A. Introduction

1. Over the past 25 years, the unemployment rate in Canada has ratcheted up over successive business cycles. This paper looks at this development from various angles and examines the roles played by modifications to the unemployment insurance system, changes in unionization, demographic effects, and changes in the cost of labor in accounting for the rise in unemployment. The paper concludes with a discussion of the effects of changes in payroll taxes on unemployment.

2. Unemployment rates in Canada have risen across all demographic groups, industries and regions, although young and less-educated workers and workers in agriculture and primary industries have been most severely affected. In terms of aggregate factors, it appears that the rise in payroll taxes may have hindered employment opportunities. The empirical estimates suggest that the effects of an increase in payroll taxes on employment and unemployment are largely offset over a six-year period. Thus, successive increases in payroll taxes during the 1980s may have had a sustained impact on unemployment.

B. Labor Market Developments

3. The unemployment rate in Canada has ratcheted up from 6 percent in the early 1970s to around 10 percent currently. The rise in the unemployment rate in the 1970s and in the early 1980s was accompanied by a sharp increase in the participation rate, as women entered the labor force in large numbers (Chart 1). In the late 1980s, the participation rate declined initially reflecting cyclical influences and it has continued to fall despite a modest recovery in economic activity. The slow rise in the employment-population ratio since the late 1970s partly reflects the substitution of capital for labor. Since the late 1970s, there has been an upward trend in the capital-output ratio in Canada (controlling for cyclical effects) reflecting substitution of capital for labor to take advantage of the cost difference between both inputs (Chart 2). Since the early 1990s, the behavior of the employment-population ratio has been influenced by cyclical developments, falling during the 1990–91 recession and rising with the rebound in activity in 1994. The ratio has been essentially flat since then.

View Full Size

CANADA: COMPONENTS OF THE UNEMPLOYMENT RATE

Citation: IMF Staff Country Reports 1997, 020; 10.5089/9781451806847.002.A002

Sources: Statistics Canada; and Fund staff estimates.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

CANADA: THE CAPITAL-OUTPUT RATIO AND RELATIVE COST OF CAPITAL

Citation: IMF Staff Country Reports 1997, 020; 10.5089/9781451806847.002.A002

Sources: Statistics Canada; Data Resources Inc.; and Fund staff estimates.1/ Real user cost of capital minus real wage.

• Download Figure
• Download figure as PowerPoint slide

4. Following the decline in aggregate employment during the early 1990s, private sector employment has picked up. However, the aggregate employment increase has remained weak on account of more recent restructuring in the public sector. During 1993–95 and during the first 11 months of 1996, employment in the private sector grew by over 2 percent (per annum). In contrast, employment in the public sector declined by an annual average of 1 percent over the 1993–95 period and declined further by 2 percent during the first 11 months of 1996.

5. The duration of unemployment also provides a measure of economic restructuring because it indicates changes in the number of core unemployed as opposed to workers who are temporarily unemployed. The mean duration of unemployment has risen from roughly 6 weeks during the 1970s to 24 weeks in the early 1990s (see Chart 1). In the early stages of a recovery, the mean duration of unemployment often declines as discouraged workers re-enter the labor force and commence their job search. However, this has not been the experience in Canada over the past two business cycles. Uncertainty about employment prospects may be reflected in the recent decline in the participation rate, particularly among youth (Chart 3). Poor employment opportunities and the recognition that education qualifications have a high financial payoff have encouraged youth to remain in school.

View Full Size

CANADA: PARTICIPATION AND UNEMPLOYMENT RATES BY AGE AND SEX ^1/

Citation: IMF Staff Country Reports 1997, 020; 10.5089/9781451806847.002.A002

Source: Statistics Canada.^1/ Shaded areas indicate recessions.

• Download Figure
• Download figure as PowerPoint slide

6. Although unemployment rates for all skill levels have risen since 1990, they have risen significantly less for individuals in the highest skill categories. As is the case in most industrialized countries, unemployment rates are inversely related to years of schooling, which is often used as a measure of skills (see tabulation below). In particular, the unemployment rate for workers with less than eight years of schooling is about three times that of workers with a college degree. Moreover, from 1990 to 1995, the average unemployment rate of workers who did not graduate from high school increased by 2½ percentage points, compared with a 1 percentage point increase in the unemployment rate of workers with a college degree.

Unemployment Rates by Education Levels

Unemployment Rates by Education Levels

Education Level	1990	1991	1992	1993	1994	1995
Less than eight years	12.6	15.4	16.1	16.7	15.7	15.0
Some secondary	12.2	15.4	17.6	17.0	16.4	15.2
Secondary graduate	7.7	10.3	10.9	11.4	10.0	9.5
Some post-secondary	8.0	10.2	11.6	11.6	10.8	10.1
Post-secondary	6.3	8.2	9.4	9.5	8.9	7.9
University degree	3.7	4.9	5.5	5.7	5.4	4.9
Total	8.1	10.4	11.3	11.2	10.4	9.5

View Table

Unemployment Rates by Education Levels

Education Level	1990	1991	1992	1993	1994	1995
Less than eight years	12.6	15.4	16.1	16.7	15.7	15.0
Some secondary	12.2	15.4	17.6	17.0	16.4	15.2
Secondary graduate	7.7	10.3	10.9	11.4	10.0	9.5
Some post-secondary	8.0	10.2	11.6	11.6	10.8	10.1
Post-secondary	6.3	8.2	9.4	9.5	8.9	7.9
University degree	3.7	4.9	5.5	5.7	5.4	4.9
Total	8.1	10.4	11.3	11.2	10.4	9.5

7. Over the past 25 years, there has been a secular decline in the employment shares of goods producing industries, particularly in the manufacturing sector. In 1970, the manufacturing sector accounted for 22 percent of total employment, but this ratio had fallen to 15 percent by 1995 (Chart 4). In contrast, the employment shares in service industries have risen over time, with the most noticeable increase from 26 percent to 38 percent of total employment in community business and services. The unemployment rates in the manufacturing and services sectors have essentially remained trendless over the 1976–95 period, whereas the unemployment rates in agriculture and primary industries have experienced a secular rise of 2 percentage points.² The unemployment rate in the construction industry has also risen dramatically since 1989; however, this rise can be attributed to cyclical weakness in the industry.

View Full Size

CANADA: EMPLOYMENT AND UNEMPLOYMENT BY INDUSTRY

Citation: IMF Staff Country Reports 1997, 020; 10.5089/9781451806847.002.A002

Source: Statistics Canada.

• Download Figure
• Download figure as PowerPoint slide

8. The unemployment rates in the Atlantic provinces have remained consistently above the Canadian average over the past 25 years, whereas in all the other regions they have remained consistently below the average (Chart 5). However, with the exception of Newfoundland, there has been little change in relative unemployment rates over time in Canada. It appears that the introduction of regional extended unemployment benefits in the late 1970s has not appreciably affected the aggregate unemployment problem in Canada.³ Recent work by Corak and Jones (1995) argues that a major shift in the proportion of unemployment insurance recipients receiving regional extended unemployment benefits did not occur in the 1980s and that the number of beneficiaries of regional extended unemployment benefits is not more persistent than the number of other unemployment insurance beneficiaries.

View Full Size

CANADA: PROVINCIAL UNEMPLOYMENT RATES

Citation: IMF Staff Country Reports 1997, 020; 10.5089/9781451806847.002.A002

Source: Statistics Canada.

• Download Figure
• Download figure as PowerPoint slide

9. It is possible, however, for the regional extended benefits system to affect economic behavior by reducing the incentives for inter-provincial migration. If workers know that their unemployment experience is cushioned by a longer duration of benefits, they are less likely to search for employment in other provinces. In Canada, data on inflows and outflows from each province are available for the working-age population. The data indicates that British Columbia and Alberta have been net recipients over the past 25 years, with average annual net migration rates of 0.8 and 0.4 percent, respectively. These provinces have gained potential labor force participants at the expense of Newfoundland, Québec, Manitoba, and Saskatchewan.

10. Changes to the regional extended benefits system in 1977 created considerable disincentives to search for employment across provinces. To determine whether a structural break in the net migration rate across provinces ocurred following this change, a relationship for net migration using the panel of migration data for the provinces was estimated allowing for fixed effects for each province. Standard migration models assume that the decision to migrate depends on the relative return to work in the current and prospective region and on the likelihood of obtaining employment in both regions. The estimated equation included the relative real wage and the unemployment rate as explanatory variables and was used to test whether the responsiveness of migration to changes in these variables declined following changes to the regional extended unemployment benefits system in 1977.

11. The results in Table 1 indicate a significant positive relationship between the relative real wage and the net migration rate and a significant negative relationship between the relative unemployment rate and the net migration rate. Moreover, there is a significant decline in the responsiveness of migration to changes in relative wages and relative unemployment rates after 1977, suggesting that the changes made to the benefit criterion of the regional extended unemployment benefits system have adversely affected labor mobility.

Table 1.

Canada: Determinants of Net Migration Across Canadian Provinces ^1/

^1/An asterisk denotes a variable or test statistic that is significant at the 10 percent level.

Table 1.

Canada: Determinants of Net Migration Across Canadian Provinces ^1/

Variables with the same coefficient across provinces
Relative wage	0.04	*
Relative wage (1977–94)	−0.026	*
Relative Unemployment rate	−0.20	*
Relative Unemployment rate (1977–94)	0.07	*
Net migration (−1)	0.84	*
Net migration (−2)	−0.20	*
Youth population share	0.26	*
Provincial fixed effects
Newfoundland	−0.23	*
Nova Scotia	−0.04
New Brunswick	−0.08
Quebec	−0.13	*
Ontario	−0.02
Manitoba	−0.28	*
Saskatchewan	−0.27	*
Alberta	0.08
British Columbia	0.19	*

^1/An asterisk denotes a variable or test statistic that is significant at the 10 percent level.

View Table

Table 1.

Canada: Determinants of Net Migration Across Canadian Provinces ^1/

Variables with the same coefficient across provinces
Relative wage	0.04	*
Relative wage (1977–94)	−0.026	*
Relative Unemployment rate	−0.20	*
Relative Unemployment rate (1977–94)	0.07	*
Net migration (−1)	0.84	*
Net migration (−2)	−0.20	*
Youth population share	0.26	*
Provincial fixed effects
Newfoundland	−0.23	*
Nova Scotia	−0.04
New Brunswick	−0.08
Quebec	−0.13	*
Ontario	−0.02
Manitoba	−0.28	*
Saskatchewan	−0.27	*
Alberta	0.08
British Columbia	0.19	*

^1/An asterisk denotes a variable or test statistic that is significant at the 10 percent level.

12. The provision of regional extended unemployment benefits after 1977 was made dependent on the unemployment rate in a particular province relative to a threshold level of the unemployment rate. Therefore, during cyclical downturns, it is possible for the benefit system to affect aggregate migration without affecting net migration by reducing both inflows and outflows. To determine this disincentive effect of the system, it is necessary to aggregate inflows and outflows. Gross migration rates were calculated for each of the Canadian provinces by dividing each inflow and outflow by the donor populations and aggregating both ratios (Chart 6). Throughout the 1972–95 period, the gross migration propensities of the Atlantic Provinces have remained higher than the corresponding propensities for British Columbia, Ontario, and Québec. However, there has been a systematic decline in the gross migration rates of all the provinces over time which may be related to the aging of the population.

View Full Size

CANADA: POPULATION MIGRATION AND UNEMPLOYMENT

Citation: IMF Staff Country Reports 1997, 020; 10.5089/9781451806847.002.A002

Source: Statistics Canada.

• Download Figure
• Download figure as PowerPoint slide

13. A comparison of the population weighted average gross migration rate with the aggregate unemployment rate in Chart 6 suggests that the gross migration behavior across provinces is pro-cyclical, contrary to the usual theoretical assumption that the relationship is counter-cyclical. Therefore, workers primarily move across provinces when the jobs climate is healthy.

14. To test for a structural break in the gross migration rate across provinces following the changes the regional extended unemployment benefits system, the gross migration rate was regressed on a segmented trend with a break in 1977, the aggregate unemployment rate, and province specific constants. The results in Table 2 indicate a significant negative relationship between the unemployment rate and the gross migration rate and a significant declining trend in the migration propensities of all the provinces since 1977. In particular, the coefficient on the lagged unemployment rate indicates that a 5 percentage point rise in the unemployment rate leads to a 0.1 percent reduction in the gross migration rate. The significant declining trend coefficient since 1977 and the insignificant coefficient prior to this date indicate that a structural change took place at this time. Moreover, by 1995 this structural change had caused a 0.2 percent decline in the gross migration rate.

Table 2.

Canada: Determinants of Gross Migration Across Canadian Provinces ^1/

^1/An asterisk denotes a variable or test statistic that is significant at the 10 percent level.

Table 2.

Canada: Determinants of Gross Migration Across Canadian Provinces ^1/

Variables with the same coefficient across provinces
1972–76 trend	0.003
1977–94 trend	−0.009	*
Unemployment rate	−0.02	*
Gross migration (−1)	0.68	*
Provincial fixed effects
Newfoundland	1.05
Nova Scotia	1.07	*
New Brunswick	1.00	*
Quebec	0.53
Ontario	0.75	*
Manitoba	1.10	*
Saskatchewan	1.18	*
Alberta	1.26	*
British Columbia	0.90	*

^1/An asterisk denotes a variable or test statistic that is significant at the 10 percent level.

View Table

Table 2.

Canada: Determinants of Gross Migration Across Canadian Provinces ^1/

Variables with the same coefficient across provinces
1972–76 trend	0.003
1977–94 trend	−0.009	*
Unemployment rate	−0.02	*
Gross migration (−1)	0.68	*
Provincial fixed effects
Newfoundland	1.05
Nova Scotia	1.07	*
New Brunswick	1.00	*
Quebec	0.53
Ontario	0.75	*
Manitoba	1.10	*
Saskatchewan	1.18	*
Alberta	1.26	*
British Columbia	0.90	*

^1/An asterisk denotes a variable or test statistic that is significant at the 10 percent level.

C. Factors Underlying the Rise in the Aggregate Unemployment Rate

15. The most common explanations for the rise in the aggregate unemployment rate over the past 25 years revolve around modifications to the unemployment insurance system, changes in unionization, demographic effects, and changes in the cost of labor (including changes in the minimum wage and payroll taxes).

16. The generosity of unemployment insurance affects the labor market in a number of ways. On the supply side, longer entitlement benefit periods are thought to increase the duration of unemployment by reducing the intensity with which the unemployed search for work. On the demand side, the absence in Canada of setting unemployment insurance charges for firms based on their use of the unemployment insurance system encourages them to rely on temporary layoffs rather than on changes in wages or hours worked in response to shocks. However, although the theoretical rationale for the existence of disincentive effects from a generous insurance system is clear, it has been difficult to find strong supporting empirical results.

17. Sargent (1995) has recently developed a new index of unemployment insurance generosity.³ Chart 7 plots the unemployment insurance (UI) index and the aggregate unemployment rate for Canada. The Chart clearly shows the sharp increase in unemployment insurance generosity in 1971–72 and the endogenous component of unemployment insurance generosity since the late 1970s, when eligibility requirements and duration were made dependent on the unemployment in a particular province relative to a fixed unemployment rate (the regional extended unemployment benefits system). Since the early 1990s the UI index has declined in response to successive reforms by the government and an improving economic climate; its current level is now comparable to the level prior to the major changes to the system in 1970.

View Full Size

CANADA: UNEMPLOYMENT INSURANCE INDEX AND THE UNEMPLOYMENT RATE

Citation: IMF Staff Country Reports 1997, 020; 10.5089/9781451806847.002.A002

Sources: Statistics Canada; and Sargent (1995).

• Download Figure
• Download figure as PowerPoint slide

18. Unionization can affect employment by preventing outsiders from bidding down wages in recessions. Using a generic model, Blanchard and Fischer (1989) demonstrate that the greater the importance of unions in wage and employment decisions, the more persistent are the effects on unemployment of aggregate shocks to the economy. Riddell (1993) has examined the unionization experience in Canada and the United States since the mid-1960s. Over this period, the unionization rate in Canada rose from 30 percent in 1965 to 39 percent in the mid-1980s, after which it fell to 37 percent in 1995.⁵ In contrast, the unionization rate in the United States has declined from approximately 30 percent to 15 percent over the same period. Riddell argues that union density in Canada is approximately double that in the United States across a wide range of industries, occupations, and types of workers.

19. There is considerable debate at present about the effects of minimum wages and payroll taxes on employment. In particular, Card (1992) and Card and Krueger (1994) have argued that it is difficult to find any adverse effect of minimum wage increases on employment using microeconomic data. Neumark and Wascher (1995) argue that the absence of any effect on employment from minimum wages is the result of using samples in which minimum wages are binding for some individuals and not for others; isolating observations in which minimum wages are binding yields an adverse effect of minimum wages on the employment of teenagers and young adults. Moreover, Canadian studies generally find that unemployment rises as a result of a rise in the minimum wage relative to the average wage.

20. The debate about payroll taxes has focused on the adjustment of the labor market to changes in payroll taxes and the length of time required for this adjustment to take place. Theoretical models suggest that changes in payroll taxes eventually are borne by the employees so that over time effects on employment are expected to diminish. The payroll tax rate in Canada is usually defined as the ratio of supplementary labor income to total compensation. Supplementary income includes five major components: unemployment insurance premiums, Canada/Quebec Pension Plan contributions, private pension plan contributions, workers compensation, and various benefits under the umbrella of “welfare” benefits such as accident, health, and life insurance premiums. Beach, Lin, and Picot (1996) argue that employers’ contributions to employees’ private pension plans and the majority of welfare benefits are not proper payroll taxes because employers pay these benefits either on a voluntary basis or as a result of collective bargaining. Applying Beach et al.’s adjustment to the standard estimate of payroll taxes, indicates a fairly continuous rise in payroll taxes from 2 percent in the early 1960s to 11.6 percent in 1993. The expansion of unemployment insurance premiums is responsible for the largest share of the rise in the tax rate, accounting for 53 percent of the overall rate hike in the 1970s, 43 percent in the 1980s, and over three quarters since 1990.

21. A number of studies have attempted to derive empirical estimates of the effects of all of the above factors on the unemployment rate in Canada, According to Coe (1990), the increase in the generosity of the unemployment insurance system in the early 1970s caused a rise of 1 percentage point in the natural rate of unemployment over the 1970–88 period. This was offset by a decline in the minimum wage relative to the average wage. However, a rise in unionization and in payroll taxes caused a further 1 and 2½ percentage point increase, respectively, in the natural rate. Sargent and Sheikh (1996) also find that movements in the natural rate are driven by changes in unionization and Coté and Hostland (1996) argue that unionization and payroll taxes are the only variables that cointegrate with the unemployment rate. In Cole et al. (1995) the rise in the unionization rate accounts for roughly 2 percentage points of the rise in the unemployment rate since the early 1970s. Moreover, this effect is insensitive to the inclusion or exclusion of the level of the payroll tax rate. In contrast, Fortin, Keil and Symons (1995) in an analysis of Canadian regions and demographic groups over the 1966–93 period, find no significant effect on unemployment from unionization and argue that the effect of taxation is only temporary because changes in taxation are expected to be borne by the employees in the long run. They find that the rise in unemployment is associated with changes in demographic pressure and a more generous unemployment insurance system which more than offset the effects of the decline in the minimum wage relative to the average wage.

22. In a recent paper, Prasad and Thomas (1996) use Sargent’s UI index in a simple empirical model of the labor market to determine whether the generosity of the unemployment insurance system prolongs labor market adjustment in Canada. They find that the quantitative impact of the endogenous component of the unemployment insurance system does not have a large effect on the persistence of aggregate unemployment. Moreover, although the noncyclical component of unemployment insurance has played an important role in increasing unemployment over the 1970s and 1980s, it cannot explain in isolation the continued high level of unemployment, since the generosity of the unemployment insurance system has now fallen to levels experienced prior to the rise in structural unemployment. However, it is possible that other factors may have overshadowed the effects of the generosity of the unemployment insurance system over the recent period.

D. An Empirical Model of the Labor Market

23. There is considerable discussion about the impact of payroll taxation on the labor market in Canada, as well as in many other OECD countries, and there is no clear consensus about the duration of the impact. For example, in the studies on Canada discussed above, the importance attached to payroll taxes in explaining unemployment varies considerably. A drawback of a number of these studies is that they relate the unemployment rate to the payroll tax rate, and, therefore, ignore how the effects of changes in payroll taxation on unemployment are transmitted through changes in gross wages. The purpose of this section is to estimate a labor market model that incorporates the transmission of changes in payroll taxes into wages and, subsequently, into changes in employment and unemployment. This will help to determine the extent and the speed with which changes in payroll taxation are absorbed by workers.

24. In labor market models which allow for unemployment in equilibrium, the extent to which a change in payroll taxes leads to a change in employment depends on the wage setting process. Under the assumption that wages are determined in a union bargaining framework, an upward sloping wage-setting curve can be generated. This wage-setting curve, combined with a downward sloping labor demand curve and an upward sloping labor-supply curve, defines a labor market model with unemployment. Figure 1 presents a graphical depiction of such a labor market prior to and following a rise in payroll taxes. The initial equilibrium is at A with unemployment L_s0 -L_e0. In response to a rise in payroll taxes, firms will hire less labor at the initial net wage W₀ because gross wage costs to the firm will have risen by the amount of the tax. The resulting short-run equilibrium at B at the intersection of LD₁ and WS₀ is characterized by a slight reduction in the net wage to make workers more attractive to firms. However, this does not represent a long-run equilibrium because over time unemployed workers are expected to bid down wages further. For example, the wage-setting locus may shift downwards from WS ₀ to WS _x leading to a new equilibrium at C. The relative importance of the unionized employees versus the unemployed workers in the wage setting process is a key factor in determining the ultimate degree of wage adjustment and the extent to which a rise in payroll taxes reduces employment.

View Full Size

Canada: The Labor Market

Citation: IMF Staff Country Reports 1997, 020; 10.5089/9781451806847.002.A002

• Download Figure
• Download figure as PowerPoint slide

25. The degree of wage adjustment also affects labor supply decisions because the decline in the net wage received by employees generated by a rise in payroll taxes will lead to a reduction in the labor force. For example, labor supply at the long-run equilibrium C is lower than at the inital equilibrium A. Therefore the extent to which a rise in payroll taxes affects unemployment depends both on the behavior of labor demand and supply.

The estimated labor demand function comparable to LD₀ is as follows:

where E is employment, W is the aggregate real wage per hour worked inclusive of payroll taxes, and PTY is aggregate labor productivity per hour worked. The productivity variable is included in the specification because in the standard production fonction with labor and capital as the factors, the firm’s choice of whether to increase employment depends on the capital stock (or, equivalently on labor productivity), as well as the real gross wage.

26. The wage-setting equation is an upward sloping relationship between the level of real wages and employment, with labor productivity and payroll taxes included to proxy productivity developments and to isolate the effects of taxes, respectively.

where τ is the payroll tax rate.

27. The labor supply equation is an upward sloping relationship between the level of net wages and the participation rate. It also includes the child dependency ratio and a time trend to account for the rise in the participation rate over the past 25 years as women entered the labor force in greater numbers. The child-dependency ratio proxies the facility with which women were able to enter the labor force, and the time trend proxies improvements in the skill level of women.

where PR is the labor force participation rate, child-dep is the ratio of working-age females to the total working-age population, and t is a time trend.

28. The above equations were estimated using annual date over the period 1966 to 1993. The starting point for the data was constrained by the availability of employment data, whereas the end point was constrained by the availability of the modified payroll tax rate calculated by Beach et al. (1996).⁶ Weighted Symmetric t test statistics indicate that all of the above variables with the exception of the child-dependency ratio, have trended upwards over time, and therefore, it is necessary to consider the extent to which the relationships are cointegrated (Table 3 provides statistics on the stationarity of the data series). Cointegration tests based on the Johansen methodology indicated that labor demand, wage-setting, and labor supply equation each generated one cointegrating vector (Table 4). Moreover, the labor demand and wage setting curves were properly identified as downward and upward sloping relationships between the real wage and employment, respectively.

Table 3.

Canada: Unit Root Tests ^1/

^1/See text for data definitions. The Weighted Symmetric T test involves a weighted double-length regression in which the dependent variable is regressed on leads and lags of its own changes. See Pantula (1994) for more details. The Dickey-Fuller τ test involves regressing the dependent variable on its own lags and its own lag level; asymptotic probability values for the DF τ test were obtained from Mackinnon (1994). An asterisk denotes significance at the 10 percent level.

Table 3.

Canada: Unit Root Tests ^1/

Variable	Weighted Symmetrie τ Test		Dickey-Fuller τ Test
emp	−0.7		−0.7
Δ emp	−3.3	*	−2.2
real wage	−0.2		−4.2	*
Δ real wage	−2.8		−1.2
prod	−0.5		−7.8	*
Δ prod	−2.9		−7.2	*
ptax	−1.4		1.5
Aptax	−3.4	*	−1.3
partr	−0.2		−0.1
Δ partr	−2.7		−1.6
child-dep	−3.4	*	−8.0	*
Δ child-dep	−2.9		−2.2

^1/See text for data definitions. The Weighted Symmetric T test involves a weighted double-length regression in which the dependent variable is regressed on leads and lags of its own changes. See Pantula (1994) for more details. The Dickey-Fuller τ test involves regressing the dependent variable on its own lags and its own lag level; asymptotic probability values for the DF τ test were obtained from Mackinnon (1994). An asterisk denotes significance at the 10 percent level.

View Table

Table 3.

Canada: Unit Root Tests ^1/

Variable	Weighted Symmetrie τ Test		Dickey-Fuller τ Test
emp	−0.7		−0.7
Δ emp	−3.3	*	−2.2
real wage	−0.2		−4.2	*
Δ real wage	−2.8		−1.2
prod	−0.5		−7.8	*
Δ prod	−2.9		−7.2	*
ptax	−1.4		1.5
Aptax	−3.4	*	−1.3
partr	−0.2		−0.1
Δ partr	−2.7		−1.6
child-dep	−3.4	*	−8.0	*
Δ child-dep	−2.9		−2.2

^1/See text for data definitions. The Weighted Symmetric T test involves a weighted double-length regression in which the dependent variable is regressed on leads and lags of its own changes. See Pantula (1994) for more details. The Dickey-Fuller τ test involves regressing the dependent variable on its own lags and its own lag level; asymptotic probability values for the DF τ test were obtained from Mackinnon (1994). An asterisk denotes significance at the 10 percent level.

Table 4.

Canada: Johansen Maximum Likelihood Tests of the System of Equations

(Cointegration likelihood ratio test based on trace of the stochastic matrix)

Table 4.

Canada: Johansen Maximum Likelihood Tests of the System of Equations

(Cointegration likelihood ratio test based on trace of the stochastic matrix)

Null	Alternative	Test Statistic	95 Percent Critical Value
Wage Setting Equation
r=0	r≥1	49.8	48.3
r≤1	r≥2	23.1	31.5
r≤2	r≥3	7.1	18.0
r≤3	r≥4	2.6	8.2
Labor Demand Equation
r=0	r≥1	32.5	31.5
r≤1	r≥2	13.6	18.0
r≤2	r≥3	4.0	8.2
Participation Rate Equation
r=0	r≥1	32.3	31.5
r≤1	r≥2	17.8	18.0
r≤2	r≥3	5.4	8.2

View Table

Table 4.

Canada: Johansen Maximum Likelihood Tests of the System of Equations

(Cointegration likelihood ratio test based on trace of the stochastic matrix)

Null	Alternative	Test Statistic	95 Percent Critical Value
Wage Setting Equation
r=0	r≥1	49.8	48.3
r≤1	r≥2	23.1	31.5
r≤2	r≥3	7.1	18.0
r≤3	r≥4	2.6	8.2
Labor Demand Equation
r=0	r≥1	32.5	31.5
r≤1	r≥2	13.6	18.0
r≤2	r≥3	4.0	8.2
Participation Rate Equation
r=0	r≥1	32.3	31.5
r≤1	r≥2	17.8	18.0
r≤2	r≥3	5.4	8.2

29. The existence of cointegrating vectors in the three relationships allows the equations to be specified in log levels because they can be reparameterized in log levels and log differences so that all elements on both sides of the equations are stationary (see Ericsson (1995) for a discussion of this issue).

30. The main focus in this paper is in estimating the effects on the labor market of changes in the payroll tax rate. In the model above the payroll tax rate enters directly into the wage- setting equation, which then feeds into changes in employment and in the participation rate. In order to determine the effects of changes in the payroll tax rate, a dynamic equation was specified for payroll taxes. This equation relates changes in the payroll tax rate to lagged changes in the tax rate and to the lagged change in employment. The change in employment captures the fact that until recently, the employment insurance component of the payroll tax rate depended on the business cycle.⁷

31. In the wage-setting equation, the two lags of the dependent variable, the first lag on productivity, and the second lag on employment are the only significant variables.⁸ Moreover, with the exception of the wage variable, the coefficients on the various lags (for a given variable) offset each other. In particular, changes in payroll taxes raise the gross wage considerably in the first two years, but this increase is reversed in the following year. In the labor demand equation, the first lags of employment and the real wage are significant, together with the second lag on labor productivity. The sum of the two coefficients on the real wage is small, however, suggesting that employment only responds weakly to changes in the real wage. In the equation for the labor force participation rate, both lags of the dependent variable are significant, whereas the coefficients on net wages are insignificant and very small in magnitude. In the payroll tax equation, the lagged change in employment is significantly negatively related to the payroll tax rate. Moreover, the lagged coefficients of the dependent variable are all negative, so that an initial positive shock to payroll taxes is muted over time although it remains positive.

32. To trace the dynamic path of the variables in response to changes in payroll taxes, it is useful to examine impulse response functions computed using the coefficient estimates and corresponding confidence intervals based on replications of Monte-Carlo simulations.⁹ The panels in Chart 8 show the impulse response functions and corresponding confidence intervals for the levels of employment, the real gross wage, the payroll tax rate, and the participation rate in response to a 1 percentage point rise in the payroll tax rate. The shock to the payroll tax rate is permanent, although it declines gradually over time to a 0.6 percentage point increase. The effect of the rise of the payroll tax rate on the gross wage is immediate,¹⁰ and within one year the gross wage increases by the amount of the tax increase. Over time, the effect of the tax on the gross wage gradually weakens and is eliminated by the fifth year following the shock.

View Full Size

CANADA: IMPULSE RESPONSES TO PAYROLL TAX SHOCK ^1/

Citation: IMF Staff Country Reports 1997, 020; 10.5089/9781451806847.002.A002

Sources: Statistics Canada; and Fund staff estimates.^1/ The shock is a one percent increase in the payroll tax rate.2/ The upper and lower bounds of a 95 percent confidence interval are included.

• Download Figure
• Download figure as PowerPoint slide

33. The initial sharp rise in the gross wage generates a 0.3 percent decline in the level of employment after three years which is subsequently eliminated as the gross wage returns to its initial level. The initial decline in employment is comparable to recent estimates obtained by other authors. In particular, Parker (1995) suggests that a 1 percentage point rise in the payroll tax rate leads to a reduction in employment of 0.3 percentage point, whereas Scarth estimates the effect to be slightly lower at 0.15 percentage point. A discussion paper published by the Canadian Department of Finance has a comparable estimate of the long-term employment effects (Weldon, 1993).

34. The studies of Parker (1995), Scarth (1994), and Weldon (1993) assume that the effect of changes in payroll taxes on employment is permanent, whereas this study suggests that changes in payroll taxes are largely shifted to the employee over a six-year period. However, if shocks to payroll taxes occur more than once in a six-year period, this would have the effect of ratcheting up the effect of payroll taxes on employment over time. Over the period 1966–93 changes in payroll taxes were positive in all years except for the early and late 1970s and for 1989.

35. Changes in payroll taxes also affect the supply of labor. In order to determine its effects on the unemployment rate, it is necessary to consider the impact of payroll taxes on the participation rate. The impulse response for the participation rate was derived by calculating an impulse response for the net wage based on the profiles of the payroll tax and the gross wage and feeding the resulting net wage profile through the coefficient estimates of the participation rate equation. This procedure indicates that the participation rate rises slightly in the first few years in response to the short-run increase in the net wage but changes little over the long run.

36. The net effect on the unemployment rate of changes in employment and in the participation rate can be derived by assuming that the population is fixed and using the following identity between employment, the participation rate, population, and the unemployment rate.

where POP is the population.

37. The resulting impulse response profile for the unemployment rate indicates that the effect on the unemployment rate peaks at 0.4 percentage point four years following the shock to payroll taxes but that the effect subsides six years following the shock (Chart 8). This is comparable to the effect obtained by Tyrvainen (1995) that a 1 percentage point rise in the payroll tax rate causes a 0.3 percentage point rise in the unemployment rate.

38. The above discussion has focussed on the effects on the labor market of an increase in payroll taxes, and the question arises whether these effects are symmetric. According to the empirical model developed above the effects of the rise in payroll taxes on the net wage over the first few periods is negligible because the gross wage rises by the amount of the increase in payroll taxes. However, over time the gross wage returns to its initial level and the net wage falls correspondingly.

39. It is not possible to isolate whether there are short-run asymmetric effects between payroll tax increases and decreases in Canada because payroll taxes have risen almost continuously over the 1966–93 period. However, Kesselman (1996) argues that reducing employer payroll taxes may yield much less boost to employment than the employment lost from a comparable size of tax increase. This is because there is much less resistance to wage hikes than wage cuts, making the adjustment process more rapid for the former. Moreover, Gruber (1995) has found in analyzing a recent large discrete cut in employer tax rates in Chile that the entirety of the cut was reflected quickly in higher gross pay for employees.

40. Overall, the empirical results presented in this paper should not be taken as conclusive. They primarily serve to illustrate the possible impacts of changes in payroll taxes and point the way for further empirical work on the structure of the labor market in Canada. In particular, pooling labor market data for all the provinces would be a useful test of the robustness of the results presented above.

Table 5.

Canada: Criteria for Selecting Lag Length

Table 5.

Canada: Criteria for Selecting Lag Length

	Number of Lags
Model Criteria	1	2	3
Labor demand equation:
Akaike	−5.6	−5.6	−5.8
Schwarz	−8.2	−8.1	−8.1
Wage setting equation:
Akaike	−5.3	−5.3	−6.1
Schwarz	−7.7	−7.5	−8.2
Labor supply equation:
Akaike	−7.9	−8.3	−8.1
Schwarz	−10.4	−10.7	−10.4

View Table

Table 5.

Canada: Criteria for Selecting Lag Length

	Number of Lags
Model Criteria	1	2	3
Labor demand equation:
Akaike	−5.6	−5.6	−5.8
Schwarz	−8.2	−8.1	−8.1
Wage setting equation:
Akaike	−5.3	−5.3	−6.1
Schwarz	−7.7	−7.5	−8.2
Labor supply equation:
Akaike	−7.9	−8.3	−8.1
Schwarz	−10.4	−10.7	−10.4

Table 6.

Labor Market Model ^1/

^1/An asterisk denotes a variable or test statistic that is significant at the 10 percent level.

Table 6.

Labor Market Model ^1/

	Dependent Variable
	Real Wage		Employment		Participation Rate		Payroll Tax
W (−1)	1.26	*	−0.19	*
W (− 2)	−0.46	*	−0.01
W(1-τ)(−1)					−0.02
W(1-τ)(−2)					0.02
E (−1)	0.21		0.56	*
E (−2)	−0.21	*	−0.14
PTY (−1)	−0.45	*	0.10
PTY (−2)	0.53		0.79	*
τ	0.46
τ (−1)	0.49
τ (−2)	−1.04
PR (−1)					1.20	*
PR (−2)					−0.81	*
Child-dep					−0.08
Child-dep (−1)					0.32
Child-dep (−2)					−0.34
Δτ (−1)							−0.26
Δτ (−2)							−0.06
Δτ (−3)							−0.14
Δτ (−4)							−0.17
ΔE (−1)							−0.12	*
Goodness of Fit Statistics
R2	0.98		0.99		0.99
Autocorrelation test	14.5		14.65		16.8		0.34

^1/An asterisk denotes a variable or test statistic that is significant at the 10 percent level.

View Table

Table 6.

Labor Market Model ^1/

	Dependent Variable
	Real Wage		Employment		Participation Rate		Payroll Tax
W (−1)	1.26	*	−0.19	*
W (− 2)	−0.46	*	−0.01
W(1-τ)(−1)					−0.02
W(1-τ)(−2)					0.02
E (−1)	0.21		0.56	*
E (−2)	−0.21	*	−0.14
PTY (−1)	−0.45	*	0.10
PTY (−2)	0.53		0.79	*
τ	0.46
τ (−1)	0.49
τ (−2)	−1.04
PR (−1)					1.20	*
PR (−2)					−0.81	*
Child-dep					−0.08
Child-dep (−1)					0.32
Child-dep (−2)					−0.34
Δτ (−1)							−0.26
Δτ (−2)							−0.06
Δτ (−3)							−0.14
Δτ (−4)							−0.17
ΔE (−1)							−0.12	*
Goodness of Fit Statistics
R2	0.98		0.99		0.99
Autocorrelation test	14.5		14.65		16.8		0.34

^1/An asterisk denotes a variable or test statistic that is significant at the 10 percent level.