Journal of Labor Research

, Volume 33, Issue 3, pp 287–302

The (Non) Impact of Minimum Wages on Poverty: Regression and Simulation Evidence for Canada

Article

DOI: 10.1007/s12122-012-9139-8

Cite this article as:
Campolieti, M., Gunderson, M. & Lee, B. J Labor Res (2012) 33: 287. doi:10.1007/s12122-012-9139-8

Abstract

We estimate the effect of minimum wages on poverty for Canada using data from the Survey of Labour and Income Dynamics (SLID) for 1997 to 2007 and find that minimum wages do not have a statistically significant effect on poverty and this finding is robust across a number of specifications. Our simulation results, based on the March 2008 Labour Force Survey (LFS), find that only about 30 % of the net earnings gain from minimum wage increases goes to the poor while about 70 % “spill over” into the hands of the non-poor. Furthermore, we find that job losses are disproportionately concentrated on the poor. Our results highlight that, political rhetoric not-withstanding, minimum wages are poorly targeted as an anti-poverty device and are at best an exceedingly blunt instrument for dealing with poverty.

Keywords

Minimum wages Poverty Canada 

JEL Classification

J30 J38 J80 J13 

Introduction

Curbing poverty is the primary rationale given for minimum wage legislation. This is the case historically, currently and internationally. Historically when minimum wages in the US were first introduced in the Fair Labour Standards Act in 1936, alleviating poverty was the main rationale provided. More currently, that rationale was espoused by President Clinton in his 1998 statement that minimum wage “will raise the living standards of 12 million hard-working Americans” and Senator Kennedy’s statement that “the minimum wage was one of the first – and still one of the best – anti-poverty programs we have.”1 While campaigning for the presidency, Barak Obama also endorsed raising the minimum wage so that “people who work full-time should not live in poverty” (Sabia and Burkhauser 2008 p. 5). The same primacy of the anti-poverty rationale both historically and currently prevails in Canada (Gunderson 2005) and internationally more generally (OECD 1998). In Canada, for example, the 2008 Ontario Budget (p. 46) indicated that increasing the minimum wage to $10.25 by 2010 was part of the “government’s early initiatives to help reduce poverty.” Battle (2011) indicates that the substantial increases in minimum wages that have occurred recently in every jurisdiction in Canada is due in large part to poverty reduction strategies in the various jurisdictions.

The political appeal for minimum wages as an anti-poverty device is obvious. Raising the wages of the working poor should enhance their earned income and reduce their likelihood of being in poverty. Increasing the wages of the working poor is appealing to both taxpayers and recipients since it may provide an incentive to leave welfare and it enables recipients to “earn” their income rather than receive it in the form of transfer payments. The potential downsides of minimum wages as an anti-poverty device – adverse employment effects and other subtle longer-run adjustments discussed subsequently – are less visible.

This paper examines the effect of minimum wages on poverty in Canada. Canada is a particularly good laboratory for studying this impact because minimum wages are under provincial jurisdiction with almost complete coverage, and there have been numerous minimum wage increases over different time periods across those jurisdictions, providing considerable variation from which to identify minimum wage effects. Minimum wage effects for Canada, for example, were estimated from 24 minimum wage changes in Campolieti et al. (2005a, b) based on individual micro data over the seven year period 1993–1999, and from 71 minimum wage changes in Campolieti et al. (2006) based on aggregate data over the 19 year period 1981–1999. These desirable features of Canadian data for testing minimum wage impacts have been explicitly highlighted by Hamermesh (2002, p. 716) and Neumark and Wascher (2008, p. 91).

The paper begins with a discussion of the possible mechanisms whereby minimum wages can affect poverty, with references to the literature on those mechanisms. This is followed by outlining our two empirical procedures: regression estimates of the impact of minimum wages on poverty; and simulation estimates based on expected earnings gains from those who retain their job after a minimum wage increase and subtracting from this the expected earnings loss from those who lose their job because of a possible adverse employment effect. The results are then discussed followed by a summary and some concluding observations.

Mechanisms Whereby Minimum Wages Can Affect Poverty

There are a variety of pathways through which minimum wages can affect poverty. On the positive side in terms of reducing poverty, minimum wages can increase the wages of the working poor, enhancing their earned income and reducing their likelihood of being in poverty.

This positive effect may be muted somewhat by the fact that minimum wages tend to affect the wages of students, teens, youths and multiple earners in families that are non-poor and often relatively wealthy. This is increasingly the case given the dominance of multiple-earner families and the increased number of students working in the growing service economy.2 For example, based on unpublished Labour Force Survey data, Battle (2003) indicates that fewer than 3 % of workers between the ages of 25–64 (the age group for which working poverty is likely to be of greatest concern) work in minimum wage jobs in Canada. In contrast, about 60 % of minimum wage workers are teens or youths who live with their parents, 25 % are couples (three-quarter of which have a spouse employed at a job above the minimum wage), 11 % are unattached individuals and 4 % are single heads of families. In essence, minimum wages appear poorly targeted in both dimensions of horizontal equity (helping many persons in the target group of poor persons and helping them sufficiently to escape poverty) and vertical equity (few spill-overs into the hands of the non-poor). The poor targeting of minimum wages towards those in poverty is a common theme in much of the literature on minimum wages.3

On the negative side in terms of increasing poverty, higher minimum wages may reduce the employment opportunities and hours of work of such persons. These adjustments are not likely to occur in the form of overt layoffs or dismissals; rather, they would occur in the form of more subtle slower growth of employment or hours of work in such low wage jobs.4 The adjustments may be slow and subtle in such forms as hiring freezes or changes in production processes away from using low-wage labor and into self-service (e.g., gas stations, cafeterias, check-out counters) or pre-packaged foods or on-line help services as opposed to personal services.

Minimum wages can have long-run effects on human capital formation in forms such as training and education that can ultimately affect poverty. Minimum wages can reduce training by inhibiting workers from accepting lower wage jobs in return for the training that could enhance their subsequent wage growth. In contrast, they could enhance training on the part of individuals so as to qualify for the now higher paying minimum wage jobs, and on the part of employers who retain their higher wage employees so as to enhance their productivity to be more in line with their higher wage. The evidence suggests that minimum wages reduce training albeit the effects are often small and statistically insignificant.5

Higher minimum wages can also affect education decisions. To the extent that job opportunities are reduced for teens and youths, they may remain in school. In contrast, minimum wages can encourage youths to drop out of school and queue for the jobs that now pay a higher minimum wage. Such dropouts miss the high returns from education that they otherwise would receive if they continued their education. By definition of being a dropout, they also miss the substantial credential effects associated with completing key phases of education.6 Overall, the evidence on the effect of minimum wages on schooling tends to be mixed, although most studies tend to find that higher minimum wages encourage students to leave school to try to obtain the higher-wage minimum wage jobs.7

In the longer-run, minimum wages can also make it more difficult for youths to obtain jobs when they first enter the labour market and this can have permanent scarring effects as well as inhibiting youths from acquiring informal on-the-job training and experience.8 This is especially the case since many minimum wage jobs are stepping-stones, temporarily occupied by youths who will move up the occupational wage distribution and not be trapped in minimum wage jobs that leave them in a state of long-run poverty.9

Higher minimum wages can also foster higher prices and especially the prices of goods and services consumed by the poor. The limited evidence on this is somewhat mixed but tends to suggest that minimum wages lead to slight price increases, especially for goods and services consumed by the poor.10

Clearly, there are a variety of channels through which minimum wages can have an impact on poverty and the expected impacts often work in opposite directions. As such, the impact on poverty is ultimately an empirical proposition. Our focus in this study is on the first-order effects of whether the positive effect on raising the wages of some of the working poor is offset by any adverse employment effect. Longer run effects that likely occur through the mechanisms of education, training, prices, scarring effects, wage spillovers to those earning more than the minimum wage, and social or third-party spillover effects are beyond the scope of our analysis.

Previous Canadian studies have focused on specific provinces, primarily Ontario. For example, Shannon and Beach (1995) provide simulation evidence of the effect of Ontario’s 1991 minimum wage increase on poverty More recently, Mascella et al. (2009) also provide simulation evidence based on data for Ontario to analyse the impact on poverty of the 2004 minimum wage increase in Ontario. Sen et al. (2011) provided evidence for all of Canada, but their simulations used the 1993, 1995 and 1998 waves of the Survey of Consumer Finances. Our analysis uses much more current data and is not restricted to any specific provinces. This means that our analysis will be more relevant for policy makers and other stakeholders since they will reflect the current economic environment and climate.

Estimation Procedures and Data

Regression Estimates of Impact of Minimum Wages on Poverty

Our estimates of the impact of minimum wages on poverty utilize an OLS regression specification that is common in the literature:11
$$ POVRAT{E_{{it}}} = \alpha MINWAG{E_{{it}}} + {X_{{it}}}\beta + {T_t}\delta + {R_i}\varphi + {u_{{it}}} $$
(1)

Where POVRATEit is the natural log of the poverty rate (different measures defined subsequently) in province i at time t, MINWAGEitis the natural log of the minimum wage in province i at time t, Xit is a vector of controls for observable characteristics for province i at time t, Ttis a time trend variable entered as a quadratic, year and year squared; Riis a set of province dummies; and uit is a residual. The control variables in the X vector include: the unemployment rate for prime-age males 25-to-54 (with higher unemployment expected to increase poverty); the natural log of the average adult wage rate (with higher wages expected to reduce poverty); the percentage age 54–64 and the percentage age 16–24 (to control for age cohort effects). The sign of the coefficient in the minimum wage variable is indeterminate since a higher minimum wage can reduce or increase poverty depending largely on the relative importance of the wage gains versus employment losses.

There is no official definition of a poverty rate in Canada. However, two measures, are commonly used, and both will be utilized here.12 The first is the Low Income Cut-Off (LICO) published by Statistics Canada. It is the level of income below which a family unit is likely to spend 20 % or more of its income on the basic necessities of food, shelter and clothing compared to the average family. In essence, the proportion of income the average family in a community spends on food shelter and clothing is first calculated; then 20 percentage points is added to that proportion to get the LICO. These proportions vary by the size of the family and of the size of the community in which the family lives.

The second common measure is the Low Income Measure (LIM) calculated as 50 % of median family income, adjusted for family size and composition. The adjustment for family size and composition is done by giving the first member a weight of 1 and the second member a weight of 0.4. Additional family members are given a weight of 0.4 if age 16 or over and a weight of 0.3 if under age 16. Advantages of LIM are that it is straightforward to calculate and is commonly used in other countries and hence facilitates international comparisons. Furthermore, as Osberg (2000) points out, in the Canadian context the LICO and LIM are highly correlated.

Both LICO and LIM measures of poverty are available on a before- and after-tax basis. Both are informative. The before-tax indicators illustrate how minimum wages affect poverty rates based on income before it is altered by the tax system and hence comes closer to the effect of minimum wages on actual labour market outcomes as reflected by wages and employment. It is also the measure more commonly used in the literature. The after-tax indicator illustrates how minimum wages affect poverty rates based on income after it is altered by the tax system and hence comes closer to the actual needs of families. As indicated subsequently, our results are similar whether based on a before-tax or after-tax bases, likely reflecting the fact that the poor do not pay substantial taxes.

Our estimates of the effect of minimum wages on poverty are based on data from the confidential microdata files of the Survey of Labor and Income Dynamics (SLID) for the years 1997 to 2007. Over the 11 years of our data (1997–2007) there were 78 minimum wage increases across the 10 provinces. We use the SLID because it contains detailed information on both before- and after-tax measures of LICO and LIM. Statistics Canada calculates specific family-size adjusted LICO and LIM measures to estimate the proportion of Canadians in low income and places these measures directly in the SLID. We use these estimates to group individual level data into income-to-needs ratio categories (ratio of income to the specific poverty line) and aggregate this up to the level of the province and year. We then aggregate the data by calculating the proportion of people within each income-to-needs grouping below either the LICO or the LIM from SLID. All the control variables are also aggregated to the province and year level from the confidential microdata files of the SLID.

Simulation Estimates of Impact of Minimum Wages on Poverty

Our simulations are drawn from the March 2008 confidential microdata files of the Labour Force Survey (LFS). The LFS is similar to the CPS in the United States. We generate a poverty measure by following the procedure that Statistics Canada used to generate the Low Income Measure (LIM). We first take the household weekly earnings and adjust it by family size (i.e. divide it by the square root of the number of people in the household). We then define the LIM as 50 % of the median adjusted income within each province. We use this LIM and the corresponding population weights to calculate the number of people who are in poverty within each province.

As detailed below, our simulation methodology involves estimating the expected earnings gains from those who retain their job after a minimum wage increase and subtracting from this the expected earnings loss from those who lose their job because of a possible adverse employment effect. These expected net gains (or losses) are then incorporated into measures of the income distribution and then linked to indicators of poverty status, as detailed subsequently. The estimates of the job losses from minimum wage increases are based on external information from the literature on the effect of minimum wages on employment.

As in Sabia and Burkauser (2010), we first estimate the probability of job loss for each individual affected by the minimum wage increase in their province and time period. That probability is estimated as:
$$ {p_i} = \frac{{(MW - {w_i})}}{{{w_i}}}\left| e \right| $$
(2)

Where (MW – wi)/wi is the percent increase in wages for an individual with hourly wage wi and who now receives the new minimum wage MW in their jurisdiction and time period, and |e| is the elasticity of employment with respect to changes in the minimum wage, obtained from external estimates in the literature, and applied to each individual in this analysis. The product of the percent increase in wages and the minimum wage elasticity yields the probability of a job loss pi, and (1- pi) gives the probability of retaining one’s job after receiving a minimum wage increase.

The expected net effect on earnings, ΔEi, for workers affected by a minimum wage increase is then calculated as the probability of retaining one’s job, (1- pi), times the expected earnings gain from the minimum wage increase (MW − wi)Hi, less the probability of losing one’s job, pi, times the earnings in the lost job, wiH, where H measures hours worked.13 That is,
$$ \Delta EAR{N_i} = (1 - {p_i})(MW - {w_i}){H_i} - {p_i}\left( {{w_i}{H_i}} \right) $$
(3)

We use minimum wage elasticities of −0.3 and −0.6 that reflect the range of elasticities typically found in the Canadian literature.14 We also use an elasticity of zero to illustrate the maximum potential for minimum wages to reduce poverty because the wage gains would not be offset in any way by employment losses.15

Our simulations are based on data from the March 2008 labour force survey (LFS). The labour force survey contains information on the household income and household size which are essential in the calculation of a LIM. We utilize this information to calculate a low-income threshold and adjust this threshold by dividing by the square root of the household size which is a close approximation to the household adjustment made by Statistics Canada.16 In our simulations, we take the hourly weekly earnings and the number of hours worked for the individual during the reference week and the number of people who made less than our assumed minimum wage increase in each province and simulate employment losses and net gains.

We assume no change in hours worked which means that our adverse employment effects are lower bound estimates since they don’t include possible adjustments in hours worked. Sabia and Burkhauser (2010, p.598) justify this assumption, which they also follow, on the basis that “existing estimates in the literature tend to point to either no effects or only small negative effects.” We also assume that there are no wage spillovers to workers earning more than the minimum wage. Sabia and Burkhauser (2010, p. 598) justify this assumption on the bases that “we find no evidence that minimum wage increases have important spillover effects.” Both of these assumptions imply that our estimates are conservative or lower-bound estimates of any negative impact of minimum wages on poverty.

Results

Estimates of the Impact of Minimum Wages on Poverty

Table 1 presents our estimates of the poverty rate Eq. 1 based on the before-tax LICO.17 As in Sabia and Burkhauser (2010) three indicators of poverty status are used: families whose income falls below the LICO (column 1), below 1.25 of the LICO, and below 1.5 of the LICO. As pointed out by Neumark and Wascher (2008, p. 314, fn 17) broadening the poverty definition to include the “near poor” (families whose income is in-between the poverty line and 1.5 times the poverty line) serves two purposes: it expands the target group to include families that have quite low income even though they are not poor; and it allows for a higher threshold for defining poverty to offset any concern that the existing definition is too stringent.
Table 1

Effect of Minimum Wages and Controls on Log Poverty Rates (Before-Tax LICOs) (standard errors in parenthesis)

 

Income < 1.0 LICO

Income < 1.25 LICO

Income < 1.5 LICO

(1)

(2)

(3)

Log (MinWage)

−0.169

−0.12

−0.037

[0.247]

[0.180]

[0.119]

Prime age Male Unemployment Rate

6.153***

5.430***

4.425***

[0.868]

[0.694]

[0.507]

Log (Average Adult Wage Rate)

−2.416***

−2.039***

−2.084***

[0.548]

[0.366]

[0.255]

Percentage aged 54-64

−0.613

−0.148

−1.811

[2.415]

[2.086]

[1.809]

Percentage Aged 16–24

3.963**

1.963

0.818

[1.987]

[1.698]

[1.179]

Province Effects

Y

Y

Y

Year

Y

Y

Y

Year Squared

Y

Y

Y

Mean of Dependent Variable

0.156

0.217

0.285

N

110

110

110

Data are aggregated by year and province from the 1997–2007 Survey of Labour and Income Dynamics

As indicated in column 1 of Table 1, minimum wage increases have no significant effect on poverty. This is also the case if the potential target groups are expanded to include the “near-poor” whose income falls below 1.25 of the poverty line (column 2) and 1.5 of the poverty line (column 3). As expected, the effect on reducing poverty becomes smaller as the poverty group is expanded to include various groups of the “near poor.” However, whether the concept of poverty is a narrow definition or broader one to include the near poor, minimum wages have no significant effect on poverty rates.

As indicated in Table 2, the result of no impact of minimum wages on poverty is robust across various of specifications including: the LIM as opposed to the LICO as a measure of poverty; an after-tax as opposed to a before-tax basis; based on all families or only families with workers; and a tighter definition of poverty or broader ones to include the “near-poor.” In none of the 24 cases do minimum wages have a significant effect on poverty. While minimum wages do not reduce poverty, they also do not increase the poverty rate. (Full regression results available on request).
Table 2

Effect of Log Minimum Wages on Log Poverty Rates (Various Specifications) (standard errors in parenthesis)

All Persons

Workers Only

Income < 1.0 Poverty

Income < 1.25 Poverty

Income < 1.5 Poverty

Income < 1.0 Poverty

Income < 1.25 Poverty

Income < 1.5 Poverty

(1)

(2)

(3)

(1)

(2)

(3)

LICO Before-Tax

−0.169

−0.120

−0.037

−0.288

−0.281

−0.147

[0.247]

[0.180]

[0.119]

[0.352]

[0.252]

[0.181]

LICO After-Tax

−0.298

−0.244

−0.123

−0.506

−0.450

−0.247

[0.271]

[0.267]

[0.184]

[0.423]

[0.351]

[0.249]

LIM Before-Tax

-.097

-.040

0.038

−0.138

−0.228

-.055

[0.248]

[0.174]

[0.138]

[0.349]

[0.232]

[0.204]

LIM After-Tax

−0.400

−0.151

−0.037]

−0.583

−0.244

-.162

[0.259]

[0.236]

[0.178]

[0.450]

[0.295

[.238]

Data are aggregated by year and province from the 1997 – 2007 Survey of Labour and Income Dynamics

Results for the Simulations of the Impact of Minimum Wages on Poverty

As indicated, our simulation methodology involves estimating the expected earnings gains from those who retain their job after a minimum wage increase and subtracting from this the expected earnings loss from those who lose their job because of a possible adverse employment effect. These expected net gains (or losses) are then incorporated into measures of the income distribution and then linked to indicators of poverty status.

Table 3 provides our simulation estimates of the employment losses and how they are distributed across the different ratios of family income relative to the LIM poverty line. (indicated in the left-hand column). Family incomes relative to the poverty line of less than 1 represent those in poverty, those with a ratio of between 1 and 1.24 and between 1.25 and 1.5 are “near poor,” and those with ratios above 1.5 have incomes that put them outside of poverty and outside of being “near poor.” As the column headings indicate, we use minimum wage elasticities ranging from 0.0 (which gives the maximum potential for minimum wages to reduce poverty because the wage gains would not be offset in any way by employment losses) to −0.3 and −0.6, with the later two reflecting the range of elasticities typically found in the Canadian literature. We do simulations separately for assumed minimum wage increases of $0.50 and $1.00 which covers the range of preferred hypothetical minimum wage increases used in Campolieti et al. (2005a, b, p. 85) on the grounds that “they encompass low-wage workers who are most likely to be similar in terms of employment instability to similar low-wage workers in the jurisdictions that experienced minimum wage increases.” They also encompass the minimum wage increases of $0.75 that have occurred in Canada’s largest province – Ontario – for each year since 2007.
Table 3

Simulated Employment Losses from Minimum Wage Increases

Ratio Income/ Poverty Line

% Potentially Affected

Number Potentially Affected

Employment Losses

% of Job Losses

Elasticity −0.3

Elasticity −0.6

 

(1)

(2)

(3)

(4)

(5)

Assuming Minimum Wage Increase of $0.50

<1 i.e., In Poverty

25.6 %

224,326

8,634

17,268

32.3 %

1–1.24 Near Poor

6.7 %

58,509

2,241

4,482

8.4 %

1.25–1.49 Near Poor

7.0 %

60,971

1,638

3,276

6.1 %

1.5 to 1.99

12.2 %

107,362

3,513

7,026

13.1 %

2 to 2.99

18.1 %

158,810

4,615

9,231

17.3 %

3 or Above

30.4 %

266,771

6,082

12,164

22.8 %

Total

100 %

876,749

26,724

53,448

100 %

Assuming Minimum Wage Increase of $1.00

<1 i.e., In Poverty

24.7 %

317,790

13,431

26,863

29.7 %

1–1.24 Near Poor

7.5 %

96,232

3,545

7,090

7.8 %

1.25–1.49 Near Poor

6.9 %

89,029

2,888

5,776

6.4 %

1.5 to 1.99

11.9 %

152,541

5,791

11,581

12.8 %

2 to 2.99

18.8 %

241,671

7,963

15,926

17.6 %

3 or Above

30.2 %

387,737

11,569

23,137

25.6 %

Total

100 %

1,285,000

45,186

90,373

100 %

Data are taken from the March 2008 Labour Force Survey

The first column in Table 3, gives the percent of workers “at-risk” (termed minimum-wage workers) of being affected by a minimum wage increase in that their wage falls between the old minimum wage in their jurisdiction and the new assumed minimum wage that is either $0.50 or $1.00 higher than their old minimum wage. As indicated in column 1, approximately 26 % of workers who would be potentially affected by a typical minimum wage increase of $0.50 fall below the poverty line. A further, 13.7 % (6.7 % + 7 %) are “near poor” so that a total of approximately 40 % of such minimum wage workers are poor or near poor. Almost one-third (30.4 %) are in families that earn three times or more the poverty line appropriate for their family.

The fact that about one-quarter of workers who are potentially affected by a minimum wage increase is poor suggests that minimum wages could potentially alleviate their poverty by raising their wage. However, this would be mitigated and possibly even more than offset by the fact that many are likely to experience an adverse employment effect and hence lose their wages.

As indicated in the last column of Table 3, based on the typical minimum wage elasticities of −0.3 to −0.6 found in the Canadian data, job losses are distributed disproportionately in the hands of the poor (32.3 %) and the near-poor (a further 14.5 %) so that almost half of the job losses occur for the poor and near-poor. This reflects the fact, of course, that they tend to be low-wage workers.

About one-quarter of workers who are likely to have their wages increase because of minimum wages are poor, but about one-third of the job losses also fall on the poor who will lose their wage. The wage gains to the 224,326 workers who are poor and whose wage should increase because of the minimum wage (column 2, row 1 of Table 3) will experience a small wage increase because of the minimum wage, but the wage losses for the 17,268 workers who are poor and who will lose their jobs (column 4 row 1) will be substantial since they will have lost their job. In essence, the wage gains are dispersed while the job and wage loses are concentrated in the hand of a few. In that vein, the impacts on the poor are horizontally inequitable in that they involve the “unequal treatment of equals” – some poor persons gain slightly but a few lose substantially.

If the minimum wage were to increase by $1.00 (bottom panel) the impacts are larger in that more persons including the poor would receive the wage increase but more would also experience job loses. The distributional pattern, however, is similar in that about 25 % of those whose wage will increase slightly are poor, while about 30 % of the job losses fall on the poor.

Table 4 presents the expected net effect on earnings for workers potentially affected by a minimum wage increase of $0.50 (top panel) and $ 1.00 (bottom panel) calculated as in Eq. 4 as the expected earnings gain from the minimum wage workers who retain their job less the earnings loss from those who lose their job. For the top panel, based on a minimum wage increase of $0.50, the net gains are generally positive suggesting that the earnings gains to minimum wage workers who retain their jobs exceed the earnings losses to those who lose their jobs. The net gains are highest when there are no assumed adverse employment effects (column 1 when the minimum wage elasticity is 0), the net gains are cut approximately in half when the elasticity is −0.3, and they are reduced to a very small amount when the elasticity is −0.6. Although not shown in the table, the net gains become negative when the elasticity is −0.7 suggesting that the “break-even” elasticity is between −0.6 and −0.7.
Table 4

Simulated Net Earnings Gain ($ million) from Minimum Wage Increases

Ratio Income/Poverty Line

Elasticity −0.0

Elasticity −0.3

Elasticity −0.6

Earnings Gain

Distribution of Gain

Earnings Gain

Distribution of Gain

Earnings Gain

Distribution of Gain

 

(1)

(2)

(3)

(4)

(5)

(6)

Assuming Minimum Wage Increase of $0.50

<1 i.e., In Poverty

200

29.7 %

110

29.6 %

20.7

30.2 %

1–1.24 Near Poor

79.9

11.9 %

38.3

10.3 %

−3.3

−4.8 %

1.25–1.49 Near Poor

52.0

7.7 %

28.8

7.8 %

5.5

8.0 %

1.5 to 1.99

91.6

13.6 %

50.8

13.7 %

9.9

14.4 %

2 to 2.99

11.4

16.9 %

63.2

17.0 %

12.1

17.6 %

3 or Above

13.6

20.2 %

80.0

21.6 %

23.7

34.5 %

Total

674.0

100 %

371.0

100 %

68.6

100 %

Assuming Minimum Wage Increase of $1.00

<1 i.e., In Poverty

313

27.5 %

177

27.2 %

40.2

24.5 %

1–1.24 Near Poor

121

10.6 %

61.5

9.4 %

1.8

1.1 %

1.25–1.49 Near Poor

91.3

8.0 %

52.8

8.1 %

14.2

8.7 %

1.5 to 1.99

156

13.7 %

89.6

13.8 %

23.5

14.3 %

2 to 2.99

196

17.2 %

113

17.4 %

29.6

18.0 %

3 or Above

259

22.7 %

157

24.1 %

55.1

33.6 %

Total

1140

100 %

651

100 %

164

100 %

Data are taken from the March 2008 Labour Force Survey

The separate entries for the different rows illustrate how the net gains are distributed across the different income-to-needs ratio (ratio of income to the poverty line). Based on the mid-range elasticity of −0.3 (column 4), only about 30 % of the net earnings gain from minimum wage increases goes to the poor (first row). Conversely, slightly over 70 % of the earnings gains “spill over” into the hands of the non-poor, although about 18 % are near-poor (rows 2 and 3). This highlights that minimum wages are poorly targeted towards the poor. As indicated by Card and Krueger (1995, p. 285): “The minimum wage is evidently a ‘blunt instrument’ for redistributing income to the poorest families.” Nevertheless, the target is not completely missed and when it is missed, it is often close in that it often helps the near-poor.

If the minimum wage were to increase by $1.00 (bottom panel), larger numbers receive wage gains but larger numbers also experience job losses and hence a complete loss of their wage. The distributional pattern across the different income-to-needs ratios is fairly similar as for the $0.50 minimum wage increase. That is, slightly less than 30 % of the net earnings gain from minimum wage increases goes to the poor while slightly over 70 % of the net earnings gains “spill over” into the hands of the non-poor.

Summary and Concluding Observations

Based on regressing poverty rates for various provinces over time on minimum wage increases, we find no significant impact of minimum wages on poverty across 24 different specifications: the LIM as opposed to the LICO as a measure of poverty; an after-tax as opposed to a before-tax basis; based on all families or only families with workers; and a tighter definition of poverty or broader ones to include the “near-poor.”

We also provide estimates based on simulations from estimating the expected earnings gains from those who retain their job after a minimum wage increase and subtracting from this the expected earnings loss from those who lose their job because of a possible adverse employment effect. These expected net gains (or losses) are then incorporated into measures of the income distribution and then linked to indicators of poverty status. Our simulation results are based on data from the March 2008 labour force survey (LFS). They indicate that only about 30 % of the net earnings gain from minimum wage increases goes to the poor while about 70 % of the net earnings gains “spill over” into the hands of the non-poor. Furthermore, the job losses are disproportionately concentrated on the poor.

Both our regression estimates and simulation estimates highlight that, political rhetoric notwithstanding, minimum wages are poorly targeted as an anti-poverty device and are at best an exceedingly blunt instrument for dealing with poverty. The bad news is that they do not alleviate poverty; the good news is that they do not exacerbate poverty, at least in the short-run.

Assisting working poor persons is a legitimate social objective. If society deems it appropriate that workers should be paid a certain minimum even if their productivity is below that minimum, however, it would seem appropriate for that cost to be borne by society in general rather than by a small subset of employers. This could be accomplished, for example, through policies such as the Earned Income Tax Credit (EITC) in the US (Burkhauser et al. 1996; Burkhauser and Sabia 2007; Sabia and Burkhauser 2008, 2010) or its equivalent of the Working Income Tax Benefit (WITB) in Canada. Such policies basically involve wage subsidies targeted towards the working poor, administered through the tax system so they can be based on family need, with the subsidy reduced as income increases so as to minimize spillover benefits to persons with higher income. Since the higher wages are paid by the state rather than employers, they do not have an adverse employment effects as can minimum wages. As pointed out by Sabia and Burkhauser (2010, p. 612) minimum wage increases are not likely to reduce poverty because most workers who are affected are not poor, many poor workers already earn more than the minimum wage, and minimum wage increases are likely to cause adverse employment effects for the working poor. Earnings subsidies through such programs like the EITC or WITB are much better targeted to the working poor without benefits spilling over substantially to the non-poor, and without adverse employment effects.

Footnotes
1

These sources are cited in Neumark and Wascher (2008, p. 141). Burkauser et al. (1996), Burkhauser and Finegan (1989), Card and Krueger (1995) and Sabia and Burkhauser (2008) also document and discuss the history and political support for minimum wages as an anti-poverty device in the US.

 
2

This weakening over time of the already weak relationship between low wages and poverty is documented and discussed, for example, in Burkhauser and Finegan (1989), Burkhauser et al. (1996), Burkhauser and Sabia (2004, 2007).

 
3

The poor targeting of minimum wages towards those in poverty in the US is documented and discussed, for example, in the studies mentioned in the previous footnote as well as in Freeman (1996), Gramlich (1976), Neumark and Wascher (2002, 2008), Smith and Vavrichek (1992) and Veeder and Gallaway (2001), and for Canada in Benjamin (1996, 2001), Campolieti and Gunderson (2010), Gunderson (2005), Mascella et al. (2009) and Sen et al. (2011).

 
4

This may be one of the reasons for the generally low level of political opposition to minimum wages. The adverse effects are likely to occur in the form of not getting a job or working fewer hours, and this is not likely to attract political attention as much as dismissals or layoffs.

 
5

Negative effects of minimum wages on training for the US are found in Grossberg and Sicilican (1999), Hashimoto (1982), Leighton and Mincer (1981) and Neumark and Wascher (2001), although Acemoglu and Pischke (2003) find no effect. For the UK, Arulampalam et al. (2004) also find no effect. For Canada, Baker (2005) generally finds a negative effect of minimum wages on training, but he states that no firm conclusion can be made because the results are not robust.

 
6

Summaries of the literature documenting high returns to education, especially for those who are otherwise likely to drop out and miss the substantial credential effects associated with completing key phases of education, are provided in Card (1999) and Gunderson and Oreopoulos (2010).

 
7

US evidence on the negative effects of minimum wages on education is found in Neumark and Wascher (1995a, b, 1996) and for some race-sex groups in Cunningham (1981) and for teens in low-income families in Ehrenberg and Marcus (1980, 1982) although that study found that it increased the education for white teenagers from high-income families. Card (1992), however, find no effect on enrolment and Mattila (1981) finds positive effects. For Canada, Campolieti et al. (2006) find no effect on enrolment.

 
8

Evidence on the more permanent scarring effects from initial bouts of unemployment is provided, for example, in Beaudry and Green (2000) and McDonald and Worswick (1999) and references cited therein.

 
9

US evidence that most minimum wage jobs are temporary stepping-stones is given, for example, in Carrington and Fallick (2001), Long (1999), Schiller (1994) and Smith and Vavrichek (1992). For Canada, Battle (2003) indicates that more than half of all minimum wage workers had been in their current job for less than one year, and only about 1 % of persons had been in their job for more than five years.

 
10

Evidence that minimum wage increases tend to lead to price increases is found in Card and Kruger (1995), MaCurdy and McIntyre (2001) and Wessels (1980) but not in Katz and Krueger (1992).

 
11

Studies using the method of regressing measures of poverty on minimum wages include Addison and Blackburn (1999), Burkhauser and Sabia (2007), Card and Krueger (1995), Neumark and Wascher (2002), Sabia and Burkhauser (2008, 2010) and Vedder and Gallaway (2001, 2002) for the US, and Sen et al. (2011) for Canada.

 
12

These poverty measures are discussed in Osberg (2000). More details of the calculation of LICOs are given in http://www.statcan.gc.ca/pub/75f0002m/2009002/s2-eng.htm.

 
13

Since our focus is on is on the impact of minimum wages on poverty holding constant other factors that can affect poverty, we do not incorporate information on transfer payments (e.g., unemployment insurance or welfare payments) that may be received by those who lose their job.

 
14

The recent Canadian evidence based on different data sets and methodologies includes Baker (2005), Baker et al. (1999), Campolieti et al. (2005a, b), Campolieti, et al. (2006) and Sen et al. (2011).

 
15

Simulation studies that focus on the net effect of minimum wage increases on poverty assuming no reduction in employment and hours include Burkhauser and Finegan (1989), Burkhauser et al. (1996), and Burkhauser and Sabia (2007) for the US, and Mascella et al. (2009) for Canada. Simulation studies that incorporate assumptions about employment and hours adjustments include Sabia and Burkhauser (2008, 2010) and Mincy (1990) for the US, Shannon and Beach (1995) for Canada and Leigh (2007) for Australia. Only Mincy (1990) reports reductions in poverty from minimum wages.

 
16

See http://www.statcan.gc.ca/pub/75f0002m/2009002/s3-eng.htm for the household adjustment for the LIM.

 
17

Results for after-tax measures are given subsequently in Table 2 for the key minimum wage impacts. Results for the control variables in Table 1 for after-tax measures (available on request) are very similar to those based on the before-tax measures.

 

Copyright information

© Springer Science+Business Media, LLC 2012

Authors and Affiliations

  • Michele Campolieti
    • 1
    • 2
  • Morley Gunderson
    • 3
  • Byron Lee
    • 4
  1. 1.Department of Management (Scarborough Campus)University of TorontoTorontoCanada
  2. 2.Centre for Industrial Relations and Human Resources at the University of TorontoTorontoCanada
  3. 3.Centre for Industrial Relations and Human Resources and Department of EconomicsUniversity of TorontoTorontoCanada
  4. 4.Renmin Business School, Department of Organization and Human ResourcesRenmin University of ChinaHaidian District, BeijingChina

Personalised recommendations