The influence of minimum wage on unemployment:
classical or monopsonistic effects?
Bachelor thesis Economics Karel Boxhoorn
Student number: 10004494
Supervisor: Francisco Gomez Martinez 07/2014
Abstract
In this study, the influence of minimum wage on employment was investigated. According to classical theory, minimum wage should have a negative effect on employment. Other
models include the possibility of monopsonistic effects, whereby a rise in minimum wage can have positive employment effects. This study analyses the effects of minimum wage on employment in the Netherlands in the period 1984-2013 using regressions derived from Slonimczyk and Skott (2012). For this period no significant effect of changes in the minimum wage on employment were found.
1. Introduction
Many industrializes counties have a minimum wage set by the government as part of their labour policy. Out of the 28 countries in the European Union, 21 currently have a nationwide minimum wage (Eurostat, 2014). Some notable exceptions are Italy, the Scandinavian
countries in the E.U. and Austria. Germany also currently has no minimum wage but will enact one in 2015 (BBC, 2014). It is clear that many European countries feel the need to have a minimum wage for economical, social or ideological reasons. Because the minimum wage places a constraint on the wages in a country, it is possible that this has impact on many aspects of the national economy.
The effect of minimum wage on employment and unemployment is an important side effect of the wage restriction. Minimum wage might, for example, be raised with the intent of strengthening the economic situation of people with a low income. Although this policy will raise wage for a part of the labour force, the rise in the minimum wage might reduce employment possibilities for this group and therefore the policy may actually be self-defeating. If, on the other hand, the rise in minimum wage also raises employment, there are double benefits for the labour force. A change in minimum wage could have different effects; what these effects are might have a big influence on the usefulness of minimum wage as a labour market policy. Therefore, this study aims to investigate the influence of changes in minimum wage on employment.
To analyse the effects of a change in minimum wage on employment, this study looks at the Dutch situation between 1984 and 2013. A longitudinal analysis has been done with annual data on minimum wage, number of high skilled workers, number of low skilled workers, unemployment in these two groups, and overeducation. Analysis on these data shows no clear effect of minimum wage on unemployment. This study consists of five sections. Section 2 deals with previous theoretical and empirical literature. Section 3 describes the methodology behind this study. In Section 4 the results are presented and section 5 consists of the conclusion and discussion.
2. Literature review
In the literature review two types of theories on minimum wage will be addressed: a model with classical effects and a model with the possibility of monopsonistic effects. After this some evidence from previous studies on the subject is presented.
2.1 Theory on minimum wage
The classical view on minimum wage is that it has a negative effect on employment. The underlying assumption of this view is that there is perfect competition in the labour market. There are many people supplying labour and many firms demanding this labour. The market will gravitate towards an equilibrium where wages equal the marginal product of the labour. The rationale behind this is that firms will keep hiring workers as long as their added value to the production process is greater than their wage cost. This supply- and demand equilibrium is based on an underlying assumption of perfect competition. In this case both the wages and the level of employment are optimal because the total benefits are maximized (Pindyck & Rubinfeld, 2009). In this case a minimum wage that raises the earned wage (for a part) of the workers will cause involuntary unemployment. This will happen because on the one hand supply will increase but on the other hand demand will drop. This drop in demand is caused by the fact that the minimum wage is above the marginal product of a part of the labour supply. The higher labour supply is caused by the raise in wage for people who could previously only get a job that paid less than minimum wage. Ultimately, the establishment of a minimum wage or an increase in minimum wage will result in an increase in
unemployment.
If there is no perfect competition, things will not work out this way. In the case of a monopsony, employment in equilibrium will be lower than in a competitive market. This is because in the case of a competitive market, the level of employment and the wage will be determined by the point of intersection between the marginal product of labour and the wage. On the other hand, in the case of a monopsony wage and employment will be determined by the point of intersection of the marginal cost and the marginal product of labour. This will result in a lower level of employment and a lower wage than in perfect competition. This case has many similarities to a monopolist on a goods market. In both
cases the monopolist or monopsonist will keep the quantity of a good or employment low because this yields maximum profits.
In the case of a monopsonistic labour market, a minimum wage above the current market wage will not result in unemployment. This is caused by a change in the cost structure of the wage. In the monopsonistic case without a minimum wage, the employer has to raise wage to attract new workers. If there is a minimum wage in place, the part of the wage curve that was originally below the minimum wage will now be flat. This will cause the firm to employ more people because there is no longer a trade-off between employment and wage. A shortcoming of this approach is that in many cases it is unlikely that the labour market is actually a monopsony. In some locations or industries there will be only one firm hiring workers, but in many other sectors there is a multitude of companies that people could work for. There are also models which imply that the labour market can be monopsonistic in its effects without being a monopsony. By using one of these models questions about the composition of the labour market can be avoided. An example of such a model is the model created by Rebitzer and Taylor (1995). They assume that workers will shirk depending on, amongst other things, their salary and the chance of being caught. They also assume that companies have fixed monitoring resources. In this case a company would have to pay more than the market wage when hiring new employees to prevent shirking. This is because the probability of detecting shirking drops when there are more employees. This causes the effects to be comparable to the monopsony case.
Another model where there can be monopsonistic effects without there being an actual monopsony is the model made by Slonimczyk and Skott (2012). In this model
monopsonistic effects are caused by heterogeneity in the labour force and the presence of mismatch between labour and education. This mismatch presents itself in the form of overeducation: people that are higher educated than is necessary for their job. The
difference with the model made by Rebitzer and Taylor (1995) is that monopsonistic effects can arise even when the probability of detecting shirking is given exogenously.
In the Slonimczyk-Skott model, the labour market consists of two types of workers: high skilled workers and low skilled workers. The skill level of workers is determined by their level of education. These two types of workers correspond with two types of jobs: high-skill jobs and low-skill jobs. Again, whether a job is high-skill or low-skill is determined by the level of education that is suitable for the job. Low skilled workers will only be able to work in
low-skill jobs; high skilled workers on the other hand will be able to work in both low and high-skill jobs. An important factor explaining the effects in their model is the assumption that overeducation is present in the labour market; not all high skilled workers will be working in high-skill jobs.
In their modelling of the labour market Slonimczyk and Skott use a Shapiro-Stiglitz model. In this model workers will either shirk or not shirk depending on a number of factors like wage and the chance of getting caught when shirking. Because there is heterogeneity in the labour market there will be three different no-shirking conditions, depending on the combination of the skill level of the worker and the skill level of the job.
Slonimczyk and Skott simplify the wage for low-skill jobs to the minimum wage. In their model, a rise in the minimum wage must cause a decrease in the wage paid for high-skill jobs when assuming constant returns to scale. Otherwise “real labour costs must have increased which contradicts the assumption of a constant markup on labour costs”
(Slonimczyk and Skott, 2012: 250).
In this situation two cases are possible: the first one is a case in which low skilled workers are preferred by employers for low-skill jobs. In the second case high skilled workers are preferred in low-skilled jobs. In the situation where firms prefer low skilled workers in low-skill jobs, high skilled workers will only be hired for low-skill jobs “if the no-shirking condition is binding for low skilled workers” (Slonimczyk and Skott, 2012: 251). Because the minimum wage is one of the constraints of the no-shirking condition for low-skill jobs, a rise in the minimum wage will relax this constraint, making it possible to hire more people in these jobs. The higher minimum wage may also have a negative effect on the number of low-skill jobs but that does not matter for low skilled employment because some of these low-skill jobs are filled by the less favoured high skilled workers. The effects on high skilled jobs could be either negative or positive in this case, depending on a number of factors. Because there is less employment for high skilled people in low-skill jobs, the no-shirking condition for high skilled people is relaxed, raising employment for high skilled workers in high-skill jobs. In this case, total employment will rise when the minimum wage rises, just like in the case of a monopsonistic labour market.
In the case where firms prefer high skilled workers in low-skilled jobs, the effects are not monopsonistic. As in the previous case, a rise in minimum wage will lead to a decline in high-skill wage. This causes the wage-premium to fall and the ratio of high skilled employed
workers to low skilled employed workers to rise. In the model high skilled employment must fall, and because of the rising employment ratio, low skilled employment must also fall. The degree of overeducation rises because of the rising number of high skilled workers in low-skill jobs.
2.2 Empirical evidence from previous studies
Previous research has not been able to come to a consensus about the employment effects of changes in the minimum wage. From a classical economic viewpoint a minimum wage will decrease employment because people who have a lower marginal product of labour than the minimum wage will not be hired. A number of studies have found effects that indicate such an influence of minimum wage on employment. Nguyen (2013) found that in Vietnam an increase in minimum wage had a negative impact on the employment in the official sector. In 2005 the minimum wage in Vietnam was increased. Nguyen found that the increase had a negative effect on employment because a portion of the people who previously earned less than the 2005 minimum wage went from the labour force to being self-employed. Majchrowska and Zółkiewski (2012) found that increases in the minimum wage had a negative effect on employment in Poland in the period between 1999 and 2010. Furthermore their analysis indicates that this negative effect was strongest in the poorest regions of the country. These results, as well as the results from Nguyen seem to indicate that the classical model is indeed a good predictor for the relation between minimum wage and unemployment.
Other studies have found effects of the establishment or rise in minimum wage that cannot be explained by the neo-classical viewpoint. These studies have found that increases in minimum wage can have positive effects on employment for the general labour forces as well as workers in specific low-wage sectors. An example of this is a study by Dickens, Machin and Manning (1999). In their model employers are not in perfect competition; the market is more monopsonistic than competitive. Therefore in their model a raise in minimum wage can either have a positive or a negative effect on employment. Dickens, Machin and Manning test their model empirically by studying the effects of the industry-wide minimum wages in Britain between 1975 and 1992. They found that these minimum wages do not lower employment. This is contrary to what one would expect based on a classical model with a competitive labour market.
Another study that found effects that are not compatible with a classical labour market model was conducted by Machin, Manning and Rahman in 2002. Here the effect of the national minimum wage on residential care homes in the United Kingdom was analyzed. This was done by looking at the industry before 1999, when there was no minimum wage in comparison with the situation after 1999, when a national minimum wage was set. The study found that the establishment of minimum wages had a small negative effect on employment, but implies that this effect is less than one would expect in classical model.
Slonimczyk and Skott (2012) test their own model, which has been mentioned in section 2.1, with a regression on quarterly data from the United States in the period 1979-2002 and panel data for individual states. They conclude that the U.S. data support the hypothesis of monopsonistic effects. Their empirical research shows that a fall in the
minimum wage between 1979 and 2002 had negative effects on the employment and wage of low skilled workers. They also found an increase in underemployed high skilled workers.
3. Methodology and Data
3.1 The modelThe influence of changes in minimum wage on employment is tested using the regressions stated below. These regressions are derived from the theoretical model that has been developed and tested by Slonimczyk and Skott (2012:254).
In these regressions is low-skill unemployment and is high-skill unemployment. is the degree of overeducation. is the minimum wage. H and L are the number of high- and low skilled workers in the total labour force. The variable t is the time in years. Based on the two theoretical cases identified by Slonimczyk and Skott the parameters are expected to be shaped according to the following restrictions (2012: 255):
3.2 The data
These regressions have been done on annual Dutch data in the period from 1984 to 2013. The third regression has been done on the period from 1986 to 2011. This is because there is no reliable data source on overeducation in the Netherlands that contains data from before 1986. Firstly, it is important to define what the distinction between low skilled and high skilled workers is. Slonimczyk and Skott define low skilled workers as those that have had a high school education or less than that. High skilled workers are those that are college-educated. Because of the differences between the Dutch and the U.S. schooling systems, a comparison is made using the International Standard Classification of Education (ISCED). The U.S. Department of Education classifies college education as having an ISCED level of at least 5 (2014). In the Dutch case an ISCED level of 5 or higher corresponds to either Higher
people who have finished a HBO or have Bachelors or Masters Degree are defined as high skilled. People with a lower educational attainment are defined as low skilled.
The data on high- and low skilled workers in the labour force, as well as the data on the unemployment in these two groups come from the CBS Statline website. The CBS
collects these data through a number of methods, mostly panel surveys (2014b). The data on the yearly minimum wage come from Towers Watson, a company that has these data freely available on its website. The minimum wage is corrected for inflation using data from the CBS.
Overeducation is measured through a combination of two variables. The first one is a workers level of education and the second one is the level of the occupation a worker has. The level of occupation is defined on a scale with five groups using the Standard
Occupational Classification 1992 (SBC ’92). The SBC ’92 is a comprehensive list of occupations in the Netherlands and also classifies these occupations in five groups
depending on the level of education that is normally necessary for the specific occupation (CBS, 2001). The highest two groups in this classification consist of jobs for which a HBO or university education is needed. Occupations in these two groups are defined as high-skill jobs. Occupations in the three lower groups are defined as low-skill jobs. People are
considered to be overeducated if they are high skilled but work in a low-skilled job. The data on overeducation come partly from the CBS Statline and partly from a Dutch labour force survey called Arbeidsaanbodpanel (AAP). The CBS Statline has information on the
occupational level of workers grouped by their highest attained education from 1996
onwards. For the years from 1986 until 1996 the AAP is used. This panel survey is held every two years by the Netherlands Institute for Social Research (SCP). The survey is done on a sample of between 4500 and 5000 people in the potential labour force; this is the
population between 16 and 65 years of age (SCP, 2014). In the 1994 survey the level of the occupation of working respondents is classified according to the SBC ’92. By using this in combination with the data on the highest attained education of respondents it is possible to see the fraction of working respondents that are high skilled but work in a low skilled job. In the years before 1994 a similar process is used, but in this case the occupations were classified using the Standard Occupational Classification 1984. The SBC ’84 has no clear indication of the occupational level. Therefore the data are converted from SBC ’84 to SBC
’92 using the conversion lists made available by the CBS.1 For the uneven years between 1986 and 1996 (in which there was no survey), an average of the overeducation in the previous and next year has been used.
With these data the variables are defined in the following way: Low-skill
unemployment ( ) and high-skill unemployment ( ) is the proportion of unemployed low- or high skilled people in the low skilled or high skilled part of the labour force.
Overeducation ( ) is the fraction of high skilled people in low-skill jobs as part of the total working labour force. The minimum wage ( ) is measured in Euro’s per year and includes the holiday allowance. Lastly, time (t) is measured in years starting in 1984.
1This conversion is not always exact because the SBC ’92 is more elaborate and detailed than the SBC ’84.
Therefore one occupation in the SBC ’84 might recode into a number of different occupations in the SBC ’92. In this paper the SBC ’84 occupations are recoded into the SBC ’92 equivalent with the lowest code. This has little or no effect on the measure of overeducation because the SBC ’92 codes that are equivalent to one SBC ’84 code almost always fall within the same occupational level.
4. Results
4.1.1 Explanatory variables: minimum wage and skill ratio
Before the results of the regressions are presented, some descriptive results on the changes in real minimum wage, high- to low skill ratio and overeducation are shown. The first
descriptively presented statistic is the minimum wage. The nominal Dutch minimum wage has been raised on an annual or biannual basis between 1984 and 2013. Exceptions to this are the periods ’84-’87, ’87-’89, ’92-’95 and ’03-’05. In these periods the minimum wage was not altered. When taking inflation into account, the changes in minimum wage look
somewhat different. Figure 1 shows the development of the real minimum wage on the 2013 price level. As can be seen in the graph, the real minimum wage drops significantly in the 1984-2013 period. The difference on a yearly basis between 1984 and 2013 is roughly €2000. On the basis of the classical view as well as a model with monopsonistic effects, this drop in minimum wage should have employment effects.
Figure 1. Development of minimum wage in Euro, 2013 price level
Figure 2 shows the skill ratio, this is the ratio of high skilled people to low skilled people in the labour force. The graph clearly shows a rising skill ratio through time; a rising percentage of the labour force is educated at HBO or university level. The small drop in the skill ratio around 2001 is caused by a difference in the way the CBS uses their data starting that year (2014c). 17000 17500 18000 18500 19000 19500 20000 20500 21000 21500 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
Development of real
minimum wage
min wageFigure 2. Development of ratio of high skilled people to low skilled people in the labour force
4.1.2 Dependent variables: unemployment and overeducation
The unemployment levels, as seen in figure 3, show no clear rising or falling pattern. Both high- and low skill unemployment move in a similar manner that at first glance appears to somewhat resemble the business cycle. Examples of this are the clearly higher
unemployment levels during the crises of the 1980’s and after 2008.
Figure 3. High- and low-skill unemployment as fraction of the total high skilled and low skilled labour force
As shown in Figure 4, the degree of overeducation shows a rising line through most of the investigated period, with small drops in 1987, 1994 and 2001. This shows that in the last 30
0 0,1 0,2 0,3 0,4 0,5 0,6 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
Development of ratio H/L
H/L 0 0,02 0,04 0,06 0,08 0,1 0,12 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 uL uhyears a rising number of people is employed in jobs that are below their level of education, according to the occupational classification of the CBS.
Figure 4. Overeducation as fraction of the total labour force from 1986 until 2011
4.2.1 Regression for low-skill unemployment
After the regression analysis had been done, it appears to be that the regression with low-skill unemployment as dependent variable has generated insignificant results at a 10% level. The results of this regression are shown in table 1. When looking at the estimated effect of minimum wage on low-skill unemployment the parameter hints at a positive effect of rises in the minimum wage on unemployment, although this effect is not significant in the estimated model. This result goes against both the possible outcomes of the Slonimczyk-Skott model,
0 0,01 0,02 0,03 0,04 0,05 0,06 0,07 0,08 0,09 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
Overeducation
Overeducation Table 1 Regression for[Standard errors in brackets]
Significance Time trend -0.003 [0.003] 0.244 Ln 0.157 [0.120] 0.201 Ln 0.076 [0.088] 0.396 Constant -1.350 [1.143] 0.248 Obs 30 R-squared 0.350
as they expect changes in minimum wage to have either a positive or negative effect on low-skill unemployment.
The ratio of high- to low skilled people in the labour force also has no significant effect on the rate of low-skill unemployment. This lack of a significant effect matches the case from the Slonimczyk-Skott modelin which employers prefer low skilled people in low-skill jobs. The time trend also has no significant effect. From the perspective of the
Slonimczyk-Skott model this would indicate that the technical change between 1984 and 2013 has had neither a positive nor a negative effect on low-skill unemployment.
A problem with using this model on the data is the high degree of collinearity between the time trend and the logarithm of the high-to low skill ratio. Because of this collinearity it is difficult to separate the effects of the skill ratio and the effects of technical changes as measured by the time trend. If the time trend is removed from the model the effect from ln becomes -0.028 with a significance of 0.053. This would indicate a -0.00028 change in low-skill unemployment with a 1% change in skill ratio. This small but negative effect is predicted by the case in which employers prefer high skilled employees in low skilled jobs. If this is the real effect of the skill ratio is unsure because of the high collinearity with the time trend; when the time trend is removed the missing variable bias will alter the effect of the skill ratio.
4.2.2 Regression for high-skill unemployment
Table 2
Regression for
[Standard errors in brackets]
significance Time trend -0.002 [0.002] 0.139 Ln 0.091 [0.065] 0.173 Ln 0.048 [0.048] 0.329 Constant -0.770 [0.621] 0.226 Obs 30 R-squared 0.524
Table 2 shows the results of the regression on high-skill unemployment. In this regression also none of the effects of the dependent variables are significant at the 10% level. The effect of minimum wage on high-skill unemployment follows the prediction from the case of the Slonimczyk-Skott model where employers prefer low skilled people in low-skilled jobs. Based on this case the effect of minimum wage should be equal to, or greater that zero. With a positive but statistically insignificant effect this is the case in the estimated model, as a insignificant effect should be seen as an effect of zero.
The ratio of high- to low skilled people in the labour force has a positive effect on high-skill unemployment but this effect is not significant. The insignificant positive deviation from zero of this effect can be explained by both cases of the Slonimczyk-Skott model. Therefore this result does not say anything about whether employers prefer low- or high skilled people in low-skill jobs. The effect of the time trend shows a similar effect to the first regression: it is negative but insignificant. This regression suffers from multicollinearity much in the same way the regression for low-skill unemployment does. When the time trend is removed from the model the effect of ln is -0.025 with a significance of 0.003. This would mean that a 1% change in the skill ratio has an effect of -0.00025 on high-skill
unemployment. This effect cannot be explained with either of the two cases from the Slonimczyk-Skott model. Again, this outcome can very well be the effect of omitted variable bias and should be looked at critically.
4.2.2 Regression for overeducation
Table 3
Regression for
[Standard errors in brackets]
Significance Time trend 0.002 [0.001] 0.022 Ln 0.030 [0.039] 0.460 Ln 0.001 [0.029] 0.968 Constant -0.273 [0.383] 0.485 Obs 26 R-squared 0.931
The regression for overeducation shows some mixed results. As can be seen in table 3, the time trend has a small but significant effect on the amount of overeducation. The minimum wage has a positive but insignificant effect on the measure of overeducation. This effect is not predicted by either of the two theoretical cases; if employers prefer low skilled workers in low-skill jobs the effect should be negative, and in the case where employers prefer high skilled workers in low-skill jobs, the effect should be positive. Because the result is
insignificant, it is neither of these two,although the result hints at the second case. The effect of the skill ratio on overeducation is very insignificant. This effect coincides with the expectation derived from the second case in the Slonimczyk-Skott model, but it is very likely that multicollinearity also has a part in this. If the time trend is removed from the regression the effect of ln is 0.073 with a significance level smaller than 1%. This indicates roughly a change of 0.00073 in overeducation with a 1% change in the skill ratio, but this result must be looked at critically. The extremely high R-squared of this model is the result of the high combined prediction power of the time trend and Ln . This may be the result of a spurious effect; this will be addressed in the discussion.
5. Conclusion and discussion
5.1 ConclusionBased on the results of the regressions the conclusion can be drawn that fluctuations in the real minimum wage have had no significant effect on unemployment in the Netherlands in the period from 1984 to 2013. In the regressions based on the study by Slonimczyk and Skott neither the minimum wage nor skill ratio in the labour force had any significant effect on either low-skill or high skill unemployment. The same results show in the regression for overeducation: neither minimum wage nor the skill ratio has a significant effect on overeducation. The insignificant effect of minimum wage on either high- or low-skill
unemployment shows a clear discrepancy between the data and the theory Slonimczyk and Skott use. The effect of minimum wage on low-skill unemployment and overeducation is either positive or negative in their model, but not zero. The only regression where minimum wage could have zero effect according to their model is in the regression for high skill
unemployment in the case where employers prefer low skilled workers in low-skill jobs. While this absence of an effect is also found in the data, one cannot say that this effect supports the Slonimczyk-Skott model as the other predictions based on the model do not hold.
Drawing a conclusion about the effect of the skill ratio on unemployment and overeducation is problematic because of the high collinearity between the time trend and the skill ratio. If the time trend is removed from the regression the skill ratio has a small negative effect on both high- and low-skill unemployment and a small positive effect on overeducation. The problem with these results is that they suffer from omitted variable bias because of the absence of the time trend. Overall the model does not seem to work as a good predictor of real effects in the Dutch situation. The model does not generate significant effects and also does not seem to fit the data very well. Regarding the main question of this study, the analysis shows that minimum wage does not have an effect on unemployment. This result does not coincide with either the classical model or a model with monopsonistic effects, as in these models minimum wage should affect unemployment.
5.2 Discussion
The fact that this study found no significant effects could, aside from a true absence of an effect, also be due to a number of factors regarding the model and the data. The first of these possible factors is the relatively small number of observations. The data in this study only contain 30 respectively 26 observations, this causes the results to be insignificant more quickly. Another likely cause of the insignificance of the gathered results is that the used model and underlying theory are microeconomic while many of the variables, like
unemployment, are macroeconomic. For example, the graph of the unemployment clearly shows a move together with the business cycle, something that is not captured in this model. Future studies looking at the effect of changes in minimum wage on unemployment might get better results by controlling for macroeconomic effects that effect unemployment but are not captured in the microeconomic theory of this model.
The regressions used in the study see the skill ratio as a determinant of
overeducation. Indeed, the two variables seem to rise together trough time at roughly the same rate. It could also be arguedthat overeducation creates credential inflation which in turn causes the normal level of education for a certain job to rise (Van de Werfhorst & Andersen, 2005). Hereby a person with a certain level of education and work in period t is seen as overeducated, but a person with the same level of education and the same job in period t+1 is no longer seen as overeducated. In this case the classification of occupations should also move along as time progresses to capture this effect. The use of an occupational classification with a static occupational level might underestimate the level of education needed for an occupation some time after the creation of the classification (and vice versa for occupations in a period before the classification was made). This in turn skews the
measurement of overeducation. Because research into the effects of minimum wage is often longitudinal, an adjustment in occupational levels might give a more real image of
References
BBC (2014). Angela Merkel approves Germany's first minimum wage. Found on: http://www.bbc.com/news/business-26851906 (11/06/2014).
CBS (2001). Standaard Beroepenclassificatie 1992. Voorburg: CBS –Sector Reproductie en Afwerking.
CBS (2014a). Inpassing van het Nederlandse onderwijs in ISCED 2011. Found on:
http://www.cbs.nl/NR/rdonlyres/4E6FB3F2-44E4-4517-8E9E-99AE77E76AEC/0/2011iscedinpassingnedond.pdf (19/06/2014). CBS (2014b). Beroepsbevolking; vanaf 1800. Found on:
http://statline.cbs.nl/StatWeb/selection/?DM=SLNL&PA=71882NED&VW=T (19/06/2014)
CBS (2014c). Enquête Beroepsbevolking (EBB). Found on: http://www.cbs.nl/nl- NL/menu/themas/arbeid-sociale-zekerheid/methoden/dataverzameling/korte-onderzoeksbeschrijvingen/onderzoeksbeschrijving-ebb-art.htm (22/06/2014). Dickens, R., Machin, S., & Manning, A. (1999) The Effects of Minimum Wages on
Employment: Theory and Evidence from Britain. Journal of Labor Economics, 17 (1), 1-22.
Eurostat (2014). Minimum wage. Found on:
http://epp.eurostat.ec.europa.eu/tgm/graph.do?tab=graph&plugin=0&pcode=tps00 155&language=en&toolbox=data (11/06/2014).
Machin, S., Manning, A., Rahman, L. (2002). Where the minimum wage bites hard: Introduction of minimum wages to a low wage sector. Journal of the European
Economic Association, 1 (1), 154-180.
Nguyen, C.V. (2012). The impact of minimum wages on employment of low-wage workers Evidence from Vietnam. Economics Of Transition, 21 (3), 583-615.
Pindyck, R., Rubinfeld, D. (2009). Microeconomics. Pearson: London
Rebitzer, J.B., Taylor, L.J. (1995). The consequences of minimum wage laws. Some new theoretical ideas. Journal of Public Economics, 56, 245–255.
SCP (2014). Arbeidsaanbodpanel (AAP). Found on:
http://www.scp.nl/Onderzoek/Bronnen/Beknopte_onderzoeksbeschrijvingen/Arbeid saanbo dpanel_AAP(24/06/2014)
Slonimczyk F., Skott P. (2012) Employment and distribution effects of the minimum wage
Journal of Economic Behavior & Organization, 84, 245-264.
U.S. Department of Education, 2014. EDUCATION INDICATORS: An International Perspective. Found on: http://nces.ed.gov/pubs/eiip/eiip1s01.asp (19/06/2014).
Van de Werfhorst, H.G., Andersen, R. (2005). Social background, credential inflation and educational strategies. Acta Sociologica, 48: 4, 321-340
