Chapter

CHAPTER 5 Exchange Rate Policy in Advanced Commodity-Exporting Countries: Australia and New Zealand

Author(s):
International Monetary Fund
Published Date:
September 1990
Share
  • ShareShare
Show Summary Details
Author(s)
Adrian Blundell-Wignall and Robert G. Gregory* 

I. Introduction

Exchange rate policy in Australia and New Zealand in the 1970s and 1980s was characterized by extremes. During the 1970s and the first half of the 1980s, both countries operated pegged or managed exchange rate regimes (see Table 1) while the major countries moved to floating rates. Australia and New Zealand were in good company, in that a majority of relatively small countries (and some large ones) continued to fix or manage their exchange rates. However, in the mid-1980s both moved to freely floating exchange rates (Australia in December 1983 and New Zealand in March 1985). This move occurred shortly before the larger countries began placing greater emphasis on exchange rate stabilization through policy coordination.

Table 1.Exchange Rate Regimes and Exchange Rate Variability

(variance of quarterly percent change)1

Bilateral U.S. dollar

exchange rate
Effective

exchange rate
Australia
pegged to a single currency

with periodic realignment (sterling until Nov. 1971 and the U.S. dollar until Sept. 1974)
10.286.12
fixed to a basket of currencies (Sept. 1974 to Nov. 1976)13.8913.25
crawling peg vis-à-vis the U.S. dollar (Nov. 1976 to Dec. 1983)6.866.16
overall pre-float (to Dec. 1983)11.669.30
floating (Dec. 1983 to present)37.5138.39
New Zealand
pegged to a single currency (sterling until Nov. 1971 and the U.S. dollar to July 1973)5.040.75
fixed to a basket of currencies with periodic step adjustments (July 1973 to June 1979)19.359.67
crawling peg vis-à-vis a basket of currencies (June 1979 to June 1982)5.990.66
fixed to a basket of currencies with periodic step adjustments (June 1982 to March 1985)41.4728.78
overall pre-float (to March 1985)19.779.81
floating (March 1985 to present)38.8628.05
Source: OECD Balance of Payments Section.

Bilateral exchange rates vis-à-vis the U.S. dollar and in trade-weighted terms (OECD weights), quarterly averages.

Source: OECD Balance of Payments Section.

Bilateral exchange rates vis-à-vis the U.S. dollar and in trade-weighted terms (OECD weights), quarterly averages.

This paper reviews the experiences of Australia and New Zealand within the context of the international literature on exchange rates and policy.

Two quite different sets of exchange rate policy issues may be distinguished:

  • (1) the extent to which the exchange rate should influence the broad settings of monetary policy; and

  • (2) the role of day-to-day intervention in the foreign exchange market. Authorities can choose the extent to which these transactions affect domestic monetary conditions through open market operations (that is, the extent to which short-run intervention is sterilized).

The first of these relates to the broad issue of the choice of exchange rate regime (floating or “light” management, as opposed to pegging or “heavy” management). The second relates to strategies for countering excessive volatility and “disorderly” markets. This can be of a very short-term nature involving the central bank giving “tone” and “breadth” to the market. At times the market may be subject to occasional periods of “thinness”—characterized by few active traders—leading to erratic swings in the exchange rate as the random arrival of new orders meets little demand. An occasional role for more substantial intervention policies might also arise when markets lack direction and where inefficiencies result in “bubbles” or bandwagon behavior.

The structure of this paper reflects these broad distinctions. Section II examines fluctuations of exchange rates and the weight that should be attached to these in the broad formulation of monetary policy. Section III examines short-run volatility and the role of intervention in countering disorderly markets. In Section IV, we review some historical episodes concerning exchange rate policy in Australia and New Zealand. Finally, in Section V, we offer some concluding remarks.

II. Exchange Rate Fluctuations and Monetary Policy

A Brief Historical Perspective

Trends in the real and nominal exchange rates of Australia and New Zealand from March 1970 to December 1988 are shown in Chart 1. Two alternative definitions of the real exchange rate are employed, one based on the consumer price index and the other on relative unit labor costs. Each measure is broken down into its component parts: the nominal exchange rate and the ratio of the two price indexes.

Chart 1.Real and Nominal Exchange Rates, and Relative Prices

Note: The first panel of chart above and on preceding page shows the bilateral exchange rate against the U.S. dollar, domestic consumer and U.S. consumer price indexes. The second panel shows the effective exchange rates and domestic and weighted foreign consumer price indexes. The third panel is as in the first, except domestic relative to U.S. unit labor costs are used to deflate the real exchange rate. The fourth panel is as in the second, except domestic relative to trade-weighted foreign unit labor costs are used.

Source: OECD, Balance of Payments Division.

Mussa (1986) pointed out that movements in relative goods prices between most countries are typically smooth, and that short-run variability in the real exchange rate mainly reflects movements in the nominal rate. This pattern is also reflected in the data for Australia and New Zealand, both before and after the floating of exchange rates. In both periods, the variance of the real rate and the nominal rate exceeds that for relative price levels. This is true regardless of whether the consumer price index or unit labor cost definition is used. The trend decline in the ratio of foreign to domestic prices in both countries reflects relatively higher inflation in Australia and New Zealand during most of the period.

The period of generalized floating within the Organization for Economic Cooperation and Development (OECD) area as a whole, from early 1973 on, was associated with wider fluctuations in nominal and real exchange rates than had been expected by the early advocates of floating (Shafer and Loopesko, 1983). Attempts to manage Australian and New Zealand exchange rates in a more volatile world environment met with only limited success. For both countries, the nominal and real exchange rate has been subject to pronounced and sustained movements that, in broad terms, do not appear consistent with purchasing power parity (PPP). It is also noteworthy that some of the largest exchange rate swings in the two countries are quite similar:

  • Both Australia and New Zealand’s nominal and real exchange rates were revalued in the early 1970s, particularly in 1972–73, when both countries pegged rates against a single currency.

  • Subsequently, both countries managed to stabilize some definition of the nominal or real exchange rate until the early 1980s. In Australia this was the bilateral real rate against the U.S. dollar. This stability partly reflected the formal arrangement of a crawling peg against the dollar. But it is particularly notable that the real effective exchange rate was not stabilized. This implied sharp movements in Australia’s bilateral exchange rates against other major trading partners: in particular, Japan and Europe. New Zealand, on the other hand, stabilized its effective exchange rate. To the extent that exchange rates were adjusted to offset inflation differentials, it appears to have been with consumer price movements in mind. Greater loss of competitiveness on the unit labor cost definition of the real exchange rate in the late 1970s or early 1980s was apparent in both countries.

  • Exchange rates have nevertheless been more volatile—quarter to quarter—since the floating of rates in both countries (see Table 1). Real exchange rates fell sharply in the mid-1980s and rose markedly from the second half of 1986 on.

The Choice of Exchange Rate Regime

Two strands of thought can be distinguished in the literature on choice of an optimal exchange rate regime: one based on the economic characteristics of countries, and one based on the nature of the stochastic shocks faced by different economies.

The first of these is associated with the theory of optimum currency areas and is related to whether fixed or floating rates generate the most efficient allocation of resources. Countries will typically choose to float their exchange rates if they have a large gross domestic product (GDP); a low degree of openness (share of traded goods in GDP); a high inflation differential with other countries; a high degree of integration in international capital markets; and substantial diversification in traded goods (Heller, 1978; and Holden, Holden, and Suss, 1979).

This literature is not particularly useful for considering the optimal exchange rate regime for Australia and New Zealand. While New Zealand may well possess many of the characteristics suited to an optimal currency area arrangement with Australia,1 the situation is unclear for either country vis-à-vis the rest of the world. Having a small GDP, an average degree of openness, and concentration of trade in commodities constitutes a weak case—according to standard resource allocation arguments—for a free float. On the other hand, inflation performance and a high degree of integration in world capital markets do argue for greater flexibility in Australia and New Zealand.2

However, Australia and New Zealand differ from most other countries in the OECD in their heavy concentration in commodity trade. This has more in common with the structure of a number of non-OECD developing countries. Studies of exchange rate regimes for these countries—for example, by Dreyer (1978)—reverse the finding on trade diversification: higher concentration makes it more likely for the country to float its exchange rate. While foreign price disturbances will influence the exchange rate and affect the allocation of resources between tradable and nontradable industries in a presumably “costly” manner,3 there may be macroeconomic inflation considerations that argue against pegging the exchange rate. If macroeconomic considerations do go in the opposite direction—and if they carry a greater weight with policymakers—it would not be surprising to find periods in which commodity-exporting countries favored a greater degree of flexibility of rates regardless of the resource allocation costs.

This brings us to the second strand of the literature: the nature of stochastic disturbances in open-economy macro models. This literature is diverse, with much of it being motivated by the important contribution of Poole (1970).4 When looked at in the context of the optimal degree of exchange rate management in open economies, however, there are some contrasts in the literature. Boyer (1978) showed that in a model with only traded goods and money:

  • use of monetary policy to fix the exchange rate is optimal for stabilizing output if shocks arise in the money market;

  • use of fiscal policy is more optimal for output stabilization if shocks arise only in the goods market; and

  • it will always be optimal to follow a managed float if the economy is subject to both goods and money market shocks.

Frenkel and Aizenman (1982) examine this issue in a model based on the monetarist approach to the exchange rate, which assumes that PPP holds. In their model, the objective is to stabilize the discrepancy between consumption and expected income. They find that the desirability of exchange rate flexibility increases the larger the variances of shocks to the demand for, or supply of, money relative to foreign prices and to PPPs. Frenkel and Aizenman extend this analysis to distinguish an economy that produces traded and nontraded goods, and that is a price taker in the world market for traded goods. They show that the desirability of exchange rate flexibility diminishes the higher the relative share of nontraded goods in the economy, and the lower the elasticities of demand and supply for the two goods. Lower relative price elasticities imply higher internal price adjustment to return the economy to PPP, which, in turn, requires less exchange rate flexibility.

None of these papers, however, deals precisely with an economy that has broad characteristics matching those of Australia and New Zealand:

  • both are commodity-exporting countries (the share of commodities in exports for both countries was about 70 percent in the late 1980s) that import mainly manufactured goods;

  • both are mainly price takers on world markets (with the exception of wool in the case of Australia), and hence may be subject to external terms-of-trade changes between importables and exportables; and

  • neither economy is subject to high price flexibility in the nontraded sector. Neither Boyer nor Frenkel and Aizenman assume sticky price adjustment, for example.

Nor have the above papers dealt explicitly with monetary policy and exchange rate management where the authorities’ objective is to stabilize inflation. The issue of optimal exchange rate management in an economy with these characteristics is dealt with in the following section.

Disturbances in a Commodity-Exporting Economy

The basic features of a small commodity-exporting country are captured in the following model:

Small-Country Commodity-Exporting Model

Supply

Demand

Prices

Asset markets

Reaction function

Notation

Yx = the supply of exports

Yn = supply of nontraded goods

Y = total output

Dn = nontraded goods demand

D = total demand

Dm = import demand

P = the goods price

Pd = the demand deflator

Pn = the price of nontraded goods

Px = export prices

Pm = import prices in foreign currency (assumed to be always equal to unity in subsequent analysis)

E = the exchange rate (domestic currency per unit of foreign currency)

M = money

R = the interest rate

R* = the foreign interest rate

π = the constant rate of growth of the money supply (which is assumed to determine inflation expectations exogenously, as underlying or core inflation)

Eo = a constant nominal exchange rate target

Mo = a constant term

Lower case letters denote natural logarithms (except for time t).

All parameters are positive.

A bar denotes an exogenous variable.

A hat denotes an equilibrium value.

Constant terms other than those to be used in subsequent analysi (for example, in linearizing equation (2) in logarithms) are ignored

Equations (1) and (2) define the supply of exports and nontraded goods in terms of relative prices. The domestic economy produces one good that is sold on world markets at a fixed price and in the domestic market at a monopoly price. Equation (3) explains overall demand in terms of disposable income and interest rates, while equations (4) and (5) explain the split between nontraded goods demand and import demand in terms of relative prices. Equations (6) and (7) define output and demand prices, where λ is the equilibrium share of nontraded goods in output. Equation (8) explains nontraded goods prices in terms of excess demand. Equations (9), (10), and (11) are standard equations for money demand, asset market equilibrium, and real exchange rate expectations. Finally, equation (12) represents the authorities’ money supply reaction function. If the intervention parameter δ is set at zero, the money supply grows at the constant rate π. AS δ approaches infinity, the exchange rate is set at E0.

The small commodity-exporting model has an important feature with respect to the relationship between the real exchange rate (defined in terms of the import price index) and the relative price of traded goods. For simplicity, we assume initially that β = 0 and that the real exchange rate is normalized by setting Pm = 1. The assumption that β = 0 is motivated by two considerations: (1) that we are more interested in the international aspects of the analysis, with a primary focus on relative prices and the role of the exchange rate in the transmission of monetary policy; and (2) that recent experience suggests that it may not be a highly unrealistic assumption for economies with deregulated and competitive financial systems, such as Australia and New Zealand.5 The normalization of import prices to unity implies that terms of trade changes will be equated with export price movements in subsequent analysis—a realistic feature in light of the more volatile commodity price cycle that drives the latter. The relationship between the terms of trade and the real exchange rate can be seen by considering the steady state of the model, where real money balances and the real exchange rate are constant, domestic inflation grows at the rate π, and the exchange rate depreciates at the rate π (foreign inflation is zero since Pm = 1).

From equation (8), we see that the steady state requires the goods market to clear (dn = yn). Substituting equations (1), (2), (3), (4), (6), and (7) into equation (8) and setting domestic inflation equal to π, the equilibrium real exchange rate is given by:6

That is, the equilibrium real exchange rate is determined by the terms of trade, given the elasticities of demand and supply for importables and exportables and the share of nontraded goods in the economy.7 If the terms of trade strengthen, the equilibrium value of the real exchange rate must appreciate (competitiveness s = e − pn falls) and vice versa. This, in turn, will have implications for the nominal exchange rate, the supply and demand for non-traded goods, and the domestic rate of inflation, depending on various parameters and speeds of adjustment in the model.8 This result motivates much of the discussion in Section II.

So far the discussion has been in terms of the properties of the deterministic version of the commodity-exporting country model. Two further steps are necessary to explore exchange rate policy in the context of the stochastic shocks literature: (1) the introduction of relevant stochastic shocks into the model’s equations; and (2) the specification of an objective for the authorities, who use exchange rate management to minimize the loss function derived from deviations from this objective.

For illustrative purposes the economy is assumed to be subject to stochastic shocks from two sources: changes in the exogenous terms of trade and shifts in the money demand function. The constant term in the money demand equation is assumed to adjust according to:

The terms of trade are assumed to adjust according to:

We assume the authorities’ objective is to ensure that domestic inflation is in line with their inflation target π, the “normal” rate of monetary growth in the reaction function equation (12). In this sense, it is the objective function of a central bank charged with containing inflation at some predetermined core rate. The loss function is assumed to be given by:9

The problem for the authorities can be viewed from the perspective of the base case of the economy moving along a steady-state path, but occasionally being subject to stochastic shocks.

The solution for the divergence of inflation from the target rate π for the two stochastic disturbances to the steady state is:10

Notice that the value of φ depends positively on the value of the intervention parameter δ, the degree to which the authorities attempt to achieve their exchange rate objective. The expected value of the loss function (equation (16)) using expression (17) is then:

where ρ is the correlation coefficient between the shocks, denoted by σ12/(σ1, σ2)1 and where σ 12 is the covariance between the shocks. Differentiating expression (18) and setting it equal to zero, we find the optimal value of the intervention term:

If the exogenous shocks are independent ρ = 0, the optimal value of the intervention reduces to

Equations (1820) contain a number of interesting results, the intuitive content of which is worth pursuing further.

Exchange Rate Policy in Response to Disturbances

The results are easiest to interpret for the case of zero covariance between the shocks (ρ = 0, that is, equation (20)). It is also important to bear in mind that we are considering the limiting case of a competitive, financially deregulated economy, where the exchange rate is the main transmission mechanism of monetary policy (β = 0). The implications of relaxing this assumption will be touched upon later.

The value of ϕ* varies positively with the ratio of the variance of money demand shocks to the variance of terms of trade shocks. Changes in the stance of monetary policy aimed at stabilizing the exchange rate should be stronger when money demand shocks are relatively more important. In the limit, as the variance of terms-of-trade shocks approaches zero, ϕ* approaches infinity, as does the intervention parameter δ, so that fixed rates would be the optimal exchange rate policy to avoid inflation disturbances. If foreign interest rate shocks are included in the analysis, they would be equivalent to money demand shocks because of the assumption of zero interest elasticity of expenditure in the domestic economy. In this case the impact of the foreign interest rate change on the exchange rate may be offset by altering the stance of domestic monetary policy. Since the exchange rate is the only transmission mechanism of monetary policy, there would be no impact on domestic inflation.

Where the variance of terms-of-trade shocks is relatively more important, the value of ϕ* declines. But it is interesting to note that in the limiting case where the variance of money demand shocks approaches zero, the value of ϕ* remains positive (Λ21. This may be consistent with zero intervention in the face of terms-of-trade shocks (δ = 0), leaning against the wind (δ > 0), or what we might call leaning with the wind (δ < 0).

To understand this range of possibilities, it is worth exploring how a terms-of-trade shock would lead to an inflation disturbance, and what the optimal policy response might be. Consider the case of a terms-of-trade increase. There are three fundamental mechanisms through which the impact on domestic inflation arises: (i) the direct income effect of the terms-of-trade rise (increase in the purchasing power of domestic income), which increases with the relative size of the traded goods sector (1 - λ); (ii) the substitution effect in domestic supply, as more goods are sold on the foreign market and less on the domestic market; and (iii) the substitution effect of domestic demand toward imported goods, a potentially offsetting influence that derives from the exchange rate appreciation induced by the shift in the terms of trade (which also moderates the substitution effect in supply discussed above).

The first two effects are inflationary. The induced appreciation of the exchange rate is deflationary. Indeed, for countries where terms-of-trade shocks are relatively more important, this latter mechanism is at the crux of preferring a floating rate to a pegged exchange rate. A pegged rate would leave the first two effects dominant—a situation that has posed problems historically for both Australia and New Zealand (see below). If the inflation impact is offset exactly by the deflationary impact of the rising exchange rate, no intervention would be needed to maintain the inflation target (the case of δ = 0). However, this would be something of a coincidence. Even if terms-of-trade effects are dominant, it will normally be optimal to manage the exchange rate, at least to some extent, if the monetary authority has an inflation target.

Where the inflation impact of the terms of trade is large and the induced appreciation of the exchange rate small, it may pay to lean with the wind: to drive the exchange rate higher than would be the case under pure floating to offset the inflationary effect of the terms of trade (enforced overshooting). The inflationary impact of the terms of trade will be greatest when:11

  • the share of traded goods in GDP (1 - λ) is high, implying a higher direct income effect and a higher relative importance of the substitution effect in supply, both of which act to increase the demand for and reduce the supply of nontraded goods;

  • the elasticity of substitution in supply η is high, which increases the extent to which the supply of nontraded goods declines; and

  • the substitution effect in demand (α) induced by any tendency for the exchange rate to appreciate is small.

The appreciation of the exchange rate will be smaller the lower the impact on interest rates of the initial rise in income, that is, the lower the income elasticity of money demand (γ).

Conversely, where the inflation impact of the terms-of-trade rise is small and the induced overshooting of the exchange rate is large, it would pay to lean against the wind to avoid disturbances to the inflation objective in the downward direction (an unnecessary decline in activity).

If the assumption that β = 0 is relaxed, so that interest rate changes affect private demand directly, the exchange rate is no longer the sole transmission mechanism of monetary policy. We do not focus on this case because complexity diminishes the transparency of the results, and our analysis seeks to show how the parameters of monetary policy might look if the exchange rate were becoming the more dominant transmission mechanism. Nonetheless, the thrust of our earlier results would change in the following ways:

  • The optimal degree of intervention would still depend on the relative importance of terms-of-trade and money demand shocks.

  • Where money demand shocks are dominant, it is not clear that a fixed exchange rate would ever be optimal for achieving an inflation target. This is because the interest rate level necessary to fix the exchange rate in response to a money demand shock would have its own direct impact on the excess demand for nontraded goods in the short term, which may be inconsistent with the inflation target.

  • Similarly, fixing the exchange rate in response to foreign interest rate shocks would no longer avoid inflation disturbances. Higher or lower domestic interest rates necessary to achieve the fixed exchange rate would again have their own impact on domestic demand.

  • Where terms-of-trade shocks are dominant, leaning with or against the wind may still be optimal, but the extent of this intervention would be less than where β = 0.

Floating or Pegging for Australia and New Zealand?

In this section we explore empirically the broad question of whether, in the cases of Australia and New Zealand, a floating exchange rate (including policies of leaning with or against the wind) is likely to be more appropriate than policies of pegging or heavy exchange rate management. We proceed on the assumption that the stochastic shocks literature is relevant and we follow up the earlier theoretical discussion, which suggests that the issue depends on the relative importance of the variance of monetary versus terms-of-trade shocks. We then return to the question of resource allocation criteria for an exchange rate regime.

The optimal approach to exchange rate policy is always likely to be a managed float, where the source of shocks is identified and the optimal degree of intervention determined according to known trade and other behavioral elasticities. Unfortunately, the actual behavior of the real exchange rate is more akin to that of a random walk:

where s(t) is the real exchange rate in period t, and u(t)an independent and identically distributed error term at time t. If the error term were composed of monetary shocks u1(t), terms-of-trade shocks u2(t), as well as other possibilities, the initial shock problem for the authorities is one of signal extraction. The second task is to determine the optimal degree of intervention (or resistance) to exchange rate movements given the information about the shocks and the behavioral parameters of the economy.

Neither of these steps is easy in practice, particularly when policy (of necessity) must be conducted on a day-to-day basis. It is important in these circumstances to have a view on the basic nature of real exchange rate dynamics in order to formulate an overall approach to policy: free floating, or leaning with or against the wind, and opposed to heavy management or pegging. Deviations from the basic regime can then be made if signal extraction problems can be solved.

The basis of the theoretical result in the earlier section was that terms-of-trade shifts induced by the commodity price cycle change the equilibrium real exchange rate, whereas monetary shocks do not. Fixing the nominal exchange rate in the face of a commodity price shock leads to prolonged adjustment through (sticky) domestic inflation, which is necessary to ensure that the real exchange moves toward its new equilibrium level. Permitting the exchange rate to float freely in the face of a domestic monetary shock causes a deviation of the real exchange rate from its existing equilibrium (say, purchasing power parity) level. This leads to inflation as a part of the process of reversion to PPP. In choosing a broad exchange rate regime with an inflation objective in mind, it is important to know the likely relative importance of these different types of effects. Is the equilibrium real exchange rate subject to important changes because of the variance of commodity prices, or is the assumption of a constant equilibrium real exchange rate a reasonable guide for policy in practice?

We investigate this issue by testing two simple alternative hypotheses about exchange rate dynamics against the null hypothesis that the real exchange rate follows a random walk, as in equation (21):

Hypothesis 1: that deviations from PPP are cumulative and there is a tendency toward reversion to a constant equilibrium level.

Hypothesis 2: that the real exchange rate reverts to a moving equilibrium determined by the terms of trade.12

These hypotheses can be tested with procedures suggested by Engle and Granger (1987). The first is a test of the null hypothesis of no co-integration against the alternative that domestic prices, foreign prices, and the nominal exchange rate are co-integrated. We first pre-test to ensure that all the components of the real exchange rate possess a unit root. We then impose the theoretical restrictions suggested by PPP on the co-integrating vector for these three variables—hence defining the real exchange rate—and test the null hypothesis that it possesses a unit root. The augmented Dickey-Fuller (ADF) unit root statistic is employed to test this null hypothesis. Since u(t) can be driven by an ARMA (p,q) process, one estimates:

where v(t) is iid(0, σ2) and tests the null hypotheses that a = 0 using critical values reported in Fuller (1976, Table 8.5.2).

Results for tests of the first hypothesis are shown in the first column of Tables 2 and 3, using quarterly data for both Australia and New Zealand. Four lagged innovations are chosen for the results shown in the tables, although lags between 1 and 12 were also tested. The rejection region for a one-tailed test of a unit root at the 5 percent level—where a constant term is included in the regression—is given by smaller (more negative) values than −2.9 for most sample periods shown in the tables, and -3.0 for the shorter post-float sample period for Australia. In none of the cases is the t-statistic smaller than the critical value. The results for the ADF statistic suggest that the logarithm of the real effective exchange rate can be characterized as an integrated process of order one. Over both managed and floating exchange rate regimes, the data do not appear to display any tendency for the real exchange rate to revert to a constant PPP equilibrium level in either Australia or New Zealand.

Table 2.Regressions of Real Exchange Rate and the Terms of Trade: Australia(full sample period 1970 Q1-1988 Q4)
Dep. Var.ΔssΔsΔsΔs
χ1.05(11.55)
χ(−1).05(2.7)
s(−1)−.07(−1.7)−.04(−3.1)
ε(−1)−.044(−3.1)−.046(−3.5)
Δs(−1).52(1.0).07(.6).08(.7)
Δs(−2).38(.7).00(0).02(.2)
Δs(−3).11(1.8).18(1.3).21(1.6).21(1.8)
Δs(−4)−.37(−.6)−.12(−.8)−.09(−.7)
Δχ(−1).00(0).00(.1)
Δχ(−2).00(.6).00(.7)
Δχ(−3)−.15(.3)−.00(−.30)
Δχ(−4).00(.1)−.00(−.1)
const..31(1.6)4.55(452.0).20(3.1).00(.1)
σ.04195.07968.00901.00896.00854
DW1.95.301.951.951.84
χ.63(8.44)
χ(−1).03(2.0)
s(−1)−.12(−1.9)−.06(−2.8)
ε(−1)−.055(−2.8)−.063(−3.6)
Δs(−1).54(.8).04(.27).04(.3)
Δs(−2).73(1.1).05(.36).05(.4)
Δs(−3)1.28(1.8).22(1.4).21(1.5).25(1.9)
Δs(−4)−.66(−.9)−.13(−.8)−.13(−.9)
Δχ(−1).00(.9).00(.9)
Δχ(−2).00(.4).00(.4)
Δχ(−3).00(.5).00(.5)
Δχ(−4).00(0)−.0(−.0)
const..54(1.9)4.60(517.7).25(2.8).001(.8)
σ.02981.04939.00619.00612.00578
DW1.96.382.022.021.85
χ1.44(4.7)
χ(−1)−.17(−1.5).22(2.1)
s(−1)−.11(−2.6)
ε(−1)−.10(−2.6)−.10(−3.3)
Δs(−1).60(−5).11(.4).17(.6)
Δs(−2).33(.3)−.33(−.9)−.16(−.5)
Δs(−3).96(.8)−.89(−.3).11(.3)
Δs(−4)−.27(−.2)−.32(−.9)−.26(−.8)
Δχ(−1)−.00(−.2).00(−.5)
Δχ(−2).00(1.3).00(.6)
Δχ(−3).00(.2)−.00(−2)
Δχ(−4).00(.7).00(.7)
const..76(1.5)4.51(175.3).49(2.6)−.002(−.4)
σ.06733.09459.01471.01469.01173
DW2.03.342.272.292.09
Note: t-values in parenthesis.s (quarterly) for Australia is the logarithm of the nominal effective exchange rate deflated by the domestic relative to trade-weighted foreign consumer price index.χ (quarterly) is the logarithm of the terms of trade, National Accounts definition, for Australia.χ (monthly, commodity price version) for Australia is the logarithm of the Reserve Bank of Australia Commodity Price Index, defined in terms of SDRs, total index rural and nonrural components.Source: OECD Balance of Payments Division.
Note: t-values in parenthesis.s (quarterly) for Australia is the logarithm of the nominal effective exchange rate deflated by the domestic relative to trade-weighted foreign consumer price index.χ (quarterly) is the logarithm of the terms of trade, National Accounts definition, for Australia.χ (monthly, commodity price version) for Australia is the logarithm of the Reserve Bank of Australia Commodity Price Index, defined in terms of SDRs, total index rural and nonrural components.Source: OECD Balance of Payments Division.
Table 3.Regressions of Real Exchange Rate and the Terms of Trade: New Zealand(full sample period 1970 Q1-1988 Q3)
Dep. Var.ΔssΔsΔsΔs
χ.42(7.9)
χ(−1).03(3.3)
s(−1)−.13(−1.7)−.05(−26)
ε(−1)−.055(−3.1)−.06(−3.4)
Δs(−1).47(.8).26(2.1).29(2.4).27(2.3)
Δs(−2).37(.6)−.21(−1.6)−.16(−1.2)
Δs(−3)41(.7).23(1.8).26(2.1).26(2.2)
Δs(−4)−1.25(−2.2)−.39(−3.1)−.34(−2.8)−.36(−3.4)
Δχ(−1)−.0009(−3.0)−.001(−2.8)−.001(−2.5)
Δχ(−2).0007(2.2).0007(2.2).0005(1.8)
Δχ(−3)−.0006(−2.0)−.0005(−1.8)−.0005(−1.7)
Δχ(−4)−.00(−1.0)−.00(−1.0)
const..61(1.8)4.59(599.1).22(2.5)−.004(−.4)
σ.03819.05512.00709.00717.00715
DW1.86.461.931.881.79
χ.37(7.4)
χ(−1).03(3.2)
s(−1)−.08(−1.0)−.05(−2.5)
ε(−1)−.058(−2.9)−.066(−3.5)
Δs(−1).98(1.4).19(1.4).26(1.8).20(1.6)
Δs(−2)−.85(−1.2)−.23(−1.6)−.14(−1.1)
Δs(−3)1.57(1.8).19(1.0).27(1.5)
Δs(−4)−.80(−.9)−.21(−1.1)−.11(−.6)
Δχ(−1)−.00(−.5)−.00(−.2)
Δχ(−2)−.00(−.0).00(.1)
Δχ(−3).00(.6).00(.7)
Δχ(−4)−.00(−1.8)−.00(−1.6)
const..35(.9)4.59(612.4).24(2.5)−.001(−.7)
σ.03225.04881.00644.00655.00644
DW1.91.411.841.852.00
Note: s (quarterly) for New Zealand is the logarithm of the nominal effective exchange rate deflated by the domestic relative to trade-weighted foreign consumer price index.χ (quarterly) is the logarithm of the terms of trade, National Accounts definition, for New Zealand.Source: OECD Balance of Payments Division.
Note: s (quarterly) for New Zealand is the logarithm of the nominal effective exchange rate deflated by the domestic relative to trade-weighted foreign consumer price index.χ (quarterly) is the logarithm of the terms of trade, National Accounts definition, for New Zealand.Source: OECD Balance of Payments Division.

The second hypothesis is a test of the null hypothesis of no co-integration against the alternative that the real exchange rate and the terms of trade are co-integrated or, equivalently, are subject to an error correction process. A simple approximate method is the co-integrating regression Durbin-Watson test (GRDW), shown in the second column of Tables 2 and 3. The co-integrating regression of the real exchange rate and the terms of trade is:

where x(t) is the logarithm of the quarterly terms of trade. If the Durbin-Watson statistic exceeds 0.386, the null hypothesis (of non-co-integration) is rejected in favor of finding co-integration at the 5 percent level, while the critical value is 0.322 at the 10 percent level (see Engle and Granger, 1987). For the full sample period using quarterly Australian data, co-integration is almost accepted at the 10 percent level. For the pre-float and post-float sample periods, co-integration appears to be accepted at the 10 percent level. In the case of New Zealand, the CRDW test finds co-integration in all cases at the 5 percent level.

However, given the low power of the CRDW test for a unit root against a highly autoregressive alternative, a related approach is to estimate an error correction mechanism in the dynamic model: error correction and co-integration being broadly equivalent concepts. To establish that the joint distribution of the real exchange rate s and the terms of trade ϳ is an error-correction system, three models were estimated.

First, we estimated an unrestricted vector autoregression (VAR) of the change in the logarithm of the real exchange rate on lags of the dependent variable and changes in the logarithm of the terms of trade, as well as the lagged levels of the real exchange rate and the terms of trade. Results for this model are shown in the third columns of Tables 2 and 3. Four lagged innovations are tested for both Australia and New Zealand (the results not being sensitive to the inclusion of more lags). It is remarkable that the lagged levels of the terms of trade and the real exchange rate were all of the appropriate signs and sizes for an error correction term for both Australia and New Zealand. Individual parameters were significant at the 5 percent level for both countries.

The second model imposes the error correction restriction (the lagged level of the residual from the co-integrating regression, EC in Tables 2 and 3) and includes all lagged innovations and the constant term.13 These results are shown in the fourth columns of Tables 2 and 3. The third (and final) model includes the error correction term and only those lagged changes that are significant in the second model, while the insignificant constant terms are excluded. These results are shown in the final column of Tables 2 and 3. Once a satisfactory representation of the data generation process has been found in this way, the error correction term can be satisfactorily estimated for both Australia and New Zealand in all cases. If the real exchange rate and the terms of trade were not co-integrated, the estimated coefficients would tend rapidly toward zero. In fact, the error correction term is of the appropriate sign and is significant at the 1 percent level for both Australia and New Zealand.14

Support for the error correction model over both fixed- and floating-rate periods demonstrates the fundamental importance of the long-run equilibrium relationship between the terms of trade and the real exchange rate in both Australia and New Zealand. Blundell-Wignall and Thomas (1987) suggest that the exchange rate regime may also matter in the strength of the terms-of-trade effect. This earlier preliminary result was also confirmed with the more general approach adopted here. In the case of Australia, the size of the estimated error correction coefficient virtually doubles in the post-float period.

In summary, there appears to be firm evidence in favor of accepting the null hypothesis of no co-integration, when the alternative hypothesis is PPP (real exchange rate reversion to a constant equilibrium level). However, the null hypothesis of no co-integration is rejected once the terms of trade is included. Indeed, it proved possible to estimate a robust error correction model of the real exchange rate in relation to its long-run equilibrium level dominated by the terms of trade. The evidence is not consistent with the characterization that monetary shocks (nominal shocks that leave the equilibrium real exchange rates unchanged) are relatively more important than commodity price shocks (real shocks that shift the equilibrium real exchange rate) in generating real exchange rate dynamics in both Australia and New Zealand. In these circumstances, it is difficult to envisage how a pegged exchange rate could ever have been optimal for either Australia or New Zealand. Since there is evidence that adjustment toward the (moving) terms-of-trade-determined equilibrium may be faster under floating rates, this view carries even more force in the late 1980s and beyond.

Chart 2 compares the real exchange rate series for both countries with the terms of trade from the beginning of the 1970s. In the early 1970s, during the fixed-exchange-rate period, both the terms of trade and the real exchange rate rose strongly in Australia and New Zealand. While the exchange rate was revalued in Australia, this action was delayed and relatively small, so that much of the adjustment occurred via inflation. In the case of New Zealand there was little nominal exchange rate response. This period was associated with strong activity and inflation pressure that exceeded those in the rest of the OECD area (see below).

Chart 2.Real Effective Exchange Rates and Terms of Trade

Note: Variables as defined in Tables 2 and 3 and in Chart 1.

Over the floating-rate period the real exchange rate has been permitted to move more in line with the terms of trade (Chart 2) and, from month to month, with commodity price developments (Chart 3). This is particularly clear in the case of Australia, where commodity prices fell sharply in 1985 and recovered strongly after 1986. The real exchange rate moved in a parallel fashion. In the case of New Zealand the terms of trade and commodity prices declined from 1983 to early 1986, recovered in 1987, and weakened again in late 1988 and early 1989. The real exchange rate again evinces a broadly similar profile. If the authorities have an inflation objective and maintain a broadly floating exchange rate, these fluctuations should be an integral part of the adjustment to goods market equilibrium. Within this environment a commodity-exporting country reduces the risk of compromising the inflation objective, in contrast to a policy of pegging or heavily managing the exchange rate. Inflation risks that arise through domestic monetary shocks can, if signal extraction permits, be reduced by giving exchange rate objectives greater weight when monetary shocks are identified.

Chart 3.Weekly Bilateral Exchange Rates

(percent change)
(percent change)

Note: Variables as defined in Tables 4 and 5.

Much greater inflation risks arise when resource allocation (real) policy goals become confused with inflation (nominal) policy objectives. This problem is discussed in the following section.

Resource Allocation versus Inflation?

It was noted above that the choice of exchange rate regime could be considered in terms of optimum currency area considerations; these depend on the structural characteristics of countries, as opposed to the nature of stochastic shocks, which has been the main focus of this paper. Fundamentally different policy objectives underlie these alternative approaches: resource allocation goals versus such macro objectives as inflation. The potential for conflict between these objectives is particularly important for commodity-exporting countries. This is because swings in commodity prices drive the real exchange rate to ensure that goods market equilibrium prevails—but this equilibrating process works through the reallocation of resources between the commodity-exporting and noncommodity sectors.

Consider the case of strong demand in the OECD area together with rising world commodity prices. The exchange rate in the small commodity-exporting country is driven upward to ensure that goods-market equilibrium prevails. The income performance of the commodity sector is buffered, since even though the exchange rate is higher, so is the foreign price of exports. The noncommodity sector, however, is greatly disadvantaged by the rise in the exchange rate. The import-competing sector loses market share, as does the noncommodity export sector. If exchange rate appreciation is severe enough and prolonged, firms may go out of business and, given set-up costs, may not be able to re-establish themselves quickly when the commodity cycle reverses itself. With respect to investment in new industries, it is difficult to diversify away from the commodity sector because of real competitiveness considerations. At precisely those periods when demand in the OECD area is strongest, the noncommodity sectors are at their greatest relative price disadvantage compared with the industries of other countries.

In the reverse situation of weak OECD demand and falling commodity prices, the income performance of the commodity sector is again buffered. However, the improved competitiveness of the noncommodity sector occurs at a time when OECD demand is likely to be at its weakest, and less encouraging of new investment. Furthermore, if the real exchange rate is significantly more variable than in other countries because of the commodity price cycle, uncertainty effects may further inhibit production and trading possibilities.15 New firms may be unwilling to set up in the belief that improved competitiveness is transitory.

If resource allocation objectives were to be given priority, it is not difficult to see how inflation objectives might be compromised. Excessively easy monetary policy to avoid real exchange rate appreciation in a commodity price boom will reinforce the inflation effects of the latter. Conversely, when the terms of trade decline, policies to resist depreciation may contribute to a slump in the domestic economy. From the viewpoint of a monetary authority with a mandate to contain inflation in a commodity-exporting country, it is difficult to see how exchange rate stabilization could be given priority in such circumstances. This is not to argue, however, that resource allocation objectives may be unimportant. Rather, it is to suggest that other (real) policies should be used to promote real competitiveness—competition policy, reduced border protection, work practices, and perhaps other labor market reforms. The importance of a flexible and dynamic noncommodity traded-goods sector capable of developing competitiveness less dependent on relative prices would seem to be the main challenge in these areas.

III. Volatility and Short-Run Exchange Rate Management

The main conclusion of Section II is that a floating exchange rate regime is essential in a commodity-exporting country subject to terms-of-trade shocks if an inflation target is to be achieved. This section focuses on the case for exchange rate management in the face of short-run volatility of the exchange rate under floating—which may result from market inefficiencies or “thinness”—in Australia and New Zealand.

The weekly percent changes in bilateral exchange rates against the U.S. dollar for Japan, the Federal Republic of Germany, the United Kingdom, Canada, Australia, and New Zealand, are shown in Chart 3 for the periods over which the currencies of the latter two countries were floating. Weekly percent changes in effective exchange rates—including for the U.S. dollar—are shown in Chart 4. The standard deviation for each series is calculated, and the straight horizontal lines reflect one standard deviation on either side of zero.

Chart 4.Weekly Effective Exchange Rates

(percent change)
(percent change)

Note: Variables as defined in Tables 4 and 5.

The following points stand out from an inspection of the charts:

  • The volatility of the bilateral exchange rates for Japan, Germany, the United Kingdom, and Australia vis-à-vis the U.S. dollar are remarkably similar. On the other hand, the volatility of the bilateral rate for the Canadian dollar is notably smaller than for other countries, while that for the New Zealand dollar is substantially larger.

  • The volatility of effective exchange rates differs much more widely among countries. That for Germany and Canada is remarkably low—presumably a consequence of German ties within the European Monetary System and the heavy weight of the United States in Canadian trade. Japan, the United Kingdom, and the United States fall within an intermediate range of effective exchange rate volatility—presumably the result of not having natural optimum currency area characteristics. Australian and New Zealand effective exchange rate volatility, on the other hand, is substantially greater than elsewhere, with the New Zealand dollar even more volatile than the Australian dollar.

From the analysis in Section II it is not surprising that overall variability (captured by the standard deviation lines in the charts) should be greater for Australia and New Zealand. This is because of the importance of the commodity price cycle in these countries. Volatility can be thought of as the sum of normal volatility about the equilibrium real exchange rate and volatility of the equilibrium rate itself. The latter is likely to be less important in the other countries considered. It is surprising, however, that New Zealand’s volatility should exceed that of Australia by substantial margins.

A number of studies using Australian data attempt to test whether observed short-run volatility is “excessive.” Tease (1988) examines speculative efficiency in the foreign exchange market by testing whether the forward rate is an unbiased predictor of the future spot rate, and whether forecast errors are correlated with other freely available information. This is a joint test of informational efficiency and zero (or constant) risk premia. He tests the 15-day, 30-day, and 90-day forward rates sampled at weekly intervals from December 1983 to February 1986. Recognizing the unusually large decline in the exchange rate in 1985, the sample period is split in early February of that year. The results suggest a break in forward market behavior at that time. The joint hypothesis of speculative efficiency and zero or constant risk is rejected for the post-February 1985 period in all of the three markets tested. These results provide some support for the view that authorities can, in principle, intervene successfully to prevent or burst price bubbles or help the market move quickly to a new equilibrium. Such intervention can help to smooth fluctuations in the exchange rate. Lowe and Trevor (1986) also examine whether there is evidence of informational inefficiencies by analyzing expectations of exchange rate movements held by financial market dealers. They find that simple, extrapolative rules of thumb dominate.

Excessive volatility in the short run may also be endemic to small countries that float in a world where capital markets have become increasingly integrated. “News” about shifts in economic fundamentals—particularly the commodity price cycle or the overall stance of monetary policy—may be quickly translated into portfolio readjustment. The sheer size of international capital market portfolios is such that relatively small readjustments may have large implications for short-run demands on the markets for the Australian dollar and the New Zealand dollar. Large orders to buy or sell the domestic currency at such times may be met with insufficient demand from currently active traders. Since the market for the New Zealand dollar is “thinner” than that for the Australian dollar, these considerations may help explain the greater observed volatility of the former. To the extent that thinness is a factor, official intervention may be necessary to ensure that adjustment toward equilibrium values of the real exchange rate is orderly.

Market intervention to combat volatility related to inefficiencies or thinness need not involve changes in the overall stance of monetary policy. It is generally accepted that sterilized intervention has little impact on fundamental exchange rate movements.16 This was the conclusion of the working group set up at the Versailles Summit (Report, 1983). But such intervention may help give the market more breadth in periods of large portfolio transactions, or more direction when market inefficiencies dominate (provided that intervention objectives are clear to market participants).

In principle there are no major practical problems with this type of exchange rate management. The market can be “tested” with sterilized intervention. If it is ineffective it is likely that more fundamental elements are at work. Practical difficulties are more likely to arise in deciding on the extent to which monetary policy itself should be adjusted in response to fundamental pressures on the exchange rate. Some of these difficulties are illustrated with historical examples in the next section.

IV. Experiences with Terms-of-Trade Changes

Some of the practical problems of exchange rate policy in commodity-exporting countries are best explored by examining specific episodes characterized by large changes in the terms of trade. For Australia we briefly sketch some of the key features of the Korean War boom and slump (a fixed exchange rate regime), the 1972–75 period (an adjustable exchange rate peg), and the 1985–89 period (a floating exchange rate regime). We attempt to give some impression of the magnitude of the income effects of terms-of-trade changes that have been associated with inflation pressures in Australia. Some of the resource allocation implications are also touched on. We also look at New Zealand’s response to terms-of-trade changes over the last few years.

There are a number of notable features of the Australian experience (see Table 4). First, terms-of-trade shocks are often very large. During the Korean War period the terms of trade increased 46 percent in 1950–51, only to fall by 50 percent during 1951–52. Changes during other periods were less sharp, but two-year upswings were of the order of 15 to 30 percent. The direct contribution of these changes to the real purchasing power of the income generated by domestic production can be measured by revaluing exports by the import price index to calculate a new constant price export series, recalculating constant price GDP data by substituting the new for the old export series and comparing the new “adjusted” GDP (column 3) with actual GDP (column 2). During 1950–51, for example, the constant-price GDP increased by 6¼ percent, but after terms-of-trade adjustment the increase was 15 percent; this represents approximately three years of strong growth in one year. Changes during other episodes are less spectacular, but terms-of-trade upswings usually add about 50 percent to GDP growth rates.

Table 4.Terms-of-Trade Shocks Australia: Selected Economic Indicators(1949–50 to 1952–53: 1971–72 to 1974–75; 1984–85 to 1988–89)
YearTerms-of-

trade

index
GDP1Adjusted

GDP1
Col 3-Col 2Employment

share of

manufact.1
M31Average

weekly

earnings1
Consumer

prices1
Real

wages1
Nominal

exchange

rate1
Percent

inflation

gap

Australia/

overseas2
Real

exchange

rate
123456789101112
I
1949–501007.80.414.39.67.91.70.09.6109.63
1950–511466.215.18.9−1.118.919.613.27.40.010.0120.5
1951–52922.5−6.5−9.0−3.2−1.022.422.8−0.40.014.0137.3
1952–53100−0.80.31.12.29.19.29.00.20.04.0142.7
II
1971–721004.94.6−0.3−1.910.510.16.83.3−0.40.7100.3
1972–731203.96.82.9−0.425.79.05.93.15.71.3107.3
1973–741304.96.01.1−5.914.516.213.13.19.91.7119.7
1974–751222.00.5−1.5−3.715.425.416.78.7−6.22.7115.5
III
1984–851005.55.0−0.5−2.117.57.65.81.8−7.21.794.5
1985–86914.32.8−1.5−4.013.06.28.4−2.2−17.65.182.7
1986–87872.82.0−0.8−3.012.87.19.3−2.2−13.66.877.1
1987–88954.45.81.40.613.36.27.31.11.94.381.9
1988–891003.36.12.8−0.626.77.27.40.212.44.095.3

Percent change.

Mean of inflation rates of the United Kingdom, the United States, Germany, and Canada (I), the United Kingdom, the United States, Germany-Canada, and Japan (II), eight major trading partners, trade weighted (III).

Nominal exchange rate adjusted by differential inflation.

Sources: Australian Economic Statistics, 1949–50 to 1986–87, W.E. Norton and C.P. Aylmer, Occasional Paper No. 8A, Reserve Bank of Australia, and Economic Round-Up, October 1989, The Treasury, AGPS, Canberra.

Percent change.

Mean of inflation rates of the United Kingdom, the United States, Germany, and Canada (I), the United Kingdom, the United States, Germany-Canada, and Japan (II), eight major trading partners, trade weighted (III).

Nominal exchange rate adjusted by differential inflation.

Sources: Australian Economic Statistics, 1949–50 to 1986–87, W.E. Norton and C.P. Aylmer, Occasional Paper No. 8A, Reserve Bank of Australia, and Economic Round-Up, October 1989, The Treasury, AGPS, Canberra.

Second, as suggested in Section II, the manufacturing employment (import-competing) share falls in response to terms-of-trade improvements (column 5). This is especially notable in the first two periods, when the employment share of manufacturing fell 3¼ percent in 1951–52 and 6 percent in 1973–74. In both episodes quotas were introduced to protect manufacturing, underlining the importance of resource allocation issues. After the Korean War boom, quotas were widespread and they remained in place until the end of the decade. After 1972–75 they were more limited and introduced for footwear, clothing, textiles, and motor vehicles. For motor vehicles, quotas lasted 12 years; for other industries they are still in place. During the third period, by which time manufacturing employment had fallen to 16 percent of total employment, the response was quite small.

Third, what can be said about real exchange rate adjustments under different exchange rate regimes? In each period money growth rates accommodated income gains (column 6) and, regardless of the exchange rate regime, Australia has been unable to escape inflation, which typically peaks one year after the terms of trade (column 8). But as the nominal exchange rate has become more flexible in each episode, the inflation response has been reduced. In the first period, with a fixed nominal exchange rate, the real exchange rate change was achieved through variations in the relative price level between Australia and overseas. From 1950 to 1953 the additional price level increase in Australia—over and above its major trading partners—was about 37½ percentage points, a substantial real exchange rate appreciation. During the second period, world inflation again accelerated, but price increases were greater in Australia by an additional 6½ percentage points, even though more of the adjustment was borne by nominal exchange rate appreciation. In the final more recent period, real exchange rate changes were achieved much more by nominal rate changes. Domestic inflation did not accelerate, at least until monetary policy was eased, and the inflation gap between Australia and overseas trading partners narrowed relative to the years immediately preceding the terms-of-trade upswing.

Fourth, although most of this reduced inflation responsiveness to terms-of-trade changes may be attributed to changes in the exchange rate regime, some of the improved performance is related to labor market changes. Terms-of-trade improvements initially add to Australian real income by increasing export receipts, which usually lead to large changes in income distribution. For example, during the first two periods, workers sought to increase nominal wages substantially to maintain fixed income shares and to enable the rest of the community to share quickly in real income gains. The result was substantial inflation of nominal wages (column 7), real wages (column 9), and an overflow into imports. When the terms of trade fell back to earlier levels, workers were unwilling to reduce nominal wages, and under the Australian centralized wage-setting system, full wage indexation was introduced for previous price changes, thus preventing real wage falls. Nominal and real wage asymmetries impart an inflation bias. Upward flexibility of nominal wages when the terms-of-trade and income effects are strong reinforces inflation pressures; downward nominal and real wage rigidity means that these are not reversed to the same extent in periods of weakness and exchange rate depreciation. Asymmetrical real wage adjustment also led to the widespread import quotas after the first period, and to high unemployment after the second.

During the third period, when the unemployment rate was four to five times the average of the 1950s and 1960s, the wage response was very different. The terms of trade fell sharply in 1985–86, and the flexible exchange rate depreciated by about 30 percent in two stages during these years. The centralized wage fixing authority, in cooperation with the trade union movement, did manage to reduce real wages by adopting less-than-full wage indexation. Social wage considerations and income tax cuts were also important. Employment growth thus remained strong in the face of a slowdown in GDP. The problem for monetary policy was to ensure that the inflation pressures from depreciation were not excessive. Nominal wages were not only rigid in the downward direction, but were likely to be subject to upward pressure through indexation; the extent to which the wage negotiating process could continue to quarantine exchange rate effects from wages was limited. Monetary policy at first leaned heavily into the wind, with the interest differential rising sharply in Australia’s favor (Chart 5). However, monetary policy was then eased prematurely in early 1986—that is, while the terms of trade were still falling—which contributed a further sharp depreciation and to increased inflation pressures. Money market conditions had to be tightened again in the second half of the year.

Chart 5.Interest-Rate Differentials and Exchange Rates

Note: Exchange rates as defined above. Interest rates for Australia and New Zealand refer to 90-day commercial bills and 3-month Treasury bills, respectively. The Australian interest differential takes the rate on the U.S. 90-day commercial paper as the foreign rate. The New Zealand interest differential is vis-à-vis Australia.

When the terms of trade improved—adding 1½ percent to real GDP in 1987–88 and 3 percent in 1988–89—the exchange rate rose in the manner predicted in Section II, and the labor movement did not respond by demanding excessive real wage increases. The response to the third large upward terms-of-trade shock, therefore, seems to have been more satisfactory in that there was both greater real wage and nominal exchange rate flexibility. The trade union movement appears to be increasingly willing to allow income distribution to respond more appropriately over the terms-of-trade cycle. This contributed to improved inflation and employment performance in 1987 and 1988.

However, while the floating rate regime has shown its worth in permitting the exchange rate to rise automatically in response to the terms of trade-thereby offsetting inflation pressures, the recent episode also illustrates problems with policies of leaning against the wind in these circumstances. Since consumer price inflation continued to decline in response to the higher exchange rate, and since a marked reduction in the budget deficit suggested fiscal restriction, the authorities were tempted to ease money market conditions in 1987. The exchange rate rose, but by much less than would have been the case if monetary conditions had not been eased. The combination of income effects from rising terms of trade and easier monetary policy contributed to a resurgence of domestic demand and inflation in Australia in 1988. Leaning too heavily against the wind in a commodity price boom entails substantial inflation risks. Monetary policy had to be sharply tightened from April 1988.

Finally, given the importance of terms-of-trade shocks, the question naturally arises as to whether the policy instruments available are appropriate. For example, to the extent that there are nonsymmetric labor market responses—and since employers have been only too willing to pay large wage claims in boom periods—could a system of taxes and subsidies levied on the export sector to offset income distribution effects be useful? Australia has experimented with this type of system to a minor degree (coal export levies and a wool marketing authority to smooth price fluctuations are examples), but there are two practical problems with it. First, taxes and subsidies should only be used to moderate temporary terms-of-trade variations. But it is difficult, in advance, to distinguish between temporary and permanent changes. Second, the government may find it difficult to save receipts when the terms of trade are strong and to fund subsidies when they are weak. Any scheme must thus be self-financing. Furthermore, as short-run export supply elasticities are low, a policy to smooth exchange rate fluctuations would need to be accompanied by a scheme to invest tax receipts overseas when the exchange rate was high in response to terms-of-trade improvements, and to repatriate the revenue in the reverse situation.

We now turn to a discussion of the New Zealand experience during the 1987–88 terms-of-trade cycle, which seems to have been less inflationary than in Australia. In both countries the real exchange rate responded to the terms-of-trade cycle. But relative to past inflation trends, the New Zealand economy seems to have absorbed more of the adjustment by permitting the nominal exchange rate to appreciate, dampening demand pressures, than was the case in Australia. There appear to be three reasons for this:

  • (1) New Zealand’s monetary policy has recently been more firmly directed against inflation. The interest differential in New Zealand’s favor vis-à-vis Australia during 1987 and early 1988 led to a stronger exchange rate and less demand pressure there (Chart 5). There was less attempt to lean against the wind than in Australia. Consequently, although there was considerable suppressed inflation in New Zealand at the beginning of the terms-of-trade cycle, owing to previous wage and price controls, the tight monetary policy has been quite successful at reducing inflation over the last two years.

  • (2) When Australia began to tighten monetary policy and narrow the interest differential between the two countries from April 1988, the New Zealand terms of trade had already “leveled off”—thus reducing income effects and the upward pressure on the real exchange rate.

  • (3) The Australian institutional wage-setting process since 1983 adopted close to full wage indexation for past price changes in most decisions. Although wage policy succeeded in achieving considerable real wage reductions at times, any system with indexation over long periods of time inevitably makes winding back inflation more difficult, unless there are downward price shocks.

V. Concluding Remarks

This paper has sought to identify key problems of exchange rate policy in commodity-exporting countries, both from a theoretical and empirical perspective. The issues have been addressed in the context of deregulated and competitive domestic financial systems and highly integrated world capital markets. In this environment the exchange rate may have become a more important transmission mechanism for monetary policy, relative to the channel of direct interest rate effects on domestic demand. The major focus has been from the perspective of a monetary authority responsible for an inflation objective, although it was explicitly recognized that optimal exchange rate policy within this framework could well conflict with resource allocation goals and that other policies might be required to address these.

Our analysis suggests that the world commodity price cycle may be a major influence on exchange rate dynamics in small countries with export concentration in commodities. Specifically,

  • commodity prices play a central role in determining the equilibrium real exchange rate in a small-country importables/exportables model;

  • when tested using Australian and New Zealand data, this influence was found to be important in practice; and

  • these findings have strong implications for exchange rate policy in these economies.

In general terms, small commodity-exporting countries should exhibit greater exchange rate volatility relative to other industrial countries. But it is not clear that this volatility (over very short horizons) should be a concern for monetary policy. Intervention that is fully sterilized can be used to promote orderly behavior when markets are particularly thin or lack direction. These issues are of secondary importance relative to the need to decide on the weight that should be attached to the exchange rate in the broad formulation of monetary policy.

The choice of a broadly floating exchange rate is essential for low-inflation monetary policy in a commodity-exporting country. Nevertheless, some form of exchange rate management is always likely to be necessary for effective monetary policy. This management should be strongest at times when monetary shocks can be identified. But even when movements in the equilibrium real exchange rate are dictated by terms-of-trade shifts, it is not clear that the optimal policy is to permit the exchange rate to find its own level with no response from the monetary authorities. Practical experience suggests that excessive concern with declining competitiveness may pose problems for achieving an inflation target when the terms of trade and the exchange rate are strengthening—leaning against the wind is risky, and it may even be necessary to lean with the wind. Conversely, when the terms of trade and the exchange rate decline, labor market rigidities may improve the case for leaning against the wind.

Comment

Jeff Carmichael

The paper by Blundell-Wignall and Gregory highlights two themes that I found particularly relevant. The first is the role of the exchange rate as a transmission mechanism for monetary policy. The second is the importance, to a commodity exporting country, of allowing adjustments in the real exchange rate. While these ideas will not necessarily be new to policymakers, they are nonetheless easily overlooked in formulating policy.

Despite finding myself in general agreement with the broad thrust of the paper, I do have a number of minor quibbles about points of interpretation. In the paragraphs that follow, I will start by summarizing briefly what I see as the authors’ main arguments. I will then offer some comments on their model and their analysis. My focus is, of course, on the Australian experience.

The Authors’ Argument

Blundell-Wignall and Gregory ask essentially three questions:

  • (1) how should a commodity exporter manage its exchange rate?

  • (2) how have the experiences of Australia (and New Zealand) matched up with the authors’ views on efficient exchange rate management? and

  • (3) is there a role for short-run (sterilized) intervention independent of the longer-run objectives of monetary policy?

To address the first question, the authors construct a model with an explicit distinction between traded and nontraded goods. They conclude that a monetary policy objective of minimizing inflationary disturbances requires considerable exchange rate flexibility. The essence of their result is that terms-of-trade shifts change the equilibrium real exchange rate, while monetary shocks do not. Since commodity exporters typically face a preponderance of terms-of-trade shocks relative to monetary shocks, an inflexible exchange rate forces the real exchange rate to adjust through domestic inflation—a process that is inconsistent with a stable inflation objective.

With respect to Australia’s exchange rate experience over the past two decades, Blundell-Wignall and Gregory note that:

  • movements in real exchange rates have been dominated by movements in the nominal exchange rate (the contribution of changes in relative inflation rates has been minor);

  • the real exchange rate has been relatively volatile, and in ways not necessarily consistent with simple movements in purchasing power parity; and

  • terms-of-trade movements have been substantial.

The authors find, through a series of rigorous statistical tests, that there is a strong relationship between Australia’s real exchange rate and its terms of trade. They also find that this relationship is stronger in the post-1983 period of floating than in the pre-float period. They interpret this as evidence that commodity price shocks have dominated monetary shocks in generating real exchange rate dynamics and, therefore, that a pegged exchange rate—such as was maintained prior to 1983—is inappropriate if controlling inflation is the main objective of monetary policy.

The third question is very much a minor focus of the paper. Using weekly data, the authors find that Australia (and New Zealand) have experienced above-average effective exchange rate volatility. They argue that there may be a role for short-term intervention within a broad policy of allowing flexible exchange rate adjustments. They base their case on market “thinness” and inefficiencies.

The Model

I have no quarrel with the choice of model for the paper. Simplifying assumptions are made for tractability and, by and large, these do not detract from the model’s relevance. Subject to my overall agreement with the model, the following are some minor thoughts on three aspects of the model.

Current Account Imbalance

The model rests on the hypothesis that inflation is the sole objective of monetary policy. While economic theory suggests that this is a logical approach, central banks do, from time to time, include current account considerations in their determination of monetary policy. This occurs despite the well-known conflict that arises in trying to influence the current account through monetary policy.

The model used by Blundell-Wignall and Gregory assumes perfect capital mobility, thereby effectively ruling out current account or debt problems. This is not unreasonable in analyzing short-run responses to random shocks. However, to the extent that the model is used to analyze history, it avoids an issue that has played a major role in shaping policy (monetary and fiscal) in Australia.

For example, in their discussion in Section IV of the “premature” easing of monetary policy in early 1986, the authors fail to mention that policy adjustments at the time were motivated much more by current account considerations than by inflation. Of particular importance was the timing of monetary and fiscal actions in that period and the perception (subsequently found to be incorrect) that the adjustments in fiscal policy and international competitiveness occurring at that time would be sufficient to slow the economy and reverse the deterioration in Australia’s external accounts.

The second area in which external debt could be relevant is in the analysis of the real exchange rate. The authors’ emphasis on the terms of trade as the primary determinant of the real exchange rate follows from their analysis of perturbations of their model from a long-run equilibrium. Australia, however, is at a particular point of disequilibrium along a dynamic path toward that steady state, and the history of that position is not irrelevant. We could ask, for example, whether the equilibrium real exchange rate would be the same (for given terms of trade) for Australia if it had, in the first case, net external debts of $A120 billion and, in the second case, net external assets of $A120 billion. Changes in Australia’s balance sheet, over the data period considered, are likely to have contributed to changes in the real exchange rate.

The Choice of Price Indexes

In their model, the authors define the objective function purely in terms of the price of nontradables. Extending this to a broader measure, such as the consumer price index, would add a fourth effect to the three they note for terms-of-trade shocks—namely, a direct price effect arising from exchange rate movements. Like the demand substitution effect, this direct price effect would tend to dampen the inflationary impact of a rise in the terms of trade. The effect will be greater, the more open the economy.

In terms of their analysis, allowing for direct price effects would reduce the case for leaning with the wind (that is, for policy to amplify exchange rate movements beyond those that would occur in a free market) and increase the case for leaning against the wind. Indeed, in terms of the criteria outlined in Section II, leaning against the wind would seem to be the most likely optimum for a country such as Australia.

Deregulation and Financial Markets

The authors argue that their analysis is in the context of deregulated markets, with competitive financial systems and highly integrated world capital markets. On reflection, I wonder whether enough is made of this aspect.

Certainly, their capital markets are integrated. A second aspect, that monetary policy has more impact on economic activity through the exchange rate than through interest rates, is imposed by assumption. Sorting out these types of issues is one of the toughest problems currently facing central banks, and Australia is no exception. The issue requires more than dismissal by assumption; to be fair, however, this paper is probably not the place to analyze the question fully.

In my view, deregulation requires a complete rethinking of the way we model the financial sector. Simple monetary mechanisms such as those described in equation (9) and the money supply reaction function in equation (12) are relics of a regulated world. In a deregulated financial system, the central bank operates almost exclusively through short-term interest rates with monetary and credit aggregates being determined endogenously. The link from monetary aggregates to economic activity and inflation has become noisier, more contemporaneous, and much less causal than under financial regulation.

Short-Run Exchange Rate Management

This section of the paper deals essentially with sterilized intervention. Oddly enough, despite the widespread use of the sterilization terminology, it is not one that many central bankers use in practice. This given, it may be useful to spell out some practical aspects of intervention as they relate to sterilization. In the first place, all foreign exchange transactions by the central bank are absorbed into the opening cash position of the banking system on settlement day. Since they are not distinguished from other sources and uses of cash, the central bank money market desk faces only one decision: how much to intervene in the cash market. In this sense they are simultaneously sterilizing (or choosing not to sterilize) the net of all cash transactions (foreign exchange, security settlements, tax payments, and so on).

The picture is further complicated by transactions undertaken by the central bank for its clients (mostly government). For example, in 1984 the Reserve Bank chose not to transact directly in the foreign exchange market. The Bank hoped the market would develop more quickly and efficiently if it were unhindered by official intervention. At the same time, it funded the Government’s overseas commitments of around $1.5 billion by running down foreign reserves. This was effectively sterilized intervention, even though the Bank did not enter the foreign exchange market. In 1988/89 the Bank purchased $5.5 billion directly in the market—all for onsale to the Government to meet its needs. Again, the transactions were effectively sterilized in the sense that they did not lead to changes in the Bank’s stance in the domestic money market.

The question raised by Blundell-Wignall and Gregory is whether the central bank can influence the exchange rate by the way it carries out transactions of this type.

Like the authors, I believe that short-term direction and depth can be added to the foreign exchange market without having to alter monetary policy. Also, like the authors, I believe the impact to be short-lived. In this respect, the authors could have drawn a contrast between the clear use of monetary policy to stop the precipitous fall of the Australian dollar in July 1986 and the prolonged attempt to slow the rise of the rate in 1987 through sterilized intervention (that is, unaccompanied by a change in monetary policy).

Grant Spencer

The paper by Blundell-Wignall and Gregory addresses exchange rate policy in a small, commodity exporting country from the perspective of a monetary authority with an inflation objective. It concludes that, in order to maintain a reasonable degree of price stability under external terms-of-trade shocks, a “broadly floating” exchange rate regime is essential. The authors also conclude, however, that some degree of exchange rate targeting is likely to be necessary, particularly where:

  • shocks to money demand can be identified; and

  • downward movements in the terms of trade create strong inflationary pressures.

As a central banker involved with monetary and exchange rate policies in a small, commodity exporting country, I found this paper informative and highly relevant to the current policy debate in New Zealand. Furthermore, I agree with the main conclusions of the paper.

Before offering a number of comments on the body of the paper, I would like to give a brief sketch of the experience with exchange rate policy in New Zealand in recent years.

Exchange Rate Policy in New Zealand

In March 1985, the exchange rate regime in New Zealand shifted from an adjustable peg to a free-float. The reasons were essentially twofold:

  • (1) The newly elected Government in mid-1984 made a commitment to reducing inflation. It became clear in late 1984 that the tight monetary policy necessary to put pressure on inflation could not be achieved under a pegged rate. Attempts to push up real interest rates were quickly sterilized through the foreign exchange window as capital inflows responded to the initial increase in domestic interest rates.

  • (2) In the post-1984 deregulatory environment, there was a desire to see greater flexibility in the real exchange rate, so as to encourage a greater movement of real resources in response to external terms-of-trade shocks. This objective should be seen in the context of the preceding decade, during which economic policy in New Zealand was generally aimed at insulating the economy from the major (permanent) commodity price shocks of the 1970s.

The New Zealand experience with a floating exchange rate has certainly achieved the first objective of allowing an independent discretionary monetary policy. The Reserve Bank was able to push real interest rates above world levels and thus put downward pressure on inflation. In addition, the direct influence of the exchange rate on prices introduced a potent new transmission channel for monetary policy.

With respect to the second objective, however, there is little evidence that the real exchange rate has been responsive to the underlying competitiveness of the traded goods sector. While current account developments may have had an important influence on the exchange rate in Australia, it is apparent that monetary policy and subsequent capital account developments have been dominant in the case of New Zealand (see Chart 1).1 The strength of the New Zealand dollar during 1985–88 appears to have resulted from both the normal “overshooting” effect of a monetary tightening in a Dornbusch-type model, and also from the interaction of a tight monetary policy with the major financial deregulation occurring in 1984–85.

Chart 1.New Zealand: Nominal Trade-Weighted Exchange Rate, Commodity Prices, and Yield Gap Between 5-Year Bonds and 90-Day Bills

The main lesson I would take from the New Zealand dollar exchange rate experience during the float period is that monetary policy probably should have responded more to the exchange rate as an indicator of monetary conditions. Such an approach would not have avoided the substantial pressure that the traded goods sector came under from the high average real exchange rate during 1985–88. However, in hindsight, it probably would have yielded a more consistent degree of disinflationary pressure while creating a more stable overall trading and investment environment.

Indeed, since August 1988, the trade-weighted New Zealand dollar exchange rate index has become the Reserve Bank’s principal monetary indicator. As a consequence of this shift in emphasis between the indicators, the New Zealand dollar exchange rate has become more stable over the past year, moving within a range of about 3 percent on either side of its average.2 However, while acknowledging a greater role for the exchange rate in setting monetary policy, the Reserve Bank has not gone as far as announcing a target band for the exchange rate. This is due to recognition of the main point in the paper by Blundell-Wignall and Gregory: an effective monetary policy in a small, commodity producing country must be prepared to allow exchange rate flexibility in the face of terms-of-trade shocks.

The Blundell-Wignall and Gregory Paper

I will comment on the paper’s model-based optimal policy analysis, empirical co-integration work, and on short-run exchange rate volatility.

Optimal Policy Framework

In this central part of the paper, the authors use a Poole-type analysis to assess alternative monetary policy rules in a small-country model subject to money demand and terms-of-trade shocks. They find that, in order to stabilize inflation in nontraded goods prices, monetary policy should aim to stabilize the exchange rate when money demand shocks are dominant. In the alternative situation, where terms-of-trade shocks dominate, the authors find that monetary policy should be less responsive—but not totally unresponsive—to exchange rate movements.

This basic result is certainly intuitively plausible and is supported by similar analyses elsewhere.3 Of particular interest is the finding that—under terms-of-trade shocks—monetary policy may be required to lean either with or against the wind, depending on the parameters of the model.

I suspect that this aspect of the optimal policy prescription could be clarified in the paper by more specific reference to the source (export or import) and direction of terms-of-trade shocks. Furthermore, I would expect the analysis to prescribe a policy of leaning against the wind in the majority of cases, particularly if the model is modified to include:

  • an asymmetric adjustment process for nontraded goods prices in equation (8) to allow for the relative inflexibility of real wages in the downward direction; and

  • the use of expenditure price inflation p˙d rather than nontraded price inflation p˙n in the loss function (equation (16)).

In the case of a terms-of-trade reduction, an assumed downward inflexibility in real wages will tend to accentuate the nominal depreciation required to regain equilibrium. If the direct exchange rate effect on import prices is then felt in a loss function based on expenditure rather that nontraded price inflation, it seems clear that monetary policy will need to lean into the wind, that is, tighten. This will almost certainly be the case when the terms-of-trade decline results from higher import prices.

In the case of a terms-of-trade increase, the direct import price effects of an appreciating exchange rate will help to offset the impact of higher nontraded goods prices in the loss function. Where terms of trade have risen because of an initial drop in import prices, the inflation risk will be further reduced, making it unlikely that a policy tightening will be required to drive the exchange rate up further. On the other hand, if export price rises are the cause of the shock (the example used in the paper), the direct pressures on domestic expenditure prices will be greater, particularly if export products are consumed domestically—as occurs with food commodities in New Zealand, for example. This may be the one case where it may be necessary for monetary policy to lean with the wind following a terms-of-trade shock.

Empirical Analysis

In order to assess Australian and New Zealand exchange rate policies in the context of the optimal policy analysis, Blundell-Wignall and Gregory test two hypotheses for each country. In the first test, the real exchange rate is found to be I(1) in both countries, suggesting that there is no tendency to purchasing power parity (PPP). In the second test, the authors attempt to assess the influence of terms-of-trade fluctuations on the real exchange rate by testing for co-integration between these two variables in each country.

I found this section to be relatively weak. It is not clear how the co-integration results relate back to the theoretical policy analysis, and the empirical analysis does not appear to be sufficiently thorough.

Based on the analysis in Section II, the optimal monetary policy approach hinges on the relative size of money demand versus terms-of-trade shocks. One would therefore expect the empirical analysis to include estimates of these variances for the two countries. The co-integration analysis can certainly be used to support the theoretical model, which implies a long-run relationship between the real exchange rate and the terms of trade (equation (13)). However, given that the rejection of PPP says nothing about the significance of monetary shocks in a sticky-price model, the co-integration results tell us nothing about the relative importance of money demand and terms-of-trade shocks in Australia and New Zealand.

While monetary variables have no impact on the equilibrium real exchange rate in the theoretical model, they certainly have an important impact in the “short term,” and in practice—as observed recently in New Zealand—monetary shocks can affect the real exchange rate over a number of years. For this reason, the regressions in the paper may well be improved by including monetary variables, at least in the short-term, error-correction model, if not in the co-integrating relationship itself. In work done recently at the Reserve Bank of New Zealand, Shelley Cooper found it necessary to include monetary variables4 in order to establish a co-integrating relationship between the New Zealand dollar exchange rate and New Zealand’s terms of trade. The authors may find that this also applies in the case of Australia.

On a more technical note, I was surprised to see co-integration and the error correction model discussed as “equivalent concepts.” The finding of co-integration allows consistent estimation in the error-correction form, but the successful estimation of an error-correction model does not imply co-integration. With respect to the Durbin Watson (GRDW) statistic used to test for co-integration, this statistic is known to have low power. The more powerful augmented Dickey-Fuller (ADF) test, used in the paper to test the first hypothesis for a stationary real exchange rate, could also have been usefully employed in the co-integration tests. Finally, the authors need to be careful about employing co-integration tests over the relatively small, post-float sample periods; while the tests are only valid asymptotically, some of their results in Table 2 appear to be based on very few degrees of freedom.

Volatility and Exchange Rate Intervention

It is fairly clear from Charts 3 and 4 in the paper that New Zealand wins first prize for short-term exchange rate volatility during 1985–88. As the authors point out, the relative thinness of the New Zealand foreign exchange market is probably the main reason for this relatively high volatility. Other relevant factors have been a quantity-based approach to monetary policy, a changing political environment, and the initially unsettled state of financial markets following the float and financial deregulation. However, I would not agree that the terms of trade have made a significant contribution to volatility; terms-of-trade movements are more likely to generate long-term swings rather than large weekly movements in the exchange rate.

The relevance of the monetary policy approach to the observed level of exchange rate volatility in New Zealand is supported by evidence of a reduction in volatility since August 1988 (Chart 2 of this note); this reduction has coincided with the shift of emphasis toward the exchange rate as a guide to monetary policy settings.

Chart 2.Monthly Exchange Rate Volatility

*Average of Swiss franc, pound sterling, deutsche mark, yen, and Australian and Canadian dollars.

Note: Mean absolute percent changes in U.S. dollar bilateral exchange rates (from maximum to minimum for the month).

Blundell-Wignall and Gregory suggest that in view of the relatively high volatility of the New Zealand dollar, the New Zealand authorities should perhaps consider sterilized foreign exchange market intervention with the aim of promoting smoother adjustment toward equilibrium.

Such a policy would effectively reduce the demand for private insurance against exchange rate movements (via futures, options, and so on) by introducing an element of compulsory state-provided insurance. In order to make a case that the public-provision alternative is more efficient, it would seem to be a minimum requirement that the authorities have more market information than private market participants. If it is assumed that the authorities have inside or expert knowledge of monetary policy intentions or terms-of-trade developments, there is clearly some basis for managing medium-term exchange rate movements through the monetary policy stance—as per the policy-optimization framework discussed in Section II of the paper.

However, the Reserve Bank of New Zealand does not consider that it has greater knowledge or foresight of short-term exchange rate movements than private market participants. Other factors that have kept the New Zealand authorities out of the foreign exchange market include the desire to avoid complicating the interpretation of exchange rate movements for monetary policy purposes, and the observation that short-term “smoothing operations” in other countries have often developed into expensive “trend management” exercises. Certainly the New Zealand Government is eager not to repeat the costly support operation that occurred just prior to the change of government in 1984. Accordingly, the Reserve Bank has avoided any foreign exchange intervention since the float in March 1985, reserving the right to intervene solely for the purpose of counteracting extreme market disorder.

References

    AkhtarM.A. andR. SpenceHilton“Effect of Exchange Rate Uncertainty on German and U.S. Trade,”Federal Reserve Bank of New York Quarterly ReviewNo. 9 (New York: Spring1984).

    BaileyM.J.et al.“Exchange Rate Variability and Trade Performance: Evidence From the Big Seven Industrial Countries,”Weltwirtschaftliches ArchivesNo. 122 (Tubingen: 1987).

    Bank for International SettlementsRecent Innovations in International Banking (Basel: BIS1986).

    Blundell-WignallAdrian andMarilynThomas“Deviations from Purchasing Power Parity: The Australian Case,”Reserve Bank of Australia Discussion Paper No. 8711 (Sydney: 1987).

    Blundell-WignallAdrian andMarilynThomas andPaulMasson“Exchange Rate Dynamics and Intervention Rules,”Staff PapersVol. 32No. 1 (Washington: International Monetary FundMarch1985).

    BoyerRusselS.“Optimal Foreign Exchange Market Intervention,”Journal of Political EconomyVol. 86No. 6 (Chicago: 1978).

    CushmanDavidO.“Has Exchange Risk Depressed International Trade? The Impact of Third-Country Exchange Risk,”Journal of International Money and FinanceVol. 5 (Guilford, England: September1986).

    CushmanDavidO. (1988a) “U.S. Bilateral Trade Flows and Exchange Risk During the Floating Period,”Journal of International EconomicsVol. 24 (Amsterdam: North-Holland: May1988).

    CushmanDavidO. (1988b) “The Impact of Third-Country Exchange Risk: A Correction,”Journal of International Money and FinanceVol. 7 (Guilford, England: September1988).

    DornbuschRudigerOpen Economy Macroeconomics (New York: Basic Books1980).

    DreyerJacobS.“Determinants of Exchange Rate Regimes for Currencies of Developing Countries: Some Preliminary Results,”World DevelopmentVol. 6 (London: Pergamon1978).

    EdwardsSebastian (1987a) “Tariffs, Terms of Trade and the Real Exchange Rate in an Intertemporal Optimizing Model of the Current Account,”National Bureau of Economic Research Working Paper No. 2175 (Cambridge, Massachusetts: NBERMarch1987).

    EdwardsSebastian (1987b) “Anticipated Protectionist Policies, Real Exchange Rates and the Current Account,”National Bureau of Economic Research Working Paper No. 2214 (Cambridge, Massachusetts: NBERApril1987).

    EngleRobertF. andC.W.J.Granger“Cointegration and Error Correction: Representation, Estimation and Testing,”EconometricaVol. 55No. 2 (Chicago: University of Chicago PressMarch1987).

    FischerStanley“Stability and Exchange Rate Systems in a Monetarist Model of the Balance of Payments,”in The Political Economy of Monetary Reformedited byR.Z.Aliber (London: Macmillan1977).

    FrenkelJacobA. andJoshuaAizenman“Aspects of the Optimal Management of Exchange Rates,”Journal of International EconomicsVol. 13 (AmsterdamNorth-Holland: 1982).

    FullerWayneA.Introduction to Statistical Time Series (New York: John Wiley1976).

    GoturPadma“Effects of Exchange Rate Volatility on Trade: Some Further Evidence,”Staff PapersVol. 32 (Washington: International Monetary FundSeptember1985).

    GregoryRobertG.“Some Implications of the Growth of the Minerals Sector,”The Australian Journal of Agricultural EconomicsVol. 20No. 2 (Sydney: 1976).

    HellerH.Robert“Determinants of Exchange Rate Practices,”Journal of Money Credit and BankingVol. 10 (Columbus, Ohio: Ohio State University PressAugust1978).

    HoldenPaulMerleHolden andEstherSuss“The Determinants of Exchange Rate Flexibility: An Empirical Investigation,”Review of Economics and StatisticsVol. 61 (Cambridge, Massachusetts: August1979).

    International Monetary FundExchange Rate Volatility and World Trade (Washington: IMFJuly1984).

    KenenPeterB. andDaniRodrik“Measuring and Analysing the Effects of Short-Term Volatility in Real Exchange Rates,”Review of Economics and StatisticsVol. 68 (Cambridge, Massachusetts: May1986).

    LowePhillip andRobertTrevor“The Performance of Exchange Rate Forecasts,”Reserve Bank of Australia Discussion Paper No. 8609 (1986).

    MundellRobertA.“Towards a Better International Monetary System,”Journal of Money Credit and BankingVol. 1No. 3 (Columbus, Ohio: Ohio State University PressAugust1969).

    MussaMichael“Nominal Exchange Rate Regimes and the Behaviour of Real Exchange Rates: Evidence and Implication,”Carnegie-Rochester Conference Series on Public PolicyVol. 25 (AmsterdamNorth-Holland: Autumn1986).

    NearyPeter“Determinants of the Equilibrium Real Exchange Rate,”American Economic ReviewVol. 78No. 1 (Nashville, Tennessee: March1988).

    OstryJonathanD.“The Balance of Trade, Terms of Trade, and Real Exchange Rate,”Staff PapersVol. 35No. 4 (Washington: International Monetary Fund1989).

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

    “Report of the Working Group on Exchange Market Intervention,”Board of Governors of the Federal Reserve System (Washington: 1983).

    ShaferJeffreyR. andBonnieLoopesko“Floating Exchange Rates After Ten Years,”Brookings Papers on Economic ActivityNo. 1 (Washington: The Brookings Institution1983).

    TeaseWarren“Speculative Efficiency and the Exchange Rate: Some Evidence Since the Float,”Economic RecordVol. 64No. 184 (Australia: March1988).

The authors are grateful to Paul Masson, Steve Grenville, Ian MacFarlane, David Gruen, Paolo Manasse, and Val Koromzay for helpful comments, and to Janice Callaghan for statistical assistance. The views expressed in this paper are those of the authors and do not necessarily reflect those of their respective employers.

The issue of whether New Zealand should peg its exchange rate to the Australian dollar is not addressed explicitly in this paper.

To this could be added the lack of geographical proximity or an obvious third currency against which to peg.

The influences would take the form of transaction costs and the effects of uncertainty in decision-making processes.

Other seminal contributions include Mundell (1969) and Fischer (1977).

This assumption makes the model more analytically tractable and transparent, and for this reason it is commonplace in the literature. It may not be unrealistic in the case of Australia and New Zealand because these economies have highly deregulated financial systems with relaxed liquidity constraints. The Cross Committee (see Bank for International Settlements, 1986) has argued that an important effect of deregulation on innovation is to weaken the direct transmission of monetary policy on economic activity via interest rates and to strengthen the transmission through the exchange rate. In Australia, in particular, the economy has appeared to be considerably more insensitive to higher interest rate levels in recent years. Nevertheless, some comments will be offered on the implications of relaxing this assumption at a later stage.

Equation (2) is linearized in logarithms by using the first terms in a Taylors series expansion about the steady state. For simplicity, it is assumed that in the steady state the shares of both exports and imports in output are equal to λ. Thus it can be shown that

Similar findings may be found in Dornbusch (1980) and, in an intertemporal model context, in Ostry (1989), Edwards (1987a and 1987b), and Neary (1988). In the case of Australia the idea that competitiveness might be linked to the performance of the commodity-exporting sector is found in Gregory (1976), who explored implications of minerals discoveries. The present focus on relative prices is a more general application.

Expression (13) is substituted into equation (11) and is used in the solution of the model below. The expected change in the real exchange rate therefore has a forward-looking aspect. We leave open the question of whether the value of θ ensures consistent expectations, an issue not directly of interest to the problem at hand.

By choosing domestic inflation as the objective, we assume the authorities take account of exchange rate effects only to the extent that they are reflected in excess supply or demand for nontraded goods. This reduces the probability of “leaning against the wind” in the subsequent discussion. Excessive focus on the consumer price index when the terms of trade and exchange rate are rising can lead to more ingrained inflation problems if the authorities ease monetary policy in the belief that the consumer price index is favorably affected, as recent Australian experience has shown. We wish to focus on these domestic demand issues in the subsequent analysis.

The model may also be solved for disturbances to foreign interest rates, the supply of money, and the supply of output. By and large, foreign interest rate and money supply shocks can be treated as similar to a money demand shock, with qualifications related to the value of the interest elasticity of domestic expenditure discussed further below. Output shocks will be similar to terms-of-trade shocks. The two shocks chosen are thought to be more representative in practice.

These relationships are best seen from the expression for the optimal intervention parameter δ* in the case where:

It is lower and possibly negative the higher (1 蜢 λ), and the lower α and γ. The size of η is ambiguous and likely to be of small importance, since the two substitution effects (in response to px and e) offset each other.

The empirical work in this section constitutes a more general treatment of the econometric issues than the earlier work of Blundell-Wignall and Thomas (1987).

This restriction is tested with the chi-square statistic based on the log-likelihood values. The restriction is accepted at the 5 percent level in most cases and at the 1 percent level for the quarterly post-float model for Australia.

As these tests are valid asymptotically, data covering longer periods than here would help to verify the robustness of the results. Given the relatively small number of observations, the post-float regressions for Australia should be treated with particular caution. They are presented only as a preliminary indication of how the relationship might be affected by a change in the exchange rate regime.

Resource allocation costs of exchange rate variability are surveyed in International Monetary Fund (1984). Subsequent studies include Akhtar and Hilton (1984); Gotur (1985); Kenen and Rodrik (1986); Bailey, et al. (1987); Cushman (1986) and (1988a and 1988b). A number of studies have found some costs of exchange rate variability for international trade.

It is also possible that sterilized intervention may destabilize adjustment toward steady-state inflation as in Blundell-Wignall and Masson (1985).

Chart 1 compares real exchange rate trends since 1985 with the terms of trade and a measure of monetary policy tightness—the yield gap between 90-day bills and five-year bonds.

Chart 2 shows that, in terms of within-month movements, the NZ dollar has become more stable than the major currencies, on average, since October 1988.

See, for example, Grimes A., “The Choice of Intermediate and Operating Targets for Monetary Policy,” Reserve Bank of New Zealand Discussion Paper G89/3 (1989); and Glick R., and M. Hutchison, “Exchange Rates and Monetary Policy,” Federal Reserve Bank of San Francisco Economic Review (Spring 1989).

The monetary variables used—real money balances—were found to be I(1) and were therefore included in the co-integrating regression.

    Other Resources Citing This Publication