The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
THE PERSISTENCEOF INFLATION Thomas M. Humphrey Among the most exasperating and puzzling of recent economic phenomena is the apparent intractability of the inflation rate. Once started, an inflation becomes difficult to subdue. It seems to develop a momentum of its own, independent of other basic economic conditions. It resists or at best responds only sluggishly to traditional restrictive policies. Its persistence in the face of high unemployment and excess capacity has resulted in the addition of the term stagflation to the economist’s lexicon. What accounts for the stickiness of the inflation rate? Why is it so policy-resistant and difficult to control? Why is it so slow to decelerate even when demand is slack? Many economists believe that the answers lie in the mechanism through which inflationary impulses are transmitted through the economy.’ Embedded in this mechanism are certain delays or lags that may slow the spread of inflation over the total price structure and may also prolong its duration. Particular prices that lag behind general inflationary movements have to catch up later to reestablish their relative position in the price structure. This lag/catch-up characteristic of the inflationary transmission mechanism is offered by some as an explanation of why strong upward pressures on prices persist long after demand slackens. The most important lags in the process by which inflation is diffused through the economy are the price-adjustment lag and the expectations-formation Zag. The first refers to the delayed response of the rate of price increase to shifts in aggregate demand. Demand pressure is transmitted to prices via a complicated and circuitous channel that runs from output to inputs to input prices to costs and lag finally to product prices. The price-adjustment accounts for the time it takes for demand to affect prices through the channel of costs. It should be noted that many of these costs themselves adjust slowly, partly because they are influenced by sticky price anticipations. These sticky anticipations are described by the second lag, which refers to the slowness with which expectations of future inflation are revised when individuals realize that actual inflation has turned out to be different than was ex1 See, for example, the studies by Cagan [l, 2, 3], and Laidler [5, 6]. See also the articles by Friedman [4] and Selden [8]. The present article draws heavily from Cagan and Laidler. pected. For example, immediately following periods of rapidly accelerating inflation, expectations about the future behavior of prices continue to reflect the preceding price experience even though the current rate of inflation may be decelerating. Some analysts point to these lags as the reason that inflation is so persistent and hard to subdue, even in the face of high unemployment and excess capacity. The purpose of this article is to examine the priceadjustment and expectations-formation lags and to indicate how they may affect the speed, pattern, and duration of inflation. The article proceeds in the following manner. First, it identifies the location and describes the operation of the lags in the inflationary transmission mechanism. Second, it provides an explanation of the existence of the lags. Third, and most important, it analyzes the policy implications of the lags. Finally, the appendix contains a brief description of how the lags are treated in simple analytical models of the inflationary process. Sketch of the Inflationary Process The first objective is to describe the operation of the lags in the inflationary mechanism. As a necessary preliminary, a brief description of the inflation process is offered, with emphasis on the time sequence or chronological order in which key economic variables (spending, output, costs, prices, expectations, etc.) The hypothetical adjust to inflationary pressures. example presented below may not conform to all of the inflationary episodes experienced in the U. S. although it probably typifies most of them. As a first approximation, the sequence of events in a typical inflation may be described as beginning with an increase in aggregate demand to a level in excess of the economy’s capacity to produce.2 Such 2 In starting with demand, the example does not deny that supply shocks-crop failures, strikes, natural disasters and the like-may also play a role in inflation. Given the random, transient, and often reversible nature of these shocks, however, it is unlikely that they could produce a continuing (sustained) inflation. What they can do is to temporarily intensify an inflation generated This is the problem of price blips by demand forces. illustrated by the transient double-digit episode of 1974 when supply constraints caused inflation to deviate temporarily from its basic trend rate determined by excess demand. When the supply shocks abated, inflation returned to its basic path. Note also that supply shocks can be treated as an excess demand phenomenon, albeit one in which the excess arises from falling supply rather than rising demand. FEDERAL RESERVE BANK OF RICHMOND 3 a step-up in the rate of spending might stem from any of a variety of causes including : (1) an increase in the government’s budgetary deficits ; (2) an episodic expansion in private spending associated, say, with a major innovation ; and (3) an acceleration in the rate of monetary growth either actively engineered or passively permitted by the monetary authorities. Whatever the cause, the increased rate of spending, if it is to continue, must be supported by a sustained higher rate of monetary growth. Without this concomitant monetary growth, rising prices would simply reduce the real value of existing cash balances below the level that people desire to hold. In an effort to rebuild their price-eroded real balances, cashholders would cut expenditures, thereby bringing the inflation to a halt. An increased rate of spending cannot be maintained for long without the monetary expansion necessary to finance it. Assuming such expansion occurs, however, the corresponding increased rate of spending leads to a buildup of excess demands on the economy. Businessmen initially respond to this rising demand not by raising prices but rather by reducing inventories and expanding output. That is, their initial adjustments are to quantities; price adjustments come later-hence the so-called price-adjustment lag. The stimulus to final output resulting from these quantity adjustments is transmitted back to earlier stages of production via increases in the demand for labor and material inputs, as well as for intermediate (semi-finished) products. Demand pressure passes downward to lower stages of production where prices -especially those of raw materials and other basic commodities-are particularly sensitive to increases These prices begin to rise almost imin demand. mediately as do certain competitive (non-union) wages whose adjustment is not delayed by long-term union contracts or collective bargaining agreements. Since the output of earlier stages of the productive process constitutes the input of later stages, these rising wages and prices pass back up through the interindustry structure in the form of increased costs. In short, demand pressure is transmitted downward thus invoking cost increases that ripple upward until they finally reach the finished product stage. Here businessmen, operating with fixed percentage markups over costs, pass these cost increases on in the form of higher prices. These price increases induce workers operating under collective bargaining agreements to bargain for cost of living increases in the next wage contract. The resulting increases in labor costs are passed through into still higher prices. Such cost-induced price increases, of course, constitute the delayed price response to the prior shift in 4 ECONOMIC REVIEW, demand. These cost and price increases provoke further rounds of wage and price increases that add impetus to the inflation through the channel of costs. It should be noted that the expectations lag is in operation throughout the early and intermediate stages of the inflation. Price anticipations are based on perceived trends and so do not change quickly. Because these trends are estimated largely from past experience, it takes time to adjust expectations to higher current rates of inflation. Thus when the actual rate of inflation begins to rise, the expected rate is not immediately affected. Only after the actual rate has exceeded the expected rate for a time will the latter start to rise. But it will rise slowly at first because it continues to reflect the lower past rates of inflation. Eventually, however, the expected rate rises faster and faster as people become acclimated to the higher actual rate and adapt their expectations to it. Once aroused, inflationary expectations feed back into the current rate of inflation as firms and unions seek to raise prices and wages at the same rate as they expect prices in general to rise. At this stage of the inflationary process, where cost increases are passing through into price increases and the expected rate of inflation is rising to catch up with the actual rate, the main impact of the inflationary stimulus shifts from quantities to prices. The temporary output and employment effects diminish, and purely inflationary price effects dominate.3 The lags, however, may be long; and the price system may take several years-years that may even encompass a business Particular recession-to complete all adjustments. 3 Note that the expectations lag temporarily influences real economic activity by causing the nominal and real prices of factor inputs to move (i.e., inflation-corrected) Put differently, the expected in opposite directions. inflation component of nominal wage and interest rates may not adjust sufficiently to offset actual inflation thus causing changes in the corresponding real (price deflated) values of those variables. For example, when an inflationary stimulus begins, nominal wage and interest rates adjust upward by the amount of the expected inflation. Because the expected rate of inflation lags behind the actual rate, however, the adjustment is incomplete and consequently real wage and interest rates fall. Businessmen take advantage of the falling real wage and interest rates by hiring more labor and borrowing from banks in order to expand production. These effects, together with decreases in the demand for money (and hence rises in spending) induced by higher nominal interest rates, result in a temporary rise in real output The stimulus ends, however, when and employment. expected inflation catches up with actual inflation. When this happens, nominal wage and interest rates fully reflect the actual rate of inflation and real wage and interest rates return to their initial levels. Similar real effects operate on the downside when actual inflation falls below expected inflation. In this case the expectations lag results in a temporary rise in real wage and interest rates and these rising real costs act to temporarily depress real economic activity below its equilibrium level. SEPTEMBER/OCTOBER 1979 costs and prices that have lagged behind the others will have to catch up in order to restore the equilibrium price relationships disturbed by the inflation. Also, at the end of a serious inflation, inflationary psychology may continue to exert upward pressure on wages and prices even after other basic pressures have begun to diminish. Wages and prices continue to mount as inflationary expectations adjust to catch up with the accelerating price movements of the recent past. This catch-up process may extend well into an ensuing recession, thus resulting in the anomaly of rising prices despite slackening demand. In the process just described, the chain of causation or pattern of adjustment runs from spending to output to costs to prices to expectations and back again to input and product prices. This sequential process is represented schematically in the chart, which also shows the location of the price-adjustment and expectations lags. It should be emphasized, however, that an important constraint exists to the continuous operation of this process. Specifically, the process cannot continue unless it is constantly refueled with additional supplies of money. A determined stand by the monetary authorities to deny this fuel will eventually bring inflation to a halt. inflationary expectations continue to mount. But as demand pressures slacken throughout successive stages of production, the resulting weakening of costs eventually leads to an easing of prices of final goods and services. As the actual rate of price increase falls and remains for a time below the expected rate, expectations of further rises are gradually revised downward. Since the expected rate is a determinant of the actual rate, the decrease of the former will produce a further deceleration of the latter. But the price-adjustment and expectations lags draw out the winding-down process and the actual rate of inflation will converge on its new equilibrium level only slowly. Factors Retarding Price Adjustment What characteristics of a modern economy account for the lags described in the foregoing sections? In the case of the price-adjustment lag, the answer is fairly straightforward. The lag arises from institutional arrangements and behavioral practices that operate to make many costs and prices relatively unresponsive to short-run shifts in demand.4 For example, long-term contracts fix some wages and prices for substantial intervals of time. In addition, regulated rates of The Winding-Down Process The lags described in the preceding section continue to operate even after an effective program of monetary restraint is implemented. Spending slackens, output falls, and input demands decline; but prices nevertheless continue upward because recent cost increases are still working their way through the system and because THE INFLATIONARY PRICE-ADJUSTMENT A 4 Note that the rigidities or inflexibilities described in this section refer not to levels of particular prices and costs but rather to rates of change of those variables. The concept of rigid price levels, of course, implies the complete absence of price change and is thus irrelevant to the analysis of inflation, which deals with continually rising, not constant, prices. The problem of sluggish inflation is not one of downward inflexibility of price levels, but rather the resistance of rates of price increase to departures from the established trend. TRANSMISSION MECHANISM LAG EXPECTATIONS LAG The inflation process is set in motion by a rise in aggregate demand or spending to a level in excess of the economy’s long-run capacity to produce. Businessmen initially respond to the excess demand by drawing down inventories and stepping up production. Output expands and the demand for resource inputs increases. The resulting upward pressure on resource pricer pusher up unit production costs upon which product prices are bared. Pricer therefore rise. There price increases induce resource owners to bargain for cost-of-living pay increaser to restore real earnings eroded by inflation. This feedback of product prices into resource costs generates additional rounds of inflation. The price-adjustment log is the time it takes for demand pressure to affect product prices through the complicated channel of factor costs. If the inflation persists, it eventually induces upward revisions of expected future rates of inflation, which feed back into wages and prices to add further momentum to inflation. Slow to build up and hard to reverse, price expectations may keep wages and prices increasing long after demand slackens. The expectations lag accounts for the slow adjustment of the expected future rate of inflation to changes in the actual current rate. FEDERAL RESERVE BANK OF RICHMOND 5 public utilities can be changed only after lengthy administrative and judicial proceedings. Moreover, there exists a variety of legal restrictions on wage and price flexibility, including minimum wage laws, rent controls, oil price ceilings, interest-rate ceilings, and the like. There are also numerous governmentsanctioned’ trade-restriction and price-fixing arrangements, such as production and marketing quotas for certain agricultural products, import quotas, licensing and other restrictions to entry of professions, and resale price maintenance agreements. All these contribute to the stickiness of the prices affected. An important factor retarding price adjustment is the inflexibility of many wages. Owing to long-term labor contracts, delays in the adjustment of price expectations, and perhaps also workers’ money illusion (i.e., failure to distinguish between nominal and real wage increases), some wages tend to lag behind inflation during certain phases of the business cycle. When inflation accelerates in the upswing, these wages are often slow to respond. In later stages of a sharp inflation, however, wage increases may outstrip price increases. Wages may then tend to rise ahead of prices to restore real earnings eroded by past price increases and to protect real earnings from anticipated future price increases. But the lag becomes noticeable again when inflation is subsiding as the rate of wage increase shows little tendency to decelerate even though the rate of rise of other prices is diminishing. Another important factor contributing to price rigidity is the price-setting behavior of large-scale firms operating in manufacturing industries. In the typical imperfectly competitive market, the price mechanism works in the long run as prices adjust to clear the market. In the short run, however, prices are relatively inflexible. Firms do not typically alter their prices in response to short-run shifts in demand. At least three explanations of this behavior have been offered. According to the first, firms have difficulty distinguishing between real demand shifts specific to a particular industry and shifts in nominal aggregate demand for the output of all industries. Because they alter only the composition and not the overall level of aggregate demand, specific demand shifts may occur without inducing a corresponding rise in input prices. By contrast, an economy-wide demand increase leads to an equiproportional bidding-up of input prices. In the former case, industry cost and supply curves remain unchanged. In the latter case, however, cost and supply curves shift sharply upward, just matching the rise in demand. The rational entrepreneurial reaction to the former is a change in 6 ECONOMIC REVIEW, some quantity whereas the rational reaction to the latter is a change in prices. But when aggregate demand alters there is a good likelihood that each producer will tend to regard the shift in demand for his product as special to him and so adjust quantity Only later, when cost changes rather than price. become widespread, will producers correctly perceive the demand shift as an economy-wide phenomenon. Then and only then will they start to change their prices. A second explanation stresses the administrative inconvenience and costs of frequent price changes-for example, the expense of printing and disseminating new price lists-as a reason for sticky oligopolistic prices. Still a third explanation of price inflexibility in imperfectly competitive markets begins with the observation that in the complex and dynamic modern industrial economy there is always much uncertainty about the equilibrium or market-clearing price. Firms operating in this kind of environment try to avoid the market disruption, confusion, and perhaps even outright price warfare that could result if each sought individually to determine the equilibrium price. In order to prevent such confusion from developing, firms seek ways to coordinate price changes. Such coordination, if successful, will assure that firms raise prices in unison and that price changes will not occur when demand shifts are thought to be temporary and reversible. Firms have devised several techniques to facilitate price coordination. Infrequent price changes are perhaps the simplest of these. Price leadership constitutes another such technique. In this case, one firm-often the largest in the industry -initiates price changes, and the rest more or less automatically follow that price change. Perhaps the most widely used mechanism for coordinating price behavior, however, is to base selling prices on unit labor and material costs that are the same for all firms in the industry. This practice is accomplished by the use of so-called unit cost or mark-up pricing formulas. Unit-cost pricing is thought to be characteristic of many of the large oligopolistic firms that operate in American industry. The mark-up technique of pricing involves setting prices on the basis of a constant percentage markup applied to production costs per unit of output at some standard level of plant operation or capacity utilization. The chief cost components are unit labor and unit material costs. Included in the markup is the firm’s profit margin per unit of output. This profit margin is usually set to provide a fixed target rate of return on equity. Since percentage markups or profit margins are treated as fixed constants in unit-cost pricing formulas, it follows SEPTEMBER/OCTOBER 1979 that formula-based price changes are strictly costdetermined, that is, they result solely from changes in unit labor and material costs. Prices respond to costs, not to demand. Moreover, since standard unit labor and material costs are roughly the same for all firms in the industry, unit-cost formulas insure that price changes will be uniform throughout the industry, thereby minimizing the risks of competitive price undercutting. Thus unit-cost pricing is consistent with the slow response of prices to shifts in demand, the dependence of prices on costs, and the coordination of prices in concentrated industries. A Source of Confusion The long delay in the adjustment of many prices to demand pressure makes it difficult to distinguish cause from effect in the inflationary sequence and contributes to confusion in popular understanding of the source of inflation. The sequence of cost-price responses observed in concentrated industries, for example, may suggest that inflation is initiated by autonomous increases in costs. The use in such industries of mark-up or unit-cost pricing formulas, and the resulting dependence of prices on costs, means that firms do not raise prices unless there occurs a prior increase in costs. This cost-price sequence, with costs rising first and prices later, makes it appear that inflation is caused by rising costs when in fact excess demand is usually the culprit. For, as pointed out earlier, both the cost increase and the ability of firms to pass on this increase are due to a prior expansion in aggregate demand for final goods and services. Similarly, the long lag in the response of oligopolistic prices to prior inflationary pressures may make it appear that large firms in concentrated industries play an important independent role in generating inflation. Due to the price-adjustment lag, catch-up price increases in such industries are often delayed until well after restrictive policies start to turn the inflation process around. The apparent perverse behavior of prices when markets are slack and unemployment is rising fosters the notion of giant firms arbitrarily exercising monopoly power by effecting autonomous increases in prices totally independent of economic conditions.5 But it should be 5 This simplistic cost-push view is implausible and at odds with the orthodox theory of monopoly behavior. According to the latter view, a monopolist sets a relative price for his product that maximize; profits in real term; and maintains that real price by adjusting his nominal price to allow for inflation. The logical implication is that given the degree of monopoly power monopolists would have no incentive to raise prices other than to keep pace or catch up with inflation. With real prices already established at profit-maximizing levels, any further upward adjustment would reduce profits. On the other FEDERAL RESERVE BANK noted that the price increases in such situations can be interpreted as constituting a delayed reaction to prior economic conditions of expanding demand and demand-induced cost increases and a near perfect example of the price-adjustment lag. They might well be interpreted as a delayed manifestation of the effects of a general inflation rather than of inflationinitiating price behavior. From this interpretation it follows that big firms are not inflation starters. Nevertheless they may play an important role in the working out of the inflationary process. Specifically, they may be inflation prolongers. It follows logically from the mere existence of the price-adjustment lag that administered prices do not usually initiate inflation. Instead they tend to slow it down. By doing so, however, they extend the duration of inflation and protract the period necessary for its reversal. Administered prices are a problem because they act to prolong inflation once it gets started and because they delay the success and exacerbate the adverse side effects of anti-inflationary stabilization policy. Just as in an upward price spiral sticky administered prices retard the spread of inflation, so when the spiral is unwinding they delay its deceleration and impede the return to price stability. The Expectations-Formation Lag As noted earlier, the complicating effects of the so-called priceadjustment lag are reinforced by a comparable lag in price expectations, which tend to adjust slowly to the actual rate of inflation. Why does this expectations lag occur and what determines its length? The expectations lag occurs because people have imperfect foresight and cannot predict the future with certainty. If the future were known with absolute certainty, there would be no forecasting errors and no expectations lag. Anticipations would adapt themselves instantaneously to realized outcomes, and the expected rate of inflation would always be the one that actually occurred. Although the future is emphatically uncertain, people nevertheless try to make informed guesses about it, often on the basis of analyses of the past. Thus one longstanding explanation of the lag holds that price predictions are based on perceived trends as estimated from past price experience perhaps modified by current information. hand, if prices are being raised to exploit hitherto unexploited monopoly potential, the question arises as to why In either case, those gains were foregone in the past. rising real prices imply non-rational behavior. It is true that rising real prices would be consistent with profitmaximizing behavior if the degree of monopoly power were increasing. But there is little empirical evidence that monopoly power is on the rise. OF RICHMOND 7 Because they reflect the past, these trends and the predictions derived from them change slowly over time. Transitory deviations from the trend have little impact on expectations. Thus if the current rate of inflation departs from the recent trend, the expected rate will remain unchanged for a time. It requires the cumulative influence of a sustained change, or at least a very large change, in the actual rate to produce a change in the expected rate. And even then the adjustment will not be instantaneous or complete. The expected rate will continue to lag behind the current rate. As time passes, however, and the new rate persists, it will eventually begin to dominate the trend. The price experience of the more distant past gradually will be forgotten and the expected rate will finally converge on the actual rate. The length of time required for this convergence cannot be specified with any degree of precision. It depends on what interval or intervals of the past people consider in formulating their expectations and, in particular, on what relative weight they assign to the price experience of the more distant past. In general, the greater the weight assigned to the distant past, the longer will be the lag. The lag may indeed be quite long because people, in formulating their expectations, may look not merely at a chronological succession of prior years but rather at the relevant phases of a succession of past inflation-recession cycles and at the public policy response to these episodes. Looking back at a succession of what might be regarded as stop-go policies, for example, people may expect the recession phase to be brief and to be followed immediately by an expansionary phase that experience suggests may bring new inflationary pressures. To use a popular expression, they may “look across the valley.” A second factor that may affect the length of the expectations lag is the variability of the rate of inflation about the trend. The greater the variability of the actual rate of inflation the greater is the financial incentive to forecast it more accurately, i.e., to predict the variations, not just the trend. Accordingly, forecasts of the inflation rate will be revised more frequently and will rely more heavily on current and recent information. The whole time frame employed in observing past price behavior and past forecasting errors will contract and shift toward the present. Correspondingly, the expectations lag will shorten. Time Lags and Demand Management Policies This article has focused on the role of time lags in the inflationary transmission mechanism. These lags raise several potential problems for demand manage8 ECONOMIC REVIEW, ment (i.e., monetary and fiscal) policy. First, if the lags are variable and hard to predict, policymakers may experience difficulty in accurately forecasting when their policy actions will take effect on the rate of inflation. In this case discretionary policymaking becomes a potentially risky undertaking. That is, unpredictability of the lag effects of policy may render the latter destabilizing rather than stabilizing. For this reason, some analysts have advocated the abandonment of discretionary policy in favor of fixed policy rules. Other observers, however, argue that the lags are not so variable and unpredictable as to defeat effective policy forecasting and control. Actually, little is known about the variability of the lags. The evidence is simply not sufficient to settle the issues. Whether rules would be superior to discretion in the conduct of stabilization policy remains an open question. Additional policy problems arise from the length rather than the variability and unpredictability of the lags. As previously mentioned, the price-adjustment and expectations lags extend the time it takes for a policy-induced change in spending to work its way through to the rate of inflation. Moreover, they influence the pattern of response of quantities and prices to inflationary stimuli. Specifically, they cause the quantity response to precede the price response. When spending changes, output and employment adjust first, prices only later. What are the policy implications of such consequences? First, owing to the slow response of inflation to a spending change, anti-inflationary monetary-fiscal policy can be expected to produce observed results only after a corresponding lag. Quick monetary remedies for inflation are not likely to be found. Moreover, since the initial impact of a change in spending is on output and employment rather than on prices, a move to monetary restraint would almost surely entail a recession or at least a marked retardation in the rate of expansion of the economy. In short, a temporary but protracted period of unemployment, idle capacity, and sluggish growth might have to be endured if restrictive policy were to ‘be successful in permanently lowering the rate of inflation. Second, due to the difference in timing of the response of output and prices to a spending change, anti-inflationary policy may appear impotent or even Because inflationary movements tend to perverse. subside so slowly, prices may continue to rise long after output and employment have turned down. Thus inflation can persist even in slack markets, a condition variously known as inflationary recession, stagflation, or slumpflation. During such periods, SEPTEMBER/OCTOBER 1979 monetary restraint may be wrongly blamed for causing both the slump and the accompanying inflation. At such times, monetary authorities, anxious to achieve quick and relatively painless results, may be tempted to abandon the policy of restraint as ineffective at best and harmful at worst. Third, the same asymmetrical pattern of response -output first, prices only much later-may create the dangerous illusion that expansionary policy in the upswing can achieve permanent gains in output and employment at the cost of little additional inflation. If inflation proceeds as described in the foregoing sections, this view may have unfortunate consequences. For it is virtually impossible to peg output and employment above their natural or equilibrium levels without continuously accelerating the rate of inflation. In any case, time lags may well compound the problem of curbing inflation by leading to the undue prolongation of expansionary policy, thus increasing the momentum behind inflation when it finally occurs. Indeed, these time lags, together with society’s commitment to full employment, may bias monetaryfiscal policy toward inflation over the entire policy cycle. The lags cause output and employment to adjust before prices. Consequently, on the upside the desired output results of demand expansion occur before the undesired inflationary results thus encouraging the prolongation of policy stimulus. On the downside, however, the painful effects come first as restrictive policy produces undesired and costly rises in unemployment with little initial effect on the rate of inflation. Not surprisingly, the appearance of painful rather than desired results leads to impatience with restrictive policies and often to their premature reversal. The policies are abandoned before they have a chance to eliminate the inflation generated in The net result is that rethe preceding upswing. strictive policies reduce the inflation rate less than expansionary policies raise it and inflation ratchets upwards. Direct Controls If orthodox demand-management policies cannot curb inflation without causing painful rises in unemployment, then what other means can be utilized? One suggestion is to reimpose direct Several arguments have been wage-price controls. advanced in behalf of controls. The most naive is that controls can constrain the rate of inflation while the authorities pursue demand-expansion policies necessary to insure full employment. This argument assumes that the rate of inflation can be permanently pegged in the face of persistent excess demand. In fact, however, the rate of inflation tends to gravitate to an equilibrium level where excess demand is zero. Controls can delay the adjustment to that level but they cannot stop it. The equilibrium rate of inflation will inevitably be reached either via evasion, i.e., through black markets, or after the controls are lifted. In the final analysis, controls will have no effect on the equilibrium rate of inflation, which is determined by aggregate demand. It logically follows. therefore, that a decrease in aggregate demand, not the application of controls, is the sine qua non for the reduction of the inflation rate. Controls and the Expectations Lag A more sophisticated argument for controls calls for using them as a supplement to monetary and fiscal policy. The prescription here calls for first employing a monetary-fiscal policy sufficient to eliminate excess aggregate demand and then using controls to speed the actual rate of inflation to its new equilibrium level. In this view, a controls program properly coordinated with restrictive demand policy could help short-circuit the painful process of winding-down a stubborn inflation. The foregoing argument rests on the belief that controls can exert an independent influence on otherwise sticky price expectations. Eradication of inflationary expectations is of course a prerequisite to the elimination of inflation, since the former is a determinant of the latter. The problem is how to dampen The orthodox method is to create expectations. slack (excess supply) in the economy, thus causing the actual rate of inflation to fall below the expected rate, inducing a downward revision of the latter. Owing to the expectations lag, however, this adjustment may be a slow process and a prime example of how sluggish price anticipations can impede the return to price stability. Direct controls are viewed as a means of speeding the expectations adjustment and thus reducing the duration and severity of the economic slowdown necessary to bring inflation to its equilibrium level determined by aggregate demand. How could controls influence expectations ? Two mechanisms have been suggested. First, the mere announcement of controls might alter expectations in the desired direction. Second, by freezing prices or at least severely constraining their rate of rise, controls would cause the gap between the expected and actual rates of inflation to be greater than it would otherwise be. Assuming that the rate at which expectations are revised varies directly with the size of the gap, controls would thus accelerate the downward revision of expectations. FEDERAL RESERVE BANK OF RICHMOND 9 This view, however, probably overestimates the influence of controls. To have anything more than a temporary impact on price expectations, the controls must convince the public that the trend of prices when controls are in force is a reliable indicator of the likely future trend of prices after they are lifted. The public may be hard to convince, especially if the controls have failed to stop inflation in the past. Aside from this point, it is hard to understand why controls should have a stronger impact on expectations than, say, an announced policy of a permanent reduction in the growth rate of the money stock. What counts is not the means by which the government announces its determination to permanently reduce inflation but that those intentions be believed. Controls and the Price-Adjustment Lag It should be noted, moreover, that even if controls could reduce the expectations lag, they would tend to lengthen the price-adjustment lag, and this would impede the return to equilibrium. The attainment of economic equilibrium requires that two conditions be satisfied. First, inflation must be correctly anticipated, i.e., the expected rate of inflation must equal the actual rate. Second, the equilibrium structure of relative prices must be restored. Lagging prices and costs must be allowed to catch up to and reestablish their equilibrium relationships with other prices that have already adjusted to inflation. Controls interrupt this catch-up process by arbitrarily freezing all prices and costs. As a result, catch-up adjustments are postponed until the controls are lifted and account for the observed tendency for prices to rise sharply when controls are terminated. Note that this problem would not arise if controls could distinguish between legitimate catch-up and unwarranted anticipatory price increases. In practice, however, the two are virtually indistinguishable and controls prohibit both. By forcing the postponement of catch-up price increases, controls probably protract the inflationary process and lengthen the interval required for the rate of inflation to reach its equilibrium level. In the interim, the controls-distorted price structure generates inefficiency, resource misallocation, and income redistribution. Clearly, in the context of the explanation presented here, controls offer no solution to the problem of persistent inflation. Other Proposed Solutions One of the more promising, and yet untried, solutions to the lag problem is indexation, i.e., the inclusion of purWith all chasing power clauses in all contracts. contractual prices tied to automatic cost of living escalators, inflation would be transmitted more 10 ECONOMIC REVIEW, quickly and evenly throughout all markets, thereby permitting faster restoration of relative price equilibrium. On the downside, indexation could help reduce expectations more quickly, thus shortening the time required to remove inflation. Critics argue that indexation would intensify inflation. This criticism confuses the level of inflation with the speed with which inflationary impulses are propagated through the economy. There is no reason to believe that escalator clauses by themselves would have any effect on the rate of inflation. The speed of inflation would be increased, to be sure, but the level of inflation would not necessarily be any higher. Another partial solution to the lag problem would be to shorten the length of contracts. For example, trade-union collective bargaining agreements could be renegotiated more frequently as could long-term contracts for raw materials and energy supplies. A distinction should be made here between reforms designed to increase the speed of response of inflation to changes in aggregate demand and reforms aimed at reducing the equilibrium rate of unemployment, i.e., the unemployment rate that, given the inevitable frictions, rigidities, and imperfections in the economy, is just consistent with zero excess demand and stable (nonaccelerating, nondecelerating) rates of inflation. Indexation falls in the former category while socalled structural reforms fall in the latter. Structural reforms refer to microeconomic policies directed at improving the efficiency of labor and product markets. True, these policies may increase the responsiveness of inflation by eradicating market imperfections that inhibit price flexibility. But their chief purpose is to reduce the equilibrium rate of unemployment at which the demand for and supply of labor are in balance and at which any stable rate of inflation (including a zero rate) is possible. By so doing, structural policies may make it easier in at least two ways to bring inflation under control. First, they may render the equilibrium rate of unemployment and the corresponding zero or other desired steady-state rate of inflation a more acceptable policy option. By bringing equilibrium unemployment down to a socially tolerable level, structural policies reduce the risk that political pressure will be put on the authorities to peg unemployment at even lower levels that can only be maintained by a constantly accelerating rate of inflation. Second, they reduce the severity of the recession necessary to achieve price stability. Since a dampening of inflation requires a temporary rise in unemployment above its equilibrium level, it follows that any policy that reduces the latter also reduces the transitional level of unemployment. SEPTEMBER/OCTOBER 1979 Conclusion The policy implications stemming from the preceding analysis are straightforward. Admit that there are no quick and easy solutions to the inflation problem. Realize that any serious antiinflation program faces the formidable obstacles of entrenched inflationary expectations and eroded public confidence in the government’s ability and determination to fight inflation and to tolerate the resulting economic slack. Comprehend that, obstacles notwithstanding, the costs of removing inflation may be far less than the costs of accepting it as a permanent way of life.6 On the basis of this cost-benefit analysis establish a permanent target rate of inflation. Rely on demand-management (i.e., monetary and fiscal) policies to bring inflation down to the desired target level. Assuming that level is zero, reduce the rate of monetary expansion and trim budget deficits until the growth rate of total spending (aggregate demand) is equal to the long-term trend growth rate of real output (aggregate supply). Eschew controls. Recognize that a reduction of inflation necessitates a transitional rise in unemployment above its equilibrium level with the extent and duration of the rise depending upon the speed with which inflation is to be removed. In other words, recognize that quick eradication of inflation requires high excess unemployment for a relatively short period whereas slower eradication requires a lower level of excess unemployment for a prolonged period. Choose the desired path to price stability always realizing that the choice is between higher excess unemployment for a short time or lower excess unemployment for a long time. Acknowledge also that the severity and duration of the recession necessary to eliminate inflation depends upon the responsiveness of wages and prices to deflationary pressure and upon the adjustment speed of price expectations. Use indexation, if possible, to increase the downward flexibility of wages and prices. Similarly, seek to influence expectations by preannouncing the inflation target and by adhering to a path consistent with achieving it. Once inflation is removed, maintain price stability by avoiding expansionary policies that generate excess demand. Balance the budget each year and maintain a money stock growth rate roughly equal to the long-term growth rate of real output. If society is unwilling to tolerate the equilibrium unemployment rate associated with price stability, design structural policies to lower that rate by improving the efficiency of labor and product markets. If society is still unwilling to accept that rate, face up to the fact that price stability may be impossible to achieve. 6 A policy of accommodating permanent inflation is likely to lead to a pattern of accelerating and highly erratic (variable) inflation. Such inflation, by virtue of its variability and unpredictability, imposes substantial costs on the economy. Not only will there be repeated falls in output and employment whenever the inflation rate drops, but the very unpredictability of volatile inflation increases business uncertainty, makes capital investment decisions riskier. diverts energies and skills from industry and production into speculative activities designed to beat the inflation, and reduces the information content of market prices thereby making the price system a less efficient mechanism for coordinating economic activity. All this makes for a wasteful and inefficient use of resources that results in a lower level of output than the economy is capable of producing. The costs of inflation can be measured in terms of this lost output. Similarly, the benefits of removing inflation can be reckoned in terms of the increased output that would result. APPENDIX THE TREATMENT OF THE OF LAGS IN SIMPLE ANALYTICAL INFLATIONARY TRANSMISSION The inflationary mechanism and its constituent lags can be summarized in the form of simple analytical models.1 Economists have long used such models to study how lags affect the speed and duration of inflation. More recently, such models have been employed to estimate the impact of lags on the effectiveness of anti-inflationary policy. These 1 For a thorough review of inflation models see the article by Laidler and Parkin [7] especially pp. 774-81 that describe models similar to the one presented in this appendix. FEDERAL RESERVE BANK MODELS MECHANISM models specify the chief determinants of the current They also and expected future rates of inflation. specify the lags linking the variables and determining the pattern of their interaction over time. Comprising these models are a price-adjustment equation and The pricean expectations-formation equation.2 adjustment equation explains how the current rate of inflation responds to inflationary expectations and to 2 Some models of the inflationary process contain equation that explains how inflation-generating aggregate demand is determined. OF RICHMOND a third excess 11 lagged excess demand-the lag on the latter variable representing the price-adjustment delay. The expectations-formation equation explains how price anticipations are generated and revised in the light of past price experience. As shown below, the latter equation expresses the expected future rate of inflation as an exponentially-declining weighted average of past rates of inflation. A crude version of this two-equation model is presented in the following paragraphs.3 It should be strongly emphasized, however, that the model constitutes a severe oversimplification of a complex process and thus should be interpreted with some skepticism. Presented solely as an illustration, the model purposely omits many of the variables and behavioral relationships that a more complex, sophisticated, and realistic model would contain. The Price-Adjustment Equation Most models of the inflationary mechanism contain an equation that explains how the current rate of inflation is determined, i.e., the rate at which businessmen mark up theirprices. One such equation shows the rate of price inflation p varying directly with lagged excess demand x-1 and with the expected rate of price increase pe. The equation is written as follows: (1) p = ax-l + pe-1 where p is the current rate of inflation (expressed as a percentage rate), x-1 is excess demand lagged one period, pe-1 is the present period’s expected rate of inflation forecast at the end of the preceding period, and a is a coefficient specifying how much each unit of lagged excess demand contributes to the rate of price increase. The excess demand variable x is measured in terms of real output since businessmen initially respond to changes in demand by altering quantity produced. More specifically, excess demand x is represented by the difference between actual and capacity real output. Actual output can exceed capacity output because the latter is defined not as the absolute physical limit or maximum ceiling level of output but rather as the output associated with the economy’s normal or standard level of operation. Equation (1) states that if aggregate demand and supply are equal so that there exists no excess de- mand (x = zero), then actual price inflation p will just equal expected inflation pe-1, i.e., businessmen will be raising their prices at the rate at which they expect other businessmen to be raising theirs. If, however, an expansion in demand raises x above zero, businessmen will eventually react to the excess demand by raising prices at a rate in excess of the expected rate of inflation. This price response, however, is not instantaneous. For a while, quantities rather than prices tend to absorb the impact of excess demand as businessmen temporarily expand output and perhaps allow their inventories to be depleted. These quantity changes affect demands for and prices of factor resources and ultimately invoke cost increases that signal the desirability of raising the rate at which prices are marked up. Later, therefore, businessmen respond to the excess demand by raising prices. The same price-adjustment lag operates on the downside. Thus if a subsequent slackening of spending causes excess demand x to become negative (i.e., a situation of excess supply) the actual rate of price increase p will eventually fall below the expected rate pe-1. The key word here is eventually because the lag prevents prices from responding immediately to shifts in demand. The Price-Response Lag The one-period delay on the excess demand variable symbolizes the tendency for price adjustments to lag behind shifts in demand.4 This price-adjustment lag is meant to account or the time it takes for demand pressure to work backward through the interindustry structure and for costs to work forward. To summarize, the association of the price-adjustment lag with the excess demand variable x implies that the impact of a shift in demand is initially registered on x. That impact is not immediately transmitted to prices, however. Instead it is transmitted first to quantities and subsequently to costs. Prices do not respond until rising costs induce them to do so. The Expectations-Formation Equation The second equation of the model is the expectations-formation equation. It is written as follows: (2) or, alternatively, (2a) 3 The model presented here is adapted from similar models developed by Phillip Cagan and David Laidler. See Cagan [2; pp. 94-6] and Laidler [5, 6]. For an elementary description of Laidler’s model, together with a diagrammatic illustration of its dynamic properties see Laidler and Parkin [7; pp. 776-8]. 12 ECONOMIC REVIEW, pe = bp + (l-b)pe-1 as pe - pe-1 = b(p-pe-1). 4 More sophisticated models would express the delayed price adjustment as a distributed lag, i.e., a lag spread over a number of time periods. SEPTEMBER/OCTOBER 1979 Equation (2a) states that the change in the expected rate of inflation pe-pe-1 is proportional to the amount by which the period’s actual inflation p deviated from expected inflation as forecast at the end of the preceding period pe-1 with the factor of proportionality b having a value between zero and unity. Embodied in the equation is a particular theorythe so-called adaptive-expectations or error-learning hypothesis-of how inflationary expectations are formed. According to the error-learning hypothesis, people formulate expectations about the inflation rate, observe the discrepancy between the actual and anticipated rates, and then revise the anticipated rate by some fraction of the error between the actual and anticipated rates. Expectations are revised in proportion to the error associated with the previous level of expectations. It can also be shown that the adaptive-expectations hypothesis is equivalent to the theory that people formulate price expectations from prior price experience by looking at a geometrically-weighted average of past rates of inflation with the weights diminishing exponentially as time recedes. This alternative interpretation of the adaptive expectations hypothesis is written as follows : the speed of adjustment of expectations. A steep slope represents a short weighting scheme, implying swift adjustment, and conversely for a relatively flat slope. The slope itself is determined by the magnitude of the fraction (1-b). A value of (1-b) close to zero implies that the weights decline rapidly as time recedes, and so future price expectations depend primarily on recent experience. On the other hand, if (l-b) is closer to 1 in value, rates of inflation from the more distant past enter the equation with higher weights, and recent price information is discounted more heavily. Econometricians who have attempted to fit equation (2b) to the statistical data have found the fraction (l-b) to be both significantly greater than zero and less than 1. These findings imply that while people generally assign higher weights to more recent phenomena, these weights do not dominate the cumulative weight of all past price experience. In short, price anticipations continue to reflect past price experience, which explains why the expected future rate of inflation does not adjust instantaneously to the current rate. To summarize, equation (2b) states that the expected future rate of change of prices is based on a geometrically declining weighted average of past rates of change of prices. The equation therefore constitutes a precise specification of the commonsense notion that expectations are based on past experience with more emphasis given to recent, rather than distant, experience. Here is the summation operator indicating the mathematical operation of adding a succession or series of terms, in this case the weighted past rates of inflation. The summation index i represents each past time period starting with the most recent (i=O) and extending backward to the most distant (i = ). The variables p-i are the past rates of inflation, one for each of the i periods stretching backward into time. Attached to each past rate of inflation p-i is a corresponding weight that measures the degree of influence that each p-i has on the formation of price expectations pe. The weights are expressed as (l-b)i, one for each of the i time periods. Since, as mentioned previously, the coefficient of expectations b is a fraction whose magnitude lies between zero and one, it follows that the term (l-b) will also be a fraction. And since any given fraction raised to progressively higher integral powers yields successively smaller numbers, it follows that the weights (l-b)i must decrease as the exponent i increases, i.e., the weights must diminish the further back in time one looks. the equation collapses to pe = p, i.e., anticipated and actual inflation are always identical. Graphically, the weights are distributed along an exponentially declining curve whose slope reflects On the other hand, the lag will be longer the closer b is to zero. In the extreme case where b FEDERAL RESERVE BANK The Expectations Lag The preceding discussion clearly implies that the length of the expectations lag can be defined in terms of the coefficient of expectations b. The coefficient b itself measures the speed of adjustment of expectations to experience, i.e., the quickness of response of pe to realized actual rates of inflation p. The average length of the expectations lag is the counterpart of the speed of adjustment. This lag is expressed as (l-b)/b. The closer b is to 1, the shorter the lag. In the extreme case where b equals 1, the lag is nonexistent, and the expected rate of inflation adjusts instantaneously to the current rate. Thus when b is set equal to 1 in the expectationsformation equation (2) OF RICHMOND pe = bp + (1-b)pe-1 13 equals zero, the lag is infinitely long, i.e., the expected rate of inflation never changes. This result can be demonstrated by setting b at zero in equation (2), which yields pe = pe-l, showing that the expected rate of inflation always remains unaltered from the preceding period. In short, if b is zero, the lag is of infinite length and expectations never change regardless of what is happening to the actual rate of inflation. One possible shortcoming of the adaptive expectations or error-learning model is that it regards the speed of adjustment or coefficient of expectations b as a fixed constant. And since b determines both the length of the expectations lag and the slope of the weighting pattern used to distribute the lag, it follows that the model, by implication, also treats these phenomena as given constants. This treatment is surely too restrictive. Some analysts think, contrary to the model, that the coefficient b is capable of being influenced by outside information and by the behavior of the rate of inflation itself. For example, it has been suggested that the sensitivity of price anticipations is greater and the corresponding adjustment lag shorter for high and volatile rates of inflation than for low and steady rates. Finally, some observers believe that the expectations coefficient can be influenced by government policy. In fact, this idea constitutes one rationale for direct wage and price controls.5 The Complete System Taken together, the price-adjustment and expectations-formation equations summarize the operation of the inflationary transmission mechanism. These two equations explain the mutual determination of actual and expected rates of price increase. They also indicate the iterative interaction process whereby the expected inflation rate influences the actual current rate, which in turn becomes a determinant of next period’s expected rate, which feeds back into next period’s actual rate, etc. Moreover, the model demonstrates how inflationary expectations operate to lengthen the lagged adjustment of prices to short-run shifts in demand. This latter result is obtained by substituting the expectations-formation equation into the price- 5 The argument here is that, by directly altering the expectations coefficient, controls could speed the downward adjustment of price expectations necessary for the removal of inflation. Of course controls might also speed the adjustment process simply by constraining the actual rate of inflation below the level that would otherwise occur at a given level of unemployment. In the latter case, controls would take the expectations coefficient as a given constant. 14 ECONOMIC REVIEW, adjustment equation and then solving recursively for p. The resulting “reduced form” expression is: where the second term on the right-hand side of the equation represents the delayed price impact of excess demand attributable to the operation of the expectations lag. Equation (3) states that once price anticipations enter into price-setting behavior, they tend to prolong the inflationary process. They cause current prices to respond not only to last period’s excess demand but also to excess demand in the more distant past. Finally, the model identifies excess demand as the proximate source of inflation. Specifically, the model implies the following causal chain. (1) Inflation is determined by excess demand and inflationary expectations. (2) Inflationary expectations are generated by previous inflationary experience and hence by previous excess demand. (3) Therefore, excess demand-past and present -is the proximate cause of inflation. The inflation-generating role of excess demand is made explicit in the reduced-form equation (3) where past levels of x constitute the sole independent variables. The model does not explain how excess demand itself is generated. Such an explanation would require an additional equation expressing the relation between excess demand and the independent variables that determine it. At a very minimum, the list of independent variables would include the money stock since excess demand cannot be long sustained without the monetary growth necessary to support it. Note, however, that there is at least one situation in which excess demand would properly be treated as an independent variable and the money stock as a dependent variable. Such would be the case if society were committed to a full-employment objective in excess of the natural or equilibrium level of employment. Here the policymakers would be expected to pursue the target employment rate (or level of excess demand), passively permitting the money stock and the rate of inflation to adjust so as not to inhibit attainment of the full-employment goal. In this case the level of excess demand would enter the system as a datum to determine the size of the money stock. Thus, depending upon the policy regime, excess demand may appear either as an endogenous or an exogenous variable. SEPTEMBER/OCTOBER 1979 EXCHANGE RATE POLICY AND THE DUAL ROLE OF EXCHANGE RATE MOVEMENTS IN INTERNATIONAL ADJUSTMENT Marvin Goodfriend I.INTRODUCTION 1. The Problem of International Adjustment and Its Relevance Over the past few years, the foreign exchange value of the dollar has gradually emerged as a primary concern of United States monetary authorities. This growing concern is well illustrated by a sampling of Wall Street Journal headlines. For example, on January 10, 1978, a headline read “Federal Reserve Moves to Tighten Credit in Escalating Effort to Bolster U. S. Dollar.” By November 1 of that year, the dollar had depreciated 35 percent against the Swiss franc, with most of the depreciation occurring after midyear. Other major exchange rates also moved substantially against the dollar with the result that U. S. officials came under heavy pressure to intervene in the foreign exchange market to prevent a further decline in the value of the dollar. On November 1, U. S. authorities publicly announced their intention to support the dollar with appropriate monetary policies, including direct intervention in the foreign exchange market. This policy was reported on November 1 in the headline “Dollar Dilemma-Bold Currency Support Announced by the U. S. Raises Recession Risks.” Although there was no official commitment to a particular rate, the decision represented a significant change of policy. Never since World War II had the U. S. publicly committed itself to intervene directly in the foreign exchange market on such a large scale to stabilize the exchange rate. While exchange rates fluctuated considerably in the months following November 1, on the whole, the policy action succeeded in arresting the dollar depreciation. In July, however, the renewed depreciation of the dollar began to concern U. S. officials. This time the Federal Reserve responded directly to events on the foreign exchange market by raising the Federal funds rate. This action was reported in a July 7 headline reading “Tighter Credit Policy of Fed Facing Test on the Foreign Exchange Market.” 16 ECONOMIC REVIEW, What is the reason for this “dollar dilemma” as it was called in the November 1 headline? Why should a sudden large exchange rate movement concern policymakers? And even if there are valid reasons for concern, what prevents the monetary authority from pegging the exchange rate if it so desires? These are the questions the article seeks to address. The sources of this policy dilemma stem from the operation of the international adjustment mechanism. This mechanism refers to the manner in which interaction among world goods markets and the foreign exchange market produces terms of trade, trade balance, and exchange rate responses to economic disturbances. It must be emphasized that the word response here refers not only to the immediate effects of disturbances but also to any persistence of the effects that may be produced by the adjustment mechanism itself. An important issue related to persistence concerns how long it takes for adjustment to occur, and whether that adjustment is, in fact, stable. If noninterference in the foreign exchange market is to be an acceptable policy alternative, then the exchange rate must automatically converge to a stable equilibOtherwise, the exrium following a disturbance. change rate may in fact require official management. This article is partly concerned with examining potential adverse effects of allowing exchange rate Consemovements to play a role in adjustment. quently, the body of this article is devoted to investigating the function of exchange rate movements in the adjustment process. It turns out that the source of the dilemma involved in holding the exchange rate fixed is easily recognized once the role of exchange rate movements in equilibrating the foreign exchange market is made clear. Exchange rate movements are shown to function in two ways in the adjustment mechanism: first, by changing the relative price of domestic and foreign produced goods in international trade, and second, by revaluing stocks of domestic relative to foreign assets in portfolios. Analysis of the terms of trade effect of SEPTEMBER/OCTOBER 1979 exchange rate changes is contained in Part II, in a general discussion of the effect of exchange rate The revaluchange on goods market equilibrium. ation effect of exchange rate change is developed in Part III, where the focus is shifted to portfolio balance and foreign exchange market equilibrium. offset by reduced expenditure on other goods, total national expenditure need not be affected. In other words, an increase in oil imports, if it is simply a consequence of individuals switching expenditure from food or clothes to gasoline, is not necessarily associated with an increase in total expenditure or a trade deficit. The partial equilibrium results of Parts II and III may be condensed into two graphical relationships called the GM and FX curves, respectively. In Part IV, these two curves are brought together to show how the exchange rate and trade balance are determined in general equilibrium. The two roles of the exchange rate are thereby synthesized in a simple graphical framework which is then used to investigate some features of the adjustment mechanism as a whole. In particular, the free exchange rate mechanism is shown to automatically restore international economic equilibrium after a disturbance. The article concludes with a discussion of the sources of the policy dilemma faced by central banks operating under a flexible exchange rate. II. THE GOODS 3. Tradable Goods Prices and the Terms of Trade Sections 3 and 4 are devoted to establishing a relationship between the trade balance and the exchange rate that must hold in the context of goods market equilibrium. The relationship to be derived is represented graphically by the GM curve in Figure 1, where the U. S. trade balance is measured on the vertical axis and the dollar/pound exchange rate is measured on the horizontal axis. MARKETS 2. A Macroeconomic Framework for Analysis of the Trade Balance The national balance of trade is defined as the value of national exports (of goods and services) minus the value of national imports. The trade balance is also the excess of the value of national production in a given period over the value To see this, note that (1) of national expenditure. expenditure by domestic residents consists of purchases of domestically produced goods plus imports and (2) domestic output consists of goods sold to domestic residents plus exports. Therefore the difference between domestic output and expenditure is exports minus imports or the trade balance. All goods produced provide owners of factor resources (land, labor, and capital) with income in wages, rent, and profit equal to the value of the output produced. Since the value of national output equals the value of national income, the trade balance can also be thought of as the difference between national income and national expenditure, which, of course, may be positive, negative, or zero. This last relation has an important implication for analysis of the trade balance. A complete explanation of movements in the trade balance must be given in terms of factors affecting total national income or total national expenditure. For example, it is not sufficient to explain a movement of the trade balance into deficit by pointing to the fact that oil price rises have led to an increase in the value of oil imports. To the extent that the increased expenditure on oil is As shown in the graph, the GM curve will turn out to be upward sloping. The curve may be interpreted as follows. Let an exchange rate change occur for reasons outside the goods markets, e.g., due to The speculation in the foreign exchange market. maintenance of goods market equilibrium implies that an exchange rate rise (fall) is associated with a movement of the U. S. trade balance toward surplus (deficit). It bears emphasizing that the determinants of the slope of the GM curve are entirely unrelated to the foreign exchange market. Since the GM curve involves only the goods markets, it may be called a partial equilibrium relationship, It is incomplete in the sense that it can only determine the trade balance for a given exchange rate. If the exchange rate is to be determined also, another curve needs to be provided. This second FX curve comes out of the FEDERAL RESERVE BANK OF RICHMOND 17 partial equilibrium discussion of the foreign exchange market undertaken in Sections 5 and 6. In Section 8 the two curves are superimposed on the same graph and the simultaneous determination of the trade balance and exchange rate is discussed in what is called general equilibrium, i.e., equilibrium analysis involving both foreign exchange and goods markets. Proceeding with the analysis at hand, suppose the world consists of two countries, the U. S. and U. K. Residents in the U. S. produce corn and U. K. residents produce wool. Both goods are traded. The price of corn in dollars is denoted by $Pc and the price of wool in pounds is expressed as £Pw. Further, let be the exchange rate between the two currencies in dollars per pound. Suppose the trade balance were in equilibrium at the initial prices. But now suppose the exchange rate rises due to foreign exchange speculation. At initial goods prices $Pc and £Pw, the higher exchange rate raises the so-called commodity terms of trade, i.e., the relative price of wool in terms of corn. The terms of trade is the number of bushels of corn it takes to buy one bale of wool, i.e., bushels of corn given up to obtain one bale of wool, or simply bushels of corn The terms of trade can be written bale of wool using the three basic prices as Note that the dollar and pound units cancel in the foregoing expression, verifying that it does yield the commodity terms of trade. Because the relative cost of wool compared to corn is increased, world residents are induced to switch expenditure from wool to corn, and an incipient excess demand for corn and excess supply of wool develops at initial goods prices. Therefore, domestic currency goods prices must adjust to eliminate these incipient excess demands and supplies so as to maintain goods market equilibrium. In particular, (i.e., the ratio of the pound price of wool to the dollar price of corn) must fall to eliminate the incipient excess supply and demand at the higher exchange rate. But how are these price changes going 18 ECONOMIC REVIEW, to affect national income, national expenditure, and the trade balance? This question is addressed next. 4. Goods Prices and the Trade Balance Suppose increased demand for domestic output raises domestic currency output prices. National expenditure depends on the level of national assets. If the domestic currency value of national assets rises proportionally with domestic output prices, then expenditure rises proportionally with income. However, one component of assets is domestic money which by definition has a fixed domestic currency unit value. Therefore, even if the domestic currency value of other components of wealth rises proportionally with output prices, the value of total assets rises less than proportionally with output prices. Consequently, a rise in the price of output raises national income more than national expenditure, and induces a trade surplus. Analogously, a price fall induces a trade deficit:. In the two-country model analyzed here, the U. S. trade surplus (deficit) must be associated with a U. K. trade deficit (surplus) of equal magnitude. This means that when $Pc and £Pw adjust in response to an exchange rate rise, they must do so by moving in opposite directions, since only in this way can the associated trade imbalances remain equal and opposite in sign as required. This requirement, taken together with the requirement that must fall, implies that $Pc must rise and £Pw fall to maintain goods market equilibrium following an exchange rate rise. It also follows that the U. S. trade account, consistent with goods market equilibrium at the higher exchange rate, moves into surplus matched by an identical movement of the U. K. trade balance into deficit. Does the terms of trade return to its initial level after the exchange rate rise? Not if the residents of each country spend the predominant share of their income on goods produced domestically. Because the U. K. is now importing more than it exports, its residents now account for a larger share of world expenditure on goods than formerly. Therefore, world demand is shifted toward wool and away from corn. Since the quantities of these goods produced have not changed, the goods markets must have been brought into equilibrium by a rise in the relative price of wool in terms of corn. The terms of trade may be said to have moved in favor of U. K. residents and against U. S. residents as a result of the exchange rate rise, since the number of bushels of corn (U. S. product) that can be obtained in trade for a bale of wool (U. K. product) increases, and the number of bales of wool that can be obtained with a SEPTEMBER/OCTOBER 1979 bushel of corn decreases. at least, U. S. residents living as a result of the and U. K. residents enjoy In this sense, temporarily suffer a lower standard of depreciation of the dollar a higher standard of living. U. S. residents traveling in Germany and Switzerland in summer 1973 or fall 1978 noted how the depreciation of the dollar against the D-mark and Swiss-franc produced an especially large and rapid deterioration in the foreign purchasing power of the dollar. But it is because their dollar incomes did not increase at home that the tourists rightly felt that they were less able to afford foreign goods and services. In other words, these U. S. tourists suffered a large adverse movement in the terms of trade brought about by a sudden depreciation of the dollar on the foreign exchange market. Because U. S. residents suffer a reduced ability to consume foreign goods, they might pressure U. S. monetary authorities to forestall such a sudden dollar depreciation in an effort to prevent an adverse movement of the terms of trade. Yet, as will be made more clear in Section 5 and later in Section 8, such pegging of the exchange rate requires the monetary authority to give up control of the money supply. To the extent that money growth becomes inconsistent with domestic policy objectives, the monetary authority gets caught on the horns of a dilemma. III. THE FOREIGN EXCHANGE MARKET 5. Financing a Trade Surplus or Deficit with a Fixed Exchange Rate The determinants of national expenditure and national income are necessary but not sufficient to explain the trade balance. For example, the existence of a U. S. trade surplus means that the value of goods and services being purchased by U. K. residents from U. S. residents exceeds the value of goods and services being sold by U. K. residents to U. S. residents. In order to pay for, or finance, this trade imbalance, U. K. residents must be willing and able to run down their money balances and U. S. residents must be willing to accept corresponding increases in their money balances. That is, at the end of any period of time in which there is a U. S. trade surplus, a value of money wealth equal to the value of the trade surplus must flow from U. K. to U. S. residents. Put another way, a trade surplus (deficit) is necessarily associated with an inflow (outflow) of money. But U. S. residents generally choose to hold their money wealth predominantly in U. S. dollars while U. K. residents largely hold their wealth in pounds. Since the wealth transfer increases the relative share of world money wealth in the hands of U. S. residents, it raises the FEDERAL RESERVE BANK demand for dollars relative to pounds and thereby creates an incipient excess supply of pounds and corresponding excess demand for dollars. It follows that in order to provide a complete explanation of the trade balance, the mechanism that eliminates these associated incipient excess supplies and demands for pounds and dollars must be explained. In other words, the mechanism by which the foreign exchange market clears must be explained. The wealth transfer from U. K. to U. S. residents, by creating a net demand to trade pounds for dollars at the given exchange rate, puts downward pressure on this rate. If the monetary authorities of the U. S. and the U. K. desire to prevent the exchange rate from moving, then they can enter the foreign exchange market to accommodate the net desire to trade pounds for dollars. Under the fixed exchange rate regime, the central banks are committed to passively supply dollars for pounds or pounds for dollars as required to prevent exchange rate movements. Suppose, for example, that the Bank of England desires to hold the exchange rate fixed. To the extent that the U. K. trade deficit (U. S. surplus) creates an excess pound demand for dollars, the Bank of England would passively accept the unwanted pounds and pay out dollars as required to prevent an exchange rate change. By providing the foreign exchange market (and ultimately U. S. residents) with the dollars desired in payment for the U. K. trade deficit, the Bank of England in effect finances the trade deficit. In this case the U. S. trade surplus is said to be financed by foreign official transactions. By passively accepting unwanted pounds in exchange for dollars, the Bank of England gives up control of the pound money supply, i.e., allowing it to fall because of conditions on the foreign exchange market. The reduced rate of growth of the pound money supply necessary to stabilize the exchange rate may not be consistent with domestic goals of U. K. monetary policy such as price stability and stable real output growth. When this is the case, the Bank of England is caught on the horns of a dilemma in the following sense. If there is downward pressure on the dollar/pound exchange rate, in order to hold the exchange rate fixed, U. K. authorities will have to reduce the rate of growth of the pound money supply. A sharp sustained reduction in pound money growth could reduce real output growth. On the other hand, allowing the dollar/pound rate to fall, i.e., letting the pound suddenly depreciate, would cause an immediate worsening of the terms of trade for U. K. residents. OF RICHMOND 19 6. The Role of Exchange Rate Movements in Equilibrating the Foreign Exchange Market In the absence of a commitment to fix exchange rates, central banks need not participate in the foreign exchange market. When this is the case, the exchange rate is said to be free. How does the foreign exchange market clear in this case? What eliminates the excess pound demand for dollars on the foreign exchange market associated with a U. S. trade surplus ? This section is devoted to establishing a relationship between the trade balance and the exchange rate that must hold in the context of foreign exchange market equilibrium in the absence of central bank intervention. The relationship to be derived is represented graphically as the FX curve in Figure 2. As shown in the graph, the FX curve is downward sloping. The curve may be interpreted as follows. Let the U. S. trade balance be disturbed due to events in the goods markets, for example, a switch of consumer tastes from U. S. to U. K. cars. The maintenance of foreign exchange market equilibrium implies that a U. S. trade balance movement toward deficit (surplus) is associated with an exchange rate rise (fall). It bears emphasizing that the determinants of the slope of the FX curve are entirely unrelated to the goods markets. Since the FX curve involves only the foreign exchange market, it is only a partial equilibIt is incomplete in the sense that rium relationship. it can only determine the exchange rate for a given trade balance. If the trade balance is to be determined also, the GM and FX curves must be brought together. This is done in Section 8 where the simultaneous determination of the trade balance and exchange rate is discussed in general equilibrium. Economic agents holding portfolios of both foreign and domestic money carefully regulate the relative shares of different monies that they hold. The agents may be importers, exporters, multinational corporations, or individuals living on international borders. The desired portfolio shares vary according to individual circumstances. As discussed in Section 8, shares may also vary over time with the anticipated rate of depreciation, or appreciation, of domestic relative to foreign money. However, to avoid unnecessary complication, desired portfolio shares are here assumed fixed and agents are assumed to hold predominantly national money in portfolio. To monitor his portfolio share, an individual must first calculate the total value of his money balances. 20 ECONOMIC REVIEW, This requires that he choose a unit of account in which to measure his money wealth. It is important to emphasize that choice of this measure has no economic significance; it is equivalent to choosing between yards and meters to measure distance. Let the dollar be the unit of account. Suppose the individual has $M dollars and £M pounds in his portfolio. If the current market clearing exchange rate is ,then the portfolio has a total dollar value of $M + p • £M,which is the combined sum of dollars and pounds in portfolio valued in the dollar unit of account. It follows that the share of pounds in portwhich is the fraction folio is portfolio value held in pounds. fall (rise) in the exchange rate of total It also follows that a lowers (raises) the share of pounds in total money wealth. An exchange rate fall reduces the value of pounds in portfolios throughout the world, thereby lowering the average share of pounds in world portfolios. Now that the notion of portfolio share has been formally defined, it is easy to explain why asset holders care about portfolio shares. The above discussion abstracts from income earning assets, but, To the in general, individuals hold such assets. extent that individuals are risk averse, they maintain a diversified portfolio to minimize the overall risk for any acceptable average return on their portfolios. That is, they dislike “holding all their eggs in one basket.” A rise in the current exchange rate raises the share of wealth in the “pound basket” so to speak. Therefore, when the exchange rate rises (the arguSEPTEMBER/OCTOBER 1979 ment works in reverse for a fall) asset holders have incentive to sell off some pound assets for dollar assets to bring their portfolios back into balance. The mechanism by which a free exchange rate equilibrates the foreign exchange market in a period of U. S. trade surplus works as follows. The wealth transfer from U. K. to U. S. residents lowers the average desired share of pounds in world portfolios. Therefore, to achieve foreign exchange market equilibrium, the average actual share of pounds in world portfolios must be reduced. But the argument above has demonstrated that an exchange rate fall does in fact reduce the average actual share of pounds in world portfolios. An exchange rate fall is said to clear the foreign exchange market when it brings the net excess pound demand for dollars to zero; in other words, when it leaves world residents just satisfied to hold the existing stocks of dollars and pounds. In a period of U. S. trade surplus, a free exchange rate will be driven down until the foreign exchange market clears. It is important to emphasize that the exchange rate fall equilibrates the market by revaluing pounds in terms of dollars, thereby adjusting the relative value of pound and dollar money stocks until they are compatible with (1) the current distribution of money wealth between countries and (2) the average desired share of pounds and dollars in world portfolios. Under a flexible exchange rate, a trade surplus or deficit is not necessarily financed as it is under a fixed rate. That is, no agent, private or official, need enter the market to voluntarily exchange pounds for dollars to clear the market. However, this does not mean that from time to time central banks or private agents do not choose to enter the market to stabilize the exchange rate. It has been a characteristic feature of the current experience with flexible exchange rates that central banks have frequently chosen to buy or sell foreign money to prevent or retard exchange rate movements in order to forestall sudden changes in the terms of trade that could produce hardship for domestic residents. Likewise, private agents may choose to vary the share of foreign and domestic money in portfolio to protect themselves against anticipated future exchange rate movements. Exchange rates have been influenced considerably by this type of official and private portfolio management. But the important point is that the fundamental free exchange rate equilibrating mechanism requires neither private speculation nor official intervention. It operates basically through the revaluation effect of exchange rate changes. IV. AN ANALYSIS TRADE BALANCE OF THE EXCHANGE IN GENERAL RATE AND EQUILIBRIUM 7. The Simultaneous Determination of the Exchange Rate and Trade Balance As should be apparent by now, the exchange rate plays a dual role in the adjustment process. In Section 3, exchange rate movements were shown to affect the terms of trade (relative prices of traded goods), domestic currency goods prices, national income and expenditure, and the trade balance. In Section 6, for a given trade balance, it was shown how an exchange rate movement equilibrates the foreign exchange market by revaluing pound relative to dollar assets in portfolios. In one role, exchange rate movements function in the adjustment mechanism through their effects on relative goods prices. In the other role, the exchange rate functions through its effects on the relative value of foreign and domestic assets. In this section, both roles of exchange rate movements are brought together in one graphical framework to investigate the operation of the adjustment mechanism as a whole. This analysis extends that of previous sections by considering all three marketsU. S. goods, U. K. goods, and the foreign exchange market-simultaneously. Therefore, the analysis contained in this section is called general equilibrium, in contrast to the partial equilibrium analysis of earlier sections. The basic result of Section 4 is that an exchange rate rise is associated with movement of the U. S. trade balance toward surplus under conditions of goods market equilibrium. This result is represented graphically in Figure 3 as the goods market clearing (GM) curve. The GM curve shows all trade bal- FEDERAL RESERVE BANK U.S. Trade Balance Figure OF RICHMOND 3 21 ante-exchange rate combinations consistent with equilibrium in the goods markets. The curve slopes upward indicating a direct relationship between the trade balance and the exchange rate. The curve says that a rise in the exchange rate must be accompanied by a rise in the trade balance to keep goods markets in equilibrium. This is so because, as outlined in more detail in Section 4, the exchange rate rise disturbs the terms of trade. Changes in the dollar price of corn and pound price of wool are then required to But these domaintain goods market equilibrium. mestic currency price changes raise U. S. income and lower U. K. income relative to expenditure, and thereby move the U. S. trade balance into surplus. The basic result of Section 6 is that an exchange rate fall is associated with movement of the U. S. trade balance toward surplus under conditions of foreign exchange market equilibrium. This result is represented graphically in Figure 3 as the foreign exchange market clearing (FX) curve, which shows all trade balance-exchange rate combinations consistent with equilibrium in the foreign exchange market. The downward slope of the curve indicates the inverse relationship between the trade balance and the exchange rate. For example, the curve says that a movement of the U. S. trade balance toward deficit must be accompanied by an exchange rate rise to maintain foreign exchange market equilibrium. This is so because, by analogy to the discussion in Section 5, the U. S. trade deficit implies a wealth transfer to U. K. residents which raises the demand for pounds relative to dollars in world portfolios. Given existing asset stocks, the increased demand for pounds relative to dollars in portfolio must be satisfied by an increase in the value of pound relative to dollar assets, i.e., an exchange rate rise. The intersection of these curves (at point A) represents the trade balance and exchange rate that are consistent with both goods and foreign exchange market equilibrium in the current period. The intersection of GM and FX curves may be called short run or period equilibrium. Short run equilibrium does not require a zero trade balance. But Figure 3 is drawn so that the trade balance is indeed zero. When this is the case, no wealth is being transferred between U. S. and U. K. residents since the value of U. S. exports (U. K. imports) equals the value of U. S. imports (U. K. exports). Consequently, the exchange rate need not move. In the absence of further disturbances, this short run equilibrium can sustain itself indefinitely and so it is called a steady state or full long run equilibrium. Because no wealth changes hands, world residents are implicitly satis22 ECONOMIC REVIEW, fied holding their stocks of money wealth. In this sense the long run equilibrium is also known as stock monetary equilibrium. Now suppose the GM and FX curves intersect at point B in Figure 4, so that the U. S. trade balance is in deficit. How do the exchange rate and trade: balance adjust to long run equilibrium? First, the short run equilibrium trade balance and exchange rate at point B cannot be self-sustaining because such a situation involves a transfer of money wealth from U. S. to U. K. residents. This wealth transfer causes the GM locus to shift up each period in which the U. S. trade balance is in deficit. To see this, remember that for a fixed exchange rate a transfer of money wealth from U. S. to U. K. residents raises U. K. resident assets relative to those of U. S. residents. This in turn raises U. K. national expenditure relative to income, and lowers U. S. national expenditure relative to income, thus tending to reduce the U. S. trade deficit gradually each period. Suppose the trade deficit is reduced by half each period. This is represented graphically by a vertical shift of the GM curve equal to half the distance between point B and the horizontal axis. U.S. Trade Balance Figure 4 By how much does the FX curve shift? To answer this question, suppose the exchange rate were to remain unchanged. Now ask what trade balance would be consistent with foreign exchange market equilibrium. Any trade imbalance and wealth transfer would create an incipient excess pound supply So if the exchange rate (or demand) for dollars. were to remain unchanged, the trade balance would have to be zero for the foreign exchange market to Therefore the FX curve remain in equilibrium. SEPTEMBER/OCTOBER 1979 shifts vertically each period by the full amount of the distance between point B and the horizontal axis. Putting the two shifts together shows that next period’s short run equilibrium is established at point C in Figure 4. It follows by similar reasoning that the sequence of trade balance-exchange rate short run equilibrium points during the adjustment process lies on a positively sloped path cutting through B and C. It also follows that the trade balance converges gradually to zero as the exchange rate converges gradually to its long run level. In Section 1, it was pointed out that feasibility of a free exchange rate policy requires that adjustment to a stable equilibrium be achieved automatically following disturbances. It is now possible to comment on this issue. The analysis outlined here suggests that, if the trade balance is not zero, wealth transfers associated with trade imbalance provide a force tending to bring the trade balance back to zero. That is, the system tends to return to long run stock monetary equilibrium after a disturbance. Therefore, the government can rely on the tendency of the free market adjustment mechanism to move toward a stable equilibrium. In other words, management of the exchange rate need not concern policy makers out of fear of instability of the free exchange rate. 8. Two Illustrations of the Adjustment Mechanism The graphical apparatus constructed in the previous section can be used to analyze two types of hypothetical disturbances to long run equilibrium characteristic of those that occur in international markets. The first of these originates in goods markets and, in terms of the model developed here, may be occasioned by, say, a shift of U. S. resident tastes from large domestic automobiles to smaller U. K. (foreign) cars. The second originates in foreign exchange or asset markets and involves, say, a shift of asset preferences from dollar assets to those denominated in pounds (foreign currencies). Disturbance l-An Expenditure Shift If U. S. resident preferences for goods shift toward U. K. output, an incipient excess supply of U. S. output and corresponding excess demand for U. K. output develops at the initial goods prices and exchange rate. At the initial exchange rate, U. K. prices rise and U. S. prices fall. This lowers U. S. income relative to expenditure and raises U. K. income relative to expenditure so that the goods markets come into equilibrium, with a U. S. trade deficit In other words, the GM (U. K. trade surplus). curve shifts down as shown in Figure 5. Since the disturbance does not initially affect the underlying determinants of the FX curve, that curve does not shift initially. Therefore immediately following the disturbance, the short run equilibrium point shifts from A to B, the exchange rate depreciates, and the U. S. trade balance moves into deficit. The movement down the initially fixed FX curve depicts a rise in the dollar value of pounds necessary to accommodate the increased relative demand for pound assets brought about by the wealth transfer from U. S. to U. K. residents through the U. S. trade deficit. With no further disturbances, the adjustment to long run equilibrium occurs along the positively sloped path between B and C. The adjustment path is positively sloped because throughout the adjustment process wealth is being transferred from U. S. to U. K. residents. Therefore the exchange rate must continually rise to accommodate the rising relative demand for pound assets. The adjustment process terminates at point C, when the trade balance comes back into equilibrium. At that time, the dollar will have depreciated relative to the pound, and wealth will have been transferred from U. S. to U. K. residents through the series of U. S. trade deficits. The wealth transfer means U. S. money expenditure will have fallen and U. K. money expenditure will have risen. Trade balance equilibrium means national expenditure equals national income, which further implies that U. S. money income will have fallen and U. K. money income will have risen in the new long run equilibrium. Note that the terms of trade moves in favor of U. K. residents between points A and B when demand initially shifts toward U. K. goods. Then, as equilibrium moves FEDERAL RESERVE BANK OF RICHMOND 23 from B to C and the relative share of U. K. residents in world expenditure rises, the terms of trade moves further in favor of U. K. residents. It should be emphasized that even though goods prices as well as the exchange rate have been assumed to be perfectly flexible, the adjustment to new long run equilibrium is not achieved by price movements This would be possible only if individuals alone. held no foreign assets. Then the FX curve would be horizontal at trade account equilibrium because an exchange rate rise would not put portfolios out of balance. In fact without diversification into foreign assets there would be no portfolio balance problem. In such a case, the new long run equilibrium would occur immediately on the shifted GM curve at trade account equilibrium. Since the trade account remains in equilibrium, no wealth transfer would take place. Therefore, national money expenditure and income would remain unchanged as would domestic currency goods prices. The exchange rate rise would be just sufficient to raise the terms of trade, i.e., the price of U. K. goods in terms of U. S. goods, to maintain goods market equilibrium. On the other hand, when individuals do hold foreign assets, even though the disturbance originates in the goods markets, it is transmitted to the asset markets through the revaluation effect of the exchange rate change. In this case, as the exchange rate increase raises the price of U. K. goods relative to U. S. goods, it also moves portfolios out of balance by raising the dollar price of pounds. Hence, a U. S. trade deficit is required to maintain foreign exchange market equilibrium. The deficit initiates wealth transfers that take time to run their course. In short, because of the dual role of the exchange rate (i.e., changing terms of trade and revaluing assets) the disturbance in the goods markets produces wealthtransfer effects which prolong the adjustment process even though goods prices as well as the exchange rate are assumed perfectly flexible. Disturbance 2 -Revised Anticipations and the Policy Dilemma As a second disturbance, suppose world residents receive information that the foreign exchange value of the dollar is likely to fall in the future. The anticipated future fall in the value of the dollar could be due to an anticipated future shift of world demand away from U. S. toward U. K. goods; or it could be due to a number of other possible disturbances to the equilibrium value of the dollar. But for the sake of the present discussion, the actual cause of the future fall in the equilibrium value of the 24 ECONOMIC REVIEW, dollar does not matter. What matters is that this future fall in the equilibrium value is anticipated.1 It must be emphasized that this type of disturbance, operating as it does through anticipations, can be a predominant source of short run exchange rate movements for the following reason. Individuals holding both pounds and dollars in portfolio of course wish to protect themselves against a future fall in the value of one currency relative to another. Consequently, resources are expended in gathering information on likely future disturbances to the foreign exchange rate. Because the market uses this information efficiently, it tends to react to anticipated future disturbances to the exchange rate before they take place. To the extent that disturbances can be anticipated, short run exchange rate movements are likely to be predominantly influenced by shifts in portfolio preference induced by anticipated changes in the future equilibrium foreign exchange rate. How does the market react to an anticipated fall in the foreign exchange value of the dollar? In order to avoid holding an asset which is likely to lose value, individuals try to sell dollars for pounds immediately. That is, individuals reduce the share of dollars in portfolio in advance of the anticipated fall in the value of the dollar. Therefore, at the initial exchange rate, an excess dollar demand for pounds develops on the foreign exchange market. At the initial trade balance, the excess dollar demand for pounds must be accommodated by an exchange rate rise. This follows because, for given asset supplies, the desire to hold a greater share of wealth in pound assets must be satisfied by an increase in the value of existing pound assets relative to dollar assets. Consequently the FX curve moves right with a shift in portfolio preference toward pounds, as shown in Figure 6. Now since the disturbance does not initially affect the underlying determinants of the GM curve, that curve does not shift. Therefore, when the disturbance occurs, the short run equilibrium point moves from A to B in Figure 6, i.e., the exchange rate rises and the U. S. trade balance moves into surplus. The movement along the initially fixed GM curve depicts the emergence of a U. S. trade surplus associated with an exchange rate rise. The reasons for this relation are discussed in detail in Section 3. In general terms, it is shown there that following an exchange rate rise, domestic currency goods price adjustments are required to restore consistency of 1 Goodfriend [2] contains ment to various anticipated SEPTEMBER/OCTOBER 1979 analyses of complete future disturbances. adjust- authorities insist on pegging the exchange rate, they must reduce the dollar money supply growth rate. U. S. monetary policy is thereby made subordinate to events in the foreign exchange market. The policy dilemma arises because the monetary policy necessary to stabilize the foreign exchange rate may not be consistent with domestic monetary policy goals. U.S. Trade Balance Figure The events of fall 1978 dramatically illustrate this policy dilemma. In the months prior to November, both U. S. money supply and the dollar price of foreign exchange had been rising at an increasingly rapid rate. Then after November 1, the U. S. monetary authorities, coming under intense pressure, announced a commitment to stabilize the foreign exchange rate. For the following five months the U. S. money supply grew much more slowly than before, thereby helping to precipitate the reduction in real economic growth that occurred in the second quarter of 1979. 6 the terms of trade with goods market equilibrium. These goods price movements are in turn shown to have disproportionate effects on national income and expenditure. Because of these income and expenditure effects, goods price movements that bring the goods markets into equilibrium also move the U. S. trade balance into surplus. V. This article has outlined important features of international adjustment under a flexible exchange rate. Exchange rate movements have been shown to function in two distinct roles in the adjustment process. In the goods markets, exchange rate movements have been shown to affect relative goods prices (the terms of trade) and national income relative to national expenditure. In the asset or foreign exchange market, exchange rate movements have been shown to function by changing the relative value of foreign and domestic assets. These two functions of exchange rate movements have been synthesized in a simple graphical framework which displays the simultaneous determination of the trade balance and exchange rate. The framework has been used to investigate some characteristics of the international adjustment process as a whole. Furthermore, the discussion in Section 3 shows that the exchange rate rise is associated with a rise in the relative price of U. K. output in terms of U. S. output units. In other words, the terms of trade move against U. S. residents and in favor of U. K. residents. It is also shown in Section 3 that this “speculatively” induced rise in the exchange rate tends to push U. S. goods prices up. These last two consequences of the induced rise in the exchange rate bear directly on the policy dilemma faced by- the world’s central banks operating in a flexible exchange rate environment. Because of the dual role played by the exchange rate in the adjustment process-i.e., relative goods price and asset revaluation roles-the exchange rate rise required to clear the foreign exchange market affects goods markets and acts on the GM curve as well. As mentioned above, from the U. S. resident point of view the depreciation of the dollar is accompanied by a worsening of both the terms of trade and domestic inflation. Consequently, the U. S. monetary authorities in this case may come under pressure to stabilize the exchange rate to prevent its depreciation relative to the pound. But stabilizing the dollar requires that the U. S. authorities passively supply pounds for dollars to satisfy the excess demand for pounds that exists at the pegged exchange rate. If the U. S. CONCLUSION The fundamental dynamic feature of the adjustment process has been shown to be the wealth transfer associated with trade account disequilibrium, which in general implies that the exchange rate and trade balance adjust gradually to a new long run equilibrium after a disturbance. Wealth transfer has also been identified as providing an automatic mechanism for exchange rate and trade balance adjustment to new long run equilibrium. The article concludes with a discussion of the effect of anticipated future disturbances on the current exchange rate, trade balance, and terms of trade. It has been argued that short run exchange rate movements are likely to be dominated by disturbances FEDERAL RESERVE BANK OF RICHMOND 25 Therefore, the operating through anticipations. short run policy dilemma faced by world central banks operating under a flexible exchange rate has been discussed in the context of this type of disturbance. The policy choice involves the extent to which exchange rate movements should be allowed to play a role in short run international adjustment. The dilemma arises because the monetary authorities must give up control of money growth in order to prevent exchange rate movements. If the exchange rate is pegged, then money growth may become inconsistent with domestic policy objectives. But if the exchange rate is allowed to function in the adjustment process, the consequent terms of trade change and goods price movements could prove equally inconsistent with domestic policy objectives such as price stability and steady real economic growth. Selected References 1. Dornbusch, Rudiger. “Currency Depreciation, Hoarding, and Relative Prices.” Journal of Political Economy 81, No. 4 (July/August 1973) : 893-915. Discusses the effect of a devaluation of the exchange rate (in a fixed rate context) on the terms of trade and trade balance. A useful reference for the relationships underlying the GM curve. 2. Goodfriend, Marvin. Money, the Terms of Trade and Balance of Payments Adjustment with a Flexible Exchange Rate: A Theoretical and Empirical Study of the International Adjustment Mechanism. Brown University Ph.D. thesis in progress. Contains development and thorough analysis of the complete model from which the framework employed in this article is derived. 3. Johnson, Harry G. Theory, Money, Balance of Payments and the International Monetary Problem. Essays in International Finance No. 124. Prince- 26 ECONOMIC REVIEW, ton: Princeton University Press, 1977. Contains a particularly good discussion of the implications of the monetary approach to the balance of payments theory for the stability of the adjustment process in the fixed rate context. Useful background for the discussion of the stabilizing effect of wealth transfer. Contains the argument that fixing the exchange rate implies loss of control over the money supply. Useful discussion of the exchange rate policy dilemma in the fixed rate context. 4. Kouri, Pentti J.K. “The Exchange Rate and the Balance of Payments in the Short Run and in the Long Run: A Monetary Approach. Scandinavian Journal of Economics 78, No. 2 (June 1976) : 280-304. A sample of the literature emphasizing the role of asset markets in exchange rate determination. Contains a useful analysis of the effect of wealth transfers on exchange rate movements. Useful background material for the relationships underlying the FX curve. SEPTEMBER/OCTOBER 1979