Chapter

CHAPTER 1 Choice of Exchange Rate Regime for a Smaller Economy: A Survey of Some Key Issues

Author(s):
International Monetary Fund
Published Date:
September 1990
Share
  • ShareShare
Show Summary Details
Author(s)
Victor Argy*

Introduction

Since the collapse of Bretton Woods in early 1973 a smaller country has had the option of adopting a potentially wide variety of exchange rate policies. A first option is to adopt an independent adjustable peg regime. In this regime a country pegs to a single currency or basket of currencies, but allows the peg to adjust upward or downward periodically (relatively infrequently, however, as in the old International Monetary Fund system). Australia (to 1976), New Zealand (1973–79, 1982–85), Sweden (since 1971), Norway (since 1979), Finland, and Iceland have for varying periods adopted the adjustable peg regime. The Nordic countries continue to use it.

A second option, which is a variation on the adjustable peg, is the crawling peg. In contrast to the adjustable peg, exchange rate adjustments are much smaller but are made more frequently (see Williamson, 1981). Two varieties of this regime have been used in member countries of the Organization for Economic Cooperation and Development (OECD): the purchasing power parity (PPP) variation and the discretionary variation. Between mid-1979 and mid-1982 New Zealand adopted a PPP variety of the crawling peg. The objective was to maintain competitiveness by allowing the currency to crawl upward or downward to offset the difference between the domestic rate of inflation and that of trading partners.

By contrast, Australia adopted a discretionary variation of the crawling peg regime between November 1976 and March 1983. The so-called four wise men (ultimately responsible to the Treasury) adjusted the Australian dollar daily. The basis for the crawl varied. Up until early 1981 the principal determinant was the overall balance of payments; after that increased attention was paid to the current account.

A third exchange rate option is to peg independently, on an indefinite basis, either to a major currency or to a basket of currencies. There are no examples of the latter, but in the former category we have Ireland, which was pegged to sterling until 1979, and Austria, whose currency has been effectively pegged to the deutsche mark since 1976.

A fourth option, open to European economies since 1972, has been to join in some collective exchange rate arrangements aimed at stabilizing the bilateral exchange rates of participating countries. Between 1972 and 1978 there were the “snake” arrangements; from March 1979 the European Monetary System (EMS) has been in operation.

The adjustable peg is distinct from the EMS arrangements in two ways. Under the EMS, the decision to adjust the peg is made jointly rather than unilaterally (as in the adjustable peg). Also the size of the adjustment is potentially smaller in the EMS case. This is partly because there is greater pressure to conform, but also because exchange rate changes are not allowed to offset fully any “excess” inflation. Since the inception of the EMS there have been 11 realignments, nearly all involving relatively small adjustments.

A fifth option is to adopt a dual exchange rate. In this scheme the authorities regulate the exchange rate for commercial transactions but allow the rate for capital transactions to float more or less freely. Thus there are two exchange rate quotations at any time: one for commercial and one for capital transactions.1 Belgium has had a dual exchange rate regime since 1955; France had a dual regime between August 1971 and March 1974; and Italy had one between January 1973 and March 1974.

As a final, sixth exchange rate option, a smaller economy may allow its currency to float, but with different degrees of exchange rate management. This can range from a pure float to “loose” exchange rate targets (or target zones) to a variety of rules for exchange rate management (for example, leaning against the wind). The United Kingdom (since 1972), Canada (since 1970), and Switzerland (since 1973) have had floating rates. Australia has floated since December 1983, and New Zealand since March 1985.

There has been a notable evolution in most of the countries toward increased exchange rate management. This is particularly striking in the United Kingdom where, since 1982, growing attention has been paid to loose exchange rate targets. The evolution is also evident in Australia and in Canada.

This paper reviews the key considerations that enter into the choice of an exchange rate regime by a smaller economy. The literature on the topic is considerable and it is not possible in a relatively short survey to discuss all or even most of the issues raised in this literature (see Goldstein, 1980, 1984; and Dornbusch and Frankel, 1989). I propose, therefore, to deal somewhat more extensively with three key considerations that bear, most importantly, on the choice of regime. These are:

  • (1) Inflation discipline, that is, the equilibrium rate of inflation consistent with each exchange rate regime.

  • (2) Insulation, that is, how each regime insulates the economy from disturbances, of domestic or foreign origin.

  • (3) Policy effectiveness, that is, the capacity of each regime to exploit the policy instruments available to achieve the key macroeconomic targets of policy.

I. Inflation Discipline

The key question addressed in this section is which exchange rate regime will best succeed in securing a “steady state” rate of inflation closest to the “optimal” rate.

There are two parts to the question. The first addresses the “equilibrium” rate of inflation to which an exchange rate regime will converge in “steady state,” that is, in the absence of all stochastic disturbances. (The next two sections deal with stochastic disturbances and policy reactions.) The second part addresses the optimal rate of inflation.

The analysis is essentially in three parts. First, I lay down general conditions concerning long-run monetary and fiscal policies that would be expected to hold, independently of the exchange rate regime. Second, I look at how steady-state rates of inflation are arrived at in three regimes: the permanent peg, the adjustable peg, and the float (the case of the adjustable peg is discussed in the Annex). Third, I examine the considerations that enter into the determination of the optimal rate of inflation.

Consistent Monetary-Fiscal Policies in Steady State

I draw here on the analysis first developed by Sargent and Wallace (1981). Their key argument is that it is not possible to set money growth policy and deficit policy independently of one another.

Notation

  • B = nominal stock of government debt

  • b = stock of debt (in real terms)

  • D = nominal deficit (excluding interest on debt)

  • d = deficit in real terms

  • H = high powered (base) money

  • r = interest rate on bonds (short term)

  • rr = real interest rate

  • P = price level

  • n = real growth rate

  • y = real output

  • П = rate of inflation

The Model

where rr = r − П

We can solve (4) for П,

or for dy,,

Equation (1) says that the deficit (excluding interest payments on the debt) plus interest on the existing stock of debt has to be financed by a combination of new bond issues (BB−1) and an increase in base money (H − H−1). Deflating (1) by the price level and dividing through by real GNP converts (1) into (2). Equation (2) represents the dynamics of adjustment of the debt/output ratio. Equation (3) asserts that the rate of change in nominal income is equal to the rate of change in high powered money. Equation (4) represents the stationary state counterpart of (2). Equation (5) simply rearranges (4) to solve for the rate of inflation. Equation (6) rearranges (5) to solve for the deficit output ratio.

Suppose we begin in steady state, with “consistent” values of the variables, as in (4), (5), or (6). As an experiment we can have either the deficit ratio (d/y) rise or the rate of growth of base money fall. This will lift the bond output ratio (b/y). Equation (2), a first-order difference equation, traces the dynamics of adjustment. If n > rr, the economy will reach a new higher equilibrium bond output ratio. If, however, rr > n, the bond output ratio explodes and the monetary and fiscal authorities are placed on a collision course; one party sooner or later has to give in. If the monetary authorities initiate the deflation, either the fiscal authorities are forced to reduce the deficit ratio or the monetary authorities are forced to abandon their deflationary policies. Indeed, in this latter case—as can be readily seen from (2) and (5)—if the deficit ratio is predetermined, any attempt to reduce base money growth will, for a while, lift the bond output ratio; ultimately, however, it will force the steady-state rate of inflation to rise above its initial levels (to accommodate the larger deficit from higher interest payments as seen in equation (5)).

If, then, rr>n (the case almost universally assumed), there must be consistency between monetary and fiscal policies. On the one hand, the monetary authorities may prevail (as in equation (6)); they independently determine the rate of inflation and the fiscal authorities have to conform. On the other hand, the fiscal authorities dominate (for example, as in some Latin American countries) and the monetary authorities have to fall in line (as in (5)).

To summarize, then, there are two circumstances under which monetary and fiscal policies are tightly linked. The first is where rr > n; the second is where n > rr but there is a base debt/GDP target. If n > rr and there is no target debt/GDP ratio then, in principle, monetary and fiscal policies may be independent of one another.

This framework2 establishes the important proposition that whatever inflation rate emerges in our subsequent analysis in each regime must also meet these tight conditions.

The Pegged Exchange Rate Regime

I consider first the pegged exchange rate regime, focusing on the two possibilities of a peg to a single currency and a peg to a basket of currencies. A simple framework will illustrate how the equilibrium rates of inflation and money growth are determined in these two cases. Suppose we have two large countries (B and C) and a small country (A). A can peg to B or to a basket composed of B and C. A produces three goods, two traded and one nontraded.

Notation

  • pa = log of consumer prices in A

  • pna = log of price of nontraded goods in A

  • pf = log of traded prices in A’s currency

  • wa = log of the wage rate in A

  • eab = (log) A’s bilateral exchange rate vis-à-vis B (units of A per unit of B)

  • eac = (log) A’s bilateral exchange rate vis-à-vis C (units of A per unit of C)

  • ebc = (log) B’s bilateral rate with C (units of B per unit of C)

  • pc = (log) prices in C

  • pb = (log) prices in B

  • q = (log) productivity in A

Framework

Equation (7) says that the consumer price index in A is made up of a weighted average of the price of nontraded and traded goods. Equation (8) says that the price of traded goods is a weighted average of the prices of the two traded goods (pb and pc) adjusted for exchange rate changes. In equation (9) the economy-wide wage rate wa is set in the traded goods sectors and is equal to the sum of the price of traded goods and productivity in the traded sector (the latter assumed to be the same in the two traded sectors). Equation (9) says that the price of nontraded goods is equal to unit labor costs in that sector. Equations (11) and (12) are derived from the preceding system of equations.

Suppose A’s currency is tied to B’s. Then eab is fixed and eac = ebc. In long-run equilibrium (with PPP holding) we also have ebc = pb − pc. We then have

The rate of inflation in A will be equal to B’s inflation rate (of traded goods), adjusted for productivity differences in the two sectors. So, for example, if the inflation rate in B were zero (for traded goods), A’s general inflation could exceed that if there were a (positive) divergence between productivity in the two sectors.

Suppose A’s currency is pegged to a basket composed of B and C. Call ef the effective rate, which is assumed to be fixed. Then (with ef = 0 and ef = α16eab + (1 − α16)eac).

This now gives

Equation (16) presents the result that A’s inflation rate is now a weighted average of B’s and C’s inflation rates, adjusted for productivity differences.

If the “equilibrium” rate of inflation is determined in this way, money growth needs to be consistent with this rate of inflation.

where moa is log of money, yr¯a is output in A (log), and v¯a is velocity. The last two are treated as exogenous. We also have

where r¯a and da represent the external and domestic sources of money. Combining equations (17) and (18) gives

In equation (19)pa is predetermined, as in the previous analysis; yr¯a and v¯a are also given in the long run. Under a peg the growth of international reserves will also be predetermined (targeted). This then determines the steady-state growth of domestic assets that will generate money growth consistent with the exogenously determined rate of inflation.

We can make a similar point as follows. We know that

where RA is international reserve assets, H is base money, D + rB−1 is, as previously, the deficit and BT represents total sales of bonds to the private sector, DCE is domestic credit expansion.

If long-run DCE exceeds (falls short of) the long-run demand for base money, reserve levels will keep falling (rising), which is not sustainable. (We return to this point in our discussion in Section III of balance of payments crises.)

With international reserve stocks predetermined we have the condition for sustainability:

Equation (21) can be rewritten as (1). We showed earlier that under certain conditions D and H must also be closely coordinated.

We can sum up this and the earlier section by laying down the conditions under which a permanent peg will be sustainable in the longer run.

First, as equations (116) demonstrate, wage policy must be consistent with continued profitability in the traded goods sector.

Second, over time, domestic credit must grow at the same rate as money demand, thus avoiding a drain or an acquisition of reserves.

Third, as shown in the initial framework, fiscal and monetary policies may also need to be coordinated. The fiscal deficit then must fall in line with monetary policy, which, in this case, is predetermined.

Flexible Rate Regime

The flexible rate regime is, by definition, free from the constraints operating under a peg. According to the analysis presented in the first section, monetary policy could now, in principle, be dominated by the fiscal or monetary authorities. We focus here on the equilibrium rate of inflation, which is time-consistent. (See also the Annex to this section for an extension of the analysis.) The fiscal deficit is then implicitly assumed to be compatible with this solution. Further analysis of the role of the fiscal deficit within the framework is outside the scope of the paper.

Although the basic framework here draws on Kydland and Prescott (1977) and Barro and Gordon (1983), the actual presentation of the model closely follows Fischer (1988b), which differs in some respects from the earlier frameworks. The explicit application of time-consistency to exchange rate regimes is made by Genberg (1988), Horn and Persson (1988), Giavazzi and Giovannini (1988), and Ganzoneri and Henderson (1988).

As in Barro and Gordon (1983), I begin with an analysis of the single-period case and then extend it to the multi-period case.

I begin by writing a conventional loss function,

Assuming the optimal rate of inflation is zero, the loss in utility (L) is assumed to be positively related to the square of the inflation rate (П) and the square of the excess of output over the target level. The target level of output is, importantly, represented by ky* where y* is the level of output corresponding to the natural rate of unemployment; k > 1 reflects the assumption that the “target” unemployment rate is below the natural rate. This could be so for a number of reasons, for example, the presence of household taxes, generous unemployment benefits, and an excessive real wage rate. This assumption is important in the analysis, as we will see. In equation (22)a represents the weight the authorities attach to inflation in their loss function.

Next assume that the deviation of actual output from its full employment is a function of the divergence between the actual and expected rate of inflation (П *).

Equation (23) is derived as follows. We assume the supply of labor is given; the demand for labor (nd) is a function of the real wage rate.

where b is the share of wages to profits.

Assuming there is no formal indexation provision and workers set wages during negotiations to secure full employment at the natural rate we have, from equation (25), the wage contract

where t−1Epd is the home price level in t expected during negotiations in t−1.

Inserting equation (26) into (25) we have

Remembering Π = pdpd−1 and Π* =t−1Epdtpd−1 allows us to arrive at equation (23).

Money supply policy is used to minimize (22). Underlying the unexpected inflation rate is an assumption about unanticipated monetary policy. Output can increase only if an increase in money is unanticipated.

Substituting equation (23) into (22) we have

The authorities proceed to use money supply policy to minimize (28), assuming expected inflation is given. Differentiating (28) with respect to П and setting the expression equal to zero gives the “optimizing” discretionary rate of inflation for the period.

In due course, in steady state, П = П*. The equilibrium rate of inflation under discretion (Пd) will then be

It is easily shown that while the loss for equation (29) is less than that for a rule that achieved the optimal rate of inflation (П* = 0), in steady-state equilibrium, as in equation (30), the loss exceeds that for optimal inflation.

Equation (30) is important from our standpoint. The equilibrium rate of inflation in this simple framework is determined by several factors, including the weight attaching to the inflation objective (a); second the real effects of unanticipated inflation (b); and the gap between desired output and the level of output corresponding to the natural rate of unemployment (k − 1). Obviously if a →∞ (overwhelming priority is assigned to inflation), or if target output corresponds to the natural rate of unemployment (k = 1), the discretionary equilibrium rate of inflation will equal the optimal rate of zero. The smaller the share of wages to profits (b), the lower the equilibrium rate of inflation.

Unfortunately the story is a little more complicated. We have focused on a one-period analysis, which may be unrealistic. The authorities may try to minimize a loss function that takes account not only of the loss in a single period but also of the (discounted) loss in one or more future periods. There is a price to be paid (in terms of higher inflation without any gain in employment) in the future once workers’ expectations have adjusted; taking account of these “future” losses could act as a restraint on present behavior. The equilibrium discretionary rate of inflation is almost certain to be less, depending on the authorities’ time horizon, on the duration of the “punishment” inflicted on the authorities for a transgression, and, most important, on the authorities’ discount rate. Obviously, the constraint weakens if the discount rate is high.

In Barro and Gordon (1983)—whose model is a little different from the above—punishment is inflicted for cheating; but only in the single subsequent period. Their sustainable rate of inflation is

where q=11+r, with r being the discount rate.

So, in the end, the analysis of the discretionary equilibrium rate of inflation turns out to be very complicated. In addition to the factors captured in equation (30) we also need to take account of “reputational” considerations. One would clearly expect, in this analysis, that the more independent a central bank, and the greater the importance it attaches to inflation objectives, the lower will tend to be the discretionary equilibrium rate of inflation. There is indeed evidence to support this proposition (see Parkin, 1986; and Banaian, Laney, and Willett, 1983).

On the one hand, then, since discretion is likely to lead to excessive inflation, a case can be made for implementing a simple money growth rule, which disallows discretion. On the other hand, discretion is at times stabilizing (we return to this in Section III); thus, what is needed is a compromise between the risks and potential gains of discretion.

It bears reemphasis that whatever equilibrium rate of inflation is arrived at in this way, the deficit is assumed to accommodate it. A broader analysis would also have attempted to pursue the case where the fiscal authorities are dominant. We would then need to focus on what drives the fiscal authorities.

The Optimal Rate of Inflation

In the previous discussion of the time-consistent equilibrium rate of inflation I assumed that the optimal rate of inflation was zero. This was quite arbitrary, and I now need to turn our attention to the question of what the optimal rate of inflation is. Clearly, in principle at any rate, this is also fundamentally important.

It is impossible to do full justice to this question, about which there is considerable disagreement. I simply want to identify some of the considerations that enter into any calculation or judgment about what constitutes the optimal rate of inflation (see Summers, 1981; Fischer, 1981; Feldstein, 1979; Phelps, 1973; Fischer and Modigliani, 1978; and Friedman, 1969). In effect, I am comparing different levels of fully anticipated rates of inflation.

I begin with a number of key assumptions; I then broaden the analysis a little and summarize.

  • (1) Money growth is superneutral (that is, there are no effects on capital intensity).

  • (2) There are no distortionary taxes on goods.

  • (3) Unemployment (output) costs of changes in steady-state money growth are disregarded (that is, I/we abstract from problems of making a transition).

  • (4) There is widespread indexation of wages, taxes, and financial instruments (except money).

Friedman (1969) made an important early contribution to this debate by advancing his “liquidity rule.” Efficiency requires that the real return on money be equal to the real return on capital. If interest is not paid on money, the rate of inflation (po) must be set equal to the negative of the real return on capital.

where rk is the return on capital. Money growth should proceed to the point where the nominal return to capital is driven down to zero. Without interest paid on money the optimal rate of inflation will thus normally be negative, as Friedman envisaged. Any rate of inflation above that (say, zero) is inefficient in the sense that the real return on capital will exceed the real return on money, “distorting” the demand for money balances (which are now “taxed”).

If some interest is paid on all money (but, however, not at market rates), the optimal rate of inflation rises. (If interest is paid on all money at market rates—that is, money is also indexed—any rate of inflation is, in principal, optimal.)

I now drop the first assumption. Summers (1981) shows that where the capital stock is below the “optimal” level and if accelerated inflation increases capital intensity (through the effects on the real value of outside money), there may be a trade-off at work between Friedman’s liquidity rule and the need to increase the stock of capital (assuming there is no other instrument to raise the sub-optimal capital stock). This consideration, if anything, tends to raise the “optimal” rate of inflation. Summers, however, also shows that the smaller the proportion of money held as outside money, the weaker this consideration since the capital intensity effect operates only on outside money (inside money not being part of wealth). The analysis is further complicated if interest is paid on inside money but not on outside money.

Consider now what happens when the second assumption is relaxed. Inflation imposes a tax on money balances; this is distortionary. However, taxes are also imposed on other goods. Hence, the optimal rate of inflation is one that equalizes at the margin the excess burden of the different taxes. There is now no reason why this optimal rate of inflation should not be positive. Phelps (1973) was the first to analyze in detail the revenue implications of a tax on money.

Suppose now that the third assumption is dropped. Suppose, too, that the natural rate of unemployment is independent of the path of adjustment toward a lower steady money growth rate. If the unemployment costs of reducing money growth exceed the benefits from reducing the rate of inflation, any “ongoing” rate of inflation becomes optimal. (Feldstein 1979), however, shows that in a growing economy the benefits of reducing inflation must exceed the costs if the growth rate is equal to or exceeds the discount rate because the current benefits will then approach infinity.

If there is hysteresis—the natural unemployment rate itself is a function of the actual unemployment rate (see Blanchard and Summers, forthcoming)—the unemployment costs become “permanent” and the case for maintaining the ongoing rate of inflation strengthens.

Finally, if we drop the assumption of full indexation the issues become considerably more complex (Fischer and Modigliani, 1978). With indexation incomplete and irregular—notably with respect to taxes and public expenditure—distortions will surface that may well have real effects on investment and equilibrium unemployment. Moreover, higher inflation may be associated with greater uncertainty about inflation and this added uncertainty at higher rates of inflation may raise the costs of inflation.

What conclusions can we draw from this? A more balanced and more extended analysis will recognize here a trade-off. On the one hand, progressively higher positive fully anticipated rates of inflation impose costs that take a variety of forms. First, there will be “menu” costs, that is, costs associated with more frequent price changes. More important, such price changes will not be synchronized across all sectors; thus, the higher the rate of inflation, the larger will tend to be the distortions across different sectors and the larger the incentive to exploit these distortions (which involves costly search activity).

Second, there is the welfare loss associated with the tax on money (which itself weakens with financial deregulation). Third, if unemployment costs are highly valued, and/or if there is hysteresis, the costs of an ongoing inflation rate are difficult to evaluate. Fourth, without full indexation, distortions will surface; and fifth, there may be additional costs that stem from uncertainty about inflation.

On the other hand there are two considerations that work in the opposite direction: the effect on capital intensity and the revenue from inflation (particularly the latter).

But the issues, it seems, are even more complicated than that. There is some literature that also suggests that whether or not money ought to be taxed depends on its nature: whether money serves as a store of value as a vehicle for transactions, or as an intermediate good. Our conclusions hold in the former case but in the latter it appears that money ought not to be taxed—in which case in some conditions Friedman’s full liquidity rule is reinstated. (For a summary of some of this literature see Spaventa, 1989.)

What appears to emerge from this is that not only is little known about the optimal rate of inflation but also that—since the underlying factors in the calculation vary greatly across countries—the optimal rate of inflation is certain to be different in different economies. Recent contributors to the literature have emphasized the revenue (seigniorage) aspects of inflation, arguing that since the revenue needs are different in different countries, a fixed exchange rate regime will not be optimal (Dornbusch, 1988; Drazen, 1988; and Grilli, 1988).

Choice Between a “Peg” and a Float

We have shown that whereas the steady-state rate of inflation is determined in a rather straightforward fashion under a peg, its determination is vastly more complicated under a float.

If a country’s main trading partners (or partner) have (has) low and stable rates of inflation (that is, with strong independent central banks that attach great importance to inflation) and if the country in question has a poor record of inflation (lacking, say, discipline) then, it would seem that, there is a case for pegging. But even this might be questioned. An alternative to pegging is to give the central bank independence, directing it to concern itself with inflation objectives.

Obversely, if there are no suitable candidates to which one can peg or if the country has a good disciplinary record, the case for a float becomes much stronger. A float allows a tailor-made “optimal” rate of inflation to be achieved.

Switzerland, with a strong independent central bank that attaches great importance to curbing inflation, is a good example of the latter. Ireland is probably a good example of the former. Ireland switched in 1979 from pegging to sterling (whose record was poor) to (near) pegging to the deutsche mark (whose track record was good). The inflation rates of France and Italy are constrained by the Federal Republic of Germany within the EMS. Canada, in part, floats because its dominant trading partner, the United States (with 80 percent of its trade), has a relatively unreliable record of inflation.

Choice Between the Single-Currency and Basket Peg

We focus here on some of the key considerations that enter into the choice between pegging to a single currency or to a basket.

(1) If a country pegs to a single currency (or for that matter, to a “select” basket), its trade-weighted rate will fluctuate in line with the fortunes of that currency vis-à-vis the other trading partners. This is considered a disadvantage, which will be greater the less the trade weight attaching to that currency.

Ireland’s currency is linked to the EMS but over 60 percent of its trade is with countries outside the EMS (with the United Kingdom absorbing over 40 percent). Denmark pegs to the European currency unit but conducts substantial trade outside the EMS (for example, with Scandinavian countries, the United Kingdom, Japan, and the United States). Austria, on the other hand, pegs to the deutsche mark, with Germany absorbing about 50 percent of its trade. Moreover, if we add Germany’s satellites, most of Austria’s trade is with relatively pegged currencies. Austria conducts little trade outside the EMS.

(2) If the largest trading partner also has a historically low rate of inflation, pegging to that trading partner will probably ensure a lower inflation rate than pegging to the basket.

(3) Pegging to a single currency (where the country is also a large trading partner) generates an “island” of exchange rate stability; with a peg to a basket all bilateral rates fluctuate. Pegging to a currency that is also linked with other currencies extends the island of exchange rate stability.

(4) It is probably harder to administer a basket peg than a single-currency peg.

(5) The “commitment” to peg is probably weaker with a basket peg than with a single-currency peg.

(6) If trade is highly diversified and there is no single large trading partner, or if a large trading partner has a poor inflation record, it is probably better, other things being equal, to peg to a basket.

(7) The single-currency peg will almost certainly generate larger “imbalances” than the basket peg and thus will require larger reserve needs, which represents a cost.

(8) Pegging to a single currency is likely to enhance trade and capital flows within the region. This may or may not raise welfare. To some extent stabilizing one bilateral rate acts as a “distortion” to trade and capital.

Consider in this general context the Scandinavian countries. Their trade is fairly diversified. At the same time they would like to impose some discipline on themselves; so they peg to a basket. However, they lack confidence in their capacity to sustain the discipline, so they compromise by “periodic” changes in the exchange rate.

Australia’s principal trading partners are Japan (32 percent), the United States (27 percent), the United Kingdom (9 percent), the EMS (18 percent), and New Zealand (5 percent). From the standpoint of “discipline,” Australia could benefit from pegging to the yen rather than to a basket. Other considerations raised above, however, argue against such a policy.

Annex

Equilibrium Rate of Inflation in a Simple One-Period Model Under Flexible Rates and an Adjustable Peg

I consider in this Annex the time-consistent equilibrium rate of inflation under a flexible rate regime and an adjustable peg regime. The model I use is simple and is intended to be illustrative. It assumes that there are wage contracts lasting only one period; it also ignores initially the possibility that contracts will incorporate indexation provisions. (A model with indexation is also presented in Section II.)

The model

Equations (32)(34) and (37) are familiar. We return to these equations in Section II.

Equations (35) and (36) are used in the text in Section I.

The real demand for goods is a function of the real exchange rate and the real interest rate. Perfect asset substitution holds. Real money balances are a function of the interest rate and of output.

Flexible Rates

It is readily shown that in this model

where b20=(α1+α2α15)(1+α4)(α1+α2α15)(1+α5α6+α4)+α4α6<1.(38)

Substitute (38) into (36). Also pd = b20mo + (1 − b20)t−1Emot

In steady state mo = t−1Emot

which is the same solution as in the text for equilibrium inflation (with b = α6).

Special Case of Full Wage Indexation

It is revealing to ask how these results would need to be modified if a provision for complete wage indexation were incorporated into the wage contract. We can rewrite the wage contract in equation (26) in two ways:

or

Equation (42) indexes to the home price level; (43) to the consumer price index. Although variation (43) is more widely used, (42) is more consistent with the framework used here.

Substituting equation (42) into (25), we have

Using equation (43) we have

Expectations now disappear. Monetary policy is also now completely ineffective, with only price effects (see also Section II). There is thus no incentive to “cheat” in an economy that is fully indexed (whatever form this takes). The problem posed by allowing discretion disappears completely. Wage indexation reduces the equilibrium rate of inflation to its optimal level (but, however, at some cost, for example, in the case of equation (42), if there were a supply side shock).

Adjustable Peg

We can use the model to show that equilibrium discretionary solutions for inflation will be the same in an adjustable peg regime as in the flexible rate case (see Horn and Persson, 1988).

I now focus on the discretionary as well as equilibrium rates of devaluation. Monetary policy is of course purely endogenous here. The exchange rate now replaces monetary policy as the policy instrument.

where b22=α14α2α15α6+α1+α2α15

Using similar procedures as previously we have a one-period discretionary solution.

The equilibrium rate of devaluation is then

which is the same as the rate of inflation with flexible rates.

Again, with reputation the rate of devaluation and hence the equilibrium could converge toward zero. It is easily shown that the equilibrium rate of devaluation will be zero with full wage indexation because a devaluation has no real effects (see Section III).

The results here may be interesting but they would hardly reflect reality. The notion that the monetary authorities would use the exchange rate to “cheat” workers and secure more output does not appear to correspond to real world situations.

II. Insulation

A second criterion by which to evaluate alternative exchange rate regimes is to compare their insulation properties. Insulation refers to the capacity of a regime to shelter the economy from a variety of potential short-run disturbances of either domestic or foreign origin, given the policy stance.

At the analytical level, three steps are needed in this approach. A first step is to construct a short-run macromodel of a smaller economy. A second step is to identify in the model sources of unanticipated shocks to the system. A third is to apply some criterion by which it becomes possible to evaluate the performance of different exchange rate regimes in the face of such shocks.

This approach evolves from a well-known paper by Poole (1970). In that paper Poole analyzed the choice between a money and an interest rate target over a period when there was uncertainty about the source of disturbance to the economy. Poole used a simple IS-LM model of a closed economy, with only two disturbances: a money demand disturbance and a real disturbance. Poole evaluated the monetary regimes in terms of their capacity to minimize the fluctuations in output. He showed that if the objective is to minimize fluctuations in output, fixing the nominal interest rate was the appropriate policy for money demand shocks but an inappropriate policy for a real shock, which would now be accommodated.

A similar methodological approach has been applied to the choice of exchange rate regime. The application to exchange rate policy took off from the early 1970s (Argy and Porter, 1972; and Turnovsky, 1976). In recent years the analysis has been refined and extended. Developments have taken the form of altering the underlying model, refining the nature and sources of disturbances, dealing with the intermediate case of the managed float, and changing the loss function.

All of the early literature was concerned only with the polar cases of fixed and flexible rates. Mostly it used simple variations of Mundell-Fleming and analyzed the choice of regime in the face of domestic real and monetary shocks and for select foreign shocks (Turnovsky, 1976). These foreign disturbances were assumed to come from independent changes in foreign prices, foreign interest rates, and foreign output. Marston (1982) analyzed the choice between fixed and flexible rates in the context of a model that allowed for wage and price flexibility. In the same paper, he also refined the analysis of disturbances originating abroad. Instead of treating foreign interest rates, prices, and output as independent shocks, he correctly treated all three as endogenous outcomes of foreign real and monetary shocks, thus introducing an exact parallel with real and monetary shocks originating in the home economy. Turnovsky (1983b) also had shocks originating on the supply side, as did Lachler (1984). Lachler (1984) also introduced a lagged output term in the aggregate supply equation, thus allowing disturbances to exert an impact in the next period as well. Turnovsky (1984) introduced the potentially important distinction between perceived transitory and permanent disturbances. Boyer (1978), and Roper and Turnovsky (1980) were among the early writers to analyze the case of a managed exchange rate. This literature has since exploded. Several papers address the question of the optimal degree of exchange rate management from the standpoint of insulation. Several also address the relationship between wage indexation and exchange market intervention (Turnovsky, 1983a; Aizenman and Frenkel, 1985; and Devereux, 1988).

By what criterion are we to evaluate performance? Different authors have adopted different criteria. Turnovsky (1976) takes the objective to be to minimize fluctuations in output. Fischer (1977) and Frenkel and Aizenman (1982) assume the authorities wish to minimize real consumption variations. Flood (1979) assumes the authorities wish to minimize domestic price fluctuations. Turnovsky (1984) takes the objective to be the minimization of output at its “full information frictionless level”; Lachler (1984) and Daniel (1985) assume a similar objective. In another paper Turnovsky (1983a) assumes the objective is to stabilize some weighted average of domestic real income and the domestic price level. Aizenman and Frenkel (1985) take the objective to be to minimize the difference between the “level of employment that would have prevailed under conditions of full clearance of labour markets and the actual level of employment.” In principle, a more extensive range of target variables could be accommodated in the loss function, including real interest rates, real exchange rates, as well as output and inflation (Argy, 1989b). All of this needs to be given a solid micro foundation.

As would be evident from the above discussion and from my own analysis to come, the choice between alternative exchange rate regimes depends on (i) how the economy is assumed to function (that is, the underlying model), (ii) the sources of disturbances, and (iii) the loss function adopted.

The next section presents a model of a small economy exposed to disturbances from within as well as from the rest of the world. The sections that follow evaluate the insulation properties of different exchange rate regimes, including one that allows for some exchange rate management. The analysis is undertaken for different assumptions about the degree of wage indexation. I also deal briefly with the literature concerned with optimizing monetary policy in the face of disturbances from different sources.

1. A Model of a Small Economy and of the Rest of the World

The model used in this part for the analysis of insulation properties is similar to one widely used in open-economy macroeconomics, most notably by Marston and Turnovsky. Our small economy produces a single good, which, in general, is an imperfect substitute for the good produced in the rest of the world. It absorbs some of its own production and exports the balance. It also absorbs imports from the rest of the world at a foreign price it cannot influence.

Table 1 lists the basic equations of the model for the small economy; Table 2 presents the basic equations for the model of the rest of the world. (Notations appear following Table 2.)

Table 1.Model of a Small Economy
Aggregate demand for goods
Money market
Aggregate supply and labor market
Equilibrium condition
Notations follow Table 2.
Notations follow Table 2.
Table 2.Model of the Rest of the World
Money market
Aggregate supply and labor market
Notation
y = domestic output (in logs)
rf = foreign interest rate
rd = domestic interest rate
p = consumer price index (in logs)
tEpt+1 = expected consumer prices formed in period t for t+1 (in logs)
pd = price of domestic output (in logs)
e = exchange rate—units of domestic currency per unit of foreign currency (in logs)
pf = price of foreign output (in logs)
yf = output in rest of world (in logs)
u1 = serially uncorrelated disturbance term to domestic private expenditure—zero mean
mo = stock of domestic money (in logs)
m¯o= exogenous long-run money supply
u2 = serially uncorrelated disturbance term to money demand—zero mean
u3 = serially uncorrelated disturbance term to supply - zero mean
Eet+1 = expected exchange rate formed in period t for t+1 (in logs)
w = wage rate (in logs)
wc = wage contract (in logs)
wf = wage rate abroad (in logs)
wcf = wage contract abroad (in logs)
t−1Ept = expected consumer price level formed in period t−1 for t (in logs)
u1f, u2f, u3f = equivalent disturbance terms abroad
Similar notation applies to expectations in the rest of the world.

Equation (50) asserts that the real demand for goods is a negative function of the real interest rate and a positive function of the real exchange rate and of foreign output; u1 is a disturbance term to expenditure. Equation (51) says that real money balances are a negative function of the domestic interest rate and a positive function of output; u2 represents a disturbance term to money demand.

Equation (52) represents the assumption of perfect asset substitution. The return on domestic bonds is assumed to be equal to the expected return on foreign bonds [rf + (tEte +1 − e)].

Equation (53) represents the monetary policy reaction function. It says that monetary policy may be directed at stabilizing the nominal exchange rate. Monetary policy may take the form of open market operations or unsterilized foreign exchange intervention.

Equation (53) accommodates a variety of potential exchange rate regimes. If П1 = 0, we have an exogenous money stock and a flexible rate regime. If П1 → ∞, we have a fixed rate regime with monetary management. If П l < ∞ > 0, we have a managed float, with П1 representing the degree to which the authorities “lean against the wind.” As we will see, it may be appropriate at times for the authorities to lean with the wind, in which case Пl < 0.

Equation (54) represents the aggregate supply side of the economy. The supply of output is a negative function of the real wage rate, with the nominal wage rate deflated by the price of domestic output (capital stock, technical progress variables are omitted); u3 represents a supply shock.

The wage rate in period t is assumed to be negotiated in period t−1; a single contract period is assumed. Wages are contracted on the strength of previously formed expectations about the consumer price index in the current period. At the same time some correction is allowed for errors in forecasting. П5 captures the coefficient of adjustment. If П5 = 1, we have in effect complete wage indexation and expectations become irrelevant.

Equation (56) defines the consumer price index as a weighted average of home prices and import prices, while equation (57) equates the demand for goods with the supply of goods.

The model of the rest of the world parallels the model of the small economy. The rest of the world is so large relative to our small economy that it is effectively a closed economy. Thus, output in the home economy and the exchange rate vis-à-vis the small economy are insignificant.

How are expectations determined in the model? Price expectations appear in the goods market and in the labor market. Exchange rate expectations appear in the money market. By design, in this model, the disturbances have a zero mean and are not serially correlated. They therefore exhaust their effects within the same period. The rationally expected value of these disturbances must therefore be zero by definition; given, too, the absence of carryover effects the expected value of the price level and the exchange rate must be their levels when disturbances are set at zero. These expectations can therefore be treated as exogenously given by the state of the economy, stripped of disturbances. This is also the treatment, for example, in Turnovsky (1983a) and Marston (1982).

The model, as we have seen, accommodates alternative exchange rate regimes, including managed exchange rates. It also accommodates a wide range of disturbances. There are parallel real demand, money, and aggregate supply disturbances at home and abroad, which amounts to a total of six disturbances.

Finally, what criteria do we use in evaluating performance? Earlier, we reviewed the different criteria used in the literature. Ideally, we ought to have a loss function that accommodates all the key variables of concern to the authorities. The final evaluation will then depend on the weights attaching to the variables. The advantage of this is that it enables us to say something about potential trade-offs. (For example, it is sometimes said that a switch to flexible rates increases exchange rate volatility but dampens interest rate volatility.) Unfortunately, such an analysis is difficult to undertake with a model of even modest complexity. Solutions multiply, adding to the confusion. For this reason I focus primarily on output volatility as the key (and indeed most widely used) criterion, at the same time recognizing the limitations.

2. Fixed Versus Flexible Rates for Different Assumptions about Wage Indexation

Table 3 presents some solutions for our smaller economy. All disturbances are assumed to be unanticipated. The model is solved on the assumption that all expectations are held fixed. Equation (61) is the solution for output in terms of all the disturbance terms; (61) is the solution for fixed rates since by definition “e” is constant.

Table 3.Solutions to Model of the Home Economy

> 0

> 0

Equation (62) is the final solution for the exchange rate in terms of disturbances. To obtain a solution for flexible rates, as already indicated, we set П1 → 0.

Equation (63) gives the final solution for output under flexible rates in terms of disturbances. It is obtained by substituting (62) into (61) and collecting terms.

Given the definitions of b1 and b2 (see footnotes to Table 3), it is readily seen that when indexation is perfect (П5→ 1), b1 = b2 so that the ratio b2/b1 → 1. This ratio appears in the results, so a solution to this case can be more readily obtained. If, at the other extreme, wage indexation were zero (П5→ 0), b2 = 0 and b1 → α6.

Table 4 gives the solutions for output, the price level, and the interest rate in the rest of the world in terms of the three disturbances abroad; П5 represents the degree of wage indexation abroad.

Table 4.Solutions to Model of Rest of the World

Returning now to Table 3, it is evident that solutions for output in the home economy can now also be obtained directly in terms of the same disturbances abroad. To illustrate, we can substitute the solution for rf in equation (66) from Table 4 into equation (63). We can do the same for pf and yf. By summing up the effects of any foreign disturbance on output we can arrive at the net effects on home output of any disturbance abroad. For example, a real disturbance abroad affects the home economy through its combined effects on the foreign interest rate, foreign output, and foreign prices.

I focus, to begin, only on the choice between fixed and flexible rates. I proceed in two steps. Deal first with the case of the three disturbances originating at home. This is clearly the easier case. Next, deal with the more complicated case of disturbances originating abroad.

Disturbances Originating in the Home Economy

It is convenient to undertake the analysis for the polar cases of zero and perfect wage indexation. I start with the case of zero wage indexation (b2 → 0) (b1 = α6).

Consider first a money demand disturbance when exchange rates are fixed. A domestic money demand shock leaves real variables as well as prices unchanged (u2 does not appear in equation (61)). The shock will be absorbed by an accommodating monetary policy: in this instance the potential upward pressure on the currency will trigger an open market purchase.3 When the exchange rates are flexible (П1 → 0), the same money demand shock, will, however, provoke real as well as price changes in the economy (see equation (63)).

For an expenditure disturbance, output expands by more under fixed rates than under flexible rates. (The solution for u1 in equation (63) is less than the solution in equation (61).) With flexible rates there is upward pressure on the interest rate; the currency appreciates and this crowds out some private expenditure. With fixed rates the potential appreciation is nullified by the injection of money, accommodating in part, at least, some of the increase in private expenditure.

Consider, lastly, the case of a supply shock. The effect of a supply shock on the currency is ambiguous. The reason is that with fixed rates output increases but prices fall, so that the effect on the interest rate and, hence, on the exchange rate is uncertain. Ambiguity on the exchange rate front translates into ambiguity with respect to the relative effects on output. If the currency appreciates (devalues), a flexible rate (fixed rate) regime performs better.

I turn now to the case where indexation is perfect (П5 = 1) and b2/b1 = 1. It is readily seen from equation (61) that the exchange rate term drops out, indicating that output effects are identical under fixed and flexible rates. This result is also found in Marston (1982) and in Turnovsky (1983a). A money demand disturbance has no real effect in either regime. A real or a supply-side disturbance increases output equally in the regimes.

Why do we have this result? The reason is fairly simple. Equation (53) of the model shows that the key difference between the two regimes lies in the behavior of the money supply. The money supply is adjusted to stabilize the exchange rate; however, a change in the money supply leaves all real variables unchanged.

A couple of illustrations will clarify this point. Suppose you have a real (expenditure) disturbance under flexible rates. Output will rise; there will be a real as well as a nominal appreciation; and the interest rate will rise (creating an expectation of a future devaluation of the home currency). Suppose at this point the monetary authorities try to stabilize the currency by injecting money into the economy (as in equation (53)), say by open market purchases. An increase in the money supply in this model will leave the real exchange rate and the real interest rate unchanged; will lower the interest rate (to its original level); devalue the currency; and increase home prices (the latter two less than proportionately to the money supply).4

Similarly, suppose a money demand disturbance arose under flexible rates. The effects are identical to a reduction in the money supply. The monetary authorities put money into the economy to undo the appreciation; this restores the original interest rate and price level, leaving output unchanged.

Disturbances Originating Abroad

To simplify, suppose wage indexation is the same in the home economy and abroad. I consider again first the case where indexation is zero.

Take a real disturbance abroad (with (П5 = 0)). Table 4 shows that output, prices, and the interest rate will all rise abroad. If the exchange rate is fixed we can use equation (61) to determine how this will affect output in the home economy. It is readily apparent that the combined effect of all three changes abroad on output (pfyfrf↑) is ambiguous. On the other hand it seems certain that under flexible rates output will rise on all three counts. In general, it is difficult to determine unambiguously which regime is more stabilizing.

Consider now a money demand disturbance abroad. This lowers output and prices abroad but raises the foreign interest rate (Table 4). With fixed rates output will fall (when the solutions for pf and rf from Table 4 are substituted into equation (61) the negative term attaching to pf disappears). With flexible rates the home currency will devalue, but the net effect on output at home is ambiguous. If output falls, on balance, flexible rates are unambiguously more stabilizing than fixed rates. If output rises, the position is less certain, although here, too, there remains some presumption in favor of flexible rates.

Finally, I consider the case of a supply shock abroad. As Table 4 shows this lowers prices and raises output, while the effect on the foreign interest rate is ambiguous. There are too many ambiguities here to be able to say anything definitive about the insulating properties of the exchange rate regimes.

To summarize, it is evident that little can be said about the relative insulation properties of fixed and flexible rates in the face of disturbances originating abroad (see also Argy, 1990).

With perfect wage indexation we have exactly the same results as previously for reasons already explained. In terms of real effects the two regimes perform identically.

Some Conclusions

Can anything useful be said about the insulating properties of fixed and flexible rates? The following are the more important results.

  • When wage indexation is perfect, all real effects (including effects on output, real interest rates, and real exchange rates) are identical in the two regimes (or intermediate regimes) whatever the source of disturbance. If the loss function takes account only of real variables, there is nothing to choose between the regimes.

  • For disturbances originating at home—provided indexation is imperfect—fixed rates are more stabilizing (in output terms) for money demand shocks while flexible rates are more stabilizing for real expenditure shocks. For supply disturbances, the relative outcomes are ambiguous.

  • For disturbances originating abroad—again provided that indexation is imperfect—little definitive can be said. It seems that for money demand disturbances abroad, flexible rates are more stabilizing (the opposite of the case of a parallel disturbance at home); nothing conclusive can be said about the other two disturbances.

Limitations of the Analysis

The limitations of the analysis and the conclusions (inconclusive as these are) stem from the potential inherent deficiencies of the model; from the assumption about the nature of the disturbances; and from the limited loss function used. We have already discussed the last. The limitations are as follows:

  • (1) For short-run analysis the implicit assumption that the Marshall-Lerner condition holds (there are no “J” curves) is restrictive (see Niehans, 1975).

  • (2) There are no imported inputs in the model. This would complicate the aggregate supply equations and the solution (see Marston and Turnovsky, 1985; and Gylfason and Schmid, 1988).

  • (3) There are no wealth effects. Except for valuation effects, such effects are not, however, likely to be significant in the short run.

  • (4) All lagged carryover effects have been suppressed. Clearly this is highly restrictive. For example, the contract period could be extended, and the aggregate supply equation could also have a lagged output term.

  • (5) The model assumes perfect asset substitution. While this assumption is becoming increasingly appropriate it may still not be entirely realistic (disallowing as it does potential risk premia). Moreover, in terms of historical relevance and evolution, it is clearly inappropriate since, for example, before 1979 many countries maintained controls over capital movements.

  • (6) Although the model identifies a wide, representative range of disturbances, the nature of the disturbance is still restrictive. For example, once a disturbance occurs the public may be uncertain whether it is transitory or permanent. The public may attach probability weights to these alternatives (Turnovsky, 1984). Moreover, for obvious reasons, I have dismissed all covariances that may be important in the real world.

  • (7) A related intriguing question is whether the disturbances themselves (or their covariances) are different in character or in importance under different regimes.

Extension to a Three-Country World

Argy, McKibbin, and Siegloff (1990) use a three-country model to evaluate the insulation properties of alternative exchange rate regimes, again for different assumptions about wage indexation. The three-country framework allows a richer menu of regimes and disturbances. The three-country world comprises two large countries and one small one. The small country can peg to either of the larger countries or to a basket or it can float. Real or monetary disturbances can originate in either of the larger economies or in the small economy. Many of the outcomes are unfortunately ambiguous.

The McKibbin-Sachs global (MSG2) model was used to attempt to resolve these ambiguities. The two large countries are Japan and the United States, and the small country is Australia. The conclusion reached was “the floating rate regime for Australia performs well…except for a shock to money demand in Australia. . . . Of the fixed exchange rate regimes it is better to peg to a basket of currencies, especially for foreign shocks in a world of globally floating exchange rates.”

Optimal Foreign Exchange Intervention

The symbol П1 represented the degree of foreign exchange intervention in the model. In principle there is a value for П1 that will minimize the variance of output, taking account of all disturbances to which the economy is exposed.

Technically, the problem can be reduced to the following; first is the general solution for output in terms of all disturbances:

where, for example, П1214 … can be read off from equation (63). Closer inspection of the underlying coefficients reveals that П1 appears in k2 and in k3 (in fact, more correctly in the ratio k3/k2).

I now take the variance of equation (66); disregarding, for reasons of simplicity, all covariances. We obtain

The next step is to differentiate the variance of output with respect to П1, set the expression equal to zero, and then solve for П1. This gives a П1*, which minimizes the variance of output. Evidently, the optimal level of intervention will depend on the relative importance of the different disturbances.

For each disturbance there is an optimal level of intervention, which may be positive (leaning against the wind) or negative (leaning with the wind). For example, from equation (68) to obtain this level of intervention for a real domestic disturbance one would find a value of П1 so that П12 = 0; the same would be necessary for each of the other disturbances.

The first point that needs to be made, which follows directly from our earlier analysis, is that if wage indexation is perfect—and given the loss function used—intervention serves no purpose. Changing the money supply will not alter any real outcomes; hence, intervention to achieve a better output performance will fail. It follows that only cases of imperfect wage indexation are worth discussing, since in this instance the volume of money does have real effects. The second point is that an exercise of this kind is, in the end, largely academic and hardly likely to yield solutions that could be of use to central bankers.

Despite these caveats it is worth at least indicating what the optimal level of intervention might be, where indexation is imperfect, for certain types of disturbances. In particular since central banks tend to lean against the wind one might ask whether there are circumstances under which the authorities might be better off leaning with the wind. Given the imprecision attaching to foreign disturbances, I will illustrate these points only for the three disturbances at home.

Consider first a money demand disturbance at home. We know that the fixed rate solution is the optimal one. Hence, for that disturbance П1* → ∞ there is “perfect” leaning against the wind.

Consider, next, a real expenditure disturbance. It is readily shown that П1* = - (α4 + b2/b1). This says that to stabilize output the authorities need to lean with the wind. The optimal level of intervention is also positively related to the degree of wage indexation b2/b1. Why is this?

We know that in this case output increases and the currency appreciates. To stabilize output money needs to be tightened further, so that effectively the authorities would be leaning with the wind, that is, destabilizing the exchange rate. If they leaned against the wind, increasing the volume of money, output would be destabilized.

Also, the higher the degree of wage indexation, the stronger the output effects (the appreciation forces wages down further). In equation (63) it is apparent that the larger b2/b1, the weaker the negative effects of an increase in u1. The stronger the output effects, the more restrictive monetary policy must be, that is, the stronger the leaning with the wind.

Finally consider the case of a supply disturbance. It can be shown that when П1* = -(1 + α4), output can be perfectly stabilized. Output increases so monetary policy needs to be restrictive. Equation (62) shows that when u3 increases at this level of intervention, the currency appreciates.

Optimal Monetary Policy in the Face of Disturbances

Is it possible to design a monetary policy reaction function that will simultaneously attempt to neutralize disturbances from various sources? This is the question addressed in Turnovsky (1984, 1985), and in Aizenman and Frenkel (1985). The approach raises important questions. I choose to illustrate this theme by using a simple model in (Turnovsky 1984, 1985). The model in Aizenman and Frenkel is more elaborate but raises similar issues; its principal conclusions are summarized later.

The model comprises the following equations:

Equation (69) assumes that purchasing power parity holds (that the real exchange rate is fixed). Equations (7072) have been used previously and require no explanation. As in the model used earlier we have disturbance terms to money demand (u2) and aggregate supply (u3), with similar assumed characteristics. Equation (73) is the important novel feature of the model. It assumes a reaction function that allows the monetary authorities to respond, at once, to the exchange rate (where П4 is the coefficient of foreign exchange intervention); the foreign interest rate; and foreign prices. Information on the exchange rate and the foreign interest rate is “instantly” available but the assumption is also made that information on foreign prices is quickly available (from equation (69), p must also be available).

There are four disturbances in the model: u2, u3, rf, and pf. Given the assumption about the disturbances (single period), the rational expectation about the future exchange rate is that it returns to its initial level. Thus t−1Ept is also fixed when the shock occurs. The model can readily be solved for output (variations in which we again assume the authorities wish to minimize). In terms of our four disturbances the authorities cannot observe u2 and u3, but (as already noted) they can observe pf and rf.

The objective is to stabilize y. We have four coefficients in the solution corresponding to the four disturbances. We have only three coefficients in the monetary policy reaction function corresponding to three indicators. If we drop one disturbance, we would have three coefficients in equation (74) which could be set at zero (meaning that the disturbances have no effect on output). We could then solve for the optimal coefficients (П4*, П5*, П6* the reaction function, which would yield perfect stabilization.

Suppose we drop u2 from the system. We then have5

We can solve this system of three equations for П4*, П5*, П6* We have

Alternatively—albeit in somewhat more complicated fashion—we could drop u3 from the system. We could again completely stabilize output. We would then have

It is evident that if both u3 and u2 are relevant, we have an “excess” of disturbances relative to indicators so that output cannot be completely stabilized.

Turnovsky (1984), using a model similar to the above, also analyzes the case of unanticipated disturbances, which are assumed to be permanent once they have occurred. Thus, tEet+1 can now no longer be taken as given. The model can be solved for the expected exchange rate, which can now also be accommodated as an additional indicator, that the authorities, in principle, can exploit. As it happens, however, in the model it is sufficient to set the intervention coefficient П2 alone at its optimal level to stabilize output completely. The other coefficients in the reaction function П5, П6, and the (new) coefficient attaching to the expected exchange rate, can now be set arbitrarily.

Aizenman and Frenkel (1985) develop a similar theme. Again PPP is assumed, but the aggregate supply side is developed in considerable detail; in particular, a wage indexation coefficient is also accommodated. The monetary sector is conventional, and perfect asset substitution is assumed. There are three disturbances: to the foreign price level, to the foreign interest rate, and to productivity. There are three independent indicators in the monetary policy reaction function: the exchange rate, the foreign interest rate, and the foreign price level (as in Turnovsky). As already indicated, the authors also assume that the goal of policy is to minimize the welfare loss associated with discrepancies “between the realised levels of real wages and employment and the equilibrium levels obtained when labour markets clear continuously without friction.”

In their model there are four potential policy settings: the three coefficients in the monetary policy reaction function and the degree of wage indexation. Thus there are four policy settings but only three disturbances; one policy setting is redundant. Foreign exchange intervention can therefore be traded against wage indexation. So, in principle, there are enough instruments to eliminate fully the distortion to the real wage.

These analyses are useful and interesting. In general, one would expect an excess of disturbances over indicators. In my model there were six disturbances; some financial variables (e, rf) are readily observable, but price levels are observed only with a substantial lag (and then usually with “an error,” later revised).

It is doubtful whether wage indexation is really a policy instrument on an ongoing basis (in the long run all evidence suggests “perfect” wage indexation). To implement this framework one would need a detailed knowledge of the structure of the macroeconomy; the structure would be highly complicated, and changing dynamics raise serious questions about the applicability of the approach.

These approaches have led the way toward, perhaps, empirical applications of the framework. Once the underlying coefficients of the model are known, a policy reaction function could be arrived at that responded to observable indicators.6

III. The Policy Efficiency Criterion

In Section II we focused on the insulation properties of alternative exchange rate regimes in the face of unanticipated disturbances, expected to be reversed in the next period. Except for the analysis of an optimal monetary policy, policy was essentially assumed to be nonactivist, an assumption that was not inappropriate in the circumstances.

We now extend the analysis in two ways. First, we assume that a disturbance is longer lasting (for example, it may come from a sustained disturbance abroad (such as a recession); a resources discovery (the so-called “Dutch disease”); an oil price shock; and protracted changes in the monetary system (deregulation or innovation)). Second, we now also allow for the possibility that policy will respond to the sustained disturbance. So activist policies, aimed at correcting disequilibria, are now explicitly introduced in the analysis. (This opens up a potential conflict with Section I, which concluded that discretion might expose a country to excessive inflation.)

We assume in this part that the authorities target the level of output (employment), the current account (or the foreign debt to GDP ratio), and inflation.7 With this new framework in mind we can undertake two kinds of analyses.

One question we might ask is which exchange rate regime has the more efficient built-in endogenous mechanisms for restoring equilibrium (that is, returning the economy to, say, full employment and current account balance), without invoking discretionary policy. There are two aspects of this: one concerns stability (whether there is convergence) and the other the speed of adjustment, assuming that equilibrium is ultimately restored. Many theoretical models (indeed most) assume, for example, that forces at work over the medium to long run will restore both the full employment and current account balance (see Branson and Buiter, 1983; Allen and Kenen, 1980; and Kawai, 1985). So we might be interested in the cumulative departures from equilibrium for each regime. Interesting as it clearly is, we do not pursue this question here.

A second question one might ask is how efficient policy will be in each regime in correcting disequilibria.8 Three key criteria may be applied here:

  • (1) One can draw on the Tinbergen principle that the more independent instruments available in a regime the more targets are achievable. Or, more elegantly, the more instruments available the more effective in minimizing a loss function (à la Theil).

  • (2) One can look at the “mean” size of the multiplier (which could be zero). Other things being equal, the larger the multiplier the less the change in the instrument to achieve a given change in the target variable. (This is an advantage in that it minimizes instrument variability.)

    Much of the original Mundell-Fleming literature was concerned precisely with the application of this criterion. For example, two well-known results are that monetary policy is more effective (the multipliers larger) under flexible rates than under fixed rates (whether or not there is sterilization), and that fiscal policy is more (less) effective under fixed rates than under flexible rates when capital mobility is relatively high (low).

  • (3) One can evaluate the degree of uncertainty attaching to the policy multiplier under each regime (see Frankel, 1988).

In summary, one exchange regime is to be preferred to another if (i) more policy instruments are available, (ii) these instruments have more “powerful” effects on the target variables, and (iii) there is greater certainty attaching to the effects of each instrument.

An important assumption underlying the analysis in Section III is that the authorities be able to identify (or forecast) the condition of the economy, which policy is intended to correct. Obviously if they could not do so, discretion would carry additional risks; this is an old issue in the debate over rules versus discretion (see Argy, 1988; and Frankel, 1988).

We focus, in this part, principally on four types of exchange rate regimes: the adjustable peg, the “permanent” peg, the float, and the managed float.

As background to the discussion that follows, Table 5 sets out the potential instruments available for each exchange rate regime.

Table 5.Exchange Rate Regimes: Potential Instruments Available
Regime“Wages Policy”Money SupplyFiscal PolicyExchange

Rate Policy
Adjustable Peg
Imperfect
Assetyesyesyesyes
Substitution(limited)
Perfect Assetyesnoyesyes
Substitution(limited)1
Floatyesyesyesno
Managedyesyesyesyes
(Float)(limited)(limited)
Pegyesnoyesno

See text.

See text.

The Adjustable Peg Regime

The discussion of an adjustable peg regime is in two parts. In the first, I assume perfect asset substitution; in the second part, I assume that “capital controls” are effective, and this limits the degree of asset substitution. Later, I deal with specific issues raised by this regime.

Perfect Asset Substitution (Traditional Analysis)

To begin, we assume that any policy change (notably a change in the exchange rate) is largely unanticipated. Later, we deal with the case of anticipated changes and the potential for balance of payments crises.

The model to be used here closely resembles the earlier one. It is set out in Table 6. The interest rate is predetermined. Equation (80) is the current account equation; the rest of the model is similar to the one previously used.

Table 6.The Model
Notation
yr = output (log)
e = exchange rate (log)
w = wage rate (log)
mo = money (log)
pd = domestic prices (log)
gr = government expenditure (log)
rd = domestic interest rate (fixed)
CAXo = current account as a proportion of initial value of exports
p = consumer price index (log)
edm = elasticity of demand for imports
edx = elasticity of demand for exports
XoYo = initial ratio of exports/imports to GNP (marginal - average)
c = marginal propensity to spend out of income

We focus here on the use of two potential instruments of policy (the exchange rate and fiscal policy) to achieve current account and output objectives. Solutions to the model for output and the current account are shown in Table 7 (equations (83) and (84)).

Table 7.Solutions for Output and the Current Account, Exchange Rate, Real Exchange Rate, and Consumer Prices
See bottom Table 6 for notation.
See bottom Table 6 for notation.

The idea that the exchange rate and fiscal policy could be used jointly to achieve current account and output objectives is an old one. The two principal contributions here are Swan (1955) and the New Cambridge School (Kinsella, 1975; and Smith, 1976).

We know that if there is full information, e and gr can in principle be set to achieve both internal and external balance (à la Tinbergen).

Mundell (1962) offered the important insight that where the detailed structure of an economy is not known or where policy-making is decentralized, governments may be able to proceed, on the strength of only limited information, by assigning a policy instrument to that target for which it has a relative advantage. In other words in a simple two-instrument, two-target framework one instrument would be assigned to one target and the other to the other target. The instrument would be altered, whenever its target variable was outside the acceptable range, in a way that would bring the variable closer to its targeted value.

On what basis would the assignment be made? Formally, in a two-target, two-instrument context, we ask how the two instruments must be varied conceptually to achieve the same change in one of the targets. Given the required changes in the instruments we need to determine how the other target responds to those same changes in the instruments. The instrument that produces the largest absolute change in the other target is the one that ought be assigned to that target. It has been shown that this ensures stability in the sense that assignment in this way will tend to move the economy closer to its targets. However, the reverse assignment need not be unstable; indeed it may also be stable. So the Mundellian criterion for assignment ensures that the policy is stabilizing while the reverse assignment may or may not be.

Consider now an increase in government expenditure with a devaluation that together yield an equivalent increase in output. Which policy change will yield the larger current account change?

The answer depends on key assumptions made about wage indexation and about the marginal propensity to spend on consumption and investment out of income. Suppose, at one extreme, that wage indexation is perfect. Fiscal expansion will then increase output and create a current account deficit. A devaluation, on the other hand, will leave both output and the current account unchanged. In this case we only have one instrument to play with, not two (see equations (83) and (84), where П43 = 1).

If wage indexation were imperfect, both fiscal expansion and a devaluation would increase output. At the same time the first will worsen and the second will improve the current account. The relative size of the effect on the current account will depend on the structural parameters of the model.

In the more general case we can apply the Mundell “assignment” rule to determine the relative effects on the current account of equal output effects of a devaluation and fiscal expansion. The result is

where gre represents an output effect for fiscal expansion equal to that of a devaluation. Fiscal policy would be assigned to the current account if the full bracketed expression in equation (89) exceeds the first expression for the exchange rate in (84).

If we take equation (89) less the coefficient for e in (84) we get

This indicates the “minimal” information needed to determine assignment with imperfect indexation. It is easily shown that the expression is unambiguously positive if α1 = α13. (This will be the case if c = 1 in Table 6.)

If the marginal propensity to spend on consumption and investment out of income approaches one (as was assumed by New Cambridge) a devaluation will have powerful effects on output but no effect on the current account (see equation (84)). An expansionary fiscal policy, on the other hand, will have adverse effects on the current account balance and positive effects on output. In this case the appropriate assignment is to have the exchange rate monitor output while the budget balance controls the current account, at least over the medium term.

Chart 1 illustrates the orthodox (Swan) assignment. Unbroken lines represent the orthodox assignment, and broken lines the (reverse) unorthodox assignment. The CA schedule represents the combinations of the exchange rate and government expenditure (fiscal policy) which keep the current account unchanged at a “satisfactory” level. Fiscal expansion is assumed to worsen the current account; a devaluation improves it. The yy schedule represents the combinations of the exchange rate and fiscal policy, which keep output at a satisfactory level. Fiscal expansion increases output, so that an appreciation is assumed to be needed to maintain the same level of output.

Chart 1.Orthodox Assignment

The special case where wage indexation is perfect—and particularly where wages respond rapidly to prices, (the vicious circle devaluation cycle case)—is interesting in that it raises the possibility that, for a smaller economy, the exchange rate might be used to achieve price level objectives. This case is shown in Chart 2.

Chart 2.Assigning the Exchange Rate to Inflation Objectives

Imperfect Asset Substitution (Traditional Analysis)

We assume here that capital controls are sufficiently effective to allow the authorities control over the money supply (or the interest rate) so sterilization policies are now feasible.

The model we can use here is similar to the one set out in Table 6 except that the interest rate is not predetermined while the money supply is now a “policy” instrument. The system now comprises equations (7881b), after imposing equilibrium in the goods market. These four equations together determine output, the interest rate, the price level, and the current account. There are now three policy instruments (money, the exchange rate, and fiscal policy) that allow us, in principle, to achieve three targets. How is policy to be designed?

We assume here that the three “targets” are output, the current account, and the interest rate. The interest rate as a target can be justified on one of two grounds: to achieve either target reserve levels or a change in the investment share of GNP.

The system can be reduced to:

The П coefficients represent the solutions to the model (П24 is ambiguous because with the money stock fixed, a high degree of wage indexation can generate a deflationary interest rate squeeze).

Without undertaking detailed analysis we can visualize the system working as follows. Keeping the interest rate fixed we can, as previously, use the exchange rate and fiscal policy to achieve output and current account targets. We can then change the mix of monetary and fiscal policy to achieve reserve level targets or a target share of investment without disturbing either the current account or output targets.9

Uncertainty of Effect of a Change in the Exchange Rate

I noted at the outset that one consideration in evaluating policy efficiency was the degree of certainty attaching to the outcome of a policy change. I focus here on the effects of an exogenous devaluation. Some uncertainties were implicitly built into our results and some were in fact noted. The effects of a devaluation in Table 6 depend on the values of all the underlying coefficients, with some uncertainty attached to each one of these (for example, the degree of wage indexation, the Marshall-Lerner condition (α13), and so on). The actual speed of adjustment will depend on all the underlying lags, including the trade lags (with potential “J” curves) and the lags in the adjustment of wages and prices. If a devaluation is anticipated some effects may precede the actual devaluation (to be discussed later). Other elements not accommodated, which may have a substantial bearing on the outcome, include: wealth effects, distribution effects (between wages and profits), initial trade, net creditor (debtor) positions, intermediate inputs, initial level of activity, and tax transfers to governments (see Currie, 1976; Hamilton, 1988; Krugman and Taylor, 1978; Gylfason and Schmid, 1988; and Hanson, 1983).

Balance of Payments Crises and the Fixed-Adjustable Peg Regime

There is at present a large literature focusing on the potential for balance of payments crises in a fixed rate regime in circumstances where the underlying macroeconomic policies are inconsistent with such a regime.

I illustrate the key issues here by presenting a simple base model (taken from Flood and Garber, 1984). Later, I relax some of the simplifying assumptions of the model.

The Base Model

The base model is simple. Real money balances M/(P) are a negative function of the home interest rate (rd) (equation (94)). The money stock is the sum of reserves (R) and domestic credit (D) (equation (95)). Domestic credit is growing at a rate u, which is greater than zero and exceeds the rate “abroad (equation (96)). Purchasing power parity and perfect asset substitution are both assumed to hold in equations (97) and (98), respectively. (S is the exchange rate, P* is foreign prices, and rf is the foreign interest rate.)

We can combine equations (97), (98), and (94) to arrive at

where β=a0P* − a1P* rf= > 0 and α=a1P*.

When the exchange rate is fixed (S˙ = 0), using (99) and (94)

and using (96) we also have

Equation (101) makes the point that if domestic credit expansion is excessive (that is, it exceeds the fixed money demand as in (99)), reserves are run down at the rate of credit expansion.

The model assumes perfect foresight. The excessive credit expansion is expected to continue; at the same time it is assumed that if the instant reserves are exhausted the monetary authorities will withdraw from the foreign exchange market, forcing the currency to float. The question addressed here is the timing of the collapse of the exchange rate regime. Without speculation, reserves will at some point fall to zero; anticipating this ultimate collapse, and to avoid losses at the time of the collapse, speculators will force a crisis before this point is reached. The question is, when?

The timing of the collapse is found at the point where a hypothetical “shadow floating rate,” reflecting market fundamentals, is equal to the fixed rate prevailing. So long as the fixed rate exceeds the shadow floating rate, the fixed rate regime is safe; beyond that point the fixed rate is not sustainable.

A first step, therefore, is to find the “shadow” floating rate, which could be expected to take the form,

Equation (102) says that the path of the exchange rate will follow the path of the money stock. Taking the rate of change of (102) and remembering that

Substituting equation (103) into (99) and rearranging,

Comparing equations (104) and (102) we see that

Returning to equation (102) we have

Noting that D(t) = D(o) + ut = Mt, substituting in equation (105) we have

The system collapses when the fixed rate S¯ equals the floating rate. We can find the collapse time tc from equation (106)

where R(o) is the initial stock of reserves.

Equation (107) says that the higher the initial stock of reserves R(o) the longer it will take before the collapse occurs; also the higher u, the rate of credit expansion, the quicker the collapse comes. Without speculation α/β = 0, the collapse occurs when reserves are run down to zero. As already indicated the collapse occurs sooner; how much sooner depends on the expression α/β. Why α/β? Noting the definitions here the expression reduces to

The larger the a0 the later the crisis; the larger a1 the sooner the crisis; a0 captures the base “utility” of holding money, a1 the “loss” from an expected devaluation.

Consider the speculator who can foresee reserves running down to zero. Without speculation there could be a loss imposed that speculators will wish to avoid. They avoid the loss from the discrete change by attacking the currency at the point where the transition to the float will be smooth (that is, where the shadow exchange rate equals the “fixed” rate, after which there is a steady devaluation).

To arrive at the stock of reserves just before the attack (that is, at tc—) we use equation (100)

Using equation (107) above

This gives

Extensions to the Base Model

The simple base model presented above can be extended and refined in various directions. First, more realistically, “uncertainty” can be introduced into the analysis, relaxing the assumption of perfect foresight. Second, one could also relax the assumption of perfect asset substitution. Third, one can use a more general model, with endogenous output, prices and the trade balance, accommodating at the same time various assumptions about substitution in goods and assets. Fourth, instead of assuming that once the fixed rate regime breaks down it is replaced by a float, one can work with alternative scenarios following the crisis: a new peg, a temporary float followed by a peg, and so on.

Many authors allow for uncertainty. Uncertainty can be introduced in various forms: there is uncertainty about future macro policies; about reserve limits (which could be negative in the presence of official borrowing) that will trigger the collapse; and about how lasting an exogenous disturbance is to the trade balance (for example, a terms-of-trade shock).

Willman (1987c) allows two possibilities with respect to the monetary policy rule: there is a probability of П that the monetary policy rule will be maintained and a probability 1 − П that the authorities will change the policy rule in conformity with the fixed rate regime. This modifies equation (107) to

If the probability is one we have the earlier result. If the probability is zero there is no speculative attack (u→ 0 tc→ ∞). The smaller the probability, the longer it will take before the collapse occurs. Variations on this basic theme (in the form of stochastic disturbances to domestic credit or to other determinants of the balance of payments, appear in Flood and Garber (1984), Grilli (1986), Obstfeld (1986), and Blanco and Garber (1986)). There are also circumstances when a system may be attacked (collapses) even though it is fundamentally viable (Obstfeld, 1986).

Krugman (1979) deals with one source of uncertainty: “incomplete knowledge … about how much of its reserves the government is willing to use to defend the exchange rate.” Willman (1987a) deals in detail with variations on this theme. Buiter (1986) accommodates the possibility of official borrowing to defend the exchange rate and its implications for the timing of the crisis. When full account is taken of the interest cost of servicing the debt (which is assumed to exceed the interest rate on reserves), his conclusions make good intuitive sense. “If the borrowing occurs just before the exchange rate regime would have collapsed absent the borrowing, the collapse is postponed. If the borrowing occurs long enough before the exchange rate regime would have collapsed in the absence of borrowing the collapse is brought forward” (that is, occurs earlier).

A widely cited paper by Giavazzi and Pagano (1985) poses the fundamental question: can an adjustable peg regime survive without capital controls? The authors present a simple model of two countries, one of which (A) has money stock growing at a positive rate, the other (B) at a zero rate. A’s inflation rate exceeds B’s; A devalues periodically vis-à-vis B. After each devaluation, its trade balance, activity, and its real exchange rate all improve, only to find this gradually eroded by inflation. At first they assume perfect foresight. Without capital controls A’s interest rate is equal to B’s interest rate plus the expected devaluation over the life of the asset. Obviously, interest rates on longer maturities will tend to fluctuate less. For short-term maturities at the approach of a realignment, interest rates will be “prohibitive.” Taking account of the “costs” of realignments and given the inflation divergence, the authors are able to calculate the size and frequency of exchange rate changes. Uncertainty about either inflation divergences or about the costs of realignments alters the predictable behavior of interest rates.

Capital controls protect A from the sharp fluctuations in interest rates. When applying this to France and Italy, this is reflected in divergences between interest rates in Euromarkets, which are free from control, and home rates. The former fluctuate more, rising more sharply at the approach of a devaluation.

The principal conclusions of the Giavazzi-Pagano paper are that without capital controls interest rates are likely to be more volatile. Nevertheless, in principle, the system remains viable because interest rate adjustments are an equilibrating mechanism that “eliminate the incentive for a run on reserves.” This last is a strong claim, distinguishing it from other such literature. To illustrate, as the public anticipates a devaluation, it will want to shift out of money; the authorities accommodate the public, say, by bond sales at interest rates that reflect these expectations; such bond sales avert the need to shift into foreign assets.

Wyplosz (1986) has a framework similar to Giavazzi and Pagano, with one country again expanding credit excessively. Now capital controls are in force: residents are not allowed to hold foreign currency assets (or to lend to nonresidents), but nonresidents are free from restrictions. There are only two assets, domestic and foreign money; interest rates are kept fixed, so the equilibrating mechanism of Giavazzi and Pagano is absent. Holdings by nonresidents of home currency are limited. There is again a cyclical pattern of real exchange rates and trade balances. Nonresidents monitor reserve levels, provoking a “crisis” when reserve levels equal nonresident holdings of money. The currency then devalues, setting off a new cycle.

To summarize, Wyplosz would argue that in the absence of capital controls the system would be viable only if “the monetary authorities maintain a sufficient degree of uncertainty … so as to force risk-averse speculators to commit only limited amounts of funds in the anticipation of a crisis.” Giavazzi and Pagano would concede that such uncertainty might limit speculative capital and, hence, interest rate volatility, but the regime might still be viable even with “perfect foresight” because of the interest-rate-equilibrating mechanism, which is absent in Wyplosz (see also Melitz and Michel, 1986).

Blackburn (1988) also deals with collapsing fixed rate regimes, but now in the context of a more general model in which there is imperfect substitution between domestic and foreign goods and between domestic and foreign assets. Output is fixed, but prices adjust in accordance with excess demand for goods. There is perfect foresight. Again excessive credit creation drives the system. The model is also flexible enough to accommodate various assumptions about the post-crisis exchange rate regime, including a permanent float or a temporary float followed by a new peg. The simplest model outlined above turns out to be a special case where there is (i) perfect asset substitution, (ii) perfect price flexibility, and (iii) a permanent float after the crisis.

Blackburn focuses primarily on the roles of price flexibility and capital mobility in determining the timing of the crisis, which occurs, as previously, at the point where the shadow rate equals the fixed rate. At the general level he is unable to say anything definitive—the expressions being now complicated—but he does demonstrate that when prices are highly flexible the higher the degree of capital mobility, the earlier the crisis occurs.

Willman (1987b) is also concerned with the timing of an attack. His concern, as in Blackburn, is in the context of a much more general model, in which not only is output flexible but wages are determined in several different ways. The roles of macro policy in generating crises are also explored. Willman focuses particularly on the role of wage determination in determining the timing of a collapse.

We noted earlier that Blackburn’s model incorporates various assumptions about the post-crisis exchange rate regime. Grilli (1986) and Blanco and Garber (1986) also deal explicitly with the case where, post crisis, a new fixed rate is actually established. This has the plausible characteristic that it must be greater than or equal to the rate that would have prevailed had there been a post-crisis float.

The Peg Versus the Adjustable Peg

We now compare these two regimes from the perspective of policy efficiency.

  • (1) With perfect asset substitution other things being equal, there is likely to be greater interest rate volatility under an adjustable peg than under a peg. Indeed, there is some question whether an adjustable peg regime can survive with free capital movements.

  • (2) If there is imperfect asset substitution with an adjustable peg but, on the other hand perfect or near-perfect asset substitution with a peg—a not unrealistic representation of the situation (compare, for example the Nordic countries with the Netherlands, Austria, or Denmark)—then:

    • the interest rate volatility gap will be less, other things being equal

    • monetary policy is now more independent with the adjustable peg.

  • (3) It is difficult to operate an adjustable peg regime if the going rate of inflation is persistently and significantly different from that of the trading partners. Such differences would bring continuing pressures for exchange rate adjustment, creating tensions and uncertainty. In such conditions a crawling peg or a float would be the more appropriate regime.

    One precondition, therefore, for the effective functioning of an adjustable peg regime is that the country maintain its inflation rate within manageable reach of its trading partners. Exchange rate adjustments could then be used, in principle, to correct for ex post unanticipated price disturbances or for fundamental imbalances.

  • (4) The fact that the exchange rate can be used periodically as a policy instrument can itself be disruptive to the economy. In most cases such exchange rate changes are anticipated, provoking large outflows or inflows of capital.

    Frequently, therefore, such movements of capital turn out to be one-way bets at the expense of central banks. A related point is that rumors of changes or suspicions of change may trigger large flows of capital, occasionally forcing the authorities into an unnecessary exchange rate adjustment.

    These comments suggest that an adjustable peg regime is likely to function more efficiently if such flows of capital can be monitored; hence, the perceived need to support such a regime with some control over capital movements. Without restrictions over capital movements, authorities might at times lose control over the money supply or over the exchange rate. For example, an expected devaluation, leading to huge outflows, might place the authorities in a serious dilemma: to restrict sharply the money supply and raise interest rates, or to go along with market sentiment, which may be misguided, and devalue.

    It is significant that in the countries where an adjustable peg has been in operation, restrictions on the movement of capital have been in force (for example, in Sweden, Norway, Finland, Ireland, Italy, France, Australia, and New Zealand).

  • (5) Policy efficiency is enhanced in a regime with capital controls in several distinct ways: the adjustable peg is more viable, there is wider latitude in the use of monetary policy, and interest rate volatility will be lessened. On the other hand there are costs associated with capital controls. The recognition that free capital movements may not, on balance, be beneficial to an economy has provoked proposals to limit short-term capital flows, notably by Tobin (1978) and Dornbusch (1982).

  • (6) In the adjustable peg regime the authorities are required to make the difficult judgment of when and how much to adjust the currency. If the market disagrees with their judgment on the size of the adjustment, further uncertainty is created. Even if the market goes along with the decision, there remains a question of whether the amount of the exchange rate adjustment will ultimately be adequate—taking account of relevant elasticities and spillovers into wages—to correct the imbalance.

  • (7) The role of real wage flexibility is critical to this analysis.

I assume in what follows perfect asset substitution. I begin with the assumption that there is symmetrical real wage flexibility in the sense that any change in the real wage rate (given unemployment) secured under, say, an adjustable peg regime can as readily be secured as under a peg. Later this assumption will be relaxed.

Consider first the case where there is no real wage flexibility (perfect wage indexation). In this case the exchange rate has no real effects. The economy is trapped in a classical dilemma: there is a single instrument (fiscal policy) but two targets, output and the current account balance. The adjustable peg regime has little apparent advantage, with two possible (but minor) exceptions: the exchange rate could be directed at price objectives (as noted earlier) or, if wages lag prices indefinitely, the authorities could, in principle, cheat workers; this indefinitely lowers the real wage rate, although at the expense of a permanently higher rate of inflation (see, however, the Annex to Section I).

Suppose that for one reason or another real wages fall (rise) with a devaluation (appreciation) and that the same outcome is possible through a wages policy under a peg. The wage rate now replaces the exchange rate as a policy instrument.

In the model used, the exchange rate—given the wage rate—and the wage rate—given the exchange rate—are perfect substitutes for one another to achieve real objectives. The solutions are:

or

Again it is evident that an adjustable peg regime offers no real advantage. A possible (but unconvincing) exception to this is that with such a regime a combination of wage and exchange rate policies might be used to stabilize the price level.

The idea that, if there is real wage symmetry in the two regimes, the adjustable peg has little if any advantage over the peg is, of course, well-known.

Friedman (1953), making his classic case for flexible rates, based one argument in favor of flexible rates on an asymmetry. Keynes also thought that real wages could be reduced by raising prices, but not so by lowering wages. Asymmetry could come from straight money illusion (particularly if a wage cut is required) or from the fact that wage policies are harder to negotiate and take longer than an adjustment to the exchange rate. On the other hand, it would be possible to argue that since a wage policy is at the heart of a peg, the authorities are in a good position to negotiate changes in real wages when appropriate.

To sum up, if there is asymmetrical real wage flexibility and a fundamental imbalance in the current account emerges, the adjustable peg regime has a clear advantage over the peg. With a peg, output will have to be sacrificed to secure current account objectives.

The Flexible Rate Regime

Output and the Current Account

I now consider the possibility of combining monetary and fiscal policy, under flexible rates, to achieve both output and current account objectives (see Genberg and Swoboda, 1987; and Boughton, 1989). I assume perfect asset substitution (see Table 6 for model solutions).

Suppose wage indexation were zero and prices were fixed (the Mundell-Fleming case). Fiscal expansion thus has no real effects on output, but at the same time it weakens the current account (on a one-for-one basis). On the other hand, monetary expansion has real effects on output but improves the current account. This case is represented in Chart 3 (Q being the static solution).

Chart 3.Assignment Under Flexible Rates: The Mundell-Fleming Case

Applying the assignment test: for a monetary expansion and a fiscal contraction that generates an equal current account surplus, output will increase with monetary expansion but remain unchanged for the fiscal contraction. Since fiscal policy cannot be used to secure output objectives, the only viable policy is to assign fiscal policy to the current account and monetary policy to the level of output. This assignment, using unbroken lines, is shown in Chart 3. The reverse, represented by broken lines, appears unstable.

Consider now the other extreme where indexation is perfect. Now monetary policy has no real effects; at the same time, it leaves the current account unchanged. Fiscal expansion, on the other hand, increases output and worsens the current account. In this extreme case we are left with a single instrument but two targets. This has an exact parallel with the adjustable peg regime with perfect wage indexation. Monetary policy, as exchange rate policy, has to be assigned to inflation.

The intermediate case is shown in Chart 4. The case where fiscal policy is assigned to the current account and monetary policy to the level of output is indicated by unbroken lines; the reverse assignment is indicated by broken lines. Q1 represents the case where the economy is fully employed but is running a current account deficit. An easy monetary policy and tight fiscal policy could keep output unchanged, but both policies serve to improve the current account. Q2 represents the case where the current account is in balance but the economy is in recession. Monetary and fiscal expansion can reduce unemployment while leaving the current account undisturbed.

Chart 4.Assignment Under Flexible Rates with Imperfect Wage Indexation

The conclusion would appear to be that we could not go too far wrong by assigning fiscal policy to the current account, at least over the medium term, and, provided there is imperfect indexation of monetary policy (that is, money stock changes) to output.

The conclusion is reinforced by empirical simulations for Australia of the effects of monetary and fiscal expansion, shown in Table 8. The table shows that an equal-output monetary and fiscal expansion will leave a large trade deficit for fiscal expansion while leaving the trade balance virtually unchanged for monetary expansion. (The “empirically based” current account schedule would thus be virtually horizontal, while the YY schedule would be negatively sloped.)

Table 8.Monetary and Fiscal Expansion Effects on Exchange Rates, Output, Prices, and Trade Balance in First Three Years (Australia)
Years
123
Fiscal expansion
Output0.440.600.65
Trade balance−0.72−0.69−0.67
Inflation−0.32−0.18−0.06
Exchange rate $/Aust.3.062.952.83
Monetary expansion
Output−0.630.260.03
Trade balance−0.020.000.01
Inflation0.360.360.22
Exchange rate $/Aust.−0.98−1.02−1.04
Source: Argy, McKibbin, Siegloff (1989).
Source: Argy, McKibbin, Siegloff (1989).

Output and Inflation

Instead of using monetary and fiscal policy to achieve current account and output objectives, the two policy instruments could, alternatively, be combined to achieve inflation and output objectives.

The idea that monetary and fiscal policies might be used for inflation-output objectives was first proposed by (Mundell 1971). He suggested that monetary policy should be assigned to the price level and fiscal policy to output. The theme was subsequently picked up in several papers (Casas, 1977; Argy and Salop, 1979; Perkins, 1979; and Sachs, 1985).

To develop this theme we replace the CA schedule with a PP schedule (which represents the combinations of monetary and fiscal policies that keep the price level unchanged).

The analytical basis can be summed up as follows:

Output (yr) and the consumer price index (p) are both written as functions of the two policy instruments money supply (mo) and government expenditure (gr). The slope of the YY schedule is П3031, while the slope of the PP schedule is -П3233 (see Table 6).

What are the signs of the coefficients? Consider again at one extreme a Mundell-Fleming type model with fixed domestic prices; in this case П30 > 0 П31 → 0. At the same time П32 > 0 while П33 < 0. (This owes to the fact that the consumer price index is determined only by the exchange rate, and monetary expansion forces down the currency while fiscal expansion strengthens it.) This case is illustrated in Chart 5, Case 1. With full information, Q1 will represent the appropriate policy mix. Applying the assignment rule it is evident in this case that monetary policy has to be assigned to output and fiscal policy to the price level. This assignment, which is stable, is demonstrated by the unbroken lines.

Chart 5.Monetary/Fiscal Policies for Output/Inflation Objectives

Consider the case where there is some, but not perfect, wage indexation. With perfect asset substitution П30 > 0 П31 > 0 П32 > 0 П33 < 0. The last result can be directly demonstrated from the definition of the consumer price index and from the money market.

Fiscal expansion raises output and appreciates the currency. From equation (115) with mo and rd¯ given, the home price level (pd) must fall. At the same time from equation (114), the appreciation reinforces the fall in the consumer price index (p).

This case is shown graphically as Chart 5, Case 2, where Q1 again represents the static full-information solution. Determining the appropriate assignment applying the rule proves to be complicated. The way the graph is drawn, either assignment would appear to be stabilizing.

If the economy were placed around Q2—where there is close to full employment but some inflation—the appropriate mix is to adopt a restrictive monetary policy and an easy fiscal policy. Such a combination maintains full employment, but on both counts reduces prices. On the other hand, if the economy were placed around Q3—where the inflation is about right but there is a recession—the appropriate policy mix is to adopt an easy monetary and fiscal policy. This mix maintains the price level but on both counts increases output.

At the other extreme, if wage indexation is perfect we have П30 = 0 П31 > 0 П32 = 1 П33 < 0. This case is represented graphically as Chart 5, Case 3. Now monetary policy has to be assigned to inflation, and fiscal policy to output.

Perhaps the principal conclusion to be drawn from our analysis is that as wage indexation increases, the YY schedule shifts from a vertical position toward, ultimately, a horizontal position; the higher is indexation, the greater the likelihood that the appropriate assignment will involve linking monetary policy to inflation and fiscal policy to output.

Table 7 sheds light on the potential signs of the coefficients, at least over a limited time horizon. It is noteworthy that fiscal expansion—at least in this model—lowers the price level, so the PP schedule is positively sloped. Both monetary and fiscal policies increase output. In the first year—for equal output monetary and fiscal expansion—the price fall for fiscal expansion exceeds the corresponding price rise for monetary expansion. This, however, is reversed by the second year.

The Managed Float Regime

Target Zones

Many rules for intervention have been proposed (see Argy, 1982; and Shelburn, 1984). In recent years the one most widely discussed and analyzed has been Williamson’s target zone proposal (Williamson, 1985) and its more recent version (the blueprint), jointly authored by Williamson and Miller (1987). We focus here on a simpler version of this last, with primary interest in its application to a smaller economy. (Section II evaluated management from the perspective of insulation.)10

There are two key proposals in Williamson-Miller that we can use here. First, exchange rate target zones would be announced. The center of the zone would be calculated on the basis of long-run, sustainable, full-employment, current account balances. Monetary policy would be used to defend these zones. Second, fiscal policy would be directed at achieving nominal income growth targets (domestic demand in their original version).11 Our model can easily be adapted to accommodate these two proposals.

Chart 6, based on the underlying equations of the model, attempts to represent this framework; YYN is the nominal income schedule (y + pd).

Chart 6.The Managed Float

What will the shape of this schedule be? Interestingly, monetary expansion will raise nominal income—defined in this way—but fiscal policy will leave nominal income unchanged, if the output coefficient in the money demand equation is unity (this is easily seen in equation (115)). The reason is simply that while fiscal expansion increases real output, it decreases prices proportionately. (If the target nominal income were defined as yr + p, where p is the consumer price index, fiscal expansion would have a perverse effect on nominal income because the exchange rate appreciation would exceed the fall in pd.) Thus, one could draw the YYN schedule vertically, as we have done.

The e − pd (C¯A/y¯) schedule shows the combinations of monetary and fiscal policies that maintain the target real exchange rate. The zones are indicated by broken lines; (a) represents the case of imperfect indexation, and (b) the case of perfect indexation.

The unbroken arrows represent a WM type of assignment (monetary policy for real exchange rates, fiscal policy for income), while broken arrows represent the reverse assignment.

In both cases the reverse assignment is the appropriate one. This is because fiscal policy has no effect on income but has effects on the real exchange rate, while monetary policy has effects on income and on the real exchange rate (provided indexation is imperfect). If indexation is perfect as in (b), assigning monetary policy alone to the real exchange rate may be self-defeating (as we have already seen in Section II) and may expose the economy to unbounded inflation or deflation.

Allowing, as well, fiscal policy to adjust to nominal income dampens this explosion, but the system remains unstable. Consider the case where the currency is strong and there is excessive income. Monetary expansion will not correct the real appreciation; fiscal contraction, however, will, so the economy is driven away from its income target but toward its real exchange rate target. If the currency is weak and income is excessive, a tight monetary policy will be combined with a tight fiscal policy; this drives the economy away from its real exchange rate target toward its income target.

The reverse assignment is stabilizing on both counts. For example, if the currency is strong and income excessive, a tight monetary policy combined with a tight fiscal policy will drive the economy closer to its two targets. The result here is that monetary policy only affects income (not so fiscal policy), while fiscal policy only affects the real exchange rate (not so monetary policy).

To sum up, the reverse assignment conforms better with the thinking that fiscal policy ought to be directed at current account objectives and monetary policy at “income objectives.” Yet, surprisingly, there is some econometric evidence—at least for the very large economies—that the “blueprint” assignment tends to outperform the reverse assignment (see Frenkel, Goldstein, and Masson, 1989; and Currie and Wren-Lewis, 1989). The issue remains a live one.

The idea of assigning monetary policy to an income objective is an old one (Tobin, 1983). For an aggregate demand “shock” (originating either in an expenditure or a money shock) one would expect policy in this form to be stabilizing. An aggregate supply shock that leaves nominal income unchanged (as it would in this model, with α9 = 1 rd = rf and mo fixed) would reduce employment and increase prices; this may not correspond with policy preferences. (On nominal income targeting, see Bean, 1983; and Frankel, 1989.)

A More General Evaluation

Disinflation, Credibility, and the Exchange Rate Regime

Suppose a country’s rate of inflation is “excessive” and the authorities have decided to adopt a restrictive money growth policy to reduce it to acceptable levels. In implementing such a policy, their key objective will be to minimize the so-called sacrifice ratio (that is, the ratio of the percent of GNP lost to the reduction in the inflation rate). What contribution can exchange rate policy make, in the disinflation process, toward minimizing this sacrifice ratio?12 This issue has been an important one in several stabilization programs, notably in Latin America, Israel, and in the context of the EMS (see Giavazzi and Giovannini, 1988; and Fischer, 1988a).

What exchange rate options does a country have in this context? A country may tie its exchange rate to a low (lower) inflation country, either by joining a union (such as the EMS)—as, for example, Ireland—or tie its exchange rate independently (say to the U.S. dollar); it can also allow the exchange to float during the disinflation (thus taking a sharp real appreciation); or it can manage the exchange rate in other ways, that is, by keeping the real exchange rate fixed or by reducing its rate of currency devaluation.

Consider the case of a country running a relatively high rate of inflation (say about 15 percent) that announces it will permanently tie its currency to a country with an inflation rate of close to zero. Suppose the authorities are totally credible; domestic credit expansion drops; in a one-period contract model wages also adjust to the changed expectation; in principle, the sacrifice ratio could be zero.13 With two-or-more period contracts, or if credibility is weak, the sacrifice ratio will be positive.

Fischer (1988a), using an extended open economy macro model, examines in detail the role of the exchange rate in a disinflation policy. Fischer’s model allows for two-period contracts and intermediate imported inputs. He compares the sacrifice ratio for the case of a market-oriented exchange rate and the case where the real exchange rate is stabilized. It turns out that the difference between the two sacrifice ratios depends on the parameters of the model (for example, the effect of the real exchange rate on supply—imported inputs—and on demand and the interest sensitivity of money demand) so no general conclusion is possible.

To conclude, structural coefficients are important in calculating the sacrifice ratio for different regimes, but the most important factor is credibility. If pegging substantially enhances credibility, pegging will reduce sacrifice ratios. (For evidence of how joining the EMS may have reduced inflationary expectations in some countries, see Giavazzi and Giovannini, 1988.)

Impotence of Anticipated Policies?

If policies are assigned as predictably as indicated in the analysis, will such policies be anticipated and hence lose their effectiveness, beyond a short period? This represents the New Classical attack on activism.

There are at least three well-known counterattacks one may mount. First, governments may have privileged or better information, either about the structure or about forecasting. Second, and more important, the private sector is unlikely to know the policy rule. Third, and most important, wage-price stickiness allows even fully anticipated monetary and fiscal policies to have real effects for some time.

To some extent we have accommodated some of the New Classical objections by dealing in detail with cases where wage indexation is perfect. The results here would parallel New Classical cases. The reality is almost certainly that while fully anticipated policies will have weaker real effects than unanticipated policies, they nevertheless will continue to have some real effects for a while.

Policy Effectiveness and Exchange Rate Regime

At the start of Section III we noted that one exchange rate regime would be preferred to another if (a) more policy instruments are available, (b) these instruments have more powerful effects on the target variables, and (c) there is greater certainty attaching to the effects of each instrument.

We now attempt to summarize our findings, recalling these three criteria. We focus on the adjustable peg, the peg, and flexible rates.

Consider (a). Suppose we are concerned with achieving output and current account objectives (that is, real objectives), and suppose perfect asset substitution holds. If wage indexation is perfect only fiscal policy is, in principle, available in all three regimes. The exchange rate (adjustable peg) and money (flexible rates) will have only price effects.

At the other extreme, suppose there is some real wage flexibility and this is symmetrical, whatever the regime. Now the peg is almost certainly superior to the adjustable peg (since a wages policy can achieve the same results). With flexible rates, it can be shown that a reduction (an increase) in nominal wages will have identical real effects as a monetary expansion (contraction), so a wages policy is an alternative to a monetary policy not a supplement. Each regime has thus two potential instruments: the adjustable peg regime has the exchange rate (or wages) and fiscal policy; the peg has wages and fiscal policy; and the flexible rate regime has fiscal and monetary (or wages) policy.

If there is some real wage flexibility but an asymmetry, the adjustable peg comes into its own. The peg is now clearly disadvantaged, losing one instrument. If there is imperfect asset substitution with the adjustable peg, it has three potential instruments—monetary, fiscal, and exchange rate policy. It is, thus, at an advantage.

Consider now (b). Over a policy horizon of say two to three years fiscal policy is almost certainly stronger under fixed rates than under flexible rates; on the other hand, monetary policy is almost certainly stronger under flexible than fixed rates (even with imperfect asset substitution).

Finally, consider (c). The medium-to-longer-run effects of both monetary and fiscal policies remain uncertain. Once allowance is made for portfolio balance, wealth effects, and interest payments on foreign and government debt, the outcomes can be ambiguous and complicated. Over time, real exchange rate outcomes tend to reverse themselves for both monetary and fiscal policies.

Over a shorter time horizon, it seems almost certain that there is greater certainty about the effects of monetary and fiscal policies (most particularly the latter) under fixed than under flexible rates. Multipliers are more complicated in the latter case. (This is easily verified from Table 6.) The additional effects of exchange rate changes confound the outcomes. (As an exercise, one could compare fiscal policy outcomes across econometric models for fixed and flexible rates. The range of results should be greater in the latter case.)

One conclusion that emerges is that fiscal policy appears to come into its own under fixed exchange rates, being both more powerful and more certain in its effects. Recalling, however, the Mitterrand experiment in 1981–82 one cannot be too certain. There are short-and medium-term consequences especially for the current account, which sooner or later will need to be addressed. Of course, if the exchange rate is also available as an effective instrument—it was only available to a limited degree to Mitterrand—some of these constraints might not apply.

Finally, one may also want to “check” to see if policy has been stabilizing. Some empirical studies attack this question (see Argy, 1988). A recent econometric study by Frenkel, Goldstein, and Masson (1988), which simulates the effects of more stable policies in large countries, is encouraging from this perspective (that is, more stable policies appear to weaken performance).

Comment

Morris Goldstein

Victor Argy has characteristically provided us with a most useful paper. It surveys a wide range of theoretical and empirical studies; develops propositions from clearly specified models, rather than merely asserting them; and addresses a host of relevant policy issues. Moreover, the survey is highly readable, which makes it an attractive teaching tool.

Fortunately—at least in my role as discussant—I was able to find several issues that I thought either deserved more attention or a different twist. I will discuss these issues in the reverse order from that of Argy’s paper—dealing first with policy efficiency, next with insulation, and then with inflation discipline.

I. Policy Efficiency and the Assignment Problem

In analyzing policy efficiency, Argy emphasizes the so-called “assignment problem,” that is, which policy instruments should be assigned to which policy targets. He also endorses Mundell’s well-known “principle of effective market classification,” which directs us to assign the policy instrument to the target on which it has the largest relative effect.

I would like to suggest that the conventional treatment of the assignment problem can be misleading for policy on at least three counts. One difficulty lies with treating the current account as an undifferentiated target. Non-zero current account positions arise from a variety of sources, some of which are “good” imbalances that require no policy intervention, and some “bad” imbalances that do require intervention. For example, intercountry differences in the age distribution of populations can be expected to yield different life-cycle-induced private saving patterns; if not offset by parallel domestic investment opportunities, these will show up in current account imbalances. Yet there is no presumption that these underlying private saving decisions are suboptimal. Contrast this with a situation where the government is borrowing abroad to finance a consumption spree and is thereby establishing an unsustainable net liability position. The point is that one needs to know the origin of a current account imbalance before one can decide if it needs correction.

The second difficulty flows directly from the first. If the origin of the current account imbalance matters, the first-best policy is to correct bad imbalances at their source. Simple assignment rules—whether of the traditional or reversed variety—run the risk of barking up the wrong tree. For example, if household saving in country X is too high because of a favorable tax incentive, then a policy rule that says to adjust government expenditure to eliminate the current account imbalance misses the root cause of the problem. I understand that a key assumption in the assignment literature is that governments operate with only limited information and make decisions on the basis of the kind of disturbances that they expect to dominate on average. Still, there will be situations where policy can and should aim for correcting a bad imbalance at the source.

My third quibble is that assignment rules do not typically consider the flexibility and reversibility of policy instruments. This is crucial for fiscal policy. The fact is that fiscal policy is much less flexible than monetary policy in virtually all industrial countries. In addition, almost all large industrial countries—except the United Kingdom—have experienced significant increases in the government debt-to-GNP ratio over the past ten years. This means that both the feasibility and desirability of using fiscal policy for short-run stabilization purposes is questionable—at least until longer-run fiscal discipline is better assured.

I might also add that taking account of the relative inflexibility of fiscal policy can alter the results of simulation-based studies of alternative policy rules. Specifically, my own work with Jacob Frenkel and Paul Masson, using the Fund’s MULTIMOD model, suggests that those policy strategies—like the Williamson-Miller blueprint—that assume very activist fiscal policy perform less well when fiscal policy is allowed to achieve internal or external balance only with, say, a one-year lag.

II. Insulation

On the broad issue of insulation, I agree with Argy that it is difficult to say much that is definitive. Again in our own simulation studies, we have found that policy rules that do well in the face of certain types of disturbances usually do poorly against other types—with no single rule clearly dominating against the typical basket of disturbances of the past two decades.

Having said that, we also find not surprisingly that when countries differ markedly from their neighbors in economic structure and are subject to real economic shocks, the nominal exchange rate can be a valuable ally in changing relative prices and in facilitating adjustment. In this connection, a key issue for the European Monetary System (EMS) is how real economic shocks will be accommodated without capital controls and within a presumably strengthened commitment to greater stability of nominal exchange rates. Presumably, if the nominal exchange rate does less, other adjustment mechanisms will have to do more. Will factor mobility—or real wage flexibility more generally—take on an expanded role? Will the tax and transfer system be called on to offset partially country-or region-specific shocks?

III. Inflation Discipline

Last but not least, I come to the effects of the exchange rate regime on the authorities’ ability to deliver low inflation. Given time and space constraints, let me just touch on two aspects of this issue.

One interesting question is what the authorities of a relatively high-inflation country can do to enhance the credibility of their exchange rate commitment—once they’ve already made the decision to “tie their hands” by pegging to the currency of a country with a better inflation-fighting reputation. A popular answer is that credibility for such a “hard-currency policy” depends on convincing the private sector that the authorities are willing to bear any output and employment losses associated with relying exclusively on reductions in wages and prices to achieve needed changes in real exchange rates. This can be a slow and costly process—albeit in the end perhaps a less costly method of disinflation than the alternatives. But a second important factor may come into play. This is convincing the public that the authorities have a lot “at stake” beyond the exchange rate in keeping their commitment. Recent work by Giavazzi and Giovannini (1990) suggests, for example, that exchange rate commitments in the EMS are strengthened by the implications of exchange rate variability for the survival of the Common Agricultural Policy and for intra-European Community trade flows—as well as by larger regional and political integration objectives. This line of reasoning might also suggest that exchange rate commitments should be softer across than within regional currency areas because there is less at stake across zones in keeping them. I wish the paper had given more attention to what determines the credibility of an exchange rate peg.

My second point concerns the effect of the exchange rate regime on fiscal policy discipline. While I appreciated the Sargent and Wallace perspective on links between monetary and fiscal policies, I was not convinced that it illustrated the possibility of how an exchange rate target defended by monetary policy can send a “false signal” for fiscal discipline. Suppose we are in a target zone scheme and assume that the home country adopts an expansionary fiscal policy that puts appreciating pressure on its currency. To keep the rate from leaving the zone, monetary policy must then loosen—in effect, monetizing the deficit and sending a false signal. Indeed, when we performed some counter-factual historical simulations to shed light on how a target zone scheme would have performed in the 1970s and 1980s, we found that this false-signal problem at a time of rising U.S. fiscal deficits produced a poor U.S. inflation outcome. This potential problem has now been recognized: whereas first-generation target zone proposals spoke only of monetary policy, second-generation ones have added a rule to rein in fiscal policy. I would also note that a policy coordination regime where authorities negotiate the stance and mix of monetary and fiscal policies directly—rather than passing through the filter of the exchange rate—can in principle avoid this false-signal problem.

References

    AizenmanJoshua andJacob A.Frenkel“Optimal Wage Indexation, Foreign Exchange Intervention and Monetary Policy,”American Economic ReviewVol. 15No. 3 (Nashville, Tennessee: June1985) pp. 40223.

    AllenPolly R. andPeter B.Kenen“Asset Markets, Exchange Rates and Economic Integration” (Cambridge, United Kingdom: Cambridge University Press1980).

    ArgyVictor“Exchange Rate Management in Theory and Practice,”Princeton Studies in International Finance No. 50 (New Jersey: Princeton UniversityOctober1982).

    ArgyVictor“A Post-War History of the Rules vs Discretion Debate,”Banca Nazionale del Lavoro (Italy: June1988) pp. 147–77.

    ArgyVictor (1989a) “International Financial Deregulation: Some Macroeconomic Implications,”Australia-Japan Research Centre Pacific Economic Papers No. 168February1989.

    ArgyVictor(1989b) “Choice of Exchange Rate Regime for a Smaller Economy: A Survey of Some Key Issues,”original paper presented at Brussels ConferenceOctober1989.

    ArgyVictor“The Transmission of Foreign Disturbances under Different Exchange Rate Regimes,”Journal of Banking and Finance (Amsterdam: North-Holland1990).

    ArgyVictor andJoanneSalop“Price and Output Effects of Monetary and Fiscal Policy under Flexible Exchange Rates,”Staff PapersVol. 26No. 2 (Washington: International Monetary FundJune1979) pp. 22456.

    ArgyVictor andJoanneSalopWarwickMcKibbin andEricSiegloff“Exchange Rate Regimes for a Small Economy in a Multi-Country World,”forthcomingPrinceton Studies in International Finance (Princeton, New Jersey: Princeton University1990).

    ArgyVictor andJoanneSalop andMichaelPorter“The Forward Exchange Market and the Effects of Domestic and External Disturbances Under Alternative Exchange Rate Systems,”Staff PapersVol. 19No. 3 (Washington: International Monetary Fund1972) pp. 50332.

    BanaianKingLeroy O.Laney andThomas D.Willett“Central Bank Independence: An International Comparison,”Federal Reserve Bank of Dallas Economic Review (Dallas: March1983) pp. 113.

    BarroRobert J. andDavid B.Gordon“Rules Discretion and Reputation in a Model of Monetary Policy,”Journal of Monetary EconomicsVol. 12 (Amsterdam: North-Holland1983) pp. 10121.

    BeanCharles R.“Targeting Nominal Income: An Appraisal,”Economic JournalVol. 93 (Cambridge, England: 1983) pp. 80619.

    BlackburnKeith“Collapsing Exchange Rate Regimes and Exchange Rate Dynamics: Some Further Examples,”Journal of International Money and FinanceVol. 7 (Guilford, England: 1988) pp. 37385.

    BlanchardOliver J. andLawrence H.Summers“Hysteresis and the European Unemployment Problem,”in National Bureau of Economic Research Macroeconomics Annualed. byStanleyFischerforthcoming.

    BlancoHerminio andPeter M.Carber“Recurrent Devaluation and Speculative Attacks on the Mexican Peso,”Journal of Political EconomyVol. 94No. 11 (Chicago, Illinois: 1986) pp. 14866.

    BoughtonJames“Policy Assignment Strategies with Somewhat Flexible Exchange Rates,”inBarryEichengreenMarcusMiller andRichardPorteseds.,Exchange Rate Regimes and Macro Economic Policy1989.

    BoyerRussell“Optimal Foreign Exchange Market Intervention,”Journal of Political EconomyVol. 86 (Chicago, Illinois: 1978) pp. 104556.

    BransonWilliam H. andWillem H.Buiter“Monetary and Fiscal Policy with Flexible Exchange Rates,”inJ.S.Bhandari andBluford H.Putnameds.,Economic Interdependence and Flexible Exchange Rates (Cambridge, Massachusetts: MIT Press1983).

    BrunnerKarl“Fiscal Policy in Macro Theory: A Survey and Evaluation,”inR.W.Hafered.,The Monetary vs. Fiscal Policy Debate (Roman and Allanheld1986).

    BuiterWillem H.“Borrowing to Defend the Exchange Rate and the Timing and Magnitude of Speculative Attacks,”Centre for Economic Policy Research Discussion Paper No. 95 (London: CEPRFebruary1986).

    BuiterWillem H. andMarcusMiller“Costs and Benefits of an Anti-Inflationary Policy: Questions and Issues,”inVictorArgy andJohn W.Nevileeds.,Inflation & Unemployment Theory Experience and Policy Making (London: Allen and Unwin1985).

    CanzoneriMatthew B. andDaleHenderson“Is Sovereign Policymaking Bad?”Carnegie-Rochester Series on Public PolicyVol. 28(1988) pp. 93140.

    CasasFrancisco R.“Capital Mobility and Stabilization Policy under Flexible Exchange Rates: A Revised Analysis,”Southern Economic JournalVol. 43(April1977) pp. 152837.

    CurrieDavid“Some Criticisms of the Monetary Analysis of Balance of Payments Correction,”Economic JournalVol. 86 (Cambridge, England: September1976) pp. 50822.

    CurrieDavid andSimonWren-Lewis“Evaluating Blueprints for the Conduct of International Macro Policy,”American Economic Association Papers and ProceedingsVol. 79No. 2 (Nashville, Tennessee: May1989) pp. 26469.

    DanielBetty C.“Optimal Foreign-Exchange Policy for a Small Open Economy,”Journal of International Money and FinanceVol. 4No. 4 (Guilford, England: December1985) pp. 52336.

    DarbyMichael R.“Some Pleasant Monetarist Arithmetic,”Federal Reserve Bank of Minneapolis Quarterly Review (Minneapolis, Minnesota: Spring1984) pp. 1521.

    DevereuxMichael“The Optimal Mix of Wage Indexation and Foreign Exchange Market Intervention,”Journal of Money Credit and BankingVol. 20No. 3 (Columbus, Ohio: August1988) pp. 38192.

    DornbuschRudiger“Equilibrium and Disequilibrium Exchange Rates,”Zeitschrift Für wirtschafts-und sozialwissenschaftenVol. 102(1982) pp. 57399.

    DornbuschRudiger“Special Exchange Rates for Capital Account Transactions,”National Bureau of Economic Research Working Paper No. 1659 (Cambridge, Massachusetts: NBERJuly1985).

    DornbuschRudiger“The European Monetary System, the Dollar and the Yen,”inFrancescoGiavazzi andS.Micossieds.,The European Monetary System (Cambridge, England: Cambridge University Press1988).

    DornbuschRudigerJeffreyFrankel“The Flexible Exchange Rate System: Experience and Alternatives,”inS.Bornered.,International Finance and Trade in a Polycentric World (London: Macmillan1989).

    DrazenAllen“Monetary Policy Capital Controls and Seigniorage in an Open Economy,”conference proceedings “Monetary Regimes and Monetary Institutions: Issues and Perspectives in Europe”1988.

    DrazenAllen andElhananHelpman“Stabilization with Exchange Rate Management,”Quarterly Journal of Economics (Cambridge, Massachusetts: November1987) pp. 83555.

    FeldsteinMartin S.“The Welfare Cost of Permanent Inflation and Optimal Short-Run Economic Policy,”Journal of Political EconomyVol. 87No. 4 (Chicago, Illinois: 1979) pp. 74968.

    FischerStanley F.“Stability and Exchange Rate Systems in a Monetary Model of the Balance of Payments,”inR.Alibered.,The Political Economy of Monetary Reform (New Jersey: Allanheld, Osmun, and Montclair1977).

    FischerStanley F.“Towards an Understanding of the Costs of Inflation: II,”The Costs and Consequences of Inflation Carnegie-Rochester Conference Series on Public PolicyKarlBrunner andAllanMeltzereds.,Vol. 15 (Amsterdam: North Holland1981) pp. 542.

    FischerStanley F. (1988a) “Real Balances, the Exchange Rate and Indexation: Real Variables in Disinflation,”Quarterly Journal of Economics (Cambridge, Massachusetts: February1988) pp. 2749.

    FischerStanley F. (1988b) “Rules Versus Discretion in Monetary Policy,”National Bureau of Economic Research Working Paper Series No. 2518 (Cambridge, Massachusetts: NBERFebruary1988).

    FischerStanley F. andFrancoModigliani“Towards an Understanding of the Real Effects and Costs of Inflation,”Weltwirtschaftliches Archiv Band 114 (Tübingen: 1978) pp. 81133.

    FloodRobert F.“Capital Mobility and the Choice of Exchange Rate System,”International Economic ReviewVol. 20 (Philadelphia, Pennsylvania: 1979) pp. 40516.

    FloodRobert F.PeterGarber“Collapsing Exchange-Rate Regimes,”Journal of International EconomicsVol. 17 (Amsterdam: 1984) pp. 113.

    FrankelJeffrey“Obstacles to International Macroeconomic Policy Coordination,”Princeton Studies in International Finance No. 64 (Princeton, New Jersey: Princeton UniversityDecember1988).

    FrankelJeffrey“An Analysis of the Proposal for International Nominal Income Targeting,”unpublished manuscript1989.

    FrenkelJacob A. andJoshuaAizenman“Aspects of the Optimal Management of Exchange Rates,”Journal of International EconomicsVol. 13 (Amsterdam: 1982) pp. 32156.

    FrenkelJacob A. andJoshuaAizenman andAssafRazin“The Limited Viability of Dual Exchange Rate Regimes,”National Bureau of Economic Research Working Paper No. 1902 (Cambridge, Massachusetts: NBERApril1986).

    FrenkelJacob A. andJoshuaAizenman andAssafRazinMorrisGoldstein andPaulMasson“International Coordination of Economic Policies: Scope, Methods and Effects”in Economic Policy Coordination (Washington: International Monetary Fund1988).

    FrenkelJacob A. andJoshuaAizenman andAssafRazinMorrisGoldstein andPaulMasson“Simulating the Effects of Some Simple Coordinated Versus Uncoordinated Policy Rules,”National Bureau of Economic Research Working Paper No. 2929 (Cambridge, Massachusetts: NBER1989).

    FriedmanMilton“The Case for Flexible Exchange Rates,”in Essays in Positive Economics (Chicago, Illinois: University of Chicago Press1953) pp. 157203.

    FriedmanMilton“The Optimal Supply of Money,”in The Optimal Supply of Money and Other Essays (Chicago, Illinois: 1969).

    GenbergHans“Exchange Rate Management and Macroeconomic Policy: A National Perspective,”paper for the 90th Anniversary Symposium of the Scandinavian Journal of EconomicsHanasaari (Helsinki, Finland: June12–141988).

    GenbergHans andAlexanderSwoboda“The Current Account and the Policy Mix Under Flexible Exchange Rates,”International Monetary Fund Working Paper 87/70 (Washington: IMFOctober1987).

    GiavazziFrancesco andAlbertoGiovannini“The Role of Exchange Rate Regime in a Disinflation: Empirical Evidence on the European Monetary System,”inFrancoGiavazziS.Micossi andMarcusMillereds.,The European Monetary System (Cambridge, England: Cambridge University Press1988).

    GiavazziFrancesco andAlbertoGiovanniniMarcoPagano“Capital Controls and the European Monetary System,”Euromobiliare Occasional PaperNo. 1 (Milan: June1985).

    GoldsteinMorris“Have Flexible Exchange Rates Handicapped Macroeconomic Policy?”Special Papers in International Finance No. 14 (Princeton, New Jersey: Princeton University1980).

    GoldsteinMorris“The Exchange Rate System: Lessons of the Past and Options for the Future,”International Monetary Fund Occasional Paper No. 30 (Washington: IMFJuly1984).

    GrilliVittorio“Buying and Selling Attacks on Fixed Exchange Rate Systems,”Journal of International EconomicsVol. 20 (Amsterdam: 1986) pp. 14356.

    GrilliVittorio“Seigniorage in Europe,”National Bureau of Economic Research Working Paper No. 2778 (Cambridge, Massachusetts: NBER1988).

    GrosDaniel“Dual Exchange Rate Regimes in the Presence of Arbitrage Activity: A Stochastic Framework,”Centre for European Policy Studies Working Paper No. 33 (Brussels: CEPSFebruary1988).

    GuidottiPablo E.“Insulation Properties Under Dual Exchange Rates,”Canadian Journal of EconomicsVol. 21No. 4 (Toronto: November1988) pp. 799813.

    GylfasonThorvaldur andMichaelSchmid“Does Devaluation Cause Stagflation?”Canadian Journal of EconomicsVol. 16 (Toronto: November1988) pp. 64154.

    HaaparantaPertti“Dual Exchange Markets and Intervention,”Canadian Journal of EconomicsVol. 21No. 4 (Toronto: November1988).

    HamiltonClive“Contractionary Devaluation: A Review of the Literature,”Working Paper No. 158The Australian National University(March1988).

    HansonJames A.“Contractionary Devaluation, Substitution in Production and Consumption and the Role of the Labour Market,”Journal of International EconomicsVol. 14 (Amsterdam: 1983) pp. 17989.

    HornHenrik andTorstenPersson“Exchange Rate Policy, Wage Formation and Credibility,”European Economic ReviewVol. 32 (Amsterdam: October1988) pp. 162136.

    KawaiMasahiro“Exchange Rates, the Current Account and Monetary Fiscal Policies in the Short Run and in the Long Run,”Oxford Economic PapersVol. 37No. 3 (Oxford, England: September1985) pp. 391425.

    KinsellaR.P.“A Note on the New Cambridge Economics,”Rivista Internazionale Di Scienze Economiche E Commerciali No. 2 (Italy: February1975) pp. 14351.

    KrugmanPaul“A Model of Balance-of-Payments Crises,”Journal of Money Credit and BankingVol. 11No. 3 (Ohio: Ohio State University PressAugust1979) pp. 32531.

    KrugmanPaulLanceTaylor“Contractionary Effects of Devaluation,”Journal of International EconomicsVol. 8 (Amsterdam: 1978) pp. 44556.

    KydlandFinn E. andEdward C.Prescott“Rules Rather Than Discretion: The Inconsistency of Optimal Plans,”Journal of Political EconomyVol. 85No. 3 (Chicago, Ilinois: June1977) pp. 47391.

    LachlerUlrich“Stabilization Policy in an Open-Economy Equilibrium Model,”Weltwirtschaftliches ArchivBand 120, Heft 3 (Tübingen: 1984) pp. 40323.

    LevinJay H.“The J-Curve, Rational Expectations, and the Stability of the Flexible Exchange Rate System,”Journal of International EconomicsVol. 15 (Amsterdam: 1983) pp. 23951.

    MarstonRichard C.“Wages, Relative Prices and the Choice Between Fixed and Flexible Exchange Rates,”Canadian Journal of EconomicsVol. 15No. 1 (Toronto: February1982) pp. 87103.

    MarstonRichard C. andStephen J.Turnovsky“Imported Materials Prices, Wage Policy and Macroeconomic Stabilization,”Canadian Journal of EconomicsVol. 18No. 2 (Toronto: 1985) pp. 27384.

    MelitzJacques andPhilippeMichel“The Dynamic Stability of the European Monetary System,”Centre for Economic Policy Research Discussion Paper No. 96 (London: CEPR1986).

    MillerPreston J. (1983a) “Budget Deficit Mythology,”Federal Reserve Bank of Minneapolis Quarterly Review (Minneapolis, Minnesota: Fall1983) pp. 113.

    MillerPreston J. (1983b) “Higher Deficits Lead to Higher Inflation,”Federal Reserve Bank of Minneapolis Quarterly Review (Minneapolis, Minnesota: Winter1983) pp. 819.

    MinfordPatrick andDavidPeelRational Expectations and the New Macroeconomics (Oxford, England: Martin Robertson1983).

    MundellRobert A.“The Appropriate Use of Monetary and Fiscal Policy for Internal and External Stability,”Staff PapersVol. 9 (Washington: International Monetary FundMarch1962) pp. 7079.

    MundellRobert A.“The Dollar and the Policy Mix: 1971,”Essays in International Finance No. 85 (Princeton, New Jersey: Princeton UniversityMay1971).

    NiehansJurg“Some Doubts About the Efficacy of Monetary Policy Under Flexible Exchange Rates,”Journal of International EconomicsVol. 5No. 3 (Amsterdam: 1975) pp. 27581.

    ObstfeldMaurice“Rational and Self-Fulfilling Balance-of-Payments Crises,”American Economic ReviewVol. 76No. 1 (Nashville, Tennessee: March1986) pp. 7281.

    ParkinMichael“Domestic Monetary Institutions and Deficits,”in DeficitsBuchananJ.M.RowleyC.K. andTollisonR.D.eds. (1986) pp. 31037.

    PerkinsJames O.N.The Macroeconomic Mix to Stop Stagflation (London: Macmillan1979).

    PhelpsEdmund S.“Inflation in the Theory of Public Finance,”Swedish Journal of EconomicsVol. 75 (Stockholm: 1973) pp. 6782.

    PooleWilliam“Optimal Choice of Monetary Policy Instruments in a Simple Stochastic Macro-Model,”Quarterly Journal of EconomicsVol. 84 (Cambridge, Massachusetts: May1970) pp. 197216.

    RoperDon E. andStephen J.Turnovsky“Optimal Exchange Market Intervention in a Simple Stochastic Macro Model,”Canadian Journal of EconomicsVol. 13 (Toronto: May1980) pp. 296309.

    SachsJeffrey“The Dollar and the Policy Mix 1985,”Brookings Papers on Economic ActivityVol. 85No. 1 (Washington: The Brookings Institution1985) pp. 11785.

    SargentThomas J. andNeilWallace“Some Unpleasant Monetarist Arithmetic,”Federal Reserve Bank of Minneapolis Quarterly Review (Minneapolis, Minnesota: Fall1981) pp. 117.

    ShelburnMarsha R.“Rules for Regulating Intervention Under Managed Float,”Studies in International Finance No. 55 (December1984).

    SmithR.P.“Demand Management and the ‘New School’,”Applied EconomicsVol. 6 (1976) pp. 193205.

    SpaventaLuigi“Seigniorage: Old and New Policy Issues,”European Economic ReviewVol. 33 (Amsterdam: 1989) pp. 55763.

    SummersLawrence H.“Optimal Inflation Policy,”Journal of Monetary EconomicsVol. 7 (Amsterdam: March1981) pp. 17594.

    SwanTrevor W.“Longer-Run Problems of the Balance of Payments,”in The Australian Economy: A Volume of ReadingsH.W.Arndt andW. MaxCordeneds. (Melbourne: Cheshire Press1955).

    TobinJames“A Proposal for International Monetary Reform,”Cowles Foundation for Research in Economics Discussion Paper No. 506 (New Haven, Connecticut: Yale UniversityOctober1978).

    TobinJames“Monetary Policy: Rules, Targets, and Shocks,”Journal of Money Credit and BankingVol. 15No. 4 (Ohio: Ohio State University PressNovember1983) pp. 50618.

    TurnovskyStephen J.“The Relative Stability of Alternative Exchange Rate Systems in the Presence of Random Disturbances,”Journal of Money Credit and BankingVol. 8 (Ohio: Ohio State University Press1976) pp. 2950.

    TurnovskyStephen J. (1983a) “Wage Indexation and Exchange Market Intervention in a Small Open Economy,”Canadian Journal of EconomicsVol. 21No. 4 (Toronto: 1983) pp. 57492.

    TurnovskyStephen J. (1983b) “Exchange Market Intervention Policies in a Small Open Economy,”inJ.Bhandari andL.B.Putnameds.,Economic Interdependence and Flexible Exchange Rates (Cambridge, Massachusetts: MIT Press1983) pp. 286311.

    TurnovskyStephen J.“Exchange Market Intervention Under Alternative Forms of Exogenous Disturbances,”Journal of International EconomicsVol. 17 (Amsterdam: 1984) pp. 27997.

    TurnovskyStephen J.“Optimal Exchange Market Intervention: Two Alternative Classes of Rules,”inJ.S.Bhandaried.,Exchange Rate Management Under Uncertainty (Cambridge, Massachusetts: MIT Press1985).

    WilliamsonJohned.,Exchange Rate Rules (London: Macmillan1981).

    WilliamsonJohn“The Exchange Rate System,”Institute for International Economics Policy Analyses in International Economics No. 5 (Washington: IIE1985).

    WilliamsonJohn andMarcus H.Miller“Targets and Indicators: A Blueprint for the International Coordination of Economic Policy” (Washington: Institute for International EconomicsSeptember1987).

    WillmanAlpo (1987a) “The Collapse of the Fixed Exchange Rate Regime with Sticky Wages and Imperfect Substitutability Between Domestic and Foreign Bonds,”Monistettuja Tutkimuksia Research Papers (Helsinki, Finland: Bank of FinlandJune1987).

    WillmanAlpo (1987b) “Speculative Attacks on the Currency with Uncertain Monetary Policy Reactions,”Monistettuja Tutkimuksia Research Papers (Helsinki, Finland: Bank of FinlandAugust1987).

    WillmanAlpo (1987c) “Balance-of-Payments Crises and Monetary Policy Reactions in a Model with Imperfect Substitutability Between Domestic and Foreign Bonds,”Monistettuja Tutkimuksia Research Papers (Helsinki, Finland: Bank of FinlandNovember1987).

    WyploszCharles“Capital Controls and Balance of Payments Crises,”Journal of International Money and FinanceVol. 5 (Guilford, England: 1986) pp. 16779.

    GiavazziFrancesco andAlbertoGiovannini“Can the European Monetary System Be Copied Outside Europe?”inWilliam H.BronsonJacob A.Frenkel andMorrisGoldsteineds.,International Policy Coordination and Exchange Rate Fluctuations (Chicago: University of Chicago Press1990).

This is a shortened version of the original paper presented at the October 1989 Brussels Conference. The original version may be obtained from the author at Macquarie University, Sydney, New South Wales 2109, Australia.

There has been renewed interest in recent years in the dual exchange rate regime (see Guidotti, 1988; Haaparanta, 1988; Frenkel and Razin, 1986; Dornbusch, 1985; and Gros, 1988).

This is not the place to undertake an evaluation of the framework. For further analysis see Miller (1983a, 1983b), Minford and Peel (1983), Darby (1984), and Brunner (1986).

With unsterilized intervention, there would have been an offsetting capital inflow.

This is easily seen from equations (51) and (52). With y fixed,

With pd = e we have

rd will fall but p will rise equally, leaving the real interest rate unchanged.

Strictly, we would take variances of equation (74), but nothing much is altered by this simpler procedure.

For a review of empirical work bearing on insulation, see Argy (1989a).

We could also have included the budget balance as a target, but this would have imposed severe restrictions on the use of fiscal policy.

Obviously the more “equilibrating” a regime is (in the earlier sense), the less the need for policy adjustment.

For evidence bearing on assignment and on the effects of a devaluation on the trade balance, see Argy (1989a).

For an analysis of how an intervention policy of leaning against the wind might affect exchange rate performance in an environment where “J” curves operate, see Levin (1983) and Argy (1989a).

In the original proposals for the large countries there is a third leg: the average level of world interest rates should be adjusted so as to achieve a target growth of nominal demand for the participating countries.

We do not address the question of the role of fiscal policy in the disinflation process. (See Section I; see also Buiter and Miller, 1985; and Drazen and Helpman, 1987.)

A one-off adjustment for a stock increase in money demand must also be made.

    Other Resources Citing This Publication