1
Darren Hoover
10653813
June 2016
The Effects of State Minimum Wage Differences
on Cross-state Commuting
Estimates using border straddling counties and fractional logit
models
Bachelor Thesis Economics and Finance
Supervised by:
2
Statement of Originality
This document is written by Darren Hoover who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Abstract
While several studies have examined the minimum wage’s effect on migration, virtually no research has been conducted on the minimum wage’s relationship with commuting. This paper aims to fill this gap by investigating the effect of minimum wage differences between state borders on cross-state commuting rates of low-skilled workers. Using county pairs that straddle state borders with minimum wage policy discontinuities between the years 2004 and 2014, we apply a fractional logit model to estimate our results. We find significant estimates that suggest workers along state borders tend to commute to the state with the lower minimum wage. An argument is subsequently made that low-skilled workers do not have the luxury of choosing where to work based on their preferences, and instead react to the decisions made by the labor demanding firms in response to minimum wage changes.
4 Table of Contents 1. Introduction 5 2. Literature Review 6 2.1. Minimum wage 6 2.2. Low-skilled workers 7
2.3. Minimum wage’s effect on migration decisions 7
3. Methodology 9
3.1. Determinants of commuting 9
3.2. Data collection 10
3.3. The model 11
4. Results 12
4.1. Analysis and interpretations 12
4.2. Discussion 14
5. Limitations & Further Research 14
5.1. Proxy for low-skilled workers 14
5.2. Internal validity 15
5.3. Further research 15
6. Conclusion 16
5
1. Introduction
Ever since the establishment of the 1938 Fair Labor Standards Act (which gives the United States Congress the ability to set a nationwide federal minimum wage), the topic of minimum wage has been a contentious subject among U.S. policymakers and economists (Neumark & Wascher, 2007, p. 1). Whenever this topic finds its way back into public discussion, minimum wage opponents make the argument that an increase in the minimum wage will increase unemployment. This probably explains why the majority of research on the minimum wage in the United States aims to test the validity of this claim, by researching the relationship between minimum wage changes and unemployment. So has been the case since at least 1981, when the Minimum Wage Study Commission published its initial report (Neumark & Wascher, 2007, p. 2).
Although there have been many studies in the past few decades on the minimum wage’s effect on employment, there has yet to be a consensus reached. While many papers (Neumark & Wascher, 1992; Neumark & Wascher, 2002; Lee & Saez, 2012) find that the minimum wage has negative employment effects for low-skilled workers, several studies (Card, 1992a; Katz & Krueger, 1992; Card & Krueger, 1994) find positive employment effects, and others (Card, 1992b; Lemos, 2004) suggest that no conclusions about the employment effects can be drawn.
One aspect that many of these conflicting studies share is the importance they place on the high degree of labor mobility between U.S. states, and the role it plays in the flexibility of the low-skilled labor supply (Cadena, 2013, p. 1). However, relatively little research has been conducted on the relationship between minimum wage and labor mobility itself. Of this research, the bulk has been focused on the minimum wage’s relationship with migration, while virtually no research has been conducted on the minimum wage’s relationship with commuting flows. This paper aims to add to existing literature by studying the latter relationship, asking the question: how do minimum wage differences between bordering states affect the percentage of low-skill workers that commute across state borders? This question will be answered by applying a fractional logit model to the commuting rates of low-skilled workers between county pairs that straddle state borders during the time period of 2004 to 2014.
In the following section, previous literature relevant to this paper will be discussed, from which a theoretical framework will be built to facilitate the analysis of the relationship in question. Subsequently, the methodology of this paper will be presented, along with expected results, information about the collection of data, and the specific model used. Next, the results of the regression performed will be discussed, as well as the implications these results have. Then, the limitations of this paper will be addressed and questions will be suggested for further research. Finally, a conclusion of this paper’s studies will be presented.
6
2. Literature Review
2.1. Minimum wage
While the United States Congress is able to set a nationwide federal minimum wage (and periodically raise it to adjust for inflation), individual states have the capability to set their own minimum wage standards that may deviate from the federal minimum wage. Since the federal minimum wage is binding, a state’s minimum wage will only be observed (with few exceptions) if it is above the federal minimum wage (Koçer & Visser, 2009, p. 362). This allows there to be minimum wage policy discontinuities between states, which may have an implication on where low-skill workers decide to work. Between the years of 2004 and 2014, 33 of the 50 U.S. states and District of Columbia observed minimum wages that differed from the federal rate at least once. The following figure depicts how the discrepancy between state minimum wage rates and the federal minimum wage rate has changed over time:
Figure 1
In this section, previous relevant literature, and their relation to this paper’s research question, will be discussed. Since there has been virtually no research conducted on the relationship between minimum wage and commuting, this paper will discuss and draw mainly from similar studies of the minimum wage’s relationship with migration.
0% 10% 20% 30% 40% 50% 60% 70% 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 % of states with minimum wages higher than the federal rate
7
2.2. Low-skilled workers
The previous literature studying the effects of minimum wages mainly focuses on results for low-skill groups. This is done in order to distinguish the minimum wage effects from the many other factors that may influence the supply and demand of low-wage labor (Neumark & Wascher, 2007, p. 82). Throughout these studies, several different worker groups have been used as a proxy for low-skilled workers earning the minimum wage. One of the most common proxies used in the analysis of minimum wage’s effect on employment is teenage workers. While there are many arguments in favor of using this proxy group, there are some potential problems. Because many teenage workers are secondary earners from non-poor families, the estimated minimum wage effects may not be representative of the minimum wage effect on low-skill adult workers (Neumark & Wascher, 2007, p. 82).
Another common approach used by previous literature is focusing on the effects on certain industries in which minimum-wage workers are prevalent. The fast-food and restaurant service industries are typically used in this approach (Neumark & Wascher, 2007; Card & Krueger, 1994; Dube et al., 2007).
The proxy used by Neumark & Wascher (2002) is the closest in style to the proxy that will be used in this study. Neumark & Wascher estimate their model using employment data on workers aged 16 to 24 years old. This paper will similarly use commuting data for workers aged 29 years old and younger.
2.3. Minimum wage’s effect on migration decisions
In 1995, Bharati Basu released a paper evaluating migration policies for an economy facing unemployment due to a minimum wage, and their effects on welfare. While the general scope of the paper is outside the scope of this paper, Basu makes arguments regarding the influences of labor migration that is relevant. She argues that migration is influenced not by the market wage rates of an economy, but instead the expected wage rates, which is determined by both the market wages as well as the unemployment level (p. 102). This inclusion of unemployment implies that the level of unemployment benefits enters the analysis of workers’ migration decisions. However, since a worker’s decision to commute will not change where the worker lives, the level of unemployment benefits will remain the same for the worker regardless of where they choose to work. Therefore, considering the argument made by Basu as valid, this paper will similarly assume that commuting is influenced by expected wage rates, which is determined by the market wage rates and the unemployment level.
A study conducted by Boffy-Ramirez (2013) asks whether an increase in the state minimum wage induces migration into the state. To answer this question, Boffy-Ramirez looks at the migration decisions of working-age immigrants, a proxy he uses for low-skilled workers. He first finds evidence that minimum wages are a significant component of the compensation for low-skilled immigrant workers. Additionally, he finds precedent from previous studies establishing the connection between minimum wages and market wages.
8
Lee (1999, pp. 998-1000) and Dube et al. (2007, p. 530) find strong effects of minimum wages on hourly earnings for many portions of the United States population, especially for the low-skilled segments. This paper will also use this precedent set by previous literature, and argue that an increase in the minimum wage will increase the market wage for low-skilled workers.
Boffy-Ramirez (2013) goes on to find that an increase in the minimum wage does not significantly decrease immigrant employment. Using Basu’s definition for expected earnings, Boffy-Ramirez consequently asserts that an increase in the minimum wage leads to an increase in expected earnings for low-skilled workers, and furthermore, an expected positive migration effect (p. 2). The results of his study corroborate the findings of other previous literature, that state minimum wage increases can induce in-flow migration. However, he finds that this effect also depends on the amount of time an immigrant has lived in the United States. For immigrants who have lived in the United States for two to four years, Boffy-Ramirez finds a statistically significant estimate that an increase in the minimum wage by a dollar will lead to a 26% increase in immigrant counts. For immigrants outside of that window, the results he found were insignificant. Boffy-Ramirez (p. 12) explains these results by suggesting that immigrants’ first arrival to the United States is into states that are traditional migration hubs. After obtaining information about the U.S. labor market from within the U.S., an immigrant then reevaluates the decision of which state to reside in. Immigrants are then less likely to move as time goes on as they may have roots set within their community. As our dataset uses all U.S. workers under 29 years old, we assume that a large portion of these workers will be non-immigrants and will have more roots set within their communities compared to recent immigrants, and thus will be less inclined to migrate. Therefore, a worker along a state border’s decision of whether to work in the neighboring county will most likely be a commuting decision rather than a migration decision.
Brian Cadena (2013) conducted a similar study on the minimum wage’s effect on low-skilled immigrants’ location decisions. His findings are similar to those of Boffy-Ramirez, in that low-skilled immigrants tend to locate themselves in areas with higher minimum wages. However, it is important to note that these two studies share a limitation: they only consider how non-native low-skilled workers respond to minimum wages in the United States. As mentioned previously, our study includes native workers, who are likely more rooted in their communities and may be more resistant to relocating decisions. Therefore, Cadena and Boffy-Ramirez’s predictions do not necessarily apply to our group of interest.
Martin and Termos’ 2015 study of minimum wage’s effect on low-skilled emigration found results different to those of Boffy-Ramirez and Cadena. Martin and Termos state that, assuming competitive labor markets (i.e. a high minimum wage is linked to job loss), states that increase their minimum wage should see an increase in out-flow migration. With data collected from the American Community Surveys, they find results that confirm their prediction. They estimate that a one dollar difference in the minimum wage of two locations will lead to 0.155% of the low-skilled labor force emigrating to the lower minimum wage
9
state (p. 201). Martin and Termos suggest that this result is mainly due to the resulting decrease in employment rates associated with minimum wage increases (p. 202).
The methodology of this paper (the model, the control variables, etc.) will mainly take inspiration from the studies of Martin & Termos (2015) and Dube et al. (2010). While the latter study is of the relationship between state minimum wage differences and employment, their approach of using counties that straddle borders of states with different minimum wages will be applied in a similar manner in this study.
3. Methodology
3.1. Determinants of commuting
The relationship of interest in this study is minimum wage differences’ effect on commuting. Therefore the dependent variable of this model will be the out-flow commuting rate. The out-flow commuting rate will be defined as the percentage of workers living in the origin county that commute to a primary job located in the destination county. For the purpose of this study, we will look at the out-flow commuting rate for low-skilled workers. To decide which variables are determinants for the out-flow commuting rate, we will consider the variables that influence the costs and benefits associated with a worker’s decision to commute across a state border.
The main independent variable of this model will be the absolute difference in the nominal minimum wages between the two states of interest. As mentioned in the literature review, an increase in the minimum wage (holding unemployment levels constant) leads to an increase in the expected earnings of low-skilled workers. Therefore, if the minimum wage of the neighboring county is raised, it follows that the benefits of commuting to said neighboring county will also increase. As the compensation low-skilled workers will receive in the neighboring county is now higher, it is expected that more low-skilled workers will decide to commute to the neighboring county to earn the higher wages. From this, the minimum wage differential (the destination county minimum wage minus the origin county minimum wage) is expected to have a positive effect on the out-flow commuting rate.
Since the decision to commute is made by a worker weighing the costs and benefits of commuting, our model will include several independent variables to control for factors that influence the value of commuting other than the minimum wage differential.
As explained in the literature review, the other component of a worker’s expected earnings is the level of unemployment. Therefore, our model will include the unemployment rates of both the origin and destination counties as independent control variables. Since a county’s unemployment rate gives an implication for the abundance of available jobs, which in turn can be used as a proxy for the relative easiness to find a job in that county, we expect workers to commute to the county with the relatively lower unemployment rate (i.e. the relatively easier county to find a job in). Thus, the origin county’s unemployment rate is expected to have a positive effect on the out-flow commuting rate. Likewise, the destination
10
county’s unemployment rate is expected to have a negative effect on the out-flow commuting rate.
Benefits of working in another area other than difference in wages include the possibility for a higher number of opportunities and potential social connections. Using Davies et al.’s (2001) assertion that an area’s population can be used to measure the area’s number of opportunities and potential social connections, we control for this difference in opportunity and potential social connections levels by including the populations of the origin and destination counties in the model. Assuming that an increased population is linked with more opportunities, which is a benefit of working in said area, an increase in the origin (destination) county population is expected to have a negative (positive) effect on out-flow commuting rates.
The costs of commuting include the travel costs endured by a worker when working further away from home. These travel costs can take form in either a monetary sense (cost of gas, car maintenance, public transportation, etc.) or a time sense (a longer commute is less desirable). As both the distance between two counties and the cost of gas used for the commute have a positive relationship with travel costs, an increase in either of these variables will lead to an increase in the costs associated with commuting. Therefore, these two variables will be included as control variables, and are expected to have a negative effect on out-flow commuting rates. Our proxy for the cost of gas used for a commute will be the average price of gas in the origin county multiplied with the distance between the two counties’ centers. Furthermore, since there is a fixed portion of the traveling costs associated with commuting (car ownership costs, car insurance, etc.), we expect that the marginal effect of distance on travel costs will be smaller at larger distances. To account for this diminishing marginal cost aspect of travel costs in the model, a quadratic term of the distance will be included as an independent variable in addition to the linear term.
3.2. Data collection
As mentioned previously, this paper will approach the research question in a manner similar to Dube et al.’s (2010) approach. For the bordering states (including the District of Columbia) that had differences in minimum wages at some point throughout the time period of 2004 to 2014, pairs were made between the closest counties on either side of the state border. Of the 109 state border pairs in the continuous United States, 87 had minimum wage policy discrepancies across their border at some point during the time period. These 87 state pairings produced 848 counties along the state borders. These counties were paired together to make 814 county pairs. 768 of these county pairs had the necessary information available. With each county pair producing two out-flow commuting rates (origin and destination counties swap) and an observation for each year that all information was available, a total of 16,507 observations were used.
The commuting rate information was obtained from the U.S. Census Bureau’s Longitudinal-Employer Household Dynamics Program’s Origin-Destination Employment Statistics (LODES). The data allows you to segregate the data based on age, with the
11
youngest segregation being the commuting rates for workers 29 years old and younger. Therefore, the group of workers aged 29 years or younger will be used as our best available proxy for low-skilled workers. The commuting rates are inputted as decimal fractions, so they are bound between zero and one.
The nominal state minimum wages over the years 2004 to 2014 were collected from the U.S. Department of Labor’s Wage and Hour Division. The unemployment rates of the counties over the time period were collected from the U.S. Bureau of Labor Statistics’ Local Area Unemployment Statistics (LAUS). The rates are inputted as percentages, so they are bound between zero and one hundred. Populations of counties throughout the years were obtained from the U.S. Census Bureau’s Population Estimates Program. The data for the distances (in miles) between the centers of two counties were collected from the National Bureau of Economic Research’s County Distance Database. The historical average prices of gasoline for each state were collected from the U.S. Energy Information Administration’s State Energy Data System (SEDS) and are in the form of dollars per gallon.
Below is a table containing summary statistics for each variable that will be used in the regression.
Table 1
It should be noted that since both directions between a county pair are considered separate observations, the average minimum wage differential is around zero. However, the mean is not exactly zero as some sides of certain county pairs are missing necessary commuting rate information leading to an asymmetrical number of observations on either side of a few county pairs. This reasoning also explains the slight difference in summary statistics between origin and destination variables.
3.3. The model
Since the dependent variable of this study, commuting rates, are fractions bound between zero and one (as is the migration rates dependent variable of Martin & Termos’ 2015 study),
Variable Observations Mean Std. Dev. Min Max
CommutingRate 16507 2.7168% 0.0506999 0 55.2% MinimumWageO 16507 6.687308 1.044493 5.15 9.67 MinimumWageD 16507 6.685068 1.043881 5.15 9.67 MinWageDifference (Destination - Origin) 16507 -0.0022403 0.6937857 -2.78 2.78 UnemploymentO 16507 6.909608 2.81 1.1 28.9 UnemploymentD 16507 6.906331 2.813483 1.1 28.9 PopulationO 16507 109391.8 253349 40 5302421 PopulationD 16507 108412.3 252218.4 40 5302421 Distance 16507 33.5754 16.97526 4.837095 99.89611 GasO 16507 2.846624 0.5675637 1.651937 3.88525
12
this paper will similarly use the fractional logit model of Papke and Wooldridge (1996). With the previously mentioned determinants of commuting being used as the independent variables, the model takes form in the following nature:
In which, is the out-flow commuting rate for workers who live in origin county
and commute to work in the destination county at time . is the absolute
difference in the nominal minimum wage of the destination and origin counties at time .
and are the unemployment rates of the origin and destination counties at time
respectively. and are the populations of the origin and destination counties at
time respectively. is the distance between the geographical centers of the origin
and destination counties in miles, and is the average price of gasoline (in dollars per
gallon) in the origin county’s state for the year.
As explained in the first section of the methodology, is expected to
have a positive effect on commuting rates, is expected to have a positive effect, is
expected to have a negative effect, is expected to have a negative effect, is
expected to have a positive effect, is expected to have a negative effect that is
expected to be diminishing at larger distances, and our proxy for gas costs, is expected to have a negative effect. In other words, we expect and .
4. Results
4.1. Analysis and interpretations
Using a fractional logit model regression in Stata, we receive the following results displayed in Table 2.
Table 2
Coefficient Standard Error Z P-value -.068 .0200 -3.38 0.001 .038 .0074 5.21 0.000 -.081 .0074 -11.00 0.000 -.574 .0116 -49.30 0.000 .799 .0117 68.50 0.000 -2.867 .3413 -8.40 0.000 .309 .0475 6.50 0.000 .089 .0660 1.35 0.176 Constant -.335 .6156 -0.54 0.586
13
In line with what was predicted in the methodology, the coefficients found for the origin county’s unemployment rate and the destination county’s unemployment rate are positive and negative, respectively, and are significant at the 0.1% level. They can be interpreted as such: an increase of the origin county unemployment rate by 1 percentage point will lead to an expected 3.8 percentage point increase of the out-flow commuting rate. A 1 percentage point increase of the destination county unemployment rate will lead to an expected 8 percentage point decrease of the out-flow commuting rate.
Similarly expected from the methodology, the coefficients for the origin county population and the destination county population are negative and positive, respectively. This is in line with Davies et al.’s (2001) assertion that an area’s population correlates with the number of opportunities and potential social connections it has. These estimated coefficients are significant at the 0.1% level and can be interpreted as follows: a 1% increase in the population of the origin county leads to an expected 0.574 percentage point decrease of the out-flow commuting rate to the destination county on the other side of the state border. Similarly, a 1% increase in the population of the destination county leads to an expected increase of the out-flow commuting rate by roughly 0.799 percentage points.
The estimated coefficients for and are in line with what was
predicted and are significant at the 0.1% level. The signs of the coefficients provide evidence in favor of the argument that an increase in these parameters are linked to an increase in the costs of commuting, which give rise for a decrease in the commuting rate. Furthermore, the fact that the estimated coefficient for is positive helps confirm the assertion that the marginal costs of distance are diminishing. For the mean distance between county centers of 33.5754 miles, the estimated coefficients for distance can be interpreted as such: a 1% increase in the distance between county centers is estimated to see a percentage point change in the out-flow commuting rate between the counties. The estimated coefficient for our proxy for a commute’s gas costs, , were the opposite
sign of what we expected. The estimated coefficient has a value of 0.089, but is not significantly different from zero with a p-value of 0.176.
While the signs for the coefficients of the control variable were as expected, the coefficient for the main independent variable of interest, , was contrary to what was predicted. We assumed that in the case of minimum wage differences between state borders, controlling for unemployment rates and other determinants, more low-skilled workers would choose to work in the state with the higher minimum wage. With a negative estimated coefficient of -0.068 (significant at the 0.1% level), it seems that the opposite is the case. If the neighboring destination county has a higher minimum wage by one dollar, the model predicts that the out-flow commuting rate to the higher minimum wage county will decrease by roughly 6.8 percentage points. This is quite large considering that the average out-flow commuting rate for the sample was just 2.7168%.
14
4.2. Discussion
When predicting the determinants’ effects on the out-flow commuting rate, we assumed that a higher minimum wage, holding the unemployment rate constant, would result in higher expected earnings for low-skilled workers. It follows that if low-skilled workers make their decision of where to work rationally, then an increase in a state’s minimum wage, after accounting for control variables, would see an increase of low-skilled workers commuting to said state. Instead, the regression results suggest that if an origin county raises their minimum wage, after controlling for unemployment rates and other determining variables, more low-skilled workers will commute out of this origin county to the state with the lower minimum wage. These results of this study could possibly be explained with the assertion that low-skilled workers do not have the luxury of choosing where to work. Instead of weighing the benefits and costs to make a rational decision of whether to commute to another state, low-skilled workers will instead take whatever job they can find. Where these available jobs are offered is determined by the firms on the demand side of the labor market. Thus, the effect of minimum wage differences on commuting rates of low-skilled workers is mostly decided through the low-skill labor demanding firms’ reactions to the minimum wage.
This line of rationalization sets a precedent for the assertion that the inner workings of the United States low-skilled labor market is dictated by the decisions of the demand side, as opposed to the preferences of the low-skilled workers on the supply side. The results of this study suggest that a worker’s decision to commute is not a decision at all, and rather a reaction to the relative amounts of low-skilled labor demanded by firms. One could possibly extrapolate these findings regarding commuting decisions to other aspects of the low-skilled labor market, such as the number of hours worked by a low-skilled worker or the number of jobs a low-skilled worker will have.
5. Limitations & Further Research
5.1. Proxy for low-skilled workers
As mentioned previously in the literature review, commonly used proxies for low-skilled workers include teenagers and fast-food employees. Unfortunately, the commuting rate data provided by the U.S. Census Bureau’s Longitudinal-Employer Household Dynamics Program doesn’t allow isolating the data to workers with these characteristics. Instead, the best proxy that this dataset would allow for was workers 29 years old or younger. While Neumark & Wascher (2002) used a similar proxy of workers aged 16 to 24 years old, including 25 to 29 year olds to this subset of workers could make the proxy group less representative of low-skill workers. This study could have potentially been more effective at isolating the effects caused by the minimum wage if a more appropriate proxy for low-skilled workers was available.
15
Furthermore, it is possible that the sensitivity of commuting rates to wage differentials may be dependent on the type of job or the skills of the workers. It is expected for there to be a lot of skill heterogeneity within our group of workers aged 29 and younger. Using a group of low-skilled workers working in the same industry or having the same skill sets may account for this variation in commuting rate sensitivity.
5.2. Internal validity
One of the largest threats to this model’s internal validity is the potential for an omitted variable bias. It is possible that the used list of commuting rate determinants was not exhaustive. Ignoring potentially important determinants of commuting rates could make the estimated coefficients biased
Another possible internal validity problem this model faces is that it omits the possible lagged effects of minimum wage changes. The model used aspired to analyze the influence of minimum wage differences on commuting rates through the decisions made by the supply side of the labor market. As commuting decisions do not require the workers to relocate their homes, it was believed that the commuting aspect of the low-skilled labor supply would be flexible and able to react quickly to changes in the minimum wage. However, as argued for in the analysis, the firms on the demand side of the low-skilled market also play a large role in whether or not a worker commutes. Since these firms are naturally less flexible to changes in the minimum wage than the workers are, it may be wise to include a component in the model to allow for minimum wages to have lagged effects on commuting rates. Although many economists argue that effects from minimum wage changes should materialize relatively quickly (for reasons such as high turnover rates of minimum wage workers and advanced announcements of minimum wage policy changes), due to firms’ longer-run substitution between capital and labor, the possibility that the full effects of minimum wage changes may take some time to materialize cannot be completely ignored (Neumark & Wascher, 2007).
5.3. Further research
This paper’s methodology uses Dube et al.’s (2010) approach of using county pairs that straddle state borders with minimum wage differences at some point during the studied time period. This method of using local county pairs helps reduce the possibility of an omitted variable bias, due to the presence of unobserved heterogeneity in low-skilled labor market conditions that is spatial in nature (Dube et al., 2010, p. 945). The low-skilled labor market in a certain county is more likely to be similar to its neighboring county than to another county chosen at random. However, it may be worthwhile for future research to deviate from this methodology slightly by also including data on workers commuting between states with no minimum wage differentials over the time period. It may be possible that the inclusion of these observations in the data set would change the results found.
16
Due to the lack of prior research on the minimum wage’s relationship with commuting, this paper took liberties in assuming the variables that determine the percentage of workers that commute. The field of labor economics may benefit from further research of what influences the rate of low-skilled workers commuting to another state. Is it the worker weighing the costs and benefits of commuting to a further location as initially assumed, or is it decided mainly by the job availability as a product of the firms decisions? A deeper understanding of what factors are actually at play may give additional insights into the inner workings of low-skilled labor markets.
As suggested in the discussion of the results, the minimum wage’s effects on the low-skilled labor market may be dictated primarily by how firms react to minimum wage changes. If this is indeed the case, it may be beneficial for further research to study how firms respond to changes in minimum wage policy. Furthermore, as the unemployment rate was found to have a significant estimated effect on commuting rates, reaching a consensus on how minimum wage changes influence the unemployment rate may give additional insight on the minimum wage’s overall effect on commuting rates. Ultimately, a deeper understanding of how the minimum wage affects low-skilled labor demands would allow for more comprehensive estimates of the overall effects of minimum wage policy changes.
6. Conclusion
With the constant debate over the minimum wage in the United States, it is not surprising that have not yet reached a general consensus regarding the effects the minimum wage has. While many studies have researched the effects of minimum wages specifically on employment, this paper aimed to expand the existing literature by analyzing the minimum wage’s relationship with commuting. Using commuting rates of workers 29 years old and younger between county pairs straddling state borders with minimum wage policy discrepancies over the years of 2004 to 2014, several possible determinants were measured for a total of 16,507 observations. A regression was subsequently performed using the fractional logit model of Papke and Wooldridge (1996).
The results of the regression estimate that a minimum wage difference of one dollar will see a 6.8 percentage point shift of the cross-county commuting rates towards the county with the lower minimum wage. To answer the research question of this paper, an increase in a state’s minimum wage significantly leads to more workers commuting to the area with the lower minimum wage. While this result was contrary to what was initially expected, the offered possible explanation is that a worker’s decision on where to work is made mainly in response to firms’ decisions regarding the relative amounts of low-skilled labor demanded.
In conclusion, although this paper sets precedents regarding what influences a worker’s decision on work location, more research on the determinants of worker commuting rates is required to reach a definitive conclusion. A deeper understanding of
17
these determinates may give insights on the inner workings of the low-skilled labor market, which could allow for extrapolation to broader effects of minimum wage changes.
18
7. Bibliography
Basu, B. (1995). Minimum Wage, International Migration And Their Effects on Welfare.
International Economic Journal, 9(2), 101-120.
Boffy-Ramirez, E. (2013). Minimum wages, earnings, and migration. IZA Journal of
Migration, 2(1), 1-24.
Cadena, B. (2013). Recent immigrants as labor market arbitrageurs: Evidence from the minimum wage. Journal of Urban Economics, 80, 1-12.
Card, D. (1992a). Do minimum wages reduce employment? A case study of California, 1987-1989. Industrial and Labor Relations Review, 46(1), 38-54.
Card, D. (1992b). Using regional variation in wages to measure the effects of the federal minimum wage. Industrial and Labor Relations Review, 46(1), 22-37.
Card, D., & Krueger, A. (1994). Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania. The American Economic Review,
84(4), 772-793.
Davies, P., Greenwood, M., & Li, H. (2001). A Conditional Logit Approach to U.S. State-to-State Migration. Journal of Regional Science, 41(2), 337-360.
Dube, A., Lester, T. W., & Reich, M. (2010). Minimum wage effects across state borders: estimates using contiguous counties. The Review of Economics and Statistics, 92(4), 945-964.
Dube, A., Naidu, S., & Reich, M. (2007). The economic effects of a citywide minimum wage.
Industrial and Labor Relations Review, 60(4), 522-543.
Katz, L. F., & Krueger, A. B. (1992). The effect of the minimum wage on the fast food industry. Industrial and Labor Relations Review, 46(1), 6-21.
Koçer, R. G., & Visser, J. (2009). The Role of the State in Balancing the Minimum Wage in Turkey and the USA. British Journal of Industrial Relations, 47(2), 349-370.
Lee, D. (1999). Wage Inequality in the United States during the 1980s: Rising Dispersion or Falling Minimum Wage? The Quarterly Journal of Economics, 114(3), 977-1023. Lee, D., & Saez, E. (2012). Optimal minimum wage policy in competitive labor markets.
Journal of Public Economics, 96(2), 739-749.
Lemos, S. (2004). Minimum wage policy and employment effects: Evidence from Brazil.
Economia 5(1), 219-266.
Martin, D., & Termos, A. (2015). Does a high minimum wage spur low-skilled emigration?
Economics Letters, 137, 200-202.
Neumark, D., & Wascher, W. (1992). Employment effects of minimum and subminimum wages: Panel data on state minimum wage laws. Industrial and Labor Relations
19
Neumark, D., & Wascher, W. (2002). State-level estimates of minimum wage effects: New evidence and interpretations from disequilibrium models. Journal of Human
Resources, 37(1), 35-62.
Neumark, D., & Wascher, W. L. (2007). Minimum wages and employment. Foundations and
Trends® in Microeconomics, 3(1-2), 1-182.
Papke, L., & Wooldridge, J. (1996). Econometric methods for fractional response variables with an application to 401(k) plan participation rates. Journal of Applied
