1. The upward trend in the unemployment rate in Canada during the 1970s and 1980s and the persistence of high unemployment experienced by Canada in the 1990s has generated a broad debate, with several competing hypotheses trying to explain these developments (Table 1 and Chart 1). This paper reviews the factors that may explain high and persistent unemployment in Canada, with particular emphasis on the role that a decline in the relative cost of capital may have had on trend unemployment.

Abstract

1. The upward trend in the unemployment rate in Canada during the 1970s and 1980s and the persistence of high unemployment experienced by Canada in the 1990s has generated a broad debate, with several competing hypotheses trying to explain these developments (Table 1 and Chart 1). This paper reviews the factors that may explain high and persistent unemployment in Canada, with particular emphasis on the role that a decline in the relative cost of capital may have had on trend unemployment.

I. The Persistence of Unemployment in Canada1

1. The upward trend in the unemployment rate in Canada during the 1970s and 1980s and the persistence of high unemployment experienced by Canada in the 1990s has generated a broad debate, with several competing hypotheses trying to explain these developments (Table 1 and Chart 1). This paper reviews the factors that may explain high and persistent unemployment in Canada, with particular emphasis on the role that a decline in the relative cost of capital may have had on trend unemployment.

Table 1.

Canada: Stylized Facts on Unemployment and Its Structural Factors 1976–96 1/

article image
Sources: Statistic Canada; Bank of Canada; Human Resources and Development Canada; and Fund staff estimates.

The data definition is as follows:

Output gap: calculated by Bank of Canada.

Unemployment Insurance benefits: approximated by Sargent’s index.

Unionization: the percentage of the labor force that belongs to a union.

The quarterly data is derived based on annual data provided by Human Resources Development Canada.

Payroll tax rate: ratio of wages, income and supplementary income to wages and income.

Relative price of equipment is the trend component of the ratio of the price deflator for equipment to the GDP deflator. The series was filtered using the Hodrick and Prescott filter.

Long-term corporate real bond yield: the Scottia McLeod interest rate divided by the annual rate of GDP inflation. The series was filtered using the Hodrick and Prescott filter.

Contemporaneous correlation for firsts differences.

CHART 1
CHART 1

CANADA: UNEMPLOYMENT RATE

(In percent)

Citation: IMF Staff Country Reports 1998, 055; 10.5089/9781451806915.002.A001

Source: Statistics Canada; and Fund staff estimates.

2. Numerous studies have tried to explain the rise in trend unemployment, the natural rate of unemployment, or the nonaccelerating inflation rate of unemployment (NAIRU)2 based on the role played by a variety of institutional and structural factors. These studies have considered the generosity of unemployment insurance benefits, the minimum wage, the degree of unionization of the labor force, payroll taxation, and the demographic structure of the labor force. The results have been mixed, with these structural factors not providing a completely adequate explanation of movements in the unemployment rate in Canada. A more recent line of research has explored the long-run effects on employment resulting from job reallocation following extensive technological change. The analysis presented here suggests that in Canada a declining trend in the cost of capital, associated with technological changes and innovations, has been an important factor in explaining the rise and persistence of unemployment.

3. Caballero and Hammour (1996, 1997) show that job creation and destruction tend to be concentrated during recessions, when the opportunity cost of reallocating labor tends to be the lowest.3 They suggest that certain market failures, together with policies that protect jobs, may disrupt a smooth reallocation of labor and increase unemployment. In examining European unemployment, they explore the relationship between capital/labor substitutability and stagnant employment. In their model, firms will tend to substitute capital for labor in the long run as a means of avoiding labor demands for higher incomes as well as restrictions and regulations imposed on the terms and conditions of employment, since the supply of capital is more elastic than labor. This implies that policies which tend to favor labor in the short run will likely be “self-defeating” in the long run.

4. In the case of Canada, some of the institutional reforms, policy measures, and relative price shocks that have taken place since the early 1970s may have contributed to increasing the cost of labor relative to capital. Generous, although declining, unemployment insurance benefits, higher rates of unionization, and a downward trend in the cost of capital may all have played a role. Moreover, during the 1980s and early 1990s, the Canadian economy has undergone widespread technological change and corporate restructuring. Several sectors may have experienced a process of “self-recreation,” associated with substantial increases in fixed investment. As noted by Freedman (1995), Canadian industry has faced a major restructuring as a result of increased globalization and free trade with the United States. The extension of the free trade agreement to Mexico has added to the competitive pressures on Canadian firms. Therefore, a process of substituting capital for labor may have disrupted the reallocation of labor from those sectors that have been affected significantly by increased competition in the domestic economy, especially in the presence of strong unions, generous unemployment insurance benefits, and increased payroll taxation. These factors may have contributed to the persistence of unemployment.

5. Since the mid-1970s, Canada has experienced an upward trend in the use of capital (Chart 2). The capital-labor ratio and the capital-output ratio increased by about 13 percent and 25 percent, respectively.4 Moreover, the capital-labor ratio increased sharply during the 1981–82 and 1990–91 recessions, supporting the hypothesis that the timing and efficiency of the capital-labor substitution process is heavily influenced by the opportunity costs for labor shedding, so that an economy will tend to experience more reallocation of labor during recessions.

CHART 2
CHART 2

CANADA: CAPITAL USE

Citation: IMF Staff Country Reports 1998, 055; 10.5089/9781451806915.002.A001

Source: Statistics Canada; and Fund staff estimates.

6. With respect to the cost of capital in Canada, its two main components—the real price of equipment and the real interest rate—have followed very different trends since the mid-1970s (Chart 3), The price of equipment relative to the GDP deflator has declined by about 20 percent, while the trend long-term real yield for Canadian corporate bonds increased by about 35 percent.5 At the same time, the real price of capital equipment relative to real producer wages has fallen sharply.6 Therefore, most of the impetus to fixed investment and the increase in the capital-labor ratio may be attributable to the decline in the real price of equipment. The situation has changed somewhat during the 1990s, as the decline in the real price of equipment has slowed, while the real producer wage has fallen by 3 percent since early 1992. However, partly offsetting these developments, the trend component of the long-term real yield on corporate bonds also resumed a downward path, dropping from an annualized rate of 7.8 percent in late 1991 to 5.8 percent in the fourth quarter of 1996.

CHART 3
CHART 3

CANADA: COST OF CAPITAL

Citation: IMF Staff Country Reports 1998, 055; 10.5089/9781451806915.002.A001

Source: Statistics Canada; and Fund staff estimates.

7. A trend decline in the cost of capital, together with rising labor costs, helps to explain the upward trend in the capital-labor ratio between 1976 and 1996, as suggested by the equation presented in Table 2.7 The equation explains relatively well the upward trend in the capital-labor ratio, especially during the 1980–82 and 1990–91 periods (Chart 4). In addition to cyclical factors, the downward trend in the structural component of the price of equipment relative to producer wages was significant in explaining the substitution of capital for labor. The equation also shows that shocks to the price of equipment had a larger impact than the long-term real yield on corporate bonds on the capital-labor ratio.

Table 2.

Canada: Capital-Labor and Capital-Output Ratios 1976-96

article image
Source: Fund staff estimates, t-statistics reported in parenthesis.

These variables represent the structural component after being filtered using the Hodrick and Prescott filter. Output gap

CHART 4
CHART 4

CANADA: CAPITAL-LABOR RATIO

Citation: IMF Staff Country Reports 1998, 055; 10.5089/9781451806915.002.A001

Source: Statistics Canada; and Fund staff estimates.

8. The capital-labor ratio equation suggests that the elasticity of substitution between capital and labor was about 0.2–0.3 between 1976 and 1996 (Table 2).8 Although significant, this coefficient is low and seems to suggest that, since the mid-1970s, capital and labor behaved more as complements than substitutes in the production process. However, the specification of the model chosen may result in a biased parameter of the elasticity of substitution. In measuring the capital-labor ratio, labor was approximated using employment data. Since the level of employment tends to vary less than hours worked during the business cycle, the underlying degree of substitution between capital and labor may be underestimated, since firms tend to adjust more hours worked rather than employment. Therefore, if employment was adjusted for changes in the number of hours worked, the elasticity of substitution between capital and labor may be even higher. For this reason, an alternative equation was estimated using the capital-output ratio as a dependent variable, which specifies this ratio as a function primarily of the real price of equipment and the long-term real yield on corporate bonds. The results, also presented in Table 2 (and Chart 5), show that the elasticity of substitution increases markedly to 2–2.25.9 These results support the suggestion that there has been a strong substitution of capital for labor in Canada since the mid-1970s, reflecting the rise in the relative cost of labor.

CHART 5
CHART 5

CANADA: CAPITAL-OUTPUT RATIO

Citation: IMF Staff Country Reports 1998, 055; 10.5089/9781451806915.002.A001

Source: Statistics Canada; and Fund staff estimates.

9. Following Cotè and Hostland (1996), an unemployment equation was estimated that explains unemployment on the basis of structural factors, using cointegration theory.10 In addition to unionization, the generosity of unemployment insurance benefits, payroll taxes, and demographic changes in the labor force, the trend decline in the cost of capital (specifically, the relative price of equipment and the long-term real yield on corporate bonds) is included in the equation. Estimated using quarterly data for the period 1976-96, the main equation suggests that most of the traditional structural factors had significant effects on trend unemployment (Table 3 and Chart 5). The cost of capital also appears to have contributed significantly to the persistence of unemployment in Canada.

Table 3.

Canada: Structural Unemployment 1976:3-1996:3

article image
Source: Fund staff estimates. Probability values are reported in parenthesis. These represent the probability of accepting the null hypothesis that the estimated coefficient is zero. A probability value less than any significance level indicates rejection of the null hypothesis. The probability values may also be interpreted as the lowest level of significance that can be used to reject a given null hypothesis.

10. Based on the regression presented in Table 3, an estimate of the trend unemployment rate was derived. After declining in the late 1980s, the trend unemployment rate rose to about 9½ percent in late 1993, reflecting a continuing decline in the real cost of capital as the price of equipment and long-term interest rates fell in real terms. The trend unemployment rate has subsequently declined to about 8 percent by end-1996, largely on account of a sharp reduction in the generosity of unemployment benefits.11 The recent stabilization of the downward trend in the cost of capital (in particular with respect to the real price of equipment relative to the real producer wage), together with the recent reforms of the unemployment insurance system, should contribute to bring down the trend unemployment rate over the near term.

CHART 6
CHART 6

CANADA: UNEMPLOYMENT

(In percent)

Citation: IMF Staff Country Reports 1998, 055; 10.5089/9781451806915.002.A001

Source: Statistics Canada; and Fund staff estimates.

List of References

  • Caballero, R. And Hammour, M., 1997, “Jobless Growth: Appropriability, Factor Substitution, and Unemployment”, NBER Working Paper No. 6221 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Caballero, R. And Hammour, M., 1996, “On the Timing and Efficiency of Creative Destruction”, Quarterly Journal of Economics, Vol. 2, pp. 805852.

    • Search Google Scholar
    • Export Citation
  • Card, D. and Riddell, C., 1993, “A Comparative Analysis of Unemployment in Canada and the United States”, in “Small Differences that Matter: Labor Markets and Income Maintenance in Canada and the United States,” edited by D. Card and R. Freeman (Chicago, University of Chicago Press).

    • Search Google Scholar
    • Export Citation
  • Cerisola, M., 1998, “The Persistence of Unemployment in Canada: Is Too Much Capital Too Cheap?,” IMF Working Paper, (Forthcoming).

  • Cotè, D. and Hostland, D., 1996, “An Econometric Examination of the Trend Unemployment Rate in Canada,” Bank of Canada, Working Paper 96-7.

    • Search Google Scholar
    • Export Citation
  • Cox, W. and Alm, R., 1992, “The Churn,” Federal Reserve Bank of Dallas Annual Report, pp. 518.

  • Freedman, C., 1995, “The Canadian Experience with Targets for Reducing and Controlling Inflation” in “Inflation Targets,” edited by Leiderman and Svensson.

    • Search Google Scholar
    • Export Citation
  • Laidler, D., 1997, “The Unnatural NAIRU,” C.D. Howe Insitute Backgrounder, November.

  • Phillips, P. and Hansen, B., 1990, “Statistical Inference in Instrumental Variables Regression with 1(1) Processes,” Review of Economic Studies, 57, pp. 99125.

    • Search Google Scholar
    • Export Citation
  • Poloz, S., 1994, “The Causes of Unemployment in Canada: A Review of the Evidence,” Bank of Canada, Working Paper 94-11.

  • Sargent, T., 1995, “An Index of the Generosity of Unemployment Insurance,” presented at the CEA Meetings, Montrèal (June).

  • Siebert, H., 1997, “Labor Market Rigidities: At the Root of Unemployment in Europe,” Journal of Economic Perspectives, Vol. 11, No. 3, (Summer), pp. 3754.

    • Search Google Scholar
    • Export Citation
  • Stock, J. and Watson, M. (1993), “A Simple Estimator of Cointegrating Vectors in Higher-Order Integrated Systems,” Econometrica, Vol. 61, No. 4 (July), pp. 783820.

    • Search Google Scholar
    • Export Citation
1

Prepared by Martin Cerisola. A more complete discussion of this issue and the econometric estimates is provided in Cerisola (1998). The empirical analysis was conducted before a revision of Canada’s national accounts was implemented during November-December 1997. This revision was comprehensive, including significant changes to several price deflators. These changes may have had some impact on the results presented in this paper.

2

Although closely related concepts, the trend unemployment rate, the natural rate of unemployment, and NAIRU have been treated in different ways in the literature, without full agreement on their interpretations. Cotè and Hostland (1996) note that conceptually trend unemployment may lie between the natural rate and the NAIRU. The natural rate is related to unemployment when the economy is at its steady state equilibrium, where flows into and out of the unemployment pool tend to be similar on average, and where agents are not “surprised” by inflation. The NAIRU is defined somewhat more narrowly, relating the unemployment rate to short-run excess demand in the labor market and the impact of excess demand on wages and inflation.

3

For a discussion and evidence on “creative destruction” in the U.S. economy, see Cox and Alm (1992). For alternative evidence in the Canadian case, Picot and Finn (1997) find that both hiring and separations fall in Canada during periods of labor market slack. These results would provide some evidence against unemployment being driven by creative destruction.

4

The capital-labor ratio was approximated by the net capital stock to total seasonally adjusted employment for workers aged above 15 years old. These figures refer to the percentage change between the quarterly averages for the periods 1976–79 and 1990–96.

5

Both figures refer to the percentage change between the quarterly averages for the periods 1976–79 and 1990–96.

6

The real producer wage is defined as hourly compensation deflated by the producer price index.

7

Assuming a constant elasticity of substitution (CES) production function, the first order

log(KtLt)=σlogΘσlog(rtwt)+κt

conditions from a static maximization problem result in the following specification for the capital-labor ratio:

where Kt is the capital stock, Lt is labor, Θ is a constant which represents the standard allocation parameter δ in the CES production function, σ is the elasticity of substitution between capital and labor, rt is the real cost of capital, and wt is the real producer wage.

8

This result is robust to different equations and specifications for the capital-labor ratio. Two sets of regressions were conducted. The first used the Fully-Modified OLS (FMOLS) procedure proposed by Phillips and Hansen and included the variables in levels. The FMOLS estimates the long-run parameters using a procedure which corrects for possible serial correlation in the residuals without having to specify the dynamics of the model. It is appropriate for estimation and inferences whenever there exists a single cointegrating relation between a set of nonstationary variables. The second specification estimated the equation in first-differences with the Newey-West long-run variance-covariance matrix. The Newey-West procedure computes a serial correlation and heteroskedasticity consistent estimate of the variance-covariance matrix of the parameters. Both sets of regressions included the output gap to account for cyclical variations in the capital-labor ratio, and the long-term real yield on corporate bonds. The regressions also were run using the structural component of the capital-labor ratio as estimated using a Hodrick-Prescott filter. The estimated elasticity of substitution was also robust to specifications which excluded the output gap and the long-term real yield on corporate bonds.

9

This equation was estimated with the Stock and Watson procedure, which allows for lags as well as leads of the first differences of the structural factors in order to account for potential Granger causality from the error term governing the set of structural factors to the unemployment rate. The standard errors were computed following the Newey-West procedure.

10

Since the unemployment rate and the set of structural factors are nonstationary series, the cointegration approach allows the estimated long-run parameters to have well-defined statistical properties, avoiding the “spurious regression” problem.

11

The predicted gap between the actual and structural unemployment rates in 1988–89 appear to be too wide (in absolute terms) relative to those observed in 1982–84 or 1991–93, where labor market slack was particularly significant. This raises some concerns on the appropriateness of the econometric specification presented, which would deserve further research.

Canada: Selected Issues
Author: International Monetary Fund