Labour market snapshot #29 July 2016 Jeff Borland Department of Economics University of Melbourne

A brief review of US studies of the effect of the minimum wage on employment Key points • US studies of the effect of the minimum wage on employment have evolved from time-series analysis of the effect of the federal minimum wage on total employment to panel analysis of the effect of state-level minimum wage variation on state or regional employment. • Time-series studies found an elasticity of employment to the minimum wage of between -0.1 and -0.3. The new minimum wage research that used state-level variation in the minimum wage widened the range of estimates of the elasticity – including positive and larger negative effects. • The most recent minimum wage research has been able to introduce extra cases of inter-state variation in the minimum wage and to control in a more detailed way for differences in US regional labour markets that might affect employment outcomes. This research has narrowed the range of estimates of the minimum wage elasticity. So, while debate continues, today the argument is about where the minimum wage elasticity lies in an interval from zero to -0.2.

A brief history The history of US studies of the effect of the minimum wage on employment can usefully be divided into three phases. In the first phase until the early 1990s the main type of US study was time-series analysis of the relation between aggregate employment and the federal minimum wage. It was increasingly recognised, however, that this type of study was not ideal for identifying the effect of changes to the minimum wage. Kennan (1995, p.1955) noted that predicted changes in teenage employment due to changes in minimum wage were small relative to month-to-month changes in teenage employment, and argued that therefore testing for minimum wage effects in aggregate employment data is like ‘looking for a needle in a haystack’. In the second phase of ‘new minimum wage research’, dating from the early 1990s, US studies have used state-level variation in the minimum wage (for example, between states where the federal minimum was binding and states where a higher state-level minimum was binding) to identify the effect of the minimum wage on employment. The scope for state-level variation in minimum wages to identify employment effects grew in the late 1980s and early 1980s as initially the federal minimum wage remained constant in nominal terms while some states began to raise their own minimum wage above the federal minimum, and then with increases in the federal minimum wage in 1990 and 1991.

Two main types of studies using statelevel variation in the minimum wage have been undertaken. A first group of studies has used panel regression models to examine the effect of changes to statelevel minimum wages on state-level employment. A second group of studies use a case study approach by applying quasi-experimental methods. For example, changes in employment from before and after an increase in the minimum wage are compared between a treatment group of establishments where the wages of some workers needed to be increased to match a rise in the minimum wage and a control group of establishments in the same state or a nearby state that did not need to increase their workers’ wages. Both the panel model and case study approaches have focused on low-wage workers most likely to be affected by the minimum wage, such as teenagers or workers in the fast food sector. In a third phase from the mid-2000s a variety of US studies have sought to address unresolved econometric issues with the state-level research. This has primarily involved paying greater attention to controlling for selection effects; in particular, the role of geographic variation in labour market conditions. The objective has been to choose control regions as controls for treatment regions so that underlying labour market conditions are as similar as possible to treatment regions where an increase in the minimum wage has occurred; or how to correct for differences in underlying conditions

between geographic labour markets where increases in the minimum wage do and do not occur. It is now recognised that obtaining valid estimates of the minimum wage elasticity depends fundamentally on having sufficient exogenous variation in the minimum wage in order to identify its effect on employment. The development of the US literature on the employment effects of the minimum wage is best understood as a progressive search for new methods of empirical analysis that can do this; for example, the shift from aggregate analysis of the effect of the federal minimum to disaggregated analysis of state-level minimum wages. A cautionary note on interpretation The main objective of US minimum wage studies has been to estimate the elasticity of employment for a specified group of workers to a change in the minimum wage. That elasticity equals:


=
 ∗  ∗ (

%∆ ) %∆

Where: %∆


 = %∆ : Equals the percentage change in employment in response to a 1 percent increase in the minimum wage for a sample of workers all of whom earn the same amount less than the new minimum wage;

 : Equals the proportion of the sample of workers whose wages are affected by the change to the minimum wage; and

(

%∆ %∆

): Equals the average

percentage change in the wages (or wage costs) of workers who earn below the new minimum divided by the percentage change in the minimum wage. What is of most interest to policy-makers is the direct effect of the minimum wage on employment – that is, percentage change in employment in response to a 1% increase in the minimum wage for a group of workers all of whom earn the minimum wage. But any estimated elasticity from empirical analysis will confound that direct effect with the proportion of workers in the sample whose wages are not affected by the minimum wage and with the size of adjustment in wages for workers who are affected by the minimum wage. It follows that the estimated elasticity from different minimum wage studies can vary, even where the direct effect of the minimum wage on employment is the same. For example, suppose that the direct effect is -0.2, that %∆is 5 percent and that %∆ is 2.5 percent. One study might be able to focus on a group of workers all of whose wages are affected by the minimum wage. That study would estimate an employment elasticity of -0.1. Another study, however, might need to examine a group of workers only one-half of whose wages are affected by the minimum wage. That study would estimate an elasticity of -0.05. There is an important lesson here. Estimates of the elasticity of employment to the minimum wage cannot necessarily

be compared between studies. This is because any difference in the estimated minimum wage elasticity may reflect differences in the proportion of the sample of workers whose wages are affected by the minimum wage or in the distribution of wages around the minimum wage – rather than differences in the employment impact of the minimum wage for workers who earn that minimum wage. Clemens and White (2014) provide an illustration of this point in their analysis of the employment effect of increasing the US federal minimum wage from $5.15 to $7.25 between July 2007 and July 2009. Focusing on workers who were directly affected by the increase in the minimum wage they find that the employment rate fell by 6 percentage points. But looking at the broader group of teenagers and food service workers they find that the employment rate fell by a smaller amount, 4 percentage points. This difference is interpreted as evidence that ‘estimates of the minimum wage’s effects on employment will scale with the extent to which an analysis sample contains unaffected workers’ (p.3). Findings from time-series studies The major review of time-series studies by Charles Brown (1999, pp.2115-16) characterises evidence from those studies done prior to the early 1980s as being that a 10 percent increase in the minimum wage would cause a 1 to 3 percent decline in teenage employment, an effect which was statistically significant. Subsequent

evidence from the 1980s and 1990s is suggested, however, to have found smaller effects – uniformly less than a 1 percent decline in employment in response to a 10 percent increase in the minimum wage, with many estimates being insignificant. Against this apparent decline over time in the effect of the minimum wage, a later review by Neumark and Wascher (2008, pp.79-81) argues that time-series analyses in the 2000s suggest continuing significant negative employment effects of the minimum wage. Findings from state-level panel studies The benchmark state-level panel study is by Neumark and Wascher (1992). They find an elasticity of teenage employment with respect to the minimum wage of 0.19. Later studies using the same methodology obtained similar findings (for example, Burkhauser et al., 2000). However, some other studies, such as Card et al. (1994a) using an alternative approaches to estimate the employment effects of minimum wages in state-level panel models found no significant effect of minimum wages on employment. At the time it was concluded that the critical factor in reconciling these different findings was whether the state-level panel model included a control for year effects (Brown, 1999, pp.2123-30). Studies that had identified the effect of the minimum wage using just cross-state variation found no significant effect on employment; whereas studies that also incorporated identification from time-

series variation tended to be more negative. Since the initial research and debate, the story of mixed findings from state-level panel models has continued. It seems that there is no result that cannot be reversed on further analysis. One line of research has argued that extending the state-level panel analysis to include more recent time periods, which encompass extra states raising their minimum wages above the federal minimum, provides greater variation in the minimum wage from which the employment effect can be identified, even when year and state effects are controlled for. When this is done it is argued that there is a significant negative effect of the minimum wage on employment similar to the original Neumark and Wascher study (for example, Sabia, 2009). Other research, which has sought to further refine the state-level panel model method, continues to show no effect of minimum wages on employment. For example, Allegretto et al. (2011) conclude that the negative significant effect of the minimum wage on teenage employment found in the original Neumark and Wascher study becomes insignificant when year-specific fixed effects for sets of states or state-specific linear trends are included as controls. Inclusion of statespecific linear trends, however, has been shown to be potentially problematic where main effect of increase in minimum wage is on growth rate of employment – and allowing for such longer-term effects

restores finding of a significant negative elasticity of -0.2 (Meer and West, 2013). Findings from case studies The case study that has attracted most attention was by Card and Krueger (1994b). That study compared changes in employment in fast food restaurants in Pennsylvania and New Jersey before and after an increase in the New Jersey minimum wage. It found that the increase in the minimum wage had a zero or small positive significant effect on employment (see also Katz and Krueger, 1992). There has also been much debate over the findings from case studies. The most substantive critique is by Neumark and Wascher (2000), to which Card and Krueger (2000) provide a response. Reviews of this debate have concluded that, for the case of Pennsylvania, Card and Krueger got it right (for example, Brown (1999, p.2138); and Belman and Wolfson (2014, p.27)). However, it’s now also recognised that the Pennsylvania-New Jersey comparison is only part of the story. Dube et al. (2010) undertake a large-scale replication of the Card-Krueger case study approach for the entire set of possible case studies of minimum wage changes affecting the US restaurant sector using contiguous counties across borders of states. They find that there is a distribution of estimated minimum wage effects from the case studies which is centred on zero.

Hence, the findings of positive employment effects from the minimum wage in the initial Card and Krueger study can be interpreted as being part of the distribution of effects from the set of possible case studies, where that distribution is centred on zero but includes some positive effects, as well as some large negative effects (for example, Sabia et al., 2012). More on geographic variation Neumark et al. (2014a) dispute the findings from recent panel model studies that find no effect of minimum wages on employment, as well as the Dube et al. (2010) analysis of multiple case studies. Their main contention is the best possible control groups for states where increases in minimum wages occur may not be other states in the same Census division (in a state-level panel mode)l or contiguous border counties (in a case study). When they apply what they argue are superior control groups the finding of a significant negative elasticity of employment with respect to the minimum wage re-emerges. This issue of the appropriateness of limiting the choice of control group labour markets to geographically proximate state , as well as the value of constructing a synthetic counter-factual control group labour market, is taken further in recent studies by Hoffman (2016) and Sabia et al. (2016). Debate on geographic variation has continued, without coming to a resolution, in studies by Allegretto et al. (2013) and Neumark et al. (2014b). The

former study finds (p.3): ‘…the employment elasticities for teens and restaurant workers range between -0.06 and +0.13’. While the latter study concludes (p.22): ‘We see the evidence as still pointing to disemployment effects for low-skilled workers from raising the minimum wage, with elasticities that are often around -0.2 for the teenagers on whom we focus.’ New methods and issues Several recent studies have applied longitudinal data on individual workers and establishments to develop alternative perspectives on how increases to the minimum wage affect employment. Using longitudinal data on labour market outcomes for individual workers Clemens and Wither (2013) are able to identify exactly which workers are affected by increases to the minimum wage and to track their employment outcomes following those increases. And Aaronson et al. (2015) use longitudinal data on establishment-level employment in the restaurant industry to argue that the main impact of increases to the minimum wage on employment is through the exit of labour-intensive establishments rather than changes to employment in on-going establishments. Endogeneity of the minimum wage to geographic variation in labour market conditions in state-level studies is addressed by Baskaya and Rubinstein (2012). They demonstrate that the statelevel effective minimum wage is procylical

– but only for those states where the effective minimum wage is the state minimum rather than the federal minimum. To deal with this endogeneity of the minimum wage variable they use an IV approach – instrumenting the state effective minimum wage with a variable that is an index for the likelihood that the federal minimum wage is binding in a state. IV estimates in a standard panel model find a large negative elasticity of employment with respect to the minimum wage of about -1. What does this all mean? In the US debate continues on the minimum wage elasticity of employment. Today that debate is about whether the elasticity of teenage and low-wage employment with respect to the minimum wage is at either end of a relatively small range: zero or around -0.2. So while as yet there may not be a consensus about the size of the elasticity, the narrow range being argued over does have something of the feel of ‘much ado about nothing’. That debate would continue to be so strong no doubt indicates that there is more at stake than the number. Nevertheless, it is important to remember that the direct elasticity may be larger than even the upper estimate of -0.2. Equally, however, from a public policy perspective it seems important to recognise that the size of the group directly affected by the minimum wage is relatively small. What is the explanation?

The finding that the employment effect of the minimum wage is close to zero or small has prompted a search for explanations.

minimum wage will cause no change (or even a rise) in employment (Bhaskar et al., 2002). References

One response has been that there is nothing to explain when the estimated elasticity is interpreted in the correct way. Even a small negative estimated elasticity might imply a large negative direct elasticity. A related response is that the estimated effect fails to properly reveal the impact of the minimum wage – for example, because the elasticity is derived from a study of total employment which misses the effects on hours of work or masks substitution of high skill for low skill workers; or because the study did not consider a sufficient time period for the long-run impact of the minimum wage to be observed. Other responses take seriously the small estimated elasticity. One possible explanation is that it could reflect that adjustment to the minimum wage is occurring in ways that keep relatively constant the cost of employing covered workers – for example by employers raising required levels of worker effort and productivity. Another approach is to explain how a small elasticity could occur in a competitive labour market – noting, for example, that low-pay workers are concentrated in non-traded sectors and account for only a small share of total in those sectors (Manning, 2013, pp.59-60). A further explanation that has been suggested is that the labour market is monopsonistic or oligopsonistic, in which case it is possible that an increase in

Aaronson, D., E. French and I. Sorkin (2015), ‘Industry dynamics and the minimum wage: A putty-clay approach’, mimeo. Allegretto, S., A. Dube and M. Reich (2011), ‘Do minimum wages really reduce teen employment? Accounting for heterogeneity and selectivity in panel data’, Industrial Relations, 50, 205-40. Allegretto, M., A. Dube, M. Reich and B. Zipper (2013), ‘Credible research designs for minimum wage studies’, IZA Discussion Paper no.7638. Balaska, Y. and Y. Rubinstein (2012), ‘Using federal minimum wages to identify the impact of minimum wages on employment and earnings across the US states’, mimeo. Belman, D. and P. Wolfson (2014), What Does the Minimum Wage Do? (WE Upjohn Institute for Employment Research). Bhaskar, V., A. Manning and T. To (2002), ‘Oligopsony and monopsonistic competition in labor markets’, Journal of Economic Perspectives, 16, Spring, 155-74. Brown, C. (1999), ‘Minimum wages, employment and the distribution of income’, pages 2101-2162 in O. Ashenfelter and D. Card (eds.) Handbook of Labor Economics Volume 3 (Elsevier). Burkhauser, R., K. Couch and D. Wittenburg (2000), ‘A reassessment of the new economics of the minimum wage literature with monthly data from the Current Population Survey’, Journal of Labor Economics, 18, 653-80. Card, D., L. Katz and A. Krueger (1994a), ‘Comment on David Neumark and William Wascher, ‘Employment effects of minimum

and subminimum wages: Panel data on state minimum wage laws’, Industrial and Labor Relations Review, 47, 487-96. Card, D. and A. Krueger (1994b), ‘Minimum wages and employment: A case study of the New Jersey and Pennsylvania fast food industry’, American Economic Review, 84, 772-93. Card, D. and A. Krueger (1995), Myth and Measurement: The New Economics of the Minimum Wage (Princeton University Press). Card, D. and A. Krueger (2000), ‘Minimum wages and employment: A case study of the fast food industry in New Jersey and Pennsylvania: Reply’, American Economic Review, 90, 1397-1420. Clemens, J. and M. Wither (2014), ‘The minimum wage and the Great Recession: Evidence of effects on the employment and income trajectories of low-skilled workers’, National Bureau of Economic Research, Working Paper no.20724. Dube, A., W. Lester and M. Reich (2010), ‘Minimum wage effects across state borders: Estimates using contiguous counties’, Review of Economics and Statistics, 92, 945-64. Hoffman, S. (2016), ‘Are the effects of minimum wage increases always small? A reanalysis of Sabia, Burkhauser and Hansen’, Industrial and Labor Relations Review, 69, 295-311 Katz, L. and A. Krueger (1992), ‘The effect of the minimum wage on the fast food industry’, Industrial and Labor Relations Review, 46, 621. Kennan, J. (1995), ‘The elusive effects of minimum wages’, Journal of Economic Literature, 33, 1950-65. Manning, A. (2013), ‘Minimum wages: A view from the UK’, Perspektiven der Wirtschaftspolitik, 14, 57-76.

Meer, J. and J. West (2013), ‘Effects of the minimum wage on employment dynamics’, National Bureau of Economic Research, Working Paper no.19262. Neumark, D. and W. Wascher (1992), ‘Employment effects of minimum and subminimum wages: Panel data on state minimum wage laws’, Industrial and Labor Relations Review, 46, 55-81. Neumark, D. and W. Wascher (2000), ‘Minimum wages and employment: A case study of the fast food industry in New Jersey and Pennsylvania: Comment’, American Economic Review, 90, 1362-96. Neumark, D. and W. Wascher (2008), Minimum Wages (MIT Press). Neumark, D., J. Salas and W. Wascher (2014a), ‘Revisiting the minimum wage-employment debate: Throwing out the baby with the bathwater?’, Industrial and Labor Relations Review, 67, 608-48. Neumark, D., J. Salas and W. Wascher (2014b), ‘More on recent evidence on the effects of minimum wages in the United States’, National Bureau of Economic Research, Working Paper no.20619. Sabia, J. (2009), ‘Identifying minimum wage effects: New evidence from monthly CPS data’, Industrial Relations, 48, 311-28. Sabia, J., R. Burkhauser and B. Hansen (2012), ‘Are the effects of minimum wages always small? New evidence from a case study of New York State’, Industrial and Labor Relations Review, 65, 350-76. Sabia, J., R. Burkhauser and B. Hansen (2016), ‘When good measurement goes wrong: New evidence that New York State’s minimum wage reduced employment’, Industrial and Labor Relations Review, 69, 321-19.