ESSAYS ON SKILL AND WAGE DISTRIBUTION By ACHALA ACHARYA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2014
2014 Achala Acharya
To my parents
4 ACKNOWLEDGMENTS This dissertation would not have been possible without the help of man y individuals I am thankful to Dr. David Denslow for int roducing me to urban economics. I am grateful to Dr. David Denslow and Dr. Steven Slutsky for their guidance throughout the entire process of writing this dissertation. I would like to thank Dr. Larry Kenny and Benjamin Smith for serving on my committee and providing great feedback on my work. I would like to thank Katsufumi Fukuda and Elias Dinopoulos for their helpful comments. I am incredibly thankful t o the great friends that I have made during my stay here in Gainesville. Amanda Phalin, Lindsey Woodworth and Fan Li have made my stay at the University of Florida memorable and supported me like family. Special thanks to Ro sha Pokharel for welcoming me at her home during my visits to Gainesville and being a great friend. I would also like to thank Martha Shaw and Shawn Lee for taking care of all the paperwork and motivating me to move forward. I am grateful to my parents Keshav Acharya and Lalita Acharya for supporting me in all my endeavors. I would li ke to thank my brother Shikhar Acharya and sister in law Shristy Acharya for their encouragement, and my niece Juneli Acharya for bringing much joy and happiness in my life. Finally, I would like to thank my husband Nirmal for his unconditional love and support.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 LIST OF ABBREV IATIONS ................................ ................................ ............................. 9 CHAPTER 1 INTERNATIONAL TRADE AND THE POLARIZATION OF WAGE EARNINGS ..... 11 Introduction ................................ ................................ ................................ ............. 11 The Model ................................ ................................ ................................ ............... 14 Educational Levels ................................ ................................ ........................... 15 Factor Endowment in Steady State ................................ ................................ .. 19 Preferences ................................ ................................ ................................ ...... 21 Equilibrium ................................ ................................ ................................ .............. 21 Ability Cut Off Conditions ................................ ................................ ................. 23 Full Employment Conditions in S ectors X, Y and Z ................................ .......... 24 International Trade ................................ ................................ ................................ .. 25 Conclusion ................................ ................................ ................................ .............. 28 2 WAGE AND SKILL I NDEX FOR U.S. CITIES ................................ ......................... 31 Introduction ................................ ................................ ................................ ............. 31 Literature Review ................................ ................................ ................................ .... 32 Data and Methodology ................................ ................................ ............................ 37 Methodol ogy ................................ ................................ ................................ ..... 38 Skill Index ................................ ................................ ................................ ......... 39 Occupational Licensing ................................ ................................ ........................... 44 Data and Methodology ................................ ................................ ..................... 48 Results ................................ ................................ ................................ ............. 49 Conclusion ................................ ................................ ................................ .............. 50 3 APPLICATION OF THE WAGE INDEX ................................ ................................ .. 72 Medicare Wage Index ................................ ................................ ............................. 72 Allocati on of Medicare Funds Based on OES Index ................................ ............... 76 Conclusion ................................ ................................ ................................ .............. 79 APPENDIX A DERIVATION OF EQUILIBRIUM WAGE OF LOW SKILLED WORKERS ( ...... 92
6 B DERIVATION OF PROPOSTION 2 ................................ ................................ ........ 93 LIST OF REFERENCES ................................ ................................ ............................... 94 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 98
7 LIST OF TABLES Table page 2 1 Definition of skills ................................ ................................ ............................... 53 2 2 Return to skills ................................ ................................ ................................ ... 54 2 3 Occupations with highest skill indexes ................................ .............................. 55 2 4 Occupations with the lowest skill indexes ................................ .......................... 56 2 5 Cities with highest and lowest skill indexes ................................ ....................... 57 2 6 Wage i ndex ................................ ................................ ................................ ........ 58 2 7 Wage index and population ................................ ................................ ............... 68 2 8 Wage index on housing price and amenities ................................ ..................... 69 2 9 Specification 1 occupational licensing ................................ ............................... 70 2 10 Specification 2 occupational licensing ................................ ............................... 70 2 11 Specification 3 occupational licensing ................................ ............................... 71 3 1 CMS OES regression ................................ ................................ ........................ 80 3 2 Comparison to CMS wage index. ................................ ................................ ...... 81 3 3 Reallocation of Medicare hospitals payments using OES index ........................ 81
8 LIST OF FIGURES Figure page 1 1 Ability and wage earnings ................................ ................................ .................. 30 3 1 Comparison to CMS wage index ................................ ................................ ....... 91 3 2 Distribution of the ratio of CMS index to OES index ................................ .......... 91
9 LIST OF ABBREVIATIONS CLEAR FHFA MSA OES O*NET SOC Council on Licensure, Enforcement and Regulation Federal Housing Finance Agency Metropolitan Statistical Area Occupational Employment Statistic Occupational Information Network Standard Occupational Classification
10 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ESSAYS ON SKILL AND WAGE DISTRIBUTION By Achala Acharya May 2014 Chair: Steven M. Slutsky Major: Economics This dissertation examines the pattern of wage and skill d istribution from the early 1980 s and explores possible causes that might be contributing to the observed pattern of wages and skill d istribution. In chapter one I introduce a theoretical model that explains how international trade is contributing to the polarization of income and employment. In chapter two I investigate the distribution of wage and skills across US cities using data fro m Occupational Employment Statistics and the Occupational Information Network. the wage index I first develop a skill index for each city. This skill index is not dependent upon educational ac hievement and is computed using actual measures of skills required for occupations. In chapter three, I compare the wage index creates using OES data to the area wage index developed by Centers for Medicare and Medicaid Services and find that at higher wage levels the two w age indices show a greater degree of divergence. In addition, I also find that an improvement in the quality of healthcare services would close the gap between the OE S and CMS wage indices.
11 CHAPTER 1 INTERNATIONAL TRADE AND THE POLARIZATION OF WAGE EARNINGS In t his chapter I develop a model of international trade and human capital acquisition with three factors of production; low skilled workers, medium skilled workers and high skilled workers. In addition a low skilled intensive non traded good is introduced to analyze the effects of trade on wage earnings of the three categories of workers. The results show that for a small open economy which is relatively high skill abundant, an increase in the price of the high skill intensive good causes the wage earnings of the high skilled workers to increase; the wage earnings of medium skilled workers to decrease and the wage earnings of low skilled workers to increase leading to a polarization of wage earnings in t he labor market. A similar polarizing effect on the shares of employment also results with increases in the employment shares of low skilled and high skilled workers and a decline in the employment share of medium skilled workers. Introduction The rising t rend of wage inequality since the 70s and the associated changes in employment has been the subject of great interest and in the last two decades a large volume of literature, both empiri cal and theoretical, on the employment and wage patterns has emerged. One of the major contributors to the rising inequality has been identified be to skill biased technical change (SBTC) 1 but recently international trade and offshoring has been believed to play a significant role In this context, this paper 1 Epifani P. and Gancia G.A., (2008),
12 attempts to h ighlight a mechanism through international trade that is contributing to the polarization of earnings and employment in the labor market. Autor (2010) conducted an empirical analysis on the structure of job opportunities in the United States over the last two decades. The study shows that employment in the US has been polarizing with growth in employment opportunities in high skill, high wage occupations and low skill, low wage occupations, coupled with contracting opportunities in medium wage, medium skill white collar and blue collar jobs. Employment and earnings are rising in both high education professional, technical, and managerial occupations and, since the late 1980s, in low education food service, personal care, and protective service occupations. Conversely, job opportunities are declining in both medium skill, white collar clerical, administrative, and sales occupations and in medium skill, blue collar production, craft, and operative occupations. Autor (2010) discusses two major factors that may middle of the skill distribution that are repetitive in nature and can be performed with a set of instructions are being replaced by comput ers as the price of computers are on a declining trend. The second potential c ause discussed in Autor(2010) is international trade and o ffshoring of goods and services With declining transportation and communication costs tasks that are repetitive in natu re such as bill processing, data entry and tax preparation are being increasingly offshored to countries with lower wages. Job Polarization is a phenomenon occurring not only in t he United States but is pervasive across Europe as well. Goos, Manning and S alommons (2010) show that the employment structure of 16 European countries has been polarizing in recent years with
13 both wages and employment increasing at the lower and upper ends of the income distribution and declining in the middle which has been term This chapt er builds on the Findlay and Kierzkowski (1983) and Borso ok (1987) papers which added human capital acquisition to the theory of international trade. Both of these papers make the assumption that the economy is endowed with a fixed stock of educational capital. In t he FK paper the workers are identical with regards to their ability. However, the productivity of those t hat choose to become skilled is positively related to the capital/student ratio at the t ime they are educated. But in a steady state all skilled workers are identical. Borsook assumes an exogenous distribution of individual ability. In the Borsook model the amount of schooling undertaken by individuals is related to their inherent abil ity. The length of time spent in school is the same for everyone but more able s tudents receive more education, bec ause the optimal capital/student ratio is increasing in ability. In this chapter there are assumed to be two levels of education and the dec ision to acquire human capital is based on the ability of workers and on the relative wages. Falvey et al (2010) construct a two factor (skilled and unskilled labor), two sector Heckscher Ohlin trade model with an education sector in which uses skilled lab or and time are used to convert unskilled workers into skilled workers This model is used to explain the short run effects on labor supply of anticipated and unanticipated trade liberalization. In the ir paper, they show that the decision to upgrade to bec ome skilled depends on the age of the individual and that younger workers benefit from liberalization while the older workers are at a loss.
14 In this chapter, I introduce another education sector to the Falvey model where the low skilled can become medium s killed and the medium skilled can become high skilled. In addition there is the introduction of a non traded sector. The low skilled workers are used only in the production of a non traded good. The remaining two goods are traded and follow the Heckscher Ohlin structure. There are two educational sectors for training low skilled to become medium skilled and medium skilled to become high skilled. Instead of a having a fixed amount of exogenously given educational capital, medium skilled workers train low sk illed workers and high skilled workers train medium school workers. This framework is used to show that there is polarization of wages and employment shares following a terms of trade improvement through a rise in the price of the high skilled intensive go od. The Model The economy produces three goods X, Y and Z. There are three factors of production ; lo w skilled labor, medium skilled labor and h igh skilled labor. Let denote the wage for low skilled labor, let denote the wage of a medium skilled worker with and denote the wage of a high skilled worker. Sector X is non trad e d and uses only low skilled labor for its production 2 It is assumed that one efficiency unit of low skilled worker can produce one unit of good X. Hence, the zero profit condition in this sector is given by ( 1 1) 2 This classification is based on Autor, Katz and Kearney (200 0 ) where low skill work requires hand eye coordination and has to be performed on site Goos and Manning (2003) identify low skill occupation such as fast food service, waiting, retail cash register attendant, c leaners etc. as services that fall under that category. Mazzalori and Ragusa (2010) find that the least skilled workforce in the US is disproportionately em ployed relative to skilled workers in non traded services like food preparation, cleaning, repair and delivery.
15 Sector s Y and Z need both medium skilled and high skilled labor for its production. Goods Y and Z are traded internationally. Goods Y and Z are produced with constant returns to scale cost fu nct ions that take the Cobb Douglas f orm. The t otal cost function for Y is given by ( 1 2) Similarly the total cost of producing good Z is given by ( 1 3) The zero profit conditions for goods Y and Z are given by the equations below. This model is of a small open e conomy so the prices of traded goods Y and Z are exogenously set in the world market Good Y is the numeraire in this model so its price has been set equal to unity. ( 1 4) ( 1 5) Educational Levels There are t wo education sectors. The medium education sector trains low skilled individuals to become medium skilled and the highe r education sector trains medium skilled individuals to become high skilled workers It ta kes units of time for a low skilled worker to become medium skilled and units of time for a medium skilled worker to become high skilled. Medium skilled workers train low skilled workers and high skilled workers train medium skilled workers. is the u nit labor requirement to train one low skilled worker to medium skilled labor per time period Therefore the co st to a low skilled worker per time period to become a medium skilled worker is Similarly the cost to a medium skilled worker to become a high skilled worker is .
16 The total number of wo rkers does not change over time. The supply of labor into each of the sectors is determined endogenously based on the level of skills acquired by the workers. The workers are heterogeneous with respect to ability denoted by F ( ) =1 Ability ( is distributed uniformly along the interval Each individual has an exogenously given working lifespan of T. I t takes time periods of training for a low skilled individual to become medium skilled. Unlike in the Falvey model where workers can decide to become skilled at different points of time, workers make their decision to gain training at the beginning of their working life A low skilled worker will enter the medium skill training at time 0 and a medium skilled worker will enter the high skill training at time right after completing medium skilled training Therefore the net benefits discounted to the present, to an individual of upgrading at the current time 0 from low skilled to medium skilled is given by ( 1 6) The first term in the integral (1 6) ( is the wage earnings per time period after upgrading and the second term is what the worker would have earned without upgrading. The total cost of acquiring medium level skills discounted to time 0 is given by ( 1 7) The first term in the integral ( is the direct cost paid to the trainers whereas is the opportunity cost of foregone wages. There is a critical level of ability beyond which it is profitable to invest in becoming a medium skilled worker. The critical level of ability ( is found by equating
17 the costs and benefits associated with becoming a medium skilled worker. The workers will start working without acquiring any training. For workers whose ability is gr eater than ( 1 8) Solving the above equation for yields the solution: where ( 1 9) is positive as long as and T > T is total working life of an individual and it is assumed to be longer than the time it takes for an individual to acquire medium skilled training and is assumed to be positive. Similarly in the high skill educational sector an individual who has completed medium skilled training will decide to become high skilled if the net earnings from being high skilled exceeds the costs. This decision is made at time as soon as the worker completes medium skilled training. ( 1 10) is the number of time periods the worker has to remain in the education sector to complete the training. is the time spent on training to become medium skilled. The left side of the inequality is the total cost to become high skilled and the expression on the right side of the inequality is the net benefits to the individual discounted to time from being high skilled. We can solve for another cutoff ability level above which it is profitable to upg rade to become high skilled. where ( 1 11)
18 is positive if and T > is the time it takes a low skilled worker to become high skilled and I assume that the wor king life T is greater than the time spent in training to become high skilled. is the time it takes a medium skilled worker to become high skilled and is positive by definition. An important point to note is that, i n order for the equilibrium to exist i t must be that If this condition does not hold then it is possible for low skilled workers to directly become high skilled without becoming medium skilled. However, the production of the goods Y and Z require both medium skilled and high skilled workers. In the absence of medium skilled workers there would be no production of either good Y or Z. To rule out that possibility, I assume The low skilled workers only earn the wage whereas for the medium skilled and high skilled workers the wages are proportion ate to their ability level 3 Therefore medium skilled workers earn and the high skilled workers earn The wage earnings for the three types of workers is shown i n Figure 1 Individuals with ability below choose to remain unskilled and all the individuals in this ability range earn wages equal to Individuals in the ability range [ will undergo training to become medium skilled and their e arnings are proportional to their ability. There is a discontinuity at because the net benefits from upgrading ( equals zero even after accounting for the cost of upgrading. This implies that is positive which is the wage premi um accruing to a worker as a result of acquiring 3 This is consistent with the empirical evidence from Lemieux, MacLeod and Parent (2009) where it is showed that there is wage disparity among people with the same level of education and experience due to the ability of employers to pay according to their performance. Since performance is a function of education and ability I have assumed that ability is observable.
19 more human capital. The discontinuity at is also arises due to the fact that workers in the ability range [ earn a wage premium for undergoing the training to become high skilled. When solving for t he equilibrium wages it is shown that .T herefore, the earnings slope in this range is steeper than in the ability range [ Factor Endowment in Steady State As in Falvey et al (2010) at each point in time an exogenous number of agents (n) are born and die and each agent is replaced by another with the same level of ability. The total workforce at any given time is equal to Tn =N. In the steady state agents constitute the supply of low skilled labor working in sector X. In the abil ity range [ there are three categories of workers. The first group works to produce goods Y and Z, the second group is in training to become medium skilled and the third group consists of the teachers who are training the low skilled workers to b ecome medium skilled. The number of working in the production of goods Y and Z and teachers working in the medium education sector is given by the following equation ( ( 1 12) The right hand side of this equation is the fraction of workers who have already completed the training to become medium skilled in the ability range The number of agents in the ability range who are training to become medium skilled is given by ( ( 1 13)
20 This is the difference between all the workers in the ability range ( and those that are working as in the production of goods Y and Z and in the medium skill education sector. is the unit labor requirement per time period to train one low skilled worker to become medium skilled .Therefore the number of agents who are training the low skilled to medium skilled is given by ( 1 14) The total number of medium skilled agents working in sectors Y and Z is given by ( 1 15) This is the difference between equations ( 1 12 ) and ( 1 13 ) Similarly for the agents in the ability range [ the supply of workers in production of Y and Z and in the education sector is given by ( 1 16) This is the fraction of workers in the ability range who have already completed the training to become high skilled. The number of students training to become high skilled is given by ( 1 17) Therefore the number of agents working in the high skill education secto r is given by times the number of students ( 1 18)
21 And the total number of high skilled agents working in sectors Y and Z is then given by the difference between equations (1 16) and (1 17) ( 1 19) Preferences The utility of the consumers is represented by the Cobb Douglas function. ( 1 20) To find the demand function for each good, workers maximize utility subject to the budget constraint Max s.t. I = Maximizing the utility function subject to the income constraint yields the demand for the non traded good X ( 1 21) Equilibrium There are nine unknowns in this model which are the wages of the three categories of labor ( the quantity of each of the goods produced (X, Y and Z) the ability cutoff for medium skilled and high skilled workers and the price ( of the non traded good X ( Zero p rofit Conditions fo r Goods X, Y and Z are listed below. These expressions are reproduced below from equation (1 1), (1 4) and (1 15) respectively.
22 The equilibrium values of can be found b y solving the zero profit conditions for good Y (equation 1 4) and good Z (equation 1 5). ( 1 22 ) ( 1 23 ) Proposition 1 : if either and or and I assume that good Z is the high skill intensive good such that and the price of high skil intensive good is greater than the price of medium skill intensive good. Proof: From equations ( 1 2 2 ) and ( 1 2 3 ) if then Then, (1 22) and (1 23) imply that if To solve for the equilibri um price and quantity of good X, the demand for good X must equal the supply of good X because this good is not traded internationally. The demand for good X is given by equation (1 21) Assuming full employment the supply of good X is equal to the supply of low skilled workers since one low skilled worker produces one unit of good X. ( 1 24 ) ( 1 25 ) Income is the sum of earnings of all workers where is the average ability of the m edium skilled workers and is the average ability of high skilled workers. The average ability is the midpoint for each category of labor since ability follows a uniform distribution.
23 I = + ( 1 2 6 ) Substituting and in equation (1 27) I = ( 1 2 7 ) Rearranging equation (1 27), I = + ( 1 2 8 ) Substituting income function into equation ( 1 25 ) yields the following, ( 1 29 ) Solving equation (1 29) for gives the equilibrium wage for the low skilled workers. The detailed derivation of this expre ssion can be found in Appendix A (1 30 ) Where a = b = < 0, and c = > 0 Ability Cut Off Conditions The equilibrium values for the ability cut off conditions are listed below. These equations are derived by equating the costs and benefits associated with acquiring medium skill and high skill education. The ability cut off conditions are reproduced below from equations (1 9) and (1 11) respectively.
24 where where Full Employment Conditions in S ectors X, Y and Z The full employment condition in the good X sector is that the number of low skilled workers exactly equals the amount of good X produced in the economy. X is the equilibrium output of good X and is the number of low skilled workers. ( 1 31 ) Equations (1 32) and (1 33 ) are the full employment conditions in sectors Y and Z. The left hand side of equation (1 32) is the number of medium skilled workers required to produce one unit of good Y multiplied by the total output of Y added to the number of medium skilled workers required to produce one unit of good Z multiplied by the total output of good Z. The ful l employment condition is that the medium skilled workers in these two sectors equals the total supply of middle skilled workers minus those who are trainers. ) ( 1 3 2 ) Equation (1 33 ) is the full employment condition for high skilled workers. The left hand side of equation (1 33 ) is the total number of high skilled workers employed in the production of goods Y and Z The total number of high skilled workers in sectors Y and Z is set equal to the total number of high skilled workers in the economy who are high skilled minus those high skilled workers who are trainers. ( 1 3 3 )
25 International Trade In this section I will analyze the effects of an increase in the price of good (Z) on the wages of the workers. In the Hecksher Ohlin model of international trade, an increase in the price of the good that a country has a comparative advantage in, is used to represent tr ade. I assume that the US exports high skil l intensive goods. Th e result derived I in Proposition 1 is an application of the Stolper Samuelson theorem in international trade which describes the relationship between the relative output prices and relative f a good will increase the real return to the factor used intensively in that good, and reduce the real return to the other fact or Proposition 2 : An increase in t he price of good Z the high skill intensive good raises the wages of the high skilled worker; lowers the wages of the medium skilled worker and raises the wages of the low skilled worker leading to a polarization in the wage earnings. Proof: The effect o f an increase in price of good Z on the wages of the medium skilled worker is given by the derivative in equation (1 22) < 0 ( 1 34 ) Since this expression is negative. Similarly the effect of an increase in price of good Z on the wages of high skilled worker is given by the derivative in equation (1 23)
26 = ( 1 35 ) T his expression is positive because g ood Z is the high skill labor intensive good ( To analyze the effect of the increase of price of good Z on the wages of the low skilled worker I differentiate the equilibrium solution for in equation with respect to This is done in a number of steps. The values of a, b and c are the same as in equation (1 30) when solving for the equilibrium 4 Using the quotient rule the derivative of with respect to is given by ( 1 36 ) This results in the overall derivative having a positive sign. This result implies that an increase in the price of the high skill intensive good raises the wage of the low skilled workers. The utility function was Cobb Douglas so there are no cross price effects and this mechanism works strictly through the income effect. Increas ing the price of the high skilled intensive good raises the wages of a high skilled worker who demand s more of the non traded goods provided by the low skilled workers and thus results in an increase in the low skilled wage. Proposition 3 : An increase in the price of the high skill intensive good Z raises the supply of high skilled workers, lowers the supply of medium skilled workers and increases the supply of low skilled workers measured in efficiency units. 4 The detailed derivation of this derivation is in Appendix B
27 The effect of the terms of trade improvement on the supply of labor is determined by the effect on the cutoff ability levels and respectively. The signs of the partial derivatives and are calculated which will then be used to compute the resulting changes in labor supply. >0 ( 1 37 ) That t he derivative of with respect to is positive implies that the ability cut off for medium skilled workers increases when the price of high skilled good increases This increases the s upply of low skilled workers as more workers will choose to remain low skilled. From Proposition 2 we know that and >0 Hence, ( 1 38 ) Since and there is a reduction in the ability range [ which is the ability range for medium skilled workers. This leads to a decline in the supply of medium skilled workers. The ability range gets larger leading to a higher supply of high skilled workers and an increase in the employment share of high skilled workers. At the other end, the ability range [0, gets bigger due to the increase in This will lead to an increase in the supply of low skilled workers because some fraction of low skilled workers who would otherwise have become medium skilled do not stand to gain by becoming medium skilled. This is the skilled workers as reported in the empirical literature.
28 Conclusion This paper has developed an international trade based explanation behind employment and earnings polarization through the introduction of a low skilled labor intensive non traded good and three factors o f production: low skilled work ers, medium skilled workers and high skilled workers. The argument presented in this paper is not an alternative explanation to the skill biased technological change explanation for income and employment polarization instead it s upports another force may a lso be at work in addition to it. The mechanism at work here is that if there is an increase in the price of the high skill intensive good, the aggregate income of the economy increases which raises the demand for the non traded low skilled intensive good thus raising the wages of the low skilled worker. The other result on polarization of employment shares follows from the fact that the decision to enter the educational sector is dependent upon the relative wage earnings. As the wage earnings of high skil led workers increases relative to medium skilled workers more people find it profitable to invest in high skill education sector and as the wage earnings of the medium skilled workers declines relative to low skilled wage earnings fewer people choose the e nter the medium skill education sector. I have derived the result of wage and employment polarization under the restrictive assumption of Cobb Douglas preferences However, it is important to note that the result is not dependent on the utility function an d as will hold as long as all three goods are normal goods. The choice of utility function was based solely on the keeping tractable. In this chapter I examined how trade in goods can lead to polarization. In my future work I would like to study how trade in tasks is affecting the labor market. I would like to be able to estimate the magnitude of the changes in wages for all three groups of
29 workers. In this chapter I am only able to say that the high skilled and low skilled workers would receive higher wag es while the middle skill workers would receive lower wages. I would also like to consider a model where the ability of workers depends on the quality of education they receive at the middle skill level. It might be the case that who receive better quality education at the middle level will eventually have the ability become high skilled compared to those who receive a low quality education at the middle skill level.
30 Figure 1 1. Ability and wage e arnings
31 CHAPTER 2 WAGE AND SKILL INDEX FOR U.S. CITIES This chapter develops an annual wage index for over 300 metropolitan areas in the US for the years 1997 2012. The index was created using data from Occupational Employment Statistics, Federal Housing Finance A gency and County and City Data Book 2007. It is shown that after controlling for detailed occupation status variation in housing price and amenities explains a large part of the variation in wages across cities. Introduction The compensation paid to worke rs depend s not only on their education, skills and experience but also on the cost of living in the city where the work is performed. Similar workers living in different cities earn different wages even after controlling for their observed skills and exper ience. Many urban economists and regional scientists, including Graves (1976), Rosen(1979 ) Roback (1982) and most recently Glaeser (2008) and Albuoy (2009) argue that urban wage levels, housing prices, and populations are simultaneously determined by deci sions of firms to hire workers, Workers can relocate easily, so the utility of similar workers will tend toward equalization across urban areas. In these models the utility is composed not only of nominal (or gross) wages but also spending on housing and the availability of local amenities. In such a spatial equilibrium a worker cannot be made better off by moving from one city to another. The insight from these models in relation to wages is that similar workers will earn different wages in different cities based on the cost of housing and the availability of amenities in the city that they choose to locate in.
32 In this chapter I wil l develop a wage index that measures the wage differential across U.S. cities at an annual frequency starting in 2000. There have been many empirical studies that have estimated the wage differential across cities. However, most of the indices that measure wage differentials have be en constructed only at one instance in time. A brief discussion of some of the empirical papers that have estimated wage differentials is included in the next section of this chapter. The advantage of the index that I develop in this chapter is that it wil l also provide information about changes in wage differe ntials over time. An index that measures wage differential across cities at an annual frequency would be useful to study different aspects of spatial equilibrium. In this paper an annual measure of th e wage differential across 282 US metropolitan statistical areas is constructed for the years 1997 2012. The wage index is computed using wage data from Occupational Employment Statistics (OES) which is available annually starting in 1997. The advantage of using this dataset is that it is available annually and it is payroll data provided by employers as opposed to self reported surveys about wages and other compensation. In addition to the OES data, I will also use data from O*NET (Occupational Information Network) to estimate the skill measures for different occupations. Literature Review There are two broad categories of studies that have measured the wage differential across different regions in the U.S. The first category of papers measure the cost dif ferences across school districts and hospital areas which are used for school and hospital budgets. Taylor (2005) has created a Comparable Wage Index (CWI) which measures regional variations in the salaries of non educators for years 1997 2003. Lori Taylor uses Census Data to create a baseline index by regressing log wages
33 on demograp hic characteristics, occupation, industry and place of work. The analysis is restricted to workers with a college degree. Taylor (2005) uses OES data for the years the Census d ata is not available assuming that more detailed occupational information compensates for the absence of demographic variables. The wage index that I develop in this chapter differs from the CWI in three ways. First, I control for the job skills for the wo rkers in a different way Taylor ( 2005) uses a number of demographic variables such as race, gender, and educational attainment and occupational dummies. She includes four categories to measure educational attainment professional degree, degree, degree and Doctorate degree. In the wage index that is developed in this chapter the job skill is measured with data on skills relevant to a particular occupation. Using numerical data on the level of skill required for different occupations from the O*NET database I create a skill index for each of the 900 occupations in the OES. The skill index is a good indicator of the earning ability of workers as it can capture the variation of sk ills within an education group. Second, Taylor (2005) fir st creates baseline CWI for the year 2000 and then extends it to 1997 2003 using OES data. The index developed in this chapter will span a longer time period starting with 1997 and going up till 2012. The third difference is that while Taylor constructed th e index to be used as a measure for teacher pa using only college graduates excluding teachers, the index that I create is for the purpose of adjusting for regional differences in cost of living for all educational levels The second category of studies h as tested the existence of spatial equilibrium to see if wages are equal across regions after adjusting for housing costs and the amenities. These studies examined the relationship between wages and housing prices,
34 controlling for amenities. According to t hese models once amenities is controlled the wage price elasticity should be equal to 1. Roback (1988) estimated an elasticity of 0.97, Dumond (1999) estimated an elasticity of 0.46. Glaeser and Mare (2001) compare the nominal and real wages of 37 metropol itan areas in 1990.They estimate real wages by dividing average nominal wages by the ACCRA Cost of Living Index, a measure produced by The Council for Community and Economic Research. Glaeser and Mare conclude that firms in large cities pay higher nominal wages, but after accounting for inter metropolitan differences in the cost of living, the observed wage premium in large cities disappears. Winters (2009) used different measures of housing price rents and hous ing value s When house prices were measured by rents the wage price elasticity was almost equal to 1. This paper also contributes to this literature by providing another check on the existence of spatial equilibrium. If wages do tend to equalize across cities after controlling for housing prices and amenities the wage index that I compute should get much smaller after controlling for housing cost and amenities. Next, I will consider some mo dels of wage determination that provide a justification of the method of wage index construction used in this pa per. Wage determination is one of the most widely studied subject s in economics. There are a number of models that explain how wage s are determined for workers. The most widely used models that have led to a large empirical literature on wage determination are the human capital model, the Roy model, the compensating differentials model and the search model. Papers using the human capital model also referred to in the literature as Mincerian earnings equation regress log wages on measures of human capital s uch as education and experience and a number of demographic variables. In the Roy model,
35 workers earn different wages as a result of different levels of skills at the time of labor market entry. The worker is able to earn the highest possible wages by choo sing a job that best compensates the skill sets that the worker has a comparative advantage in. The compensating wage differential is different from the Roy model in that workers with identical skills can earn different wages due to their preference for a certain type or place of work. According to the search model the wage of workers is dependent upon situations beyond their control, like the state of the economy at the time of job entry and other search frictions that is different for different workers. The objective of this paper is not to estimate the effects of variables that affect wages but to use the ideas in these models to develop a measure of wage and skills across space. This measure of wage dispersion across US cities relative to the national a verage will be useful in determining the average compensating wage differentials across cities. In this cha p t er I will be creating a wage index for each city base d on the occupational employment statistics collected by the BLS that provides wage and employ ment data for 300 US metropolitan statistical areas. Most of the studies that measure the skills in cities have used the level of education as a proxy for skill. A number of papers measuring the urban wage premium have resorted to the use of education as a measure of skills 5 In this paper the distribution of skills that is not dependent on educational attainment will be used to create a wage index that measures the compensating differentials that arise due to differences in housing prices and amenities acr oss US metropolitan areas. 5 Some papers using this approach are Glaeser and Mare (2001), Wheeler (2001), Lee (2005), Combes et al (2008), Rosenthal and Strange (2008),
36 The motivation to not use education as a measure of skill comes from the review of literature on returns to education that show wide variations in returns to education ranging from 3.5 20% per year of schooling depending on the data used and the estimation methodology. In recent years there have also been extensions to the human capital model where educational attainment is believed to be a signal to employers due to increase s in jobs that require prob lem solving and decision making employers are not solely relying on education and test scores to screen their employees. Google, an employer known to attract highly talented people has found from its own data analytics division that there is no correlati on between test scores or GPA and job performance. The share of employees who have never been to college is on the rise. 6 Based on th is evidence, my attempt in this paper is to create a skill index for occupations that exhibits significant variation inst ead of the binary classification of high skill jobs that are typically performed by college graduates and low skill jobs that are performed by non college graduates. Autor and Dorn (2013) have argued that such a skill premium throughout the twentieth century but fails to explain the non monotone growth of employment and wages by skill level. In contrast to the high skill and l ow skill categories used extensively in th e literature Autor and Dorn (2013) classify occupations in nature and can be performed following a number of steps and these jobs can be done 6 Based on the interview with Laszlo Bock, senior vice president of people opera tions at Google, that appeared in New York Time, June 19, 2013
37 by computers. The non routine task require more abstract skills such as creativity, problem solving and coordination and other low skill service occupations that have to be performed onsite. Autor, Levy and Murnane (2003) classified occupations into routine and non routine cat egories based on the task component of occupations using data for each occupation using measures of importance and level of skill required. The skill index is exp ected to be increasing in wages so the skills that are used to create it consists of skill sets that are similar to the high and Murnane (2003) paper The goal here is to create a measure of how the wage to skills ratio for a particular city deviates from the national wage to skill ratio. For each occupation a measure of basic skills require d to enter t he occupation provides the basis on which the skill index i s computed. The skill index is created from a composit e of skills which include complex problem solving skills, critical thinking and judgment skills and decision making skills that are required in each occupation. Data and Methodology The data used for this study comes from the Occupational Employment Statis tics (OES). The OES is a Bureau of Labor Statistics (BLS) database which contains wages by occupation for states and metropolitan areas for every year starting in 1997. For each MSA, OES provides aggregate wage by occupation at the 10th, 25th, 50th, 75th a nd 90th percentiles. In the remaining part of the paper the terms city and MSA will refer to the same geographic unit For most occupations both hourly and annual wages are reported. The total number of workers in each occupation is also included. The d ata is collected through a mail in survey of employers in the MSA. Each year, the BLS samples and contacts approximately 400,000 civilian, nonfarm establishments for the
38 OES survey. Every firm in the United States with at least 250 employees is included in the sample with near certainty each year. Smaller firms are sampled every three years. The rate of response to the survey is typically quite high. The response rate has been in the range of 75 85 percent for each year the data has been collected. The majo r advantage of using this data is that it is available annually which facilitates the creation of an annual wage index which was not possible using data from Census years. The data on skill requirement for jobs comes from Occupational Information Network (O*NET) which is online database with very detailed information about skills, education, type of work related to occupations. The website is sponsored by the US Department of Labor/ Employment and Training Administration (USDOL/ETA). The O*NET is a success or to the Dictionary of Occupational Titles (DOT) and was developed as the economy shifted towards information and services from manufacturing. For each occupation the O*NET provides information in five areas : (i) skills and knowledge (ii) abilities inte rest and values (iii) training and level of licensing and expertise needed (iv) the work activities and context, including the physical, social and organizational factors involved in the work and (v) the occupational outlook and pay scale. The data is collected by conducting surveys among workers in the occupation and occupation experts. Methodology The primary intent of creating a wage index for cities is to create an annual measure of the dispersion of wages across cities to get an estimate of the co mpensating wage differentials arising from the differences in prices of housing and the availability of amenities and to look at the trend of this dispersion over time. In order to achieve that, it is necessary to control for the productivity of workers in different cities
39 There is overwhelming theoretical and empirical evidence that since on skill biased technical change, which is defined as a change in production technology that favors high skilled worke rs over low skilled workers by increasing the relative productivity of high skilled workers. This leads to a higher relative demand for high skilled workers thus increasing the price of high skilled labor Autor Katz and Kearney ( 2006) Goos and Manning ( 2009) provide empirical evidence that workers in high skill occupations that require problem solving and decision making skills earn higher wages than those workers in occup ations where analytical skills are not so important. Skill Index For the wage inde x to capture the variation in skills across occupations, I first create a skill index for each occupation listed in Occupational Employment Statistics. In order to create the skill index, I utilize occupational skill data from the O*NET. T he O*Net data base has a value for which is a measure of proficiency of the skill and an which is a measure of important the skill is to the job responsibilities associated with the occupation. This data is available for all the occu pations that are listed in the OES Both the level and importance measures have values ranging from 0 to 100. I will utilize this data in the estimation of the skill index for different occupations. In order to construct the skill index first I need to kn ow which skills are important in predicting wages. To answer that question first I estimate the relationship between wages and different skills. I estimated the returns to skill based on the following model from Welch (1969) :
40 where the are different skills that a worker possesses, and is an idiosyncratic error term. The are the returns to each skill component k. is a base payment that a worker receives regardless of skills. While this model is for an individual worker, I run this regression on the aggregate wage and skill data from OES and O*NET. The equation I estimated was of the form: where, the dependent variable is the log of the m edian hourly log wage for occupation i for the year 2012, is the return to skill k at time t and is the error term. The dependent variable is log wages instead of absolute wages for a more intuitive interpretation of the coefficient. T he skills that were used in the regression were chosen because there was data available for at least 800 occupations and there was good variation in the skills required across occupations. The nine skills used for the regression were active learning, compl ex problem solving, critical thinking, judgment and decision making, learning strategies, negotiation, persuasion, speaking and writing. These nine skills fall under the four broad categories of Basic Skills, Complex Problem Solving Skills, Social Skills a nd Systems Skills. The definition of each skill c ategory is reported in T able 2 1 The definitions in Table2 1 have been compiled from the O*Net database verbatim. Th e result of the regression is reported in Table 2 2 The estimated values for the returns to skill show that critical thinking complex problem solving and judgment and decision making skills are statistically significant at the 5% level in determining wages.
41 The skill index for each occupation was calculated based on the following formula 0.4( ) + 0.133( ) (2 1) i is occupation and complex is the product of level and importance score for complex problem solving skills, cthink is the score for critical thinking skill and decisionmaking is the score for judgment and decision making skills. These skills were statistically significant in the wage regression reported in Table 2 2. Another reason t hese three skills were chosen over others is due to their positive relation with wage and a wide variation across occupations. The occupation with the highest skil l index was Chief Executives with an index of 61.6 and the occupation with the lowest skill index was Order Fillers Wholesale and Retail Clerk with an index of 5.25 Table 2 3 lists 20 occupations with the highest skill index and Table 2 4 lists 20 occupat ions with the lowest skill index. for each city. The OES data provides information on the employment share of each occupation for all cities. The skill index for each city was calculated as the weighted sum of skill index where the weights were equal to the employment share of occupation i in city c. = (2 2) The variation in the skill index for cities arises from the varying employment shares of different occupations. Due to the data, it was not possible to measure skill differences between the same occupations across cities. The data from O*NET is a measure o f skills to be able to enter the profession so this skill index is not an exact for
42 measuring skill differences across cities but it serves the purpose of comparing how it has changes over the years and how much it deviates from other cities. A national skill index was computed using the same methodology as for the city skill index for comparing the skill index of a city relative to the national skill index. = (2 3) The 20 least skilled and most skilled MSA based on this calculation is presented in Table 2 5 Similar to the relative skills, a measure of relative wage for each city was calculated. The relative was the average hourly wage for all occupations in a city c divided by the average hourly wage for all occupations for the entire country. Both of these data were available in the OES. = (2 4) The final step in computing the wage i ndex was comparing the relative wage to the relative skills for each MSA Wage Index = (2 5) The wage index for 404 U.S cities is reported in Table 2 6. The wage index can be interpreted as the wage differential across cities arising from the variation in housing prices and amenities. The skill measure of each occupation was the minimum required skills to work in a particular occupation so from the construct of the data we have ruled out the possibili ty of urban wage premium or productivity gains from agglomeration. Gla e ser agglomeration econ omies it is possible that workers in the same occupation may have different productivity depending on where they are located. The skill measure from the
43 O*Net database does not allow workers of the same occupation to vary in their skill levels across citie s. Due to this reason the resulting wage index only provides an estimate of the compensating differentials due to location of the job. Since the data does not accommodate varying skill level within an occupation across cities the wage index does not accoun t for higher productivity in larger cities due to agglomeration economies. I regressed the wage index against The result of this regression can be found in Table 2 7 This analysis shows that the wage index is not posit ively correlated to the size of the city. The wage index developed in this chapter measures the geographical variation in wages after controlling for wage skill. This wage index is supposed to measure the differences in pay across geographic regions that arise from variation in housing prices and amenities across U.S. cities. To test how the wage index responds to housing cost and amenities, I estimate the following equation: (2 6) This was a cross sectional regression for the year 2012. The housing cost index data for cities from the ACCRA (American Chamber of Commerce Researchers Association) was used for this analysis. The data source for average January temperature, average July temperature, number of cooling days and number of heating days was County and City Data Book: 2007. The data on violent crimes per 100, 000 residents was from collected from the FBI website 7 Although this data is from 2007, all 7 This dataset is available for download at http://www.fbi.gov/about us/cjis/ucr/crime in the u.s/2011/crime in the u.s. 2011/tables/table 6
44 the variables from Country and City Data were related to natural amenit ies which I a ssumed have not changed significantly in five years. The estimated result for equatio n (2 6) is reported in Table 2 8 The variables housing cost, heating degree days and violent crime were statistically significant. As expected, the housing cost index was positively correlated to the wage index. A percent increase in housing cost index increase the wage index by 0.034 percentage points. Of the four variables used to measure differences in natural amenities, the variable heating degree days was positively s ignificantly related to the wage index. The variable violent crime per 100,000 residents was negatively correlated to the wage index. Occupational Licensing The other line of inquiry in this chapter will be to look at the effect of o ccupational licensing on wages. I also test whether occupational licensing has differential effects on workers at differe nt levels of wage distribution. If occupational licensing has significant effect on wages the wage index will not be accurately measuring the differences in wages in arising from differences i n housing prices and amenities. In this section, I will test if occupational licensing has a positive relationship with wages. If it does the wage index will need a correction that adjusts for the rigor of occupational licensing across occupation requires the permission of the government, and the state requires some to earn a license a worker must complete a certain amount of training and education or pass an exam administered by the licensing board for a particular occupation. According to the CLEAR
45 (Council on Licensure, Enforcement and Regulation) currently 18 per cent of the labor force works in licensed occupations with over 800 occupations requiring a license in at least one state. The number of workers in licensed occupations is significantly higher than those affected by minimum wage or unionization which is re spectively around 10 and 12 percent of the labor force. Licensing is believed to protect the consumers by ensuring that the professionals who provide them service are competent and up to date on the development in their respective fields. However, since o ccupational licensing also creates a barrier to entry into those professions, it is expected that workers in licensed occupations earn higher wages than those who work in occupations that do not require licensing. Friedman (1962) argued that the reason for higher wages in licensed occupations was that licensing creates an entry barrier fewer workers enter which enables them to earn wage premiums equivalent to monopoly rents. The other reason is that people who are not able to work in an occupation due to li censing requirements, enter occupations that do not require licensing and exert a downward pressure on the wages in those occupations. There are a number of papers that have examined the empirical relationship between licensing and the quality of service p rovided by people in licensed occupations. However, empirical research on the effect of licensing on wages is not as abundant. Kleiner and Krueger ( 2010) characterize occupational licensing as one of the fastest growing yet least understood labor market in stitutions in the U.S. Some of the more recent ones that I will discuss in this paper are Gittleman and Kleiner (2013), Kleiner and Krueger (2013), Kleiner and Krueger (2010).
46 Kleiner and Krueger (2010) estimated the effect of labor unions and licensing r egulation on wages using Survey data collected by the Gallup Organization from May to August of 2006. The survey had data on demographics, industry, occupation, earnings and education. The response rate for this survey was 37%. Using education and demograp hic variables as controls, both licensing requirements and unions had an effect of 15% and if a worker was both in a union and a licensed occupation the wages could be higher leading the wage premium up to 25%. Kleiner and Kruger (2013) analyze the labor market effects of occupational licensing using data from a telephone survey carried out by Westat. The questions for the survey were developed by Princeton Data Improvement Initiative and the response rate for the survey was 17.9%. Log Wage estimates with demographic variables, controls for education and experience and union membership and a dummy for licensed occupation was estimated. The results were similar to the effect of licensing found in Kleiner and Krueger(2010) with licensed workers earning a wage premium of 17%. Kleiner and Krueger also test to see if the effect on wage is different if the licensing is at the local, state or federal level. The result from this analysis was that wages were higher by 17% if the licensing was at the state level and i f both local and state or state and federal licensing was required the wages were higher by 25%. Gittleman and Kleiner (2013) estimate the effect of licensing on wages using data on licensing statutes and National Longitudinal Survey of Youth from 1979 to 2010. This survey data was linked with occupational licensing data available at CareerOneStop, sponsored by the U.S. Department of Labor. The data used in this study spanned over many years in contrast to the cross sectional data used in most of the other papers on
47 licensing. With the availability of panel data Gittleman and Kleiner were able to use panel methods to estimate the effect of licensure on wages of workers. In contrast to the findings from the previous two papers, Gittleman and Kleiner found tha t the effect of licensure on wages was much smaller than reported in earlier papers. A variety of estimation strategies were employed to arrive at this conclusion. In a baseline cross section model they estimate an effect of 0.281% on wages when licensure requirement was the sole explanatory variable in the log wage regression. When other variables were added to the regression such as years of schooling, potential experience, indicators for union coverage, government employment, self employment part time st atus and sets of dummy variables for major industry the coefficient further declined to 0.123%. The second estimation strategy was to compare the average growth of log wages between different categories of workers. The first category of workers were those who moved out from licensed occupations, the second category was the group that moved into licensed occupations and the third category was the control group who worked in non licensed occupations the entire time. The result from this estimation strategy sh owed that there were no wage premiums for those who worked licensed occupations. And finally with a fixed effects panel, in the specification with the standard controls the effect is 3.8%. When detailed occupational controls were added the coefficient on t he effect was statistically insignificant. In addition to log wages this paper also examines if licensing has any effect on non wage compensation like health insurance and retirement benefits. Consistent with the results for direct wage compensation, Gittl eman and Kleiner do not find any evidence to support the hypothesis
48 that people in licensed occupations are more likely to have health insurance or retirement benefits. Since the results in this analysis show that occupational licensing do not have a sign ificant effect on wages, it is not necessary to correct the wage index for the rigorousness of occupational licensing across different labor market areas. Data and Methodology The number of licensed occupations has been growing over the years and the research on the effect of licensing on wages has been limited. All the studies cited in this paper have utilized survey data to examine the labor market implications of occupational licensing. In this section I will test this hypothesis about licensing le ading to higher wages using data from Occupational Employment Statistics which is payroll data reported by employers. The data on licensing requirements of occupations by state was obtained from CareerOneStop (the same source used in Gittleman and Kleiner (2013) 8 This website lists the occupation license name and licensing agency for each state. The occupations listed are based on the Standard Occupational Classification (SOC) which is the same classification used in the OES data. Using this information, a dummy variable for licensing requirement for each occupation in each MSA (based on the state the MSA is in) was created. The analysis using Occupational Employment data has the added benefit of large sample size and it is different from self reported su rvey data used in other studies since it is payroll data reported by employers. This affords the opportunity to analyze if 8 The data can be accesses at http://www.acinet.org/licensedoccupations/ For this paper the data was retrieved from the website in June 2103.
49 occupational licensing requirements have differential effects on workers at different ends of the income distribution To test the effect of licensure on wages equations of the following form were estimated: = + ( 2 7 ) = + ( 2 8 ) = + ( 2 9 ) In the first specification, I regress the average hourly log wage for occupation i in city c on a dummy variable that takes a value of 1 if that occupation i requires a license in city c, 875 occupational dummy variables and a dummy variable for each of the 400 cities. In the second specification, instead of using the dummy variables f or cities I control for city level variation in wages using the wage index developed in this chapter. In the third specification, I add a variable percent_licensed which is the percentage of workers in city c in licensed occupations. This variable is anoth er test for the effect of licensing on wages to see if wages are higher in cities that have higher employment shares in licensed occupations. Results For each specification 5 different regression results are reported at different levels of the wage distrib ution. Each regression was weighted with weights being equal to the number of emp loyees in occupation i in city c The results from these regressions are reported in tables 2 9, 2 10 and 2 11 The result from the first sp ecification reported in table 2 9 is that licensure has a minimal effect on wages. At each level of the wage distribution the effect of licensure was less than 1.05%. In the specification that includes
50 wage index as a control variable instead of a dummy variable for each city the effect of licensure on wages diminishes further with an effect of 0.03 % on the wage earners at the 75 th percentile. In the third specification that includes the percentage of workers in licensed occupation in each MSA the effect of licensing on wages was around 1%. The effect ranged from 0.8% to 1.05% depending on the level of wage distribution. The percent_licensed variable itself was not significantly correlated to the wage except. F or all three specifications licensing had a bigger effect on wage earners at th e higher end of distribution (median, 75 th percentile and 90 th percentile) compared to the wage earners at the lower end of wage distribution (10 th percentile and 25 th percentile) With the use of Occupational Employment Statistics data the result regardin g the effect of licensure on wages is quite different from the one that has been usually reported in the literature. With the exception of a few studies like Gittleman and Kleiner (2013) most papers estimate the effect of licensing between 15 20 percent. T he lower result from using large sample size data like the OES and longitudinal data used by Gittleman and Kleiner suggest that more research using different data sources and methodologies is necessary before we can conclusively determine the effect of lic ensing on wages. Conclusion In this chapter, I developed a wage index for 300 cities from the year 1997 2012.I tested the effect of occupational licensing on wages to see it influenced the wage index that I created would be influenced by occupational lic ensing laws and found no statistical evidence to support the hypothesis. Finally I compared to the wage index to an existing wage index to analyze how Medicare reimbursement would be affected if the OES index was used. I find that hospitals in large cities would receive a lower payment
51 for services provided to Medicare enrollees if the OES index is used. While there have been wage indices for the Census years, this index contributes to the literature by providing a measure of skill and wage which can be use ful to understand changes across cities or in the same city over time. With the use of Occupational Employment Statistics data, it is possible to construct a wage index that is helpful in learning about annual movement of wage dispersion across U.S. cities Even though the absolute wage index may not be fully accurate, the annual changes constructed from the wage index can be helpful to study housing markets, migration and employment trends over time One of the problems with the OES wage index developed in this chapter is that workers in the same occupation are assumed to have the same skills across all the cities which may not be a realistic assumption. Economies of agglomeration suggest that workers are more productive in larger cities. However, the data o n skills for occupation maintained by O*NET does not allow for workers in the same occupation to have different levels of skills based on the location. In my future research I would like to more closely examine skill biased technical change across US citi es. The wage index would not be based on occupational but it would be based on the cost difference of tasks across us cities. Autor and Dorn (2013) uses classify tasks into categories based on whether s particular task can be automated or not, in order to provide evidence of polarization of the U.S. labor force across cities. Using a similar classification, an index that measures the comparative advantage of different cities in completing a particular category of task can be computed. The wage index calcula ted in this paper using detailed occupations is a good start and the same data can be used to analyze the wage differential across different occupations.
52 The wage index developed in this chapter has opened up different avenues of future research. It would also interesting to study how the returns to different skills has changed over the years using data on skills from the O*NET database. Understanding the pattern of how skills are being rewarded by market over time, might be important in explaining wage pol arization.
53 T able 2 1 Definition of s kills Skill Type Skill Definition Basic Skills Developed capacities that facilitate learning or the more rapid acquisition of knowledge Active Learning Understanding the implications of new information for both current and future problem solving and decision making Critical Thinking Using logic and reasoning to identify the strengths and weaknesses of alternative solutions, conclusions or approaches to problems Leaning Strategies Selecting and using training/instructional methods and procedures appropriate for the situation when learning or teaching new things Speaking Talking to others to convey information effectively Writing Communicating effectively in writing as appropriate for the needs of the audience Complex Problem Solving Skills Developed capacities used to solve novel, ill defined problems in complex, real world settings Complex Problem Solving skills Identifying complex problems and reviewing related information to develop and evaluate options and implement solutions. Social Skills Developed capacities used to work with people and goals Persuasion Persuading others to change their minds or behavior Negotiation Bringing others together and tryin g to reconcile differences Systems Skills Developed capacities used to understand, monitor, and improve socio technical systems Judgment and Decision Making Considering the relative costs and benefits of potential actions to choose the most appropriate one
54 T able 2 2. Return to s kills Log wage active learning 0.0029 (0.007) complex problem solving 0.196** (0.009) critical thinking 0.168** (0.007) judgment and decision making 0.055** (0.006) learning strategies 0.011 (0.006) negotiation 0.004 (0.008) persuasion 0.002 (0.008) speaking 0.008 (0.007) writing 0.001 (0.005) constant 2.110 (0.867) R 2 0.785 Number of Observations 867 Standard errors in bracket ** = statistically significant at 5% level
55 Table 2 3. Occupations with h i ghest skill i ndexes Occupation S kill I ndex Chief Executives 60.54 Judges, Magistrate Judges, and Magistrates 54.96 Surgeons 53.87 Physicists 53.52 Neurologists 52.27 Molecular and Cellular Biologists 52.26 Neuropsychologists and Clinical Neuropsychologists 52.25 Biomedical Engineers 51.54 Biostatisticians 50.87 Emergency Management Directors 50.43 Air Traffic Controllers 49.50 Oral and Maxillofacial Surgeons 49.47 Mathematicians 48.98 Architects, Except Landscape and Naval 48.98 Mining and Geological Engineers, Including Mining Safety Engineers 48.96 Operations Research Analysts 48.75 Internists, General 48.41 Chemical Engineers 48.18 Water/Wastewater Engineers 47.86 Foresters 47.84 Nuclear Medicine Physicians 47.71
56 Table 2 4. Occupation s with the l owest s kill i ndex es O ccupation S kill i ndex Cleaners of Vehicles and Equipment 6.91 Dishwashers 7.38 Graders and Sorters, Agricultural Products 7.98 Order Fillers, Wholesale and Retail Sales 8.04 Shampooers 8.10 Food Preparation Workers 8.30 Models 8.30 Funeral Attendants 8.32 Slaughterers and Meat Packers 8.34 Dining Room and Cafeteria Attendants and Bartender Helpers 8.97 Food Servers, Non restaurant 9.15 Packers and Packagers, Hand 9.30 Locker Room, Coatroom, and Dressing Room Attendants 9.33 Maids and Housekeeping Cleaners 10.28 Amusement and Recreation Attendants 10.44 Combined Food Preparation and Serving Workers, Including Fast Food 10.55 Cooks, Fast Food 10.60 Industrial Truck and Tractor Operators 10.74 Word Processors and Typists 10.81 Meat, Poultry, and Fish Cutters and Trimmers 10.92
57 Table 2 5. Cities with highest and l o west skill i ndex es Lowest skilled c ities Skill i ndex Highest skilled c ities Skill i ndex Wenatchee East Wenatchee, WA 0.53 Baltimore Towson, MD 1.64 San German Cabo Rojo, PR 0.55 Rochester, NY 1.65 Fond du Lac, WI 0.55 Topeka, KS 1.65 Longview, WA 0.57 Boston Cambridge Quincy, MA NECTA Division 1.66 Fajardo, PR 0.60 Worcester, MACT 1.67 Mayaguez, PR 0.60 Bethesda Rockville Frederick, MD Metropolitan Division 1.68 Sandusky, OH 0.60 Lawrence, KS 1.68 Hinesville Fort Stewart, GA 0.60 St. Louis, MOIL 1.71 Yauco, PR 0.62 Louisville Jefferson County, KYIN 1.74 Guayama, PR 0.63 Edison New Brunswick, NJ Metropolitan Division 1.74 Sherman Denison, TX 0.63 Kansas City, MOKS 1.75 Madera Chowchilla, CA 0.64 Haverhill North Andover Amesbury, MANH NECTA Division 1.76 Hanford Corcoran, CA 0.66 Detroit Livonia Dearborn, MI Metropolitan Division 1.78 Palm Coast, FL 0.66 Flint, MI 1.78 Steubenville Weirton, OHWV 0.66 Omaha Council Bluffs, NEIA 1.78 Laredo, TX 0.67 Lafayette, LA 1.80 Kokomo, IN 0.67 Syracuse, NY 1.81 Danville, VA 0.68 Fayetteville, NC 1.82 Lewiston, IDWA 0.68 Manchester, NH 1.82 Lawton, OK 0.69 New York White Plains Wayne, NYNJ Metropolitan Division 1.84
58 Table 2 6 Wage i ndex Area name Relative s kills Relative w age Wage i ndex Fayetteville, NC 1.82 0.88 0.48 Lawrence, KS 1.68 0.82 0.48 San JuanCaguasGuaynabo, PR 1.20 0.60 0.49 Ponce, PR 1.09 0.55 0.51 Lafayette, LA 1.80 0.92 0.51 Flint, MI 1.78 0.93 0.52 OmahaCouncil Bluffs, NEIA 1.78 0.96 0.54 MuskegonNorton Shores, MI 1.61 0.87 0.54 LexingtonFayette, KY 1.63 0.89 0.54 LouisvilleJefferson County, KYIN 1.74 0.95 0.54 Joplin, MO 1.45 0.81 0.56 Topeka, KS 1.65 0.93 0.56 Syracuse, NY 1.81 1.03 0.57 Lake Charles, LA 1.48 0.84 0.57 Monroe, LA 1.42 0.82 0.58 Kansas City, MOKS 1.75 1.03 0.58 St. Louis, MOIL 1.71 1.00 0.58 HaverhillNorth AndoverAmesbury, MANH NECTA 1.76 1.04 0.59 Manchester, NH 1.82 1.09 0.60 BuffaloNiagara Falls, NY 1.62 0.98 0.61 St. Cloud, MN 1.54 0.94 0.61 Rochester, NY 1.65 1.01 0.61 McAllenEdinburgMission, TX 1.06 0.65 0.62 DetroitLivoniaDearborn, MI Metropolitan Division 1.78 1.10 0.62 Jefferson City, MO 1.43 0.89 0.62 AguadillaIsabelaSan Sebastian, PR 0.87 0.54 0.62 CharlotteGastoniaRock Hill, NCSC 1.61 1.01 0.63 HagerstownMartinsburg, MDWV 1.50 0.94 0.63 Wichita, KS 1.46 0.93 0.64 BrownsvilleHarlingen, TX 1.01 0.64 0.64 Barnstable Town, MA 1.62 1.04 0.64 Bowling Green, KY 1.30 0.84 0.64 Binghamton, NY 1.42 0.93 0.65 EdisonNew Brunswick, NJ Metropolitan Division 1.74 1.14 0.65 Atlantic CityHammonton, NJ 1.40 0.92 0.66 Cape GirardeauJackson, MOIL 1.25 0.83 0.66 Greenville, NC 1.31 0.87 0.66 HuntingtonAshland, WVKYOH 1.24 0.82 0.66 HollandGrand Haven, MI 1.40 0.93 0.66
59 Table 2 6. Continued Area Name Relative s kills Relative w age W age i ndex ParkersburgMariettaVienna, WVOH 1.20 0.80 0.66 Grand RapidsWyoming, MI 1.42 0.96 0.68 Worcester, MACT 1.67 1.14 0.68 RaleighCary, NC 1.48 1.02 0.69 El Paso, TX 1.07 0.74 0.69 WarrenTroyFarmington Hills, MI 1.50 1.04 0.69 New YorkWhite PlainsWayne NYNJ 1.84 1.27 0.69 Camden, NJ Metropolitan Division 1.53 1.06 0.70 Jackson, MS 1.24 0.87 0.70 Eau Claire, WI 1.25 0.87 0.70 BaltimoreTowson, MD 1.64 1.16 0.71 Jacksonville, NC 1.09 0.78 0.72 ProvoOrem, UT 1.21 0.87 0.72 GreensboroHigh Point, NC 1.24 0.90 0.73 HoumaBayou CaneThibodaux, LA 1.31 0.95 0.73 New OrleansMetairieKenner, LA 1.30 0.95 0.73 Bismarck, ND 1.31 0.96 0.73 Wheeling, WVOH 1.06 0.78 0.74 Asheville, NC 1.16 0.86 0.74 New Bedford, MA 1.30 0.96 0.74 Myrtle BeachNorth Myrtle BeachConway, SC 0.92 0.68 0.75 Peabody, MA NECTA Division 1.46 1.09 0.75 Burlington, NC 1.07 0.80 0.75 AlbanySchenectadyTroy, NY 1.47 1.11 0.75 NilesBenton Harbor, MI 1.19 0.90 0.75 Springfield, MO 1.08 0.81 0.75 Corpus Christi, TX 1.09 0.83 0.76 LawrenceMethuenSalem, MANH NECTA Division 1.36 1.04 0.76 ShreveportBossier City, LA 1.12 0.86 0.76 Alexandria, LA 1.11 0.85 0.76 Roanoke, VA 1.16 0.89 0.77 MinneapolisSt. PaulBloomington, MNWI 1.49 1.15 0.77 OgdenClearfield, UT 1.16 0.90 0.77 Goldsboro, NC 1.03 0.80 0.78 Waco, TX 1.10 0.86 0.78 Manhattan, KS 1.10 0.86 0.78 LewistonAuburn, ME 1.17 0.91 0.78 NewarkUnion, NJPA Metropolitan Division 1.55 1.21 0.78 Rocky Mount, NC 1.04 0.82 0.79 Charleston, WV 1.12 0.88 0.79
60 Table 2 6 Continued Area Name Relative s kills Relative w age Wage i ndex BethesdaRockvilleFrederick, MD 1.68 1.33 0.79 San AntonioNew Braunfels, TX 1.09 0.86 0.79 Charlottesville, VA 1.23 0.98 0.80 Billings, MT 1.11 0.89 0.80 Wilmington, NC 1.09 0.87 0.80 Bangor, ME 1.13 0.91 0.80 Cumberland, MDWV 1.07 0.87 0.81 HickoryLenoirMorganton, NC 1.04 0.84 0.81 Lincoln, NE 1.16 0.94 0.81 Lynchburg, VA 1.03 0.84 0.81 Virginia BeachNorfolkNewport News, VANC 1.18 0.96 0.81 St. Joseph, MOKS 1.06 0.86 0.81 Logan, UTID 0.99 0.80 0.82 VinelandMillvilleBridgeton, NJ 1.23 1.01 0.82 NassauSuffolk, NY Metropolitan Division 1.37 1.13 0.82 KilleenTempleFort Hood, TX 1.04 0.86 0.82 Las Cruces, NM 0.98 0.81 0.83 BostonCambridgeQuincy, MA NECTA Division 1.66 1.38 0.83 Las VegasParadise, NV 1.12 0.93 0.83 GreenvilleMauldinEasley, SC 1.06 0.89 0.83 Amarillo, TX 1.03 0.86 0.83 Knoxville, TN 1.05 0.88 0.84 Tyler, TX 0.99 0.84 0.84 Rapid City, SD 0.98 0.83 0.84 Baton Rouge, LA 1.11 0.94 0.84 Tacoma, WA Metropolitan Division 1.29 1.09 0.84 Sioux Falls, SD 1.04 0.88 0.85 Santa Fe, NM 1.12 0.94 0.85 Anderson, SC 0.98 0.83 0.85 Wichita Falls, TX 0.95 0.80 0.85 Missoula, MT 1.00 0.85 0.85 Abilene, TX 0.95 0.81 0.85 Florence, SC 0.98 0.84 0.85 Portsmouth, NHME 1.30 1.11 0.85 Lubbock, TX 0.88 0.75 0.86 College StationBryan, TX 1.01 0.86 0.86 Columbia, MO 1.00 0.86 0.86 Clarksville, TNKY 0.97 0.83 0.86 KingsportBristolBristol, TNVA 0.98 0.85 0.86 LansingEast Lansing, MI 1.21 1.05 0.86
61 Table 2 6 Continued Area Name Relative s kills Relative w age Wage i ndex Albuquerque, NM 1.10 0.95 0.87 Oklahoma City, OK 1.04 0.91 0.87 Rochester, MN 1.30 1.13 0.87 Jackson, MI 1.10 0.96 0.87 St. George, UT 0.93 0.81 0.87 WinstonSalem, NC 1.08 0.94 0.87 Appleton, WI 1.06 0.93 0.87 MilwaukeeWaukeshaWest Allis, WI 1.19 1.04 0.87 Grand Forks, NDMN 1.03 0.90 0.87 Hattiesburg, MS 0.86 0.75 0.88 Racine, WI 1.03 0.90 0.88 Johnson City, TN 0.93 0.82 0.88 Yauco, PR 0.62 0.54 0.88 Morgantown, WV 1.00 0.88 0.88 Kingston, NY 1.09 0.96 0.88 Spartanburg, SC 1.01 0.89 0.88 GulfportBiloxi, MS 0.94 0.83 0.89 Elmira, NY 1.08 0.97 0.89 La Crosse, WIMN 1.05 0.94 0.89 Salt Lake City, UT 1.12 1.01 0.90 Guayama, PR 0.63 0.57 0.90 Spokane, WA 1.09 0.99 0.90 Fargo, NDMN 1.05 0.95 0.91 Mayaguez, PR 0.60 0.54 0.91 OrlandoKissimmeeSanford, FL 0.93 0.85 0.91 Duluth, MNWI 1.05 0.97 0.92 YoungstownWarrenBoardman, OHPA 0.91 0.84 0.92 Great Falls, MT 0.88 0.81 0.92 Fajardo, PR 0.60 0.55 0.92 DeltonaDaytona BeachOrmond Beach, FL 0.87 0.81 0.93 Little RockNorth Little RockConway, AR 0.97 0.90 0.93 SaginawSaginaw Township North, MI 0.96 0.89 0.93 Victoria, TX 0.95 0.88 0.93 Farmington, NM 1.04 0.97 0.93 Toledo, OH 0.98 0.92 0.93 Longview, TX 0.92 0.86 0.94 Morristown, TN 0.86 0.81 0.94 Fort Smith, AROK 0.85 0.80 0.94 AugustaRichmond County, GASC 0.94 0.88 0.94 Green Bay, WI 1.03 0.97 0.94
62 Table 2 6 Continued A rea Name R elative s kills Relative w age Wage i ndex UticaRome, NY 1.00 0.94 0.94 Memphis, TNMSAR 0.95 0.90 0.94 KalamazooPortage, MI 0.96 0.91 0.94 North PortBradentonSarasota, FL 0.92 0.87 0.95 Fort WorthArlington, TX Metropolitan Division 1.00 0.95 0.95 Dothan, AL 0.80 0.76 0.95 FayettevilleSpringdaleRogers, ARMO 0.92 0.87 0.95 Tallahassee, FL 0.98 0.93 0.95 MiamiMiami BeachKendall, FL Metropolitan Division 0.93 0.88 0.95 Springfield, MACT 1.14 1.08 0.95 Columbia, SC 0.97 0.93 0.95 RenoSparks, NV 1.01 0.97 0.96 TampaSt. PetersburgClearwater, FL 0.97 0.93 0.96 Jonesboro, AR 0.79 0.76 0.96 Albany, GA 0.85 0.81 0.96 Pittsburgh, PA 1.04 1.00 0.96 Palm BayMelbourneTitusville, FL 0.96 0.92 0.96 Cheyenne, WY 1.06 1.01 0.96 PoughkeepsieNewburghMiddletown, NY 1.11 1.06 0.96 Ocean City, NJ 0.91 0.87 0.96 Jackson, TN 0.87 0.84 0.96 Boise CityNampa, ID 0.94 0.91 0.96 PensacolaFerry PassBrent, FL 0.87 0.84 0.96 Lima, OH 0.92 0.89 0.97 Chattanooga, TNGA 0.89 0.86 0.97 DurhamChapel Hill, NC 1.22 1.18 0.97 Altoona, PA 0.85 0.82 0.97 NashvilleDavidsonMurfreesboroFranklin, TN 0.97 0.94 0.97 Cape CoralFort Myers, FL 0.88 0.85 0.97 HoustonSugar LandBaytown, TX 1.05 1.02 0.97 TexarkanaTexarkana, TXAR 0.89 0.87 0.97 Hot Springs, AR 0.73 0.71 0.97 Fort LauderdalePompano BeachDeerfield Beach, FL 0.93 0.91 0.97 Bellingham, WA 1.05 1.02 0.97 Montgomery, AL 0.90 0.88 0.97 Janesville, WI 0.93 0.91 0.97 Columbus, GAAL 0.84 0.82 0.98 CharlestonNorth CharlestonSummerville, SC 0.96 0.94 0.98 Pascagoula, MS 1.04 1.02 0.98 BlacksburgChristiansburgRadford, VA 0.94 0.92 0.98
63 Table 2 6 Continued Area name R elative s kills Rel ative w age Wage i ndex KennewickPascoRichland, WA 1.07 1.05 0.98 Nashua, NHMA NECTA Division 1.11 1.09 0.98 TrentonEwing, NJ 1.43 1.40 0.98 BeaumontPort Arthur, TX 0.98 0.96 0.98 Valdosta, GA 0.75 0.73 0.98 BirminghamHoover, AL 0.95 0.94 0.98 San GermanCabo Rojo, PR 0.55 0.54 0.99 Mobile, AL 0.89 0.88 0.99 EugeneSpringfield, OR 0.97 0.96 0.99 CantonMassillon, OH 0.85 0.84 0.99 Macon, GA 0.85 0.85 0.99 Bend, OR 0.94 0.93 0.99 Owensboro, KY 0.83 0.83 0.99 West Palm BeachBoca RatonBoynton Beach, FL 0.93 0.92 0.99 DavenportMolineRock Island, IAIL 0.90 0.90 0.99 Columbus, OH 1.02 1.01 0.99 Dayton, OH 0.98 0.98 1.00 Fresno, CA 0.88 0.88 1.00 Tucson, AZ 0.93 0.93 1.00 Idaho Falls, ID 0.80 0.80 1.00 Huntsville, AL 1.04 1.05 1.00 Jacksonville, FL 0.93 0.94 1.00 PortlandSouth PortlandBiddeford, ME 1.00 1.01 1.01 Springfield, OH 0.87 0.88 1.01 VisaliaPorterville, CA 0.79 0.80 1.01 Ocala, FL 0.79 0.80 1.01 Yakima, WA 0.91 0.92 1.01 Monroe, MI 0.95 0.96 1.01 RochesterDover NHME 1.01 1.03 1.01 SebastianVero Beach, FL 0.80 0.82 1.02 Panama CityLynn HavenPanama City Beach, FL 0.78 0.79 1.02 Port St. Lucie, FL 0.82 0.84 1.02 Wausau, WI 0.95 0.97 1.02 Colorado Springs, CO 0.98 1.01 1.02 CrestviewFort Walton BeachDestin, FL 0.84 0.86 1.02 Sioux City, IANESD 0.79 0.81 1.02 AtlantaSandy SpringsMarietta, GA 0.99 1.02 1.03 Yuma, AZ 0.73 0.75 1.03 Lake Havasu City Kingman, AZ 0.82 0.84 1.03 Midland, TX 0.97 1.00 1.03
64 Table 2 6 Continued Area name Relative s kills Relative w age Wage i ndex PhoenixMesaGlendale, AZ 0.98 1.01 1.03 FlorenceMuscle Shoals, AL 0.75 0.77 1.03 Salem, OR 0.96 0.99 1.03 Fort Wayne, IN 0.88 0.91 1.03 Gadsden, AL 0.71 0.73 1.03 Evansville, INKY 0.86 0.89 1.03 Gainesville, FL 0.91 0.94 1.03 Savannah, GA 0.88 0.91 1.03 Harrisonburg, VA 0.82 0.85 1.04 Sumter, SC 0.77 0.79 1.04 Lancaster, PA 0.89 0.93 1.04 Coeur d'Alene, ID 0.81 0.84 1.04 Tulsa, OK 0.89 0.92 1.04 South BendMishawaka, INMI 0.88 0.91 1.04 Cleveland, TN 0.76 0.79 1.04 Erie, PA 0.82 0.85 1.04 IndianapolisCarmel, IN 0.96 1.00 1.04 LakelandWinter Haven, FL 0.82 0.86 1.05 Punta Gorda, FL 0.77 0.80 1.05 ScrantonWilkesBarre, PA 0.85 0.90 1.05 Medford, OR 0.88 0.92 1.05 Laredo, TX 0.67 0.70 1.05 OshkoshNeenah, WI 0.96 1.01 1.05 AllentownBethlehemEaston, PANJ 0.94 0.99 1.06 Richmond, VA 0.97 1.03 1.06 HarrisburgCarlisle, PA 0.98 1.04 1.06 Reading, PA 0.95 1.00 1.06 Brunswick, GA 0.77 0.81 1.06 San Angelo, TX 0.72 0.77 1.06 Dalton, GA 0.79 0.84 1.06 Gary, IN Metropolitan Division 0.88 0.94 1.06 NaplesMarco Island, FL 0.81 0.86 1.06 RiversideSan BernardinoOntario, CA 0.90 0.96 1.07 Tuscaloosa, AL 0.83 0.88 1.07 Fort CollinsLoveland, CO 0.94 1.00 1.07 BakersfieldDelano, CA 0.90 0.96 1.07 MankatoNorth Mankato, MN 0.84 0.90 1.07 Rockford, IL 0.86 0.93 1.07 ElkhartGoshen, IN 0.80 0.86 1.07 Ann Arbor, MI 1.04 1.11 1.07
65 Table 2 6 Continued Area name Relative s kills Relative w age Wage i ndex Salisbury, MD 0.85 0.91 1.08 Prescott, AZ 0.81 0.87 1.08 Des MoinesWest Des Moines, IA 0.96 1.03 1.08 AthensClarke County, GA 0.75 0.81 1.08 AnnistonOxford, AL 0.78 0.84 1.08 Anderson, IN 0.74 0.80 1.08 Williamsport, PA 0.81 0.88 1.08 Glens Falls, NY 0.86 0.93 1.09 Mount VernonAnacortes, WA 0.95 1.04 1.09 Elizabethtown, KY 0.80 0.88 1.09 Bay City, MI 0.77 0.84 1.09 DenverAuroraBroomfield, CO 1.03 1.13 1.09 Salinas, CA 0.81 0.89 1.10 DallasPlanoIrving, TX Metropolitan Division 0.94 1.03 1.10 Philadelphia, PA Metropolitan Division 1.03 1.13 1.10 ChicagoJolietNaperville, IL Metropolitan Division 0.97 1.07 1.10 AuburnOpelika, AL 0.70 0.77 1.10 CincinnatiMiddletown, OHKYIN 0.91 1.00 1.10 Decatur, AL 0.80 0.88 1.10 ClevelandElyriaMentor, OH 0.93 1.02 1.10 Madison, WI 0.98 1.08 1.11 BurlingtonSouth Burlington, VT 0.97 1.07 1.11 Pueblo, CO 0.80 0.88 1.11 Sheboygan, WI 0.85 0.95 1.11 San DiegoCarlsbadSan Marcos, CA 0.99 1.10 1.12 Terre Haute, IN 0.76 0.84 1.12 Muncie, IN 0.76 0.85 1.12 Springfield, IL 0.86 0.97 1.12 WaterlooCedar Falls, IA 0.81 0.91 1.12 Dubuque, IA 0.80 0.90 1.12 Winchester, VAWV 0.82 0.92 1.13 Lake CountyKenosha County, ILWI 0.93 1.05 1.13 Los AngelesLong BeachGlendale, CA 0.96 1.09 1.13 Pittsfield, MA 0.88 1.00 1.13 Mansfield, OH 0.75 0.85 1.14 AustinRound RockSan Marcos, TX 0.90 1.02 1.14 Peoria, IL 0.86 0.98 1.14 Battle Creek, MI 0.83 0.95 1.14 Pocatello, ID 0.72 0.83 1.14 Michigan CityLa Porte, IN 0.75 0.86 1.14
66 Table 2 6 Continued Area name Relative s kills Relative w age Wage i ndex Cedar Rapids, IA 0.89 1.02 1.15 Palm Coast, FL 0.66 0.76 1.15 Bloomington, IN 0.78 0.89 1.15 Santa AnaAnaheimIrvine, CA Metropolitan Division 0.97 1.11 1.15 YorkHanover, PA 0.83 0.96 1.15 Johnstown, PA 0.74 0.84 1.15 Rome, GA 0.74 0.85 1.15 Modesto, CA 0.83 0.96 1.16 Chico, CA 0.78 0.90 1.16 Honolulu, HI 0.92 1.07 1.16 Ithaca, NY 0.95 1.11 1.17 Columbus, IN 0.82 0.95 1.17 OxnardThousand OaksVentura, CA 0.91 1.06 1.17 KankakeeBradley, IL 0.76 0.89 1.17 ProvidenceFall RiverWarwick, RIMA 0.90 1.05 1.17 Lebanon, PA 0.76 0.90 1.18 Danville, VA 0.68 0.80 1.18 Stockton, CA 0.85 1.01 1.18 Boulder, CO 1.01 1.19 1.18 Flagstaff, AZ 0.73 0.87 1.18 Gainesville, GA 0.73 0.87 1.18 Lawton, OK 0.69 0.81 1.19 Wilmington, DEMDNJ Metropolitan Division 0.97 1.16 1.19 SacramentoArdenArcadeRoseville, CA 0.98 1.17 1.20 Pine Bluff, AR 0.70 0.84 1.20 Santa BarbaraSanta MariaGoleta, CA 0.88 1.06 1.21 BrocktonBridgewaterEaston, MA NECTA Division 0.93 1.12 1.21 Redding, CA 0.78 0.95 1.21 El Centro, CA 0.69 0.84 1.21 Waterbury, CT 0.87 1.06 1.22 SteubenvilleWeirton, OHWV 0.66 0.81 1.22 Framingham, MA NECTA Division 1.10 1.36 1.24 Greeley, CO 0.79 0.98 1.24 Casper, WY 0.84 1.04 1.24 Corvallis, OR 0.83 1.03 1.24 PortlandVancouverHillsboro, ORWA 0.89 1.11 1.25 New Haven, CT 0.95 1.18 1.25 Lafayette, IN 0.74 0.92 1.25 Grand Junction, CO 0.76 0.95 1.25 HartfordWest HartfordEast Hartford, CT 1.01 1.26 1.25
67 Table 2 6 Continued Area name R elative s kills Relative w age Wage i ndex Warner Robins, GA 0.88 1.10 1.25 State College, PA 0.79 1.00 1.26 San Luis ObispoPaso Robles, CA 0.79 1.00 1.26 Merced, CA 0.69 0.87 1.26 Akron, OH 0.77 0.97 1.27 Santa RosaPetaluma, CA 0.87 1.11 1.27 Anchorage, AK 0.96 1.23 1.28 Danbury, CT 0.86 1.11 1.29 Odessa, TX 0.74 0.96 1.30 Santa CruzWatsonville, CA 0.82 1.06 1.30 ChampaignUrbana, IL 0.78 1.02 1.31 TauntonNortonRaynham, MA NECTA Division 0.82 1.08 1.31 WashingtonArlingtonAlexandria, DCVAMDWV 1.10 1.45 1.32 NorwichNew London, CTRI 0.78 1.04 1.32 Danville, IL 0.69 0.92 1.32 BridgeportStamfordNorwalk, CT 0.98 1.31 1.33 ShermanDenison, TX 0.63 0.84 1.33 Dover, DE 0.70 0.94 1.34 Napa, CA 0.80 1.06 1.34 Lewiston, IDWA 0.68 0.92 1.35 Yuba City, CA 0.72 0.97 1.35 VallejoFairfield, CA 0.82 1.10 1.35 OaklandFremontHayward, CA Metropolitan Division 0.99 1.34 1.35 LowellBillericaChelmsford, MANH NECTA Division 0.96 1.31 1.36 BloomingtonNormal, IL 0.74 1.01 1.37 BremertonSilverdale, WA 0.87 1.19 1.37 Sandusky, OH 0.60 0.84 1.39 Carson City, NV 0.76 1.08 1.41 Kokomo, IN 0.67 0.95 1.41 Iowa City, IA 0.74 1.05 1.42 San FranciscoSan MateoRedwood City, CA 1.03 1.46 1.42 MaderaChowchilla, CA 0.64 0.91 1.42 Ames, IA 0.71 1.01 1.42 LeominsterFitchburgGardner, MA 0.71 1.01 1.43 Decatur, IL 0.70 1.00 1.44 San Jose Sunnyvale Santa Clara, CA 1.06 1.54 1.45 Olympia, WA 0.79 1.14 1.45 HinesvilleFort Stewart, GA 0.60 0.90 1.49 SeattleBellevueEverett, WA Metropolitan Division 0.85 1.33 1.57 HanfordCorcoran, CA 0.66 1.03 1.57
68 Table 2 6 Continued A rea name R elative s kills Relative w age Wage i ndex Fairbanks, AK 0.78 1.30 1.67 Fond du Lac, WI 0.55 0.92 1.69 WenatcheeEast Wenatchee, WA 0.53 0.92 1.72 Longview, WA 0.57 1.03 1.82 Table 2 7 Wage index and p opulation wage_index logpop 0.004 (0 .012 ) constant 0. 927 (0.169) R 2 0.227 N 401
69 Table 2 8 Wage i ndex on housing price and amenities Wage index housing_cost 0.034** (0.004) avg_july_temp 0.0021 (0.0019) avg_jan_temp 0.009 (0.0011) heating_degree_days 0.0051* (0.0026) cooling_degree_days 0.0012 (0.0029) violent_crime 0.0037* (0.0016) constant 1.182 (0.164) N 236 R 2 0.242 Standard errors in parenthesis *= significant at the 5% level, **= significant at 1% level
70 Table 2 9. Specification 1 o ccupational l icensing 10 th percentile 25 th percentile 50 th percentile 75 th percentile 90 th percentile license 0 .023 ** 0.0282** 0.0302 ** 0 .0286 ** 0 .02834 ** ( 0 .0017) ( 0 .0018) ( 0 .0018) ( 0 .0018) ( 0 .0018) Constant 2.122 2.325 2.715 3.156 11.1 (0.009) (0.009) (0.0094) (0.009) (0.009) N 130,893 130,528 129,761 128,655 125,840 R 2 0.80 0.832 0.84 0.839 0.827 Standard errors in parenthesis **= significant at the 1% level Table 2 10. Specification 2 occupational l icensing 10 th percentile 25 th percentile 50 th percentile 75 th percentile 90 th percentile w age index .709 ** .854 ** 1.035 ** 1.150 ** 1.13 ** (0.002) (.003) (.003) (.0030) (0.0032) license .009 ** .0010 .0063 ** 0.0003 .0037 (.001) (.001) (.0014) (0.0016) (0.0017) constant 3.011 3.311 3.608 3.897 4.179 (.002) (.0020) (0.002) (0.002) (0.0041) N 130,893 130,528 129,761 128,655 125,840 R 2 0.929 0.945 0.953 0.950 0.937 Standard errors in parenthesis **= significant at the 1% level and *= significant at the 5% level
71 Table 2 11. Specificatio n 3 occupational l icensing 10th percentile 25th percentile 50th percentile 75th percentile 90th percentile w age_index 0 .819 0.928 1.014 1.0558 1.042 (0.004) (.0043) (0.0043) ( 0 .0044) ( 0 .0046) license 0.005 0.009 0.0097 0 .0076 0 .0073 (0.0016) (0.0015) (0.0015) ( 0 .0016) ( 0 .0016) p ercent_licensed 0.0202 0.0042 0.0075 0.014 0.015 (0.0041) (0.0048) ( 0 .0047) (0.0048) (0.005) constant 2.891 3.217 3.545 3.858 4.15 ( 0 .008) ( 0 .008) ( 0 .008) ( 0 .008) (.0102) N 130,893 130,528 129,761 128,655 125,840 R 2 0.844 0.873 0.885 0 .885 0.875
72 CHAPTER 3 APPLICATION OF THE WAGE INDEX In this chapter I use the wage index developed in chapter two to examine how the Medicare payments to hospitals will be impacted if the OES wage index is used to adjust geographic variation in labor costs across different cities. I find that if the OES ind ex is used, a total of 20.3 billion dollars gets reallocated between hospitals. Medicare Wage Index A wage index that is computed annually at the MSA level similar to the OES index developed in this chapter is the Medicare Wage Index. In this section I w ill compare the Medicare Wage Index with the OES index. Before beginning the comparison I will start with a brief introduction the Medicare Wage Index and how it is computed. The Medicare program is a national insurance program, administered by the U.S. Fe deral government that guarantees access to health insurance to citizens aged 65 and older who have worked and paid onto the system. Under special circumstances such as disability and end stage renal disease people who are younger than 65 may also qualify fo r Medicare benefits. Medicare is funded solely by federal government, and is administered by the Centers for Medicare and Medicaid Services (CMS ) which is a division of the U.S. Department of Health and Human Services. Since 1983, hospitals have been reimb ursed for inpatient services provided to Medicare beneficiaries based on the inpatient prospective payment system ( PPS). Under this system hospitals are paid predetermined, fixed amounts by CMS based on a yment is also adjusted for geographic variation in wages across hospital areas. The wage index is a measure of
73 the differences in hospital wage rates among labor markets. The wage index compares the hourly wage for hospital workers in each metropolitan sta tistical area (MSA) or statewide rural area to the national average. A hospital labor market is defined to be the same as a metropolitan statistical area and residual rural areas. The wage index is based on wage data collected through a survey. This survey collects each hospital data is the n adjusted for occupational mix based on data from the Occupational Mix Survey. The wage index is revised each year based on wage data repo rted by hospitals. The CMS considers the hospital labor market to be the same as MSA so the labor markets for the CMS and OES indices are the same. One major difference between the two indices is that while the OES data adjusts for all the 900 plus occupations the CMS index only adjusts for occupational classification The nineteen occupations are grouped into seven clinical Occupational Mix Adjustment ( OMA ) : Nu rsing, Physical Therapy, Occupational Therapy, Respiratory Therapy, Pharmacy, Dietary and Medical and Clinical Laboratory. The Occupational Mix Adjustment adjusts total wages assigned to each class for the proportion of hours worked by each category within physician on a salary at the hospital as well as non clinical staff such as clinical and administrative support services staff. Initial computations are performed for e ach hospital and the result is aggregated at the MSA level. Each hospital does not contribute equally to the final adjustment. T he wages are weighted by the number of
74 hours for each occupation so that the hospitals with a greater number of employees have a greater influence in determining the wage index. T he fundamental difference between the OES and the CMS wage index is in the occupational mix adjustment. The OES index survey covers the entire spectrum of occupations whereas the CMS index is confined to wages for healthcare workers. The first difference that stands out between these two indices is the dispersion in wages across MSA s For 2012, the most recent year for which the data is available, t he OES index ranges from 0.53 to 1.53 while the CMS index ranges from 0.36 to 1.71. Table 3 1 shows the results from a regression the relationship between the CMS and OES indices were the dependent and independent variables respectively. The result of this regression shows that the two indices diffe r from each other. If these indices were perfectly correlated the coefficient on the oes_index variable would have been equal to 1. The result shows that a 100% increase in OES index is will lead to 122% increase in the CMS Index. This result shows that re lative to the OES index the CMS index overestimates the geographic differences in pay by 22%. A scatterplot of the OES index (Fig 2 1) shows that these two indices show a greater degree of di spersion at higher wage levels. Figure 2 2 shows a distr ibution of the variable cms_oes which is a ratio of the CMS Index to the OES Index. This variable has a range of 0.654 to 1.757 with a median of 0.996 The plot also shows a longer right tail with a greater dispersion between the two wage indices above the 75 th pe rcentile of 1.08. The cities where the CMS index is lower in comparison to the OES index are cities where the wages paid to healthcare providers are relatively lower than the wages paid to the aver age worker. Similarly, at the other end of the distribution where CMS Index is much greater than the
75 OES index, healthcare providers are paid a significantly higher relative wage compared to the average worker. To get a better understanding of the divergence between the two indices, I look at how the quality of c are provided by healthcare workers is related to the cms_oes variable. My hypothesis is that cities where medical wages are higher relative to the average worker should be receiving a higher quality of healthcare which implies that cms oes and the quality of care variables should be positively correlated The data for this analysis comes from the Dartmouth Atlas of Healthcare. The Dartmouth Atlas for Health Care is maintained by the Dartmouth Institute for Health Policy and Clinical Practice, an organizatio n within Dartmouth College. The Dartmouth Atlas projects has receiving appropriate ma According to the Atlas, effective care refers to services that are of proven value and have no significant tradeoffs. The benefits of these services far outweigh the risks so that all patients with s pecific medical condition should receive them. Using this variable as a measure of the quality of care I test to see if higher cms oes cities do in fact receive higher quality healthcare. To test this hypothesis, I estimated an equation of the following form: ( 3 1 ) The result of this regr ession is reported in Table 3 2 Contrary to my expectations, the results show that if the CMS index gets larger relative to the OES index the quality of care declines. The regr ession result shows that a 100% increase in the payment to healthcare relative to the workers across all occupations is associated with 4.1%
76 decline in the quality care provided. This result is counterintuitive however it is consistent with literature on healthcare cost containment by providing effective and quality care. Brody (2012) conducted a study in the cost effectiveness of breast cancer diagnosis and treatment and found that a treatment with high dose chemotherapy followed by autologous bone ma rrow transplantation which is known to have meaningful ben is still practiced in 15% of the sample that he examined. Frank et al (2010) find that despite the popularity of pay for performance as a tool for improving quality of care there is little empirical basis to support the claim. The authors used data from published performance reports of physician medical groups to compare quality of care before and after the implementation of pay for performance relative to a c ontrol group and do not find any evidence of an increased quality of care in the pay for performance group implying that higher wages for healthcare providers does not necessarily translate to higher quality of care. The result on the relationship between quality of care and the divergence between CMS and OES index support s the finding that higher wages may not lead to higher quality of care, in fact I find that there is a decline the quality of care provided. Allocation of Medicare Funds B ased on OES Inde x The CMS wage index is used by Medicare to reimburse funds to hospitals for inpatient services provided to Medicare enrollees In this section, I will examine how the allocation of funds to hospitals changes if the OES index is used to adjust geographic v ariation in wages. The payment to hospitals consists of two categories of payment: an operating payment and a capital payment. The operating payment is further divided into a labor portion and non labor portion. The labor portion is adjusted by the wage in dex. The payments to hospitals is specifically defined in equation (2 11 ):
77 = (Operating Payment + Capital Payment) Diagnosis Weight ( 3 2 ) w here the operating payment is defined to be the sum of a labor portion and a non labor portion Diagnosis weight adjusts for the type of service of provided. There are 751 DRG categories used by the Medicare. The operating payment is a standardized amount set by the Congress every year If the wage index is greater than 1, the labor s hare equals 68.8 percent. If the wage index is less than or equal to 1 the labor share is 62 percent. Next, the dollar amount of the labor portion is adjusted using the wage index. For the year 2012, the standard operating payment was set to $ 5209.74. For hospitals in areas with a wage index greater than 1, the labor portion was 3584.74 and for hospitals in areas with a wage index less than or equal to 1 the labor portion was 3230.04. The labor portion of the operating payment is then multiplied by the wag e index to get the total labor related payment for each discharge. To compute how the payment to hospitals would change if the OES index was used to adjust for differences in wages across geographical regions, I utilize data from the CMS IPPS Final Rule F iles 9 for the year 2012. The final rule files provide data on the wage index for each hospital and the number of discharges for the year. Based on this information I am able to compute the change in hospital payments using OES index instead of the CMS ind ex. I used the following formula to compute how funds would be reallocated if the OES index is used: Difference in hospital payment = ( 3 3 ) 9 The data is available for download at the CMS website: http://www.cms.gov/Medicare/Medicare Fee for Serv ice Payment/AcuteInpatientPPS/FY 2012 IPPS Final Rule Home Page.html
78 There were 3500 hospitals in the Medicare Impact files. Since the OES does not report wages for rural areas, of those 3500 hospitals, only 2428 hospitals were matched to the OES wage index. Those hospitals were then aggregated to the MSA level. The reallocation resulting from the use of OE S index is reported in Table 3 3 Of the 365 3 3 hospitals in 200 cities would receive a higher payment than they receive under the CMS index and the remaining areas will receive a lower payment. The OES index would reallocate a total of 20.5 billion dollars between hospitals in different labor market areas. Hospitals in large cities like New York, Chicago, Los Angeles, Seattle, Miami and others would receive significantly less und er the OE S wage index while midsize cities like Kansas City, Buffalo, Charlotte, and Nashville would receive a higher amount. The biggest loser is New York City, which would receive a payment that is 1.26 billion dollars. The biggest gainer is Cleveland, w hich would receive a payment that is 156.4 million dollars. The CMS wage index has been criticized for not accurately accounting for cost variations across regions. There have been large differences in wag e indices across adjacent labor market areas. Medi care has been considering revising the wage index calculations. One of the suggestions of Medicare Payment Advisory Committee (MedPAC) is to use the data from the Occupational Employment Statistics for wages. A M e dPAC report published in 2010 has identifie d that OES data is better suited for wage index construction. The report claims that there is an incentive problem when hospital wage data is used for calculation of wage index. The hospitals do not bargain with employees over compensation. A data source l ike OES that aggregates wage data across all industries is bel ieved to mitigate this problem.
79 Another advantage of using the OES data is that it covers a very broad spectrum of occupations even for the medical occupations. The CMS uses data from the Occup ational Mix Survey to adjust for the variations in occupational mix across hospitals. In this survey, wage data is reported for 19 occupations which are then used to adjust the CMS wage index. The OES reports data for 60 different medical occupations. The level of occupational detail in the OES index is better suited to handle the differences in hospital pay arising from variations in occupational mix across hospitals in different labor market areas. Conclusion In this chapter I examine how the wage index w ould reallocate funds if it were used by Medicare for hospital payments. While the wage index developed in this chapter was a measure of the wage differential across cities aggregated for all occupations, it is possible to use the same datasets to measure differences in pay for healthcare workers. OES wage data that is aggregated across all industries is less prone to year fluctuations that arise due to reporting errors by individual hospitals, and non representative wages paid by hospitals.
80 T able 3 1 CMS OES r egression CMS_Index oes_index 1.224** (0 .055 ) constant 0. 736 (0.048) R 2 0.257 N 367 Standard errors in parenthesis **= significant at the 1% level
81 T able 3 2 Comparison to CMS wage i ndex quality_of_care cms_oes 0.0412 ** (0 .018 ) constant 0. 879 (0.0192) R 2 0.176 N 153 Standard errors in parenthesis **= significant at the 5% level Table 3 3 Reallocation of Medicare hospitals payments u sing OES i ndex Area Net Redistribution Cleveland Elyria Mentor, OH 156425732.1 Seattle Bellevue Everett, WA Metropolitan Division 122551855.4 Washington Arlington Alexandria DCVAMDWV 112359631.8 Dallas Plano Irving, TX Metropolitan Division 102203569.1 Cincinnati Middletown, OHKYIN 99947029.53 Atlanta Sandy Springs Marietta, GA 90007458.75 Chicago Joliet Naperville, IL Metropolitan Division 72408755.26 Akron, OH 63993262.54 Indianapolis Carmel, IN 58094888.03 Pittsburgh, PA 57531516.73 Jacksonville, FL 55291153.75 Tulsa, OK 51841730.51 Austin Round Rock San Marcos, TX 49143945.54 Richmond, VA 38994906.57 Scranton WilkesBarre, PA 37643149.65 PortlandVancouverHillsboro, ORWA 37518385.77 LakelandWinter Haven, FL 34574010.12 BirminghamHoover, AL 32726071.86 TampaSt. PetersburgClearwater FL 28326424.99 Peoria, IL 27055424.32 Tucson, AZ 26193826.45 Philadelphia, PA Metropolitan Division 25352339.39 Springfield, IL 25168691.53 HarrisburgCarlisle, PA 23517401.04 Iowa City, IA 22601664.23 Evansville, INKY 21520453.98 Decatur, IL 21397740.07 Ocala, FL 21270848.11
82 Table 3 3 Continued Area Net Redistribution YorkHanover, PA 19907233.73 AllentownBethlehemEaston, PANJ 19645259.38 Des MoinesWest Des Moines, IA 19390614.22 Erie, PA 18771789.69 Johnstown, PA 18587902.4 ChampaignUrbana, IL 18254823.23 Huntsville, AL 17444278.14 Fort Wayne, IN 17377222.07 Gainesville, FL 15762991.79 Savannah, GA 15747470.4 Mobile, AL 15243961.26 Steubenville Weirton, OHWV 15087354.77 Gainesville, GA 14705584.86 Tuscaloosa, AL 14592694.86 Lafayette, IN 14364064.08 Sherman Denison, TX 14260740.57 Punta Gorda, FL 14240852.98 York Hanover, PA 19907233.73 Allentown Bethlehem Easton, PANJ 19645259.38 Des Moines West Des Moines, IA 19390614.22 Erie, PA 18771789.69 Johnstown, PA 18587902.4 Champaign Urbana, IL 18254823.23 Huntsville, AL 17444278.14 Fort Wayne, IN 17377222.07 Fort Wayne, IN 17377222.07 Gainesville, FL 15762991.79 Savannah, GA 15747470.4 Mobile, AL 15243961.26 Steubenville Weirton, OHWV 15087354.77 Gainesville, GA 14705584.86 Tuscaloosa, AL 14592694.86 Lafayette, IN 14364064.08 Sherman Denison, TX 14260740.57 Punta Gorda, FL 14240852.98 Wilmington, DEMDNJ Metropolitan Division 13386906.74 NashvilleDavidsonMurfreesboroFranklin, TN 13315924.2 Olympia, WA 13149745.95 WaterlooCedar Falls, IA 13043775.89 NaplesMarco Island, FL 12900578.39
83 Table 3 3 Continued Area Net Redistribution Rome, GA 12895900.69 Cedar Rapids, IA 12812265.7 BloomingtonNormal, IL 12485639.12 Lancaster, PA 12402256.75 Port St. Lucie, FL 12095792.11 Rockford, IL 11727515.6 Odessa, TX 11558879.09 FlorenceMuscle Shoals, AL 11518215.41 Panama CityLynn HavenPanama City Beach, FL 11489739.37 DenverAuroraBroomfield, CO 11450157.9 Lake CountyKenosha County, ILWI Metropolitan Division 11406876.56 Dayton, OH 11366491.59 PensacolaFerry PassBrent, FL 11360287.08 Dothan, AL 11267840.57 AnnistonOxford, AL 11234615.81 South BendMishawaka, INMI 10969642.81 Little RockNorth Little RockConway, AR 10950894.51 Terre Haute, IN 10940017.79 Longview, WA 10724397.51 Sandusky, OH 10411616.62 Gadsden, AL 10391236.3 Warner Robins, GA 10353449.71 Chattanooga, TNGA 10174938.21 Kokomo, IN 10131762.64 AthensClarke County, GA 10012622.9 WenatcheeEast Wenatchee, WA 9889906.583 Montgomery, AL 9811485.844 Knoxville, TN 9776007.206 Muncie, IN 9725811.27 YoungstownWarrenBoardman, OHPA 9599593.978 Salisbury, MD 9480095.903 Gary, IN Metropolitan Division 9450536.937 Laredo, TX 9443953.508 Decatur, AL 9171891.569 CantonMassillon, OH 9029533.545 Bay City, MI 8866050.132 State College, PA 8853018.879 San Angelo, TX 8698128.631 Lawton, OK 8684453.452 Fort Smith, AROK 8684186.98
84 Table 3 3 Continued Area Net Redistribution Ann Arbor, MI 8546316.199 Colorado Springs, CO 8431468.728 Harrisonburg, VA 8358316.915 CharlestonNorth CharlestonSummerville, SC 8232928.525 Danville, VA 8167025.985 CrestviewFort Walton BeachDestin, FL 8127180.683 Pine Bluff, AR 8090170.066 Fort CollinsLoveland, CO 7916287.681 Dubuque, IA 7791326.681 Fond du Lac, WI 7779676.9 BeaumontPort Arthur, TX 7775498.032 Dover, DE 7654382.611 Glens Falls, NY 7640918.596 Columbia, SC 7635539.588 Mansfield, OH 7595088.595 KingsportBristolBristol, TNVA 7451912.711 Pueblo, CO 7033638.365 Ames, IA 6971038.109 Bloomington, IN 6943488.937 ElkhartGoshen, IN 6880715.455 Michigan CityLa Porte, IN 6829985.315 Elizabethtown, KY 6795712.638 AuburnOpelika, AL 6763975.617 Williamsport, PA 6694262.083 HanfordCorcoran, CA 6469472.822 Lebanon, PA 6418795.603 Wausau, WI 6338673.853 Carson City, NV 6231685.843 SebastianVero Beach, FL 6189600.664 Johnson City, TN 5902160.843 BremertonSilverdale, WA 5857159.865 Owensboro, KY 5851544.649 TexarkanaTexarkana, TXAR 5829479.542 Boulder, CO 5808025.065 Anderson, IN 5726680.408 Cleveland, TN 5649039.148 Palm Coast, FL 5620251.361 Cape CoralFort Myers, FL 5612914.894 OshkoshNeenah, WI 5380757.751 Sioux City, IANESD 5338765.513
85 Table 3 3 Continued Area Net Redistribution Danville, IL 5273680.044 Dalton, GA 5177012.151 DavenportMolineRock Island, IAIL 5151239.336 Palm BayMelbourneTitusville, FL 5056893.389 Jackson, TN 5015610.145 Memphis, TNMSAR 4984717.999 Grand Junction, CO 4965279.172 Pascagoula, MS 4880231.177 Jonesboro, AR 4864764.091 North PortBradentonSarasota, FL 4825905.023 Columbus, IN 4686811.409 Greeley, CO 4655037.144 WinstonSalem, NC 4643562.025 UticaRome, NY 4556830.24 Springfield, OH 4478180.204 KankakeeBradley, IL 4478048.418 Sumter, SC 4307314.113 Brunswick, GA 4268850.19 Lewiston, IDWA 4074645.172 Yuma, AZ 4009429.305 Battle Creek, MI 3810801.267 Macon, GA 3662959.378 Tallahassee, FL 3540980.825 Columbus, GAAL 3356228.141 Valdosta, GA 3322635.975 Casper, WY 3297357.193 Morristown, TN 3251451.828 Sheboygan, WI 3251150.238 Altoona, PA 3163181.208 Hattiesburg, MS 3134117.678 Ithaca, NY 3015273.161 Pocatello, ID 2953786.946 BlacksburgChristiansburgRadford, VA 2790053.566 Columbia, MO 2789616.635 Boise CityNampa, ID 2734056.583 Winchester, VAWV 2670930.378 Toledo, OH 2655522.342 Victoria, TX 2640302.848 Longview, TX 2538579.199 SaginawSaginaw Township North, MI 2509127.565
86 Table 3 3 Continued Area Net Redistribution Albany, GA 2430894.321 Yuba City, CA 2384427.977 Monroe, MI 2342225.066 GulfportBiloxi, MS 2027748.353 Corvallis, OR 2013153.951 Idaho Falls, ID 1999886.321 Honolulu, HI 1997558.025 FayettevilleSpringdaleRogers, ARMO 1957376.698 Hot Springs, AR 1868488.223 DeltonaDaytona BeachOrmond Beach, FL 1782389.936 Fairbanks, AK 1707282.97 OrlandoKissimmeeSanford, FL 1539450.324 Morgantown, WV 1520893.487 Lima, OH 1508039.053 Florence, SC 1112583.706 MaderaChowchilla, CA 1083153.389 DurhamChapel Hill, NC 1024669.821 Clarksville, TNKY 984685.2599 Medford, OR 834741.7286 Elmira, NY 790417.5235 Janesville, WI 724948.0658 San Luis ObispoPaso Robles, CA 529536.456 Merced, CA 502787.0807 Reading, PA 484337.2384 Santa BarbaraSanta MariaGoleta, CA 456355.1631 Coeur d'Alene, ID 404229.8421 Columbus, OH 363516.9344 St. Joseph, MOKS 309380.7767 El Centro, CA 282658.8716 Midland, TX 78863.46143 Tyler, TX 43104.40689 MankatoNorth Mankato, MN 244980.4447 Logan, UTID 308584.9841 Yakima, WA 555491.354 Madison, WI 698455.7447 Green Bay, WI 699250.5782 Mount VernonAnacortes, WA 734389.9867 Manhattan, KS 778849.4989 Fort WorthArlington, TX Metropolitan Division 805857.6892
87 Table 3 3 Continued Area Net Redistribution St. George, UT 823847.6827 Augusta Richmond County, GASC 858310.9658 Abilene, TX 879897.7451 Amarillo, TX 895664.364 Jacksonville, NC 1035172.428 Appleton, WI 1057519.638 Goldsboro, NC 1106502.221 Racine, WI 1153987.228 Baton Rouge, LA 1287358.415 Flagstaff, AZ 1353888.34 Jackson, MI 1396692.41 Anderson, SC 1441815.253 Spartanburg, SC 1511207.689 Las Cruces, NM 1541022.674 Chico, CA 1689833.155 Lubbock, TX 1704625.485 Houma Bayou Cane Thibodaux, LA 1891552.225 Farmington, NM 2028013.841 La Crosse, WIMN 2041046.346 Cheyenne, WY 2071193.73 Lynchburg, VA 2080321.524 College Station Bryan, TX 2133671.836 Alexandria, LA 2166038.381 Lawrence, KS 2564828.395 Kalamazoo Portage, MI 2598161.892 Burlington, NC 2671062.767 Wheeling, WVOH 2749469.283 Charleston, WV 2836659.602 Hickory Lenoir Morganton, NC 2843017.159 Kennewick Pasco Richland, WA 3041616.171 Great Falls, MT 3106554.383 Salem, OR 3249166.262 Waco, TX 3261526.78 Lake Havasu City Kingman, AZ 3262705.666 Phoenix Mesa Glendale, AZ 3287194.587 Anchorage, AK 3403560.294 Cumberland, MDWV 3524674.437 Oklahoma City, OK 3583298.683 Santa Fe, NM 3744334.89
88 Table 3 3 Continued Area Net Redistribution Grand Forks, NDMN 3790735.448 Rocky Mount, NC 4014713.498 Salt Lake City, UT 4290169.322 HollandGrand Haven, MI 4459359.737 Prescott, AZ 4537700.723 Missoula, MT 4577552.491 Ocean City, NJ 4642706.022 NilesBenton Harbor, MI 4840606.224 Napa, CA 4976266.077 ParkersburgMariettaVienna, WVOH 5287740.508 Charlottesville, VA 5514107.24 OgdenClearfield, UT 5557539.541 KilleenTempleFort Hood, TX 5681080.43 Bend, OR 5697881.728 ProvoOrem, UT 5739092.587 Myrtle BeachNorth Myrtle BeachConway, SC 5893299.106 Jefferson City, MO 5893758.95 Wilmington, NC 6381538.472 VallejoFairfield, CA 6673117.627 Bellingham, WA 6861615.549 Rapid City, SD 6876756.725 Lincoln, NE 7040344.305 Corpus Christi, TX 7044727.672 Kingston, NY 7195646.507 Modesto, CA 7335037.953 Lake Charles, LA 7427002.816 Wichita Falls, TX 7730828.923 Fargo, NDMN 8032619.142 Cape GirardeauJackson, MOIL 8125529.637 Eau Claire, WI 8164241.999 ShreveportBossier City, LA 8201660.809 Bowling Green, KY 8281067.042 GreenvilleMauldinEasley, SC 8365311.68 VisaliaPorterville, CA 8689482.16 Albuquerque, NM 8706077.933 Vineland Millville Bridgeton, NJ 9176258.188 Stockton, CA 9237966.308 Monroe, LA 9334045.257 Bismarck, ND 9497884.231
89 Table 3 3 Continued Area Net Redistribution Hagerstown Martinsburg, MDWV 9620206.404 Binghamton, NY 9625229.529 Santa Cruz Watsonville, CA 9681679.512 Oxnard Thousand Oaks Ventura, CA 9840122.57 Bakersfield Delano, CA 10087362.87 Redding, CA 10386085.06 Springfield, MO 10733909.13 Billings, MT 11517534.17 HoustonSugar LandBaytown, TX 12264557.23 RenoSparks, NV 12266573.96 Roanoke, VA 12511645.57 Duluth, MNWI 13016633.85 EugeneSpringfield, OR 13099576.44 MuskegonNorton Shores, MI 13363629.44 Jackson, MS 13469221.03 Joplin, MO 13578496.46 Topeka, KS 13707229.88 Santa RosaPetaluma, CA 13822069 Asheville, NC 13945583.75 Greenville, NC 14925281.15 Sioux Falls, SD 15053124.61 El Paso, TX 15069647.84 AlbanySchenectadyTroy, NY 15131331.37 BrownsvilleHarlingen, TX 15423290.64 Lafayette, LA 16009843.61 Santa AnaAnaheimIrvine, CA Metropolitan Division 16590459.36 Virginia BeachNorfolkNewport News, VANC 17593366.85 Fayetteville, NC 17969633.22 GreensboroHigh Point, NC 18277030.94 LansingEast Lansing, MI 18479131.42 Salinas, CA 19087912 TrentonEwing, NJ 19137222.94 Spokane, WA 19289120.06 HuntingtonAshland, WVKYOH 19412540.27 Fresno, CA 21217941.07 Wichita, KS 22091386.15 McAllenEdinburgMission, TX 22846031.39 Rochester, MN 23425724.49 St. Cloud, MN 24959741.81
90 Table 3 3 Continued Area Net Redistribution San DiegoCarlsbadSan Marcos, CA 26084464.82 San FranciscoSan MateoRedwoo d City, CA 26484578.5 MilwaukeeWaukeshaWest Allis, WI 28265146.04 San AntonioNew Braunfels, TX 28926825.33 Grand RapidsWyoming, MI 29655086.92 New OrleansMetairieKenner, LA 30159611.19 Atlantic CityHammonton, NJ 31342860.13 RaleighCary, NC 32931538.52 West Palm Beach Boca Raton Boynton Beach, FL 33526880.76 Sacramento Arden Arcade Roseville, CA 34001755.38 Lexington Fayette, KY 34383227.11 San Jose Sunnyvale Santa Clara, CA 34487344.65 Tacoma, WA Metropolitan Division 38109150.54 Fort Lauderdale Pompano Beach Deerfield Beach, FL Metropolitan Division 40980664.24 Bethesda Rockville Frederick, MD Metropolitan Division 43506333.01 Rochester, NY 43787141.6 PoughkeepsieNewburghMiddletown, NY 45304963.5 Syracuse, NY 46716348.62 RiversideSan BernardinoOntario, CA 50586414.72 OmahaCouncil Bluffs, NEIA 56982217.36 OaklandFremontHayward, CA Metropolitan Division 59830186.99 Flint, MI 63234492.53 CharlotteGastoniaRock Hill, NCSC 71373890.42 MiamiMiami BeachKendall, FL Metropolitan Division 71452532.35 BuffaloNiagara Falls, NY 73221163.09 LouisvilleJefferson County, KYIN 85467087.45 Los AngelesLong BeachGlendale, CA Metropolitan Division 95582208.77 Las VegasParadise, NV 103789304.6 Camden, NJ Metropolitan Division 110958946.8 Kansas C ity, MOKS 113380958.4 DetroitLivoniaDearborn, MI Metropolitan Division 123314934 WarrenTroyFarmington Hills, MI Metropolitan Division 136856572.1 BaltimoreTowson, MD 162391504.8 St. Louis, MOIL 167007047.3 MinneapolisSt. PaulBloomington, MNWI 171036721.2 NewarkUnion, NJPA Metropolitan Division 179919099.7 EdisonNew Brunswick, NJ Metropolitan Division 234436427.3 NassauSuffolk, NY Metropolitan Division 302690032.9 New YorkWhite PlainsWayne, NYNJ Metropolitan Division 1261145975
91 Figure 3 1. Comparison to CMS wage i ndex F igure 3 2. Distribution of the r atio of CMS i ndex to OES i ndex
92 APPENDIX A DERIVATION OF EQUILIBRIUM WAGE OF LOW SKILLED WORKERS ( Simplifying equation ( 1 29 ) results in the quadratic equation given by =0 (A.1) Using the quadratic formula we can solve for the equilibrium value of Where a= <0 since the first term in the bracket is positive (A.2) b= (A.3) c= >0 (A.4) We know since ; where The discriminant so there are real solutions to Ruling out a negative wage for low skilled labor the only solution for is given by where a, b and c are as denoted above (A.5)
93 APPENDIX B DERIVATION OF PROPOSTION 2 The signs of the partial derivatives and are calculated first which will then determine the sign of >0 (B.1 ) In the inequality listed in (A.5) the first expression is negative since and and the second expression is also negative. (B.2 ) + >0 (B 3 ) The above inequality holds since given that since
94 LIST OF REFERENCES Albouy, D of life Acemo glu Daron 2002. Journal of Economic Literature Vol. 40, No. 1 (Mar., 2002), pp. 7 72 Autor D, Dorn D Skill Service Jobs and the Polarization of American E conomic Review 103(5): 1553 1597 Autor, David, and David Dorn 2010. Low Autor, David, Katz, Lawrence, and Melissa Kearney 200 6. US Labor American Economic Review Papers and Proceedings Vol. 96, No.2 (May 2006),189 194 Autor, David, Katz, Lawrence, and Melissa Kearney 2008. Review of Eco nomics and Statistics Vol. 90 (May 2008), 300 23. Beeson, Patricia E .1991. "Amenities and Regional Differences in Returns to Worker Characteristics." Journal of Urban Economics September 1991, 224 41. Beeson, Patricia E. and Randall W. Eberts 1989. "Iden tifying Productivity and Amenity Effects in Interurban Wage Differentials." Review of Economics and Statistics August 1989, 443 52. Blomquist, Glenn C., Berger, Mark C., and John P. Hoehn 1988. "New Estimates of Quality of Life in Urban Areas." American Economic Review March 1988, 89 107. Borsook, I Journal of International Economics 22, 281 295. Christensen, Bent Jesper, Rasmus Lentz, Dale T. Mortensen, George R. Neumann, and Axel Werwatz. the Job Search and the Wage Journal of Labor Economics 23 ( 1 ) : 31 58. Cowell, Frank. Handbook of Income Distribution. Vol. 1. ed. A.B. Atkinson and F. Bourguignon, 87 166. Amsterdam: Elsevier Science, North Holland. CLEAR (Council on Licensure, Enforcement, Regulation).2012 http://www.clearhq.org/mission.
95 David H. Autor & Frank Levy & Richard J. Murnane 2003. "The Skill Content Of Recent Technological Change: An Empirical Exploration," The Quarterly Jo urnal of Economics, MIT Press, vol. 118(4), pages 1279 1333, November. Dickie, M., Gerking, S. 1989 Migration and Labor Market Adjustment. Kluwer Academic, Dordrecht, Netherlands, pp. 111 145. DuMond, J.M., Hirsch, B.T., Macpherson, D.A Economic Inquiry 37 (4), 577 598 Eberts, Randall W. and Mark E. Schweitzer .1994. "Regional Wage Convergence and Divergence." Economic Review Federal Reserve Bank of Cleveland, 1994:2, 26 37. Epifani P. and Gancia G.A ., 2008. Economic Journal 118, 927 960 Falvey Rod, David Greenaway, Joana Silva .(2011) Trade liberalization and human capital adjustment, Journal of International Economics Volume 81, Issue 2, July 2010. Feenstra, Robert C Advanced International Trade: Theory and Evidence Princeton Univ ersity Press. Findlay, R., Kierzkowski, H ., 1983. International trade and human capital: a simple general equilibrium model Journal of Political Economy 91, 957 978. Gerking, S.D., Weirick, W.N. diff Review of Economics and Statistics 65 (3), 483 487. Gittleman M, Kleiner M. M, Research. Working Paper 19061 Glaeser, Edwa rd L 2008. Cities, Agglomeration, and Spatial Equilibrium. Oxford University Press. Glaeser, E.L., Mar, D Journal of Labor Economics 19 (2), 316 342. Goos, Maarten, Alan Manning and Anna Salomons 2009. he European American Economic Review Papers and Proceedings May 2009, 58 63.
96 Daniel J. Gottlieb, Weiping Zhou, Yunjie Song, Kathryn Gilman Andrews, Jonathan S. Skinner and Jason M. Sutherland Health Affairs 29 3: 537 543 Graves, Phillip E. 1976 Journal of Regional Science 16, 107 112. Hanushek, Er ic 1973. "Regional Differences in the Structure of Earnings." Review of Economics and Statistics May 1973, 204 13. Hwang, M., Quigley, J. 2006. Economic fundamentals in local housing markets: Evidence from US metropolitan regions Journal of Regional Science 46, 425 453. Kleiner M.M, Krueger A.B 1080, pp.676 687 Kleiner M.M, Krueger A.B Lic Journal of Labor Economics 31(2) pp. S173 S202 Lemieux, Thomas 2008 Journal of Population Economics 21(1), January, 21 48. Lemieux, Thomas 2008 Lemieux Thomas, W. Bentley MacLeod and Daniel Parent 2009. and Wage Inequality 124(1):1 49. Mazzolari F, R agusa G Skill Consumption to Low Skill Review of Economics and Statistics 95(1): 74 86 Mortensen, Dale T., and Christopher A. Pissarides. R eview of Economic Studies 61 ( 3 ) : 397 415. Mortensen, Dale T. 2005. Wage Dispersion: Why Are Similar Workers Paid Differently? Cambridge MA: MIT Press. Moscarini, Giuseppe. Econometrica 73 ( 2 ) : 481 516. Roback, Jennifer Journal of Political Economy December, 90 pp. 1257 1278 Roback, Jennifer Economic Inquiry ,26, pp. 23 41.
97 Rosen, S., Straszheim, M. (Eds.), Current Issues in Urban Economics Johns Hopkins University Press, Baltimore, pp. 74 104. Rosen, S. Handbook of L abor Economics Volume 1, Chapter 12 pp. 641 692. Roy, A. D Oxford Economic Paper 3 (2), 135 146. Sahling, Leonard G. and Sharon P. Smith 1983. "Regional Wage Differentials: Has the South Risen Again?" Review of Economics and Statistics February 1983, 131 35. Scully, Gerald W. 1969. Interstate wage differentials: A cross section analysis. American Economic Review 59: 757 73. Sullivan, P choices, and International Economic Review 51 (1), 289 317. Taylor Lor School Working Paper #540 Welch, F. 1969 Journal of Human Resources 4(3): 311 327 Winters, JV. Regional Science and Urban Economics 39, pp 632 643
98 BIOGRAPHICAL SKE T C H Achala Acharya was born in Kathmandu, Nepal. Achala graduated summa cum laude from Minnesota State University Moorhead, in 2008 with a major in economics and minor in m athematics. In August 2011 she earn ed a Masters of Arts degree in e c onomics from the University of Florida. She received her PhD. in economics from the University of Florida in the spring of 2014. Her research interest includes international trade, urban economics and labor economics.
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID ER5O7SE57_2J4MR3 INGEST_TIME 2014-10-03T22:12:24Z PACKAGE UFE0046489_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC