1 Introduction

One common finding in the empirical literature is that young individuals are relatively strongly affected by the imposition of minimum wages, both by raising their income level and by decreasing their employment—see, e.g., Clemens and Wither (2016), Kalenkoski (2016), Neumark and Wascher (2007), Neumark et al. (2014) and Laporsek (2013) for recent contributions.Footnote 1 Uniform minimum wages may also reduce on-the-job-training opportunities and discourage younger individuals from entering the labor market. Particularly since the onset of the Great Recession, this delayed entry into the labor market may have reduced individuals’ lifetime stream of earnings (Carcillo et al., 2015).

With this in mind, policymakers may consider the use of specific youth minimum wages instead of uniform minimum wages that are equal for all ages. Youth minimum wage rates are typically set as a fraction of the adult minimum wage, with systems ranging from single minimum wage rates for all young workers below a certain age (as in Greece, Ireland, Portugal) to step-wise functions of an individual’s calendar age (as in the UK, Australia, and The Netherlands) (see Table 6 in the Appendix for details on the implementation of youth minimum wage systems).

Although there is abundant empirical work on the employment effects of general minimum wages on young individuals, studies on the specific effect of youth minimum wages on youth employment are scarce. So far, findings for single countries provide inconclusive evidence; they are largely dependent on the specific context and to identifying assumptions and usually only focus on the short-term impact of changes in youth minimum wages. Typically, this literature exploits reforms or discontinuities of systems in specific countries. For the UK, Dickens et al. (2010, 2014) and Fidrmuc and Tena (2013) use regression discontinuity designs to estimate the impact of a 20% increase of the minimum wage on the 22-year-olds. Dickens et al. find positive employment effects, whereas Fidrmuc and Tena find evidence of negative or no employment effects. For New Zealand, Hyslop and Stillman (2007) perform a difference-in-differences analysis to compare the employment results of young workers who experienced changes in their relative minimum wages (16–17- and 18–19-year-olds) with young adults whose minimum wage rates remained unaffected (20–25-year-olds). Their results indicate insignificant employment effects for both affected groups. Finally, Olssen (2011) studies the impact of a 10% minimum wage increase on workers aged 15–21 in Australia. He finds no effect of the changes in youth rates on short-run youth employment.

This paper is the first to perform a cross-country level analysis that explicitly focuses on the impact of the level of specific youth minimum wages. As such, we add to the analyses of Neumark and Wascher (2004) and Dolton and Bondibene (2012), who exploit cross-country variation in general minimum wages of OECD countries to estimate employment on young and adult individuals. Both these studies find that the presence of youth minimum wages reduces the detrimental impact that general minimum wages may have on employment rates. In the current paper, we contribute to this line of research by investigating how the level of youth minimum wages—as compared to the adult minimum wage or the median wage—affects the employment rate, the labor force participation, and the unemployment rate of younger individuals. Contrasting to the earlier analyses, we thus exploit variation in the use and level of minimum wages across ages within countries.

For our analysis, we follow 30 OECD countries, out of which 21 have implemented minimum wages, between 2000 and 2014. Using information from governmental statistics and OECD publications, we are able to define the use and, if relevant, the level of youth minimum wages compared to uniform minimum wages and to the median wage. More specifically, we construct series of youth rates by recording the level of wage that is eligible at each age. We complement these data with data on the employment, labor force participation, and unemployment level of 5-year age interval categories—all derived from the official OECD database. Our focus is on cross-country differences in the age profiles of employment, labor force participation, and unemployment rates that can be explained by the use of uniform and youth minimum wages. By controlling for interacted country and year fixed effects, we effectively compare the relative labor market position of young individuals with the relative labor market position of young individuals in other countries. In doing so, the key assumption we need to make is that minimum wage settings are not correlated with other labor market institutions that determine labor market outcomes of young individuals. We therefore restrict our sample to young individuals below the age of 35; this setup corresponds to a regression discontinuity design wherein young individuals may face similar youth policies but are either located to the left or the right of the age cutoff-point that determines whether youth or adult minimum wages prevail.

To investigate the effect of minimum wages, we first compare the differences in the relative labor market performance of individuals below the age of 25 between the three country groups. In line with the analyses of Neumark and Wascher (2004) and Dolton and Bondibene (2012), we find the relative employment of young individuals below the age of 25—as compared to older individuals between the age of 25 and 35—in countries with youth minimum wages to be similar to the employment rates of their peers in countries without statutory minimum wages. However, employment rates are substantially lower for young individuals in countries that employ uniform minimum wages.

We next zoom into the sample of the 21 countries that set statutory minimum wages—either uniform or age-specific. Using fixed effects for all possible combinations of countries and years, we compare the relative labor market position of young individuals (vis-à-vis older ones) who are exposed to youth minimum wages with the relative labor market position of their peers that receive adult rates. We specify youth minimum wages as a fraction of both the adult minimum wage and/or the median wage in a particular country. Following this strategy, the use of a specific youth minimum wage rate exerts a positive and statistically significant effect on the employment rate of young individuals for whom it applies. In particular, a 1 percentage point increase in the youth minimum wage as a fraction of the median wage—i.e., the “Kaitz index”—causes the employment rate of young individuals to decrease with 0.72 percentage points. We also find positive labor force participation effects of youth minimum wages, which suggests the mitigation of discouraged-worker effects that prevail with uniform minimum wages.

The remainder of this paper is organized as follows. Section 2 provides a description of the data sources and the construction of our variables. Section 3 explains the empirical strategy, whereas the estimations results and robustness analyses are discussed in Section 4. Finally, Section 5 concludes.

2 Data

We use data derived from databases and from OECD publications. Data on labor market outcomes, including the employment to population ratio, labor force participation, and unemployment rate, are acquired from the OECD Statistics Database.Footnote 2 These variables are measured as (unweighted) averages of 5-year age intervals for 30 OECD countries followed between 2000 and 2014.Footnote 3 We complement these data with information on the incidence and level of adult and youth minimum wages, as derived from OECD Employment Outlook reports of 1998 and 2015, Eurostat and governmental information sources.Footnote 4

With these data, we compare country groups that are stratified with respect to the presence of statutory national minimum wages (or not) and the presence of youth minimum wages (or not). In effect, we have four country groups in our sample that may overlap in some cases: countries with statutory minimum wages (group 1), countries without statutory minimum wages (group 2), countries with specific minimum youth wages at some point in time (group 1A), and countries with uniform statutory minimum wages at some point in time (group 1B).Footnote 5 For each country and year, we constructed the following variables:

  1. i.

    A dummy value that indicates whether an age category is subject to youth minimum wages or not. This fraction is equal to one for the age categories that receive a youth minimum wage and zero across all age categories in countries without youth minimum wages as well as for the rest of the age categories for which the adult minimum wage applies in countries with youth minimum wages.

  2. ii.

    The level of the youth minimum wage rates as a percentage of the adult minimum wage. This variable is used in our main analysis where we concentrate on the group of countries with national minimum wages; it is equal to 1 for the age categories that are eligible for an adult minimum wage. In countries with uniform minimum wages, all age categories are assigned with “1”.

  3. iii.

    The level of the (youth) minimum wage rate relative to the median wage—i.e., the Kaitz index. Again, this variable concerns the group of countries with legally binding minimum wages and is used in the main analysis. Where a youth minimum does not exist, this variable is equal to the uniform minimum-to-median wage.

As the labor market data we use are measured at the level of 5-year age intervals, we need to transform the above variables into 5-year age interval values as well. For this purpose, we assume equal relative weights of five consecutive years in an age interval. Thus, for intervals where we observe a change of the youth into the adult minimum wage at a particular age, we compute the weighted average of variable values before and after the minimum age for which the adult minimum applies.Footnote 6

Table 1 categorizes the countries into groups 1, 2, 1A, and 1B. In our sample, 21 out of 30 OECD countries had statutory minimum wages in the period under investigation (i.e., group 1). These include most of the European countries, as well as Canada, Australia, Japan, Korea, Turkey, and the USA. Group 2 includes countries without statutory minimum wages, namely the Nordic countries, Italy, Austria, and Switzerland.Footnote 7 More relevant to our study is the lower part of Table 1, which concentrates solely on countries of group 1 and further categorizes them into those that employ lower youth rates (group 1A) and those that do not (group 1B). Two countries, Belgium and Greece, appear in both groups 1A and 1B at the same time, since they did not have specific youth minimum wages in the full time period under investigation. Belgium phased out specific youth wages during 2013–2015, whereas Greece introduced them in 2012.

Table 1 Classification of countries on minimum wage systems (2000–2014)

Figure 1 reveals substantial variation in the level of minimum wages across countries in group 1. The real average hourly minimum wage level for the period 2000–2014 was less than 3 US dollars in the Slovak Republic and Estonia, while it exceeded 10 US dollars in France, Australia, and Luxembourg. In this respect, it should be noted that taxes and other non-wage labor costs may significantly affect the net wage (or “take home pay”) as well as the cost of labor for employers. It is estimated that the total tax burden from social contributions and other obligatory payments for OECD countries reaches nearly one third of gross minimum wage, on average, with the shares being almost equally paid by employers and workers (OECD, 2015).

Fig. 1
figure 1

Average hourly adult minimum wages, $US 2014 PPP

Almost half of the OECD countries with a statutory minimum wage use specific minimum wages for young workers. Figure 2 illustrates the relative youth rate as a ratio of the adult rate and of the median wage at each age for workers aged 15 to 25, measured in 2014. In countries like Australia and the Netherlands, the minimum wage for young workers is specified as a step-wise increasing function of the worker’s age. In particular, it starts at a very low age, usually at 15 and approximately at 30% of the adult wage while it gradually increases till the age of 23. In contrast, a single subminimum for young workers exists in Greece, Ireland, and Portugal. In other countries, such as the UK or the Slovak Republic, the youth rate involves up to two youth rates for age groups of young workers for whom the adult minimum wage does not apply.

Fig. 2
figure 2

Share of youth minimum wage rate to the adult minimum wage rate and to the median wage rate as a function of age, 15–25 years old (measured as shares; 2014)

As a first eyeball test of minimum wage effects, Fig. 3 plots the employment rate per age category for our four country groups, averaged over all years in our sample.Footnote 8 The upper panel of each graph compares the average outcomes in countries with statutory minimum wages (group 1) with those without statutory minimum wages (group 2); the lower panel of each graph compares the average outcomes for countries with youth minimum wages (group 1A) with those with uniform minimum wages (group 1B). Countries without minimum wages have higher employment rates than those with minimum wages. This is particularly true for individuals below the age of 25. Interestingly, we also find marked differences between the employment outcomes of younger workers in countries with youth minimum wages and those with uniform minimum wages. Differences in employment rates between groups 1A and 1B are more prominent for younger ages than for older age groups, suggesting that youth minimum wage may have an impact on labor market outcomes.

Fig. 3
figure 3

Average employment rate per age category for countries with statutory minimum wages (group 1) and without statutory minimum wages (group 2) and for countries with youth wages (group 1A) and with uniform wages (group 1B)

3 Model specification

To estimate the effect of youth minimum wages on youth employment, we start with an exploratory investigation on the effect of uniform and youth minimum wages on young individuals’ relative labor market outcomes below the age of 25, as compared to those aged between 25 and 35. More specifically, we compare the relative employment rate, labor force participation rate, and the unemployment rate of individuals below the age of 25 in countries without statutory minimum wages to those with uniform statutory minimum wages and to those with youth-specific minimum wages. The concerning model that describes labor market outcomes is specified in a linear fashion:

$$ {\displaystyle \begin{array}{l}{y}_{ijt}\kern0.5em =\kern0.5em {\beta}_0+{\beta}_1\kern0.5em {\mathrm{UMWD}}_{jt}\times I\left(i\le 2\right)+{\beta}_2\kern0.5em {\mathrm{YWD}}_{jt}\times I\left(i\le 2\right)+{\mathrm{AGE}}_i\\ {}\kern10em +\mathrm{COUNTRY}\_{\mathrm{YEAR}}_{jt}+{u}_{ijt}\end{array}} $$
(1)

where, y ijt is the labor market outcome of interest—i.e., the employment rate, labor force participation rate, or the unemployment rate—for age category i (i = 1,..4) in country j (j = 1,..,J) in year t (t = 1,.,T). The variable UMWD jt represents a dummy on uniform minimum wages, which assigns “1” to countries that have uniform minimum wages in year t and “0” to countries without national minimum wage systems as well as to countries with youth minimum wages in year t. Likewise, YWD jt represents a dummy that is equal to 1 if a country uses youth specific minimum wages in a particular year, and 0 otherwise. Our model also includes fixed effects for all possible combinations of countries and years. As such, we cannot estimate model parameters for the effect of (youth) minimum wages across all age categories.

The main coefficients in model (1) are β1 and β2. These parameters describe the impact of having a uniform minimum wage and a youth minimum wage on the first two age categories that consist of individuals of 15–20 years and 21–25 years of age, respectively. The parameter values can thus be interpreted as the relative labor market performance of young individuals in countries with uniform and youth minimum wages, as compared to the relative labor market performance of their peers in countries without statutory minimum wages. Finally, the error term u ijt is assumed to be identically and independently distributed. We can estimate model (1) with interacted country and year fixed effects, while allowing for clustering effects at the level of countries.

In the second model, we concentrate on the group of countries that employ a national minimum wage to estimate the effect of the level of youth minimum wages on the outcome variables. We thus compare the group of countries with specific youth rates (group 1A) with the group that implements uniform minimum wages (group 1B). Again, we exploit differences in relative outcome variables between young individuals who are affected by the policy of youth minimum wages and the relative outcome variables of their peers who are not exposed to lower rates. This yields the following model for labor market outcomes:

$$ {\displaystyle \begin{array}{l}{y}_{ijt}\kern0.5em =\kern0.5em {\beta}_0+{\beta}_1\kern0.5em {\mathrm{YMD}}_{ijt}+{\beta}_2\kern0.5em {\mathrm{YMW}}_{ijt}+{\mathrm{AGE}}_i+\\ {}\kern5.5em \mathrm{COUNTRY}\_{\mathrm{YEAR}}_{jt}+{v}_{ijt}\end{array}} $$
(2)

where y ijt is the labor market outcome of interest for age category i in country j in year t. The variable YD ijt represents the “youth minimum dummy,” which assigns “1” to the relevant age categories that receive less than the adult minimum wages in those countries where lower youth rates exist, and “0” otherwise. The variable YMW ijt is the youth rate expressed as a fraction of the adult minimum wage for the age categories for which the youth minimum applies; for those for whom the youth minimum does not apply, the value is normalized to “1”. By including YMW ijt , we can test whether the youth minimum wage effects are proportional to the relative level of youth minimum wage—that is, β2 is the quasi-elasticity of youth minimum wages on the outcome measure of interest. As an alternative to the relative level of youth minimum wages as compared to the adult rate, we also present model outcomes with the youth minimum wage measured as a fraction of the median wage—i.e., the Kaitz index. Similar to Eq. (1), we also assume v ijt to be identically and independently distributed. We estimate the model with country and year fixed effects and allow for standard errors that are clustered across countries.

4 Estimation results

Table 2 summarizes the estimation results of Eq. (1) for the employment rate, labor force participation rate, and the unemployment rate as outcome measures. Since Eq. (1) does not require specific information on the level of minimum wages and the minimum age of adult minimum wages (if relevant), we can estimate this model for all countries in our sample. Table 2 suggests that the relative performance of young individuals below the age of 25 in countries with a uniform minimum wage is worse than their peers in countries without statutory minimum wages (which is used as a reference group). More specifically, they are less likely to be employed and less likely to participate in the labor market, with negative effect estimates of about 14 percentage points and 16.5 percentage points, respectively. Interestingly, these negative effects of a minimum wage are insignificant when (lower) youth minimum wages apply. Individuals below the age of 25 for which a subminimum applies perform equally well with their peers in the group of countries with no legal minimum wages (as compared to individuals older than 25).

Table 2 Estimation results of model [1]; N = 1740

In line with Dolton and Bondibene (2012), these findings indicate that a uniform minimum wage across all ages may provide an incentive to employers to cut down employment of young individuals. Moreover, the results could be explained by the existence of discouraged-worker effects; youths who expect that finding a job will be difficult because of the existence of a binding minimum wage are more likely to withdraw from the labor market, typically by prolonging their time in education.

Next, we focus on the group of countries with minimum wages—including both with youth rates (group 1A) and with uniform minimum wages (group 1B) and on individuals below the age of 35. Table 3 reports the corresponding results, which are based on the estimation of Eq. (2) for the employment, labor force participation, and unemployment rates. For each of these outcome measures, we estimate a model specification that includes youth minimum wage dummies as well as a model specification with both youth minimum wage dummies and proxies for the relative level of youth minimum wages, i.e., the youth-to-adult minimum wage and the youth-to-median wage ratios—see columns (ii), (iii) and (iv).

Table 3 Estimation results of model (2) for countries with minimum wages (N = 1200)

Again, the outcomes of model specifications with only the youth minimum wage dummy reveal that young individuals in countries with youth minimum wages generally exhibit better labor market outcomes than those in countries that employ uniform minimum wages. When including the youth-to-adult minimum wage rate or the youth-to-median minimum wage rates, however, these dummy effects turn insignificant. This suggests that employment and labor force participation effects are proportional to both of these measures. From the coefficient estimate of the youth-to-median rate, we infer that the employment rate of young individuals decreases by 7.2 percentage points if the youth-to-median rate would increase with 10 percentage points. These coefficient effects are considerably bigger than those obtained by Dolton and Bondibene (2012), but comparable to those by Laporsek (2013).

From the estimates of the model specifications that include both youth minimum wage fractions, the picture that emerges is that the Kaitz index does a better job in explaining employment rate effects than the youth-to-adult minimum wage rate. Given that the average youth-to-adult minimum wage rate is 0.81 for whom it applies in our sample, a 10 percentage point increase in the youth-to-median rate corresponds to a 12.3 percentage point increase in the youth-to-adult rate. Based on our findings, the employment rate of young individuals would decrease by 5.6 percentage point due to the raise vis-à-vis the median wage and another 2.0 percentage point due to the raise vis-à-vis the adult minimum.Footnote 9 Stated differently, the impact of a raise of the youth minimum wage would be 5.6 percentage points if adult minimum wages would be increased proportionally.

5 Conclusions

This study employs cross-national data from 30 OECD countries and for the time period 2000–2014 in order to explore the effect of introducing lower wages for young individuals on their labor market performance. Essentially, the identification follows from comparing the relative performance of individuals eligible for youth minimum wages with their peers in countries where this policy does not apply. Our results indicate that both the relative employment and labor force participation rates of individuals below the age of 25 are about 10 and 12 percentage points higher, respectively, in countries with youth minimum wages, as compared to countries with uniform minimum wages. Moreover, the effects are found to be proportional to the level of the youth wage, with an impact of an increase of the youth-to-median minimum wage of 1 percentage point that is equal to a decrease in the employment rate of 0.72 percentage point.

In line with earlier research, our results point at employment effects of minimum wages that are relatively large for younger individuals. Still, the effects we find are also considerably larger than those obtained from cross-country analyses that have been conducted by Neumark and Wascher (2004) and Dolton and Bondibene (2012). One explanation for this may be that our analysis is the first to use the exact level of youth minimum wages to explain the employment rates of younger individuals, rather than general minimum wages that are only a crude measure to detect minimum wage effects for young individuals. Accordingly, our estimates are less susceptive to attenuation bias. Related to this argument, it is important to stress once more that our research strategy exploits variation between age groups within countries and years to estimate youth minimum wage effects. By comparing young individuals below and above the age cutoff that gives eligibility to adult minimum wages, we are able to detect displacement effects that would have been left unnoticed if one would exploit variation between countries instead.

That being said, there might be room for improving the estimation youth minimum wage effects in some ways. Even though our data are the first to exploit country-level data on statutory youth minimum wage settings, there still might be variation in non-statutory or local settings that cause measurement error. In line with this, more detailed analyses that incorporate other institutional settings may help in deepening our understanding of minimum wage systems and their effect on youth labor markets. This particularly applies to labor standards, the design of education systems, opportunities for part-time work, union coverage, and employment protection legislation. All these institutions may affect the extent to which individuals and employers respond to minimum wage settings. Finally, for policymakers, it is of key importance to compare the employment effects of youth minimum wages with the income effects due to lower wage earnings for younger employed individuals.