INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6” x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. Bell & Howell Information and Learning 300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA u i v l i 800-521-0600 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. DISSERTATION THE IMPACT OF IM M IGRATION ON LOW -SKILLED NATIVES Submitted by Hannes Johannsson Department of Economics In partial fulfillment of the requirements for the Degree of Doctor of Philosophy Colorado State University Fort Collins, Colorado Summer 1999 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 9947967 Copyright 2000 by Johannsson, Hannes All rights reserved. ___ ® UMI UMI Microform9947967 Copyright 2000 by Bell & Howell Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. Bell & Howell Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. COLORADO STATE UNIVERSITY FEBRUARY 17, 1999 WE HEREBY RECOMMEND THAT THE DISSERTATION PREPARED UNDER OUR SUPERVISION BY HANNES JOHANNSSON ENTITLED "THE IMPACT OF IMMIGRATION ON LOW-SKILLED NATIVES" BE ACCEPTED AS FULFILLING IN PART REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY. Committee on Graduate Work m Advisor Department Chair- -ii- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT OF DISSERTATION THE IMPACT OF IMMIGRATION ON LOW-SKILLED NATIVES How do immigrants in the United States affect the labor market outcomes of less- skilled native workers? This dissertation addresses this question by analyzing first the effect low-skilled immigrants have on low-skilled natives in 70 metropolitan statistical areas, and then on 25 cities in which more than a quarter o f the low-skilled population are immigrants. Specifically, I evaluate the impact immigrants with less than high-school education have on their native counterparts, as these individuals are most vulnerable to such labor supply shocks. Pooled cross-section data from 1994 to 1997 from the March supplements to the Current Population Surveys of the U.S. Census are the basis for the study. This research extends the current literature in several ways. First, using a fixed- effects model on pooled cross-sectional data provides increased efficiency of the estimators and thus more reliable results. Second, this study focuses on the short-run impact o f immigration by looking at only a four year period, 1994-1997. This is important as it reduces the problem associated with factor mobility commonly observed in the migration literature. As immigrants arrive into an area, natives may simply drop out o f the labor force and/or move out of the area. This potential departure out of the -iii- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. local labor force clearly may diminish the actual impact associated with immigration. Third, in addition to looking at the impact of immigration on annual income and unemployment, this study looks at labor force participation rates as well. This is a significant addition as it provides a potential explanation as to why previous studies have failed to find a consistent negative impact on income and employment. The results o f this research indicate that immigrants have a statistically significant negative impact on the labor market outcomes of low-skilled native workers. More specifically, immigrants reduce the labor force participation rates of natives and as the immigrant intensity increases, the impact is compounded. The impact on annual income is also relatively large although not statistically significant. The results on unemployment rates are positive, suggesting that immigrants actually reduce the unemployment rates of natives, which is consistent with what one would expect given the large observed reduction in natives’ labor force participation. This result is even more profound given the time period of the study. The fact that from 1994-1997 the United States experienced sustained economic growth suggests that low-skilled immigrants do indeed have an adverse impact on low-skilled natives. Hannes Johannsson Economics Department Colorado State University Fort Collins, Colorado 80523 Summer 1999 -iv- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGMENTS I would like to thank my advisor, Dr. Steven Shulman, for his support and guidance throughout the process of writing this dissertation. He provided me with assistance and encouragement while allowing me to explore the subject matter on my own terms. My committee members, Dr. Alexandra Bemasek, Dr. Nancy Jianakoplos, Dr. Stephan Weiler, and Dr. Jun Xing, were also indispensable help throughout the process. I would especially like to thank Stephan for his uncounted hours o f help and words o f encouragement. This dissertation would not have been completed, or even started, without the unqualified support o f my parents, Johann horir Jonsson and Sigridur Vilhjalmsdottir. They gave me the necessary optimism, perseverance, and humor to follow my heart and desire to complete this degree. My gratitude will never be expressed enough. The only regret that I have, now that this long process is finally over, is that I am not able to celebrate this accomplishment with my father who was so instrumental in getting me to where I am. Pabbi, I miss you! I would like to thank my wife, Beth, for staying with me through this long and tedious process. I know it has not been easy to be around me for the last few years but hopefully it will become easier now that we can finally move on with our lives. I am truly blessed to have such a wonderful wife and a mother for our beautiful daughters. -v- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Finally, this process would not have been completed without the support of the friends that I have made during this process. Specifically I need to thank Steve Byers for his friendship and encouragements along the way. Although his encouragements were annoying at times (especially after he got done), they were well appreciated. Making a friend like Steve, makes going through the process o f a Ph.D. worth while. Thanks for allowing me to enjoy the process. I dedicate this dissertation to the memory of my father. Pabbi, we did it! Sena til nada sol er gengin svefninn strykur |n'na bra. Syng eg Ijod vid litia drenginn ljufur Gud Jjer vaki hja. Augunn |n'n fogru aftu r lokast elsku bamid sofdu rott. Nottin yfir (Mtgul (jokast pagnar Ijod mitt - goda nott. (Johann f>. Jonsson) Audur vex af visku best viska a f boka lestri lesturinn (ivi Ijosast sest l a rum a f alud mestri. (Johann l>. Jdnsson) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table of Contents ABSTRACT OF DISSERTATION .....................................................................................iii ACKNOWLEDGMENTS ..................................................................................................... v LIST OF FIGURES .............................................................................................................. x LIST OF TABLES ............................................................................................................... xi Chapter 1: INTRODUCTION ..........................................................................................1 Focus of the Research ................................................................................................1 Hypothesis ................................................................................................................ 6 Methodology ............................................................................................................7 Data .......................................................................................................................... 8 Policy Implications .................................................................................................10 Synopsis of the Dissertation ................................................................................... 12 Chapter 2: IMMIGRATION AND LABOR MARKET ADJUSTMENT .................. 14 Introduction ..............................................................................................................14 The Labor Market Adjustment Process ...................................................................17 Area Studies .............................................................................................................19 Altonji and Card’s Study .............................................................................19 Butcher and Card’s Study .......................................................................... 24 Easton’s study .............................................................................................27 Card’s Study ............................................................................................... 29 Macro Studies .........................................................................................................31 Policy Perspectives ................................................................................................. 33 Summary ................................................................................................................. 34 Chapter 3: METHODOLOGY AND DATA ...............................................................37 Introduction ............................................................................................................. 37 Analytical Framework ..............................................................................................39 The Model .................................................................................................. 39 The Dynamic Effects of Immigration......................................................... 43 The Empirical Approach .........................................................................................48 The Empirical Model ...................................................................................49 -vii- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Data ........................................................................................................................... 51 Dependent Variables ................................................................................... 52 Independent Variables .................................................................................54 Sample Size .................................................................................................60 Conclusion ............................................................................................................... 61 Chapter 4: EMPIRICAL FINDINGS AND DISCUSSION........................................ 64 Introduction ..............................................................................................................64 Descriptive Statistics ...............................................................................................65 The Larger Data Set .................................................................................... 66 The Smaller Data Set .................................................................................. 68 Results ...................................................................................................................... 70 The Larger Data Set .................................................................................... 71 Income .............................................................................................72 Unemployment ................................................................................76 Labor Force Participation R a tes ..................................................... 80 The Smaller Data Set .................................................................................. 83 Income .............................................................................................83 Unemployment ................................................................................86 Labor Force Participation Rates ..................................................... 89 Interpretation of Results...... ......................................................................................93 Chapter 5: INTERPRETATIONS AND IMPLICATIONS......................................... 97 Introduction ..............................................................................................................97 Immigration Policy Prior to 1965 ............................................................................100 Immigration Policy Since 1965 ............................................................................... 101 Consequences of the Post -1965 Policies................................................. 103 Theoretical Consideration.........................................................................................105 Immigrants as Substitutes in Production.................................................. 106 Immigrants as Complements in Production..............................................109 Results and Implications..........................................................................................110 Chapter 6: CONCLUSIONS AND FUTURE RESEARCH...................................... 114 Purpose of the Research..... ....................................................................................114 Summary o f Results ...............................................................................................115 Policy Implications .................................................................................................117 Future Research ..................................................................................................... 119 APPENDIX A: CATEGORIES OF ADMITTANCE....................................................... 121 APPENDIX B: PERCENT OF IMMIGRANTS ADMITTED BY REGION AND PERIOD ..................................................................................................123 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX C: IMMIGRANTS ADMITTED BY TOP TWENTY COUNTRIES OF BIRTH IN 1994 ......................................................................................124 APPENDIX D: IMMIGRANTS ADMITTED BY DECADE AND SELECTED COUNTRIES OF ORIGIN .................................................................... 125 APPENDIX E: LIST OF CITIES AND CENSUS CODES (All 70 Cities)................... 126 APPENDIX F: CITIES AND CENSUS CITY CODES (TOP 25 CITIES)................... 128 APPENDIX G: SUMMARY OF LITERATURE R ESU LTS....................................... 129 SELECTED BIBLIOGRAPHY .......................................................................................131 -ix- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. List of Figures FIGURE 3.1: LABOR SUPPLY AND DEMAND DIAGRAM....................................... 46 FIGURE 4.1: IMPACT OF IMMIGRATION ON NATIVES......................................... 91 FIGURE 5.1: LABOR DEMAND SHOCK AND NATIVES’ INCOM E...................... 107 -x- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. List of Tables TABLE 3.1 DESCRIPTION OF DEPENDENT VARIABLES ........................................57 TABLE 3.2 DESCRIPTION OF INDEPENDENT VARIABLES - CITY LEVEL.......58 TABLE 3.3 DESCRIPTION OF INDEPENDENT VARIABLES - INDIVIDUAL.......59 For The Larger Data Set: TABLE 4.1 RESULTS ON ANNUAL INCOME - CITY L E V E L .................................. 74 TABLE 4.1.1 RESULTS ON ANNUAL INCOME - INDIVIDUAL...............................75 TABLE 4.2 RESULTS ON UNEMPLOYMENT RATES .............................................. 79 TABLE 4.3 RESULTS ON LABOR FORCE PARTICIPATION.....................................82 For The Smaller Data Set: TABLE 4.4 RESULTS ON ANNUAL INCOME - CITY L E V E L .................................. 84 TABLE 4.4.1 RESULTS ON ANNUAL INCOME - INDIVIDUAL...............................85 TABLE 4.5 RESULTS ON UNEMPLOYMENT RATES .............................................. 88 TABLE 4.6 RESULTS ON LABOR FORCE PARTICIPATION R A T E S.....................92 -xi- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 1 INTRODUCTION McCarran Act "The blood that made this nation great Will now be tested at the gate To see i f it deserves to be Admitted to democracy. Or rather to that small elite Whose hemoglobin counts can meet Requirements o f purity Consistent with security And with that small and rabid mind That thinks itself above mankind." The Reported. July22, 1952 Focus of the Research The enactment of the 1965 Amendments to the Immigration and Nationality Act caused a disproportionate increase in the immigration o f less-skilled workers. The 1965 Amendments replaced the existing quota system with a family reunification emphasis and thereby dramatically altered the immigrant population. The primary purpose of the 1965 Amendments was to abolish discriminatory quotas and make immigration policy more - 1- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. humane. Immigration policy prior to the 1965 Act was primarily a quota-based system which granted the opportunity to immigrate to the United States to people from a few selected countries. For example, people from Asia were not allowed to immigrate to the United States while individuals from countries in the Western Hemisphere were exempt from the quotas and faced no numerical restrictions. Congress passed the 1965 Amendments at a time when the civil-rights movement was at its peak, and the legislation can be interpreted as one of a series o f civil-rights statues that were enacted during that period. In particular, the 1965 Amendments abolished the discriminatory national-origins quota system. In combination with subsequent revisions in the law, the Amendments permitted the entry of 270,000 persons per year, with no more than 20,000 immigrants originating in any particular country of origin (Boijas, 1990). Appendix A summarizes the main components o f current law and reports the number of legal immigrants admitted in 1994 under the various provisions. This change in immigration policy severely changed the national composition of the immigrant flow. Whereas earlier immigrants tended to be mostly o f European descent and were relatively skilled, more recent immigrants tend to originate from Asia and Mexico and are relatively less skilled (Boijas, 1990). See Appendixes B, C, and D for changes in both the size of the immigrant flow and the national composition since the implementation of the 1965 Amendments. Although the apparent change in the immigrant population may in part be due to changing political and economic conditions, the 1965 Amendments clearly play a significant role in the observed change. The record number of immigrants arriving during - 2- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the 1970s and 1980s revived public interest in immigration issues. Particularly, the public became increasingly concerned with the impact of the increased influx of immigrants on labor market outcomes. Though much theoretical work has been done on the labor market impact o f immigration, the first empirical study did not surface until 1982. Grossman’s 1982 study on the substitutability of immigrant and native workers represented the first empirical study o f the impact o f immigration on the wages and employment of natives. Since then, several empirical studies have appeared on the potential adverse impact o f immigration on the labor market outcomes o f native workers. Most empirical studies to date have emphasized the impact immigrants have on local labor markets. These studies have typically taken a cross-section o f standard metropolitan statistical areas (SMSA). Assuming that immigrants and natives are substitutes in production, one would expect to a find negative impact on the labor market outcomes of native workers. Furthermore, in light of the fact that immigration is very area specific, one would expect that these area studies would be more likely to find any adverse impact on labor market outcomes than would macro studies. It is peculiar that these area studies have failed to find any consistent negative impact associated with increased immigration. Some in fact have found that immigrants yield a positive impact on both wages and employment of natives, suggesting that immigrants and natives are complements in production (see Boijas, 1994). More recent studies, however, have criticized the area studies for their overly simplistic assumptions. -3- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The area studies have been noted for making two crucial assumptions: i) the location choice of immigrants is exogenous to the local labor market conditions, and ii) once immigration takes place, the local labor market represents a closed economy. The assumption that the location choice of immigrants is exogenous to the local labor market conditions suggests that immigrants do not take into account the presence o f people from their home country, their chances at getting a job, or their earnings potential as they make their location decision. As noted in recent literature, this assumption is clearly unrealistic and may bias any empirical results. This study uses the fixed-effects model to help capture any systematic differences across cities that may contribute to location choice. By doing so, this study reduces any biases this assumption introduces. The assumption that the local labor market represents a closed economy once immigration takes place is a rather ambiguous one. It is quite apparent that if natives respond to the arrival o f immigrants by migrating out o f the area, the impact o f the immigration will be seriously understated. Also this simplifying assumption will alter the results if natives migrate out of high immigrant areas, or if natives that might otherwise migrate into a high immigration area decide not to do so. Frey (1994), for example, has found that natives are more likely to migrate out of, and less likely to migrate into, high immigrant areas. Although this study does not take into account the potential out­ migration directly, it does so indirectly. By capturing the short-run impact o f immigration, the problem associated with out-migration is reduced. According to -4- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Blanchard and Katz (1992), it takes the labor market about four to eight years to recover from an adverse labor demand shock. By capturing the impact associated with immigration prior to the completion o f the adjustment process, the potential bias introduced by the closed economy assumption is reduced. This research focuses on the short-run nature o f immigration and thus partially accounts for the closed economy assumption by using a pooled cross-section, time-series data, over a relatively short time period, 1994 - 1997. Although other studies have attempted to account for the migratory response o f natives, those studies have focused on the national impact as opposed to the local impact. It seems implausible, however, that the ‘macro’ approach will yield reliable results given the fact that immigrants represent only a small fraction of the domestic labor force, while in some cases they represents a large fraction o f the growth in a given city’s labor force. This especially true when attempting to evaluate the impact on a specific subgroup within the labor force. One of the byproducts of the implementation of the 1965 Act is that the skill level o f immigrants became bipolar. Today’s immigrants are either highly educated or high school drop outs (less than 12 years o f education). Given this bipolar skill distribution, any attempt to evaluate the impact of increased immigration that does not take into account these vast skill differences be inherently biased. Furthermore, any study that uses the fraction o f immigrants in the local population as an independent variable will undoubtedly have biased results. This bias stems from the change in the immigrant population since the 1965 Amendments. If in fact immigrant skills have been -5- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. deteriorating since the passage of the Amendment, including immigrants that arrived prior to the implementation o f the Amendment will understate any potential impact. To summarize, this research corrects some of the problems that weaken the current immigration literature. Specifically, potential biases introduced by factor movements are reduced by looking at a shorter period of time. Any skill changes in the immigration population that may have resulted from the 1965 Amendments are accounted for, by including only immigrants that have arrived since 1965. Additionally, this study evaluates the short-run impact o f immigration as opposed to a cross-Census approach which is a long-term impact evaluation. Also, the increasing number o f immigrants in the population is evaluated to see if the labor market outcomes o f natives in a given city changes as the immigrant intensity increases in the city. This is accomplished by first evaluating the seventy cities in which the fraction of low-skilled immigrants is the highest, and then by limiting the sample to the twenty-five cities in which more than a quarter of the low-skilled population are immigrants. A complete list of cities included in this research are listed in Appendices E and F. All individuals in this study are high school drop outs, which reduces problems associated with the more recent immigrant skill distribution, and all the immigrants in the sample arrived in the United States after the implementation of the 1965 Amendment. Hypothesis It is the hypothesis of this study that the increasing number o f less-skilled immigrants arriving in the United States have an adverse impact on the real incomes and - 6- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. employment opportunities o f less-skilled natives. The true impact of immigration should be found in area studies that properly identify the issues at hand. For example, the impact o f immigration will differ according to the economic conditions in a given city. Immigrants arriving during a recessionary period will cause a greater adverse impact than immigrants arriving during an economic boom. Additionally, it is apparent that labor market conditions will differ across metropolitan areas and time. Any study that does not allow for differences in location and time periods will surely come up with very ambiguous results. Not allowing for the possibility that natives may respond to the influx o f immigrants by simply migrating out of the metropolitan area will undoubtedly underestimate the true impact associated with immigration. Methodology The empirical analysis used in this study diverges considerably from the methods used in previous area studies. Instead of using the standard cross-sectional analysis, this study utilizes the fixed-effects model. The fixed-effects model provides for the opportunity to evaluate cross-sectional time-series data which in turns offers the possibility to perform short-run analysis of the impact o f immigration. Also, using the fixed-effects model provides the opportunity to allow for structural differences across cities. That is, the local labor market conditions are not required to be the same across cities for this analysis. Aside from the differences in methodology, the regression variables used are consistent with those used in other immigration studies. The standard control variables o f -7- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. race, education, age, and gender are included to account for local labor force characteristics. Any systematic differences in labor markets across cities is captured in the fixed-effects. Additionally, time dummies are used to account for changes in cities over time and the cities are allowed to experience differences in economic conditions over the time period o f the study. This is important as it is unlikely that different cities’ labor markets grow or experience the same changes over time. Data The data used here are drawn from the March supplements to the Current Population Surveys of the Census Bureau. The study covers the time period from 1994 to 1997. The dependent variables are: real annual income (i.e. the natural logarithms of annual income), labor force participation rates (in natural logarithms), and the unemployment rate (not in natural logarithms) o f low-skilled native workers. The income variable is in natural logarithms so that it captures the earnings profile o f individuals. This is a standard practice in the labor literature. The analysis was performed on two different sets of variables. First, the regression was performed at the city level (i.e. city averages), and then at the individual level. Because of the way the questions on "weeks worked" and "weeks looking for work"1 are asked in the March supplements, only one regression was performed at the 1 These questions did not ask "how m any did you work last year", but rather "weeks was .... looking for work or on layoff from a job" and "during 19.. In how many weeks did .... work even for a few hours, include paid vacation and sick leave as work," without specifying the time period. - 8 - Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. individual level. Because the Census Survey did not phrase these questions properly, truncation problems were introduced. In light o f this shortcoming o f the Survey, the only dependent variable at the individual level is annual income. At the city level, however, all three dependent variables are used. It is important to note that the unemployment rate reported here is not the official unemployment rate in a given city for a given year. This study specifically evaluates the impact on low-skilled natives, which implies that the unemployment rate reported here is only relevant to this subgroup. Since some of the cities report having only a few low-skilled people, it is possible to have either full employment or no one employed in a given city in a given year. If the data are collected properly, this should not pose any serious problem to the analysis, because as long as the data represent a true random sample of the general population, the reliability of the results should not be impacted. Since some of the cities in the study report a zero unemployment rate for some of the years, the unemployment rate is not in logarithms. The immigration variable is defined as immigrants with less than a high school diploma who have arrived since 1970. An immigrant in this study is defined as a foreign-born person who identified themselves as such in the 1994-1997 March supplements to the Current Population Surveys. A clear shortcoming of the data is how an immigrant is defined. As the Census does not differentiate between legal and illegal immigrants, this study does not attempt to make that difference either. Since illegal immigrants (or illegal aliens as they are not technically immigrants) may choose not to identify themselves in the Census for obvious reasons, the data only includes some of the illegal population presently residing in the -9- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. United States. As illegal aliens are not immigrants, including them in an analysis of immigration policy will potentially overstate the adverse impact (or understate the positive impact) in the analysis. Also, some selection bias may occur due to the fact that the illegal alien sample may not be a representative sample o f the illegal population. This is a problem that any study using Census data faces. Policy Implications The question of what immigration policy should or should not be is a complex one. Since the implementation of the 1965 Amendments to the Immigration and Naturalization Act, several adjustments have been made. These adjustments have almost without exception been geared towards increasing the number of high-skilled immigrants, i.e., immigrants with twelve or more years of education. The first significant immigration policy change since the 1965 Act came in 1986. The 1986 Immigration Reform Act gave illegal aliens residing in the U.S. prior to 1982 the opportunity to become legal residents. In 1990, the immigration policy was expanded to allow for more skilled immigrants. Any changes since 1990, including the 1990 adjustment, have focused on allowing more skilled immigrants to enter the United States rather than decreasing the number of admitted low-skilled migrants. Should immigration be limited to only highly skilled individuals or should it take more of an account of humanitarian perspectives as the 1965 Act did? The fact that this analysis evaluates only a small portion of the immigration issue, it is not the purpose of this study to make judgement calls on what U.S. immigration policy should or should not - 10- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. be. However, the study performed here will shed some additional light on the policy debate. In order to suggest specific policy measures, a more comprehensive evaluation of the impact o f immigration is warranted. For example, a recent study by the National Research Council concludes that immigration yields a positive net economic gain to the native-born (NRC, 1997, pp. 219). Although the NRC report concludes that the overall impact o f immigration is positive, does not mean that all natives benefit equally or that some subgroups of natives do not actually become worse off as a result o f immigration. This research will focus on evaluating the impact associated with the arrival o f low- skilled individuals on the labor market outcomes of their native counterparts, which is clearly only a small part of a bigger puzzle. Most immigration studies to date have demonstrated that immigrants tend to have little or no impact on the labor market outcomes o f natives. These studies may therefore be interpreted to suggest that the current immigration policy requires little adjustment. The continued influx of low-skilled labor and the clear bipolarization o f the immigrant flow, however, leaves one to wonder if change is not warranted. How much longer can the U.S. economy sustain the continued record numbers o f low-skilled immigrants? Does the impact o f low-skilled immigrants change as their intensity increases in a given locale? These are some of the questions answered in this study. If the continued massive influx o f low-skilled individuals leads to worsening labor market conditions for low-skilled natives, the current immigration policy may require some reevaluation. - 11- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Synopsis of the Dissertation The following is a synopsis o f the dissertation: Chapter 1: Chapter 2: Chapter 3: Chapter 4: The introduction includes an explanation o f the focus o f the research, the hypothesis and the methodology that is utilized. The Immigration and Labor Market Adjustment chapter gives the study its relevance by discussing the impact of a labor market demand shock on regional economies. As immigration is an example of a labor demand shock, this discussion is a lead into the immigration literature. The discussion of the literature focuses on the works that are most relevant to this research. The approaches taken by other area studies that have evaluated the impact o f low-skilled immigrants on their native counterparts are discussed, along with a reporting of their findings. Macro studies are also introduced and evaluated. The results o f these studies are also introduced and discussed. The Methodology and Data chapter discusses and justifies the particular empirical procedures that are utilized in this study. The fixed-effects model is explained along with a detailed discussion o f the data. The Empirical Findings and Discussion chapter evaluates the findings of the study by properly identifying the theoretical foundation underlying the immigration analysis. The results are evaluated in light of the outcomes o f other related studies. - 12- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 5: Chapter 6: The Interpretations and Implications chapter interprets the result and discusses potential implications on immigration policy. It discusses how labor markets adjust to labor demand shocks using immigration as a cause of that shock. The Conclusion summarizes the key findings of the research. Possible future research topics are also discussed and evaluated. -13- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 2 IMMIGRATION AND LABOR MARKET ADJUSTMENT “...the general history o f the United States disproves the theory that immigration tends to lower the wage scale. No country has admitted so large a number o f immigrants, yet no country has so high a wage scale or standard o f living. " William S. Bernard, Economic Effects o f Immigration Introduction The first empirical study of the impact of immigration on the labor market outcomes of natives was done by Jean Grossman in 1982. Since Grossman’s study came out, several other empirical studies have been done on the impact of immigration. Only four o f those studies, however, have concentrated on the impact on low-skilled native workers in local labor markets (Card, 1990; Altonji and Card, 1991; Butcher and Card, 1991, Easton, 1996). Because of the disproportional increase in the arrival of low-skilled immigrants, it is important to attempt to understand how these immigrants impact their closest substitutes, the low-skilled natives. Although other studies have been performed - 14- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. on the impact on low-skilled workers none of them are area studies (see, for example, Boijas, Freeman, and Katz, 1996; Jaeger, 1996). Since immigrants tend to cluster in certain cities, it is important to focus any study aimed at understanding the potential adverse impact o f immigration at the city level. Altonji and Card (1991), did an extensive study o f the impact o f immigration on the labor market outcomes of low-skilled natives. This study indicates that immigrants do indeed have an adverse impact on the wages o f low-skilled natives. Card (1990) and Butcher and Card (1991), find little impact on the wages o f low-skilled natives as a result of immigration. While all three studies represent area studies, Altonji and Card’s (1991) and Butcher and Card’s (1991) studies represent a cross-city evaluation. The results of these area studies are rather inconclusive. The failure o f these studies to come up with consistent results can perhaps be attributed to the rather ambiguous assumptions made in this literature. There are two critical assumptions in particular that are problematic: 1. the location choice of immigrants is exogenously determined, and 2. the metropolitan areas represent a closed economy once immigration takes place. The first o f these assumptions implies that immigrants arrive into a city at random. That is, the immigrants’ location choice is not driven to a significant degree by preexisting labor market conditions, such as wages and employment opportunities. The second assumption implies that native workers and capital do not respond to the influx of added labor by migrating out of the area in the case o f labor, into the area in the case of capital. Both Filer (1992), and Frey (1995) have found that native workers do indeed -15- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. react to the influx o f immigrants by migrating out o f the local area. Furthermore, Boijas, Freeman, and Katz (1996), found that less-educated natives are the ones who are more likely to react negatively to the influx of immigrants. In order to avoid the problems associated with the first assumption, some studies have used what has been called “natural experiments” or exogenously determined immigrant flow. A natural experiment is immigrant flow that is not based on local specific factors, such as wages and employment rates. Card’s study (1990) represents a natural experiment. He evaluated the impact of the 1980 Mariel Boatlift on Miami’s labor market. Another natural experiment was performed by Hunt (1992) on the impact of the 1962 Repatriates from Algeria on the French labor market. The advantage of a natural experiment is that the exogeneity assumption is validated. However, the closed economy assumption is not resolved in these studies. One o f the primary problems associated with current immigration studies is the assumption that the local economy represents a closed economy once immigration takes place. What this assumption suggests is that firms and native labor do not respond to the immigrant influx by moving into or moving out of the area. This is a rather ambiguous assumption as Blanchard and Katz (1992) found that both firms and labor respond to adverse labor demand shocks by migrating out o f the area (in the case o f labor), and migrating into the area (in the case of firms). This study attempts to avoid making this assumption by evaluating the impact associated with immigration over a very short time period, four years. By evaluating the impact of immigration over a relatively short time -16- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. period, it may be possible to capture the impact prior to the suggested adjustment taking place, and thus reduce any potential biases associated with this assumption. To better understand the problems associated with the closed economy assumption, it is necessary to go through the labor market adjustment process and how it relates to immigration in some detail. Following that discussion, the immigration literature as it pertains to the low-skilled labor market is examined and some of the main weaknesses o f these studies pointed out and evaluated. The Labor Market Adjustment Process The arrival of immigrants impacts the labor market of natives in terms of both labor demand and labor supply. Assuming that natives and immigrants are substitutes in production, the arrival of immigrants reduces the demand for natives workers. The impact o f this initial adverse labor demand shock will set into a motion an adjustment process comprised of firms moving into, and workers out of, the area. This adjustment mechanism is described in detail by Blanchard and Katz (1992). What they found is that it takes income in a local labor market up to a decade to recover from the initial shock, and five to seven years for the unemployment rate and the labor force participation rates to adjust. What these results suggest is that evaluation of the impact of immigration needs to focus on relatively short time periods in order to find any impact. To get reliable results on the impact of immigration, attempts must be made to capture the impact prior to the completion of the labor market adjustment. -17- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The adverse shock to labor demand drives down both wages and employment. The Blanchard and Katz model of labor market adjustment shows that two adjustment mechanisms come into play in response to an adverse shock in demand. First, the lower wages induce firms to move into the area as input costs are decreased. As firms enter the local market, new firms are created which will slow down the out-migration o f labor. Second, people will begin to migrate out of the area as they observe their wages and employment opportunities diminishing. It is this adjustment mechanism that makes it difficult to measure the impact associated with the arrival of immigrants. By focusing on a relatively short time span, however, one can better capture the impact by capturing it prior to the completion of the adjustment process. Although this is not an ideal way to account for the apparent labor market adjustment, it is clearly better than simply ignoring any such adjustment process. Most studies on the impact immigrants have on the labor market outcomes o f natives have failed to account for this labor market adjustment process. If natives do in fact migrate out of heavy immigration areas, as Frey (1995) and Filer (1992) have concluded, any study that does not account for the migratory response of natives will underestimate any impact associated with the arrival o f immigrants. Also, in light o f the relatively short labor market adjustment period, any study that does not successfully account for the short-run effect of immigration will further understate any potential impact. Thus, by capturing the impact prior to the completion of the labor market adjustment period, one can expect to produce more reliable results. -18- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Area Studies As mentioned above, the area studies conducted so far have relied on either cross- sectional analysis or on natural experiments. Neither of these approaches have provided convincing evidence of the impact o f immigrants on the less-skilled natives. To assist in understanding why these studies have been unsuccessful in finding any significant impact, a closer look is taken at each o f them, beginning with the cross-section studies. The results o f the area studies are summarized in Appendix G. Altonji and Card's Study Altonji’s and Card study has been considered by some to be the most comprehensive and theoretically sound study done on the issue. Their study uses variations in the fractions of immigrants across different cities to measure the effects of immigration on the labor market outcomes o f less-skilled natives. They assemble information from the 1970 and 1980 Censuses on labor market outcomes o f natives in 120 major cities. By using information from two consecutive Censuses, they are able to correlate changes in immigrant fractions with changes in native outcomes within cities. This is important because it enables them to abstract differences across cities that might bias a simpler cross-section analysis. Another interesting addition provided by their study is the information on the industry distributions of natives and immigrants and how these distributions have changed in cities with higher and lower immigrant shares. On a theoretical note, their study departs from earlier models in two ways. First, they disaggregate labor along skill lines rather than along the lines of national origin. -19- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Sedond, they allow for demand side effects associated with increases in the local population and for supply side effects associated with the possible crowding out of native workers in response to lower wage rates. Their model leads to a simple empirical specification in which the wage and employment outcomes o f less-skilled natives (either in cross-section or within cities over time) vary with the share and skill composition of immigrants in the local labor market. Altonji and Card’s empirical approach diverges from previous studies in at least two important ways. First, as stated above, they look at the changes in the outcome variables across Censuses. This enables them to look at the relationship between the change in immigrant shares and changes in outcome variables (wages, participation, and employment) across major cities. They furthermore compare the results from the first- difference analysis to that o f standard cross-sectional analysis. Although the results o f the two approaches vary somewhat, they argue that the first-difference analysis is less likely to be contaminated by city-specific factors that affect immigrant densities and native outcomes. The second important addition to the literature provided by this study is the usage o f an instrumental variable to account for the location choice of immigrants. For the four race/sex groups that they consider, the instrumental variable estimates suggest that an inflow of immigrants equal to one percent of the population of a standard metropolitan statistical area (SMS A) reduces average weekly earnings of less-skilled natives by about 1.2 percent (statistically significant). The least squares estimates, by comparison, imply that a one percent increase in the fraction of immigrants yields a modest 0.3 percent reduction (not statistically significant). - 20- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Altonji and Card’s basic empirical approach is reasonably straight forward. They regress SMS A averages of the labor market outcome variables for their four race/sex groups against measures of the immigrant fraction in the SMSA and a variety o f controls for the characteristics for each city. The first step in their analysis is to construct SMSA- specific means o f the outcome variables that are purged o f differences in the observable characteristics of the native population across different cities. Due to the limited information collected in the Census, this step amounts to regression adjusting the outcome variables for differences in age and education. Such an adjustment has two potential advantages: 1. it should reduce the sampling variation associated with the means o f the outcome variables across different cities; 2. it should eliminate any bias arising from correlations between the fraction of immigrants in a city and the age and educational attainment o f natives. For each race/sex group in each of the two Censuses, they regress each o f the outcome variables against a full set of SMSA dummies and a flexible function of age and education. Specifically, they include a cubic polynomial in age, a detailed set o f dummy variables for different education levels, and a full set of interactions of age and education up to the second order. They then use the estimated SMSA dummies as their regression- adjusted outcome measures. In their second step, they include as explanatory variables the fraction o f immigrants in each SMSA and three additional control variables: the logarithm of SMSA population, and SMSA-specific means of age and education for the particular race/sex - 21- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. group under consideration. Although their outcome variables are adjusted for age and education, they found in preliminary work that the mean o f adjusted weekly earnings is correlated across cities with the mean of education, particularly for blacks. Thus the cross-section regression equation has the form where YNj is the adjusted labor market outcome for native group N in city j , XNj- is a vector o f control variable for the race/sex group and city (the mean o f age and education for the group and the logarithm o f SMSA population), f j is the fraction of immigrants in the city, and e NJ is a residual term. Similarly, their first-difference estimating equations have the form where AZy refers to the change in the variable Z in city j between 1970 and 1980. Altonji and Card attempt to control for any potential correlation between immigrant inflow and local economic conditions in their first-difference analysis by including an instrumental variable. They follow Bartel’s work (1989) and use the fraction (1) Y nj ~ X Njb + f jC + e Nj, (2) A Y nj = b X f/Jb + A f j C + A e Nj, - 22- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. o f immigrants in a city in 1970 to predict the change in the fraction o f immigrants over the following decade. Bartel’s analysis (1989) suggests that economic conditions have relatively small effect on the destination city chosen by immigrants. Instead, Bartel’s findings suggest immigrants are mainly attracted to cities with a large concentration of previous immigrants from the same country (see also Greenwood and McDowell, 1986). One important short coming of this analysis is the fact that although dynamic issues are addressed, short-run evaluations are not performed. Although Altonji and Card do mention their suspicion that the short-run effects of immigration on employment of less-skilled natives will be larger than the long-run effects due to labor market adjustment, their model has no way of capturing this difference. As immigrants enter a city, natives are faced with increased competition for jobs. The increased competition for jobs drives down wages and causes some natives to lose their jobs (assuming immigrants and natives are substitutes in production). As natives lose their jobs and see that the monetary return for their skills has declined, they drop out o f the labor force and/or migrate out of the area. Furthermore, as labor costs drop, firms enter the area to take advantage o f these cost reductions. This labor market process causes the short-run or the initial impact o f immigration to be greater than the long-term impact. As Blanchard and Katz (1994) found, it takes a local labor market five to eight years to adjust to the initial labor demand shock. Thus first-differencing across Censuses is not an appropriate way to capture the short-run impact o f immigration. By using the first-difference across the two Censuses, one is using the change in the immigrant population and the change in native labor market -23- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. conditions over a ten year period which can hardly qualify as short-run. In light o f Blanchard and Katz’s study (1992), studying the labor market over a ten year period will not capture the primary impact associated with immigration. It should be quite clear that the primary impact o f immigration will be experienced in the short-run as opposed to in the long-run. In summary, although Altonji and Card’s study (1992) represents a dramatic improvement in the way in which the impact of immigrants on less-skilled natives is measured, it is hardly complete. Most of the impact felt by immigrants would be expected to be noted within the first 5 years or so. As immigrants enter an area, the labor market competition increases, causing people either to move out of the area or to become unemployed. Using the first-difference across Censuses as the basis for measuring the short-run impact of immigration is hard to accept. As immigration into an area begins (or increases), those who are (or expect to be) hurt the most will migrate out of the area relatively early, especially if they anticipate immigration into to the area to continue. Butcher and Card’s Study Butcher and Card (1991), using data from the merged files of the 12 monthly Current Population surveys administered in 1979, 1980, 1988, and 1989, found little impact on the wages of less-skilled workers as a result of immigration. They evaluated the impact of immigration on the hourly wages of less-skilled native workers living in 24 major cities. For each city and each year they calculated the 10th and 90th percentiles of the log wage distribution. Their empirical analysis revealed large differences across cities -24- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in the relative growth rates of wages for low- and high-paid workers. However, these differences do not appear to be related to the size of the immigrant inflows. Their results thus confirm the findings o f previous studies, based on 1970 and 1980 Census data, that suggest that immigration has little impact on the labor market outcomes o f natives. It is interesting to note that contrary to their expectations, their data suggests that higher immigration is associated with more rapid increases in the 90th percentile of wages, rather than with any relative decline in the 10th percentile o f wages. This is particularly interesting in light o f the educational distribution o f recent immigrants. As they note, the education distribution o f newly arriving immigrants is relatively dispersed: the fraction o f college graduates is about the same as in the native population, but close to one-quarter of recent immigrants have less than an elementary education. As a result, recent immigrants make up seventeen percent of the population in these cities with six of fewer years of schooling, and ten percent of the population with less than an eighth-grade education. The fact that the fraction of unskilled immigrants exceeds that of the native population while the fraction of skilled immigrants matches that of the native population would suggest that wages of skilled labor should increase while lowering the unskilled wage. That is, as the competition for jobs increases more for low-skilled workers than for high-skilled workers, one would expect the wages of low-skilled workers to decline relatively more than the wages o f the high-skilled. However, the raises in the wages of high-skilled workers could be attributed to the complementarity in production that exists between low- and high-skilled workers. As relatively more low-skilled workers enter the -25- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. labor market, the demand for high-skilled labor increases, driving their wages up while the increase in the supply o f low-skilled workers drives their wages down. The fact that high-skilled immigrants increased the supply o f high-skilled workers is offset by the fact that relatively more low-skilled workers entered the workforce as well, increasing the demand for high-skilled workers. The end result for the high-skilled workers depends therefore on the relative size o f the shift of each curve. I f the demand increases by more than the increase in supply, the net impact would be higher wages, which is what their study suggests. Butcher and Card’s major contribution to the literature is that they actually attempted to account for the potential difference in the cost o f living. They tested the possibility that the cost o f living rose more rapidly in high immigration areas, by constructing wage changes relative to city-specific cost o f living indexes using data from the Bureau of Labor Statistics. After adjusting for the local cost o f living, the resulting regression coefficient relating the change in the 10th percentile o f wages to the fraction of recent immigrants in 1985 became slightly negative. Thus after adjusting for city-specific changes in the cost o f living, they find less evidence of a positive correlation between immigration rates and the growth of high-skilled wages, and more evidence o f a negative correlation between immigration and the growth of low-skilled wages. Neither o f these correlations were large or statistically significant, however. -26- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Easton’s Study Easton (1996, unpublished manuscript) evaluated the impact o f low-skilled immigrants on nominal wages of low-skilled natives. This study corrects for the endogeneity problem by using a Two Stage Least Squares approximation (2SLS). This study also attempts to evaluate the short run impact associated with immigration by looking at the impact associated with immigrants arriving between 1985 and 1990. In doing this, he uses mostly data from the 1990 Census and analyzes the impact on the largest 86 cities (cities with population greater than 500,000). Easton uses a standard wage equation for employed natives in the sample, with three additional variables. His estimation equation includes human capital variables, along with dummy variables to measure a person’s race or ethnicity. It also includes a variable intended to measure the strength of labor demand and the cost o f living in each labor market. To accomplish this he uses a measure of the labor demand and living cost that is exogenous to the model of wage determination used. The metropolitan wage level variable is the coefficient of dummy variables identifying each metropolitan area included in a regression estimating wages earned in 1985. It measures average wages in a metropolitan area, relative to other metropolitan areas, controlling for the personal characteristics of people with wages. He chooses this variable as it is presumably influenced by the strength of labor demand, by living costs, and by the attractiveness of each location. The findings of his study basically provide no evidence that previous studies have missed the negative impact of immigration by examining immigration over a long time -27- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. horizon. He actually finds that native workers in 1989 in high immigration cities actually earned higher nominal wages than native workers in cities receiving less immigration. Although this positive impact may appear surprising at first, it may not be inconsistent with what one would expect given the possibility of out-migration o f natives. Easton’s study focuses on the short-run impact of immigration by evaluating the impact associated with recent immigrants. However, looking at the impact of recent immigrants is a rather ambiguous way to account for the short-run impact. In order to evaluate the short-run impact, one needs to look at the natives’ wages over a short period of time as opposed to a specific point in time, 1989 in this case. Another short coming of this study is the way that cities were chosen. Basing the analysis on the 86 largest U.S. cities does not necessarily capture the cities in which the majority of low-skilled immigrants reside. Thus the analysis may possibly understate the true impact by ignoring smaller cities where significant numbers of low-skilled immigrants may reside. The fact that this study fails to find a significant negative impact associated with immigration does not imply that no such impact exists. Like most other studies on the impact of immigration, this study focuses on the wages of natives which may not be the appropriate variable to evaluate. Although Easton’s study clearly provides a contribution into the current immigration literature, it falls short in capturing the reasons why immigrants appear to have little impact on natives’ wages. To be able to explain why most studies have been unsuccessful in their mission, one needs to look at other labor market variables, such as natives’ labor force participation rates. If natives do indeed respond to the arrival of immigrants by dropping out of the labor force or migrating out o f -28- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the area, the impact on both wages and employment will clearly be lost. Thus in order to evaluate the true impact associated with immigration, one must evaluate the impact on variables such as natives’ labor force participation rates. This way, one can at least attempt to better explain why the impact may be less significant than one would expect. Card’s Study As mentioned above, one of the problems that have plagued all area studies discussed here is the assumption of exogenous location choice. Neither Butcher and Card’s (1991), nor Altonji and Card’s studies (1991), managed to deal with this issue in an adequate manner. Card (1990), however, was able to avoid this issue in his breakthrough study on the impact of the Mariel Boatlift on the Miami labor market. Virtually over night the Miami labor force grew about 7 percent and the percentage increase in the labor supply o f less-skilled workers was even greater because most o f the immigrants were relatively unskilled. From May to September 1980, some 125,000 Cuban immigrants arrived in Miami on a flotilla o f privately charted boats. About fifty percent of the Mariel immigrants settled permanently in Miami, resulting in a 7 percent increase in the labor force and about a 210 percent increase in the number of Cuban workers in Miami. Since the timing of the immigration was independent o f Miami’s employment and wage rates, the endogeneity problem associated with location choice was avoided. In spite o f this tremendous influx of Cuban immigrants over a very short period, Card was unable to find any significant impact on either wages or employment on non- -29- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Cubans in Miami’s labor market. What is even more surprising is the conclusion that the Boatlifi had no strong effect on the wages o f other Cubans either. These results, however peculiar they may seem, are consistent with another natural experiment performed by Hunt (1992) on the impact of the 1962 repatriates from Algeria on the French labor market. The one significant difference between the two studies is in the skill level of the immigrants. Most o f the repatriates were high-skilled while most of the Mariels were low-skilled. Card suggests three potential explanations for these surprising findings: 1. Miami’s industry structure was well suited to make use o f an influx of unskilled labor. This structure, particularly the high concentration of textile and apparel industries, evolved over the previous two decades in response to earlier waves o f immigrants, and may have allowed the Mariel immigrants to take up unskilled jobs as earlier Cuban immigrants moved to better ones. 2. Because of the high concentration of Hispanics in Miami, the lack o f English-speaking ability among the Mariels may have had smaller effect than could be expected for other immigrants in other cities. 3. Miami’s growth rate, compared to other cities in Florida, indicates that the net migration rate o f natives and earlier immigrants into the Miami area slowed considerably after the Boatlift. This would suggest that the Mariels may have displaced other migrants from within the United States who could have been expected to move to Miami. -30- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The proposed explanations for Card’s surprising findings, brings us back to the fundamental problems associated with the area studies. Instead o f approaching these problems directly most of the recent literature has chosen to evaluate the impact immigrants may have on the rising wage inequality in the United States. More specifically, it appears as if the area approach has been abandoned and replaced with ‘macro’ studies. These macro studies have managed to find some indirect negative impact associated with immigration on the wages of natives. In light of the fact that about 60 percent of all immigrants arriving to the United States arrive to only six cities, it is hard to believe that a significant impact of immigration cannot be found from area studies while at the macro level significant impact can be found. M acro Studies The macro approach, or factor of production approach, assumes that the effects of immigration and trade are sufficiently diffused across areas due to native migration or capital responses that it is best to examine the effect of immigration through its effect on the national supplies of labor with different skills. Most of the studies using this approach have evaluated the change in relative wages of low- and high-skilled workers (or high-school and college graduates). Most of these studies find that immigration has significant impact on the relative wages of less-skilled workers (Topel, 1994; Boijas, Freeman, and Katz, 1996). Boijas, Freeman, and Katz (1996) used the factor-proportion approach to evaluate how immigrants affect the labor market outcomes of natives. This approach requires - 31- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. estimates o f the changed number of immigrants with different skill levels, the implicit change in labor supply due to net trade, and the elasticities of relative wages to relative labor supplies. They obtain the number of immigrants for their evaluation from the 1980 and 1990 Censuses o f Population. Their sample includes workers aged 18-64 years from the 5 percent Public Use Microsamples (PUMS) o f the 1980 and 1990 Censuses o f Population. They acknowledge three problems with these data: 1. the data may include some but not all illegal immigrants, 2. immigrants with a certain level o f schooling may be imperfect substitutes for natives with the same nominal level, and 3. the question on education changed between the two Censuses, causing a classification problem. They concentrated on the comparison o f high school dropouts versus all other workers and high school equivalents versus college equivalents. Their study demonstrates that the influx of less-educated immigrants appears to dominate the impact on high school dropouts, whereas trade and immigration have similar effects for high school versus college equivalents. Their factor-proportion approach demonstrates that immigration has been important in reducing the pay of high school dropouts, while immigration and trade have contributed modestly to the falling pay of high school- equivalent workers. Jaeger (1996), evaluated the impact of immigration on both the relative wage levels and real wages of less-skilled workers (or high school dropouts). He reports that immigration accounts for as much as 3 percentage points o f the decline in their real -32- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. wages. He further notes that the effects o f immigration are unlikely to be uniformly distributed across the United States since immigrants tend to locate in relatively few metropolitan areas. It is precisely this point that this research is focusing on by evaluating the impact o f increased immigration on local labor markets. By allowing for local labor market differences, the true impact of immigration should be discovered. Policy Perspectives What should immigration policy be? In 1965, the U.S. government decided to make the immigration policy more family oriented. Family reunification became the new emphasis as opposed to skill based immigration policy or the national quota that had preceded the 1965 Amendment. One of the main reasons for the implementation o f the 1965 Amendments was to make immigration policy more humanitarian in nature. Prior immigration policies had been quota based and immigrants from countries deemed undesirable were simply not allowed to enter. The majority of the empirical work done on the impact of immigration on the labor market outcome of natives suggests that the current immigration policy does not have a noticeably harmful effects on natives. This result indicates that perhaps no changes in immigration policy are necessary. Furthermore, an immigration policy primarily based on a humanitarian perspectives as opposed to economics appears to have little, if any, adverse impact on the labor market outcomes of natives. However, Blanchard and Katz’s study (1992) suggests that these studies have failed to account for an important aspect associated with immigration. By not accounting -33- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. for the migratory response of natives to an immigration influx, a study will underestimate the impact o f immigration. As immigrants arrive to a given area, the labor demand curve for native workers shifts inwards, driving income and employment down. Blanchard and Katz (1992) show that this adverse impact on the labor demand for natives will throw into action two adjustment mechanisms. First, firms will migrate to the area to take advantage of the lower labor cost. Second, workers will out-migrate as they find themselves unemployed and the return for their skills has decreased. Any study on the impact o f immigration that does not account for this reaction process will yield a rather suspect result. It takes the labor market up to a decade to adjust from these demand shocks, suggesting that any study that does not evaluate the impact o f immigration over a shorter time span will underestimate any potential impact. A more detailed discussion on policy implications will be done in Chapter 5. Summary Utilizing the fact that immigration is geographically concentrated, area studies contrast the level or change in immigration by area with the level or change in the earnings of non-immigrant workers. Like other area studies, the area studies that have concentrated on the impact of increased immigration on the wages of low-skilled labor have found little evidence of adverse impact associated with immigration. That is, virtually all area studies have found only a slight difference in native earnings between the gateway cities that receive immigrants and other parts o f the country. -34- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The failure of the area studies to find any significant impact of immigration on the wages of nativesare commonly contributed to two critical assumptions: 1. the location choice of immigrants is exogenously determined, and 2. the metropolitan area represents a closed economy once immigration takes place. It is quite apparent that these critical assumptions will cause the impact o f increased immigration to be undervalued. Although there is some conflicting evidence that natives react to an increased inflow of foreigners by locating elsewhere, thus dissipating the adverse impact of increased immigration, the majority of the evidence appears to point in that direction (see for example, Chiswick, 1992, 1993; Filer, 1992; Butcher and Card, 1991). Furthermore, failure to allow the impact of immigration to differ across cities will clearly underestimate the true impact of immigration. For example, as Boijas, Freeman, and Katz (1996) found, the 1980 and 1990 cross section on real wages for males had negative and positive signs respectively. They explain their findings in the following way: “The changes in coefficients over time suggest that the cross sectional calculations are dubious structural relations o f the effect o f immigrants on the native wages. One possible reason for the different relations over time is that, for exogenous reasons, demand changed markedly across regions. The 1980’s were a period of economic boom in the coastal areas that receive most immigrants and o f rust-belt problems in many interior areas.” (Boijas, Freeman, and Katz, 1996) This observation strongly suggests that in order to find the true impact of immigration, one must take into account the different economic conditions across different metropolitan areas. It is the purpose of this research to attempt to shed some -35- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. further light on this issue by allowing for different labor market conditions in each labor market and by using the fixed effect model. This approach allows different metropolitan areas to have different labor market conditions by incorporating any structural differences into the fixed-effects. -36- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 3 METHODOLOGY AND DATA "The injury o f unrestricted immigration to American wages and American Standard o f living is sufficiently plain and is bad enough, but the danger which this immigration threatens to the quality o f our citizenship is far worse. " Henry Cabot Lodge, Congressional Record. March 16. 1896 Introduction All area studies on the impact of immigration on the labor market outcomes o f natives have used a cross-sectional analysis. Cross-sectional analysis, however, has the shortcoming o f being static in nature and they only capture the impact of immigration at a specific point in time. Furthermore, such analysis assumes that there are no structural differences across U.S. metropolitan areas. For example, Pinschke and Veiling (1994) suggested that the impact o f immigration would be sensitive to the general economic conditions. This would suggest that if the economic conditions at any given time differ -37- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. across cities, the impact of immigration will differ as well. For example, the type of jobs, the presence of labor unions, and the degree o f unemployment all differ across cities. Those structural differences, along with several other differences, will cause the impact associated with the influx of immigrants to vary across cities. It is the goal of this study to attempt to capture to some extent the problems associated with the assumption that the labor market is homogenous across Metropolitan Statistical Areas (MSAs). Furthermore, this study attempts to capture the impact o f immigration before the factor migration adjustment is complete. This is accomplished by evaluating the impact over a short period o f time, four years. This chapter introduces and explains the methodology and the regression model used in this study. The regression equation is also introduced and justified along with the data source and type. Using the fixed-effects model on pooled cross-sectional, time- series data provides new ways to evaluate the impact o f immigration. Most studies on the issue to date have utilized a more standardized cross-sectional approach. Furthermore, this study takes advantage of the March supplement to the Census which did not include questions on immigrant status until 1994. Hence this study provides new insights into the way in which natives are affected by the continued arrival of immigrants by using a different methodological approach as well as different data sources. -38- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Analytical Framework How immigrants will impact natives will clearly depend on the economic conditions prevalent in the locale as well as the type of immigrants arriving into a given area. This study aims to account for the impact immigrants have on low-skilled natives (i.e., natives with less than a high school diploma). In order to accomplish this, the labor force is split up into two skill levels, high-skilled and low-skilled. The model presented here is similar to that of Altonji and Card’s (1991), except that this study makes a distinction between recent and "old" immigrants. For the purposes o f this study, recent immigrants are considered to be individuals who arrived since the implementation of the 1965 Amendments to the Immigration and Naturalization Act. The Model Suppose we view a labor market as a closed economy where a single competitive industry uses a linear homogeneous production function to produce Q units of a good. The production process uses both high-skilled and low-skilled workers. The wage rate of high-skilled and low-skilled workers are ws and wu respectively. The cost function in this industry is then given by Qc( ws, w j , where c( ws, w j is the unit cost function. Perfect competition implies that long-term profits are zero, which suggests that the price of the product equals the unit cost of production. Thus the price of the output, p, equals the unit cost of production, or p = c( ws, w j . Let us further assume that both skilled and unskilled workers purchase the good. Each type-/ worker (/ = s, u) has an output demand function given by Dj(w;, p). There are -39- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ns high-skilled workers and Nu low-skilled workers, and the fraction of low-skilled workers in the population is b. Product-market equilibrium requires: 0 ) Q = N sD s(w s, p ) + N uD u( w u,p ) . To close the model, suppose the labor-supply function o f each type-/ worker is Li(w„ p). Labor-market equilibrium implies: (2) N sL s( w s,p ) = Q c s( w s, w u) 0 ) N uL u( w u, p ) = Q c u( w s, w u), 5c where c. = 5W; Consider now what would happen if A N immigrants enter the labor market exogenously. Suppose that the fraction of low-skilled workers in the immigrant flow equals |3 . Under some simplifying conditions, Alton] i and Card (1991, pp.204-205) show that the resulting change in the wage of skilled and unskilled workers is given by:2 2 To derive these equations, differentiate the labor market equilibrium conditions, the product market equilibrium condition, and the zero profit condition assuming that d D i / d w i = 0 , dLj / d p - 0, and that the cross-elasticity of factor demand are zero, so that the demand for skilled (unskilled) workers is independent o f the unskilled (skilled) wage. -40- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (4) A logw s = e s - 5 s P - b AN b ( l — b)' N = a AN N 1 - A. b - P AN AN e „ - 5 u b ( l - b ) ' N N ’ where A. = N uD u( w u,p) / Q; s ; > 0 is the labor supply elasticity of type-/ workers; and A.; < 0 is the labor demand elasticity for type-/ workers. Equations (4) and (5) give the reduced-form impact of immigration on the skilled and unskilled wage. Suppose that the fraction o f unskilled workers in the immigrant flow ( P ) equals the fraction of unskilled workers in the native population (b). The linear homogeneity o f the production function then implies that neither the skilled nor the unskilled wage changes as a result of immigration. Alternatively, if the fraction of unskilled workers in the immigrant flow exceeds the respective fraction among natives (P > b), immigration increases the skilled wage and decreases the unskilled wage. This conceptual experiment, therefore, indicates precisely how the impact o f immigration on native employment opportunities can be measured. Practically all empirical studies in the literature, beginning with Jean Grossman (1982), have attempted to replicate this experiment by treating a city or metropolitan areas as the empirical counterpart o f the closed labor market in the theoretical analysis. Equations (4) and (5) clearly point out the importance of taking into account the skill distribution o f immigrants -41- (5) A log w = Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. relative to that of natives. Although some empirical studies have evaluated the impact o f immigration on less-skilled natives, only a few of them have actually accounted for the differences in the fraction of native unskilled workers across cities. By using the fixed- effect model on pooled cross-section time-series data, this structural difference across cities is accounted for. By allowing for structural differences, the results become more consistent with reality. This study also accounts for the possibility of an in- and out-migratory response by natives by evaluating the impact of immigration over a relatively short time span of four years. This study attempts to capture the impact of immigration before the labor market adjustment process is completed. Butcher and Card (1991), and White and Liang (1993), found that states with high levels of low-skilled immigrants have a relatively high level o f in-migration o f high-skilled natives. This result suggest potential complementarity between low-skilled workers and high-skilled workers which is consistent with theoretical predictions. Frey (1994), concluded that less skilled native workers residing in states which receive large immigrant flow are relatively likely to out- migrate. These studies would suggest that the arrival of immigrants set into motion two forces. Low-skilled natives move out of high-immigrant areas while high-skilled natives move into such areas. The combination of these two forces, would suggest that the immigrant flow decreased the unskilled wage (i.e. P > b ). The objective o f this study is -42- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. to better account for the structural differences that exist in the local labor markets across the United State, as well as account for the adjustment process that the influx o f immigrants sets into motion. The Dynamic Effects o f Immigration In evaluating the impact immigrants have on the labor market outcomes of natives, a basic supply/demand model is used. This simple framework captures the basic impact associated with immigration. As low-skilled immigrants arrive into an area, the labor supply of low-skilled natives shifts out, causing real income to decrease and employment to increase. However, the noted increase in employment may not go entirely to native workers. How the new jobs are distributed between natives and immigrants will depend to some extent on how substitutable in production the two groups are. The initial impact of immigration is to put downward pressure on income, ceteris paribus. The greater the immigrant concentration, the greater the negative effect on income. Natives may respond to the declining income in several ways. For example, as income decreases in an area, some natives may opt to drop out of the labor force as income has fallen below reservation income level. Second, as immigration into an area intensifies, some natives lose their jobs due to increased competition (assuming natives and immigrants are substitutes in production). To put it simply, as immigrants arrive into a locale, some natives will be directly replaced by immigrants while others will opt to drop out of the labor force due to lower income. -43- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Some of the natives that lost their jobs (regardless of the reason) may choose to move out of the area to areas where their income is maximized. In any case, the end result on the unemployment rate o f natives may be somewhat ambiguous. If income falls a lot, more natives may opt to drop out o f the local labor force, causing the end result on the unemployment rate to be negative (i.e., lower it). As unemployed natives find themselves faced with increased competition for jobs as well as diminished returns for their skills, they may simply drop out of the labor force. This would reduce the labor force participation rate of natives as well as the unemployment rate. Therefore, the end result on the unemployment rate of natives will depend to a large extent on the exact degree o f the reduction in income. In light o f this, one would expect to find a greater negative impact on income, lower labor force participation rates, and possibly reduced unemployment rates in cities reporting the greatest amount of low-skilled immigrants. At the same time, one would expect to find somewhat smaller negative impact on natives’ income, labor force participation, and unemployment rates in cities where immigration is less concentrated. To capture this initial impact associated with immigration, a short-run analysis must be used. In order to further justify, or establish, the importance of capturing the short-run impact o f immigration, it is useful to go through a simple exercise that captures the dynamic impact of immigration. Blanchard and Katz (1991) have, for example, found that it takes local labor markets less than a decade to recover from a labor demand shock. What they found is that people tend to move out of an area that just experienced a labor -44- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. demand shock while firms tend to move in to take advantage of the lower labor costs. As immigrants enter an area, the demand for native workers decreases (assuming natives and immigrants are substitutes in production). This decrease in labor demand for natives drives their wages down as the competition for jobs increases. Furthermore, as natives lose their jobs, or simply quit as income has now dropped below their reservation level, they leave the labor force. Some of those people actually migrate out of the area, while firms may enter the market. Eventually, or five to eight years later, the market will be back in equilibrium. Blanchard and Katz’s study suggests that studies which use first- difference analysis across Censuses will come short o f capturing the initial impact of immigration. To further demonstrate the importance of analyzing the short-run impact, let us consider the impact o f immigration on wages and employment. Most of the focus below will be on what happens to MSAs that face an increase in labor supply, caused by immigration. There are at least two initial impacts of immigration into an area. First, immigration will increase the local supply of labor and hence decrease nominal wages while increasing employment. Second, immigration will cause an increase in the demand for goods and services. This increase in goods and services will drive up prices and thus reduce real income. The lower incomes and higher prices will set into effect an adjustment process that will cause income to begin to rise again and employment to decrease towards its original levels. Thus, the impact of increased labor supply should clearly be greater in the short run than in the long run. Let us assume that initially the labor market is in equilibrium at the macro level. That is, factor prices are equal across the country. As immigrants enter certain -45- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. metropolitan areas, the labor supply curves of these specific labor markets shifts outwards. This initial increase in the local labor supply will lower local income and increase local employment. The decrease in local income will have two effects. First, workers will leave the area and move to areas that compensate their skills better. Second, firms will observe the decline in labor costs in the area, causing them to move in to take advantage of the lower costs (Blanchard and Katz, 1992). As these forces continue, income begins to rise and the initial rise in employment begins to shrink. This process can be illustrated with a simple labor demand - labor supply diagram. Real Incom e S S1 W0 ----------------- * -------- c W , -------------------------------- B D Em ploym ent Figure 3.1 Suppose that initially the economy is at point A in figure 4.1, with inelastic labor supply S, and a downward sloping labor demand D. At time 0, a positive shock to supply shifts S to S1; under the assumption of full employment the new equilibrium will be at point B in figure 4.1. Employment has increased while real income has decreased. However, this only applies to the overall labor market. The natives will experience -46- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. increased competition for jobs (assuming low-skilled natives and low-skilled immigrants are substitutes) which will drive down their labor demand. As the demand for low-skilled natives is decreased, natives’ income and employment opportunities in the city are reduced. This sets two things into motion: lower income causes people to move out of the metropolitan area, shifting S1 steadily to the left; and lower labor costs lead to net in- migration of firms, shifting D steadily to the right. Whether the final impact on real income or employment will be greater depends on the relative speeds at which the two curves adjust. That is, the precise location of the long run equilibrium depends on the relative speeds at which workers leave and firms enter. If firms arrive at a greater speed than workers leave, the employment effect will be greater than if it were the other way around (Blanchard and Katz, 1992). It should be apparent by this simplified exercise that the greatest impact o f immigration into an area will be felt in the short run. That is, in light o f the results from studies made by Filer (1992), White and Hunter (1993), and Frey (1994), any impact associated with immigration should be primarily found in the short run. Thus, by using data that allows for true short-run analysis, problems that have plagued the literature to date, such as factor mobility, are reduced. -47- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The Empirical Approach In order to evaluate the impact low-skilled immigrants have on low-skilled natives, a fixed-effects model is used. There are several advantages associated with choosing the fixed-efFects model. First, it allows the labor-market conditions to vary across cities. The fixed-effect model accounts for any structural differences that exist in the labor-market conditions across cities by incorporating them into the intercepts. This is a very important aspect o f the fixed-effects model, as it would be unrealistic to assume no structural differences across U.S. cities. Although this study focuses on a relatively short time span, 1994 - 1997, time dummies are included to account for any structural changes across cities over time. Furthermore, it forces fewer restrictions on the model to be estimated. That is, the fixed-effects model does not require that there be no structural differences across cities. Along the same lines, any omitted variable bias which is introduced by not including variables that are systematically related to the cities in the study is avoided as any such variable is accounted for in the error terms. Accounting for the problem associated with the omitted-variable bias provides an increased confidence in the results. Although the specific regression equation employed in this study does not differ from previous studies in terms o f the independent variables chosen, using pooled cross-section time-series data enhances the efficiency of the estimators. Using pooled data increases the number of observations per variable and thus increases the efficiency of estimators. Furthermore, the shortness of the time period used adds to the reliability o f the results as any bias introduced by out-migration is reduced. -48- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. By capturing the short-run nature of the labor market, and allowing for structural differences across labor markets (as well as across cities over time), the analysis that follows provides an important new insight into the existing immigration debate. Altonji and Card (1991), captured the dynamic nature of the labor market by first- differencing the data across the 1980 and 1990 Censuses. Although this approach captures the dynamic nature of the labor market, it fails to captu