Does Minimum Wage Lower Employment for Teen Workers? Kevin Edwards Abstract This paper will look at the effect that the state and federal minimum wage increases between 2006 and 2010 had on the employment levels of teenage workers. Recent research finds a variety of effects. In this paper, a fixed effect model is used to control for state and year. The employment data used is from each county in the United States. It is split into high and low wage industries for comparison. The hypothesis was that there would be a negative effect on teenage employment, but the regression shows that there is in fact a positive significant relationship in all low wage industries and two high wage industries. These findings did not hold true when the double-log was taken. Ultimately, based on this research, increasing minimum wage actually helps teenagers get jobs.
Edwards 1 I. Introduction The original Fair Labor Standards Act of 1938 was originally meant to help struggling workers during the Great Depression. Its intent was to protect workers from substandard wages that were relatively common at the time. It was also an effort to eliminate child labor and poor conditions for women. Today, the threat of these substandard conditions and child-labor has virtually disappeared. As such, the focus of the minimum wage has shifted to keeping workers above the poverty line. That brings up the question, does it? Does the minimum wage policy benefit workers as a whole? Or does the increase in wage cause employment to drop, resulting in a net loss? The threat of employment loss is concerned especially with teenage workers. With teenagers making up roughly 23 percent of workers earning minimum wage, effects of minimum wage increases would be strongly felt by this age group (BLS). This paper will use a fixed effect model to examine the effect that minimum wage increases between 2006 and 2010 had on the quarterly employment of teenagers between the ages of 14 and 18 at the county level. Since the majority of minimum-wage workers are teenagers, it is expected that the greatest effect of a minimum wage increase would be seen in teenage employment. II. Theoretical Framework Traditional labor economics views the minimum wage as a price floor. In a competitive labor market, if the minimum wage is set above the market equilibrium wage, then it will result in a decreased number of jobs available, but an increased number of workers seeking jobs. It is
Edwards 2 possible that these new job seekers are teenagers. They were not in the labor market because they were not required to be. Most teenagers live with other family members that support them. As such, these teenagers would opt to spend their time in other ways until they felt the minimum wage was worth sacrificing free time for extra spending money. In this situation, it is likely that teenagers will receive jobs rather than older workers competing for the same positions. Teenagers don t have the responsibilities outside of work that an older worker might have. As such, it is possible that the teenage employment could increase with a minimum wage hike, while causing unemployment in other age groups. Generally speaking though, employment should drop with an increase in minimum wage (Brown, Gilroy and Kohen, 1982). - Graph from Federal Reserve Bank of Cleveland However, if the market is monopsonistic, then there may not necessarily be any unemployment created. In a monopsony, employers face an upward sloping marginal cost of labor curve. This is because they hire the cheapest workers first. When they hire new workers that want a higher wage, they have to increase the wages of their old employees too. A minimum wage allows for the employers to see a flat marginal cost of labor curve until the
Edwards 3 minimum wage equals the marginal product of labor. If the minimum wage increase isn t drastic, employment may benefit by the increase (Brozen, 1969). III. Literature Review Previous literature on the topic of minimum wage and employment has not yielded a single uniform answer to the question at hand. Most work done before 1990 reports a negative relationship between minimum wage and teen employment. Over the last few decades, some research has found no significant effect between minimum wage and employment. Some even report a small positive effect. The works that find a negative result use a variety of techniques to reach that conclusion. Sen, Rybczynski and Van De Waal (2011) use a panel analysis on Canadian employment and minimum wage data from 1981 to 2004. They find that a ten percent increase in the minimum wage results in a decrease in teen employment by about three to five percent. They also find that not only does it hurt teen employment levels; it causes an increase of four to six percent of families living below the poverty line. The authors also speculate on the accuracy of various techniques. They say that there are shortcoming and benefits to their method. They report that while the use of time-series cross-sectional models can offer ways to identify changes over time and across state borders, they are very subject to endogeneity. However, they also say that casestudies can have drawbacks of their own. They provide a great comparison effect of very specific scenarios, but they are often criticized for inadequate control groups, small sample size and short time frames (Sen, Rybczynski and Van De Waal, 2011). This article helps to point out the strengths and weaknesses of the various techniques used to analyze the topic.
Edwards 4 A study by Rohlin (2011) uses a difference-in-difference model to look at the effect that minimum wage has on the establishment of new businesses. The study looks at the states between 2003 and 2006 that increased their minimum wages. It compares establishments near the state borders to examine the difference between the side that increased minimum wage and the side that did not. The study finds that a minimum wage increase negatively impacts new establishments that rely on low-wage employees (Rohlin, 2011). This research suggests that a minimum wage increase would not be enough to cause existing business to lay off workers or to move from an area, but it is enough to deter potential new businesses from choosing the area in which the minimum wage increase occurred. In that way, it causes harm to the employment level by limiting its growth (Rohlin, 2011). The research that finds no significant relationship, or in some cases a positive relationship, uses similar techniques. Perhaps the most well-known study was by Card and Krueger. They did a case-study in 1992 on the New Jersey-Pennsylvania border. They looked at 410 fast-food restaurants and compared employment before and after a minimum wage hike in New Jersey (Card and Krueger, 1994). The study looked at the employment trends in New Jersey and Pennsylvania as well as New York and the United States as a whole. They found that the minimum wage increase had no significant negative effect, and in some cases, even a small positive effect (Card and Krueger, 1994). Dube, Lester and Reich (2010) expand upon Card and Krueger (1994) study by looking at employment in 1380 counties nationwide. They arrange these counties into cross-border pairs. The counties were also examined in a sample that did not pair them off by contiguous borders. This was done in an attempt to explain the unobserved heterogeneity associated with fixed effect models. The authors find that there is no significant effect of minimum wage on employment,
Edwards 5 but there is a strong positive effect on earnings. For a one percent increase in minimum wage, there is roughly a 0.21 percent increase in wages across the various treatments (Dube, Lester and Reich, 2010). An early work by Adie (1973) looks at the impact that the federal minimum wage has on teenage unemployment rates. The study splits the teenage demographic by ethnicity and gender. All groupings were found to have a positive relationship. More importantly, the study found that a one percent increase in minimum wage 0.362 percent increase in unemployment for all teenage workers. IV. Model The employment data used in this research came from the Quarterly Workforce Indicators from the United States Census Bureau. This data gives total employment counts for each state further broken down by county. Employment was chosen for several reasons. First, the theoretical models for competitive labor markets and monopsonistic labor markets relate wages to employment. Employment was used to more directly relate the results of this paper to the theory. Second, employment more accurately depicts the availability of jobs (Brown, Gilroy, and Kohen, 1982). Finally, most of the literature reviewed looks at employment rather than unemployment. For this study, the numbers from 2006 to present were used. The county employment gives data for multiple age groups and industries. The data is split into eight age groups as well as a cumulative all ages grouping. For the purpose of this study, only the all ages group and the 14-18 grouping were used. The summary statistics for the employment can be seen in Tables 1 through 4.
Edwards 6 The minimum wage data came from the Department of Labor. The years between 2006 and 2010 were chosen because of several federal minimum wages shifts. There was an increase in 2007, 2008 and 2009. There are also a number of state minimum wage increases. Using these years with the QWI data provides a variety of changes in the minimum wage to observe. Yearly minimum wages for each state as well as the federal minimum wage were provided. Each state has the ability to adopt its own minimum wage. The federal minimum wage covers all the states. If the state minimum wage does not exist or is lower than the federal minimum wage, then the federal minimum wage is used. These wages were sorted into a variable that represented the effective minimum wage in each state. The effective minimum wage is the higher of the state and federal minimum wage. The model uses a fixed effect model that controls for state and year in which the observation took place. It looks at the effect of the minimum wage on the employment counts. Emp = β₀ + β₁mw + S + T A second regression looks at the double logs of minimum wage and employment to find elasticity. Descriptions are available in the appendix. Ln(Emp) = β₀ + β₁ln(mw) + S + T The coefficients for both MW and ln(mw) are expected to be negative assuming a competitive labor market. Three generally low-wage industries and three generally high-wage industries were analyzed. The reason for doing this is to compare the effect of high and low wage industries. It is hypothesized that the low-wage industries will be negatively affected by the minimum wage
Edwards 7 increases. The high-wage industries are not expected to have any effect. Data for all industries together was also examined. The three low-wage industries analyzed are accommodations and food services, manufacturing and retail. The high-wage industries chosen are Finance, Information and Professional. When the regression was run for all ages, it found that the effect of minimum wage on the employment counts was not significant for any industry. However, when the regression was restricted to teenagers, it revealed that minimum wage does have significant positive effect on employment in almost all industries. For the all industries regression, a minimum wage increase of one dollar resulted in an employment increase of 181.17. In the accommodations, retail and manufacturing industries, a one dollar minimum wage increase results in an employment increase of 42.26, 72.31 and 7.34 respectively. In the high wage industries, the information industry yielded a positive effect of 3.55. The finance industry yielded an effect of 2.29. These results are displayed in tables 5 through 7. The positive relationships found in the counts do not carry into the regression for the natural log of minimum wage and the natural log of employment. The observations that would have resulted in taking the log of 0 were discarded. These situations were a small proportion of the total number of observations. To cover all bases, one was added to all of the employment totals before the logs were run, but the results still showed no significant effect. Some of the signs were opposite of those found in the regression of the employment counts. The r-squared values are higher for the double log regression, indicating that more variation is explained. It is possible that this regression better accounted for factors that were omitted by the counts model. However, none of the relationships were found to be significant on any level. These results can be viewed in Tables 8 through 10.
Edwards 8 V. Conclusion Based on the data, it can be concluded that an increase in the minimum wage has a positive effect on the employment counts of counties in the United States. However, the second model that examines the elasticity of employment with regards to minimum wage reveals that there may be no significant effect at all. The positive effect found in the count regression is the opposite of what was expected. However, as explained earlier, the positive effect can be explained in the monopsonistic model and the competitive market model. The positive effect in the high wage industries comes as a surprise. Since no effect was expected, it is possible that the model doesn t account for current trends or state and local policies. It is also possible that a minimum wage increase could lead to a ladder effect. In this situation, the people who make above minimum wage see the raises that those at minimum wage increase get and decide they want one as well. This effect can rise through a number of wage levels. Even the insignificant effects found in the log model fall within the possibilities of the theoretical models. Based on the research in this paper and that of previous literature, it doesn t appear that current minimum wage policies have no negative effect on teenage employment levels. It is possible that it even has a positive effect. Previous literature warns against a minimum wage policy that goes over the top. In both the competitive and monopsonistic models, setting a minimum wage that is too high could result in employment loss and a rise in unemployment. This research is far from perfect. Even though effect from state and year are accounted for in the fixed effects, the fact that the data used is at the county level means there are still a number of county and local area affects that might be causing an omitted variable bias. Sen, Rybczynski and Waal (2011) offer a number of possible variables to include in this model, like
Edwards 9 population, unemployment rate and others. Because the data used was quarterly county data, I was unable to find usable data for additional regressors. Future research could go several routes to build upon this paper. Obviously, adding regressors and limiting the omitted variable bias would be one route. Another would be to implement more advanced methods for the regressions than those used in this paper. Some of the previous literature used difference-in-difference models or IV models.
Edwards 10 Appendix Variable Descriptions Variable Description Source Emp Quarterly employment Quarterly Workforce Indicators (US Census Bureau) MW Effective Minimum Wage (Greater of State US Department of Labor MW and Federal MW) S State Fixed Effect Variable N/A T Year Fixed Effect Variable N/A Table 1 - Summary Statistics for Ages 14-18 (Employment Counts) Industry Mean Median Minimum Maximum Standard Deviation All Industries 1304.60 314 0 144139 4241.57 Accommodations 479.72 129 0 39115 1361.24 Retail 330.10 73 0 42582 1088.26 Manufacturing 48.04 14 0 10884 214.86 Finance 13.78 4 0 2035 57.57 Information 27.30 6 0 5699 121.47 Professional 30.85 6 0 4359 120.82 Table 2 - Summary Statistics for All Ages (Employment Counts) Industry Mean Median Minimum Maximum Standard Deviation All Industries 38801.18 7863 17 4459531 145264 Accommodations 3329.79 641 0 340279 12285.99 Retail 4551.78 964 0 495340 15388.98 Manufacturing 4418.02 1264 0 474352 13936.24 Finance 1831.70 216 0 331789 9545.71 Information 1075.56 108 0 302815 7041.40 Professional 2355.40 170 0 315197 12445.04
Edwards 11 Table 3 - Summary Statistics for Ages 14-18 (Employment Logs) Industry Mean Median Minimum Maximum Standard Deviation All Industries 5.8151 5.7526 0.6931 11.8785 1.5879 Accommodations 3.2794 4.8903 0.6931 10.5743 1.6114 Retail 4.4564 4.3175 0 10.6592 1.5140 Manufacturing 2.9844 2.8904 0 9.2950 1.2434 Finance 2.1700 1.9459 0 7.6183 1.0523 Information 2.7364 2.5649 0 8.6480 1.2172 Professional 2.4374 2.3026 0 8.3800 1.2530 Table 4 - Summary Statistics for All Ages (Employment Logs) Industry Mean Median Minimum Maximum Standard Deviation All Industries 9.0687 8.9699 2.8332 15.3106 1.6053 Accommodations 5.4438 6.4646 1.0986 12.7375 1.7665 Retail 6.9194 6.8721 1.0986 13.1130 1.6833 Manufacturing 6.9791 7.1460 1.0986 13.0697 1.8082 Finance 5.5892 5.3753 0 12.7123 1.6431 Information 4.9137 4.6913 0 12.6209 1.7480 Professional 5.3620 5.1358 0.6931 12.6610 1.9596 Table 5 - All Industries (Counts) All Industries All Ages 14-18 MW 1140.33 (2013.55) 181.17* (58.39) Adj-R 2 0.0844 0.0973 N 55044 54910 * Significant at 1% **Significant at 5% *** Significant at 10%
Edwards 12 Table 6 - Low Wage Industries (Counts) Accommodations and Food Service All Ages 14-18 MW 21.04 (171.16) 42.26** (19.01) Adj-R 2 0.0849 0.0887 N 53941 53574 Retail MW 209.38 (212.95) *Significant at 1% **Significant at 5% ***Significant at 10% Table 7 - High Wage Industries (Counts) 72.31* (15.08) Adj-R 2 0.0907 0.0955 N 54599 54097 Manufacturing MW 231.11 (198.73) 7.34** (2.18) Adj-R 2 0.0807 0.0750 N 51050 46865 Information All Ages 14-18 MW 48.75 (103.61) 3.55*** (1.85) Adj-R 2 0.0455 0.0599 N 48646 44716 Finance MW 100.43 (136.83) * Significant at 1% **Significant at 5% *** Significant at 10% 2.29** (0.93) Adj-R 2 0.0505 0.0721 N 53122 39650 Professional MW 78.47 (177.80) 2.66 (1.84) Adj-R 2 0.0612 0.0882 N 52834 43619
Edwards 13 Table 8 - All Industries (Logs) All Industries All Ages 14-18 MW 0.0112 (.1193) -0.0367 (0.1201) Adj-R 2 0.2521 0.2264 N 55044 54910 * Significant at 1% **Significant at 5% *** Significant at 10% Table 9 - Low Wage Industries (Logs) Accommodations and Food Service All Ages 14-18 MW -0.0051 (0.1341) -0.1463 (0.1256) Adj-R 2 0.2291 0.2028 N 53941 53574 Retail MW 0.0225 (0.1268) 0.1380 (0.1154) Adj-R 2 0.2335 0.2293 N 54599 54097 Manufacturing MW -0.0759 (0.1394) 0.152 (0.1073) Adj-R 2 0.2385 0.1870 N 51050 46865 * Significant at 1% **Significant at 5% *** Significant at 10%
Edwards 14 Table 10 - High Wage Industries (Logs) Information All Ages 14-18 MW 0.1812 (0.1429) 0.2142 (0.1256) Adj-R 2 0.1664 0.1147 N 48646 44716 Finance MW -0.0751 (0.1307) * Significant at 1% **Significant at 5% *** Significant at 10% 0.0578 (0.1134) Adj-R 2 0.1691 0.1468 N 53122 39650 Professional MW 0.0149 (0.1528) 0.1978 (0.1229) Adj-R 2 0.2069 0.1392 N 52834 43619
Edwards 15 Works Cited Adie, Douglas K. Teen-Age Unemployment and Real Federal Minimum Wages. Journal of Political Economy. Mar-Apr 1973, 81(2), 435-441. Brown, Charles, Gilroy, Curtis, and Kohen, Andrew. The Effect of the Minimum Wage on Employment and Unemployment. Journal of Economic Literature, June 1982, 20(2), 487-582. Brozen, Yale. The Effect of Statutory Minimum Wage Increases on Teen-Age Employment. Journal of Law and Economics, April 1969, 12(1), 109-122. Card, David and Krueger, Alan B. Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania. The American Economic Review, September 1994, 84(4), 772-793. Dube, Arindrajit, Lester, T. William, and Reich, Michael. Minimum Wage Effects Across State Borders: Estimates Using Contiguous Counties. The Review of Economics and Statistics, November 2010, 92(4), 945-964. Rohlin, Shawn. State minimum wages and business location: Evidence from a refined border approach. Journal of Urban Economics, 2011, 69(1), 103-117. Sen, Anindya, Rybczynski, Kathleen, and Van De Waal, Corey. Teen Employment, poverty and the minimum wage: Evidence from Canada. Labour Economics, January 2011, 18(1), 36-47.