After a period of accelerating inflation in the early 1970s, the industrial countries in 1974–75 entered their most severe recession since the 1930s. The recession was brought on primarily by restrictive monetary and fiscal policies coupled with the dramatic rise in petroleum prices during 1973–74. Since early 1976 there has been some recovery, but unemployment continues to be a problem while inflation rates, although gradually moderating, remain at historically high levels. In this environment, governments have acted with caution in formulating their policies. In many of the industrial countries, monetary targets have been maintained at fairly modest levels, and fiscal policy, which was expansionary during 1975, reversed itself in 1976 and remains generally conservative. Over all, the desire to stimulate production has been tempered by concern over the inflationary consequences.
This paper addresses itself to this central policy dilemma. To this end, it presents a formal analysis of the effects of monetary and fiscal policy under flexible exchange rates, with the main emphasis on the division of the effects between output and prices. It focuses on the relatively small industrial country and treats only unilateral adoption of more expansionary policies by that country, disregarding foreign repercussions.1 Five alternative policies are contrasted: monetary expansion, a cut in income taxes, an increase in government expenditure, a balanced budget expansion, and, finally, an employment subsidy. 2 Constraining each of these policies, where feasible, to secure the same addition to output, we then ask which of the policies will have the smallest impact on the price level.
Surprisingly, despite the evident importance of this issue, it has received little attention in the literature.3 Early literature on policy effectiveness under floating exchange rates generally abstracted from price changes induced by these exchange rate changes. In the classic papers by Mundell 4 and Fleming,5 the effects of expansionary monetary policy are analyzed under the assumption of constant wages and prices. In their analyses, with capital perfectly mobile, expansionary monetary policy increases output and employment, while expansionary fiscal policy does not. 6 In neither case are the price effects of the respective exchange rate changes pursued. More recently, economists 7 have focused their attention on the dynamic process by which the economy moves toward its post-policy-change equilibrium. This literature distinguishes markets by the speed at which they adjust to a shock, and assumes that asset markets clear more quickly than do goods markets, thereby giving rise to the possibility of exchange rate “overshooting.” In one paper, Dornbusch 8 imposes this assumption on the Fleming-Mundell fixed price framework and traces the path of the exchange rate and nominal aggregate demand in response to a monetary expansion. In another 9 he allows prices to vary but constrains output to return ultimately to its initial full-employment level. In this context, it is not surprising that monetary policy is found to have no real effects in the long run.
Our approach to the study of policy effectiveness under floating rates is different. We have sacrificed the elegant dynamics of this recent work for a clearer perspective on the relative effects of the various macroeconomic policy measures on prices and output via their differential impact on the marginal productivity of labor and wage demands. Hence, in our analysis, the effect of a policy change depends very much on its direct and indirect impact on the labor market.
The formal model consists of an aggregate demand sector and an aggregate supply sector. In essence, the aggregate demand sector is an amalgam of the conventional IS and LM schedules after appropriate allowances are made for international substitution effects and appropriate deflators. On the other hand, since it is the aggregate supply sector of the economy that ultimately determines the division of effects between prices and output for given variations in nominal demand, we have paid more careful attention to this sector in constructing our macromodel. First, we allow taxes to play a role in the determination of wages. With the size of the government sector as well as the tax rate increasing in nearly all industrial countries, workers and trade unions have become conscious of the real value of their aftertax increases. This relationship has been stressed in a number of contributions.10 Second, we emphasize the fact that, in their respective employment decisions, the relevant price to the producer is the price of domestically produced goods, while the relevant price to workers is the overall price level, which is itself influenced directly by the exchange rate.11 Third, we modify the classical labor demand function to allow for the imposition of an employment subsidy. This serves to differentiate the wage received by the workers from the net wage paid by the employer. These three extensions mean that a tax cut will now directly affect wage demands and, hence, domestic prices,12 while an employment subsidy will directly affect the demand for labor. At the same time, a change in the exchange rate will directly influence nominal wage demands without directly affecting the demand for labor.
Whenever a static construct is used to model a dynamic process, the formulation of the market-clearing equations largely determines the time period for which the analysis is relevant. To take account of the fact that economies differ in the speed with which their labor markets adjust fully to shocks as well as in the magnitude of the wage adjustment to price changes, we allow for different degrees of “money illusion” to characterize the labor supply response. Moreover, because significant adjustments to nominal income are forthcoming over the time frame of interest to us, roughly 12–18 months, the theoretical foundations for any exchange rate overshooting are seriously mitigated. This allows us to make the simplifying assumption that the prevailing exchange rate equals its expected future value. We further assume a degree of financial integration sufficiently high that interest rate differences are eliminated by arbitrage.13 On the other hand, we do not constrain our solutions so as to satisfy purchasing power parity, which we believe to be only weakly operative at the level of aggregation with which we are dealing. Finally, we assume that an improvement in the country’s price competitiveness improves its trade balance, despite possible perverse J-curve effects in the very short run.
The results of the model may be conveniently summarized according to whether or not money illusion is assumed. If there is no money illusion, monetary policy affects only prices and equivalently the exchange rate but has no output effects. Expansionary fiscal policy increases output 14 and decreases prices (i.e., below their trend path), while a balanced budget expansion in government spending, interestingly, reduces output and raises prices (i.e., generates stagflation).15 This suggests that a contractionary expenditure shock can induce unemployment that only pure fiscal expansion or a balanced budget contraction in spending can reverse. Attempts at real expansion via monetary policy will be futile and, if repeated, will take on the appearance of a vicious circle.
These results can be explained with reference to the effects of the various policies on real wage demands vis-à-vis the marginal product of labor. With the interest rate fixed, monetary policy leaves them both unchanged. On the other hand, fiscal expansion opens up a divergence between the change in the foreign price level and the change in the domestic price level, permitting both the real wage rate to be maintained in terms of the overall price level and the real demand for labor to increase. For example, where fiscal expansion produces exchange rate appreciation, as well as a rise in the domestic price level, the relevant real wage rate to workers can stay constant even while the real wage rate to employers falls, the latter inducing some increase in production. In the balanced budget case, these tendencies toward expansion are more than offset by the contractionary influence of the increased tax rate on the nominal wage.
As the degree of money illusion increases, the importance of real considerations in labor’s supply response wanes, and the effect of a policy on output tends to approach its effect on nominal aggregate demand. As a consequence, the familiar Fleming-Mundell results about policy effectiveness under floating rates are increasingly replicated. Moreover, in this instance, for the same output effect, we can rank the price effects of the various policy injections. For an increase in government spending, a decrease in the tax rate, and imposition of an employment subsidy, the resulting consumer price index will be the same, and uniformly less than that resulting from expansionary monetary policy. This is somewhat counterintuitive because it means that, even though tax cuts and employment subsidies will both act directly to reduce domestic prices, the ultimate overall price effect, for any given expansion in real output, is no different from that of an increase in government expenditure. The three fiscal policies, however, will differ in their impact on the components of the consumer price index. While we cannot rank the subsidy’s price effects a priori, we can say unequivocally that the producer price index as well as the exchange rate will be higher from an increase in government spending than from a cut in taxes.
The organization of the paper is as follows. Section I outlines a model that, we think, captures the current situation confronting small industrial economies with a minimum of excess baggage. Discussion of our particular specifications and their appropriateness is included in this presentation. Section II shows how the model can be solved graphically by requiring the simultaneous clearing of the goods market, the money market, and the labor market. Using the same apparatus, Section III examines the effects of the alternative policies on prices and output, assuming that exchange rates are fully flexible. Section IV summarizes our results and briefly considers some limitations of the analysis of the main text. Appendix A provides a mathematical solution to the model, and Appendix B casts the model is terms of the “assignment problem.”
A. Mathematical Supplement
Setting initial values of s equal to zero and P, Pd, e, and Pf equal to one, we solve for the determinant of the coefficients of the endogenous variables.
The sign of D is positive because the parameters C’, t, M2, s, f1 and f2 all lie between zero and one. In addition, a is positive and b is negative. This can be verified with reference to footnote 17. The last two terms contain proxy expressions for the Marshall-Lerner condition. In the former, it appears as X’ – M1 + M and in the latter, as –X’+ M1, –f2X – f1 M. It is assumed that this condition holds, and that the former expression is negative and the latter positive.
Differentiating the system, we ascertain how output is affected by the various policies. BB is used to denote a balanced budget expansion.
In each case except for monetary policy and the balanced budget expansion, output increases with expansionary policy regardless of the degree of money illusion. Equation (20) indicates that if there is no money illusion, that is, n = 1, output is invariant to the money supply. Moreover, in this case the numerator of equation (23) reduces to
Hence, the “balanced budget multiplier” is negative.
In view of the Fleming-Mundell analysis of fiscal policy under floating exchange rates, it is especially interesting to see how the effectiveness of fiscal policy varies with the degree of money illusion. To this end, we differentiate equation (21) with respect to n. Simplifying a rather cumbersome expression, we get
whose sign is positive. Hence, as the degree of money illusion rises (n falls), the expansionary impact of fiscal policy diminishes. However, in a world of no money illusion, fiscal policy clearly has output effects.
We now present the effects on the domestic price of the various changes.
The exchange rate changes in accord with the following:
Now we proceed to calculate the impact of the various policies on the domestic price for a given output effect. As noted in the text, inspection of the money equation indicates the relative rankings of effects on the overall price index, so that algebraic manipulation is unnecessary. For the individual components, however, such a route is not open, so that explicit consideration of the algebra is required.
First, we note that the actual change in Y (ΔY) for a finite change in policy variable (ΔL, ΔG, Δt, Δs) can be calculated as follows:
Since we want to compare price effects for equal output effects, we set ΔYi = ΔYj, and arbitrarily set ΔG = 1. Now we can calculate the actual policy change required to increase output by the same amount as an increase of one unit in fiscal spending.
It remains only to rank the price effects given by equations (45) to (48). A convenient starting place is to consider the relative Δ Pds for complete money illusion. This is represented by setting N = 0. In this case, ΔPdL = ΔP.dG = ΔPdt. This is so because the AS schedule is invariant to price and tax considerations; hence, a given ΔY requires a common ΔPd. As the degree of money illusion is reduced, ΔPdL rises while ΔPdG and ΔPdt get smaller. Moreover, it can be shown that, except with complete money illusion, ΔPdt < ΔPdG.
This expression is negative in general, but equals zero if N = 0. Thus, our ranking is ΔPdL > ΔPdG > ΔPdt, for all cases except complete money illusion.
The AS schedules shift with the imposition of a subsidy even if labor has full money illusion. Hence, for high degrees of money illusion ΔPds is below the others. As money illusion decreases, and ΔPdt and ΔPdG become negative, it is possible that the always negative ΔPds will come to exceed one or both of them.
B. The Assignment Problem
In the light of our finding that monetary and fiscal policies have different inflationary effects for a given expansion in output, the question may now be asked whether this result implies anything about the assignment of our instruments to our two targets—output and inflation.
Figure 7 illustrates the assignment problem. The PP schedule combines the volume of money and tax rates in such a way as to maintain a constant overall price level.40 Since an expansion in money raises the overall price level while a reduction in the tax rate lowers the price level, the slope of PP must be negative. The YY schedule reflects the combinations of monetary and fiscal policies that keep real output constant. Its slope is positive, since an increase in money requires an increase in tax rates if output is to be maintained. In the extreme case where there is no money illusion, the slope will be infinite (YY will be vertical), since an infinitely large increase in the volume of money will be required to offset any reductions in output from an increase in the tax rate.
Following convention, four zones are identified corresponding to recession/inflation (stagflation), excess demand/inflation, excess demand/deflation, and recession/deflation. These are derived as follows. The area to the right of the PP schedule represents a situation of relatively high prices (inflation): with the volume of money given, tax rates are higher than required to ensure the short-term price target. The area to the left of the PP schedule represents relatively low prices (deflation): tax rates are below those required to meet the price target. The area to the right (left) of YY represents recession (excess demand), since, at a given volume of money, tax rates are now higher (lower) than those required to meet the short-term demand target.
In the conventional assignment literature, which is concerned with the appropriate assignment of the monetary and fiscal instruments to the targets of internal and external balance,41 the assumption is made that both surpluses and deficits in the balance of payments are undesirable and hence need to be corrected by an appropriate mix of policy instruments. It is not quite so evident, when the price level is the target, why the areas to the left of PP, where the price level is below “target,” should be considered undesirable. For obvious reasons, we feel it would not be meaningful to have a government authority manipulate one of its instruments so as to push the price level up toward its target. Hence, we assume in what follows that the instrument assigned to the price level is activated only when the price level is too high and, asymmetrically, not when it is too low. In any event, it is evidently only the area to the right of PP that is of interest to us.
Now consider the case where there is some money illusion, and assume that fiscal policy is assigned to the output target while the money supply is assigned to the price level. This particular assignment is shown in the form of broken arrows in each of the four zones. For example, in the zone labeled recession/inflation (stagflation), this would require that the volume of money be reduced (to reduce the inflation) at the same time that taxes are reduced (to stimulate output). It is evident from the diagram that this particular assignment is stable in the sense that, starting from any disequilibrium situation, either the point of intersection of PP and YY or some point on YY to the left of this point would ultimately be reached.
The alternative assignment, gearing the volume of money to output and the tax rate to the price level, is shown by the unbroken arrows. Now, in the face of the same stagflation conditions, the volume of money would be allowed to increase (to combat the unemployment) while the tax rate would be lowered to combat the inflation. It turns out, as the diagram clearly demonstrates, that this assignment is also stable. The conclusion, then, is that either assignment appears to be stable in terms of the model.42
The diagram shows that if an economy finds itself in the stagflation zone (the most relevant in today’s world), it would be possible to reduce the volume of money and lower the tax rate so as to maintain the level of output while lowering the overall price level, or, alternatively, to raise the volume of money and lower the tax rate so as to stimulate output while leaving the overall price level unchanged.
Mr. Argy, Consultant in the Fund’s Research Department when this paper was prepared, is Professor of Economics at Macquarie University, Sydney, Australia.
Ms. Salop, economist in the Special Studies Division of the Research Department, is a graduate of the University of Pennsylvania and Columbia University.
For an analysis of the large-country case, see Victor Argy and Joanne Salop, “Price and Output Effects of Monetary and Fiscal Policy in a Two-Country World Under Flexible Exchange Rates” (unpublished, International Monetary Fund, May 1979).
An employment subsidy, in a variety of forms, has recently been adopted by a large number of industrial countries. It has been viewed as a means of simultaneously increasing employment, reducing prices, and reducing the budget deficit. See John Burton, “Employment Subsidies—The Cases For and Against,” National Westminster Bank, Quarterly Review (February 1977), pp. 33–43.
Among the few contributions in the area are Robert A. Mundell, The Dollar and the Policy Mix: 1971, Essays in International Finance, No. 85, International Finance Section, Princeton University (May 1971); Thomas F. Dernburg, “The Macroeconomic Implications of Wage Retaliation Against Higher Taxation,” Staff Papers, Vol. 21 (November 1974), pp. 758–88; Francisco R. Casas, “Capital Mobility and Stabilization Policies under Flexible Exchange Rates: A Revised Analysis,” Southern Economic Journal, Vol. 43 (April 1977), pp. 1528–37; Rudiger Dornbusch and Paul Krugman, “Flexible Exchange Rates in the Short Run,” Brookings Papers on Economic Activity: 3 (1976), pp. 537–75. A very recent and relevant exception to the statement in the text is J. Sachs, “Wage Indexation, Flexible Exchange Rates, and Macroeconomic Policy.” (This is scheduled for publication in the Quarterly Journal of Economics.)
Robert A. Mundell, “Flexible Exchange Rates and Employment Policy,” Canadian Journal of Economics, Vol. 27 (November 1961), pp. 509–17.
J. Marcus Fleming, “Domestic Financial Policies Under Fixed and Under Floating Exchange Rates,” Staff Papers, Vol. 9 (November 1962), pp. 369–80.
It is evident that with the interest rate and the price level unchanged, monetary expansion must generate a proportionate effect on real output. At the other extreme, with prices, interest rates, and the money supply all fixed, fiscal expansion cannot change output—the associated revaluation fully offsetting the fiscal stimulus. These results are of interest because they replicate some monetarist propositions (including complete real crowding out for fiscal policy) with what is basically a Keynesian model.
Stanley Warren Black, International Money Markets and Flexible Exchange Rates, Princeton Studies in International Finance, No. 32, International Finance Section, Princeton University (1973); William H. Branson, “Asset Markets and Relative Prices in Exchange Rate Determination,” Sozialwissen schaftlich Annalen, Vol. 1 (1977); Rudiger Dornbusch, “The Theory of Flexible Exchange Rate Regimes and Macroeconomic Policy,” Scandinavian Journal of Economics, Vol. 78 (No. 2, 1976), pp. 255–75; Pentti J.K. Kouri, “The Exchange Rate and the Balance of Payments in the Short Run and in the Long Run: A Monetary Approach,” Scandinavian Journal of Economics, Vol. 78 (No. 2, 1976), pp. 280–304.
Rudiger Dornbusch, “Exchange Rate Expectations and Monetary Policy,” Journal of International Economics, Vol. 6 (August 1976), pp. 231–44.
Rudiger Dornbusch, “Expectations and Exchange Rate Dynamics,” Journal of Political Economy, Vol. 84 (December 1976), pp. 1161–76.
See Dernburg, op. cit.; Dudley Jackson, H.A. Turner, and Frank Wilkinson, Do Trade Unions Cause Inflation? Two Studies: With a Theoretical Introduction and a Policy Conclusion (Cambridge University Press, 1974); Alan S. Blinder, “Can Income Tax Increases Be Inflationary? An Expository Note,” National Tax Journal, Vol. 26 (June 1973), pp. 295–301; G. Brennan and D.A.L. Auld, “The Tax Cut as an Anti-Inflationary Measure,” Economic Record, Vol. 44 (December 1968), pp. 520–25.
This asymmetry is emphasized in Joanne Salop, “Devaluation and the Balance of Trade Under Flexible Wages,” in Trade, Stability, and Macroeconomics: Essays in Honor of Lloyd A. Metzler, ed. by George Horwich and Paul A. Samuelson (New York and London, 1974), pp. 129–51.
This is also true in the contribution by Dernburg, op. cit. Dernburg’s paper deals only with the closed economy case; moreover, he allows wages to be influenced by taxes but not by prices.
It is sometimes argued that the degree of capital market integration will itself be reduced by flexible exchange rates. It is not, however, clear what meaning should be attached to this. Is it that the response to the uncovered interest differential is weaker? Or to the covered differential? Or is it simply that the speculator’s schedule in the forward market is now less elastic?
The evidence on capital market integration remains somewhat inconclusive. While it is generally agreed that integration is very nearly perfect as between Eurocurrencies, it appears to be less than perfect as between national markets. See Eurocurrencies and the International Monetary System, ed. by Carl H. Stem, John H. Makin, and Dennis E. Logue, American Enterprise Institute for Public Policy Research (Washington, 1975).
In the two-country version, fiscal expansion raises output at home but reduces it abroad. Without money illusion, the system as a whole functions in the classical manner. Thus, total output cannot rise, but its distribution among countries can change. See Argy and Salop, op. cit.
For an argument along these lines applied to the United Kingdom, see Robert Bacon and Walter Eltis, “How Growth in Public Expenditure Has Contributed to Britain’s Difficulties,” Ch. 1 in The Dilemmas of Government Expenditure: Essays in Political Economy by Economists and Parliamentarians, Institute of Economic Affairs (Sussex, 1976), pp. 1–21.
The subsidy affects disposable income in the following way. As output increases above some initial value Y0, the value of the subsidy will equal the subsidy (s) times labor’s share in the increase in GNP. Because our production function is of the form
labor’s share equals ɑ2. The exponent b, to be used in equation (8), equals
Equation (8) can be derived from the profit-maximizing conditions for the competitive firm and the Cobb-Douglas production function Y = ɑ0Kɑ1Lɑ2 for which case,a = K ɑ0(ɑ2/ɑ1)2 and b = –ɑ2/ɑ1
For some econometric work supporting this hypothesis, see Michael Parkin, “The Causes of Inflation: Recent Contributions and Current Controversies,” in Current Economic Problems: The Proceedings of the Association of University Teachers of Economics, Manchester 1974, ed. by Michael Parkin and A.R. Nobay (Cambridge University Press, 1975), pp. 243–74, especially pp. 254–55; Robert J. Gordon, “Recent Developments in the Theory of Inflation and Unemployment,” Journal of Monetary Economics, Vol. 2 (April 1976), pp. 185–219, especially p. 212.
Rewriting the numerator, assuming that the average and marginal propensities to consume are equal, we have
The coefficient of C’ is less than one, as is From equation (1) we know that
This implies that the numerator is less than X’ –M1, + M. The negative of this expression is a proxy for the Marshall-Lerner condition. Following convention in assuming this to hold, we find that the numerator is less than some negative quantity.
To minimize the amount of graphical presentation, we show only the case where the slope of LM is steeper than the slope of IS, except where the reverse situation is critical to the argument.
As drawn, the IS schedule shifts to the right from the combined effects of the increase in government spending and the revaluation. It is less likely, but conceivable, that it will shift to the left. In that event, Y rises and Pd f falls.
This entails setting N = 0.
This follows directly from the money market condition.
In this case,
With sticky prices, Keynesian unemployment would result even if real wages were rigid. Essentially, the demand component of the shock would cause producers to be “demand constrained” in their production and sales decisions. Hence, they would employ fewer workers than their marginal productivity conditions would admit. With diminishing marginal productivity, labor’s marginal revenue product will exceed the wage. With both prices and real wages sticky, unemployment occurs on two counts—Keynesian unemployment and rigid real wage unemployment.
A decrease in real wage demands shifts the AS schedule to the right. The exchange rate appreciates if LM is steeper than IS and depreciates if IS is steeper. In either case, output rises and unemployment falls.
Note that expansionary remedies can eliminate the portion of unemployment that is Keynesian in origin. By relaxing the demand constraint, such remedies allow producers to expand and to hire more workers. The marginal revenue product falls as this happens, until it equals the wage. Once this point is reached, however, further expansion goes into prices.
If financed by money creation, then a vicious circle will ensue at a higher level of output. In the first period, the monetary and fiscal effects on the exchange rate tend to offset each other. In subsequent periods, however, the money created would still be in circulation and unmatched by additional fiscal stimulus.
Because the AS schedule is stationary, the change in Pd for a given change in Y is invariant to the source of the expansionary impulse.
Since neither fiscal policy alters the money supply, for the same output, the consumer price index must be the same to satisfy LM.
See Burton, op. cit. See also Towards Full Employment and Price Stability: A Report to the OECD by a Group of Independent Experts, Organization for Economic Cooperation and Development (Paris, June 1977). While this Report says it is “attracted” by such schemes, it warns against the administrative complexities and possible “displacement” effects. (See paragraph 349 of the Report.)
Since the employment subsidy has taken many different forms in different countries, these objections clearly do not apply in equal degree to all variations of this subsidy. For a good discussion, see George F. Kopits, “Wage Subsidies and Employment: An Analysis of the French Experience,” Staff Papers, Vol. 25 (September 1978), pp. 494–527.
As far as the assignment issue is concerned, there is no difference between the three fiscal policies, since we have shown that they have the same overall price effect for any given expansion in output.
See Robert A. Mundell, “The Appropriate Use of Monetary and Fiscal Policy for Internal and External Stability,” Staff Papers, Vol. 9 (March 1962), pp. 70–79; Jay H. Levin, “International Capital Mobility and the Assignment Problem,” Oxford Economic Papers, Vol. 24 (March 1972), pp. 54–67.
As the degree of money illusion decreases, however, this assignment becomes increasingly inappropriate. At the limit where YY is vertical, the assignment is clearly less efficient than the reverse assignment.