Common menu bar

• Français
• Home
• Contact Us
• Help
• Search
• canada.gc.ca

Breadcrumb

1. Home >
2. Employment Standards

Archived - Minimum Wages in canada : theory, evidence and policy

Previous Table of Contents Next

Canadian Studies

Kan and Sharir 1996

Kan and Sharir (1996) estimate pooled cross-section (9 provinces) time-series (1975-91) regressions on the impact of minimum wages. Although they do not report the magnitudes of the effects, they indicate (p. S 56) that minimum wage increases had “no effect, or to have a negative effect, on employment … [and] a negative effect on labour force participation, especially for males.” Contrary to most other empirical studies they also found “no effect on the employment of the two youngest groups.” It is difficult, however, to assess why their results appear different from most others, because they only report the conclusions and not the actual results.

Shannon and Beach 1995

Shannon and Beach (1995) provide simulation evidence of the potential impact of Ontario’s proposed 35% increase in the minimum wage (from $5.00 to $6.75) that was to be phased-in between 1991 and 1995. They base their analysis on the 1989 Labour Market Activity Survey (LMAS) where the individual is the unit of observation.

They first determine who would potentially be affected by the minimum wage increase in that their wage falls between the old and the new minimum. Such individuals tend to be young, part-time workers, less educated, in the retail, accommodation and food industries and from families with low earnings. For these workers they then calculate their wage gap as the amount that their wages would have to increase to be in compliance with the new minimum wage. This averages 23.3% -- that is, on average the wages of those affected by the minimum wage would rise by 23.3%. This is 2.3% of the wage bill across all workers, not just those affected by the minimum wage. This wage increase of 23.3% is then multiplied by conventional labour demand elasticities obtained from the literature (-0.33 for employment and –0.51 for hours) to get the expected impact on employment and on hours worked. These calculations yield a reduction in employment of 7.7% for those affected by the minimum wage increase or 1.5% across all workers, and a reduction in hours of work of 11.9% for those affected by the minimum wage increase or 1.3% across all workers. Adjusting these figures for an 80% compliance and coverage rate yields a reduction in employment of 6.2% for those affect by the minimum wage increase or 1.2% across all workers, and a reduction in hours of work of 9.5% for those affected by the minimum wage increase or 1.1% across all workers. These are effects that would result from a proposed 35% increase in the minimum wage – converting them to elasticity equivalents from a 10% increase in the minimum wage would entail dividing them by 3.5.

Shannon and Beach also indicate that about 20% of the Ontario workforce would have been affected by a minimum wage increase of that magnitude, albeit only about 11 % of hours worked would be affected since most minimum-wage workers work part-time. Of those affected, 60% would be youths (under 25) and about 50% would be full-time students who work part-time. Of the additional earnings emanating from the higher minimum wage, only 28% would go to low-income families, 41% to middle-income families and 31% to high-income families. They indicate that the minimum wage increase would reduce the proportion of workers who are in families that had family earnings below the Statistic’s Canada low-income cut-off rate by a meagre amount, from 16.9 percent of workers to 16.6 percent. This leads them to conclude (p. 300) that “the policy suffers from a targeting problem … a better targeted policy focusing more exclusively on the working poor (e.g., skills upgrading) may be advisable.”

Shannon 1996

Shannon (1996) utilises the Labour Market Activity Survy (LMAS) for 1986 to estimate the impact of minimum wage legislation on the male-female earnings gap. This entails estimating the hypothetical male-female wage gap that would prevail in the absence of the minimum wage, utilising the procedure outlined in Meyer and Wise (1983a, 1983b) for estimating the impact of minimum wages on the wage distribution.

Shannon finds that for youths age 16-24 the male-female wage gap would have been almost twice as large as it was were it not for minimum wages. About half of this can be attributed to the fact that minimum wages had a disproportionate impact on raising the wages of low-wage females compared to males, and the other half due to the fact that it correspondingly had a disproportionately larger adverse employment effect on females so that they were no longer in the wage distribution. For adults age 25 and over, minimum wages had only a small effect (almost zero) on reducing the male-female wage gap and all of it was due to the larger adverse employment effect for females. Overall, it appears that minimum wages can reduce the male-female earnings gap for youths (16-25) but not for adults, and that slightly over half of the effect for youths comes about because of the greater adverse employment effect for females – they no longer contribute to the gap because they are no longer employed.

Baker, Benjamin and Stanger 1999

Baker, Benjamin and Stanger (1999) use pooled cross-section (9 provinces) time series (1975-93) data to estimate the impact of minimum wage changes that occurred both over time and across the different jurisdictions. They estimate a wide range of specifications and filter the time series data in a way that approximates short-run (1-2 years) and long-run (4-6 years) adjustments to the minimum wage changes.

Their results indicate that the short-run employment effects are generally small, statistically insignificant and sometimes even positive. However, the longer run employment effects are negative and statistically significant, with a 10% increase in minimum wages generally leading to approximately a 2.5% reduction in employment. This minimum wage elasticity is towards the higher end of the earlier consensus

They suggest that this may partially reconcile some of the differences in the US results that have emerged, as discussed previously. That is, the natural experiment results which generally yielded no adverse employment effects and possibly even positive effects were essentially measuring short-run adjustments. The other studies, and especially those with lagged minimum wage effects, were estimating longer-run responses, and these were generally negative and more substantial. In essence, an adverse employment effect occurs after employers have had sufficient time to adjust through attrition and turnover (of both firms and employees) as well as through altering their production processes and the substitutions of other inputs.

Goldberg and Green 1999

Goldberg and Green also use pooled time-series (1976-97) cross-section data (Quebec, Ontario, Alberta and British Columbia). They find that a 10 percent increase in the minimum wage leads to a decrease in the employment of 1.6% for female teenagers, and 1.4% for male teenagers, but both are statistically insignificant. For young adults (20-24) the impacts are smaller (1%) but statistically significant for males although essentially zero for females. For adults 25-54 they are essentially zero for males and 1% (and significant) for females. They also find that minimum wages did increase the total wages paid to low-wage workers.

Fortin and Lemieux 2000

Fortin and Lemieux (2000) use the 1988 Labour Market Activity Survey (LMAS) and the 1995 Survey of Work Arrangements (SWA) to simulate the effect of minimum wage changes on the distribution of wages and of family income. They find that minimum wages do reduce wage inequality and that (p. 240): “individuals in the lower half of the family income distribution benefit the most from the minimum wage. Individuals in this part of the wage distribution account for 70 percent of the earnings of all minimum-wage workers.”

Benjamin 2001

Benjamin (2001) finds that there are little or no spillover effects, based on estimates of the effect of minimum wage changes on average wages in various industries over the period 1975-1993. Using the 1990 Labour Market Activity Survey he indicates that the bottom decile of the family income distribution has no low-wage workers since these families have no workers. However, 17% of families in the second lowest decile have a low-wage worker, and this proportion generally declines with each higher family income decile. Overall, he concludes (p. 217) that “Minimum wages have only a limited scope for improving the welfare of the lowest income households, since most of these families have no full-time earners. Similarly, many of the benefits of higher minimum wages are transferred to teens who are distributed relatively evenly across the income distribution. Nevertheless, the benefits would flow disproportionately to poor working adults.”

Yuen 2003

Yuen (2003) uses the Labour Market Activity Survey (LMAS) panel data from 1988-90 to estimate the impact of minimum wage changes over that period. The methodology involves estimating the employment transition probabilities for those “at risk” of being affected by a minimum wage increase (treatment group) compared to those not at risk (control group).

When the control group is all individuals whose wage is not in the range of the minimum wage increase (thereby consisting mainly of high-wage individuals) persons affected by a minimum wage increase (which averaged 8.4% in the different jurisdictions over that time period) were 7% less likely to be employed in the subsequent time period in the case of teens and 15% less likely in the case of young adults. These results were fairly similar to US results using the same methodology and based on the “high-wage” control groups. The US studies had to rely on high-wage control groups since most minimum wage changes in the US are the result of federal changes and hence are uniform across the country so there are no comparable low-wage groups that did not have a minimum wage change. The reliance on high-wage control groups has been a criticism of these studies since the results may simply reflect how other factors may have a different effect on high- and low-wage groups (Card and Krueger 1995, p. 228).

To ascertain if this is a concern, Yuen exploits the fact that Canada does have low-wage control groups in other provinces who are not affected by minimum wage changes since minimum wage legislation in Canada is under provincial jurisdiction. When he restricts the control group to low-wage individuals the effects become statistically insignificant and quantitatively small, suggesting that the criticisms were correct; that is, the large adverse employment effects were the result of common forces having a different impact on the low-wage treatment group affected by minimum wage increases and the high-wage control groups.

However, Yuen further divides his low-wage control groups into those who are “transitory” low-wage workers in that they have less than 3 quarters of low-wage employment in the three-year sample period, and “permanent” low-wage workers who have 3 or more quarters of low-wage employment in the sample period. The “temporary” low-wage workers tend to be full-time students working in low-paid summer jobs or normally high-wage workers who are in a temporarily low-paid job. In such cases, wages are not likely to reflect their normal longer-run productivity. When he restricts his analysis to the permanent low-wage control group, he gets larger and significant adverse employment effects that are similar to the conventional estimates that include the high-wage control groups. Specifically, he finds that workers whose wages are affected by a minimum wage increase (which averaged 8.4%) are 7% less likely to be employed in the subsequent period in the case of teens and 10% less likely in the case of young adults, with the effects being statistically significant. These are substantial adverse employment effects for those who are affected by the minimum wage increase, although they will be less prominent for all teens and young adults since most are not affected by the minimum wage increases.

Campolieti, Fang and Gunderson 2005a

Campolieti, Fang and Gunderson (2005a) also estimate employment transition probabilities based on the “at risk” methodology. They utilize the longitudinal nature of the Master File of the Survey of Labour and Income Dynamics (SLID) for the period 1993–1999 to compare transitions from employment to non-employment for individuals affected by minimum wage changes with appropriate comparison groups not affected by minimum wages. This is based on the large number (24) of minimum wage changes that have occurred across the different provincial jurisdictions in Canada over the 1990s. The results indicate that the minimum wage increases have increased the transition from employment to non-employment of employed low-wage youths, who are at-risk of being affected by a minimum wage increase, by around 6 percentage points (ranging from 4 to 8 percentage points). These dis-employment effects in turn imply fairly high ‘minimum wage’ negative elasticities of about - 0.4 (ranging from -0.3 to - 0.5).

Campolieti, Fang and Gunderson 2005b

Campolieti, Fang and Gunderson (2005b) also use the longitudinal data from the Master File of the Survey of Labour and Income Dynamics (SLID) for Canada for 1993-99 to estimate multinomial logit equations of the effect of minimum wages on the probability of being in one of four schooling-employment states as well as transitions across the states. They find that minimum wage increases led to large and statistically significant reductions in the employment of teenagers but had no net effect on their school enrolment (in contrast to the more recent US studies which found that it reduced school enrolment). They also find no substantial substitution of students for non-students or students leaving school to queue for the higher minimum wage jobs.

Baker 2005

Although the focus of his analysis is on the impact of minimum wages on human capital formation (education and training), Baker (2005) also updates earlier estimates of the employment effect from Baker, Benjamin and Stanger (1999). His replication for the same time period of 1983 – 1993 yields a statistically significant elasticity of employment with respect to minimum wages for teens of -0.32, very close to the earlier estimates -0.37 (Statistics Canada revisions to the data accounting for the difference). Updating the data for the full time period 1983 to 2000, however, yields a much larger negative elasticity of -0.57. Alternative specifications yield an elasticity of -0.49. Since this is for the full period 1983-2000, and the elasticity for the earlier period of 1983-1993 was -0.37, this implies that the negative elasticity was considerably higher than -0.57 over the more recent period of the 1990s; that is, the adverse employment effect increased substantially over the 1990s. Taking an elasticity of -0.5 as a reasonable estimate over the full period implies that a 10 percent increase in the minimum wage led to a 5 percent decrease in the employment of teens over that full period 1983 to 2000. For youths age 20-24, the elasticities are smaller but also much larger in the full 1983 to 2000 period to include the more current data from the 1990s (-0.09 in the earlier 1983-1993 period, and -0.17 to -0.20 in the full 1983 to 2000 period).

Baker (2005) attributes the greater adverse employment effect to the increasing proportion of youths affected by the minimum wage increase. Baker, Benjamin and Stanger (1999) reported that in the 1983-1993 period, 13 percent of employed teens held jobs paying within 5 cents of the minimum wage; by 1997 (half-way through the updated period), Baker (2005) calculates that 25% of employed teens were earning within 5 cents of the minimum wage in their jurisdiction.

Baker (2005) argues that this adverse employment effect has the largest indirect but “first-order” effect on reducing the training opportunities for youths since they miss out on work related training and experience. He then estimates the more direct effect of minimum wage on training using the 1992, 1994 and 1998 Adult Education and Training (AETS) Surveys. He generally finds negative effects on training, sometimes small and statistically insignificant, and sometimes large and statistically insignificant. He concludes, however, that the data are not adequate to make precise statements of this more direct impact of minimum wages on training, although he argues that the more indirect effect through employment is substantial and negative.

Baker (2005) also estimates the effect on education enrolment. The results are mixed. He concludes that minimum wage laws have no effect on schooling for teens age 15-16 in part because compulsory school laws inhibit them from leaving school to queue for the minimum wage jobs. They do, however, reduce the proportion of such teens who work while in school although the effects are small. For the older age groups of youths who can leave school, the effects of minimum wages are modestly positive on staying in school. For 17-19 year olds, this is also associated with a reduced likelihood of working while in school, while for 20-24 year olds it is associated with an increased likelihood of working while in school. The magnitudes of the effects, however, are generally small.

Campolieti, Gunderson and Riddell (forthcoming)

Based on the methodology of a pre-specified research design, Campolieti, Gunderson and Riddell (forthcoming) use pooled cross-section time series from the Labour Force Survey (LFS) and the Survey of Consumer Finances (SCF) for years 1981 through 1997. This yields considerable time series and cross-section variation in provincial minimum wages (71 changes) over the time period. They find substantial adverse employment effects for youths (16-24) with negative minimum wage elasticities typically in the range of -0.14 to -0.44, with -0.30 being a reasonable point estimate, and with the effects being larger after lagged adjustments.

Conclusions from Canadian Studies

While there are substantial differences across the different Canadian studies, the following generalisations emerge:

• The earlier Canadian studies (based on data prior to the 1980s) tended to find adverse employment effects that were in the range of US consensus estimates, and sometimes higher, where a 10% increase in the minimum wage would give rise to a 1-3% reduction in employment.
• Studies based on data to include the 1980s tended to find smaller effects that were at the lower end of the consensus range, and possibly zero, as was often also the case in the US[21].
• However, some more recent studies using different and more sophisticated methodologies as well as more recent data (e.g., Baker, Benjamin and Stanger 1999, Yeun 2003, Baker 2005, Campolieti, Fang and Gunderson 2005a, b, Campolieti, Gunderson and Riddell, forthcoming) find larger adverse employment effects at the higher end and beyond the consensus range, especially in the longer run. The elasticities typically range from -0.3 to -0.6 for teens (slightly lower for young adults), implying that at 10 percent increase in the minimum wage would lead to a 3 to 6 percent reduction in the employment of teens. The fact that they use different data sets and methodologies suggest that these results are robust.
• Overall it appears that the Canadian studies tend to find adverse employment effects that are at least as large and likely larger than US studies; certainly none find positive employment effects as occasionally occurs in the US.
• Minimum wage increases also tend to reduce the labour force participation rate inducing some to leave the labour force and this means that not all of the employment reductions get translated into unemployment rate increases.
• There is some evidence of wage spillover effects but not in all studies.
• There is no substantial impact on schooling, although there may be some weak positive effect for older youths.
• Although data problems preclude estimating robust results, the effects on training are generally negative, although sometimes small and statistically insignificant. The most likely negative effect, however, is indirect resulting from the more substantial adverse employment effect that precludes accumulating on-the-job training and experience.
• Minimum wages tend to reduce wage inequality and disproportionately benefit low-income families. As an anti-poverty device, however, they are an exceedingly blunt instrument and not well targeted towards the poor for various reasons:
• many of the poor do not work;
• those that do often work few hours;
• there is the risk of an adverse employment effect;
• minimum wages disproportionately affect teens who are distributed throughout the family income distribution; and
• minimum wages affect individual wages while poverty is defined in terms of family income and need.
• There is, perhaps surprisingly, no direct published evidence on the differential impact of raising minimum wages in times of high unemployment or low unemployment. Basic theoretical reasoning (and the experience of Britain discussed previously) would suggest that any adverse employment effect would be mitigated if minimum wages were raised in periods of low unemployment in part because it would more likely not to be a binding constraint. As stated by Sussman and Tabi (2004, p. 6) in commenting that only 1.1% of workers in Alberta were working at the minimum wage compared to 4.1% across all of Canada and 8.5% in Newfoundland and Labrador: “more opportunities in Alberta may have translated into greater bargaining power for workers.” Certainly, any adverse employment effect would be masked by a tight labour market where its manifestation would be only slower employment growth.

Previous Table of Contents Next

Primary navigation (left column)

Date Modified:

2011-11-22