This paper examines economic developments and policies in Canada during 1990–95. Spurred by the robust growth in the United States and the easing of monetary conditions between 1991 and 1993, economic growth in Canada continued to strengthen during 1994. Real GDP grew by 4.5 percent in 1994 after growing by 2.2 percent in 1993 and 0.6 percent in 1992. Economic growth in 1994 was led by exports and investment in machinery and equipment. However, growth was more broadly based in 1994; private consumption strengthened, and there was a rebound in residential and nonresidential construction.

Abstract

This paper examines economic developments and policies in Canada during 1990–95. Spurred by the robust growth in the United States and the easing of monetary conditions between 1991 and 1993, economic growth in Canada continued to strengthen during 1994. Real GDP grew by 4.5 percent in 1994 after growing by 2.2 percent in 1993 and 0.6 percent in 1992. Economic growth in 1994 was led by exports and investment in machinery and equipment. However, growth was more broadly based in 1994; private consumption strengthened, and there was a rebound in residential and nonresidential construction.

III. Estimates of the Structural Rate of Unemployment 1/

Numerous studies have attempted to explain the large rise in the unemployment rate in a number of industrialized countries over the past 20 years. However, as emphasized in a recent survey by Bean (1994), the competing theories of unemployment often do not provide clear cut explanations for the unemployment experiences of both Europe and North America. 2/ Nonetheless, estimates of the level of structural unemployment, or the nonaccelerating-inflation rate of unemployment (NAIRU), are an important element in gauging wage and price pressures in the economy.

Updated estimates of the NAIRU for Canada are presented below. 3/ The methodology adopted is to estimate long-run values for the employment/population ratio and the participation rate separately, and use these to calculate the NAIRU. In particular, the employment/population (E/P) ratio is assumed to be explained by the change in payroll taxes (Δptax), unemployment insurance benefits (uben), and the unionization rate (union), defined as the ratio of union membership to nonagricultural workers.

It is assumed that only the change in payroll taxes affect the employment/population ratio because, in the long run, wages are assumed to offset the level of payroll taxes. 4/ The unionization rate is included as a determinant of E/P because unionization is assumed to affect the bargaining process between firms and workers, causing wages to rise above their long-run levels, and employment to fall below its long-run level. The participation rate is also included as a determinant of the ratio of employment to population, since it is assumed that the ability of firms to create employment is limited in the long run by the availability of labor. 1/ The output gap (Y-Ypot) is also assumed to affect the employment/population ratio on the assumption that short-term shocks to the economy cause permanent effects on the level of employment. 2/

The participation rate is assumed to be principally determined by demographic variables. For example, the large rise in the female participation rate over the last 30 years coincided with a decline in the male participation rate. 3/ Rather than attempting to explain this phenomenon, it is simply assumed that the aggregate participation rate is a positive function of the female population share (F/P). Similarly, the population share for the 15-24 year-old age bracket (Y/P) is expected to have a negative influence on the participation rate because young people delay entry into the labor force in order to receive additional education and training. 4/ In addition, unemployment insurance benefits (uben) are assumed to encourage participation, while the output gap is also included to control for persistent effects of cyclical output movements on the participation rate. 5/

As a preliminary step, tests were conducted to determine whether the data were stationary. These tests indicate that the data were integrated of order 1 (with the exception of the female/population ratio and the change in payroll taxes, which were stationary in levels). When data are nonstationary, standard significance tests using the t-statistics are not valid.

Therefore it is necessary to find a long-run, linear relationship whose residuals are stationary in order to conduct statistical inference. This relationship is termed a cointegrating relationship. As a result, the existence of long-run structural relationships for E/P and L/P was tested by estimating the following equations (t-statistics are in parenthesis). 1/

(E/P)t=35.5(22.7)+0.68(17.3)(L/P)t+0.26(11.7)(YYpot)t0.53(5.7)(union)t0.49(0.9)ubent0.14(0.6)Δptaxt+t
R2=0.96ADFstatistic=3.39
(L/P)t=7.6(3.2+1.7(30.0)(F/P)t0.39(14.7)(Y/P)t+0.17(9.6)(YYpot)t+0.46(4.3)d1.ubent
R2=0.97ADFstatistic=3.16

The results were generally satisfactory. The estimated coefficients were of the correct sign and augmented Dickey-Fuller (ADF) tests indicated that the hypothesis of cointegration--i.e., the existence of a long-run relationship--could not be rejected. 2/ When the variable measuring the effect of UI benefits (uben) was included in the equation explaining the participation rate, the estimated coefficient had the incorrect sign. However, when uben was multiplied by a dummy for periods in which real GDP declined (d1 = 1 when GDP growth is negative, 0 otherwise), the estimated coefficient was positive, presumably capturing the reluctance of the unemployed to leave the labor force during cyclical downturns because of the generosity of unemployment insurance.

The estimated coefficient on the participation rate in the equation for the employment/population ratio is considerably less than unity, suggesting that an increase in the participation rate tends to be associated with an increase in the. unemployment rate. The female/population ratio also has a strong effect on the participation rate; the results indicate that a 10 percentage point rise in the ratio results in a 17 percentage point rise in the participation rate. The predicted effect of a change in the output gap on the employment rate exceeds the effect on the participation rate, suggesting that the NAIRU tends to rise during prolonged economic downturns. Finally, estimates indicate that a rise in unemployment compensation during cyclical downturns causes a rise in the participation rate- (and unemployment), after controlling for movements in the output gap.

Using the results above, an estimated long-run employment/population ratio is derived by setting the output gap and the change in payroll taxes equal to zero and using the long-run participation rate, a Hodrick-Prescott filtered unionization rate, and the actual value for the generosity of unemployment compensation in the predicted equation. 1/ The estimated long-run participation rate is derived by setting the output gap and the GDP growth dummy equal to zero.

The estimates of the long-run employment/population ratio and participation rate can be used to calculate the NAIRU as follows:

NAIRU=1(E/P)1r/(L/P)1r

The long-run and actual values for the employment/population ratio, the participation rate, and the unemployment rate are presented in Chart III-1. Since 1988, when it reached its peak at 9 percent, the NAIRU has gradually declined and is estimated to have fallen to 8 3/4 percent in 1994, as a result of a slight decline in the unionization rate and the generosity of UI benefits.

CHART III-1
CHART III-1

CANADA COMPONENTS OF THE UNEMPLOYMENT RATE

Citation: IMF Staff Country Reports 1995, 046; 10.5089/9781451806823.002.A003

Source: Statistics Canada (supplied by DRI); and Fund staff estimates.

Alternative Estimates of the NAIRU

(In percent)

article image

The results appear to be within the range of recent estimates (see tabulation above). For example, OECD (1994) estimates that the NAIRU fell from about 9 percent during the late 1980s to 8 1/2 percent in 1993. Point estimates of the NAIRU are heavily dependant on the equation specification (see the analysis by Setterfield et al.). Nonetheless on the basis of the analysis above and the estimates cited below, a reasonable estimate of the NAIRU would appear to fall within the 8 to 9 percent range.

References

  • Alleyne, T., “Provisional Estimates of Potential Output and NAIRU,” (mimeographed, International Monetary Fund, 1993).

  • Bean, C., “European Unemployment: A Survey,” Journal of Economic Literature Vol. XXXII (June 1994).

  • Bean, C. and R. Layard, “Why Does Unemployment Persist?” Scandinavian Journal of Economics (1989: 2).

  • Blanchard, O. and L. Summers, “Hysteresis and the European Unemployment Problem”, NBER Macroeconomics Annual (1986).

  • Blanchard, O., “What Can Competing Theories of Unemployment Say About Cross Country Unemployment Rate Differences: The Case of Spain and Portugal,” mimeo, MIT (January 1995).

    • Search Google Scholar
    • Export Citation
  • Card, D. and C. Riddell, “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, ed. by D. Card and R. Freeman, (Chicago: University of Chicago Press 1993).

    • Search Google Scholar
    • Export Citation
  • Coe, D.. “Structural Determinants of the Natural Rate of Unemployment in Canada,” Staff Papers, (March 1990).

  • Coté, A. and D. Hostland, “An Econometric Examination of the NAIRU in Canada Using a Cointegration Approach,” Bank of Canada Technical Report (forthcoming).

    • Search Google Scholar
    • Export Citation
  • Feldstein, M., “The Effect of Unemployment Insurance on Temporary Layoff Unemployment,” American Economic Review (March 1978).

  • Fortin, P., “The Unbearable Lightness of Zero Inflation Optimism,” Canadian Business Economics (Spring 1993).

  • Layard, R., How to Beat Unemployment, Oxford University Press (1986).

  • Nickell, S. and B. Bell, “Unemployment Policy: How Should Governments Respond to Unemployment,” mimeo, (University of Oxford, 1994).

    • Search Google Scholar
    • Export Citation
  • Poloz, S., “The Causes of Unemployment in Canada: A Review of the Evidence,” mimeo, Bank of Canada (1994).

  • Robidoux, B. and C. Dea, “The Natural Rate of Unemployment in Canada: A Statistically-Robust Estimation,” Department of Finance Canada (1994).

    • Search Google Scholar
    • Export Citation
  • Rose, D., “The NAIRU in Canada: Concepts, Determinants and Estimates,” Bank of Canada Technical Report No. 50, (1988).

  • Setterfield, M., D. Gordon and L. Osberg, “Searching for a Will o’ the Wisp: An Empirical Study of the NAIRU in Canada,” European Economic Review (January 1992).

    • Search Google Scholar
    • Export Citation
  • Van Rijckeghem, C., “Endogeneity in Structural Unemployment Equations: The Case of Canada,” International Monetary Fund Working Paper 93/94.

    • Search Google Scholar
    • Export Citation
1/

Prepared by Alun H. Thomas.

2/

For example, Blanchard (1995) examines unemployment in Spain and Portugal and finds that none of the existing theories of unemployment can explain the wide difference between the unemployment rates in these two countries. Card and Riddell (1993) also are unable to explain most of the recent increase in the gap between Canadian and U.S. unemployment rates. Setterfield, et al. (1992) illustrate that the confidence interval around estimates of the nonaccelerating inflation rate of unemployment (NAIRU) for prime-age males in Canada spans almost the entire range of unemployment rates experienced by this group over the 1956-87 period.

3/

For a recent discussion of determinants of structural unemployment see Poloz (1994). Estimates of Canadian structural unemployment include those by Rose (1988), Coe (1990), van Rijckeghem (1993), and Robidoux and Dea (1994).

4/

Underlying this assumption is the view that the long-run supply of labor is relatively inelastic. While the empirical evidence on the link between labor taxes and long-run unemployment is mixed, a recent analysis by Nickell and Bell (1994) suggests that the effect should only be present in the short run.

1/

For example, many authors have noted that Europe’s inability to create jobs over the past 20 years is partly a result of its weak labor force growth (see Layard 1986). Bean and Layard (1989) emphasize that over the period 1900-85 the unemployment rate in the United Kingdom showed little trend because employment adjusted to the massive labor force changes. They argue that “in the long-run employment follows the labor force, and any meaningful model of the economy must reflect this tendency.”

2/

The inclusion of an estimate of the output gap is somewhat questionable since it means that the estimate of the NAIRU depends in part on the estimate of the output gap. However, the apparent ratcheting up of the unemployment rate in Europe and Canada following economic downturns during the past 20 years suggests that cyclical developments have longer term effects. This phenomenon has been termed “hysteresis” by Blanchard and Summers (1986).

3/

The female participation rate rose from 35 percent in 1966 to 58 percent in 1993, whereas the male participation rate fell from 80 percent to 73 percent over the same time period.

4/

This effect has been particularly strong in recent years. Over the period 1989-93 the youth participation rate declined by 6 percentage points.

5/

The uben variable includes an index of the generosity of unemployment compensation across provinces because in Canada, the maximum number of weeks an individual is eligible for UI benefits and the minimum number of weeks of work needed to qualify for benefits depend on provincial unemployment rates. The use of this type of index can lead to an estimation bias and therefore we use the aggregate replacement rate as an instrument for uben.

1/

The employment/population, female/population and youth/population ratios and participation rate were obtained from DRI, the unionization rate and the unemployment benefit ratio were obtained from the Department of Human Resources, and payroll taxes were obtained from the Bank of Canada.

2/

The ADF test statistic indicates that the linear combination of variables in both the employment-population and participation rate equations are stationary at the 10 percent significance level. The ADF test failed when the output gap variable was excluded from the equations, indicating the importance of hysteresis effects.

1/

Filtering the unionization rate corrects for its countercyclical behavior: unionization tends to rise during recessions, as unionized workers are better able to keep their jobs during recessions than nonunionized workers (see van Rijckeghem (1993), and Robidoux and Dea (1994)).