150. The sale price of the imported small four-wheel tractor is around $3,650. In general, there is more variation in the cost and design of four-wheel tractors than of two-wheel tractors.
Marketing of local farm machinery is very important. Because product differentiation among two-wheel tractors is low, firms have to use special techniques to gain popularity. Machines manufactured outside the Bangkok Metropolitan Area were produced predominantly for the local market and sold directly rather than through dealers. The large firms in Bangkok and one firm in Prathum Thani sold for cash or on credit through dealers throughout the country. Dealers were crucial to the firms because they also provided useful customer feedback on potential design changes. Products from large firms were usually cheaper because of economies of scale in mass production. Smaller firms maintained their market share by having close contact with the customers.

Table 7. Estimated cost of production of two-wheel and four-wheel tractors.a
Item Two-wheel tractor Four-wheel tractor
Cost ($) % Cost s) %
Steel plate 66 34 167 15
Steel rod 29 15 47 4
Cast iron 24 2
Cast aluminum 123 11
Lubricated oil 15 7 10 1
Bearing 16 8 61 5
Chains 17 9 37 3
Hydraulic pump 126 11
Other parts 24 12 233 21
Wage 11 6 87 8
Electricity and fuel 2 1 52 5
Depreciation 1 1 52 5
Other expenses 7 3 70 6
Production tax 7 3 39 3
Tax on machinery 24 0
Subtotal (without engine) 195 100 1132 100
Diesel engine 654 1037
(8 hp and 15 hp)
Total cost 849 2169
a Source: estimates by Industrial Service Institute 1981.
At the time of the study, there was no serious competition among four-wheel tractor manufacturers because there was excess demand and each firm was producing close to its full capacity. Most four-wheel tractors were sold direct to farmers growing broadcast rice in the Central Plains because such tractors were too heavy to perform well in wet paddy fields. Farmers who transplant paddy prefer two-wheel tractors.
The foreign market is the next logical step for large manufacturers. Some have sent trial shipments to Indonesia, but improvements are necessary before they are accepted.
Two government agencies are involved in the extension of farm machinery (Mongkoltanatas 198 1). The Industrial Service Institute (151), Ministry of Industry, assists manufacturers by providing technical training and advisory services to owners, managers, and workers. ISI also organized the Forum of Tractor Manufacturers of Thailand in 1977. The objectives of the forum are to support local farm machinery manufacturers and to improve production techniques. At present the 40 members are mostly in the central region. The second agency the Agricultural Engineering Division (AED) of the Ministry of Agriculture and Cooperatives assists manufacturing firms in testing and modifying new products. The survey showed 40% of the firms received government assistance in the form of investment credit, machine design, tax exemption for equipment and management, and technical training.

Major problems facing the local farm machinery industry and solutions suggested by producers were as follows:
I Financial. Forty percent of the respondents, especially smaller firms, reported
financial problems, in particular, the difficulty of obtaining low'interest loans from the government. It was suggested that the government should encourage small firms by providing sufficient funds to help reduce production costs and
lower the price of machinery to farmers.
2. Technological difficulties. Although there are no patent laws and firms could
copy designs, the unavailability of suitable technology was considered an
important problem.
3. Lack of skilled labor. Most workers gained experience through on-the-job
training. Many skilled workers had been attracted to the Middle East and
training was needed for young unskilled workers.
4. Marketing. The demand for machinery depends on crop yield and price.
Because these are variable, manufacturers must be ready to respond to changes.
There were also minor problems, such as lack of proper equipment and unfair taxation (Industrial Service Institute 1981). For example, imported materials are taxed higher than imported farm machinery (13.55% and 11.2790' tax for local two-wheel and four-wheel tractors, and 8.85% tax for imported machines).
In conclusion, the farm machinery industry in Thailand has been developed by private rather than government effort, but requires government assistance and protection for its continuation. If the goal of the government is to induce mechanization, it should support local farm machinery manufacture because these machines are inexpensive, simple, and designed to fit local conditions. Standardization of farm machinery for export markets, may be necessary but this will raise production cost to some extent. Taxation of imported machines and materials should also be revised if the government wants to generate employment and reduce the trade deficit.
Tractor contractors have played an important role in farm mechanization. Farms average 4 ha and most farmers cannot afford to own tractors. Contractor services, using full-size tractors, are mainly for cultivation of dryland for upland crops and broadcast rice. Contract services are also available for transporting and threshing, while farmers use their own bullocks for planting and weeding.
Contractor characteristics and services
Some tractor owners offering contract services travel more than 100 km for plowing during the off-season. A contractor survey was carried out during April and May 1980 in Phra Buddhabaht, Pak Chong, and Bang Pa In. The first two are major upland crop growing areas and the third is the broadcast rice area where most contractors are located. Thirty-four contractors, owning 48 tractors, were contacted. Of the 48 tractors, 22 were purchased new. Seventy-six percent of the contractors were rice or maize growers, 15% were upland crop middlemen, and the rest were

general merchants. The average landholding was 14 ha. Two contractors operated only contract services.
Most contractors serviced neighboring areas after finishing their own work. Contracts were usually made directly with farmers and the custom rate was determined by soil conditions and the contract market situation. Some were made through commission agents, who looked for customers in nearby areas. The number of tractor owners is increasing and some contractors travel in groups to other provinces to find new customers.
Fuel is the most important factor in the contracting business. Because only a part of this increase can be passed on to the farmers, rising prices and shortages that increase operating cost discourage contractors from traveling long distances for customers.
Work and revenue from owning a tractor Tractors plowed an average of 176 ha in 1979,29 ha (17%7) of which was owned land and 147 ha (83%o) was contract (Table 8). The work averaged 432 hours. Contractors in Bang Pa In utilized their tractors more than those in the two other areas; however, the total revenue per tractor was less because of lower custom rates. The 1979 custom rate for plowing was $22/ ha for upland and $11 /ha for paddy land. That year, the contractors received an average total revenue of $3,077, 88% from plowing and the rest from activities such as threshing and transportation of farm produce. In Phra Buddhabaht, 96%/ of the revenue was from plowing.
Cost of operation and profitability
The cost of operation can be divided into cash and noncash costs. Cash costs, which refer to actual outlays such as fuel, oil, driver, maintenance, etc. accounted for 75% of total cost (Table 9). The cost of maintenance and repairs was the largest (4 1%7)
Table 8. Average amount of work and revenue per tractor in 1979.
Item Phra Buddhabaht Pak Chong Bang Pa In Average
Area plowed (ha)
Own land 25 25 40 29
Service 167 93 196 147
Total 192 117 237 176
Length of plowing (h)
Owniland 47 85 61 67
Service 352 319 446 365
Total 399 404 506 432
Revenue from plowing($
Own land 347 575 466 465
Service 2672 1933 2059 2229
Total 3019 2508 2525 2694
Revenue from other activities()
Own land 61 111 21S 121
Service 45 627 40 262
Total 106 738 255 383
Total revenue($
Own land 408 685 681 586
Service 2718 2560 2099 2491
Total 3126 3245 2780 3077

Table 9. Cost structure and profitability of tractors in 1979.
Item Costs and returns Whol
Phra Buddhabaht Pak Chong Bang Pa In Whople Non-cash costs
Depreciation 120 3130 34 176
Opportunity cost of4746 5348
investment (8%f4746 5348
Opportunity cost of owner 29 s0 53 43
Subtotal 620 843 617 703
Cash costs
Repair and maintenance 1264 1242 673 1096
Fuel 988 692 593 782
Oil 62 143 81 94
Driver 109 139 168 136
Others 14 13 76 31
Subtotal 2437 2229 1591 2139
Total cost 3057 3072 2208 2842
Total revenue 3125 3245 2781 3077
Net return over cash cost 688 1016 1191 938
Net profit 68 173 573 236
Return from investment (%) 8.3 12.7 14.5 11.5
a Prevalig interest rate on bank savings accounts.
expense because most tractors had been operating for more than 5 years in poor soil conditions and required regular repairs. The cost of fuel accounted for nearly 28% of the total cost. As mentioned earlier, plowing land for broadcast rice is easier than for upland crops. So, tractor operation cost was much lower in Bang Pa In than in the two other areas, and therefore even though the revenue within the year was lower, the contractors made more profit.
Break-even point analysis
Annual tractor use required for economic viability can be evaluated by break-even point analysis (Fordson 1959). Because the tractors surveyed were used for a wide range of activities, each with varying contract rates, no common unit could be defined for measuring output. The analysis was therefore limited to use for plowing and fixed costs were determined in proportion to total use. Break-even points were estimated for each region and for the entire sample.
The results (Table 10) show the whole sample break-even point as 116 ha, while the actual average area plowed was 176 ha. For each region, the break-even area plowed was also less than the actual. It is reasonable to conclude that contractors are, on the average, making a profit from their operations.
Table 10. Break-even point of using tractor for land preparation in 1979.
LoainTotal fixed Average Average Break-even Actual
Loaincost ($) variable cost revenue point (ha)
($/ha) ($/ha) (ha)
Pbra Buddhabaht 509 12.20 15.74 146 192
Pak Chong 593 15.30 21.35 98 117
Bang Pa In 499 6.65 10.65 131 237
Whole sample 538 11.04 15.30 116 176

The tractor contractor has helped increase agricultural productivity by expanding new cultivated area and speeding up cultivation. This is beneficial to tractor owners and users because few owners can economically justify tractor ownership without contract operations. At present, tractor owners are making profits. However, with increasing competition and rising costs of operation, investment in tractors may not be worthwhile in the future unless sufficient demand for contractor services can be assured. Cost increases for fuel and spare parts have increased custom rates, and therefore costs of production. It is expected that there will be some substitution of buffalo for machines. Unless ways can be found to either increase yields or reduce other costs, there will be pressure on the government to increase crop prices.
Fordson, J. C. 1959. Break-even points for harvesting machines. University of Georgia,
Georgia. 91 p.
Industrial Service Institute. 1981. Small tractor industry [in Thai]. Ministry of Industry,
Bangkok, Thailand. 12 p.
Kitdhakorn, M. C. S. 1971. The use of farm machinery in Thailand [in Thai]. Pages 132-153 in
Society of Social Science, ed. A report by and about M. C. Sithiporn Kitdhakorn.
Sivaporn Press, Bangkok, Thailand.
Mongkoltanatas, J. 1981. Survey of agricultural machinery manufacturers. Agricultural
Engineering Division, Ministry of Agriculture and Cooperative, Bangkok, Thailand.
9 p.
Pinthong, J. 1974. Economics of small tractor production in Thailand. MS, Faculty of
Economics, Thammasat University, Bangkok, Thailand. 76 p.
Taenkam, P. 1980. The small tractor industry. MS, Faculty of Economics, Thammasat
University, Bangkok, Thailand. 211 p.
Thailand Office of Agricultural Economics, Ministry of Agriculture. 1980. Small tractor
industry in Thailand, Bangkok, Thailand. 25 p.

Paitoon Wiboonchutikula
The rate of total factor productivity growth (TFPG) of the agricultural machinery industry in Thailand from the initial year of production to 1979 was estimated using data from two sources: 1) survey and interview of large and medium-sized firms, and 2) aggregate census data. Estimation from the first source showed the TFPG rates of all firms increased until 1976, then declined.
Estimation from the second source showed the rate was small compared with that in developed countries and other manufacturing industries in Thailand.
One immediate impact of farm mechanization in Thailand has been the growth of the tractor manufacturing industry. The number of firms grew from a few in the mid-1960s to more than 100 in 1980, employing over 2,000 workers. Production grew about 16% annually. Growth came from two sources: the increase in resources used and an increase in the productivity of these resources which can be measured as total factor productivity growth (TFPG). This paper measures and analyzes TFPG of the agricultural machinery industry in Thailand. Basically, TFPG can be measured by subtracting the average of the rates of growth of real inputs, properly weighted, from the rate of growth of real output. For developed countries such as the USA (Solow 1957, Christensen and Jorgenson 1970, Kendrick 1973, and Denison 1974) and Japan (Jorgenson and Ezaki 1973, and Nishimizu and Hulton 1978), measured TFPG accounts for more than 50% of the rate of growth of real output of all manufacturing industries. Due to data limitations, there are few TFPG studies for developing countries.
Small farm tractors were introduced from Japan in the mid-1950s. In the mid- 1960s, local firms modified and started to manufacture the Japanese designs. They were Department of Economics, University of Minnesota.

readily accepted because they were more suited to local soil conditions, easier to operate and repair, and less expensive than imported tractors. Since the late 1960s, with increases in farm incomes, higher food prices, and increases in agricultural wages, the industry has expanded rapidly. Nowadays many farmers own tractors and about 90% of total sales are produced domestically.
Tractor production in Thailand is labor-intensive and the market is free from government intervention. Other than the common business taxes, which are applied to all industries, subsidies and international trade protection are negligible. Local production is also competitive. Firms are numerous, products are similar, and entry is relatively easy. With some experience in machine shops and an initial investment of about $4,000 on machinery and equipment, production can start.
Both two-wheel power tillers and small four-wheel tractors are produced in Thailand. The two-wheel power tillers are equipped with 5- to 9-hp imported diesel engines 'and the four-wheel tractors with 10 hp and larger engines. About 85% of tractors used by Thai farmers are power tillers. In 1980, the survey year, there were about 99 tractor-producing firms, 79 producing two-wheel tractors only, 2 producing four-wheel tractors only, and 18 producing both. Eighty percent of the firms were located in the central region, and more than 70% of these were in the greater metropolitan areas, which provide both a better infrastructure and better access to raw materials.
The firms were placed into small, medium, and large categories. Most of the 62 small firms, producing less than 300 units!/year and employing 2-10 workers, were located near farming areas. Their principal activity was assembling parts and components bought from others. Twenty-three medium firms produced 300-1,000 units annually and employed about 10-50 workers. They usually produced other farm machines, such as threshers and irrigation pipes in addition to tractors. Fourteen large firms produced more than 1,000 units annually and employed 50-200 workers. They were normally more vertically integrated, producing many parts and components themselves.
Most of the medium and large firms were located in the central region. Although the number of medium and large firms was increasing, the number of small .ones was decreasing. All the firms were owned by Thais.
Measurement of TFPG is based on an unrestricted, linear homogeneous, smooth, aggregate production function which allows substitution possibilities between all inputs. The inputs are categorized as labor, capital, and raw materials. The i-th firm's production at time Imay be defined as:
Q (t) =fi fL (t, K, (t), M (), tI
Q, (t) is real output of the i-th firm at time t,
Li (t) is labor employed by the i-tb firm at time t,
K, (t) is physical capital used by the i-th firm at time t,
M, (i) is real intermediate inputs used by the i-th firm at time t,
t is a "shift" variable subject to time,

and the subscripts
i= 1,2,..., n and t= 1,2,..., T.
Totally differentiating equation I with respect to time we obtain:
Ki(t) -(t)+f K(')
- ,(t L ,()*i (t) f
ON(t) Q(t) Li(t) Qj(t) Ki(t)
+ M,() ,(t) 1l,(t + L (2)
Qi(o mio QONt
where (') denotes the rate of change of the variables over time.
Equation 2 shows that the rate of growth of real output can be decomposed into the weighted averages of the rates of growth of labor input, physical capital, and raw materials, where the weights are the output elasticities of the corresponding factors of production and the rate of growth of the shift variable.
Under competitive equilibrium, where each input is paid according to the value of its marginal product, equation 2 can be written as:
&~it) wi(t) Li(t) L.i(t) r~)X(t) k w~t pim(t) miot A (t) q ., (t.)3
-+ (3)
Q,(t) P,(t) Q,(t) Li(t) P,(t) Q,(t) Ki(t) Pi(t) Q,(t) M1(t) Q10t)
wi(t) is the nominal wage rate of the i-th firm at time t,
ri(t) is the nominal rental rate of capital of the i-th firm at time t,
Pp'(t) is the price of raw materials of the i-th firm at time t,
Pi(t) is the price of products of the i-th firm at time t.
By letting a and/3 represent labor and raw material shares of the value of the i-th firm's total production, respectively, Euler's theorem, which assumes a linear homogeneous production function, implies that the capital share is (l-a-/3), and equation 3 can be rewritten as:
&it) = ,(I) Li(O + -llaft)-# k(t) + 00) X) i+O -F I (4)
QONt L,(t) 0 Ki(t) M1t) O(t)
Notice that, given these assumptions, the TFPG of the i-th firm can be measured from the following differential equation, or accounting identity, without estimating directly the production function or the output elasticities.
fota) i(&) () o ", ) ki(t) + li(t) (5)
ON [ Li(t) t I K)(t) II
Estimation of TFPG of tractor firms from equation 5 requires measures of real output, real inputs, and shares of all factors of production for successive years. This is possible given the availability of annual data on units of production, values of intermediate inputs, numbers of workers, capital stock, and factor shares. Data were obtained from surveys and interviews of firms in the peak producing season, February to April 1980. Only medium and large firms were included in the surveys because small firms did not maintain records on inputs and output. Also, it would have been expensive to survey the small firms they were too numerous and scattered throughout the country, while accounting for less than 15% of total production.

The initial year of production, yearly production, raw materials used, labor employed, and investment in structures, machinery, and equipment were recorded in the interviews. Most medium firms started production in the mid1960s, half of the large firms started in the early 1970s. The latest year of records was 1979. Details of the survey and the methods of estimating all variables from these data are described below.
Physical output
The physical output of single-product firms was the number of tractors produced each year. For firms producing more than one product, the physical output was estimated by aggregating all products weighted by relative prices. The weighting procedure was as follows. Let
Q~i(t) be the units of the j-th product produced by the i-th firm at time t, and
Pi,(t) be the price of the j-th product of the i-th firm at time t,
where i = 1, 2,... n,j = 1, 2 ..... ,.. m and t = 1, 2,... ,T. The aggregate
physical output of the i-th firm at time t, Qi(t), in terms of the 1-th product can be written as:
m QYt) (6)
it) = Pi t)
The continuous growth rate of physical output of time t, qi(t), is found by subtracting the logarithm of the output of the previous period from the logarithm of the output of the current output. That is:
Q,(t) = In Q,() In Q,(t-1). (7)
Real intermediate inputs
Intermediate inputs comprised raw materials, electricity, and fuel. The raw materials included sheet steel, steel rods, angle steel, chains, bearings, seals, gears, pulleys, steering sets, wheel rims, bolts, and nuts.
More than half of these were produced domestically. If the number of units of each of the raw materials used each year were available, they could have been aggregated using relative prices as weights in the same way as for physical output. Unfortunately, such data were available for only a few large firms and then only for 1979. No data were available for the medium firms; however, all firms could provide the total value of raw materials used each year. We therefore estimated the values of real intermediate inputs by disaggregating the total values in proportion to the value shares of the major purchased inputs. The major purchased inputs were defined as those accounting for at least 1% of total purchased inputs, subject to at least 90% of the value of purchased inputs. The value shares were computed from the 1975 input-ouput table, which is presently the only one available (NESDB 1980). Because the interviews indicated no major changes in purchased input coefficients and product composition over time, the 1975 value shares were applied to all years.

Real capital stock
Capital stock was classified into 1) buildings and structures, and 2) machinery, equipment, and vehicles. For each type in any year, real capital stock was obtained by adding current investment at constant prices to the real capital stock of previous year minus real depreciations accumulated up to that year.
That is, if we let
Kik., be the real capital 5tock of the k-th type for the i-th firm at time t, L*., be the investment flow of the k-th type of capital for the i-th firm at time t,
evaluated at constant prices, and
6ik be the rate of replacement of the k-th type of capital stock by the i-th firm, where i 1,,2,... ,n,k= 1,2, and t= 0, 1,2,.... ,T, then
Kk,= lik., + ( l-6k)Kk.,1-. (8)
By a process of iterative substitution of Kik,,.I
/<,., = (1-liKik.o + (1-64 34, (9)
where Kk.o is the initial capital stock of the k-th type for the l-th firm.
For estimating real capital stock from equation 9, we need an estimate of initial real capital stock, flows of real investment, and rates of replacement. The initial real capital stock and annual real investment were obtained by deflating investment data from the surveys. The rate of replacement was defined as the reciprocal of the economic life of capital. The methods of measuring investment deflators and economic life of capital follow.
Investment deflators
Investment deflators were obtained from the Ministry of Commerce. The wholesale price indexes of construction materials were used for buildings and structures. The weighted averages of the indexes of machinery and equipment and transportation equipment were used for machinery, equipment, and vehicles. The weights were the value shares of total investment and amounted to about two-thirds for machinery and equipment and one-third for transportation.
Economic lives of assets
Buildings and structures are more durable than machinery and equipment. Krueger and Turner (1980) provide some estimates for Turkey. Assuming that structures, machinery, and equipment of Thai manufacturers have a life which is not significantly different from the Turkish, we have used an average structure life of 33 years and machinery and equipment life of 15 years.
Structure life was assumed to be invariant across industries, -whereas machinery and equipment lives were made specific to each industry, based on estimates for U.S. industries (Park 1973). Because the average fife of machinery and equipment in the U.S. is longer than in Thailand, Park's estimates were scaled down so that the weighted average for the entire manufacturing sector was 15 years.
Labor input
The labor input of a firm is the number of workers employed by that firm during the year. Initially, we intended to take account of quality differences. In the surveys, we

asked firms to categorize the number of workers by sex, age, and education. After analyzing the data, we decided to use the total number of workers because there were no significant differences in labor quality among firms. Only one firm employed some female workers. Workers were from 15 to 50 years old. None had any formal education beyond high school and their skills were learned from working in machine shops. The proportions of workers by age and education did not vary much across firms.
Factor shares
The share of intermediate inputs was obtained for each year by dividing the value of intermediate inputs by the value of production. Labor shares were obtained likewise, where the value of labor was computed by multiplying the number of workers of each type by the corresponding wage rate, including adjustments for other benefits such as meals and lodging. The capital share was simply the residual.
Of the 30 firms interviewed only 14 (6 large and 8 medium) provided sufficiently complete and reliable data for estimating TFPG. Data were checked for reliability by comparing with reports of other financial and government institutions. Of the 14 firms, 6 were in the greater metropolitan area; the rest were in other major rice growing provinces in the central region: Ayuddhaya, Chachoengsao, and Nakomswan. Four of the 6 large firms were the major four-wheel tractor producers in the country, accounting for more than 85% of domestic production. The rest of the large firms and all of the medium firms produced 75% of the two-wheel power tillers and other farm machinery. Among the large firms, four started production after 1970 and two in the late 1 960s. Among the medium firms, six started production after the mid-1960s and two in the mid-1970s.
Table I shows the average growth rates of output, input, and total factor productivity of the surveyed firms for three periods (initial year of production to 1970, 1970-76, and 1976-79) and also from the initial year of production to 1979. From the initial year to 1979, TFPG accounted for about 19% of the rate of growth of real output while 8 1% was from input growth. Output and employment grew faster in the first period when two-wheel tillers were newly introduced and in the second period when four-wheel tractors became more popular. During the second period, large
Table 1. Output, input, and productivity growth of all farm machinery firms.
Growth rate per annum (%)
Period Oupt. Lbr Cptl Total factor'
Outpu Labr Caital productivity
Initial to 1970 34.94 23.21 9.67 5.91
1970-76 38.02 20.56 23.90 8.29
1976-79 13.23 6.18 11.69 2.10
Initial to 1979 28.02 13.81 18.96 5.33

rises in farm income, food prices, and wages occurred. Capital input increased much faster in 1970-76 due to the initial investment of some new large firms. TFPG from the initial year to 1976 was about 7%, accounting for about 10% of growth in real output. For the most recent period, TFPG was about 2%.
The growth rates of output, input, and total factor productivity of the large and the medium firms from the initial year to 1979; and for the three periods; are in Table 2. For the initial year to 1970, the sample consisted of two large and six medium firms. Output, labor, and capital of the large firms grew much faster than those of the medium ones, but TFPG was similar. During 1970-76, output and the TFPG of the medium firms increased substantially. However, the growth rate of employment for the large firms slowed down, but capital growth increased because of the entry of four new large firms. The TFPG of the large firms in this period was almost the same as in the previous one. During 1976-79, all firms showed a slowdown in the growth of output, inputs, and total factor productivity. However, the growth rate of capital of the medium firms was influenced by the entry of two new firms. For all periods, TFPG was higher for medium than large firms.
TFPG for all manufacturing industries, including machinery and equipment, is in Table 3. The industries were disaggregated in accordance with the three-digit International Standard Industrial Classification (ISIC) of manufacturing industries and the estimates of TFPG derived from secondary data for 1963-76. Details of the data sources and estimation procedures are given in Wiboonchutikula (1982). From Table 3, the TFPG of the machinery and equipment industry, which includes the manufacturers of all agricultural machines and equipment as well as other wood or metal working machines was 1.36. This is much lower than the TFPG estimated for the sample large and medium agricultural machinery firms during the same period. There are several possible reasons for the difference:
* The assumptions necessary for estimating TFPG may not be valid. It would
therefore be unreasonable to expect consistent results. This could be investigated by relaxing some of the assumptions and reestimating TFPG.
* ISIC industry 382 includes a wider range of firms than medium and large-scale
manufacturers of agricultural machinery. TFPG of these other firms may have
been even more negative than that of the aggregate.
" The data were from firms still in business, and these we would expect to be the
more successful. It is therefore likely that TFPG of the sample may have
overestimated that of the population.
Table 2. Output, input, and productivity growth of large and medium farm machinery firms.
Growth rate per annum (%)
Period Oupt LbrCptl Total factor
Outpt LborCaptal productiit
Large Medium Large Medium Large Medium Large Medium Initial to 1970 46.05 33.6 53.65 18.14 14.08 8.94 5.72 5.94 1970-76 40.05 36.86 20.73 20.46 36.40 16.76 5.74 9.63
1976-79 10.73 14.80 7.75 5.20 11.05 12.09 0.63 3.02
Initial to 1979 25.56 29.55 16.48 12.15 22.05 17.03 3.71 6.35

Table 3. The TFPG of 3-digit ISIla manufacturing industries, 1963-76.
ISCindustry Material Labor Growth rate of TFPG
ICshare share Output Labor Capital
311 Food processing .7137 .0511 12.16 9.41 8.41 1.25
312 Food products .7984 .1022 25.26 21.25 20.91 -1.30
313 Beverages .3329 .0690 16.02 12.20 12.50 1.38
314 Tobacco .4318 .0520 6.82 3.04 3.33 1.44
321 Textiles .5746 .1050 15.42 9.93 16.53 1.78
322 Wearing apparel .5990 .1682 16.30 7.74 17.49 -0.37
323 Leather & leather products .7210 .0717 16.30 14.16 13.72 1.98 324 Shoes .6878 .1880 16.30 12.61 14.35 -1.06
331 Wood and cork .6717 .1264 5.70 4.17 5.97 0.15
332 Furniture & Fixtures .7015 .1013 16.30 11.85 16.95 -1.41
341 Paper & paper products .6765 .1012 16.76 11.03 9.99 2.98 342 Printing & publishing .5595 .1493 16.30 9.92 13.26 3.00
351 Basic chemicals .5661 .0669 20.81 22.01 13.83 1.57
35 2 Chemical products .6316 .0901 11.45 7.14 8.49 1.16
353 Rubber & rubber products .6944 .0530 25.16 20.20 27.15 1.05 361 Nonmetallic minerals .6241 .2658 29.00 25.07 28.58 -1.99
362 Glass &glass products .4816 .1312 14.55 13.19 18.66 -2.99
369 Other nonmetallic .7086 .0871 11.46 5.29 11.98 -1.36
371 Ferrous metals .8308 .0609 17.37 14.76 17.33 0.04
372 Nonferrous metals .5775 .0431 26.68 12.94 14.88 3.18
381 Metal products .7059 .0655 1.29 0.25 5.49 -0.39
382 Machinery and equipment .6431 .0589 16.30 9.22 17.93 -1.36 383 Electrical machinery .6911 .0809 53.15 45.89 45.87 3.22
384 Transport equipment .8151 .0618 20.18 14.28 18.68 -2.09
390 Miscellaneous .6341 .0991 16.30 10.30 16.21 -0.32
Average .6429 .0980 17.73 13.11 15.94 0.38
a International Standard Industrial Classification.
Comparison of the aggregate estimate of TFPG for the machinery and equipment industry shows that it is well below the overall average for manufacturing industries; in fact, only four of the other industries had lower TFPG. This could be because the industry is relatively young and still developing. As yet, most agricultural machinery firms have been operated by workers without any formal training and who work with rather old or secondhand machines in plants with poor layouts. Training programs to upgrade the skills of workers, technical assistance at the plant level, and incentives to increase investments may help improve the total factor productivity of the industry.

Christensen, L. R., and D. W. Jorgenson. 1970. U. S. real output and real factor input,
1927-1967. Rev. Income and Wealth 16:19-50.
Denison, E. F. 1974. Accounting for United States economic growth 1929-1969. The Brookings Institution, Washington, D.C.
Jorgenson, D. W., and M. Ezaki. 1973. The measurement ofmacroeconomic performance in
Japan, 1951-1968. In K. Ohkawa and Y. Hayami, eds. Economic growth: the Japanese
experience since the Meiji era, Vol. 1, No. 19, Tokyo.
Kendrick, J. 1973. Postwar productivity trends in the United States, 1948-1969. National
Bureau of Economic Research, New York.
Krueger, A. 0., and B. Tuncer. 1980. Total factor productivity growth in Turkish manufacturing. University of Minnesota. (mimeo.)
NESDB (National Economic and Social Development Board), Institute of Developing
Economies, and National Statistical Office of Thailand. 1980. Basic input-output table
of Thailand, 1975, Bangkok.
Nishimizu, M., and C. R. Hulton. 1978. The sources of Japanese economic growth: 19551971. Rev. Econ. Stat. 351-361.
Park, W. R. 1973. Cost engineering analysis. John Wiley and Sons, New York. Solow, R. M. 1957. Technical change and the aggregate production function. Rev. Econ.
Stat. 312-320.
Thailand Ministry of Commerce, Price Index Division. Various years. Wholesale price
indexes in Thailand. Bangkok.
Wiboonchutikula, P. 1982. The measurement and analysis of the total factor productivity
growth of the manufacturing industries in Thailand. Ph D dissertation, University of

S. Sukharomana
Farm mechanization, especially expansion of labor-displacing technologies for rice production, reduces the use of domestic resources and increases the demand for imported inputs such as fuel, oil, engines, and spare parts. The effects of farm mechanization on the domestic resource cost of earning a net unit of foreign exchange from rice production are less than the effects of yield and, the opportunity cost of land. Mechanization has a tendency to generate profits for society even though the demand for imports of~
machinery-related items is increased.
There are two kinds of rice production technology land augmenting and labor displacing. The first increases output without expansion of the area of farmland; the second reduces the requirement of labor per unit of farmland. Examples of land augmenting technologies are improved seeds, chemical fertilizers, insecticides, herbicides, fungicides, and improved cropping patterns. Labor-displacing technologies include tractors and threshers. Adoption of a new technology affects the use of domestic resources (land, labor, and domestic capital) directly and indirectly. These effects need to be identified and measured, because they may have implications for policies relating to resource utilization and allocation.
This study investigates the direct and indirect effects of farm mechanization and measures:
* the domestic resource and foreign exchange costs of machinery production and use; and
* the effects of farm mechanization on domestic resource and foreign costs of rice production, that is, the ratio of domestic resource cost to the net foreign
exchange earned from rice production.
Economics Department, Kasetsart University.

The first stage in measuring the domestic resource cost of mechanized rice production involves estimating the per unit social cost of production for each machine type and the social cost of-operating a machine or an animal for crop production. The marginal costs of undertaking farm activities using different machine types and intensities, such as land preparation by two-wheel and four-wheel tractors, or rice threshing by animal power, two-wheel tractor, or thresher, can then be estimated. The second stage measures the domestic resource cost per unit of net foreign exchange earned from rice production at different levels of farm mechanization.
The social cost of machine production and use The social cost of machine production and use can be measured by estimating the costs of component inputs valued at their social prices. For primary or nontradable inputs (land, labor, and domestic capital), the social price was assumed to be the same as the market price. For imported and tradable inputs, the social prices were equated with their border values: c.i.f. if imported, and f.o.b. if exported. The social cost pet unit is the sum of the social costs of the individual components. This can be written as:
n m
SCk X piqi + I pjq-.SER (1)
i=2 j=n+ 1
p's are the social prices,
q's are the amounts,
SER is the shadow exchange rate (baht/dollar),
i = 2, .... n refers to primary and nontradable inputs, and
j = n + 1, .... m refers to the tradable inputs.
By convention the P's are measured in local currency and the Pj's are converted to local currency using the shadow exchange rate which reflects the tax structure on imports and exports (McCleary 1976).
Domestic resource cost of a unit of net foreign exchange earned
The domestic resource cost (DRC) of a unit of net foreign exchange earned is defined as the social cost of primary inputs used divided by the net foreign exchange earned. The net foreign exchange earned is defined as the difference between the border value of output and the value of tradable inputs at border price.
DRC- DC i=2 (2)
NF m
pjql Y. p j" SER
j= n+ 1
p, = social price of output;
q, = quantity produced; and
all other terms are as defined above

The DRC reflects a comparative advantage in rice production if its value is less than the shadow exchange rate (SER) (Akrasanee and Wattananukit 1976, Monke et al 1976, and Pearson et al 1976).
Since farm mechanization requires both imported capital input (engines) and imported operating inputs (fuel and spare parts), the quantity of foreign exchange used will increase with machine use. For a given yield, if an increase in machine use does not reduce the quantity of domestic resources used sufficiently to compensate for the DRC increase, production at that level of farm mechanization will be relatively inefficient. The effects of farm mechanization on DRC can be measured by the mechanization elasticity of DRC, defined as the percentage change in DRC resulting from a 1% change in mechanization.
The primary data used for the analysis came from the farm record-keeping activity of the Thailand component of the Consequences of Mechanization project, 1979-80 wet season.The mechanization levels for the rainfed areas were animal only (A), animal and four-wheel large tractor (A+TL), animal combined with power tiller and tractor (A+PT+TL), and power tiller and tractor (PT+TL). The mechanization levels for the irrigated areas were (A+PT), (A+TS), (A+PT+TS), (PT+TS), (PT), and (TS), where TS indicates a small locally manufactured tractor. In the rainfed areas, large tractors were usually hired by the farmers for the first land preparation. In contrast, the tillers and locally manufactured four-wheel tractors used in the irrigated areas were owned by the farmers. Although observations for each mechanization category were few, the details collected should ensure accuracy.
Secondary data on tax and tariff rates, the cost structure of farm machine production, marketing, and transportation costs were collected from the reports of Customs Department, The Bank of Thailand, and National Economic and Social Development Board (NESDB).
Social cost of machine production
The engine is the major imported component of locally made machinery. The foreign cost of locally made machinery, at the border, ranged from 56 to 60% of production cost, whereas the primary factors used ranged from 33 to 39% (Tables I, 2, and 3). For the imported large tractors, the share of foreign cost in the assembling cost was 83% and the primary input required for assembly was only 17%.
Table 1. Share of engine and body in the total value of locally made farm machines.
Body and Average investment cost
Type of machine Engine accessories Year
(%) (%) $ of estimate
Two-wheel tractor (PTf 65 35 1020 1974-79
Small four-wleel tractor (TSf 46 54 1540 1973-79
Rice thresher 51 49 1072 1978
Water pump 82 18 119.8 1973-79
aFrom farm reord-keeping data, Supanburi Province, 1979-80 wet season. b Pathnopas (1980).

Table 2. Cost breakdown of machine body by type of locally made farm implement.
Costa ($)
Power tillerD Local tractor Imported tractor Rice thresherc Material cost
Imported 41 32 83 41
Domestic 36 40 14 40
Direct labor cost 12 10 1 15
Electricity 5 5 0 2
Depreciation, interest, 7 13 2 2
and taxes
Total 211 761 10000 748
Year 1978 1978 1978 1980
aExchange rates, 1978 baht 20.25 = US$1, 1980 baht 20.40 = US$1. bSamahito and Kongkietngarn (1979). cRungroj shop. A Rice Thresher Manufacturer in Chachoengsao Province, Thailand, April 1981.
Table 3. Primary and imported input components of farm machines.
Components (%)
Item Two-wheel Small four- Large Rice Water
tractor wheel tractor tractor thresher pump
Primary input in
Machine body and
implement 19 33 nab 28 n
Engine 15 11 na 12 na
Subtotal (1) 34 44 17 40 21
Foreign cost
Machine body and 14 18 na 20 na
Engine 46 32 na 36 na
Subtotal (2) 60 50 83 56 66
Taxes levied on
Machine body and 2 3 na 1 na
Engine 4 3 na 3 na
Subtotal (3) 6 6 0 4 13
a Included all nonforeign costs of manufacturer. Costs of engine were calculated from the equation ul = (1 + aj + tj + 77) uf*. See Appendix 1. bna = not available.
Social cost of machine operation among mechanization methods The social costs of farm mechanization for land preparation and rice threshing are in Tables 4 and 5. The social cost of land preparation by a two-wheel tractor is higher than that by a locally made four-wheel tractor because two-wheel tractors use more primary and tradable inputs. This is explained largely by the slower work rate of a two-wheel tractor while still requiring one man for operation, and the higher cost per hp for smaller engines. Land preparation by buffalo, costs more per hectare than land preparation by a two-wheel tractor, but no imported inputs are required (Table 4).
Because of the considerably lower use of domestic resources, the social cost office

Table 4. Social cost of alternative methods of land preparation, 1979-80 wet season, Supanburi, Thailand.
Two-wheel Small four- Buffalo
Item tractor wheel tractor
(baht/ha) (baht/ha) (baht/ha) Primary input
Labor 56.13 28.06 168.38
Capital 57.37 24.55
Indirect labor and capital 22.33 16.63 60.00
Total primary input 135.83 69.21 228.38
Tradable inpua 47.70 17.69
Taxes 19.50 10.00
Total 203.80 96.93 228.38
aValued at shadow exchange rate: baht 23.44 = US$1. Source: Farm record keeping data, 1979-80 wet season.
Table 5. Social cost of alternative methods of threshing rice, Supanburi, Thailand, 1978 wet season.
Thresher Two-wheel Buffalo
Cost item (baht/t) tractor treading
(baht/t) (baht/t)
Primary inputs
Labor 35 97 180
Capital 15 19
Indirect labor and
indirect capital 12 37 13
Total primary inputs 63 152 193
Tradable inputsa 27 34
Tax 1 3
Total costs 92 190 193
aValued in baht at shadow exchange rate: baht 23.44 = US$1. Source: Pathnopas (.1980).
threshing with a thresher is much lower than that with a two-wheel tractor or buffalo. The use of tradable inputs is also lower for the thresher. Threshers can save domestic resources, but they require imported fuel, oil, and an engine (Table 5).
The DRC coefficients of rice production in the rainfed areas were 26-75 baht/dollar, considerably higher than the 9.40-25 baht/dollar in the irrigated area (Table 6). The DRC coefficients for the rainfed areas were all greater than the SER as estimated by McCleary (1976), indicating no comparative advantage to rice production in these areas given the yields obtained in 1979-80.
The favorable coefficients for each category in the irrigated area resulted from the much higher yields obtained. Only power tiller users did not have a comparative advantage in rice production, and this group also had the lowest yield.
From Table 7, the elasticities of DRC with respect to tractor, rice thresher, and water pump use are positive, indicating that an increase in machine use will increase

Table 6. Domestic resource cost (DRC) for selected rainfed and irrigated rice production techniques in Supanburi, Thailand, 1979 wet season.
Water Level of Observations Area Yield DRC DRC/SERb
control mechanizationa (no.) (ha) (t/ha) (baht/$)
Rainfed A 9 3.7 0.37 56.43 2.41
A+PT+TL 3 5.9 0.46 75.08 3.20
A+TL 3 4.1 1.19 31.08 1.33
PT+TL 2 4.1 0.85 26.60 1.13
Irrigated A+TL 5 3.0 2.68 16.90 0.72
A+PT 2 2.5 3.84 13.61 0.58
PT 6 4.0 1.89 25.23 1.08
A+PT+TS 2 6.2 2.28 17.50 0.75
TS 3 3.0 3.33 15.60 0.67
PT+TS 2 5.6 6.25 9.64 0.41
aA = anial, PT = two-wheel tractor, TS = small four-wheel tractor, TL = large four-wheel tractor. A shadow exchange rate (SER) of baht 23.44 = $1 was used for all calculations.
Table 7. Estimated inyut and output elasticities of domestic resource cost (DRC) coefficients, Supanburi, Thailand.
Water control Type of land DRC coefficients
preparation Rice Water
Yield Land PT T theer pm
thresher pump
Rainfed A -0.50 0.61 0.00
A+PT+TL -0.58 0.60 0.17 0.05 0.05
A+TL -0.30 0.52 0.14 0.01
PT+TL -0.44 0.62 0.30 0.0 0.01
Irrigated A+TS -0.33 0.44 0.04 0.02 0.0
A+PT -0.35 0.39 0.04 0.0
PT -0.44 0.51 0.17 0.02
A+PT+TS -0.40 0.50 0.00 0.03
TS -0.37 0.40 0.08 0.02 0.02
PT+TS -0.28 0.34 0.01 0.06
aA = animal, PT = two-wheel tractor, T = four-wheel tractor, WP = water pump, TL = large fourwheel tractor, TS = small four-wheel tractor.
the ratio of domestic resource to the net foreign exchange earned from rice production DRC. The elasticity with respect to farm mechanization, however, is less than that with respect to land.
Comparing the types of power used for land preparation, the mechanization elasticity of DRC was lower on irrigated farms which used small four-wheel tractors than on farms which employed two-wheel tractors. The main reasons appear to be the lower operating cost and the lower work hours per hectare of the small fourwheel tractor.
Where land preparation was done with a combination of tractor and animal power, the mechanization elasticity of DRC was less than where machines were used alone, regardless of whether the farm was irrigated or rainfed.
The negative sign of the yield elasticity of DRC indicates that an increase in yield will lower the DRC coefficient, whereas a yield reduction will increase the DRC

coefficient. The elasticities, ranging between -0.30 and -0.58, imply that the yield effect on the DRC value is greater than the mechanization effect.
The best alternative methods were the locally made four-wheel tractor for land preparation and the thresher for harvesting. These reduced the use of both domestic and nondomestic resources.
The estimated elasticities indicate that farm mechanization affects the DRC coefficient value less than either yield or land does, because DRC elasticities with respect to the varied types of machine are less than those with respect to yield and land cost.
Akrasanee, N. and A. Wattananukit. 1976. Comparative advantage in rice production in
Thailand. Food Res. Inst. Stud. 15, 2.
McCleary, W. A. 1976. Equipment versus employment: a social cost benefit analysis of
alternative techniques of feeder road construction in Thailand. International Labour
Office, Geneva.
Monke, E., S. C. Pearson, and N. Akrasanee. 1976. Comparative advantage, government
policies, and international trade in rice. Food Res. Inst. Stud. 15, 2.
Pathnopas, R. 1980. The economics of rice threshing machines in Thailand: a case study of
Chachoengsao and Supanburi Provinces MS thesis, Faculty of Economics, Thammasat
University, Bangkok.
Pearson, S. R., N. Akrasanee, and G. C. Nelson. 1976. Comparative advantage in rice
I production: a methodological introduction. Food Res. Inst. Stud. 15, 2.
Samahito, P., and K. Kongkietngarm. 1979. Report on farm machinery and farm equipment
production: Part I. Farm tractor and power tillers. Bangkok: Industry Section, Economic Research Department, Bank of Thailand.

Appendix 1. Social value of output and tradable inputs.
1. Social value of output in 1979
1) f.o.b. export price of 5% broken white rice = 6.8136 B/kg
2) farm price of paddy (1.5 kg of paddy = 1 kg
white rice) = 4.42 B/kg
3) Export taxes = 0.95 B/kg
4) Processing cost, transportation cost and the
profit of trader = 1.4437 B/kg
Assuming that 20% of 1.4437 B/kg (or 0.289 B/kg) is the foreign cost and the rest was unallocated primary input.
If. Social value of tradable input
Tradable inputs are seed fertilizer, insecticide, herbicide, fuel, and oil. Tradable inputs also include the tradable component of machine services.
Rice seed was valued at the farm price. The following equation was used to calculate the border value of imported inputs:
Ui= (1 +a+ti + Ti) Ui*
where U is user's cost (value in baht at official exchange rate),
a is added cost including wholesale retail and semi-processing cost,
t is transportation rate,
T is overall tax,
Ui* is border value of imported inputs (valued in baht at official exchange rate).
In this study, al is given as the share of wholesale and retail value added to gross domestic product (GDP). The t is also assumed as the share of transportation and communication value added into GDP/. In 1979 a/ and t-were 24.36 and 7.65%, respectively. The imported content in transportation cost was small. The straight line method, assuming a 10-year machine life, was used to calculate machine depreciation. The average annual utilization of two-wheel fourwheel, and large tractors used in the calculations were 246, 360, and 1,000 hours, respectively. Annual utilization of rice threshers was set at 304 hours and that of water pumps at 300 hours.
The overall tax (T) for each type of imported input was calculated from
T= t + b (1 +P) (1 + t) + mb (1 +P) (1 + t)
= t+b (1 +m) (1 +P)(1 + t)
where P standard profit,
t = tariff,
b = business tax, and
m = municipal tax.
The tax rate of each import item was obtained from the Department of Customs, Ministry of Finance, Thailand (Tariff and Business Tax, 1977). Overall tax rates for each imported input are in Appendix 2.
The primary and imported input components of farm machines used in the calculation of tradable and nontradable components for the depreciation of machines at users' costs are in Appendix 3. These included the added cost between the manufacturer and farmer. The shares of primary inputs are higher than in Table 3 which was calculated at the factory. Similarly, the share of foreign costs here is lower than in Table 3.

Appendix 2. Tax rates for imported agricultural inputs.
Tax rates M%
Import items (t) (p) (b) (in) (T)
Tariff Standard Business Municipal Overall profit tax tax tax rate
Fertilizer free 11.5 1.5 10 1.84
Insecticide 5% 11.5 1.5 10 6.84
Fungicide Herbicide
(arsenic) 10% 8.5 1.5 10 11.97
Tractor with implements free 16.0 8.0 10 3.83
Accessory input free 26.0 3.0 10 4.158
for tractor
Power engine 5% 13.0 3.0 10 8.915
Water pump15160 300 194
complete set15160 300 194
Accessory and spare
parts for water pump 15% 26.0 3.0 10 19.78
Diesel oil 0.22 B /1 value 0.7065 B /1 10 1.30 B /1
Gasoline oil 0.93 B /I value 3.045 B /1 10 3.488B/1
Engine oil and 25 B /1 value 0.30 10 25.45 B /1
Appendix 3. Component shares for farm machines.
Component share(%
Cost item Two-wheel Small four- Large Rice Water
tractor wheel tractor tractor thresher pump
Primary input in
machine body and
implements 23 38 na 34 na.
engine 15 11 na 12 na.
Subtotal (1) 38 49 37 46 21
Foreign cost in
machine body and
implements 11 13 na 15 na,
engine 46 32 na 36 na
Subtotal (2) 57 45 63 51 66
Taxes levied on
machine body and
implement 1 2 na na a
engine 4 3 na 3 na
Subtotal (3) 5 5 0 3 13

M. A. Jabbar, M. S. R. Bhuiyan, and A. K, M. Bari
Causes and consequences of power tiller utilization were examined using,- data collected from 63 tiller owners and 56 nonowners.
Timely and quick cultivation, difficulty in managing large numbers of animals, low cost and better quality tillage, and animal shortage were the main reasons for purchasing tillers. Costs were low because of distortions in the prices of tillers and fuel. Unavailability of spare parts and lack of repair facilities were major problems.
Tiller use increased size of cultivated holding, decreased regular labor, evicted tenants, changed tenure status, increased cropping intensity, and increased machine orientation of farmers. The findings indicated that mechanization of tillage would greatly benefit
rich farmers at the expense of small and marginal ones.
The level of tillage mechanization in Bangladesh is quite low, with more than 98% of the land cultivated by bullock-drawn plows. There is, however, a growing shortage of power because long-term neglect of the livestock sector has resulted in decreased availability of bullocks. Use of mechanical power has not increased fast enough to fill the gap. The power gap, along with other factors, has probably been responsible for the slow growth of output and employment, and unless that gap is immediately filled, growth of agriculture might be further constrained (Jabbar 1980). The use of mechanical power has grown very slowly. Causes of the slow growth need to be identified and eliminated before any rapid increase can take place. The consequences also need to be measured for appropriate policy making.
A survey of power tiller owners and nonowners was conducted in two selected areas to:
* Identify the reasons for buying or not buying tillers;
" Study the characteristics of the tillers and the process of their acquisition and
" Identify factors affecting the nature and extent of tiller use; and
* Measure the consequences of tiller use on selected aspects of the farm business, Particularly 1) ownership of animal power, 2) ownership of land, 3) cultivation Bangladesh Agricultural University.

of land, 4) tenure status, 5) cropping pattern and intensity, 6) employment of
labor, and 7) machine orientation.
Since the early sixties, 6,362 power tillers have been imported, of which 4,278 were reportedly sold to the private sector on highly subsidized credit provided through the Bangladesh Krishi Bank (BKB). The actual number and geographical distribution were not known but it was believed that operational tillers were scattered throughout the country, with a few small pockets of higher concentration. Mymensingh District and a cluster of four villages in Munshigonj Thana of Dacca District were selected as areas of low and high tiller density, respectively. The four Munshigonj villages were about 3 miles south of Munshigonj Thana headquarters. Mymensingh District covered over 10,000 km2 while the 4 Munshigonj villages covered about 7.7 km2.
Preliminary fieldwork revealed that a large number of tillers had been sold by the original owners, some to buyers outside Mymensingh. A large number of tillers had been out of operation for a number of years, and some operating tillers had been purchased from sources other than BKB. Complete enumeration showed 34 operational and 26 nonoperational tillers. Apparently, more than 150 tillers sold by BKB had been resold outside Mymensingh. Subsequent analyses werd based on data collected from these tiller owners plus data from 26 nonowners located in the same villages.
Of the 29 operational tillers found in the four Munshigonj villages, 73% were in 1 village, South Char Masura. Data were collected from all 29 owners plus 22 nonowners on 3 visits to each farm.
Reasons for buying tillers
Tiller owners were asked their reasons for purchasing tillers. Nonowners were asked whether they were interested in buying and, if so, why. Fifty-six percent of nonowners in Mymensingh and 82% in Munshigonj expressed interest in buying tillers. The relative importance of the reasons varied among owners and intending owners and also between the two areas. In general, frequently stated reasons included increased output, reduced costs, and reduced drudgery (Table 1).
The implications of some of the responses need explanation. First, 30% of both owners and intending owners reported a shortage of animal power. Only 8% of intending owners mentioned labor saving. This supports recent findings of a significant shortage of animal power, which machines would overcome (Jabbar 1980).
Second, 67% of the owners and 24% of intending owners expected tiller cultivation to be cheaper. At market prices, bullock cultivation was found to be 3-5 times more expensive than tiller or tractor cultivation (GOP 1970, Lawrence 1970, Mian and Hussain 1975) because the overvalued currency underpriced both tillers and fuel by 40-50%. At real prices, tractor or tiller cultivation was found to be 2-3 times more expensive than bullock cultivation (Lawrence 1970). Such distorted markets encourage substitution of animals by machines even in situations where labor is still plentiful and cheap.

Table 1. Distribution of reported reasons for purchase and intended purchase of
a tiller.
Reported Owners (%) Intending owners (%)
reasons Mymensingh Munshigoni Mymensingh Munshigonj
cultivation by tiller 100 100 45 76
Animal management
difficult 50 100 10 17
Cheaper cultivation 82 48 25 22
by tiller
Better land preparation 50 14 55 17
Animal power
shortage 15 48 30 28
Multiple uses of tiller 18 10
Tiller custom service 13 5 33
Available on credit 12 3
Good for puddling
dry hard soil 6 3
Labor saving
High death rate of
weak animals 10
Sample size 34 29 18 18
Third, it has been argued there is a technical limit to the optimum size of a farm using animal and human labor. One reason is the management problem. Seventythree percent of the owners and 14% of the intending owners indicated they had difficulty managing a large number of animals. Large landholders generally cultivate land up to that technical limit and rent out any excess. But engine power was expected to, and did, induce more self-cultivation (Jabbar 1977, 1980).
Fourth, one-third of the intending owners, mostly in Munshigonj, wanted to buy a tiller so they could provide custom services. Custom operation was already booming in Munshigonj and intending buyers might have been influenced by the prospects.
Fifth, only 7% of the owners mentioned availability of credit as a cause for buying a tiller. Credit was available for purchase of all new tillers.
Reasons for not buying a tiller
The reasons nonowners did not buy a tiller are in Table 2. More than 50% of the nonowners were interested in buying when their neighbors bought, but they could not because of capital shortage or unavailability of tillers. Most nonowners not interested in a tiller were located in Mymensingh. Their primary reasons were related to repair and maintenance. Many of the current tiller owners already face this major problem.

Table 2. Distribution of nonowners' reported reasons for not purchasing a tractor.
Reported All nonowners (%) Unwilling nonowners (%)
cause Mymensingh Munshigonj Mymensingh Munshigonj
Lack of capital 38 18
Not available 21 27
Could not manage 14 -to buy
Lack of knowledge or
experience with 6 23 14
Had/have adequate 6 14 29
maintenance, 15 5 85 50
repair problems
Did not like 12 5
Animal cheaper 6 5 36 50
Sample size 34 22 16 4
Brand and capacity
Tillers were either received as grant aid or imported on credit from Japan. Of the sample, 48% were Yanmar, 40% Mitsubishi, and 6% Kubota and Isaki. However, 70% in Mymensingh were Mitsubishi and 83% in Munshigonj were Yanmar.
Tiller capacity was 6-10 hp. Ninety-two percent of the Mitsubishi tillers were 6-8 hp and 82% of the tillers in Mymensingh were 6-8 hp. In Munshigonj, 55% were 6-8 hp and 45% were 8-10 hp.
Additional equipment and tiller
In Mymensingh, 50% of the tillers were purchased with one or more attachment. Attachments reported were pump (25%), trolley (23%), hauler (6%), and furrower
(6%). Nine percent of tillers were purchased without additional equipment, but the owners bought them later. In Munshigonj, only one tiller was purchased with a pump attachment and two other owners purchased threshing equipment later.
Most owners of tillers with no attachments wanted to buy and those with some wanted more. In Mymensingh, tiller owners wanted to buy a trolley (35%), pump (32%), hauler (24%), and equipment for threshing, seeding, and electric generation (12%). In Munshigonj, 34% wanted a trolley and 65% wanted a pump.
Source and time of purchase
The government first imported tillers for experimental purposes, but a few were subsequently sold to farmers. Later tillers were imported and distributed through the BKB and also through private dealers who generally sold on credit from BKB. During 1972-74, the Bangladesh Agricultural Development Corporation (BADC), the public corporation responsible for import and distribution of agricultural inputs, distributed tillers with BKB providing credit.

Acquisition dates suggest that tillers have been use longer in Mymensingh than Munshigonj. In Mymensingh, 32% of the tillers were purchased during 1963-68, 41% in 1969-74, and 27% in 1975-80. For Munshigonj, the corresponding figures were 7%, 24%, and 69%, respectively. Sources for purchase of tillers are in Table 3.
Secondhand purchases from other farmers have increased consistently. In Mymensingh, secondhand tillers were 44% compared to 55% in Munshigonj. Six percent were more than 6 years old at the time of purchase, 39% were 5-6 years old, and 39% were 3-4 years old. The main reasons for selling were repair, maintenance, and operational problems.
Tiller price and sources of capital
Tiller prices varied with time of purchase, machine condition, brand and capacity, number and type of accessories included, type of payment, and supply source. For new tillers brand and capacity and supply source had minimal effect. An index of average price by time of purchase, and tiller condition is in Table 4. Up to 1974, prices were quite low, and secondhand tillers were sometimes more expensive than new ones. After 1974, the price of new tillers increased about 300%, but prices of secondhand tillers remained fairly constant.
All tillers purchased on full cash were secondhand and those purchased on full credit were new. Although they were purchased from BKB, BADC, and private dealers, BKB provided the credit in all cases. With part cash payments, 38% of the price was paid in cash. Most tillers purchased with part cash were new and supplied
Table 3. Percentage of tillers purchased from different sources by period Sources (%) of supply
Period Dealer BKB Other farmers Othersb
1963-65 33 67
1966-68 50 40 10
1969-71 14 58 14 14
1972-74 7 7 29 57
1975-77 50 38 12
1978-80 10 19 71
All periods 16 31 38 15
aSource: Field survey 1980. bInclude tiller mechanics, Foreign Voluntary Agency,
and Bangladesh Agricultural Development Corporation.
Table 4. Index of average price by time of purchase and condition of tiller.
Period Index of av price
purchased Mymensingh Munshigonj
New Used New Used
1963-65 1.00
1966-68 0.90 1.20 0.78
1969-71 1.00 1.06 1.51
1972-74 0.76 1.01 1.94
1975-77 4.23 0.68 4.24 2.00
1978-80 3.05 1.63 3.58 2.23

by BKB, BADC, or a private dealer with BKB credit. Most sellers of used tillers accepted part cash payment. In two cases, full cash payment was made by borrowing from private lenders. The interest rate for BKB credit increased from 5% in the mid-1960s to 10% in the late-1970s.
Of 36 owners purchasing with credit, 64% had repaid fully, 25% partially, 6% not at all, and 5% had not yet reached the repayment stage. The number of defaulters was similar in both areas, but those in Mymensingh had been defaulting longer, some since 1966.
Characteristics of tiller operators
Respondents were asked who operated the tiller during the survey year. In Mymensingh, 12% used only family members, 32% annual hired labor and family members, 35% only annual hired labor, 12% hired tiller drivers, and 9% hired tiller drivers and family members. In Munshigonj, 14% used only family members, 45% hired tiller drivers, and 41% hired tiller drivers along with family members.
Over 90% of the family members operating tillers in Mymensingh had some secondary education and 15% were high school graduates. In Munshigonj, more than 40% of family tiller operators had no formal education and none had above secondary education. In both places, few tiller operators from thenther categories had any education; none were educated beyond primary level.
Nine percent of the tiller owners in Mymensingh and 35% in Munshigonj reported receiving no training in tiller operation and maintenance. They learned mostly from other tiller drivers. Those receiving training obtained it predominantly from sellers: 29% of owners in Mymensingh and 48% in Munshigonj reported that a member of the family, who was trained by the supplier, taught other family members and hired laborers.
Major and minor breakdowns
The number of major breakdowns of the tillers since acquisition is in Table 5. Breakdowns increased with the tiller age at the time of purchase. The main reasons given for breakdowns were overturning during operation, loose-fitting parts, irregular gasoline delivery, and excessive or insufficient oil use. No owners could specify the reason for the third and fourth major breakdowns and 54% could not name the cause of the first and second major breakdowns.
During 1978-80, 64, 23, and 13% of the owners reported doing I to 4, 5 to 8, and more than 8 minor repairs, respectively. Repair frequency was significantly higher in Munshigonj. Average repair costs for the 2 years were $24 in Mymensingh and $41 in Munshigonj, and 20 and 14 potential work days, respectively, were lost.
Table 5. Major breakdowns of tillers by region.
Breakdowns (no.) Mymensingh Munshigonj
(%) (%)
0 29 52
1 35 28
2 27 17
3and4 9 3

Repair facilities
Service guarantees, ranging from I to 3 years, were assured for 79% of the new tillers purchased in Mymensingh and 56% in Munshigonj. No guarantee was available for used tillers. Of those with service guarantees, 59% reported receiving proper service, 14% did not require service, and 27% (located in Mymensingh) did not get service apparently because of problems with suppliers.
In Munshigonj, repair facilities were available within 8 km. In Mymensingh, 30% of owners reported that the nearest repair shop was more than 64 km away, 30% reported between 32 and 64 km, and the remainder reported between 8 and 32 km. Tillers usually had to be hauled to repair shops on trains, trucks, boats or bullock carts, and then most of the repair shops did not stock sufficient parts. Sometimes transportation was not possible and mechanics, who charged very high fees, were called.
Tiller problems
The main problems in tiller use were unavailability of spare parts, lack of repair facilities, high priced spare parts and fuel, and unavailability of pure diesel (Table 6).
Information on the extent of tiller use was collected from the owners for 1976-77 to 1979-80. For 1979-80, data were collected for each operation whereas for the 3 other years, the limited records kept by the users were supplemented by their recollections. Annual use varied from 640 to 696 hours in Mymensingh and from 1, 144 to 1,432 hours in Munshigonj. Detailed analysis is based on 1979-80 only.
Various characteristics of tiller users in Mymensingh and Munshigonj are in Table 7. In Mymensingh, only 38% of the owners provided custom services, compared to 97% in Munshigonj. Custom services accounted for 9% and 59% of total operations, respectively. Possible reasons for the lower level in custom services in Mymensingh are fewer repair services and large landowner concern about loss of status in the area.
Forty-eight percent of the owners in Munshigonj traveled 16-40 kin, usually by boat to provide custom services and another 11I% traveled 8-10 km. Farmers with
Table 6. Problems in tiller use reported by owners.
Problem Mymensingh Munshigonj All areas
M%) (%) (%)
Unavailability of spare parts 94 76 86
Frequent breakdown 38 17 29
Lack of repair facilities 41 14 29
High price of spare parts 35 3 21
High price of fuel 35 7 22
Unavailability of pure diesel 15 31 22
High charge for mechanics 6 3 5
Lack of training facility 3 -2
for repair work
Lack of efficient tiller 3 -2

Table 7. Extent of tiller use in 1979-80.
Mymensingh Munshigonj
Characteristics Farms ()Hours/farm Farms (%) ours/farm
Type of work
Tillage only 62 496 100 1192
Tillage and other tasks 38 1008
Area cultivated (ha)
Under 6.0 23 176 52 872
6.1-8.0 23 536 21 1448
8.1-10.0 15 864 14 1800
10.1-12.0 8 1864 3 1960
12.1 and over 31 824 10 1224
Type of family
Single 68 568 56 1088
Joint 32 976 44 1328
Past experience in machine use
Yes 53 864 27 1304
No 47 424 73 1152
Main income source
Farming 47 504 35 1352
Farming and business 44 752 62 1096
Farming and service 9 960 3 1400
inadequate, or no, draft animals bought tiller services. Twenty-four percent of the owners reported charging lower custom rates in distant places, but still making a profit because custom services were done after finishing their own work.
Tiller use increased with size of cultivated holdings up to 12 ha and then declined sharply. This size effect was indirectly reflected through type of family because most joint families had larger holdings.
Owners with experience in handling different types of machines used the tiller longer than those without such experience. Experienced owners could do minor repairs, getting more use from the tiller. In Mymensingh, owners who had income from business or services as well as farming had better external contacts which helped them locate mechanics, manage parts, and make quick repairs.
There are three approaches to measuring the effects of tiller use. First, tiller owners and nonowners may be compared at a given time with differences attributed to tiller use. The main problem with this approach is that owners and nonowners may differ in respects unrelated to tiller ownership. The second approach involves before-andafter comparison, with any differences attributed to the tiller. Here the main problems are that other changes might have taken place simultaneously and "before" data must be collected by recall, which is less reliable. The third possibility is to combine the cross section and time series approaches (Binswanger 1978). All three approaches were used in this study.

Effect on animal ownership
Tiller ownership was expected to have an immediate effects on work animal ownership. Changes in the number of owned work animals and the number per cultivated hectare are in Table 8. On the average, 2-2.5 animals were replaced by a power tiller. The degree of substitution was much higher in Munshigonj where 53% of tiller owners completely replaced their animals. Although only 3% of the farms in Mymensingh replaced all their animals, some of the larger ones replaced 5 or 6, yet still retained several because: 1) the tiller was not considered fully reliable, 2) the tiller was not suitable for puddling in low-lying areas or for preparing dry hard soils in summer, 3) land preparation with a tiller followed by laddering with animals gave better results, 4) during short sowing or planting seasons some larger farmers needed to supplement their tiller power with draft animals, 5) some farmers had increased their operational holdings beyond the capacity of one tiller, and 6) animals were fed mostly on crop by-products and could be retained for investment at little additional cost.
Effect on land ownership and tenure Those who bought tillers in both areas normally had larger forms than those who did not (Table 9). By 1979-80, the farms in all categories were significantly larger, but the relative differences between owners and nonowners remained similar. Indeed, a substantial proportion of farmers had acquired additional land whether they had tillers or not. Tenure status of tiller owners and nonowners in both areas changed substantially. However, the impact of the tiller on and changes could not be ascertained.
Table 8. Changes in work animal ownership on farms of tiller owners and
Mymensingh Munshigonj
Owner Nonowner Owner Nonowner Number of animals/farm
Before tiller purchase 7.3 6.4 3.6 3.1
1979-80 5.3 6.8 1.1 2.4
% change -27 +6 -70 -23
Number of animals per
cultivated ha
Before tiller purchase 0.22 0.20 0.17 0.19
1979-80 0.11 0.19 0.05 0.13
% change -48 -8 -69 -33
Table 9. Changes in size of land ownership of tiller owners and nonowners.
Land ownership
Mymensingh Munshigonj
Owner Nonowner Owner Nonowner Year before tiller purchase (ha) 8.64 6.23 3.15 2.70
1979-80 (ha) 9.53 7.10 3.87 3.36
% change 10 14 23 25

Effect on cultivated area
There were negligible differences in the average size of cultivated holdings of tiller owners and nonowners in both areas at time of purchase (Table 10). By 1979-80, the average size of cultivated holdings of tiller owners and nonowners had increased in both areas. The difference between the two groups increased in Mymensingh but not in Munshigonj.
Regardless of tiller ownership, most farmers increased their areas under cultivation, and the predominant mechanism was acquiring new land (Table 11). Tiller owners in Mymensingh also increased their cultivated holdings by using land previously rented out or left fallow. Previously rented out land was used by 81% of tiller owners in Mymensingh, compared to only 33% in Munshigonj.
Effect on cropping pattern and intensity Because farmers could not be expected to recall crop areas accurately over several years, cropping pattern and cropping intensity differences of the two groups were compared using average cropping patterns for 1978-79 and 1979-80. The proportions of irrigated crop area and of area devoted to high-yielding varieties (HYV) were similar for owners and nonowners in both areas. However, contrary to the findings of Gill (1979, 1980), cropping intensity differed markedly between the two groups in Munshigonj. It is likely that the increase in cultivated area in Mymensingh and in cropping intensity in Munshigonj was the result of different initial sizes of holdings.
The increased cropping intensity in Mushigonj was accompanied by greater diversification. A decrease in the paddy area on farms with tillers was accompanied by increases in the areas for potato, jute, mustard, and wheat.
Table 10. Changes in sizes of cultivated holdings of tiller owners and nonowners.
Cultivated holdings
Mymensingh Munshigoni
Owner Nonowner Owner Nonowner Year before tiller purchase (ha) 4.75 4.80 2.92 2.63
1979-80 (ha) 7.59 5.95 3.61 3.11
% change 60 24 24 18
Table 11. Sources of additional cultivated land.
Additional cultivated land (ha)
Source of land Mymensingh Munshigonj
Owner Nonowner Owner Nonowner New land acquired 0.89 0.87 0.72 0.66
Cultivation of previously 1.42 0.35 -0.08 -0.19
rented out land
Cultivation of previously 0.60 -0.06
fallow land
Mortgaged land -0.03 -0.01 0.05
Total additional land 2.84 1.15 0.69 0.48

Table 12. Changes in regular hired labor employment by tiller owners. 0
Hired labor/farm Location, Total no./cultitime period Annual Seasonal Other annual labor Tiller Tenant Total vated ha z
plowman plowman Male Female driver farmer
Mymensingh a
Before tiller purchase (no.) 2.15 0.79 3.90a 0.90 7.00 14.74 3.11
1979-80 (no.) 1.68 0.38 3.60a 1.00a 0.20 3.00 9.86 1.31
% change -22 -41 -8 +11 + -57 -33 -58
Before tiller purchase (no.) 0.66 0.31 0.90 0.30 2.00 4.17 1.43
1979-80 (no.) 0.28 0.14 0.60 0.30 0.86 2.00 4.18 1.14
% change -59 -55 -33 + -21
aAbout 20% below 14 years old.

Effect on employment
Although the introduction of tillers may have various effects on employment, the only one measured was the effect on regular hired labor. This was done using a before-and-after comparison (Table 12). In Mymensingh, there was a negative, aggregate impact on regular employment per farm; in Munshigonj, employment per farm remained constant. In both areas, however, use of regular labor per hectare declined.
Most of the evicted plowmen were working as casual laborers on the same farms or elsewhere, and some got plowing jobs elsewhere. About 60% of the evicted tenants in Mymensingh were cultivating their own small holdings, or renting. Others became day laborers.
Tiller use has changed the pattern of family labor participation in farming. For example, the number of families having members working as plow drivers declined from 12% to 3% in Mymensingh and from 45% to 21% in Munshigonj following tiller introduction. On the other hand, 54% of the families had members working as tiller drivers in 1979-80. Other changes included increases in the number of family members participating in farm activities, hours worked per person, and increase in supervisory work.
Effect on machine orientation
After purchasing tillers, owners wanted to buy more equipment. However, lack of money and available machinery limited additional purchases. Although most desired machines are labor displacing, some have potential for indirect employment expansion.
Machine use is being encouraged by distorting market prices in a situation where animals are in short supply, but labor remains quite cheap. Such policies may be justified if tiller use increases output, but there is little supporting evidence. The large scale introduction of tillers appears likely to benefit the rich farmers, at the expense of small and marginal ones, through employment, tenancy, land accumulation, and machine orientation. Detailed studies, incorporating mechanization of other farm operations, are required if there are to be sound farm mechanization policies.
Binswanger, H. P. 1978. The economics of tractor in South Asia. Agricultural Development
Council, Inc., New York, and International Crops Research Institute for the Semi-Arid
Tropics, Hyderabad. 96 p.
Gill, J. 1979. Appropriate technology: mechanized land preparation. The ADAB News
Gill, J. 1980. Appropriate technology: mechanized land preparation. The ADAB News
7(1): 18-19.

GOP (Government of Pakistan). 1970. Farm mechanization in East Pakistan: report of the
Farm Mechanization Committee. Islamabad, Pakistan. 225 p.
Jabbar, M. A. 1977. Relative productive efficiency of different tenure classes in selected areas
of Bangladesh. Bangladesh Dev. Stud. 5(1):1 7-50.
Jabbar, M. A. 1980. Draft power shortage and mechanization of tillage in Bangladesh.
Bangladesh J. Agric. Econ. 3(1):1-26.
Lawrence, R. 1970. Some economic aspects of farm mechanization in Pakistan. 61 p.
Mian, M., and M. K. Hussain. 1975. A comparative study of the economics of cultivation by
bullock and power tiller in the production of transplanted aman paddy in some selected areas of Bangladesh. Prod. Econ. and Farm Manage. Res. Rep. I. Department of
Agricultural Economics, Bangladesh Agricultural University, Mymensingh. 62 p.

J. Hafsah and R. H. Bernsten
Minitractors have been introduced into the relatively sparsely populated province of South Sulawesi to provide additional land preparation'power. To better understand the impact of this program, 50 individuals in Sidrap and Pinrang Districts were inteiviewed. Ten of them bought minitractors in 1975-79. Information about socioeconomic characteristics and various aspects of tractor operations was collected. Analysis showed that tractor ownership was not economically viable because of breakdowns and a shorter than expected useful life. Utilization of new tractors during the first year has declined each year since 1975. This suggests that increases in the tractor population have made it more difficult for owners to
cover variable and fixed costs.
Because rice is the staple food of Indonesia, government policy has focused on increasing production to achieve self-sufficiency. During the first five-year development plans (Pelita I 1969-74, and Pelita 111974-79), rice output increased 4.7 and 3.8%/year, respectively. Yet, with population growing at 2.5%/year and incomes rising, demand still exceeds domestic supply (B. P. S. 1980).
Minitractors have been introduced into several places in the less densely populated outer islands in an effort to increase production and to remedy apparent shortages of human and animal labor. These shortages constrain area expansion, crop intensification, and synchronized rice planting.
In the first tractor feasibility study in South Sulawesi, Parussengi (1972) concluded there was a need for small tractors, because of a labor shortage. Follow-up studies by Gadjah Mada University and the Agricultural Equipment Office evaluated tractor needs in 18 of the 23 districts in South Sulawesi (Gadjah Mada University 1976, Directorate of Food Crops 1975, Agricultural Extension Service 1976, Directorate for Technical Agriculture 1979). Based on these studies, fourwheel minitractors were introduced to several districts beginning in the mid-1970s. By 1980, almost 1,300 minitractors had been sold throughout the province (Agricultural Extension Service 1980). South Sulawesi Extension Service and Central Research Institute for Food Crops/IRRI Cooperative Program, Bogor, Indonesia.

Farmers in Sidrap and Pinrang first purchased minitractors in 1969. In 1975, a major government-sponsored program, aimed at increasing productivity, provided for considerably more tractors. An understanding of the successes and weaknesses of this effort should provide guidance for planning and implementing future tractor programs.
This study focuses on:
" the characteristics of minitractor owners and their reasons for purchasing;
* minitractor use;
* problems encountered in ownership, operation, and maintenance;
" financial arrangements and credit characteristics;
" the characteristics of tractor operators; and
" the profitability of tractor ownership.
Sidrap and Pinrang Districts are about 180 km north of Ujung Pandang, the provincial capital of South Sulawesi. This is a highly productive agricultural area and has a population density of 93 persons/ikm2 (B. P. S. 1980). Seventy percent of the 92,581 ha of lowland is irrigated, mostly by a modern technical system receiving water from the Saddang diversion dam (Agricultural Extension Service 1980). Typically, two irrigated crops are grown with farmers using the latest varieties along with fertilizer and insecticides.
Sidrap and Pinrang were chosen as the study area because these districts have the most minitractors. Four villages with the largest number were selected from each district. Of the 508 minitractors in the two districts in 1979, 108 were owned by residents of these eight villages. A few adopters bought units before 1976; however, most purchases (40%) were made in 1979 when Presidential Aid (BANPRES) was extended through the small investment credit system (KIK). Before the BANPRES program almost all purchases were Kubota, but in 1979 Iseki and Sateh captured 64% of the market.
Ten tractor owners, who purchased units in each of the previous five years, were randomly selected and interviewed once during each of the 1979 wet and 1979-80 dry seasons.
Characteristics of the sample farmers
Some characteristics of the 50 minitractor owners are shown in Table 1. Over 60% were between 31 and 50 years old and only 26% had education beyond primary school. Their principal occupation was farming (56%). Their average assets were 7.2 ha of riceland, valued at $17,120, and 3.1 cows and 0.7 buffalo, together valued at $520. The average value of all assets, excluding house and tractor, was $22,080.
By comparison, nonmechanized respondents in the same area had a similar age and education, but only 1.34 ha of riceland (Consequences Team 1981).
Respondents gave several reasons for purchasing a tractor (Table 2). Because most were large landowners, the most common answer was expected to be "to

Table 1. Characteristics of 50 minitractor owner respondents.
Characteristic Number
Age (yr)
4 30 0
31-40 15
41-50 16
> 50 14
Education (yr)
Illiterate 14
Primary school 23
Junior/senior high school 7
University 6
Farmer 28
Government employee 7
Trader 10
Merit/service 5
Table 2. Reasons given by 50 minitractor owners for purchasing a tractor.
Reason Responses
To cultivate their land 71
To rent to neighbors 40
To experiment 12
Social prestige 9
Available credit 26
aSome respondents gave more than one answer.
cultivate their own land." However, 25% expected to earn money renting the tractor to neighboring farmers, especially when pressed to meet repayments.
The respondents perceived several benefits from owning a minitractor (Table 3). Most important was the potential for timely planting, whereas only 10% mentioned increased yields and 8% reduced drudgery.
Minitractor characteristics
All the minitractors surveyed were rated at 12-15 hp and had diesel engines. Their costs which increased significantly over the period, averaged $5,513 (Table 4). These dollar prices understate the real rate of increase. With the 50% devaluation of the rupiah in November 1978, the price of a tractor to the Indonesian farmer in 1979 was 32% higher than in 1978.
In both wet and dry seasons, complete land preparation involved rototilling the field twice. However, some farmers hired the minitractor for only one rototilling then harrowed the field using animal or human power. Consequently, data on land preparation were converted into two rototilling equivalents. The hectares prepared each season since tractor purchase are shown in Table 5. For each purchase year, utilization was highest in the first year and declined in each successive year. At the

Table 3. Minitractor owners' perceptions of benefits associated with tractor ownership.
Reason Farmers (no.)
Timely planting 45
Better land preparation 39
Reduced need to hire labor 39
Increased yields 15
Reduced drudgery 12
Table 4. Purchase price of sample tractors by year of purchase.
Purchase year Purchase price ($) Price index
1975 4471 100
1976 4941 111
1977 5647 126
1978 6588 147
1979 5920a 132
Mean 5513
aDolar price affected by devaluation from $1 = Rp425 to $1 = Rp625.
Table 5. Hectares plowed each season, by year of tractor purchase a
Year, season Area plowed (ha)
1975 1976 1977 1978 1979
Wet 55.5
Dry 46.6
Wet 36.8 58.0
Dry 28.9 42.1
65.7 100.1
Wet 30.5 34.9 43.6
Dry 25.7 34.3 29.8
56.2 69.2 73.4
Wet 25.4 35.3 39.6 43.6
Dry 15.8 22.0 16.2 37.7
41.2 57.3 55.8 81.3
Wet 10.3 20.8 20.5 24.6 33.9
Dry 3.9 7.8 10.8 16.5 23.3
14.2 28.6 31.3 41.1 57.2
Average 55.9 63.8 53.5 61.2 57.2
aAverage of hectares plowed by all units operating in respective seasons.

same time, it is clear that late adopters were unable to achieve the same use in the first year as early adopters, suggesting the tractor population may be reaching saturation.
The hectares prepared each year were distributed evenly between seasons, with 55-60% of the area plowed in the wet season, when tractors are used on both rainfed and irrigated land.
The relationship between tractor age and number of seasons it was broken is shown in Table 6. Reasons for not repairing the tractor were spare part unavailability, repair expense, or the repair cost equaling that of a new unit. Of 10 tractors
Table 6. Number of seasons during which minitractors operated.
Purchase Av season operating per tractor Tractors (no.) Av area b
year Potentiala Actual Seasons broken operating n plowed in 1979
(no.) (no.) M 1979-80 (ha)
dry season
1975 9.1 7.3 20 3 14.2
1976 7.3 6.6 10 5 28.6
1977 5.9 5.5 7 7 31.3
1978 3.0 2.9 3 10 41.1
1979 1.9 1.9 0 10 57.2
aThe potential number of seasons the tractor could be used takes into consideration that some tractors were purchased after the beginning of a season and not used that season. By units still operating in 1979.
Table 7. Damage requiring spare parts, reported by 50 tractor owners.
Units (no.)
Part repaired or replaced e ninobe
experiencing problem
Injector 1
Piston 19
Dynamo 4
Brake 5
Joint 7
Chain 7
Gears 10
Clutch 6
Ring 1
Pump 2
Oil cylinder 2
Joint 3
Rotary blade 11
Floating wheel 2

purchased in 1975, only 3 were still operating in 1979 and they averaged only 14.2 ha. The data in Table 6 suggest that minitractors are relatively free from major problems for the first 2 years, but thereafter use is reduced significantly. It appears that the maximum useful life of minitractors is 5 years. As shown in Table 7, breakdowns were most frequently associated with the transmission (43%1). The parts requiring replacement most often were the piston (23%o) and associated components. Transmission damage was probably aggravated by erratic stopping and starting, engine damage from late oil changes, and continuous operating beyond the recommended hours per day. These explanations are consistent with the owners' observations (Table 8). Breakdowns might be greatly reduced through better driver training and supervision.
Respondents felt that spare parts were expensive and 43% reported they were not always available when needed. Other reasons for delays in repairs are in Table 9.
Financial arrangements and credit characteristics Ninety-four percent of the minitractors were purchased on credit 74% from banks, and 20%/ from dealers. Credit purchases were made easier by the BANPRES program, extending the payback period to 72 months, providing a low interest rate of 10.5%/ year, and reducing the land required for collateral from 5 to 2.5 ha (Team on Selective Mechanization 1979). Sole owners accounted for 81% of the tractor purchases.
Arrears on minitractor loans varied by year of purchase. At the end of the 1979 dry season, loan defaults were 0% for 1975 purchases, 12% for 1976, 5.4% for 1977, 15% for 1978, and 4% for 1979. The most important reasons for difficulties in meeting loan payments are in Table 10.
Table 8. Perceptions of 50 owners as to causes of damage to their units.
Reason Respondentsa (no.)
Poor maintenance 32
Improper driving 19
Driver turnover 18
Distance from workshop 8
Low quality equipment 7
Poor quality workshop service 4
a Sme respondents gave more than one answer.
Table 9. Reasons for delays in repairing broken units.
Reason Respondents(%
Spare parts not available 43
Distance from workshop 24
No time to bring minitractor to workshop 20
No money to pay for repairs 13

Minitractor operators
Because the tractor operator is a key to the length of service life, the 70 operators were interviewed. Generally, operators were younger and more educated than the owners (Table II). About 59% were sons of the owners, 37% relatives, and only 14% from outside the family. Slightly more than 50% of the operators had no other occupation. Of those with second jobs, 88% operated their own farm. Most operators had very little driving experience and training appeared to be inadequate (Table 1I). The few organized training courses had been sponsored by dealers (63%), the government (25%), and jointly (12%). Of the 12 operators who had attended a course, 5 were taught only driving, 3 only maintenance, and 4 both. Although dealers agreed to provide training for new owners, few owners operated the units themselves and they had insufficient knowledge for training the actual operators.
Table 10. Reasons given by 50 minitractor owners for having difficulties in
repaying their tractor loans.
Reason Respondents? (no.)
Minitractor damaged 45
Cannot plow enough area 30
Crop failure 12
Renters do not pay 9
Used for other purposes 6
aSome owners gave more than one reason.
Table 11. Characteristics of 70 minitractor operators.
Characteristic Number
Age (yr)
15-30 60
21-25 23
Formal education (yr)
0 8
1-6 48
> 6 14
Driving experience (yr)
1-2 45
3-4 21
>4 2
Training courses
0 58
1 11
2 1

The most frequent reasons individuals chose tractor driving were increased prestige, supplemental income, and job satisfaction (Table 12). Abbas (1977) noted that operators regarded driving a tractor similar to driving a car, in that they can listen to the radio while working and it gives them "style."
Operators earn a high income generally 15-20% of the plowing rate. In addition, the landowner provides snacks and cigarettes to encourage good work. If 30 ha are cultivated, at $40/ha, the operator's share would be $180-240 for the 60-day land preparation season. This is a daily wage of $3-4 with meals, compared to a bricklayer's or carpenter's $1.20-2.40/day without meals.
Custom activities
The land preparation charge is established jointly by the tractor owners and the Extension Service each year, although some owners charge less than the agreed rates to increase demand. The average annual increase in custom rates between 1975 and 1979 has been 33%. About 86% of the area plowed has been custom work for local farmers but some owners traveled as far as 50 km.
Data collected from each minitractor owner were costs and returns for each season he operated the tractor.
In November 1978, the Indonesian rupiah was devalued from Rp 425 to Rp 625 to the dollar. To simplify interpretation, all results are presented in dollars using the current exchange rate of $1 = Rp 625.
Table 12. Economic analysis of minitractors by year of purchase.
Item 1975 1976 1977 1978 1979
Av area plowed (ha/yr) 58.9 63.8 53.5 61.2 57.2
Capital investment ($) 3040 3360 3840 4640 5920
Average fixed costs
Depreciation ($) 547 605 691 828 1066
Interest ($) 219 282 369 497 710
Total ($) 766 887 1060 1325 1776
Average variable costs $/ha 7.56 7.96 9.89 10.72 11.57
Average gross revenue $/ha 19.24 23.27 26.66 31.02 38.35
Break-even area (ha/yr) 65.6 57.9 63.2 65.3 66.3
Annual cash flows ($)a
1974 -3040
1975 860 -3360
1976 758 1289 -3840
1977 761 -1051 1097 -4640
1978 595 961 936 1463 -5920
1979 595 1547 2425 3991 6386
Benefit-cost ratiob b 0.91 1.06 0.92 0.98 0.97
Net present value ($)' -412 296 -388 -109 -218
a For computational purposes, capital investments are assumed to be made at the end of year 0, that is the year before the tractor was actually purchased. Estimated using a discount rate of 12%.

For breakeven point analysis, annual fixed costs were computed as depreciation plus interest. Annual depreciation was calculated as initial cost less salvage value divided by the useful life of 5 years. Salvage value was assumed to be 10% of the initial cash price. Interest was calculated at 12%/year on the declining unpaid balance and averaged over the 5-year life. Because the price of minitractors increased each year, total fixed costs have also risen for tractors bought each successive year.
Variable costs included all expenditures on fuel, oil, repair and maintenance, and the driver's salary. These costs increased each year for all tractors, partly due to increases in the price of inputs and increases in the repair and maintenance component as tractors aged. In 1979, the variable costs per hectare for new tractors were 30% less than for tractors purchased in 1975.
Gross revenue was estimated as the custom operation rate multiplied by the total area plowed (custom work and own land). This formula assumes that the shadow price of own land preparation is the same as the cost for custom operation. Although gross revenue per hectare increased each year as the contract rate rose, gross revenue per season declined due to falling utilization as the tractor aged. Gross revenue and variable costs were converted to a per-hectare basis by dividing the total value over the years of use by the total area prepared. With these data, it was possible to compute the average break-even area to be prepared per annumn over tractor life and compare this with actual tractor performance. Benefit-cost (B:C) ratios for each purchase year could also be estimated by discounting the cash flows.
Values used for estimation and results of the break-even and B: R analyses are shown in Table 12. In only one case (tractors purchased in 1976) was the average annual use larger than the break-even use. Similarly, 1976 was the only year of purchase for which (using a discount rate of 12%) the B: C was greater than unity and the net present value (NPV) was positive. Increasing the discount rate to 15% did not alter these conclusions. However, in each of the other years, the B:C was close to unity.
When estimating these results, all capital investments were assumed to be made at the end of the year before operation started. Although this makes no difference if the tractor was actually purchased at the beginning of the wet season, it will underestimate the B:C and NPV for later purchases. Computations were based on cost and return data for all tractors operating during the respective years. Yet, 1978 tractors did not operate in 3% of the season following purchase, 1977 models in 7%, 1976 models in 10%, and 1975 models in 20% (Table 6). Consequently, if seasons during which tractors were inoperable were taken into consideration, the B:C and NPV would be lower than in Table 12. In addition, the computation of B:C and NPV was influenced by the fact that only those tractors purchased in 1975 had been fully depreciated. The salvage values used for tractors purchased in other years were the depreciated values.
Minitractors have been used in Sidrap and Pinrang Districts for several years. Evaluation of data collected from 10 owners who purchased minitractors from 1975 through 1979 showed that, even when purchased with subsidized credit, these

tractors were not economically viable. Two problems appeared responsible for this situation. First, breakdowns reduced the available working time during the season and the useful life to less than the 6 years assumed in the loan repayment schedule. Second, as the number of tractors introduced increased use declined.
The low man-to-land ratio suggests a shortage of labor (human and animal) for land preparation in Sidrap and Pinrang. And because minitractors did not displace hired labor (Bernsten 198 1) there were no negative social welfare impacts. Yet, unless the life of the tractors can be extended to at least 5 full years, and their number held down to a level allowing for sufficient use of each, tractors will not be an economically feasible alternative to existing techniques.
Abbas, S. 1977. Perspective of agricultural mechanization in Indonesia and South Sulawesi.
Ujung Pandang.
Agricultural Extension Service. 1980. Internal data on tractor sales in the province. Ujung
Pandang, South Sulawesi.
Agricultural Extension Service and Association of Devotees of Agricultural Mechanization
in Indonesia. 1976. Feasibility study of tractor development in Barru, Enrekang, and
Tona Toraja Districts, South Sulawesi. Ujung Pandang.
Bemsten, R. H. 1981. Effect of minitractor mechanization on employment and labor use
intensity, Sidrap and Pinrang Districts, South Sulawesi, Indonesia. The Consequences
of Small Rice Farm Mechanization Project Working Paper 32. Los Bafios, Laguna. B. P. S. (Biro Pusat Statistik). 1980. Statistical handbook of Indonesia. Jakarta. Consequences Team. 1981. Consequences of land preparation mechanization in Indonesia:
South Sulawesi and West Java. Paper presented at the Regional Seminar on Appropriate Mechanization for Rural Development, Jakarta, Indonesia.
Directorate for Technical Agriculture. 1979. Survey for the possibility of using small tractors
in Sidrap and Pinrang Districts, South Sulawesi. Jakarta.
Directorate of Food Crops. 1975. Tractor development study in Gowa, Maros, Pangkep,
Polmas, and Soppeng Districts, South Sulawesi. Jakarta.
Gadjah Mada University. 1976. Survey for agricultural tractor development at Bantaeng,
Bulukumba, Sinjai, Wajo, Bone-Bone, Tekalar, Luwu Districts, South Sulawesi.
Faculty of Agriculture, Yogyakarta.
Parusengi, D. 1972. Possibility of using small tractor for people's farming in South Sulawesi.
Fatometa IPB, Bogor, Indonesia.
Team on Selective Mechanization. 1979. Owning and operating a tractor in South Sulawesi
(manual). Ujung Pandang.