Disentangling the Minimum Wage Puzzle:
An Analysis of Job Accessions and Separations from a
Longitudinal Matched Employer-Employee Data Set
*Pedro Portugal **Ana Rute Cardoso
Changes in the legislation in mid-80s in Portugal provide remarkable conditions for economic analysis, as the minimum wage increased very sharply for a very specific group of workers. Relying on a matched employer-employee panel dataset, we model gross job flows — accessions and separations — in continuing firms, as well as in new firms and
those going out of business, using a Poisson regression model applied to proportions. Worker behaviour is as well modelled. Employment trends for teenagers, the affected group, are contrasted against older workers, before and after the rise in the youth minimum wage.
The major effect on teenagers of a rising minimum wage is the reduction of separations from the employer, which compensates for the reduction of accessions (to new and continuing firms) and the rising dismissals from firms closing down. This result points to the relevance of supply side factors overcoming demand forces, as they indicate that job attachment for low wage youngsters rises following an increase in their minimum wage. In this sense, our results can reconcile some of the previous evidence that has been presented in the empirical literature when analysing the overall impact of the minimum wage on youth employment without looking at its sources.
Key words: minimum wage, employment.
JEL: D21, J38.
* Bank of Portugal and Universidade Nova de Lisboa
** Universidade do Minho, NIMA, Portugal
Visitor, Economics Program, RSSS, Australian National University
Data for this study were provided by the Department of Statistics, Ministry of Employment and Solidarity. Part of this work was carried out while the second author was visiting the Bank of Portugal, whose support is gratefully acknowledged. We are grateful to Lucena Vieira for her excellent research assistance.
When analysing the impact of the minimum wage, labour economists have invariably been constrained by two types of factors. On one hand, the impact of the minimum wage is hard to isolate from that of other economic forces. Ideally, the conditions for quasi-experiments should be met, in which case changes in the minimum wage would affect strictly a specific group of workers (treatment group), subject to conditions otherwise similar to other workers (control group), a situation that very rarely occurs in economics. The studies by Card and Krueger (1994, 1995) and Katz and Krueger (1992) have become famous examples of that methodology. However, even when changes in the minimum wage are large,
unpredictable and affect just a particular group of workers, the study to be undertaken still depends crucially on the data available, in particular its quality and detail.
On both of these fronts, our study relies on unusually good conditions, enabling the development of the analysis into directions left unexplored by previous studies. In 1987, the minimum wage for workers aged 17 increased in Portugal by 50% strictly due to changes in the legislation, as it was raised from half to 75% of the full minimum wage. Also in 1987, the minimum wage for workers aged 18 or 19 was raised from 75% to the full minimum wage rate, therefore increasing by 33% due to legal changes ceteris paribus. Instead of defining a priori youngsters as a
proxy for the group of workers more severely hit by a rising general minimum wage, this study is motivated by a rise in the minimum wage specifically for youngsters, and we are able to directly evaluate the impact on their employment against different control groups.
Longitudinal data matching workers and firms and covering the population of wage-earners in the manufacturing and services private sector in Portugal will be used.
Relying on the quality of this large panel data set, we can, first of all, quantify and model gross job flows at the firm level. Several studies of the impact of the minimum wage have quantified and explained net job flows at the firm level (Card and Krueger 1994, Katz and Krueger 1992, Pereira 1999). Given that we are able to follow, not just firms, but workers as well, our study adds to the previous
literature by quantifying and modelling employer behaviour regarding both accessions and separations. The analysis is not restricted to firms remaining in business, focusing as well new firms and those closing down. We can therefore disentangle the minimum wage puzzle, by identifying exactly where and how the minimum wage bites: Do employers dismiss mainly youngsters, following a rise in the youth minimum wage? Or is it the case that youngsters become under-represented among the newly hired workers? Do firms that are about to be set up, which are free to choose their labour force from the available pool of workers, hire relatively less youngsters than comparable firms before the minimum wage had been raised? Does a rising minimum wage place unbearable constraints on the firms, contributing to firm closure? Identification of the precise source for changing employment levels can reconcile some of the evidence that has previously been presented in the literature as contradictory.
Secondly, we will rely on micro data on workers to follow a line of analysis previously used by Abowd, Kramarz and Margolis (1999), Zavodny (2000) and Currie and Fallick (1996). At the individual level, we can provide a direct answer to the question: are the workers directly affected by a rise in the minimum wage less likely to keep their jobs?
Section 2 provides a detailed description of the institutional framework in Portugal and the changes that took place in the minimum wage. Section 3 overviews changes in the wage distribution and in employment. Section 4.1 models job accessions and job separations in continuing firms, whereas section 4.2 performs a similar exercise for new firms and for firms closing down. Worker behaviour is the subject of section 5 and in the last part of the paper concluding comments are presented.
2. Institutional framework
The national minimum wage was set in 1974, covering at the time workers aged 20 or older and excluding agriculture and domestic work. Ever since, it has been
1. Decisions on annually updated by the parliament, under government proposal
1 The only exceptions were 1982, when it was not updated, and 1989, when it was updated twice.
the level of the minimum wage are taken on a discretionary basis, usually taking into account past and predicted inflation and after consulting the social partners. It is defined as a monthly wage.
Since it was first enacted, the minimum wage has undergone several changes,
2: which can be summarised as follows
– Industry coverage: in 1977, it was extended to cover workers in agriculture,
though their wage rate was lower; the following year, domestic work was
covered as well, at a rate lower than agriculture. In 1991, the agricultural and
the general minimum wages were harmonised and in 1998 the same
happened to domestic work.
Raising the minimum wage for agriculture and for domestic work to the general level took place gradually and the full harmonisation was therefore highly predictable. As such, these changes in the legislation do not provide the conditions for a quasi-experiment to study the impact of changes in the minimum wage. Also, the coverage of agriculture in the data set used is low. – Firm size coverage: small firms outside agriculture with fewer than 5 workers
were allowed to pay the lower minimum set for agriculture, a possibility
introduced in 1978 and revoked in 1991. Before 1978, several regimes had
existed: in 1974, firms with fewer than 5 workers were exempted; in 1975 and
1976, firms with fewer than 10 workers were declared exempted as well; in
1977 firms with fewer than 10 workers outside agriculture were allowed to pay
the minimum set for agriculture, if they proved to be on difficult financial
On request, firms claiming that the application of the new minimum wage
would make them undergo an unbearable rise in labour costs, could as well be
allowed to pay the minimum wage set for agriculture. This possibility was
introduced in 1978. In 1987, only firms with fewer than 50 workers were
eligible for this reduction, and the firm size benchmark was lowered to 30
workers in 1988, 20 workers in 1989 and it was revoked in 1990.
2 Throughout the period, minimum wage reductions applied to handicapped workers.
Since the minimum wage for agriculture and the general one had been
gradually harmonised, the abolition of these exemptions did not lead to a
remarkable rise in the applicable minimum wage. On the other hand, the data
set does not allow identification of the firms that were requesting and using the
latter minimum wage reduction, and it is thus not feasible to study the impact
of these changes.
– Age coverage: the ages subject to exemption, as well as the percentage of
exemption, changed several times, and a clearer description can be provided
by a table. The major changes are highlighted in bold.
Table 1 – Share of the general minimum wage enforced, according to the age of the worker
age ; 15 16 17 18 19 20-25 year ；
1979 to 1986 50% 50% 50% 75% 75% 100%
100% 100% 100% 1987 75% 50% 50% (80% if apprentice) (80% if apprentice) (80% if apprentice)
100% 100% 100% 1988 75% 75% 75% (80% if apprentice) (80% if apprentice) (80% if apprentice)
Source: Portugal, Diário da República, several issues.
Figure 1 provides a longitudinal perspective, clarifying the changes in the exemptions to the minimum wage applied to a worker in Portugal, as he/she grows older.
Figure 1 – Growth in nominal minimum wage for different age categories, 1986-91
aged 15 in 1986aged 16 in 1986
311.1311.1311.1300300285.2285.2266.7266.7250248.9250248.9241.8233.3213.3213.3200200.0200200.0193.4181.3181.3178.2181.3178.2index base 1986168.0168.0155.6index base 1986150155.6150133.3120.9120.9133.3112.0112.0112.0112.0120.9100100.0100.0100.0100.0112.0112.0100100.0100.0100.0100.0
01986198719881989199019910198619871988198919901991adult minimum wagejust age effectage+legislation effectsage+legislation effects if apprentice adult minimum wagejust age effectage+legislation effectsage+legislation effects if apprentice
aged 17 in 1986aged 18 in 1986400250237.6237.6356.4356.4350207.4207.4200311.1311.1190.1300285.2178.2177.8177.8266.7266.7165.9161.2161.2250248.9155.6241.8150149.3142.2224.0213.3133.3128.9200193.4120.9119.5181.3179.2112.0112.0178.2168.0index base 1986index base 1986100100.0100.0100.0100.0155.6150133.3120.9112.0100100.0100.0100.0100.050
adult minimum wagejust age effectage+legislation effectsage+legislation effects if apprenticeadult minimum wagejust age effectage+legislation effectsage+legislation effects if apprentice
Three changes in the legislation have yielded remarkable wage increases for the ages involved:
– first of all, in 1987 the minimum wage for workers aged 17 increased by
50% due to changes in the legislation ceteris paribus, as it was raised from
half to 75% of the full minimum wage;
– secondly, also in 1987, the minimum wage for workers aged 18 or 19 was
raised from 75% of the minimum wage to the full minimum, therefore
increasing by 33% due to legal changes ceteris paribus;
– the third change took place the following year, in 1988, as the minimum
wage for workers aged 16 or less was raised from 50% to 75% of the full
minimum, thus increasing by 50% due to changes in the legal setting ceteris
Exceptions could however apply to the second change described, if the worker was an apprentice, in which case he would be entitled to 80% of the general minimum wage, instead of the full rate. This exemption could undermine the impact of the rise in the minimum wage that we intend to study, if employers decided to declare the worker as an apprentice to get away without paying him the full minimum wage.
Our study concentrates on the changes that took place in 1987. These changes in the legal framework of the minimum wage in Portugal define the setting for a quasi-experiment in economics, as first argued by Pereira, who used in her study the second change described (for more details, see Pereira (1999)). 3. Overview: aggregate data suggest that youth employment increased
following a sharp rise in the youth minimum wage
Preliminary evidence on the impact of the rise in the minimum wage, based on aggregate figures, suggests that traditional economic theory is challenged. The evidence is instead consistent with the literature indicating that a rise in the minimum wage does not inexorably lead to declining employment.
Note first of all the impact of the legal changes on the average wage of the different age groups, relying on three cross-sections of data.
Table 2 – Growth in the average wage of full-time wage earners, by age group, 1986-1989 (*)
Age 1987 1988 1989
1.21 16 1.16 1.17
1.21 17 1.14 1.15
1.21 18 1.12 1.13
1.18 19 1.12 1.13
20 1.15 1.11 1.13
21 1.14 1.12 1.12
22 1.14 1.12 1.13
23 1.14 1.12 1.14
24 1.14 1.12 1.14
25-29 1.14 1.10 1.13
30-34 1.15 1.10 1.15
35-39 1.14 1.09 1.13
40-65 1.16 1.11 1.14
>=66 1.16 1.12 1.13
Source: Computations based on Portugal, MTS (1986-1989).
wNote: (*) Computed as , where stands for the average wage of the age group a, wwaa,ta,t？1
and the subscript t denotes time.
Wages increased most sharply precisely for the groups of workers affected by the rise in the minimum wage (see highlights in bold). Between 1986 and 1987, the average wage for workers aged 17, 18 or 19 increased the most, while all the other age categories had a much lower and very homogeneous wage growth. The rise in the legal minimum wage thus seems to have had a relevant impact on the wages of the eligible workers, indicating that the possibility to declare a worker as
3an apprentice was not extensively used by employers. The same pattern holds
for the year afterwards, as the groups of workers affected by changes in the minimum wage — those aged 16 or less — saw their average wage rise the most.
Visual inspection of the wage distribution stresses the relevance of the changes that took place.
3 Note also that the share of apprentices in the age categories directly affected by the change in the legislation did not increase during this period (see appendix).
Wage distribution, teenagers, 1986 versus 1987
In 1986, the wage distribution for teenagers aged 17 to 19 presented a sharp peak at 75% of the national minimum wage. That was precisely the subminimum wage legally allowed for workers aged 18 or 19. The distribution presented a less pronounced peak at the full minimum wage. By 1987, the subminimum wage peak had almost vanished, and the share of teens at the current minimum wage had increased sharply. The kernel plot therefore indicates that compliance to the new minimum wage regulation was widespread, as the rise in the applicable minimum wage did shift subminimum teenager wages to the new minimum wage level. The impact on employment resulting from these sharp wage changes does not exhibit the pattern predicted by competitive theory.
Table 3 – Growth in number of full-time wage-earners by age groups, 1986-1989 (*)
age 1987 1988 1989
1.148 16 1.060 1.092
1.052 17 1.094 1.119
1.043 18 1.090 1.106
1.061 19 1.066 1.108
1.086 20 1.042 1.108
1.082 21 1.051 1.088
1.109 22 1.054 1.099
23 1.063 1.083 1.095
24 1.055 1.041 1.136
25-29 1.014 0.995 1.071
30-34 1.023 1.002 1.032
35-39 1.029 0.983 1.001
40-65 1.027 0.996 1.049
>=66 1.041 0.988 1.096
total 1.033 1.012 1.077
Source: Computations based on Portugal, MTS (1986-1989).
LLLNote: (*) Computed as , where stands for employment of age group a, and the aa,ta,t？1
subscript t denotes time.
While total employment increased by 3.3% between 1986 and 1987, the number of jobs taken by workers aged 17 to 19 increased more sharply, by 4% to 6%. Nevertheless, the group aged 20 to 22 presented even higher employment growth rates. The following year, while total employment increased by 1%, the youngest workers had a much higher employment growth, which compensated for the decline in the employment of prime-age workers. This level of analysis therefore dismisses the existence of any negative impact on employment resulting from the sharp rise in the minimum wage for specific age groups.
A longitudinal perspective can shed another light on the issue. Following workers over time, we can directly check whether youngsters who were earning sub-minimum wages tended to loose their jobs following the sharp increase in their legal minimum wage. Results reinforce the initial hint, as the rise in the minimum wage did not have the negative employment consequences predicted by traditional economic theory.
Table 4 – Mobility out of employment for 1986 full-time wage-earners, by age and wage categories
situation in 1987 age in wage in 1986 employed wage earner out of 1986 full-time not full-time employment
16-18 below minimum 65.96 4.24 29.80 at minimum 59.91 3.66 36.42
between min86 and min87 75.65 4.41 19.94
at/barely above min87 * 67.87 4.10 28.03 above min87 x 1.1 71.72 4.71 23.57
19-25 below minimum 56.55 5.12 38.33
at minimum 57.63 3.26 39.11 between min86 and min87 67.66 4.79 27.54
at/barely above min87 * 65.91 4.32 29.77
above min87 x 1.1 71.66 3.70 24.64
26-35 below minimum 56.67 6.32 37.02 at minimum 60.06 3.45 36.49
between min86 and min87 70.64 5.46 23.89
at/barely above min87 * 71.83 4.32 23.85 above min87 x 1.1 78.09 3.56 18.36
36-65 below minimum 45.61 22.85 31.53
at minimum 23.7 3.75 72.55 between min86 and min87 56.67 6.32 37.02
at/barely above min87 * 60.06 3.45 36.49
above min87 x 1.1 70.64 5.46 23.89
Source: Computations based on Portugal, MTS (1986-1987).
Note: (*) ―Barely above‖ is defined as not exceeding the minimum plus 10%.
Workers aged 17 to 18 and earning sub-minimum wages in 1986 were a year later eligible for the full rate minimum wage, thus experiencing a sharp wage rise. Those aged 16 had as well a notable increase in the applicable minimum wage. These workers would therefore, according to competitive theory, be at risk of loosing their job. If workers were paid the value of their marginal product, the rise in the minimum wage would drive sub-minimum workers out of employment.