IV Adjustments for Changes in Unemployment and Unemployment Compensation

Ahsan Mansur, Richard Haas, and Peter Heller
Published Date:
May 1986
  • ShareShare
Show Summary Details

Under the present fiscal impulse methodology of the Fund, the growth of government expenditures other than unemployment insurance benefits is regarded as cyclically neutral if it is equal to the growth of potential output. Unemployment insurance benefits are excluded from the base-year expenditure ratio and from actual expenditure in a given period, implying that any change in unemployment insurance benefits is treated as a wholly cyclical phenomenon.

This treatment of unemployment insurance benefits (UIBs), which was introduced in early 1983, was based on the argument that these benefits respond to the state of the economy in a manner that is symmetrical, in its aggregate demand effects, to the movement of tax revenues. As the economy moves into a recession (recovery), UIB payments increase (decrease) sharply. In constructing a measure of the cyclically neutral budget, the sensitivity of revenues to the cyclical position is considered by assuming a higher cyclically neutral deficit in a recession. The increase in unemployment insurance benefits that is also owing to the cyclical position should presumably be treated in an analogous fashion.

This approach is deficient in several respects. The present methodology implicitly assumes that an economic recovery will return the unemployment rate to the base-year level. It does not take account of significant changes in structural or noncyclical unemployment owing to the permanent disappearance of jobs because of changes in technology, foreign competition, consumer preferences, geographical relocation of industries, or demographic factors. Neither does it take account of changes in unemployment benefit payments per worker. Thus, the adequacy of the present approach may be affected by developments in the structural unemployment and benefit rates since the base year.

To the extent that changes in unemployment insurance benefit payments are owing to structural or other noncyclical factors, the expansionary impact of the budget will be understated. For example, in a period of growing noncyclical unemployment (in terms of the unemployment rate prevailing at the cyclical peak) and consequently increasing unemployment benefit payments, measured (expansionary) fiscal impulses would be biased downward. This section suggests an approach to modifying the current impulse methodology to take account of shifts in the structural unemployment rate.

The current World Economic Outlook methodology lakes account of unemployment insurance payments in the following manner. From equation (2),


UIB = unemployment insurance benefit payments


in the base year. The fiscal impulse is then derived in a manner analogous to equation (3).

As an alternative, one may attempt to explicitly disaggregate the unemployment insurance benefit payments UIBt into normal and cyclical components


  • UIB0 = U0UB0

  • U0 = number of unemployed persons who received benefits in the base year

  • I = an index of the total labor force, (I1978 = 1.0)

  • UB0 = average unemployment benefits per unemployed person in the base year

  • P = cost of living index

  • Uc = level of cyclical unemployment

  • FIS* = measure of fiscal stance derived from using this approach

The alternative methodology in equation (9) implicitly assumes that the level of unemployment at any point in time may be broken down into three categories. That part of unemployment which would remain after the economy returned to the normal capacity utilization prevailing in the base year is referred to as the “normal” level of unemployment and is equivalent to the rate prevailing in the base year (U0). The normal level of unemployment varies from one economy to another, depending on various structural, geographic, and demographic characteristics, including differences in voluntary turnover rates. The cyclical component of unemployment, alternatively called “demand-deficient” unemployment, is caused by a lack of effective demand. The third component of unemployment is caused by structural changes in the economy that have taken place since the base year and reflects both labor market mismatches and capital shortages.

Unemployment insurance benefit payments for the normal level of unemployment are measured by the term in the second set of parentheses in equation (9). In any given year, this allows for the growth in the labor force and assumes that the benefit per unemployed person remains unchanged in real terms at the base-year level. The cyclical component of the unemployment insurance benefit payments is defined analogously; additional (reduced) benefits in the form of UIB are treated as expansionary (contractionary) and are residually attributable to the fiscal stance. This equation can be reduced to

Equations (8) and (10) are identical if the entire change in the level of unemployment over the base year is attributable to cyclical unemployment Utc, so that Utc=UtU0It, and if there has been no change in average real unemployment benefits since the base year.

Movements in the structural unemployment rate for the major industrial economies may be qualitatively examined using an approach developed by Hancock (1963) and Solow (1964). According to these authors, if the unemployment rate is the same in two different years and if there is a greater number (or greater rate) of unfilled vacancies in the later year, one may assert an increase in the structural unemployment rate. The inverse relationship between the unemployment rate and job vacancies has been called the “Beveridge curve,” with shifts in the curve representing a change in the incidence of structural unemployment. However, for all countries, cyclical fluctuations can potentially explain most of the fluctuations in unemployment, since levels of unemployment and job vacancies are inversely related to each other (Table 5).

Table 5.Seven Major Industrial Countries: Indices of Unemployment and Jobs Vacant/Help Wanted Advertising, 1975–82(1978 = 100)
Help wanted advertising99.095.092.1100.0113.9123.8135.660.4
United States
Help wanted advertising53.263.378.2100.0103.785.679.256.9
Jobs vacant87.690.281.6100.081.1101.096.490.9
Jobs vacant125.3142.5119.5100.0100.0102.379.396.6
Germany, Federal Republic of
Jobs vacant95.995.593.9100.0123.6125.284.642.7
Jobs vacant
United Kingdom
Jobs vacant71.457.675.2100.0114.868.146.252.9
Source: Organization for Economic Cooperation and Development, Main Economic Indicators, various issues.
Source: Organization for Economic Cooperation and Development, Main Economic Indicators, various issues.

The Organization for Economic Cooperation and Development (1983 a) study on unemployment indicates that for most major industrial countries, shifts in the Beveridge curve appeared to occur during the 1960s and early 1970s. However, for none of the economies does the OECD study indicate any significant shift in the Beveridge curve since 1978, though it may be too early to pick up such a shift in a statistically significant way. The French data on vacancies and unemployment, in particular, suggest the possibility of increased structural unemployment.

This does not preclude the possibility of a change in the structural unemployment rate in the medium term as the industrial economies recover from the current recession and structural adjustments continue in various sectors during the period through 1987. If the rate of unemployment after the recovery is expected to be u, compared with u0 in the base year, the difference (u - u0) (which appears to be positive over this period) may be treated as a change in the incidence of structural unemployment, with the change gradually apportioned among the intervening years. Alternatively, the change may be treated as a discrete structural shift starting in the recession years.20

The discrete adjustment may be justified on the ground that during the recession years, what is observed as cyclical unemployment may be partly structural, and thus the gradual change approach may overestimate the cyclical component of unemployment and understate the measure of the fiscal stance.

Using the average rate of structural unemployment, one may estimate the potential UIB payments owing to changes in the structural unemployment rate from that of the base year. If one explicitly allows for changes in structural unemployment, equation (10) can be extended to


  • us = structural unemployment rate

  • L = size of the labor force

  • Us = usL, the level of structural employment

  • Uc′ = U – U0 I – Us the level of cyclical unemployment

and FIS* denotes the fiscal stance.

Any discretionary change in real unemployment insurance benefits per worker should also be viewed as noncyclical, with appropriate adjustments made to the cyclically neutral budget balance.21 Equations (10) and (11) treat unemployment benefits as fully indexed. Any changes which either increase or decrease the real benefit level from that of the base period may be considered as discretionary changes.22

Using the general methodology described above, two approaches can be used to estimate the base-year average unemployment benefit payment, depending on the particular choice of average unemployment benefit rates. One approach sets the base-year rate of average UIB payments as the ratio of total UIB payments to the total number of unemployment benefit recipients. Alternatively, the product of the replacement ratio and disposable income of a “typical worker”23 may serve as the rate of average UIB payments.24 Given the total number of unemployed persons receiving benefits, total UIB payments are endogenously determined if the second method is used, while total UIB payments are exogenous if the first method is used. However, the use of the typical worker concept in the second method may bias the estimated total of UIB payments, since the income and family structure of the typical worker is generally not the same as that of the average unemployed person. The bias will differ from country to country and may be very significant; any significant bias in the average value will distort the total, resulting in undesirable changes in the measures of fiscal stance and of impulses.

An unbiased estimate of the costs of unemployment benefits based on a weighted average of the occupational, marginal, and sex categories of unemployed persons requires more data and detailed analysis than the typical worker approach does. Since the replacement ratios corresponding to the “average unemployed” are not available for most of the major industrial countries, the first method will be used in the following analysis.25

The empirical effects of incorporating these modifications relating to unemployment insurance into the Fund’s fiscal impulse methodology are indicated in Table 6. Estimates are provided under the assumption of a linear trend in the structural unemployment rate. The results suggest that the alternative assumptions do not significantly alter the qualitative results on the fiscal impulse for the period under consideration. Quantitatively, the results can change the impulse by as much as 0.4 percent of GDP (as for the Federal Republic of Germany in 1978), though in general the changes tend to be less than 0.1 percent of GDP. The noteworthy difference in the case of the Federal Republic of Germany arose from a significant increase in the labor force in 1979.26

Table 6.Seven Major Industrial Countries: Fiscal Impulse of Central Government Under Alternative Treatments of Unemployment Insurance Benefits, 1976–851(As percentages of GDP)
Present fiscal impulse methodology0.041.210.98−0.65−0.37−1.091.360.71−0.08−1.19
United States
Present fiscal impulse methodology−0.870.16−0.790.350.160.522.03−0.07−0.15
Present fiscal impulse methodology0.630.190.211.04−0.01−0.31−0.64−0.45−0.23−0.68
Present fiscal impulse methodology−1.07−0.351.87−0.85−0.680.960.02−0.32−0.34−0.23
Germany, Fed. Rep. of
Present fiscal impulse methodology0.29−0.370.12−0.38−0.69−1.75−0.11−0.21−0.07
Present fiscal impulse methodology−0.65−0.775.30−2.98−0.070.590.65−0.07−0.09
United Kingdom
Present fiscal impulse methodology−2.49−2.352.740.53−2.36−2.46−1.252.43−1.22−0.54
Source: Fund staff estimates as of February 29, 1984.

Cost of living adjustments based on consumer price indices.

Source: Fund staff estimates as of February 29, 1984.

Cost of living adjustments based on consumer price indices.

    Other Resources Citing This Publication