Abstract

Karl Habermeier with Steven Symansky1

Karl Habermeier with Steven Symansky1

Following German unification, the ratios of public revenue to GDP and expenditure to GDP rose sharply (Table 4-1). Further increases in expenditure are taking place in 1994—96 as the federal government begins to service the accumulated debt from currency conversion and privatization in east Germany. Expenditure on social benefits, notably on long-term care, pensions, and health, are also likely to increase faster than GDP in the years to come. In order to reduce the general government deficit to less than 2 percent of GDP as planned, the Government has already announced that it will continue to raise taxes, notably by reimposing a 7 1/2 percent surcharge on wage and income taxes from the beginning of 1995. Contribution rates to the social security funds will also continue to increase gradually from their already high level.

Table 4-1.

General Government Finances1

(In percent of GDP)

article image
Sources: Federal Ministry of Finance: and IMF staff calculations.

Ratios for 1989 as a percentage of west German GDP; ratios for 1990-94 as a percentage of united German GDP.

The composition of fiscal measures affects growth and employment by affecting the incentives of households and businesses to save, invest, or work. Economic theory suggests that measures to raise revenue will discourage the activity being taxed, though taxes on consumption are less likely to discourage saving, investment, and work than are taxes on income and wealth. Similarly, most types of government spending, when carried beyond a certain point, are likely to dampen economic growth. In particular, generous welfare and unemployment compensation programs tend to reduce work effort and job search.

Empirical estimates of the effects of various fiscal policy variables on real activity in west Germany in 1960—90 show that lower fiscal deficits have been associated with faster growth. Furthermore, higher government expenditure and revenue ratios—that is, a larger public sector—have, on average, been associated with slower economic growth. A more disaggregated treatment supports the view that higher direct taxes have a more negative effect on growth than other means of raising revenue. Similarly, social transfer payments appear to be more detrimental to economic performance than subsidies to enterprises or public investment.

The effects of fiscal restructuring on macro-economic performance are also assessed using simulations performed with the IMF’s multi-country macroeconomic model (MULTIMOD).2 Following a brief exposition of the channels through which fiscal policy operates in MULTIMOD, the consequences of an equal reduction in the ratios of both public revenue to GDP and public expenditure to GDP are considered. There is also a comparison of the macroeconomic effects of reducing taxes.

The results suggest that fiscal policy in Germany should aim to rectify the imbalance between revenue and expenditure measures in pursuing fiscal consolidation. Both the revenue and the expenditure ratios are already higher than those that would be compatible with robust medium-term economic growth, indicating a need to reduce the size of the public sector. The Government has recognized the need for changes in the structure of public finances. As currently enunciated, the strategy in this area emphasizes strict expenditure restraint accompanied by eventual reductions in taxes, as well as measures to reduce disincentives connected with the structure of the tax system.

Review of the Literature

There is a large literature on the relationship between public finances and other macroeconomic variables such as economic activity and inflation. One branch of this literature emphasizes the short-run stimulative effects, through higher effective demand, of increased public expenditure or lower taxation on economic activity. In contrast to this “Keynesian” tradition, another branch examines the effects of public revenue and expenditure on aggregate supply and longer-run growth. These “neoclassical” studies focus on how public sector activity affects the incentives facing households and businesses.3

Research on the longer-term consequences of public sector activity on economic growth have generally supported the hypothesis that increases in the size of the public sector, or higher tax rates, adversely affect economic activity. These results appear to hold in both time-series and cross-sectional studies.

Cebula and Scott (1992), using data for the United States for the 1957-84 period, show that a larger government deficit exerts a significant negative influence on economic growth. They also show that higher tax rates are associated with slower economic growth. On the basis of data for the United States for 1929-86. Peden (1991) finds that government expenditure, when it exceeds a certain percentage of GNP, reduces economic growth. Fard-manesh (1991), in a cross-sectional study of 21 developed countries over the period 1972-81, obtains the result that foreign trade taxes have the most adverse effects on growth, followed by income taxes and domestic excises. He also finds that cuts in current expenditure have no lasting effect on growth. Ahsan and others (1989), using the Granger bivariate model, find that in a cross-sectional sample of 24 industrialized countries, there is considerable support for bidirectional causality between government consumption expenditure and national income. Somewhat mixed results on the Australian experience are reported by Grossman (1988), with government expenditure showing a positive effect on growth but taxation a negative effect. Marsden (1983) argues, on the basis of data for a large cross-section of industrial and developing countries, that there is a pronounced negative correlation between taxation and growth.4

The results obtained by two other studies are somewhat more adverse to the hypothesis that higher taxation has an adverse effect on growth. Using a large cross-sectional sample of 63 countries, Koester and Kormendi (1989) find no negative relationship between marginal tax rates and economic growth when the relationship is controlled for the effects of per capita income on the growth rate. Garrison and Lee (1992), who extend the time period covered by Koester and Kormendi’s data set, find no clear negative relationship between marginal tax rates and growth over the longer period. However, the time periods covered by both of these studies are far too short (1970-79 in the first study and 1970-84 in the second): moreover, a panel data approach would have been a more appropriate method of analysis than simply averaging the data over the sample period.

Results for Germany

There are a number of pitfalls in investigating the effects on economic growth of revenue and expenditure policy and the size of the public sector. First, economic activity is affected by many other factors besides fiscal policy. Monetary conditions, the international economic environment, the pace of technological innovation, and the age and skill structure of the population, as well as more general cultural and political factors, also influence the long-term rate of economic expansion. Most notably, the slowdown in economic growth in the 1970s is often attributed to the sharp increases in petroleum prices that occurred in 1973-74 and 1979.

Second, it is not obvious that causality runs from higher revenue and expenditure ratios to lower growth rather than the other way around. Slower economic growth may lead to calls for more government spending; the desire to contain the fiscal deficit may then lead to a higher revenue ratio as well.

It is clear that these issues cannot be definitively disentangled using econometric methods, all the more so as the growth rate of real GDP from year to year is erratic and hard to explain with any precision.5

Chart 4-1 summarizes the basic features of economic growth and public sector activity in west Germany over the sample period, which spans 1960 to 1990. A period of fast growth in the 1960s, during which the size of the public sector was relatively limited, was followed by a sharply growing public sector and considerably slower growth in the 1970s and early 1980s. These trends were reversed during the 1980s, when both the revenue and the expenditure ratios decreased and growth accelerated.

Chart 4-1.
Chart 4-1.

West Germany: Growth and Fiscal Indicators

Sources: Statistiches Bundesamt: and IMF staff calculations.1Series were smoothed using the Hodrtck-Prescott Filter (c = 100).

Simple regressions of the growth rate of real GDP and the ratios of overall revenue and expenditure to GDP, as well as of several subcategories of revenue and expenditure variables, are reported in Table 4-2. These estimates confirm the broad impression given by Chart 4-1 of a negative correlation between the scope of public sector activity and economic growth.

Table 4-2.

Effect of Fiscal Variables on Economic Growth1

(Dependent variable: DLQ)

article image
Sources: Federal Statistical Office: and IMF saff calculations.

DLQ = logarithmic difference of real GDP, west Germany. Ratios to GDP (in percent):

RFB: general government financial balance;

RR: general government revenue;

REP: primary expenditure;

REPU: primary expenditurejess social transfers;

RT: tax revenue;

RTI: indirect tax revenue;

RTD: direct tax revenue;

RTS: social security contributions;

RE: general government expenditure;

RET: transfer expenditure;

RETS: social transfer expenditure;

REC: government consumption expenditure;

RES: subsidy expenditure;

REI: investment expenditure.

In order to reduce the influence of cyclical disturbances, the regressions were also modified by including a cyclical indicator that measures the percentage deviation of real GDP from a quadratic trend.6 The results, which are presented in Table 4-3, are similar to those obtained by the simple regression. The causality tests reported below, which include a more general lag structure, represent a further approach to eliminating the influence of the cycle on the results.

Table 4-3.

Effect of Fiscal Variables on Economic Growth, with Cyclical Indicator1

(Dependent variable: DLQ)

article image
Sources: Federal Statistical Office; and IMF staff calculations.

DLQ = logarithmic difference of real GDP, west Germany.

Ratios to GDP, in percent

RFB: general government financial balance;

RR: general government revenue;

REP: primary expenditure ;

REPU: primary expenditure, less social transfers;

RT: tax revenue;

RTI: indirect tax revenue;

RTD: direct tax revenue;

RTS: social security contributions;

RE: general government expenditure;

RET: transfer expenditure;

RETS: social transfer expenditure;

REC: government consumption expenditure;

RES: subsidy expenditure;

REI: investment expenditure.

On the whole, the coefficients reported in Tables 4-2 and 4-3 have the expected sign.7 However, the coefficients on tax revenues, subsidies, and investment expenditure are not significant at the 5 percent level. Also, several of the equations show evidence of slight positive serial correlation in the residuals. When these equations are “corrected” for the serial correlation using the Cochrane-Orcutt technique, the signs and magnitudes of most of the coefficients remain largely unchanged: furthermore, the estimated coefficients of serial correlation are smaller than 0.15 in many cases. Similarly, when a lagged value of the dependent variable is included in the equations, its coefficient is in no case significantly different from zero, and the coefficients on the fiscal variables are not materially affected, although their estimated standard errors increase somewhat.

These results suggest that the expansion of the public sector, which reached its peak in the early 1980s, was associated with slower economic growth and that growth picked up again in the 1980s as the size of the public sector decreased. Moreover, the increasing deficits seen in the 1970s were associated with a deceleration in economic activity.

Although the relative magnitude of the estimated coefficients is subject to great uncertainty, the estimates suggest that among revenue instruments, direct taxes are most harmful to growth and social security contributions are less harmful: indirect taxes appear to exert a positive influence, perhaps by reducing the attractiveness of consumption relative to saving. On the expenditure side, subsidies and public investment expenditure seem to have little measurable effect on growth rates, while public consumption has a strong negative impact. Consistent with the view that incentives matter, social transfers have the largest negative effect of any expenditure category.

Looking to the future, a straightforward application of these results suggests that the sharp expansion of the public sector since unification may considerably reduce the rate of economic growth. Specifically, an increase in the ratio of noninterest expenditures or overall revenues by 1 percentage point will lead to a reduction in the growth rate of about 0.2 percentage point. If the relationship were truly linear, the almost 10 percentage point increase in the size of the public sector that is expected between 1989 and 1995 should reduce the long-term rate of economic growth by about 2 percentage points. A much more conservative estimate was obtained by subtracting two standard errors from the absolute value of the coefficients; in this case, the rate of economic growth would be reduced by between 1/4 and 1 percent annually.

Tests of “causation” were also performed to obtain a clearer sense of the dynamic relationships among real GNP and fiscal variables and to examine whether other factors, such as the real price of oil, played an important role in explaining the behavior of the growth rate. Using the Geweke variant of the Granger bivariate model (Table 4-4), it was shown that price-adjusted general government revenues both caused and were caused by real output but that there was no causation among price-adjusted general government expenditures and real output.8 However, once general government expenditures were adjusted for interest expenditure (which is related to the debt dynamics) and for social transfers (which are cyclical to a considerable extent), the remaining “structural” expenditures cause real output but are not caused by it. Finally, it was found that the real price of oil (in deutsche mark terms) neither causes nor is caused by real output.9

Table 4-4.

Geweke Tests for Causality1

article image
Sources: Federal Statistical Office; and IMF staff calculations.

The critical values for refection of the null hypothesis of “no causality” are 2.74 at the 5 percent level and 4.17 at the 1 percent level.

Expenditure excluding interest and social transfers.

Thus, the power of oil prices to explain movements in real output in Germany appears to be smaller than commonly assumed, while the relationships between fiscal variables and economic activity are considerably more pronounced. Although no firm conclusions can be drawn on the basis of such a narrow investigation, the results lend some support to the hypothesis put forward by King and Rebelo that differences in institutions and policies are a primary cause of variations in growth performance over time and across countries.

Results from an International Cross-Section

Additional support for the results presented above was obtained by estimating the effect of the ratios of current revenue and current expenditure to GDP on the growth rate of real GDP for a sample of industrialized countries covering 1964 to 1991.10 As shown in Table 4-5, the results are generally consistent with the findings for Germany: an improvement in the financial balance is associated with faster growth, while higher revenue and expenditure ratios are related to slower growth.11

Table 4-5.

Effect of Fiscal Variables on Growth in Selected Industrial Countries1

article image
Sources: OECD Economic Outlook; and IMF staff calculations.

Estimated effect of 1 percentage point increase in fiscal variable on the growth rate.

Not significant at 5 percent level.

In order to reduce the influence of the international transmission of the business cycle on the results, the data for the individual countries were aggregated using the weights from the IMF’s World Economic Outlook exercise.12 The aggregated data are shown in Chart 4-2. Again, the increase in the size of the public sector that took place from the 1960s to the early 1980s was associated with slowing growth. Unlike Germany, however, there was no reduction in the share of the public sector in the 1980s and no revival of economic growth.

Chart 4-2.
Chart 4-2.

Major Industrial Countries: Growth and Fiscal Indicators

Source: Statistiches Bundesamt and IMF staff calculations.1Series were smoothed using the Hodrtck-Prescott Filter (c = 100).

These observations are confirmed by econometric estimates. Table 4-6 reports the results obtained by regressing the growth rate of real GDP on aggregate current revenue, aggregate current expenditure, and public saving ratios. Again, the signs and magnitudes of the elasticities are consistent with the German experience.

Table 4-6.

Major Industrial Countries: Effect of Fiscal Variables on Growth1

(Dependent variable: DLQ)

article image
Sources: Federal Statistical Office; and IMF staff calculations.

DLQ = logarithmic difference of real GDP, aggregated. Ratios to GDP (in percent):

RPS: general government saving;

RRC: general government current revenue;

REC: general government current expendiuire;

Finally, the pattern of “causation” among real GDP and three other variables—real government revenue, real government expenditure, and the real oil price—was examined using the Geweke test. The results are summarized in Table 4-7. Output both “causes” revenue and is caused by it; in addition, the strength of the effect of revenues on output is significantly greater than the other way around. A similar pattern emerges for government expenditure. Finally, oil prices cause output but not vice versa; thus, the effect of oil prices appears to be more pronounced for the industrial countries as a group than for an individual country such as Germany.

Table 4-7.

Major Industrial Countries: Geweke Tests for Causality1

article image
Sources: Federal Statistical Office; and IMF staff calculations.

The critical values for rejection of the null hypothesis of “no causality” are 2.81 at the 5 percent level and 4.34 at the 1 percent level.

Incorporating Distortionary Taxes and Spending into MULTIMOD13

This section outlines the effects of tax and expenditure policies on consumption, labor and capital income, and the government budget constraint, the main channels through which taxes and spending affect economic behavior. Based on these considerations, Bartolini, Razin, and Symansky (1994) added several new equations to MULTIMOD and modified a number of others.14

The labor market segment of the revised MULTI-MOD consists of three behavioral equations governing price setting, wage setting, and unemployment. The equations describe cyclical fluctuations of unemployment around its natural level, which is taken as largely exogenous and is regarded as the long-run outcome of a search-bargain framework, such as that of Pissarides (1985). Although the framework is not specified as a full-fledged bargaining model, it preserves the intuition that there is a positive link between equilibrium unemployment and a broadly defined labor income tax, a result that is also well supported empirically.15 The model, however, does not incorporate the effect of unemployment benefits on equilibrium unemployment.

The wage-setting equation fixes the long-run growth of real consumption wages as a function of average productivity growth and other structural factors (such as relative bargaining power and the target real wage), which are absorbed into the constant term. Taxes on goods and services, such as the value-added tax (VAT), enter this equation through their effect on the consumer price level. The unemployment equation combines labor demand and labor supply elements, which in turn reflect the effects of taxes on labor income and consumption, with an element that captures the effect of taxation on the natural rate of unemployment. Finally, the price-setting equation is of a standard form and assumes that prices are a markup on unit labor costs.

In addition to affecting the supply side of the economy through their effect on real wages and unemployment, taxes and government expenditure also influence consumption and investment decisions, the dynamics of government debt, and various national income and price deflator identities. The effect of VAT on the consumer price index has already been mentioned. The intertemporal budget constraints of private households, business enterprises, and government were each modified to reflect the effects of the three categories of taxes that are distinguished in the model.

Balanced Reduction in Revenue and Expenditure

The model as modified by Bartolini, Razin, and Symansky was used to examine the macroeconomic effects of equal reductions in the ratios of revenue and expenditure to GDP. It was assumed that the revenue ratio would be reduced by 1 percentage point each year over a four-year period beginning in 1996, with the reductions equally divided between wage and consumption taxes. This change was matched by an equivalent cut in expenditure, also spread over four years. Not taking into account the feedback effect of these changes on macroeconomic variables, these actions would imply an unchanged general government deficit.

The principal results are summarized in Table 4-8, which gives deviations from the baseline projection. All in all, the medium-term effects of a balanced reduction in revenue and expenditure are highly favorable, with real GDP some 1 1/4 percent above the baseline level in the year 2000, the GNP deflator about 1 percent lower than the baseline, and the unemployment rate 1 percentage point lower.

Table 4-8.

Macroeconomic Effects of Balanced Revenue and Expenditure Reduction

(Percentage deviation from baseline levels, unless otherwise noted)

article image
Source: IMF staff calculations.

Percentage point deviation from baseline level.

Billions of deutsche mark.

Much of the positive impact of this restructuring on GDP can be attributed to the stimulation of domestic demand, particularly consumption and investment. Nonetheless, the beneficial spillover effects on other countries and the real depreciation of the deutsche mark, partly induced by lower domestic-interest rates, contribute to an increase in net exports and an improvement in the current account balance by about 1 percent of GDP.

Tax Restructuring

This section examines the macroeconomic effects of a revenue-neutral change in the tax system, one in which a reduction in one category of taxation is offset by an increase of equivalent size in another category of taxation. This type of policy change, which may also be called tax substitution, is of interest because it allows efficiency gains to be captured without increasing the budget deficit.

Table 4-9 illustrates the effect in MULTIMOD of reducing taxes on labor income by 2 percent of GDP, compensated by an equivalent increase in taxes on goods and services. The initial impact of this policy change on real GDP is negative but small. In the longer run, the effect is positive. This reflects the negative influence on consumption of higher taxes on goods and services; in conjunction with the dampening influence on wages of lower taxes on labor income, this enhances the incentives for investment. The higher level of GDP, and the lower level of wage taxes, is also reflected in a lower unemployment rate. As expected, the effect on prices is positive, but this is later reversed by the reduction in wages relative to the baseline.

Table 4-9.

Macroeconomic Effects of Balanced Revenue and Expenditure Reduction1

(Percentage deviation from baseline levels, unless otherwise noted)

article image
Source: IMF staff calculations.

Reduction in one category of taxation (by 2 percent of GDP in 1997) offset by an equivalent increase in another category of taxation.

Percentage point deviation from baseline level.

By the equivalent of 1 percent of GDP in 1997.

For the purposes of illustrating the effects in MULTIMOD of changes in tax policy, simulations were run in which each of the three categories of taxes in the model was reduced, other things equal, by the equivalent of 2 percent of GDP. The results of this experiment need to be interpreted with caution, as the simulations are invariant to the level of the deficit in the baseline. In reality, it is likely that the effect of a tax reduction depends not only on the size of the cut itself but on the credibility and sustainabil-ity of fiscal policy before the cut. With this caveat, the conclusion is that reductions in tax rates lead to increases in real GDP and lower the unemployment rate. Interestingly, an alleviation of capital taxation produces the largest positive effect on real GDP in the long run, possibly suggesting that further efforts to streamline and reduce taxes on business capital may be desirable.

References

  • Adams, Charles and David T. Coe, “A Systems Approach to Estimating the Natural Rate of Unemployment and Potential Output for the United States,” Staff Papers, International Monetary Fund, Vol. 37 (1990). pp. 232 –93.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ahsan, Syed, and others, “Causality Between Government Consumption Expenditure and National Income,” Public Finance, No. 2 (1989).

  • Barro, Robert, “Government Spending in a Simple Model of Endogenous Growth,” Journal of Political Economy (October 1990).

  • Banolini, L., A. Razin, and S. Symansky, “G-7 Fiscal Restructuring in the 1990s: Macroeconomic Effects” (unpublished; Washington: International Monetary Fund, 1994).

    • Search Google Scholar
    • Export Citation
  • Boskin, Michael, “Tax Policy and Economic Growth: Lessons from the 1980s,” Journal of Economic Perspectives (Fall 1988), pp. 71 –97.

    • Search Google Scholar
    • Export Citation
  • Cebula, Richard, and Gerald Scott, “Fiscal Policies and Growth: An Extension,” Rivista Internazionale di Scienze Econorniche e Commerciali (January 1992). pp. 91 –94.

    • Search Google Scholar
    • Export Citation
  • Coe, David T., and Thomas Krueger, “Why Is Unemployment So High at Full Capacity? The Persistence of Unemployment, the Natural Rate, and Potential Output in the Federal Republic of Germany,” in German Unification: Economic Issues, IMF Occasional Paper 75 (Washington: International Monetary Fund, 1990).

    • Search Google Scholar
    • Export Citation
  • Courant, Paul, “Fiscal Policy and European Economic Growth,” in Barriers to European Economic Growth: A Transatlantic View (Washington: Brookings Institution, 1987).

    • Search Google Scholar
    • Export Citation
  • Fardmanesh, Mohsin, “Economic Growth and Alternative Deficit-Reducing Tax Increases and Expenditure Cuts: A Cross-Sectional Study.” Public Choice (February 1991).

    • Search Google Scholar
    • Export Citation
  • Garrison, Charles, and Feng-Yao Lee, “Taxation, Aggregate Activity and Economic Growth.” Economic Inquiry (January 1992).

  • Grossman, Philip, “Growth in Government and Economic Growth: The Australian Experience,” Australian Economic Papers (June 1988).

  • Hoffman, Philip, “Land Rents and Agricultural Productivity: The Paris Basin, 1450-1789,” Journal of Economic History (December 1991).

    • Search Google Scholar
    • Export Citation
  • King, Robert, and Sergio Rebelo, “Public Policy and Economic Growth: Developing Neoclassical Implications,” Journal of Political Economy (October 1990).

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Koester, Reinhard, and Roger Kormendi, “Taxation, Aggregate Activity, and Economic Growth: Cross-Country Evidence on Some Supply-Side Hypotheses,” Economic Inquiry (July 1989).

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Landau, Ralph, and Dale Jorgenson, Technology and Economic Policy (New York: Ballinger, 1986).

  • Lockwood, B., and A. Manning, “Wage Setting and the Tax System: Theory and Evidence for the United Kingdom,” Journal of Public Economics, Vol. 52 (1993), pp. 1 –29.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Marsden, Keith, “Taxes and Growth,” Finance and Development (September 1983).

  • Masson, Paul, Steven Symanksy, and S. Meredith, MULTIMOD Mark II: A Revised and Extended Model, IMF Occasional Paper 71 (Washington: International Monetary Fund, 1990).

    • Search Google Scholar
    • Export Citation
  • Peden, Edgar, “Productivity in the United States and Its Relationship to Government Activity: An Analysis of 57 Years, 1929-1986.” Public Choice (February 1991).

    • Search Google Scholar
    • Export Citation
  • Pissarides, Christopher, “Taxes, Subsidies, and Equilibrium Unemployment,” Review of Economic Studies, Vol. 52 (1985), pp. 121 –33.

1

The MULTIMOD simulations in this section were provided by Sieve Symansky.

2

Recent work by Banolini, Razin, and Symansky (1994) has extended MULTIMOD 10 capture the incentive effects of changes in tax and expenditure policy. See also Barro (1990).

3

The rationale for the neoclassical approach is best summarized in a recent paper by King and Rebelo (1990). Their argument is that differences in the long-term growth rate of individual countries are explained by variations in the national policies that affect the incentives individuals have to accumulate physical and human capital. Courant (1987) and Boskin (1988) also discuss the interaction between fiscal policy and supply-side issues. Landau and Jorgenson (1986) examine the effects of tax policy on the incentive to innovate. Finally, see Barro (1990).

4

Looking at the effect of taxation on growth over a period of centuries, a careful historical study by Hoffman (1991) argues that sharp increases in taxation in the early seventeenth century impeded the growth of agricultural productivity, preventing a recovery from the war damage of the previous century.

5

Augmenied Dickey-Fuller tests on German real GDP in log levels fail to reject the hypothesis that the series is at least /(1). Intuitively, this means that real GDP behaves much like a random walk.

6

The inclusion of the cyclical indicator is intended to capture those effects on the growth rale that are purely transitory. An alternative approach is to smooth the growth rate of real GDP and the fiscal variables using the Hodrick-Prescolt filter. This too had little effect on the sign and magnitude of the estimated coefficients.

7

There is only one exception: the coefficient on indirect taxes is positive and significant at the 1 percent level.

8

The tests were conducted using three lags and two leads.

9

The tests reponed here were performed on log-level data; essentially the same results were obtained when log differences were used for real output and ratios to GNP were used for the fiscal variables.

10

The sample comprises Austria, Belgium, Canada. Finland. France, Germany, Italy, Japan. Norway, Sweden, the United Kingdom, and the United States. These countries account for more than half of world output for most of the sample period.

11

The magnitude of the coefficients is little affected by including cyclical indicators in the regressions, or by first smoothing the underlying data using the Hodrick-Prescott filler.

12

These weights are based on estimated purchasing power parity.

13

The model simulations were conducted by Steve Symansky.

14

A complete description of MULTIMOD may be found in Masson, Symansky, and Meredith (1990).

15

See, for instance, Adams and Coe (1990), Coe and Krueger (1990), and Lock wood and Manning (1993).

The First Five Years: Performance and Policy Issues