Anotable development in recent years has been the substantial success of most countries in reducing inflation, in many cases to the lowest levels in several decades (Chart 34). The decline in inflation is evident both across regions and among countries at different stages of development. It has been most dramatic in the developing countries and countries in transition, where inflation had risen to the highest levels. But the success of the industrial countries in reducing their average rate of inflation from double digits in the mid-1970s back to the low single-digit levels of the 1950s and 1960s is also clearly significant.

Anotable development in recent years has been the substantial success of most countries in reducing inflation, in many cases to the lowest levels in several decades (Chart 34). The decline in inflation is evident both across regions and among countries at different stages of development. It has been most dramatic in the developing countries and countries in transition, where inflation had risen to the highest levels. But the success of the industrial countries in reducing their average rate of inflation from double digits in the mid-1970s back to the low single-digit levels of the 1950s and 1960s is also clearly significant.

Chart 34.
Chart 34.


(Annual percent change)

Inflation has come down significantly in recent years.

From the rise and fall of inflation experienced over the past three decades, what are the principal lessons about inflation’s causes, its effects on economic activity and welfare, the costs of reducing it, and the policies for maintaining low inflation? And following the progress made toward the achievement of price stability in many countries, how far down should inflation rates be pushed? What constitutes price stability? Do institutional arrangements such as central bank independence and inflation targeting have roles to play in the maintenance of price stability? These are some of the issues explored in this chapter.

Sources of Inflation: Three Inflation Episodes in Industrial Countries

Sustained inflation is a relatively modern phenomenon. Until World War I, the international experience was one of long-run stability of prices: average prices essentially fluctuated around a stationary level, with periods of inflation—usually war-related—generally offset by periods of deflation (Chart 35). Following the turbulence of the interwar period, however, price levels have followed a clear upward trend, and inflation rates rather than price levels have tended to return to stationary values following episodes of increased inflation. The past quarter century has witnessed three major inflation peaks, 1973–74, around 1979–80, and again in 1989–90. Changes in the exchange rate regime, supply shocks, and macroeconomic policies have all played important roles. Because the events contributing to these increases in inflation are well-known, the following review of the three episodes concentrates on a few stylized facts and lessons.

Chart 35.
Chart 35.

Price Level in the United States and the United Kingdom

(1938 = 100; logarithmic scale)

Price levels have followed a steady upward trend in the post-World War II period.Sources: Robert P. Flood and Michael Mussa, “Issues Concerning Nominal Anchors for Monetary Policy” in Frameworks for Monetary Stability: Policy Issues and Country Experiences, ed. by Tomás J.T. Baliño and Carlo Cottarelli (Washington: IMF, 1994); U.S. Bureau of Labor Statistics; and IMF, International Financial Statistics.

The inflation peaks of 1973–74 and 1979–80 were both associated with steep increases in primary commodity prices, particularly the price of oil. But the most important lesson of these episodes is not that commodity shocks and other supply shocks can provide substantial impetus to inflation. More important lessons are that the inflation response to such shocks depends on policies and also on the behavior of wages.

The late 1960s witnessed a gradual increase in inflation in the industrial countries, partly as a result of expansionary fiscal and monetary policies. By 1970, average inflation in the industrial countries had drifted upward to the highest rates since the Korean War period of the early 1950s. Although a downturn in economic activity in 1970–71 dampened inflationary pressures in many countries, expansionary demand-management policies were already contributing to a strong cyclical upswing in the global economy when non-fuel primary commodity prices began to rise steeply in 1972. When the oil price shock hit in October 1973, many economies were around the peak of their expansions, with capacity constraints exerting clear upward pressure on wages and prices. The quadrupling of oil prices was followed by sharp inflationary spikes in the industrial countries along with severe contractions in economic activity (Chart 36).

Chart 36.
Chart 36.

Industrial Countries: Inflation and Real Output Growth

(Annual percent change)

Supply shocks were associated with sharp increases in inflation and contractions in economic activity.

In the United States, inflation had already risen from 4 percent at the beginning of 1973 to over 7 percent by the time of the oil shock. Following the oil shock, which coincided with the end of wage and price controls that had been imposed in 1971, inflation rose to above 12 percent at the end of 1974. Facing both rising prices and unemployment, the Federal Reserve initially adopted an accommodative stance. Though output and employment rebounded and monetary policy was eventually tightened, inflation remained above 5 percent, about double the rate that had prevailed during the previous two decades.

In Europe, inflation also surged following the oil shock but was better controlled in Germany and Switzerland. The Bundesbank did not accommodate the price shock and continued the restrictive monetary policy begun at the end of 1972 when inflation had risen to over 6 percent. As a result, inflation never surpassed the preshock peak and fell below 4 percent by the end of 1976. Switzerland similarly did not allow the high inflation of the early 1970s to continue and pursued a restrictive monetary policy from the beginning of 1973; inflation fell below 3 percent by the beginning of 1976.

In contrast, the inflation aftermath of the 1973 shock was severe in the other European countries, as governments generally sought to offset its adverse output and employment effects with accommodative monetary policy. Inflation in France reached 15 percent by the end of 1974 and even by the next oil shock at the end of 1978 had not fallen below 9 percent. In the United Kingdom and Italy, it surged to around 25 percent at the end of 1974, partly reflecting indexation formulas under wage policies, and remained above 10 percent until the next shock.

Japan was hit particularly hard by the 1973 oil shock, with consumer price inflation rising to a postwar high of 23 percent in 1974, despite monetary tightening that had begun in mid-1973 in response to a booming economy. This jump in inflation in part reflected the dependence of the Japanese economy on imported oil, but was also the result of preexisting inflationary pressures that had driven inflation above 10 percent in the middle of 1973. The Bank of Japan maintained a firm anti-inflationary policy following the shock, and consequently inflation abated sharply in subsequent years. By 1978, inflation in Japan had fallen below 4 percent.

The initial accommodative response to the oil shock in the United States and most European countries stood in sharp contrast to the stance of monetary policy in Germany and Switzerland, and so did the consequences in terms of inflation (Chart 37). These differences in responses to the first oil shock had important implications for the inflationary impact of the second oil price shock that hit in 1978–79.71 Countries in which monetary authorities had failed adequately to counter the inflationary pressures of the first half of the 1970s were left not only with higher inflation rates but also with a ratcheting up of expectations about future inflation that made them more vulnerable to the impact of the second oil shock than countries where the containment of inflation had been given high priority. Higher expected inflation affected wage bargaining and other contractual arrangements, with increased indexation helping to lock in higher rates of inflation.

Chart 37.
Chart 37.

Selected Industrial Countries: Inflation

(Annual percent change)

The inflation performance of Germany and Switzerland differed significantly from that of other countries following the 1973 oil price shock.

In the United States, inflationary pressures had again begun to build up prior to the second oil shock, with inflation reaching 7 percent by the middle of 1978. At the end of 1979, the inflation rate in the United States was in the double digits, with increases in inflation having outpaced increases in official interest rates. In fact, the United States was the only major industrial country in which the inflation peak in 1980 was above the peak in 1974. Faced with a sharp (but brief) recession, monetary policy was again eased in the summer of 1980, but with inflation rising toward a new postwar high by year-end, the Federal Reserve soon reversed course and embarked on a two-year span of maintaining continuously tight monetary conditions. Inflation finally broke in 1982, falling below 4 percent at the beginning of 1983 for the first time since 1972. The cost of the delayed efforts to control inflation, however, was a deep recession, in which real GDP fell by 3¼ percent from peak to trough and unemployment peaked at nearly 11 percent. Recessions also proved necessary to subdue inflation in France, Italy, the United Kingdom, Canada, and other industrial countries that similarly had not succeeded in wringing out the inflationary impulse that had begun in the early 1970s. Having brought inflation back down by the end of 1976, Germany and Switzerland did not face the same magnitude of inflationary pressures from the second oil shock. They did not escape the global downturn, but the consequences were less severe than in other countries where more dramatic contractionary measures were necessary to contain inflation. Against the experience of the surge in inflation that resulted from its hesitation in 1973, the Bank of Japan responded strongly in 1978.

The behavior of wages was also crucial to the outcomes observed. In Japan, a high degree of real wage flexibility contributed to the relative stability of employment and output in the wake of the adverse supply shocks. In fact, real wage growth decelerated more in Japan following the oil shocks than in the other major countries. In the United States, real wage growth also declined, but the degree of real wage flexibility in the short term was insufficient, relative to the deterioration in the terms of trade and the slowdown in productivity growth, to prevent a decline in employment. In Europe, real wage rigidities, in part owing to the centralization of wage bargaining and indexation of wages, had the most adverse consequences for employment and output growth.72

Two policy lessons emerge from this experience with supply shocks. First, it is important to limit the monetary accommodation of adverse supply shocks, since the inflation that is permitted tends to get built into inflation expectations, to become persistent, and to raise the cost of subsequent disinflation. Second, flexibility in labor markets—the ability of both wages and the structure of employment to adjust in response to shocks—plays a crucial role in determining the output and inflation costs of such shocks.

Faced with unacceptable rates of inflation, the industrial countries realized that bringing inflation down was crucial for achieving broader economic goals, and starting in the early 1980s strong and widespread efforts were made to restore reasonable price stability. These efforts at inflation control formed part of a general reorientation of economic policies toward medium-term objectives. The main elements of this “medium-term strategy” were monetary policies focused on medium-term price stability, fiscal consolidation, and structural reforms of labor and productmarkets.73

This strategy, although its elements were applied to different degrees in different countries, was broadly successful in gradually reducing inflation in the industrial countries to around 3 percent by 1985 (see Chart 36). This reduction of inflation, facilitated by a significant decline in oil prices, set the stage for a robust economic recovery that continued until the end of the decade. By that time, signs of overheating (especially in asset markets, as discussed below) were signaling that monetary policies had become more expansionary than intended.74 Among other developments, the unification of Germany resulted in a substantial procyclical fiscal stimulus, while the temporary increase in oil prices in 1990 also contributed to the upturn in inflation—to an average of 5 percent in the industrial countries in that year.

In response to this third episode of increased inflationary pressures in less than two decades, monetary authorities in industrial countries reacted forcefully to resist further increases in inflation. On this occasion, the United States started tightening monetary policy relatively early, which allowed the economy to cool off by 1991 at the cost of a much shallower recession than in the early 1980s. By 1992, the U.S. economy had embarked on a moderate but solid economic expansion while inflation fell below 3 percent. The relatively modest cost of the disinflation in this episode points to a reshaping of inflation expectations and to the credibility of the commitment to reasonable price stability that had been gained. German monetary policy also responded forcefully to the building of inflationary pressures, through a progressive tightening of monetary conditions. However, although inflation was also contained relatively quickly, the output costs proved significantly larger than in the United States. Inflation was even better controlled in many other European countries, as Germany’s tight monetary conditions were reinforced in the countries participating in the ERM by the need to resist pressures in the foreign exchange markets. For many European countries, the output costs in this disinflation episode were larger than in the two previous episodes, but inflation came down to rates not seen in 30 years. In the United Kingdom and Sweden, the economic cycle was more pronounced than in most other European countries because of falling prices of real estate and other assets from inflated levels.

In Japan, the inflation experience of the late 1980s was dominated by an asset price bubble that eventually collapsed when monetary policy was tightened in 1990. While consumer price inflation exceeded 3 percent in 1991, consumer prices by 1995 were actually falling, under the influence of a strong currency and a substantial output gap. The costs of this episode proved particularly large because of the serious consequences of asset-price deflation, especially in the banking system.

Although monetary authorities generally did not allow inflation to rise nearly as much in the third upturn as in the two earlier episodes, the failure in some countries to react sufficiently early to signs of overheating in the late 1980s still resulted in pronounced cyclical fluctuations. This third episode demonstrates the importance of taking action to preempt inflationary pressures before standard price indices signal a rise in inflation.

Role of the Exchange Rate Regime

The divergent inflation experiences of the 1970s and early 1980s would not have been possible under a system of fixed exchange rates. The advent of floating exchange rates among the major currencies in 1973 gave countries the monetary policy freedom to respond to adverse supply shocks with different degrees of monetary accommodation. But as the countries that were not reticent to exploit this freedom came to realize, their attempts to maintain employment tended to propagate the initial inflationary impulse and resulted in persistent inflation. This was particularly true in economies with strong wage and price inertia. One of the lessons of the 1970s, therefore, is that exercise of the monetary policy autonomy made possible by flexible exchange rates, unless sufficiently restrained, can lead to persistent inflation. This is demonstrated further by the tendency for average inflation in the industrial countries, as well as national deviations from the industrial country average, to be more persistent in the post-Bretton Woods period.75 By contrast, the nominal anchor of the gold-dollar standard of the Bretton Woods system acted as a global constraint on accommodative monetary policies as long as the United States maintained low inflation, while each other country’s commitment to maintain the exchange value of its currency in terms of the dollar limited divergences among monetary policies and inflation rates. Indeed, the system broke down in large part because the anchor country of the system—the United States—pursued policies that were at odds with the desires of other countries, notably Germany, for low inflation.

Even though the greater freedom that floating exchange rates grant to pursue independent monetary policies can be associated with higher inflation, as shown by the experience of the 1970s, the disinflation since the early 1980s shows that floating exchange rates are by no means incompatible with the achievement of reasonable price stability. Whether exchange rates are floating or fixed, it is the commitment of authorities to disciplined and responsible fiscal and monetary policies that is necessary for the control of inflation.

This is supported by a recent study of inflation in a large group of countries, both industrial and developing, over the period 1960 to 1990.76 The study does find that countries with some form of fixed exchange rate experienced generally lower and less variable inflation than countries with flexible exchange rate arrangements (Table 34). The superior inflation performance, however, was found mainly in countries that adjusted their exchange rate peg infrequently or not at all. Countries that frequently changed their exchange rate parity did not derive anti-inflationary benefits from the peg: indeed, they experienced higher and more volatile inflation rates than countries with freely floating exchange rates. Viewed together with the recent decline in inflation in many flexible exchange rate countries to their lowest levels since the late 1950s, this evidence supports the argument that simply pegging the exchange rate does not necessarily deliver low inflation. Rather, it is the pursuit of appropriate macroeconomic policies that is important, whether in enabling an exchange rate peg to be maintained, or, in the case of flexible exchange rates, in ensuring adherence to a domestic nominal anchor.

Table 34.

Exchange Rate Regime and Inflation

article image
Source: Atish R. Ghosh, Anne-Marie Guide, Jonathan D. Ostry, and Holger C. Wolf, “Does the Nominal Exchange Rate Regime Matter?” IMF Working Paper 95/121 (November 1995). Classification of 136 countries over the period 1980–90 according to the stated intention of central banks regarding their intervention policy, as summarized in the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions.

Increased international integration of goods and capital markets in recent decades has had significant implications for domestic policies and exchange rates. Increased integration has tended to enhance the transmission of economic shocks and policies across national boundaries. With greater capital mobility, interest rates, and thus output and inflation, in each country are increasingly affected by global developments in the supply and demand for capital. This is particularly important for countries that peg their exchange rates, since their monetary policies have to be directed toward resisting exchange rate pressures arising from external (as well as internal) developments. Even with floating exchange rates, however, developments in large countries have global effects on inflation through changes in world interest rates, output, and the prices of traded goods.

The increased openness of countries to international trade and the widespread adoption of flexible exchange rate arrangements have also tended to enhance the importance of the exchange rate channel of the monetary transmission mechanism, potentially amplifying the overall impact of monetary policy on economic activity (see Chapter III). Moreover, because of their effects on exchange rates, monetary policies are likely to have a more rapid and more direct impact on consumer prices. The discipline imposed on domestic financial policies by increased capital market integration has undoubtedly contributed to the declines in inflation experienced in recent years.

Guideposts for Anti-Inflationary Monetary Policy

Having lost the nominal anchor of fixed exchange rates, and with inflation at postwar highs, central banks in many industrial countries starting in the mid-1970s adopted monetary targets to guide the conduct of monetary policy. Soon, however, problems with this framework began to surface. In some countries, the abandonment of quantitative restrictions on bank credit led to reintermediation into the banking system and upward shifts in some targeted monetary aggregates. In other cases, innovations in the financial services industry, in part sparked by high rates of inflation, led to downward shifts in the demand for money, so that a given rate of monetary growth would be associated with more expansionary pressure than might have been expected judging from the historical relationship between money and income growth. This played an important role in the inflationary process of the 1970s in a number of countries, as central banks at times found themselves tightening policy but not catching up with the increasing rate of inflation. Such instability of money demand, especially in the context of supply shocks and declines in potential output growth, complicated the task of monetary authorities. As a result, during the 1980s most central banks—with some notable exceptions—either abandoned or downplayed the role of monetary targets.

A related factor that complicated the task of monetary policy in the 1980s in a number of industrial countries and contributed to the third of the inflation peaks referred to earlier was the rapid expansion of credit in the wake of financial market deregulation and liberalization.77 Much of this credit growth expanded the demand for assets such as corporate stocks and real estate, leading to a run-up in asset prices. This inflation in asset markets was for the most part not directly reflected in conventional measures of inflation relating to the flows of production and consumption of goods and services. But it contributed to the expansion of aggregate demand and subsequent inflationary pressures in goods and factor markets. The excessive growth of credit and the associated rise in the private sector’s financial leverage meant that by the second half of the 1980s, households and businesses were quite vulnerable to the effects of higher interest rates. Hence, as monetary policies were tightened in the late 1980s, the resulting balance sheet adjustments led to asset-price deflation in a number of countries, most notably Japan and the United Kingdom, but also in Australia, Finland, France, New Zealand, Norway, Sweden, and the United States. One consequence of this asset-price deflation was an increase in nonperforming loans, which was reflected in financial sector distress in the United States, Japan, France, and the Nordic countries. Another was prolonged economic slowdowns in the countries most affected.

The lesson that emerges from the experience with monetary targeting is that monetary aggregates cannot be relied upon to provide intermediate targets sufficient in themselves to guide monetary policy. So what should be the guideposts for monetary policy? For a country that pegs its exchange rate to a major currency, the monetary policy framework is greatly simplified: all of the information necessary to guide monetary policy is summarized in one variable.78 For countries that forgo a pegged exchange rate, and in particular for the three major reserve currency countries, whose currencies are freely floating, the monetary policy framework is more complex and necessarily more pragmatic. The significance of any particular indicator can vary with circumstances, including the conduct of policy and the development of financial markets. This has become particularly apparent from the breakdown of relationships between monetary aggregates and prices and real activity in the early 1980s, as well as from the experience of the latter half of the 1980s, when the formulation of monetary policy paid inadequate attention to asset-price developments.

In practice, monetary authorities have therefore found it useful to monitor a range of indicators, including monetary and credit aggregates, that have potential predictive content for the ultimate goal of policy. A variety of indicators is needed because no single indicator can adequately summarize the stance of monetary policy and because each indicator is subject to influences other than monetary policy; the chances of correctly identifying the nature of economic disturbances are therefore greater by observing several indicators. A disadvantage of this approach is that the monetary authorities do not reap the potential credibility benefits that the commitment to a single variable can bestow by serving as a focal point for expectations. It is in part to overcome this drawback that a number of central banks in recent years have adopted formal inflation targets. Not all countries, however, have the technical means to carry out and to develop the necessary credibility for such a forward-looking policy framework. The use of inflation targets is discussed further in Box 8.

Special Features of Inflation in Developing and Transition Countries

The inflationary process in many developing and transition countries differs from that in the industrial countries because of the larger role of fiscal deficits, which in turn partly reflects the limited development of domestic financial markets. Additional, albeit temporary, complications in many transition countries have been the role of price liberalization and the existence of monetary overhangs.

Fiscal pressures have contributed greatly to inflationary surges and to chronic inflation, as many countries have monetized large budgetary deficits. In many developing countries, fiscal pressures intensified in the 1980s because of the debt crisis. Unlike in the industrial countries, inflation rates in the developing countries as a group did not fall back to trend levels following the successive supply shocks, but instead drifted upward throughout the 1980s and early 1990s (Chart 38). Of course, these average (in Chart 38, median) inflation rates combine together countries with quite varied inflationary experiences, ranging from the extreme inflations in countries such as Argentina, Bolivia, and Brazil to the more moderate or even low inflation rates in many Asian countries such as Thailand, Korea, Malaysia, and Singapore. In recent years, widespread and substantial progress has been made in lowering inflation through determined stabilization efforts. In many countries, however, inflation rates remain in the double digits and continue to impose substantial social costs. Moreover, countries with inflation rates in the 20–30 percent range, or higher, may face the risk that inflation will ratchet up to even more damaging levels (see below).

Chart 38.
Chart 38.

Developing Countries: Inflation Across Regions

(Median, percent change)

Inflation has fallen recently in most regions.

Part of the reason for the persistence of inflation at double-digit rates is governments’ reliance on seigniorage for a substantial portion of their revenue—nearly 25 percent in high-inflation countries (Table 35). The underdeveloped nature of tax collection systems and institutions in many developing countries provides a powerful incentive for money creation, since the inflation tax can be levied without the need for a revenue collection infrastructure. In such countries, fundamental fiscal reforms are often necessary to prevent disinflation from leading to a deterioration in budgetary performance. On the other hand, at extremely high rates of money growth—where seigniorage equals around 10 percent of GDP79—inflation can lead to demonetization of the economy, so that transactions are made either by barter or with a foreign currency such as the U.S. dollar. In this case, seigniorage revenue can fall in real terms, as the tax base (i.e., the demand for money) shrinks by more than the increase in the tax rate (i.e., the rate of inflation). Moreover, high inflation can erode the real value of tax revenues.80

Table 35.

Importance of Seigniorage in Importance of Seigniorage in Developing Countries

(Change in monetary base as a fraction of GDP and government revenue)

article image
Note: The figures cover 71 countries.

Inflation Targets

In recent years, there has been a growing trend toward the adoption of inflation targets as the primary focus for the conduct of monetary policy.1 Countries that have adopted official inflation targets in the 1990s include (in chronological order): New Zealand, Canada, the United Kingdom, Sweden, Finland, Australia, and Spain (see table on page 110 for details). By focusing attention on the ultimate goal of monetary policy—that is, reasonable price stability, which is generally understood to mean inflation at rates low enough not to affect economic decisions—inflation targets may be argued to provide a more transparent framework for policy than alternative frameworks based on monetary targets or pegged exchange rates. Indeed, inflation targets were typically adopted after unsuccessful experiences with either monetary aggregates targeting or pegged exchange rates, or both. Inflation targeting may not be the most appropriate monetary policy framework for all countries, however. It has been most useful in cases where policymakers have needed to establish a credible commitment to low inflation policies. A number of countries that have been relatively successful at achieving and maintaining low inflation, including the United States, Japan, and Germany, have continued to eschew formal inflation targets.

Explicit inflation targets play two main roles in efforts to reduce and control inflation: first, by communicating to the public the objective that monetary policy seeks to accomplish, they serve as a coordination device in wage-and price-setting processes and in forming the public’s inflation expectations; and second, they provide a transparent guide to the conduct of monetary policy, whose commitment and credibility can then be assessed on the basis of whether policy actions are taken to ensure that the targets are achieved. Accordingly, in those countries that have adopted them, inflation targets have become a key nominal anchor. Usually, inflation targets have not been supplemented by explicit intermediate targets for other nominal variables, such as monetary aggregates, though in principle it would not be inconsistent to do so.2

The shift toward inflation targets finds support from various developments in theoretical and empirical research. One of these is the emphasis on reasonable price stability as the primary goal of monetary policy. This follows mainly from research indicating that in the long run monetary policy affects only on nominal variables, such as the price level, and has no marked impact on real economic activity. A second development is the insight that any attempt to exploit the existence of short-run nominal rigidities to lower the rate of unemployment tends to impart an inflationary bias to monetary policy. A credible commitment to an inflation target can attenuate the extent of such inflationary bias, but possibly at the expense of raising the variability of output.3 A third development is research indicating that transparency and openness of the operating framework of monetary policy can bring important credibility gains, and that institutional arrangements such as greater independence of the central bank can be helpful in this regard. Hence, it is perhaps not surprising that in many countries the adoption of official inflation targets has been accompanied by moves toward central bank independence and accountability. In addition to developments in economic research, practical considerations, such as failures with monetary targeting and pegged exchange rates, have also played an important role in the adoption of inflation targets.

Once an explicit inflation target is adopted, the authorities need to specify the price index to be used in calculating the inflation rate, the target inflation rate or range, the horizon over which the target applies, and the situations under which the target may be modified or even disregarded. In practice, countries have set targets for the rate of inflation rather than the price level, which implies that overshooting the inflation rate target need not be followed by undershooting, as would be the case if the objective were absolute stability of the price level. As discussed in the main body of the chapter, a variety of theoretical and statistical arguments have been advanced in favor of targeting a low, but non-zero, inflation rate. Most inflation targets have been specified as inflation bands with widths of up to 3 percentage points. While a single target rate might most effectively serve as a focal point for inflation expectations, a relatively narrow band may be more credible in view of the authorities’ necessarily imperfect control over inflation. As to the appropriate price index, some countries target an inflation index that excludes particularly volatile items such as food and energy prices, indirect taxes and subsidies, and interest cost components; others directly target the broadly defined consumer price index.

There are important differences across countries in the legal and institutional support for inflation targets. In New Zealand, the targets were established as part of a thorough institutional reform, including a legislated goal of price stability and an independent and accountable central bank. In some other cases, targets have been established by the central bank without an explicit commitment by the government to achieve them. In addition, accountability by the monetary authorities for their actions can vary considerably. For example, in the United Kingdom, although the Bank of England was given greater operational independence in determining the timing of monetary policy actions following the introduction of inflation targets in 1992, the Chancellor of the Exchequer (who is accountable to Parliament) retains control of monetary policy; but many other central banks are directly accountable to the legislature.

Operationally, since monetary policy affects economic activity and inflation with long lags, and since knowledge of the monetary transmission mechanism is imperfect, policy must be forward looking and policymakers rely on multiple indicators to evaluate the economy’s inflation outlook. These indicators include monetary aggregates, the yield curve, movements in asset prices, market- or survey-based expectations of inflation, forecasts of the output gap, indicators of the fiscal stance, and exchange rate forecasts. Based on their evaluation of the inflation outlook, the authorities can choose a time path for the monetary policy instruments that result in a conditional probability distribution of future inflation that is consistent with achieving the inflation target. In fact, instead of setting policy instruments on the basis of point estimates of future inflation, a more fruitful procedure has been to assess the risks and uncertainties associated with the inflation outlook and to set policy instruments according to probable inflation outcomes.

Since the decision-making process of monetary policy involves assessing current and future inflationary trends and comparing these trends with the announced inflation target, the process leaves room for discretion, the use of which in turn will affect the credibility of policy. These considerations point to the importance of the transparency and openness of the conduct of policy in relation to the operating framework. Thus, some aspects of the monetary policy process in the United Kingdom since 1992, such as the Bank of England’s Inflation Report with its carefully argued inflation forecast, and the publication of the minutes of meetings between the Governor and the Chancellor, have increased the transparency of the conduct of policy and increased the accountability of the treasury and the bank to the public.

In most countries, performance with inflation targets thus far has been satisfactory in some important respects. Despite the less-than-full credibility of the targets, which may be inferred from ex ante discrepancies between measures of the public’s inflation expectations and the targets, inflation objectives have in most cases been achieved or surpassed. While this performance may seem impressive, a more complete judgment awaits the passage of at least a full business cycle. In most countries, the targets were introduced in periods of considerable economic slack, making it difficult to judge their performance since countries without explicit targets have also experienced low inflation in recent years. The advent of greater inflationary pressures in the future may show more decisively the effectiveness of inflation targeting relative to other monetary policy frameworks. These pressures—if and when they arise—may give rise to divisions about the goals of monetary policy, which in turn may constitute a substantial test of the credibility and functioning of inflation targets. Last, it should be emphasized that formal inflation targets are only one aspect of macroeconomic policy. As such they are likely to funtion best when fiscal policy and wage behavior are compatible with the inflation targets, and when there are supportive institutional arrangements concerning the status and policies of the central bank.

Inflation Targets in Selected Countries

article image

For detailed analysis of inflation targets, including description of individual country cases, see the various contributions in Leonardo Leiderman and Lars E.O. Svensson, eds., Inflation Targets (London: Centre for Economic Policy Research, 1995); and G. Andrew Haldane, ed.,Targeting Inflation (London: Bank of England, 1995).


When the velocity of money is fully predictable, targeting monetary aggregates can be considered as a special case of inflation targeting, see Mervyn King, “Do Inflation Targets Work?”Quarterly Balletin. Bank of England, Vol. 35 (November 1995). pp. 392–94.


See John H. Green, “Inflation Targeting: Theory and Policy Implications.” IMF Working Paper 96/65 (June 1996).

The differences in average inflation rates among developing country regions in part reflect differences in fiscal performance. In particular, developing countries in Asia generally have experienced lower inflation rates than those in Latin America largely because of their greater fiscal discipline. Differences in fiscal discipline are not the only reason, however: the ability to adjust rapidly fiscal positions to external shocks is also a contributing factor. In this respect, the relatively more equal distribution of income in Asia compared with Latin America appears to make it politically easier in Asia to adjust fiscal policies rapidly.81

A further explanation for chronically high inflation is that institutions have developed that lessen the marginal costs of increases in inflation.82 In particular, the pervasive indexation of wages and prices in many countries partially insulates real activity from inflationary effects. Indexation, however, can also increase the ultimate cost of chronic inflation, since a longer lasting and deeper recession will be needed to lower inflation expectations that have become entrenched. A crucial policy lesson, therefore, is that countries should increase incentives for disinflation by modifying institutional features that reduce the costs of continued inflation and thus present obstacles to reform. In some countries, the development of financial markets and of indirect instruments of monetary control would help to insulate monetary policy from pressure to finance the fiscal deficit and would enhance the effectiveness of monetary policy.83

Since the beginning of the period of transformation in 1989, annual inflation has at some point surged to triple-digit levels in nearly all the transition countries. A large part of the initial surge can be attributed to a one-time increase in the price level, which was necessary to absorb the monetary overhang immediately following the liberalization of prices. In countries that undertook stabilization programs early in the transition, such as Hungary, Poland, and the Czech Republic, inflation quickly stabilized following the adoption of firm stabilization programs, with falls in output being halted within two years. This pattern appears to be under way also in other transition countries that have more recently embarked on the stabilization process.84

A factor that has contributed to the persistence of high inflation in many transition countries has been the continued need for relative price adjustments following the demise of central planning.85 The process of convergence to free market prices has entailed dramatic increases in the prices of goods and services, such as fuel, food, housing, medicine, and health care, all of which were formerly heavily subsidized. The products whose prices have increased sharply account for a substantial portion of the consumer price index, and for the most part these price increases have not been offset by decreases in prices of other goods. As a result, these price increases have translated into concomitantly large increases in the average price level. Moreover, the process by which prices have adjusted has been prolonged in many transition countries, leading to continuing major adjustments of relative prices and associated increases in the overall price level. As the process of liberalization proceeds and the structure of prices in the transition countries converges toward that in the rest of the world, this additional source of upward pressure on prices would be expected to subside eventually.

Stopping High Inflation

A salient feature of the worldwide disinflation of the past ten to fifteen years is that it includes major episodes of stopping high inflation. These episodes are especially interesting in view of the sharp contrast between some of the observed outcomes and what had been expected, based on conventional wisdom about the output cost of stopping inflation. Instead of resulting in deep recession and high unemployment, most recent stabilizations from high inflation were associated with a sizable expansion of output and economic activity. The analysis and interpretation of these episodes can contribute to a deeper understanding of the inflation process and the policy measures that are effective in stopping high inflation.86

Table 36 provides evidence from the various episodes of inflation stabilization during the 1980s and 1990s, defined here as a movement from a situation in which annual rates of inflation are above 40 percent for two years or more, to one in which annual rates of inflation are below 40 percent for two years or more.87 Thus, instead of dating and defining stabilizations on the basis of official announcements, a results-based definition is adopted that is applied broadly to all countries. The table shows the rate of inflation in the year prior to the disinflation, the rate of inflation in the third year of the stabilization, and the rate of inflation in 1995.

Table 36.

Inflation and Output Growth Before and After Stabilization

(In percent a year)

article image
Sources: Based on W. Easterly, “When Is Stabilization Expansionary?” (unpublished, World Bank, December 1995): and IMF, International Financial Statistics. Inflation rates are December-over- December percent changes.

The figure is for June 1995/June 1994.

Output growth two years after disinflation.

The episodes considered include cases in which annual inflation rates were at extremely high levels of above 4,000 percent a year before stabilization, as well as cases in which yearly inflation was below 500 percent. Previous work has identified important differences between hyperinflation and chronically high inflation—differences that have vital policy implications for attempts to stop both types of inflation. Hyperinflations are characterized by the eruption of very high and rapidly accelerating rates of inflation, with inflation typically reaching monthly rates of 50 percent or more until the hyperinflation comes to an end. In these circumstances, nominal contracts tend to disappear and almost all prices are set and transactions effected in a foreign currency. Historically, hyperinflations occurred in some European countries in the aftermath of the two World Wars, when war- and postwar-related hikes in government spending induced large budget deficits that were primarily financed by monetary expansion.88 In the sample in Table 36, there are four countries that underwent hyperinflation in the 1980s or early 1990s: Argentina, Bolivia, Nicaragua, and Peru.

In contrast to hyperinflations—which are shortlived—chronically high inflation may endure for years or even decades. Examples of high inflation include Colombia, Venezuela, Israel, and Turkey, all of which exhibited high double-digit, or even triple-digit, annual rates of inflation for a decade or more. Usually, chronically high inflation is associated with the existence of various indexation mechanisms for wages and nominal asset returns, and with substantially accommodative monetary and exchange rate policy, which contribute to the self-perpetuating and persistent nature of the process. They also reduce the costs of high inflation and thereby weaken the incentives for disinflation. High government budget deficits generally contribute to the development of chronically high inflation. Yet, theory and experience suggest that there are complex and dynamic relations between budget deficits and the rate of inflation, which may be reflected in apparently weak statistical links between these variables.

The differences between the two types of inflation processes have definite implications for stabilization policy. Historical experience indicates that policies to stop hyperinflation commonly include strong fiscal adjustment, drastic fiscal reforms, and the establishment of convertibility of the domestic currency in terms of gold or foreign currencies. Although in some cases convertibility of the domestic currency was set in conjunction with the adoption of a fixed exchange rate, in others it coexisted with a floating exchange rate. In the context of the chaotic situation at the peak of hyperinflations, the introduction of such fundamental fiscal and monetary reforms is likely to be perceived as highly credible, which will in turn contribute to the success of the policy. As regards the observed results, the two main stylized facts that emerge from experiences of stopping hyperinflation are, first, that there is an immediate and abrupt reduction in the rate of inflation, with price stability virtually achieved in a matter of a few months; and second, that low inflation is attained without deep recessions and sizable output costs. As a matter of fact, many episodes have exhibited negative output growth during the hyperinflation and an output recovery when the hyperinflation was stopped.

The highly persistent nature of chronically high inflation—and the fact that most countries have developed mechanisms (such as indexation and accommodation) to make living with it easier—have created unique difficulties for stopping it. In particular, disinflation by way of conventional fiscal and monetary contractions alone may be problematic in the short run because these policies alone cannot break the inertial elements of chronic inflation. Additional difficulties often arise owing to the failure of previous attempts to disinflate, which makes any new anti-inflation policies less and less credible. To overcome these difficulties, sharp adjustments in fiscal and monetary policies are typically supplemented by the adoption of nominal anchors aimed at dealing directly with the inertial elements of chronic inflation. Such “heterodox” stabilizations have usually been “exchange rate based.” with a fixed exchange rate serving as a nominal anchor, but with temporary wage and price norms typically performing a complementary function.

Among countries with chronically high inflation, recent well-known cases of exchange-rate-based disinflation are Israel (1985), Argentina 1985, 1991), Brazil (1986, 1994), and Mexico (1987). Conventional money-based stabilization policies—in which the money supply is the nominal anchor—have been implemented only in a few cases: to stop hyperinflation in Peru (1990) and, for a limited time, in Argentina and Brazil in 1990.

Most exchange-rate-based stabilizations have exhibited the following features:

(1) A marked but relatively slow reduction in the rate of inflation. Typically, it has been easier to disinflate from annual triple-digit rates to rates of between 10 and 20 percent a year than to reach the territory of single-digit annual rates of inflation. Accordingly, inflation rates in most previously chronically high inflation countries are still higher than those in industrial countries.

(2) A real appreciation of the domestic currency. To a large extent, this reflects the coexistence of a relatively stable nominal exchange rate and a positive inflation differential against trading partners.

(3) A deterioration in the trade and current accounts of the balance of payments.

(4) An expansion in the level of economic activity in the first few years following disinflation. In fact, economic growth is often weaker before than after disinflation (see Table 36). The finding that output growth is generally negative both in the year of peak inflation and right before disinflation begins supports the notion that stabilizations often are implemented only when the output losses from inflation become sizable and when some groups in society are willing to bear the fiscal costs of stopping high inflation.

Features (1) and (2) above have also characterized the less frequently used money-based programs to stop chronic inflation. But features (3) and (4) have not been as characteristic of money-based stabilizations. Indeed, some money-based disinflations have been associated with a marked slowdown in economic activity and with an improvement in the trade and current accounts of the balance of payments. The output response of exchange-rate-based and money-based stabilizations has not been definitively established, however.89

Interestingly, the broad patterns and results of reducing chronically high inflation discussed above apply also to the stabilization experiences of transition economies in central and eastern Europe in 1990–94.90 Most of these stabilization programs relied on at least two nominal anchors: the nominal exchange rate and wage norms. When coupled with fundamental (even if gradual) fiscal reforms, most of these programs brought satisfactory results, such as in Croatia, the Czech Republic, Estonia, Hungary, Poland, and the Slovak Republic. More conventional money-based stabilization programs were employed in fewer countries, and satisfactory results obtained only in some of these cases, notably, Albania, Latvia, and Slovenia. The success over the past year or so of money-based stabilization programs in most countries of the former Soviet Union, several of which informally fixed their exchange rates, however, suggests that it is the strength of and commitment to tight financial policies that matter most.

Despite the successful implementation of measures to stop high inflation in various countries in recent years, most of the countries concerned have not succeeded in eradicating double-digit inflation rates (see Table 36). As discussed below, one reason may be that it is hard to avoid some short-run output costs in bringing inflation down significantly farther. Another reason may be that in most of the moderate inflation countries, seigniorage is on the order of 2–3 percent of GDP and accounts for a significant share of government revenue. A shift to low single-digit inflation will require the fiscal problem to be tackled partly through tax reforms that make it possible to generate tax revenues at levels comparable with those of low-inflation countries and also through public expenditure restraint. Failure to pursue further disinflation may be costly, not least because of the risk that inflation may ratchet upward, with attendant costs in terms of low growth. This is most likely to be the case for inflation rates above the 30–40 percent range. When inflation exceeds that range, the probability that it will increase to triple digits rises sharply (Table 37).91 Nonetheless, the lessons learned in the design of disinflationary policies in high-inflation countries in the 1980s and 1990s should help these economies in future attempts to bring inflation down to single-digit levels at low output costs.

Table 37.

Probability of Inflation Being Above 100 Percent in Next Year at Different Rates of Inflation in Current Year

article image
Source: Michael Bruno and William Easterly, “Inflation Crises and Long-Run Growth,” NBER Working Paper No. 5209 (Cambridge, Massachusetts: National Bureau of Economic Research. August 1995).

Calculated as number of observations in this range followed by 100+ percent inflation, divided by total number of observations in this range (pooled cross-section, annual data, 1961–92).

Some Empirical Evidence on Factors Influencing the Inflationary Process

The monetary nature of sustained inflation is widely recognized. This principle is illustrated by the close relationship between the average rate of money growth and the average rate of inflation from 1960 to 1994 across a sample of 53 developed and developing countries (Chart 39). While money growth is the principle long-term determinant of inflation, the experience of the past thirty years shows that nonmonetary influences, such as commodity price increases, can play a role in triggering inflation. Charts 40 and 41 summarize the results of a study of the sources of inflation in developed and developing countries over the period from 1960 to 1992.92 The study compares the importance of various demand-side and supplyside influences on the inflationary process (Chart 40) and assesses the magnitude and duration of the response of inflation in reaction to an unexpected movement (a “shock”) in each of the determining factors (Chart 41). The supply-side influences considered are changes in oil prices, changes in non-oil commodity prices, and movements in nominal exchange rates.93 while the demand-side factors include the growth of broad money, a measure of the gap between actual output and potential output, and past rates of inflation itself. Fiscal deficits are not included as a separate influence on inflation because they are likely to be subsumed in other factors: the accommodation of fiscal deficits is a principal cause of monetary growth in many high-inflation developing economies, while deficits in low-inflation industrial countries strongly influence aggregate activity as reflected in the output gap. The results show the empirical relationships between the six factors without imposing any constraints as to the particular channels through which the factors interact. Furthermore, the results are not very sensitive to changes in the assumptions made about the direction of causality between contemporaneous movements in the six influences of inflation, most notably the direction of causality between movements in money and output.

Chart 39.
Chart 39.

Selected Industrial and Developing Countries: Inflation and Money Growth, 1960–95

(In percent change: annual averages)

There is a close relationship across countries between long-run inflation and long-run money growth.
Chart 40.
Chart 40.

Influences of Various Factors on Inflation

Share of movements in inflation, which are explained by five factors at the horizons of 0–1 year, 3–5 years, and 10 years.
Chart 41.
Chart 41.
Chart 41.

Response of Inflation to Various Shocks1

Response of inflation over a ten-year horizon to an unexpected movement (a “shock”) in the determinants of inflation.1 The shocks are each one standard deviation in magnitude, and the peak response of inflation is scaled to equal 100.

In both developed and developing countries, inflation was found to be very sticky: an unexpected increase in inflation resulted in a higher rate of inflation for four to five years in developing countries and ten or more years in the industrial countries. As discussed further below, the stickiness of inflation reflects several factors, including backward-looking expectations and indexation mechanisms. This persistence means that accommodation of inflationary shocks will have long-lasting effects, as the initial shock of a commodity price increase or an adverse movement in the terms of trade is propagated over time.

Perhaps surprisingly, oil and non-oil commodity prices directly account for only a modest portion of global inflationary movements. As discussed earlier, this may be due to the fact that the more important channels by which these shocks affect inflation are indirect, particularly through the use of monetary policy to offset the negative output effects from supply shocks that reduce potential output, and in the accommodation of the second-round effects on wages and prices. Nonetheless, the direct effects of shocks to oil and non-oil commodity prices on inflation are statistically significant, with the magnitude of the effects similar across industrial and developing countries. While commodity price shocks are not a pervasive feature of the economic environment, they clearly matter for inflation when they do occur.

The output gap accounts for a substantial portion of medium- and long-term movements in inflation in the industrial countries, with an unexpected increase in the output gap (i.e., a weakening of the economy) leading to a long-lived and statistically significant fall in inflation. This suggests that permanent changes in potential output, such as productivity movements and technological innovations, are important for inflation in the industrial countries, since fluctuations in demand over the business cycle would not be expected to have such a long-lasting effect on inflation. In contrast, the output gap explains little if any of the movements in inflation in developing countries as a group (see Chart 40). Note, however, that this represents the average effect of the output gap on inflation for a wide range of developing countries taken together, and this effect varies greatly across countries. Considered individually, there is a larger effect of the output gap on inflation in many countries in Asia, as in the results discussed in Box 3 in Chapter IV. Furthermore, in the developing countries as a group, an unexpected movement in the output gap has essentially no effect on inflation (see Chart 41).94 Since the run-up in inflation in the 1970s and 1980s and the subsequent decline were much greater in developing countries than in the industrial countries, these results support the findings discussed above that inflation can be reduced from high to moderate rates at little cost in terms of lost output, but that it may be more difficult to reduce inflation at low levels where there is a short-term trade-off between output and inflation.

Conversely, changes in nominal exchange rates and money growth account for a larger share of the movements in inflation in developing than in the industrial countries. And an unexpected depreciation of the exchange rate or an unexpected increase in the rate of money growth has a longer-lasting effect on inflation in developing countries than in industrial countries. These results support the view that a disinflationary strategy in developing countries should include a combination of fiscal adjustment and tax reform to reduce the dependence on monetary financing, along with a commitment to a stable exchange rate regime. The importance of inflation expectations as seen in the persistence of inflation suggests that there is an important role for policies that lessen the scope of institutional features, such as wage indexation, that tend to contribute to their stubbornness.

Costs of Inflation

Inflation imposes significant economic and social costs. By clouding relative price signals, generating uncertainty about future inflation, and generally reducing the information content of the price system, inflation distorts the allocation of resources and adversely affects economic efficiency and growth. Further, inflation brings about arbitrary redistributions of income and wealth. Many of the welfare costs of inflation, particularly of very high rates of inflation, have long been recognized, but the full costs, especially of moderate inflation, have been difficult to quantify and some of the costs have only more recently been widely recognized, while others may be larger than previously thought. As the public has become increasingly aware that higher inflation is associated with broadly inferior, rather than superior performance, there has been increased pressure on governments to follow policies consistent with low inflation.95 This revealed preference for low inflation has not, however, extended to a preference for pricelevel stability, reflecting perhaps an equally strong popular dislike for the costs of deflation.

Anticipated Versus Unanticipated Inflation

In analyzing the costs of inflation, a distinction needs to be made between anticipated and unanticipated inflation. It is widely acknowledged that the costs of fully anticipated inflation, which include the welfare costs of economizing on the use of money balances and the resource costs of periodic revisions of price lists (“menu” costs) are small, at least at low inflation rates. However, these costs are likely to be economically meaningful at high rates of inflation, especially at inflation rates sufficiently high to induce a demonetization of the economy. Then the costs are likely to be very high indeed, greatly exceeding estimates obtained from simple money demand calculations, because high inflation destroys social institutions and arrangements that simplify individual decisions and the coordination of activities. In particular, high inflation economies end up with a shrunken structure of intertemporal markets, with some markets disappearing altogether. In fact, this can occur not only at the very high rates of inflation that have been experienced in the postwar period only by some nonindustrial countries, but also at the more moderate rates of inflation experienced by industrial countries. For instance, the markets for long-dated, fixed-rate mortgages disappeared in some industrial countries in the late 1970s and early 1980s when inflation rates reached double digits. Because conventional estimates of the costs of inflation take no account of such disruption of market institutions and arrangements, they understate the true social cost.

The welfare costs of inflation are much higher when inflation is not fully anticipated. For inflation to be fully anticipated, all contracts, accounting systems, taxes, and other institutional arrangements would have to be fully indexed and adapted to it. In practice this is not possible; and attempts at widespread indexation have in most cases fostered higher inflation, with costs far exceeding the benefits of indexation per se. Tax systems are not fully indexed even in high-inflation countries; in industrial countries, tax systems are at best only partially indexed. Indeed, most industrial countries do not even issue government bonds indexed to inflation, though the number doing so is increasing (see Box 9). The interaction of inflation with a less than fully indexed tax system works to discourage capital accumulation, distort the allocation of capital among economic sectors, distort the allocation of lifetime consumption between early years and later years, and distort corporate and household financing and labor supply decisions.

The costs of these distortions depend on the institutional features of the economy. For the U.S. economy they have been estimated at 2–3 percent of GDP at an inflation rate of 10 percent, while a recent study has calculated that reducing the ongoing rate of inflation from 4 percent to 2 percent would result in a perpetual annual welfare gain of about 1 percent of GDP.96 These costs, however, do not include some of the costs associated with the distortion of financial decisions. For instance, in many countries during the 1980s the interaction of inflation and tax rules resulted in low or negative after-tax borrowing costs, which contributed to excessive corporate borrowing and financial instability. In a number of countries, this resulted in large banking sector losses, increased the risk of systemic crisis, and led to the use of public funds to shore up problem-ridden financial institutions.

Owing to the absence of complete indexation, the costs of inflation do not fall equally across the population. Unanticipated inflation thus redistributes income and wealth: those on fixed incomes and with little wealth in the forms of real assets and foreign assets are made worse off, and negative real interest rates ex post mean that wealth is transferred from lenders to borrowers. Also, to the extent that inflation reduces the value of government debt and the real cost of servicing the debt, current holders of government debt lose while future taxpayers gain, so that there is an arbitrary redistribution of wealth between generations. Surprisingly little is known about the empirical, quantitative effects of inflation on income distribution. However, a recent study aimed at filling this gap found that both a higher and a more variable inflation rate tended to increase income inequality and that a negative effect of inflation on income distribution is more likely in countries with a less developed financial sector (Box 10).

Indexed Bonds and Expected Inflation

Inflation-indexed bonds have been advocated by many economists—including Marshall, Irving Fisher, Keynes, and Milton Friedman—as a useful tool for debt management and monetary policy. Yet indexed bonds remain uncommon. A belief that indexation is in general likely to perpetuate inflation, for example, through wage-price spirals or effects on expectations, and through subsequent pressures for accommodation by monetary authorities, is a major reason governments have been reluctant to offer them.1 The experience of countries that adopted comprehensive indexation beginning in the 1960s and 1970s bolstered this view. Examples include Argentina, Brazil, Chile, Colombia, as well as Iceland and Israel. In these cases, indexation was seen as a way to cope with inflation, and indexation of financial markets in particular as an expedient to promote domestic saving and capital formation. By the 1980s, however, inflation performance in a number of these countries had deteriorated, in some cases to the point of hyperinflation, and a number have since taken steps to reduce the scope of indexation. Since the early 1980s, many industrial countries have also abolished wage indexation, which was widely perceived to have contributed to the escalation of inflation in the 1970s.

Advocates of the indexation of government debt instruments argue that it removes the incentive to inflate. Attempts to reduce real debt burdens by generating unanticipated inflation are self-defeating when government liabilities are indexed. Governments therefore are more likely to focus on reasonable price stability as the appropriate monetary policy objective. Indexed bonds may thus enhance monetary policy credibility and in doing so reduce funding costs by the amount of the inflation risk premium incorporated in conventional bonds. Advocates point out that indexation in itself does not foster inflation.2 Monetary and fiscal policies are at the root of inflation, regardless of indexation; should any potential inflationary effects of indexing occur, they can be countered by monetary and fiscal policy actions. Advocates generally do not argue for indexing the entire government debt, since there is still a public demand for nominal debt.

Starting in the early 1980s, a new group of issuers in the industrial countries, led by the United Kingdom in 1981 and followed by Australia, Canada, New Zealand, and Sweden, began issuing indexed bonds and integrated them into existing debt-management programs.3 In these countries, indexed bonds now trade alongside nominal bonds, albeit in much smaller volumes. The introduction of indexed bonds was justified chiefly on the basis of savings in debt-service costs, the development of capital markets and the promotion of saving, and benefits to the implementation and credibility of monetary policy. The U.S. Treasury announced in May 1996 that it too will soon offer indexed bonds, for similar reasons.

The nominal return on indexed bonds includes a real return plus compensation for the ex post erosion of purchasing power, while the nominal return on conventional bonds comprises the same expected or required real rate of return plus compensation for expected inflation and an inflation risk premium. While the real return on an indexed bond is assured at the time of purchase, the nominal return is uncertain; for nominal bonds, it is the nominal return that is certain (disregarding coupon reinvestment), while the real return is uncertain. In principle, indexed bond payments could be tied to any inflation-related index, including a consumer price index (CPI), a wage index, or an exchange rate index. In practice, most issues have been tied to a broad index of domestic prices, typically the consumer price index. Such indices are widely disseminated and understood, available with a short lag, and rarely subject to revision. A drawback, however, is that CPIs typically overstate true inflation because of measurement biases.

Indexed bonds can be viewed as a public good with a potential to improve economic welfare. By providing such instruments, governments can fill a void in financial markets, offering securities free from both credit risk and purchasing power risk. Other assets, such as real estate, equities, and commodities, which were once thought to be good inflation hedges, have proven not to be so when measured by the past correlation of returns on inflation. Indexed bonds can also be viewed as providing a form of insurance—purchasing power insurance—to investors. Buyers will be likely to be willing to pay a premium (in the form of a smaller return than on nominal bonds) in return for shifting the burden of inflation risk to the issuer. The government’s borrowing costs will then decline by the amount of the inflation risk premium, which has been absorbed by the issuer. A government issuer, however, may be neutral to inflation risk, for example, because its real budget position is less affected by inflation, or perhaps because it exerts some control over the inflation outcome.

Indexed bonds may also promote financial savings. Where inflation is high or volatile, indexed bonds may be the most promising vehicle for encouraging domestic financial savings, compared with real or foreign asset accumulation. By extension, they may help to foster the growth of domestic capital markets. Because unanticipated inflation entails unintended transfers of wealth from lenders to borrowers, indexed bonds have also been promoted on purely distributional grounds, especially because some lenders, such as pensioners or those saving for retirement, are often presumed to be the most disadvantaged and least able to protect themselves from inflation.

Indexed bonds will provide cheaper financing than nominal bonds, especially if lenders overestimate future inflation. Thus, if inflation expectations are to a large extent backward looking, indexation will be particularly attractive to a government that is confident of reducing inflation. During the 1970s, when markets consistently underestimated inflation and real returns on bonds in many countries were negative, borrowers were better off issuing nominal debt, since the real value of this debt was eroded by inflation. Governments gained from the inflation tax. By the 1980s, however, when nominal and real interest rates in many countries were stubbornly high and inflation expectations were slow to fall, governments would have profited by switching to indexed debt. But while the public’s inflation expectations may well be wrong from time to time, systematic errors are unlikely to persist for long periods. It is more likely that such forecasting errors cancel out over the long run. Nonetheless, the perception that inflation expectations were excessive was one reason indexed bonds were introduced in the United Kingdom and recently reintroduced in New Zealand. In both cases, authorities cited a lack of credence in their inflation-fighting policies, as indicated by the high level of real rates.

Another debt management argument is that a treasury would stabilize its real debt-servicing costs by issuing indexed debt, better matching government tax revenues, which tend to move in line with inflation. Issuers would also achieve a more predictable pattern of real cash outflows (at the expense of a less predictable pattern of nominal cash outflows). The result would be a more predictable overall real budgetary position.

Indexed bonds can also provide useful information on inflation expectations and real interest rates, aiding the implementation of monetary policy and providing evidence on its credibility. Real yields are quoted directly on indexed bonds, and the spread between nominal and indexed bond yields provides a readily available measure of the market’s inflation expectations. Several problems have been identified with such measures, however. First, the existence of an unobserved inflation risk factor and a possible liquidity premium on indexed bonds (the latter because issues are relatively small and trade infrequently) makes disentangling the pure expectations component difficult.4 Furthermore, lags in the application of the index and possible differences in tax treatment between the two types of bonds distort the information content on inflation expectations.

The Australian treasury has indicated that since its indexed bonds were reintroduced in 1993 the yield gap has at times significantly overstated mean inflation expectations derived from survey data or expressed by financial market commentators. This may occur because of the small stock of Australian indexed bonds on issue and their relatively thin trading, limiting the interpretation and usefulness of the yield gap. The Bank of England, however, publishes inflation expectations data derived from indexed gilts as one of its policy indicators and as a gauge of monetary policy credibility. In general, alternative measures of inflation expectations through sampling techniques or models are likely to prove costlier, less accurate, less timely, and more difficult to interpret than information provided through indexed bonds.


For a survey of the arguments for and against indexed bonds, their design, and the experience and practice of government issuers, see Robert Price, “The Rationale and Design of Inflation Indexed Bonds,” IMF Working Paper (forthcoming).


See, for example. Stanley Fischer. “Indexing and Inflation.” Journal of Monetary Economics, Vol. 12 (November 1983). pp. 519–41.


Finland. Iceland, and France (in the 1950s) are rare cases of industrial countries that issued indexed bonds before 1981. Barro has argued, however, that under the gold standard and the Bretton Woods par value system, government debt effectively was indexed, and that it is nominal bonds that are the recent innovation under paper money standards. See Robert Barro, “Optimal Debt Management,” paper presented at the Bank of England’s conference on the U.K. Index-Linked Gilt-Edged Market: Future Development, held September 14–15. 1995 in London.


If both premiums are relatively constant, however, changes in the yield spread would still provide useful information on inflation expectations.

Some of the major costs of inflation stem from the reduction in economic efficiency caused by increased uncertainty about future prices and the distortion of relative price signals. High inflation tends to be associated with increased variability not only of the aggregate price level but also of relative prices. Such increased variability, by increasing uncertainty about inflation and relative prices, has a negative effect on output.97 It appears from Charts 42 and 43 that industrial countries that on average have had higher rates of inflation have tended to have more variable rates of inflation, and that countries with more volatile inflation have tended to have more variable output growth.

Chart 42.
Chart 42.

Industrial Countries: Inflation and Inflation Variability, 1960–95

Greater volatility of inflation tends to be associated with higher levels of inflation.
Chart 43.
Chart 43.

Industrial Countries: Variability in Inflation and Real Output Growth, 1960–95

Volatile inflation is associated with volatile growth.

Effects of Inflation on Economic Growth

Inflation may not only reduce the level of output in an economy, notably by reducing the efficiency with which the factors of production are utilized, but it may also reduce the rate of growth of output, both by discouraging investment in physical and human capital, and by discouraging innovation and thus depressing the growth of total factor productivity. Even a small reduction in the growth rate will, over time, have substantial cumulative effects on output and welfare.

Empirical evidence shows consistently that inflation is negatively correlated with economic growth, and that the growth costs of inflation are economically significant. A number of estimates point to a reduction of economic growth in the range of 0.02–0.08 of 1 percentage point for a 1 percentage point increase in inflation.98 The negative correlation between inflation and growth is most strongly apparent when economies with high rates of inflation are included in cross-country data (Chart 44) or when episodes of high inflation are included in time-series data. The relationship between inflation and growth at low rates of inflation is less clear. Some studies have found a negative relationship even at low single-digit rates of inflation, whereas other studies have not found a statistically significant negative relationship for annual inflation rates below 8 to 10 percent.99 Be that as it may, a number of developing countries with average annual inflation rates of between 5 and 10 percent have been able to sustain rates of economic growth in the range of 5 to 10 percent a year. Another qualification is that the causality in the inflation-growth relationship has not been easy to establish, especially for low inflation rates. Not only may low growth be conducive to high inflation, but also the negative correlation may reflect the effects of third variables: thus, an adverse supply shock could simultaneously raise inflation and lower growth, and conversely for a favorable supply shock. Several conclusions emerge from this evidence. First, there is no support for the notion that higher inflation is beneficial for longer-term growth, although, as discussed further below, there is also no evidence that zero inflation is superior to very low positive rates of inflation. Second, the evidence is strongly indicative of a negative relationship between inflation and long-term growth, even at moderate rates of inflation of 10 to 30 percent. High rates of inflation are clearly harmful to growth. Growth falls steeply as inflation exceeds 30–40 percent and tends strongly to become negative at very high rates of inflation. The experience of the transition economies, in particular, shows that reducing high inflation has been a precondition for the revival of growth, and that stabilization leads to growth: specifically, it seems that annual inflation has to be reduced to less than 50 percent for growth to resume.100 Third, even at the lower rates of inflation of 2–10 percent experienced by industrial countries over the past three decades, the greater variability of output associated with relatively higher inflation complicates the conduct of macroeconomic policy and increases the risk that relatively deep recessions may be required to stabilize the inflation rate in the event of overheating.

Chart 44.
Chart 44.

Inflation and Real Output Growth, 1965–95

(In percent change; annual averages)

High inflation has generally been associated with poor growth.

Effects of High Inflation on Income Distribution

It has been increasingly recognized that high and variable inflation not only complicates macroeconomic policy but also tends to induce significant changes in the distribution of income and wealth. While inflation may in principle either increase or reduce distributional inequalities, depending on the circumstances, most empirical studies—single-country studies as well as a recent cross country study—find that reducing inflation from high levels usually reduces inequality. Such effects may be of considerable importance in evaluating the costs and benefits of macroeconomic adjustment.

Income distribution is determined by two factors: the allocation of income-earning assets, including human capital, and their relative returns. Inflation affects income distribution, first, by altering relative returns. A classic example is the inflation tax: if indexed assets and assets denominated in foreign currency are held primarily by upper-income groups, while lower-income groups hold mainly domestic cash, then inflation, even if expected, will increase inequality, as the inflation tax will be borne by the poorer segments of society. “Bracket creep” in schedules of income tax rates provides another mechanism: with nonindexed income tax systems, even if they are progressive, wage inflation will reduce disproportionately the real return to the human capital of workers at the bottom of the wage scale. Again, the worsening will occur whether or not the inflation is expected.

Inflation also affects income distribution by changing the relative values of different assets and liabilities. Unanticipated inflation, by lowering the real value of both nominal assets and nominal debt, favors debtors and holders of real equity over lenders and owners of nominal assets. In the absence of indexation, it also favors the buying side in long-term contracts, including employment agreements. The net distributional effects will depend on the availability and use of hedging instruments (notably including contractual indexation), the incidence of contracts, and the distribution of nominal assets and liabilities across groups. In financially less-developed economies, lower-income groups rarely hold significant nominal liabilities, such as mortgages. Consequently, they tend to suffer through erosion of their nominal assets, while the middle class may be compensated by an offsetting erosion of nominal liabilities.

High inflation may also affect income distribution indirectly by lowering output and employment through a variety of channels, including distortions in relative price signals and their effects on allocative efficiency. These effects also are likely to increase inequalities.

There are ample reasons, therefore, to expect high inflation to exert a significant regressive influence on the distribution of income. In practice, the precise effects will depend on the distribution of assets in an economy, the development of financial markets, the flexibility of prices and wages, and other factors.

The paucity of high-quality data on income distribution limits the scope for empirical work, particularly for cross-country studies requiring comparable time-series data. Schultz and Blinder and Esaki represent early pioneering work for the United States.1 Schultz found that an increase in inflation increased income inequality, and this finding was generally supported by the results of Blinder and Esaki, though they identified the middle-income groups as the main losers from higher inflation. A number of later studies of industrial economies confirmed the differential impact of inflation on different income groups. No robust pattern emerged, however, again suggesting the importance of country-specific factors, notably the stage of development of the financial sector, the degree of indexation, and the initial distribution of assets and liabilities.

While the effects of inflation on income distribution are likely to be particularly important in developing and transition countries—because of the higher and more variable inflation in these countries, and also because of the absence of widely available indexation mechanisms—the lack of time-series data has limited empirical investigation. The void has been partly filled by recent cross-country work based on panel data for some forty developing and industrial countries.2 After controlling for country effects and other determinants of income distribution, it was found that high inflation was associated with increased income inequality. The results showed, furthermore, that income inequality increased with the variability of both the inflation rate and the exchange rate. Indeed, inflation variability was more important than changes in the average inflation rate.

Overall, the evidence strongly suggests that high average inflation and high variability of inflation increase income inequality significantly. By extension, reductions in inflation and the stabilization of inflation at low levels are likely to reduce inequality. This inference is of some importance, in assessing the common criticism that the costs of stabilization are borne primarily by the poor.


Paul T. Schultz, “Secular Trends and Cyclical Behavior of Income Distribution in the United States: 1944–1965,” in NBER Studies in Income and Wealth, Vol. 33 (New York: Colombia University, 1969), pp. 663–81. Alan G. Blinder and Howard Y. Esaki, “Macroeconomic Activity and Income Distribution in the Postwar United States,” Review of Economics and Statistics, Vol. 60 (November 1978), pp. 414–23.


Aleš Bulíř and Anne-Marie Guide: “Inflation and Income Distribution: Further Evidence on Empirical Links,” IMF Working Paper 95/86 (August 1995). as discussed further below, there is also no evidence that zero inflation is superior to very low positive rates of inflation. Second, the evidence is strongly indicative of a negative relationship between inflation and long-term growth, even at moderate rates of inflation of 10 to 30 percent. High rates of inflation are clearly harmful to growth. Growth falls steeply as inflation exceeds 30–40 percent and tends strongly to become Chart 44. Inflation and Real Output Growth, 1965–95

The above discussion is only one side of the coin. The other side is that there are also costs from deflation, that is, a declining price level. Although widespread deflation has not been experienced since the 1920s and 1930s, that experience clearly revealed that deflation—unanticipated or anticipated—is detrimental to economic activity. Unanticipated deflation, by increasing the burden of nominal debt and lowering borrowers’ net worth, can lead to defaults and financial distress, which depress investment and economic activity. Even if anticipated, deflation can have a depressive effect on the economy by reducing the attractiveness of holding real capital.

Costs of Disinflation

Apart from the costs of inflation per se, inflation also brings with it the costs that have to be paid to reduce it. Indeed, the disinflation of the past few years, like the two earlier episodes of disinflation in the mid- 1970s and the first half of the 1980s, was not without costs in lost output and employment. It is expensive in the short run to fight inflation—at least at moderate rates of inflation. Many of the costs of disinflation arise from the slow adjustment of prices and wages. With slowly adjusting prices and wages, reducing the rate of inflation generally requires a slowdown in economic activity during which output falls below the economy’s productive capacity. The persistence of inflation therefore depends partly on institutional factors, such as staggered wage agreements, indexation mechanisms, and other regulatory and institutional features (particularly as they relate to the labor market) that impart a degree of stickiness to nominal wages and prices. It also depends on inflation expectations and the credibility of policies.

If expectations were entirely backward looking, then there would be a lot of inertia in inflation—current inflation would be affected only by past inflation and factors that affect current excess demand, and disinflation would tend to entail relatively large transitory output losses. Although opinions differ on how expectations are formed, most economists would agree that there is an element of forward-looking behavior in the formation of actual and expected prices and wages.101 The greater the degree to which expectations are forward looking, the greater the extent to which perceptions of policy can influence expected inflation. The credibility of policy is thus a potentially important determinant of the persistence of inflation and the costs of disinflation. The more credible is a policy of disinflation, the more quickly will inflation adjust and the smaller will be the output losses.

Policymakers’ past behavior and institutional arrangements are both likely to play a role in determining the credibility of policies. Thus, policymakers in a country with a history of price stability may enjoy a degree of credibility in dealing with inflationary shocks not enjoyed by their counterparts in countries with a less consistent history of low inflation. Similarly, institutional arrangements—for example, a central bank charter that prevents automatic borrowing by the government or that clearly specifies the central bank’s price stability objectives—may also promote credibility.

Often the single most important determinant of the credibility of an anti-inflationary policy is the coherence of the overall macroeconomic program—in particular, whether fiscal policy is consistent with the anti-inflationary stance of monetary policy. This is especially relevant for economies with high levels of public debt. In such economies, markets may become very skeptical of the ability of the monetary authorities to pursue a noninflationary monetary policy. This is most likely to be so when there is a persistent budget deficit and the real rate of interest exceeds the economy’s growth rate. In such a situation, a tightening of monetary policy will worsen the debt dynamics as higher interest rates increase the stock of debt that, barring repudiation, would have to be monetized in the future in order to stabilize the ratio of debt to GDP. Without a supportive fiscal policy, therefore, a tight monetary policy to lower current inflation could actually lead to higher inflation in the future. If agents anticipate this outcome and incorporate it in their expectations of inflation and interest rates, disinflationary policy may lead to higher interest rates, slower growth, and higher inflation—but the higher inflation does not reduce the debt burden. This vicious circle is of potential concern in a number of high-debt countries, especially those without an independent central bank. The way out, as countries have increasingly realized, is fiscal consolidation that reduces public debt.

In addition to nominal rigidities and the consistency and credibility of policies, the costs of disinflation depend on two other factors: initial conditions and the speed with which policies are implemented. As regards initial conditions, it tends to be less difficult to reduce inflation when a country starts from a higher rate of inflation than when it starts from a lower rate. In other words, it may be less costly to go from, say, an inflation rate in the high teens to one in the low teens than to go from an inflation rate in the high single digits to one in the low single digits, in part because nominal rigidities become more important at low inflation rates. The speed at which disinflation is brought about may also influence the cost of reducing inflation. Empirical evidence on this is mixed, however, reflecting perhaps the difficulty of disentangling credibility and other effects. Some researchers have found that, provided it is fully credible, a policy of gradual disinflation is less costly than a sharp disinflation—with the optimal rate of disinflation depending on the frequency with which contracts (for wages and prices) are rolled over. Others, however, have found that disinflation is less costly if it is quick.102

Asymmetric Effects of Economic Activity on Inflation

Analysis of the trade-off between inflation and output (or more precisely between inflation and the gap between actual and potential output) has often been based on the assumption that the trade-off is symmetric—that a rise in inflation caused by excess demand can be reversed without any net loss in output and employment. The implications of this are that stabilization policy cannot affect the average level of output and employment over the business cycle. Some recent research, however, finds that the relationship is asymmetric: that excess demand is more inflationary than excess supply is disinflationary.103 Thus, any output and employment gains associated with an increase in inflation resulting from excess demand will be more than offset by the losses necessary to reverse the rise in inflation. The estimated degree of asymmetry for the major industrial countries is large enough for the net costs of allowing the economy to overheat to be quite significant. These are illustrated in Chart 45, which shows simulated responses of inflation and unemployment to an increase in excess demand that temporarily reduces the unemployment rate below the nonaccelerating inflation rate of unemployment (NAIRU) by 1 percentage point. The simulations are from an estimated model of the U.S. unemployment-inflation process and assume a NAIRU of 6 percent.104

Chart 45.
Chart 45.

Asymmetric and Symmetric Model Responses to Overheating

Effects of a decline in unemployment one percent below the nonaccelerating inflation rate of unemployment (NAIRU).

When the inflation-unemployment relationship is asymmetric, a substantial increase in excess demand has a considerably larger effect on inflation than when the relationship is symmetric, and real interest rates have to rise by more in order to open up the correspondingly larger output gap needed to restore the inflation rate to its initial level. In both the symmetric and the asymmetric versions of the inflation-unemployment relationship, the unemployment rate returns to its initial level when the excess demand shock wears off. The cumulative effect on unemployment, however, is larger in the asymmetric version—by some 2 percentage points in this case. The cost of allowing the economy to overheat rises more than proportionately with the extent of overheating. The more an economy is allowed to overheat, the deeper or longer will be the recessions required to achieve a given inflation objective on average over a business cycle, and hence the higher will be the average rate of unemployment. Asymmetry becomes especially important when the economy operates at unemployment rates substantially away from the NAIRU. Close to the NAIRU, the inflation-output relationship is almost symmetric, so that the net costs of a modest degree of overheating are proportionately smaller.

Some important policy implications emerge from this analysis. First, asymmetry underscores the importance of avoiding excess demand pressures through preemptive steps to tighten monetary conditions before inflation starts to accelerate.105 In this respect, it is notable that the raising of short-term interest rates in early 1994 in the economies most advanced in the current expansion, including the United States and the United Kingdom, stemmed inflationary pressures without major costs in terms of output and employment. This contrasts with the previous cycle in the late 1980s, when failure to act preemptively to avoid overheating resulted in a major slowdown in economic activity, particularly in European countries, when policy was eventually tightened.

A second implication is that a macroeconomic policy that avoids wide boom-and-bust cycles can, over time, result in a higher average level of employment and output than a policy that does not. Thus, it is particularly important to avoid substantial overheating, since the social costs of a deep and prolonged recession exceed the benefits of a boom of approximately equal magnitude.

Yet another implication is that the need for caution in the conduct of policy is reinforced by the unavoidable uncertainty about current and prospective inflationary pressures. Knowledge of potential output and the NAIRU is imprecise, and so is knowledge of excess demand conditions at any given time. This argues for the exercise of extra caution when an economy is at, or is fast approaching, potential.

Policies to Foster Disinflation

Policies adopted to reduce the costs of disinflation fall into two broad categories: structural policies, especially for labor markets, and measures to bolster the credibility of anti-inflationary policies, particularly by strengthening policymakers’ commitment to them. Such policies aim essentially to remove rigidities in actual wages and prices and to make the formation of inflation expectations less backward looking and more reflective of current anti-inflationary policy.

Structural policies have focused mainly on labor markets, aiming to increase wage flexibility and to promote market efficiency. Measures adopted have included active labor market programs to improve the employability of unemployed workers (e.g., training and placement), and regulatory and institutional reforms to enhance wage flexibility and remove disincentives to employment (e.g., changes to minimum wage laws and to regulations that affect the costs of hiring and laying off workers). In many instances, however, the structural reforms undertaken have tended to be piecemeal, minimizing their potential impact. As emphasized in Chapter III, broad-based labor market reforms continue to be needed, especially in Europe.

Measures to bolster the credibility of anti-inflationary policies have proceeded along several lines. One avenue has been to establish an explicit nominal anchor for monetary policy. Its purpose has been to tie down inflation expectations, and also to tie the authorities’ hands and exert a measure of self-discipline. Fixed exchange rates have long served this function for numerous countries, as discussed earlier. The ERM of the European Monetary System is the most prominent example among industrial countries; by tying their currencies to the deutsche mark, members of the ERM hoped to gain some of the Bundesbank’s credibility. Inflation rates among the members of the ERM have indeed converged to substantially lower levels than in the early years of the system, but the credibility gain is hard to assess, in part because inflation has also converged at low levels in non-ERM industrial countries. The ERM experience has also demonstrated that pegged exchange rates may be more difficult to maintain than previously thought. Increased international capital mobility has increased the vulnerability of pegged exchange rates to speculative attacks induced by perceived divergences in fundamentals. Furthermore, with downward price and wage rigidity, it may be difficult to bring about the real exchange rate adjustments necessitated by asymmetric real shocks. This may lead to devaluations or abandonment of the peg, or, if exchange rates remain fixed, result in higher inflation.106

Countries opting not to fix their exchange rate often adopted targets for monetary aggregates. Monetary targets were widely adopted among the major industrial countries in the mid-to-late 1970s but were generally abandoned in the 1980s as relationships between targeted monetary aggregates and inflation broke down, partly owing to financial innovations.107

More recently, explicit inflation targets have been adopted as the nominal anchor by a number of central banks. These have tended to be countries with histories of relatively high inflation. Central banks with well-established records of maintaining low inflation have not found the need to adopt explicit inflation targets, even though they have also found it useful to quantify their medium-term inflation goals. It is too soon to assess the role of explicit inflation targets in increasing credibility and lowering the costs of disinflation, since much of the period during which inflation targets have been in place has been characterized by a widespread easing of price pressures. Although the evidence thus far does not point to a detectable influence of inflation targets on the formation of inflation expectations, they may have helped to stabilize inflation expectations and to lower inflation-risk premiums.108 The recent experience with inflation targeting shows that inflation targets will sometimes be missed. This suggests that establishing too narrow a target range may end up diminishing rather than enhancing credibility if the targets are too frequently breached by significant amounts and the overall inflation performance is unsatisfactory. Alternatively, credibility may be impaired if the public comes to believe that the costs of overly rigid adherence to an inflation target may be too high to sustain.

A second approach to enhancing credibility has been to change the institutional framework in which monetary policy operates. Here, the major development has been the general tendency toward greater central bank operating independence and toward the clear establishment of price stability as the central bank’s principal goal. Starting with New Zealand, several countries, including France, Spain, Chile, Mexico, and Venezuela have granted their central banks more independence, and discussions are under way in Japan. Also in the United Kingdom, increased transparency in the decision-making process and greater freedom for the Bank of England to make its views known publicly have served to increase somewhat the bank’s independence and power in the conduct of monetary policy. The purpose of greater central bank independence is to eliminate the possible inflationary bias of discretionary policy. The empirical evidence is that a country is more likely to have low inflation if the central bank is operationally independent (Box 11). Legal independence for a central bank is not sufficient, however, to ensure good inflation performance. A consensus is also necessary within society that low inflation is desirable. And as indicated earlier, cooperation between the monetary and fiscal authorities is essential.

Fiscal consolidation has been an important third avenue to enhance credibility. Markets tend to be skeptical of monetary authorities’ ability to pursue an anti-inflationary monetary policy in the presence of large budget deficits. Large budget deficits can lead to high real interest rates, rapidly growing public debt, and exchange market pressures. Especially in economies with high levels of public debt, as discussed above, an anti-inflationary monetary policy that is not supported by fiscal adjustment might not be credible.

Achieving and Sustaining Price Stability

With inflation now at the lowest level in decades in many countries, the question naturally arises as to what constitutes price stability. Is zero inflation preferable to a modest rate of inflation of, say, 2 percent? How far should anti-inflationary policies be pursued?

What Constitutes Price Stability?

Several reasons have been advanced as to why the optimal inflation rate may not be zero. A long-held view is that a small positive rate of inflation enhances the efficiency of interindustry adjustments by facilitating relative real wage adjustments. Declines in real wages in a particular industry—and in the relative price of the particular good or service—can be brought about more easily through wage increases in other industries than through absolute declines in nominal wages in the industry concerned. Although nominal wage cuts are rare, this may not be as important a factor as it seems. This is because with productivity growing, real wage adjustments can be effected without nominal wage cuts even while maintaining zero inflation. Furthermore, total compensation may be more flexible than nominal wages. Firms can adjust their total compensation bill by changing benefits rather than nominal wages or salaries.

Another reason why zero inflation may not be desirable, which is an extension of the one above, is that it may increase the unemployment rate and correspondingly reduce the average level of output. A recent estimate for the United States has found the difference in the sustainable rate of unemployment between operating with a steady 3 percent inflation rate and steady zero inflation to be over 1 percentage point.109 These estimates are derived from a model in which downward nominal wage rigidity constrains firms in their ability to make desired adjustments in real wages in response to changes in demand, resulting in inefficient reductions in employment. At low rates of productivity growth, the inefficiencies increase in significance as the rate of inflation approaches zero: the relation between inflation and unemployment is highly nonlinear at very low rates, so that targeting zero inflation may lead to unnecessarily high unemployment and large real costs to the economy. Since it is the sum of inflation and productivity growth that determines the number of firms constrained and the extent of the constraint, structural policies aimed at lessening nominal rigidities and boosting productivity growth will lower the optimal rate of inflation. In the absence of structural reforms, however, maintaining zero inflation may not be optimal.

Yet another reason why a policy of maintaining zero inflation may not be optimal is that a small positive rate of inflation enhances the efficacy of countercyclical monetary policy by allowing the short-term real interest rate to be negative, thereby increasing the potency of monetary policy during recessions. In a zero inflation environment, the best that the monetary authorities can do to mitigate the effects of an adverse shock to aggregate demand is to drive nominal and real short-term interest rates to zero. Such a reduction in the real interest rate, however, may not be sufficiently large to fully offset the negative effects of a large shock to aggregate demand.110 The importance of this argument depends on the nature of the monetary transmission mechanism, in particular the relative importance of the liquidity effect.

A final consideration is that official price indices are in many cases considered to overstate true inflation primarily because of unmeasured quality changes and the introduction of new goods, but also because of failure to properly account for substitution from goods whose prices have increased to other goods, substitution among retail outlets, and other biases. Estimates of the overall bias in the consumer price index range from 1 to 2½ percent a year for the United States and ½ to 1 percent for Canada, the two countries where the issue has received most attention.111 Estimates of the upward bias in consumer price inflation in Europe, Japan, and other countries are not available, but it seems very likely that consumer price increases are overstated in those countries as well.112

In brief, there are analytical grounds for thinking that the optimal rate of inflation is not zero but is rather a small positive rate. Except for the negative real interest rate argument, however, they are not independent of the macroeconomic policies adopted nor of structural features of the economy that are themselves amenable to change. Measurement bias alone suggests that it may be prudent for central banks to aim for an inflation rate of 1–2 percent, at least until some of the measurement issues have been resolved and greater efforts have been made to remove structural rigidities in labor and product markets. Indeed, none of the countries that have announced inflation targets has set a target of zero inflation, and no central bank seems to be aiming at absolute stability of the price level. Instead, most have established target inflation ranges between 1 percent and 3 percent. In fact, even in the countries that have exhibited the strongest dedication to “price stability,” inflation rates over periods as long as a decade have generally not run below about 2 percent, and in no case has zero inflation over five years been achieved for any country in the postwar era.

How Far Should Anti-Inflationary Policies Be Pursued?

Economic principles suggest that a policy of disinflation should be pursued as long as the present discounted value of the social benefits from a further reduction in inflation outweigh the present discounted value of the social costs.113 Although definitive answers are impossible, in part because they involve discounting the welfare gains of future generations, a discussion of some of the factors that affect the analysis can shed some light on the problem.

The two most relevant factors to consider are whether inflation affects the potential growth rate of output and whether the costs of disinflation are permanent or transitory. If inflation has a depressing effect on potential growth, even by a small amount, then it is most likely that the benefits of bringing inflation down to low levels (in the 1 to 2 percent range) will outweigh the costs, since the cumulative output loss will be large. Similarly, if the output losses of disinflation are transitory, it is quite likely that they will be outweighed by the permanent benefits of reasonable price stability.

If, however, the unemployment and output losses of disinflation are permanent (or at least sufficiently long lasting to be considered permanent) and substantial, then the optimal rate of inflation will depend on the types of disturbance to which the economy is prone, and the starting point of the disinflation process. Clearly, the issue is empirical and cannot be resolved by economic theory alone. Nonetheless, insofar as nominal rigidities are a function of both economic policies and institutional structures and regulations, the optimal rate of inflation can be reduced over time by appropriate policies and structural reforms.

Central Bank Independence and Inflation

Increasing the independence of a country’s central bank can help to achieve and maintain low inflation. The principal reason is that a central bank that is endowed with a mandate to pursue price stability may thereby be better insulated from political pressure for lower interest rates or monetary funding of fiscal deficits, so that monetary policy will be more clearly oriented toward the medium-term objective of reasonable price stability. With an independent central bank, any political interference in the operation of monetary policy would have to be more transparent. Thus, a government that wishes to change the primary goal of monetary policy is forced to act openly by, for example, publicly instructing the central bank that it wants it to conduct a more expansionary monetary policy than is consistent with the central bank’s goal, or that it requires the central bank to provide deficit financing.

Increasing the independence of the central bank may promote low inflation, but it is not a panacea. It is one part of a set of institutions and reforms that can support the goal of price stability.

It is useful to distinguish two forms of central bank independence: goal independence and instrument independence.1 Goal independence allows the central bank to choose the goals of monetary policy, rather than have them determined by the government or be legislated in the central bank’s constitution. The central bank could decide to pursue a goal of low inflation or a target rate of unemployment. Instrument independence lets the central bank pursue its goals (self-imposed or legislated) in the manner it deems most appropriate. The central bank is allowed to choose the timing and size of movements in its monetary policy instrument (e.g., a short-term interest rate). In a democratic society, it may be appropriate for the goals of the central bank to be set by the government, in which case the central bank should not have goal independence. To achieve price stability, however, the central bank, in addition to having price stability as its primary goal, must have instrument independence.

Greater independence of the central bank should not come at the expense of reduced accountability. Independence is meant to ensure that the operation of monetary policy is not affected by short-term political considerations. However, the central bank should be accountable to the government and the public for achieving its designated goal. In a number of countries (e.g., the United Kingdom, New Zealand, and Sweden), this accountability has taken the form of a quarterly or semiannual “Inflation Report” detailing the outlook for inflation and justifying the decisions taken on monetary policy in terms of the inflation objective.2 In other countries, including the United States, the head of the central bank is required to testify on a regular basis before legislative bodies. The planned European System of Central Banks will be required first and foremost to achieve price stability. Without prejudice to that objective, it will be required to support the general economic policies in the Community, including the objectives of sustainable noninflationary growth and high levels of employment. The European Central Bank will have a constitution similar to that of the Bundesbank. Its independence will be enhanced by the appointment of governors for one eight year term only. It will be required to publish regular reports and to testify before the European Parliament.


Industrial Countries: Central Bank Independence and Inflation

The relationship between central bank independence and inflation has been extensively investigated.3 The chart above plots for selected industrial countries the average inflation rate from 1970–95 against an index of central bank independence. The index was calculated by assessing each central bank on a number of criteria, including its goals, its financial linkages with the government, and the procedure used to appoint the central bank governor.4 A higher score on the index indicates greater independence. Various studies have calculated indices of central bank independence, and although they arrive at somewhat different rankings from those shown in the chart above, they generally show an inverse relationship between the degree of central bank independence and the average inflation rate: the more independent a central bank, the lower the rate of inflation. The statistical significance of the relationship, however, hinges on the inclusion of data for the 1970s and 1980s. For the 1960s, and the 1990s, there is no clear relationship between the index of central bank independence and the rate of inflation. This suggests that one of the main benefits of an independent central bank may be in ensuring an appropriate response to the type of inflationary shocks that hit the world economy in the 1970s.

In interpreting these results two points should be noted. First, with pegged exchange rates, as in the 1960s under the Bretton Woods system, central bank independence is necessarily limited, and inflation rates tend to converge to that of the anchor country. Second, demand by the public for the maintenance of reasonable price stability may in fact be the most important guarantee of low inflation, but an independent and accountable central bank can nevertheless make an important contribution to that objective.

The relationship is not so robust for the developing countries, as shown in the chart on the right. For these countries, the correlation is not statistically significant.5 This finding may reflect the fact that any index of central bank independence is necessarily imprecise. The indices used in the empirical literature assess the independence of a central bank on the basis of its founding legislation and institutional structure. There may be a large divergence between such a measure of independence and the degree of independence that the central bank has in practice.

It should also be noted that the simple correlation illustrated in the chart does not take into account all aspects of the interaction between the central bank and the government. In particular, the central bank’s ability to pursue a goal of low inflation may be severely curtailed if a large fiscal deficit requires monetary financing, even if it has a high degree of independence by many criteria. Increasing the independence of the central bank may be futile if there is not also fiscal reform. At the same time, a more independent central bank may exert discipline on the fiscal authorities.


Selected Developing Countries: Central Bank Independence and Inflation

The reform of the Reserve Bank of New Zealand suggests some guidelines on increasing central bank independence. The primary goal of the bank is to achieve price stability. This goal is not set by the bank itself, however, but by an agreement, the Policy Targets Agreement (PTA), between the finance minister and the governor of the bank. That is, the bank does not have goal independence. It does, however, have full instrument independence, as it has complete control over the determination of official interest rates. Moreover, the governor is directly accountable to the minister of finance for the achievement of the PTA. The performance of the governor is monitored by the board of the bank on behalf of the finance minister. The board in turn reports to Parliament. Thus, while maintaining a high degree of independence, the bank is also highly accountable for its actions.

While the New Zealand experience suggests that increased central bank independence facilitated the achievement of low inflation in recent years, a number of other industrial countries without increased central bank independence also experienced significant declines in inflation.6 Furthermore, increasing the independence of the central bank did not ensure a costless disinflation: New Zealand underwent a large recession during its disinflation. The increased independence of the bank, however, may have reduced the cost of the disinflation below what it would otherwise have been by increasing the credibility of monetary policy, and it is likely to assist in the maintenance of low inflation.


This distinction was introduced by Guy Debelle and Stanley Fischer, “How Independent Should a Central Bank Be?” Goals, Guidelines, and Constraints Facing Monetary Policymakers, Conference Series No. 38, ed. by Jeffrey C. Fuhrer (Boston, Massachusetts: Federal Reserve Bank of Boston, 1994)


In the United Kingdom, accountability is also enhanced by the publication of the minutes of policy meetings between the Governor of the Bank of England and the Chancellor of the Exchequer. The Chancellor is, of course, accountable to Parliament.


See, for example, Vittorio Grilli, Donato Masciandaro, and Guido Tabellini, “Political and Monetary Institutions and Public Financial Policies in the Industrial Countries,” Economic Policy: A European Forum, Vol. 6 (October 1991), pp. 342–91; Alex Cukierman, Central Bank Strategy, Credibility and Independence: Theory and Evidence (Cambridge, Massachusetts: MIT Press, 1992); Alberto Alesina and Lawrence Summers, “Central Bank Independence and Macroeconomic Performance: Some Comparitive Evidence,” Journal of Money, Credit and Banking, Vol. 25 (May 1993), pp. 1–14.


The index reflects institutions and legislation as of the late 1980s. Changes in some central bank statutes since then alter the index illustrated in the chart.


A negative correlation between inflation and central bank independence has been found for developing countries when the rate of turnover of the central bank governor is used as a proxy for independence. This may simply be capturing the effect of general political instability, however. See Alex Cukierman, Steven B. Webb, and Bilin Neypati, “Measuring the Independence of Central Banks and Its Effect on Policy Outcomes,” World Bank Economic Review, Vol. 6 (September 1992), pp. 353–98.


See Guy Debelle, “The Ends of Three Small Inflations: Australia, New Zealand and Canada,” Canadian Public Policy, Vol. 22 (March 1996), pp. 56–78.

Policies for Maintaining Price Stability

Inflation rates in industrial countries, as well as in a number of developing countries, have declined to their lowest levels in decades, and in a number of industrial countries are at rates close to, or approaching, reasonable price stability. With appropriately designed monetary and fiscal policies, it should be possible to keep inflation low and to reduce it further over time, while minimizing undesirable short-term fluctuations in output. The U.S. Federal Reserve in particular has emphasized the importance of preventing inflation from rising significantly during economic expansions while taking advantage of economic slowdowns to bring inflation down, and thus allowing a ratcheting down of inflation over the course of successive business cycles until price stability is achieved.114

The decline of inflation in recent years is partly the result of a changing global economy. But most important, it is the result of determined policy actions to bring inflation under control. Increased openness to trade has been a beneficial influence on inflation. Liberalization of trade restrictions typically results in lower prices of imported goods and thus provides competitive discipline for domestic firms. In countries with inefficient or monopolized industries, this can result in a lowering of price-cost markups.115 Globalization has the further beneficial effect that the introduction of advanced production techniques from foreign firms can lead to increased productivity and output growth, particularly in developing countries.116 However, these effects from openness and globalization are likely to provide one-time downward shifts in the price level rather than act as permanent restraints on the rate of inflation. Moreover, markets in many industrial countries are already quite competitive, so that added competition from foreign firms tends to have only limited effects on the pricing environment.117 Such forces as trade liberalization are thus unlikely in themselves to provide long-lasting changes in inflation, notwithstanding their beneficial effects on resource allocation and economic efficiency. Nonetheless, economies characterized by greater openness to trade have generally experienced lower rates of inflation; this suggests that institutional features, such as openness, that increase the costs of inflation can induce countries to favor policies that lead to lower inflation.118

What provides the greatest hope for the future, however, is that a large part of the recent decline in inflation can be attributed to the adoption of sustained noninflationary policies. Despite initial hesitation and policy mistakes, governments and central banks have applied the hard-learned lessons from the high-inflation decades of the 1970s and 1980s with increasing success. Foremost among these lessons is the importance of stopping inflation before it gains momentum. The global experience of the past thirty years has shown that the short-run costs in terms of unemployment and lost output of preemptive actions to arrest incipient rises in inflation are likely to be far smaller than the long-run costs of allowing inflation to rise to double or triple digits. Expectations of future inflation are important determinants of current inflation—thus, establishing credibility in the fight against inflation is crucial. A central bank that too readily trades its credibility for short-term gains in output will soon find that the short-term gains disappear, leaving only worse inflation to combat. In contrast, a central bank that firmly establishes and maintains its anti-inflationary reputation will likely garner a “credibility bonus,” so that the effectiveness of monetary policy in preventing undesirable variations in output will be enhanced. The challenge, of course, is to find the appropriate balance between determined action to stop inflation and policies that will ensure the stability of employment and output.

Monetary policy alone cannot and should not bear the burden of achieving and maintaining price stability. Structural policies, especially labor market policies, that reduce wage and price rigidities and enhance competition, and sound fiscal policies and low public debt levels are essential elements of a sustainable anti-inflationary policy strategy. In recent years, many countries—industrial, developing, and transition economies—have made substantial progress toward reducing excessive fiscal imbalances. But, as discussed in the May 1996 World Economic Outlook, budgetary imbalances remain a source of upward pressure on real interest rates in industrial countries and further progress toward fiscal consolidation is essential, especially in view of the budgetary pressures associated with aging populations. Similarly, in developing and transition countries, fiscal deficits are still too large in many cases and remain a threat to the sustainability of recent reductions in inflation.

* * *

Restoring a high degree of price stability has been a central objective worldwide. It has now been achieved or is within reach in most countries. Safeguarding this achievement is essential. It is first and foremost—albeit not exclusively—the responsibility of monetary authorities. While the rewards in terms of output growth in any one year may seem small, the cumulative effects on levels of output and living standards are potentially considerable.


See Michael M. Hutchison, “Aggregate Demand and Oil Prices: The 1990 Oil Shock in Comparative Perspective,” BIS Economic Papers, No. 31 (Basle: Bank for International Settlements, August 1991).


These differences in the responsiveness of real wages and the behavior of employment are analyzed in Charles Adams. Paul Fenton, and Flemming Larsen, “Differences in Employment Behavior Among Industrial Countries.” Staff Studies for the World Economic Outlook (IMF, July 1986), pp. l-50.


The strategy was a precursor to the guidelines set out in the Madrid Declaration of October 1994.


In particular, it appears in retrospect that the monetary policy response to the 1987 stock market crash may have been overdone.


George S. Alogoskoufis. “Monetary Accommodation. Exchange Rate Regimes and Inflation Persistence.”Economic Journal. Vol. 102 (May 1992). pp. 461–80: and Michael D. Bordo and Lars Jonung,“Monetary Regimes. Inflation and Monetary Reform.” in Inflation. Institutions, and Information; Essays in Honor of Axel Leijonhufvud. ed. by D.E. Vaz and K. Vellapilai (London: MacMillan. 1996).


See Atish R. Ghosh, Anne-Marie Guide, Jonathan D. Ostry, and Holger C. Wolf, “Does the Nominal Exchange Rate Regime Matter?” IMF Working Paper 95/121 (November 1995).


The implications of financial liberalization and asset-price in-flation for monetary policy are discussed in Annex I of the May 1993 World Economic Outlook, and in Garry J. Schinasi and Monica Hargraves, “‘Boom and Bust’ in Asset Markets in the 1980s: Causes and Consequences,”Staff Studies for the World Economic Outlook (IMF, December 1993), pp. 1–27.


This is the case provided the anchor currency does not become misaligned. If the anchor currency is misaligned, the information transmitted by the pegged exchange rate would not be useful.


Jeffrey D. Sachs and Felipe B. Larrain, Macroeconomics in the Global Economy (Englewood Cliffs, N.J.: Prentice Hall, 1993), p. 737.


See Vito Tanzi, “Inflation, Lags in Collection, and the Real Global Economy (Englewood Cliffs. N.J.: Prentice Halt, 1993), Value of Tax Revenue,”Staff Papers, IMF, Vol. 24 (March 1977), p. 737. pp. 154–67.


See Rudiger Dornbusch and Alejandro Reynoso, “Financial Factors in Economic Development,” in Policymaking in the Open Economy: Concepts and Case Studies in Economic Performance, ed. by Rudiger Dornbusch, EDI Series in Economic Development (Oxford; New York: Oxford University Press for the World Bank, 1993).


See Stanley Fischer and Lawrence H. Summers, “Should Governments Learn to Live with Inflation?” American Economic Review, Papers and Proceedings, Vol. 79 (May 1989), pp. 382–8


William E. Alexander, Tomás J.T. Baliño, Charles Enoch, and others, “The Adoption of Indirect Instruments of Monetary Policy,” IMF Occasional Paper No. 126 (IMF, June 1995).


Fischer, Sahay, and Vegh, “Stabilization and Growth.”


See Paula De Masi and Vincent Koen, “Relative Price Convergence in Russia,” Staff Papers, IMF, Vol. 43 (March 1996), pp. 97–122; and Sharmini Coorey, Mauro Mecagni, and Eric Offerdal, “Disinflation in Transition Economies: The Role of Relative Prices,” IMF Working Paper (forthcoming).


See, for example, C. A. Vegh, “Stopping High Inflation: An Analytical Overview,” Staff Papers, IMF, Vol. 39 (September 1992), pp. 626–95; L. Leiderman, Inflation and Disinflation: The Israeli Experiment (Chicago: University of Chicago Press, 1993); and Guillermo A. Calvo and Carlos A. Végh, “Inflation Stabilization and Nominal Anchors,” Contemporary Economic Policy, Vol. 12 (April 1994), pp. 35–45.


This operational definition follows W. Easterly, “When Is Stabilization Expansionary?” (unpublished; World Bank, December 1995). Based on the robustness tests reported by Easterly, most of the analysis that follows (and the empirical evidence) is not particularly sensitive to the specific definition of stabilization that is used.


The relevant cases are Austria (1921–22), Germany (1922–23), Hungary (1923), Poland (1923), Greece (1943–5), and Hungary (1945–46). Hyperinflations occurred also in China (1946–49) and Taiwan Province of China (1947–48).


For instance, while Calvo and Végh in “Inflation Stabilization and Nominal Anchors” find that output expands under exchange-rate-based programs and contracts under money-based programs. Easterly, for a larger sample of stabilization episodes, provides evidence that output expansion also holds for money-based stabilizations.


See Ratna Sahay and Carlos Végh.“Inflation and Stabilization in Transition Economies: A Comparison with Market Economies,” IMF Working Paper 95/8 (January 1995).


See Michael Bruno and William Easterly, “Inflation Crises and Long-Run Growth,” NBER Working Paper No. 5209 (Cambridge, Massachusetts: National Bureau of Economic Research, August 1995); and Rudiger Dornbusch and Stanley Fischer, “Moderate Inflation,” World Bank Economic Review, Vol. 7 (1993), pp. 1–44. This result by Bruno and Easterly is consistent with the finding of Dornbusch and Fischer that for most countries with moderate inflation rates in the range of 15 to 30 percent, there is no obvious tendency for inflation to ratchet upward.


See Phillip Swagel and Prakash Loungani, “Sources of Global Inflation,” IMF Working Paper (forthcoming). For an additional background study, see Swagel and Loungani. “Supply-Side Sources of lnflation: Evidence from OECD Countries,” International Finance Discussion Paper No. 515 (Washington: Board of Governors of the Federal Reserve System. April 1995).


Exchange rate fluctuations can be thought of as a supply shock in the sense that a depreciation of a nation’s currency increases the price of imported goods used as inputs to the production process. Of course, movements in exchange rates are influenced by the other five factors, since inflation would eventually be expected to lead to a depreciation of the currency. To take this into account, the results in Charts 40 and 41 show only the effect on inflation of a change in each of the influences after taking into account past movements in each of the six factors. For example, the effect of an exchange rate depreciation on inflation shown in Chart 41 represents the inflationary consequences of a movement in exchange rates that cannot be attributed to previous movements in oil or commodity prices, exchange rates, the output gap. monetary growth, or past inflation.


As noted above, this response of inflation to the output gap represents the effect only of movements in the output gap that cannot be attributed to changes in oil prices, money growth, or the other factors.


Robert P. Flood and Michael Mussa. “Issues Concerning Nominal Anchors for Monetary Policy.” in Frameworks for Monetary Stability: Polity Issues and Country Experiences. ed. by Tomás J.T. Baliño and Carlo Cotlarelli (IMF, 1994).


The first estimate is from Fischer (and the second from Feldsteint, Feldstein takes measured Inflation of 2 percent to represent price stability because measurement problems cause the CPI to overstate the true rate of inflation by about 2 percent. Measurement problems are discussed further below. See Stanley Fischer, “Towards an Understanding of the Costs of Inflation;II.” in The Costs and Consequences of Inflation, ed. by Karl Brunner and Allen H. Meltzer, Carnegie-Rochester Conference Series on Public Policy, Vol. 15 (New York; Amsterdam, North-Holland, 1981), pp. 5–42; and Martin Feldstein, “The Costs and Benefits of Going from Low Inflation to Price Stability.” NBER Working Paper No. 5469 (Cambridge, Massachusetts: National Bureau of Economic Research, February 1996).


See A. Steven Holland, “Comment on Inflation Regimes and the Sources of Inflation Uncertainty.” Journal of Money, Credit, and Banking, Vol. 25 (August 1992). Part 2. pp. 514–20. for a review of recent empirical studies.


Recent studies include Jose De Gregorio, “Effects of Inflation on Economic Growth: Lessons from Latin America,” European Economic Review, Vol. 36 (April 1992), pp. 417–25, and “Inflation, Taxation, and Long-Run Growth,” Journal of Monetary Economics, Vol. 31 (June 1993), pp. 271–98; Stanley Fischer, “The Role of Macroeconomic Factors in Growth,” Journal of Monetary Economics, Vol. 32 (December 1993), pp. 484–512; Robert Barro, “Inflation and Economic Growth,” Quarterly Bulletin, Bank of England, Vol. 35 (May 1995), pp. 166–76; and Michael Sarel, “Non-Linear Effects of Inflation on Economic Growth,” Staff Papers, IMF, Vol. 43 (March 1996), pp. 199–215.


On the former result see Fischer “Macroeconomic Factors,” on the latter see Sarel “Non-Linear Effects” and Barro “Inflation and Economic Growth.”


See Bruno and Easterly, “Inflation Crises and Long-Run Growth,” and Michael Bruno, “Does Inflation Really Lower Growth,” Finance Development, Vol. 32 (September 1995), pp. 35–38 for evidence on developing countries; and Fischer, Sahay, and Végh, “Stabilization and Growth” for evidence on transition countries.


For empirical evidence supporting the hypothesis that inflation in the Group of Seven countries is a function both of past and future expected inflation, see Bankim Chadha, Paul Masson, and Guy Meredith, “Models of Inflation and the Costs of Disinflation,” Staff Papers, IMF, Vol. 39 (June 1992), pp. 395–431.


See, for instance, Chadha, Masson, and Meredith “Models of Inflation” for the first result; and for the second see Laurence Ball, “What Determines the Sacrifice Ratio?” in Monetary Policy, N. Gregory Mankiw (Chicago: University of Chicago Press, 1994), and “Disinflation and the NAIRU,” NBER Working Paper No. 5520 (Cambridge, Massachusetts: National Bureau of Economic Research, March 1996).


See Douglas Laxton, Guy Meredith, and David Rose, “Asymmetric Effects of Economic Activity on Inflation: Evidence and Policy Implications, “Staff Papers, IMF, Vol. 42 (June 1995), pp. 344–74; and Dave Turner, “Speed Limit and Asymmetric Inflation Effects From the Output Gap in the Major Seven Economies,” OECD Economic Studies, No. 24, 1995/I.


The model on which these results are based is developed in Peter Clark and Douglas Laxton, “Phillips Curves, Phillips Lines and Unemployment Costs of Overheating,” IMF Working Paper (forthcoming).


In its latest annual report, the Bank for International Settlements notes that the symmetric view of the trade-off between inflation and output contrasts with that of most policymakers, who believe that there are large output costs associated with reversing inflationary forces (Bank for International Settlements, 66th Annual Report (Basle, 1996), pp. 27–28).


These points are developed in Lars E.O. Svensson, “Fixed Exchange Rates as a Means to Price Stability: What Have We Learned?” European Economic Review, Vol. 38 (April 1994), pp. 447–68.


This has been widely described in the literature. For a recent assessment of the United States’ experience see Benjamin Friedman, “The Rise and Fall of Money Growth Targets as Guidelines for U.S. Monetary Policy,” Bank of Japan, Institute for Monetary and Economic Studies, Discussion Paper 96-E-14 (March 1996). The Bundesbank’s conduct of monetary policy is analyzed in the Manfred J.M. Neumann, “Monetary Targeting in Germany,” Bank of Japan, Institute for Monetary and Economic Studies Discussion Paper 96-E-15 (March 1996); and Richard Clarida and Mark Gertler, “How the Bundesbank Conducts Monetary Policy,” NBER Working Paper No. 5881 (Cambridge, Massachusetts: National Bureau of Economic Research, May 1996).


Guy Debelle, “The Ends of Three Small Inflations: Australia, Canada, and New Zealand,” Canadian Public Policy, Vol. 22 (March 1996), pp. 56–78; and Richard T. Freeman and Jonathan L. Willis, “Targeting Inflation in the 1990s: Recent Challenges,” International Finance Discussion Papers No. 525 (Washington: Board of Governors of the Federal Reserve System, September 1995).


George A. Akerlof, William T. Dickens, and George L. Perry, “The Macroeconomics of Low Inflation,” paper prepared for the Brookings Panel on Economic Activity, March 28–29, 1996.


This argument is developed in Lawrence Summers, “How Should Long-Term Monetary Policy Be Determined?” Journal of Money, Credit, and Banking, August 1991, Part 2, pp. 625–31; and Bradford DeLong and Lawrence Summers, “Macroeconomic Policy and Long-Run Growth,” Economic Review, Federal Reserve Bank of Kansas City (Fourth Quarter 1992), pp. 5–29.


Measurement bias in price indices are surveyed in Mark A. Wynne and Fiona D. Sigalla, “A Survey of Measurement Biases in Price Indexes,” Journal of Economic Surveys, Vol. 10 (1996), pp. 55–89.


See Luc Soete, “New Technologies and Measuring the Real Economy: The Challenges Ahead,” paper prepared for the Central Statistics Institute (ISTAT) conference on Economic and Social Challenges in the 21st Century held in Bologna, February 5–7, 1996; and Nicholas Oulton, “Do UK Price Indexes Overstate Inflation?” National Institute Economic Review, No. 152 (May 1995), pp. 60–75. Soete believes that inflation in Europe might be overstated by as much as 2 to 3 percent a year; Oulton, however, argues that while the retail price index in the United Kingdom is likely biased upward, 2 percent is too high an estimate.


Peter Howitt “Zero Inflation as a Long-Term Target for Monetary Policy,” in Zero Inflation: The Goal of Price Stability, ed. by Richard G. Lipsey (Toronto: CD. Howe Institute, 1990); and Daniel Thornton, “The Costs and Benefits of Price Stability: An Assessment of Howitt’s Rule,” Review, Federal Reserve Bank of St. Louis, Vol. 78 (March/April 1996), pp. 23–38.


This approach to the conduct of monetary policy when inflation is low has become known as the opportunistic approach to disinflation—see Athanasios Orphanides and David W. Wilcox, “The Opportunistic Approach of Disinflation,” Board of Governors of the Federal Reserve System, Finance and Economics Discussion Series, Working Paper 96–24 (May 1996).


For evidence that trade liberalization in the Ivory Coast and Turkey led to increased productivity and lower price-cost margins, see Ann Harrison, “Productivity, Imperfect Competition and Trade Reform: Theory and Evidence,” Journal of International Economics, Vol. 36 (February 1994), pp. 53–73; and James Levinsohn, “Testing the Imports-as-Market-Discipline Hypothesis,” Journal of International Economics, Vol. 35 (August 1993), pp. 1–22.


Eduardo Borensztein, José De Gregorio, Jong-Wha Lee, “How Does Foreign Direct Investment Affect Economic Growth?” NBER Working Paper No. 5057 (Cambridge, Massachusetts: National Bureau of Economic Research, March 1995), show that foreign direct investment leads to increased growth in developing countries.


For the United States, for example, changes in import prices appear to have only small effects on domestic prices, implying that additional openness and foreign competitive pressure will have little effect on inflation. See Phillip Swagel, “Import Prices and the Competing Goods Effect,” International Finance Discussion Paper No. 508 (April 1995).


David Romer, “Openness and Inflation: Theory and Evidence,” Quarterly Journal of Economics, Vol. 108 (November 1993), pp. 869–903; and Philip R. Lane, “Inflation in Open Economies,” Journal of International Economics (forthcoming).

Annex I

The Difficult Art of Forecasting

The disappointing growth performance of some industrial and developing countries in 1995-96 and the successive downward revisions to the IMF staff’s growth estimates for several regions have highlighted the margins of uncertainty that attach to economic projections. In fact, these recent difficulties are not too different from the previous experience of Fund staff and other forecasters in predicting the paths and especially the turning points of business cycles. A recent study by Professor Michael Artis analyzing the Fund’s economic forecasting record from its inception in 1971 to 1994 provides some perspective on the accuracy of the World Economic Outlook projections for the industrial and developing countries.1

In principle, and like most official forecasts, those in the World Economic Outlook are formally presented as projections based on “unchanged policy” assumptions. Strictly maintaining such an assumption, however, is difficult because much of the market information available at any point in time (including such relevant indicators as interest rates, exchange rates, and business expectations) will reflect, inter alia, the anticipation that the values of policy variables may be changed in the future. For this reason, the general practice of treating unchanged policy projections as “unconditional” was followed in Artis’s study, which is summarized here.

On this basis, for the industrial country group as a whole, the average year-ahead absolute forecast errors indicate that World Economic Outlook forecasts have tended to over- or underpredict growth by about 1 percentage point, which is sizable when measured against the actual average absolute growth rate of 2¾ percent. Inflation has been over- or underpredicted on average by about ¾ of 1 percent a year, which is relatively smaller when measured against an actual average absolute inflation rate of 5¾ percent. Among the major industrial countries, growth rates in the United States and France appear to have been less difficult than the others to forecast (see Chart 46). For inflation, however, the forecasts for Germany appear to have the best record. Average forecast errors, however, represent only one of many criteria by which to judge the World Economic Outlook’s forecasting record. Errors generated by a good forecast procedure should also be unbiased and serially uncorrelated, and have no other property indicating that there is information in the data that could be used to improve the forecasts: this is to say that the forecasts should be efficient. Judged by these criteria, the World Economic Outlook industrial country forecasts of output growth are broadly satisfactory. There is little evidence of bias or serial correlation, and on the whole the forecasts are efficient. When the data for the industrial countries are pooled, some evidence of bias emerges, but this is primarily due to errors generated in the first half of the sample period. For inflation, the evidence is more mixed. Although the forecasts are generally unbiased and efficient, they appear to suffer from serial correlation, suggesting that it takes time for forecasters to recognize that the inflation environment has changed.

Chart 46.
Chart 46.

Industrial Country Real Output Growth: World Economic Outlook Forecast Compared with Actual1

(In percent)

1World Economic Outlook forecast is for the year ahead, and actual represents the first settled estimate available.

Forecasting output and inflation for the industrial countries has not become any easier over time. Artis analyzes two subsamples of the forecast errors (pre-and post-1983) and shows that there is not a great deal of difference in the accuracy of forecasts between the two periods relative to forecasts based on “naive” methods, such as the assumption of a random walk. Even though the first subsample includes the unusual economic disruptions associated with both of the major oil shocks, the environment of the second sub-sample did not prove any easier for forecasting. The study also confirms that the greatest area of weakness in forecasts for the industrial countries is predicting turning points in the business cycle. Thus during the most recent cycle, there were considerable difficulties in anticipating the strength of the upswing and subsequently the duration of the slowdown. For inflation, although a number of statistical tests reveal that the accuracy of the forecasts did not improve, there was a large decline in the average absolute value of errors, reflecting a similar decline in the actual average rate of inflation.

The World Economic Outlook forecasts for the major industrial countries, however, are not alone in accumulating rather large errors at turning points of the business cycle. In an earlier study, Artis conducted extensive comparisons between the World Economic Outlook forecasts and those produced by the Organization for Economic Cooperation and Development (OECD) and by individual national official forecasters. A notable finding was that the major forecasting errors were widely shared across all of these forecasters.2 A comparison with the private sector Consensus Forecasts in the recent study reveals that over the last business cycle (1990-94), the forecasting errors were generally about the same.3 A scatter plot of the Consensus Forecasts and World Economic Outlook forecast errors of output growth for the major industrial countries is shown in the top panel of Chart 47. Most of the observations fall on or close to the 45-degree line, suggesting that the two forecast error records are similar. Moreover, a large number of the errors fall in the upper right quadrant of the chart indicating that both forecasts tended to overestimate growth. The World Economic Outlook forecasts, however, may be seen to have tended to be slightly more optimistic than the Consensus Forecast. A similar scatter plot for the inflation forecasts (bottom panel of Chart 47) shows little overall difference between the two sets of forecast errors.

Chart 47.
Chart 47.

Major Industrial Countries: Comparison of World Economic Outlook and Consensus Forecasts Errors1

Sources: World Economic Outlook; and Consensus Economics Inc., Consensus Forecasts.1 Forecast errors are for the period 1990-94 and are defined as year-ahead forecast value minus actual realized value. Each observation “shows” the Consensus Forecasts and World Economic Outlook forecast errors for one of the seven major industrial countries for forecasts constructed at approximately the same time.

For the developing country group, the task of forecasting movements in economic activity is even more difficult. In many countries, the data on which the forecasts are based are poor in quality and lack timeliness. In addition, many of these economies experience relatively greater volatility–especially those suffering from acute domestic or external imbalances and subject to major shifts in sentiment of domestic and foreign investors, and also those sensitive to commodity price fluctuations. Even though an analysis of the aggregate performance of the developing countries should reduce the influence of these factors, the forecast performance is considerably less satisfactory than for the industrial countries.

The average forecast errors for the period 1977–94 for output growth and inflation differ considerably across regions of the developing world but are relatively large in comparison with their average absolute actual values. The average forecast errors for output growth and inflation are particularly large in the Middle East and Europe region of developing countries, where uncertainty in oil markets and events such as the Persian Gulf war complicate forecasting efforts, and in the Western Hemisphere where output and inflation have tended to be characterized by greater volatility, with developments in a few large countries often dominating developments. It is important to note, however, that for the developing countries, a significant source of the forecast errors is the fact that projections contained in the World Economic Outlook are consistent with those underlying IMF-supported policy programs, which are naturally based on the assumption that these programs will be implemented and successful.

Additional analysis of these forecast errors reveals that for output growth, there is evidence of positive bias for both Africa and the Western Hemisphere. In contrast, for inflation there is evidence of a negative bias in most regions of the developing world, and serial correlation in Africa and the Western Hemisphere. Overall, by comparison with the results for the industrial countries, these results suggest that it is much more difficult to forecast both output and inflation for the developing countries, and that there has been some bias in these forecasts historically.

* * *

The World Economic Outlook forecasting record for both the industrial and developing countries suffers from a variety of weaknesses, and evidence suggests that the task of forecasting has not become easier over the sample period. The cumulative experience of forecasting, the contribution of significant advances in data processing to the timeliness of forecasts, and competition offered by the large increase in the availability of economic forecasts around the world, are all factors that might have helped improve forecast accuracy. At the same time, however, the structure of the world economy has been changing rapidly, constantly introducing new and sometimes unpredictable sources of error that continue to complicate the task of forecasting.

Artis concludes that the recent experiences in predicting cyclical turning points have been especially disappointing, although the record obviously needs to be assessed against the difficulty of the task. The oil price shocks in 1973 and 1979 and their subsequent impact on the world economy provided the driving force behind business cycle developments in the 1970s and early 1980s, and it may seem reasonable to excuse forecasters for not having foreseen those price increases. The period since the mid-1980s has been quite different, with supply shocks not such a dominant feature. Notwithstanding “exogenous” events such as the drop in oil prices in the mid-1980s, the Persian Gulf war, the unification of Germany, and the collapse of central planning, the prolonged global upswing of the mid- to late 1980s and the subsequent slowdown appear to have been endogenous in a way that provides less obvious justification for forecast errors. If the judgment is right that the recent cycle was largely endogenous to the natural momentum of a world economy more integrated and substantially less regulated than before, forecasters will do well to learn from it: it is too early to say how successful this learning process will be.

Annex II World Oil Market: Recent Developments and Outlook

Throughout the twentieth century, but particularly since the mid-1970s, much of the discussion of developments in primary commodity markets has focused on the demand, supply, and price of oil. Despite some erosion in its share in total energy use, oil continues to provide the critical energy ingredient for almost all modes of transportation involving machines.1 It is a major fuel source in heating. In manufacturing, oil is not only frequently the main source of energy but also in many cases an important raw material ingredient. Furthermore, at least in recent decades, the value of international trade in oil has been almost as great as that of all nonfuel primary commodities combined. This annex provides a summary of recent price developments and underlying movements in oil demand and supply.2

Crude Oil Prices

The price of oil in recent years has continued to be highly volatile over short periods.3 It has been relatively stable, however, when compared with the period from the “first oil shock” in 1973 through the Iraq-Kuwait conflict in 1990–91 (Chart 48). Nevertheless, a recent unexpected climb in the oil price—from about $16 a barrel in October 1995 to about $20.50 in April 1996—attracted attention in financial markets and among policymakers in many oil importing countries. The climb was associated with strong demand for heating oil as a result of the prolonged and unusually cold winter in North America and Europe, and it was exacerbated by the fact that inventories had been reduced to low levels because the oil industry was anticipating lower prices in the near future partly as a result of the expected return of Iraq to the export market (Chart 49). Throughout the final quarter of 1995 and the first half of 1996 more distant positions on oil futures markets have been trading at substantial discounts to spot prices and to near positions (Chart 50).

Chart 48.
Chart 48.

Crude Petroleum Price in Current and Real Terms

(In U.S. dollars a barrel)

1 Current U.S. dollar price deflated by the index of U.S. dollar unit values of manufactured goods exports (1990 = 100).2 Average of spot prices for U.K. Brent, Dubai, and West Texas Intermediate crude petroleum.
Chart 49.
Chart 49.

Commercial Stocks of Petroleum

(In millions of barrels)

1 Middle distillates and residual fuel oil. that is. products for which a large part is used as heating fuel.
Chart 50.
Chart 50.

West Texas Oil Prices: Spot and Futures Contracts1

(In U.S. dollars a barrel)

1 Spot prices are monthly averages.


World demand for oil is still largely determined by developments in the industrial countries, since they still consume the greater part of the world’s oil production. Their share has declined significantly since the early 1970s, however, from about 72 percent to about 58 percent in 1995, as the share of developing countries—especially the rapidly growing Asian economies—has risen (Table 38). Since the mid-1980s, the industrial countries’ share of world consumption has remained broadly unchanged as declining consumption in the economies in transition has largely offset the continuing rise in the share of the developing countries.

Table 38.

Consumption of Petroleum by Groups of Countries

article image
Source: International Energy Agency, Monthly Oil Market Report.

Projections based on Shane S. Streifel, Review and Outlook for the World Oil Market. World Bank Discussion Paper 301 (Washington: World Bank, 1995), as revised in 1996.

Not including Japan, transition economies in central Asia and countries in the Middle East.

Industrial Countries

Oil consumption in industrial countries in recent years has grown by about 1 percent a year. There is a marked seasonal pattern in the consumption of two of the major groups of oil products (Chart 51). The seasonal pattern is especially marked for the consumption of the oil products used partly for heating (diesel, other gas oil, residual fuel oil, and liquid petroleum gas (LPG)). For these products, consumption is highest in the first quarter of the year and lowest in the second and third quarters. In the fourth quarter of 1995 and the first quarter of 1996, the demand for oil products used partly for heating was particularly strong in Europe and North America because of the unusually cold weather. A seasonal pattern is also evident in the consumption of motor gasoline, which is greatest in the third quarter of the year and lowest in the first quarter. The total demand for oil is smoothed somewhat over the year by the partly offsetting seasonal patterns in the consumption of these two groups of products. For the industrial countries as a group, overall consumption is highest in the first quarter of the year and lowest in the second quarter. In the United States, however, the overall seasonal pattern is muted, with low consumption in the second quarter being the only reliable regularity. This is a consequence of the considerably higher proportion of motor gasoline in total oil consumption than in most other industrial countries.

Chart 51.
Chart 51.

Industrial Countries: Consumption of Petroleum

(In millions of barrels a day)

1 Diesel, other gas oil, residuel fuel oil, and liquid petroleum gas (LPG), that is, products for which a large part is used as heating fuel.

Economies in Transition

Consumption of oil in the economies in transition was 6.5 percent lower in 1990 than in 1980, largely because of substitution by natural gas. Consumption then fell sharply in the early 1990s with the decline and restructuring of economic activity and the reduction of oil subsidies. The consumption decline was especially marked in the Baltic states, Russia, and other countries of the former Soviet Union, amounting to nearly 45 percent between 1990 and 1994–95, when it appeared to bottom out. In the economies of central and eastern Europe, a decline of nearly 20 percent occurred in 1991, but consumption seems subsequently to have bottomed out, with some sign of an upturn beginning in 1995. In general, consumption of motor gasoline has been buoyant relative to other oil products, with a relatively small decline in the Baltic states, Russia, and other countries of the former Soviet Union and no decline in central and eastern Europe. In contrast, the declines in fuel oil and gas oil have been large, reflecting lower economic activity including electricity generation, and also reduced allocations for household heating. The slump in consumption in the transition economies, in part matched by lower oil production, also reflects to some degree an improvement in the efficiency of energy use in response to the shift to market prices.

Asian Developing Countries

In the 1970s, China and other developing countries in Asia accounted for little more than 7 percent of world consumption of oil; by 1995 they accounted for more than 16 percent. China’s consumption more than doubled over this period, but consumption in the other Asian countries has more than quadrupled. This difference in growth occurred mainly in the years immediately following the 1979 oil price shock, when Chinese consumption was relatively stable while that of other Asian countries continued to grow, albeit at a lower rate than previously. Demand for fuel oil has been strong, both in China and other Asian countries, particularly because of rising demand for electricity. Although motor gasoline consumption has grown, it has remained below 10 percent of total consumption, with a share smaller than in other regions in the world.

Other Developing Countries

The pattern of oil consumption in other developing countries has varied from region to region, in part reflecting differences in the importance of oil exports in generating revenue and economic growth. In the countries of the Middle East, oil consumption grew rapidly in the 1970s and the early 1990s as a result of the economic activity generated by increased earnings from exports of oil. Since 1990, however, the increase in consumption has been relatively small. The pattern is similar in Africa, with little growth in consumption from 1990 to 1995. In the developing countries of the Western Hemisphere, growth in oil consumption has been more steady although there was a marked slowdown in the early 1980s associated with the impact of the 1979 oil price shock, the debt crisis, and other economic and political difficulties. In addition, fuel oil demand was reduced by increased use of hydro and natural gas in power generation.


Global oil production stood at 66.9 million barrels a day (mbd) in 1990, increased to 70 mbd in 1995, and is projected to reach about 72.8 mbd in 1996. Although total crude oil production increased by 3.1 mbd between 1990 and 1995, the evolution of supply has been uneven and major changes have occurred in the production of different countries and regions. The following sections discuss the evolution of crude oil reserves and oil technology, policies affecting crude oil supplies, and the role of the Organization of Petroleum Exporting Countries (OPEC) and non-OPEC producers in supplying the market as a whole.

Oil Reserves

Crude oil reserves form the geological basis for the development of oil fields and for the extraction of oil. Contrary to popular perceptions, global oil reserves have not been declining over time. While in some countries extraction rates have fallen and reserve-to-output ratios have declined, in others there have been substantial net additions to reserves through new discoveries, and increases in projected recovery rates from existing oil fields. On balance, new discoveries and improving recovery from known oil reservoirs have increased oil reserves substantially over time.

The size of crude oil reserves depends upon oil technology and on development and production costs. A frequently used measure is “proven reserves,” defined by the American Oil Institute as the volume of crude oil that could be recovered from oil fields in the future under existing economic and operating conditions. Over time, there has been a dramatic increase in the volume of proven reserves (Chart 52). In 1950, it was estimated that cumulative production had been about 60 billion barrels, while proven reserves were estimated to be about 90 billion barrels. With the extraction rate at 4 billion barrels a year, there were thus proven reserves sufficient for 22 years’ production. By 1980, however, proven reserves were estimated at 648 billion barrels and by 1990 they had grown to 900 billion barrels. By 1993, cumulative production was estimated at 640 billion barrels, while proven reserves had risen to 1,000 billion barrels. According to industry sources, over the last twenty years 1.7 barrels have been added to proven reserves for every barrel of oil consumed.

Chart 52.
Chart 52.

Global Oil Reserves and Consumption

(In billions of barrels)

Sources: International Petroleum Encyclopedia: and Twentieth Century Petroleum Statistics.

While investment in exploration and development has been successful in replacing and adding to proven reserves as oil has been extracted, proven reserves are not an indicator of the more general concept of oil resource availability. A broader measure of “ultimate recovery” has been put forward by the U.S. Geological Survey, which in 1994 put oil left and that potentially possible to produce economically at between 1.4 and 2.1 trillion barrels, sufficient at current consumption rates to last over 65 years. Based on an assumed recovery rate of 35 percent, unrecoverable oil would be estimated at 4 to 5.4 trillion barrels. If technological change could increase the rate of recovery still further, this would add to the general measure of oil resource availability.

Whatever the estimates of general oil resource availability, which are inherently speculative, it is estimates of potential yields from both new discoveries and known oil fields that yield additions to crude oil reserves that can be exploited over time. While reserves in the Middle East accounted for about 65 percent of the global total in 1990, there have been additions to reserves in many countries (Chart 53). The North Sea and Latin America are regions where reserves have expanded considerably in recent years and, as discussed further below, these additions to reserves have had a powerful impact on global oil supply.

Chart 53.
Chart 53.

Distribution of Crude Oil Reserves in 1990

(In billions of barrels)

Source: International Petroleum Encyclopedia.1 FSU represents the Baltic states, Russia, and other countries of the former Soviet Union.


When technological changes in the oil industry and learning about oil fields take place, the volume of reserves and future production prospects can be significantly affected. Not all oil in a reservoir can be recovered economically: recovery rates in the past have typically been in the range of 25–35 percent. A continuous challenge in the oil industry has been to improve economic recovery rates and to stabilize the projected decline in extraction rates in mature oil fields. Several methods have emerged including water injection and, where natural gas deposits are available, gas injection. In the 1990s, developments enhancing oil recovery have included horizontal well drilling, steerable drills, use of smaller pipe casings, advanced three-dimensional seismic modeling, and oil field behavior studies. These techniques have increased productivity by 10–20 percent, lengthening the productive life of many oil fields and helping maintain extraction rates. On the frontiers of technological development are air injection methods and very deep offshore drilling, the latter having been used successfully in the North Atlantic, the Gulf of Mexico, and off the coast of west Africa.

In addition, hydrocarbon resources other than crude oil—principally extra heavy oil, bitumen, and shale oil—have become increasingly economically viable. Estimates put known resources of such oil-yielding hydrocarbons at 15 trillion barrels in 1994, substantially exceeding resource estimates for conventional crude. Technological progress has been important in lowering the costs of oil production from these sources. An example is crude oil recovered from tar sands, or synthetic crude, which is produced in Canada and which has become much more competitive with conventionally extracted oil as production costs have been lowered. Canadian tar sands-based oil production was about 400,000 barrels a day in 1995 and is projected to increase over the next few years as development proceeds. Further technological improvement or higher real oil prices would probably lead to exploitation on a larger scale. In recent decades, discovery, technological improvement, and increasing knowledge of oil fields have therefore not only fostered crude oil production but also expanded the global oil resource base.

Supply-Side Policies

Policies of producing countries relating to the supply of crude oil significantly affect the flow of oil to the world market. In many countries, government policies set the broad parameters (e.g., the licensing procedures and the fiscal regime) within which oil companies determine the flow of oil to the market. While the quest for efficiency has fostered vertical integration by international oil companies, independent producers have continued to thrive: together, these companies have led the search for ways of reducing the costs of development and extraction. In a number of producing countries, however, decisions about supply are made by governments themselves through the operation of large state-owned oil companies.4 Some governments, which together control a large market share, have cooperated within OPEC to manage supply in an attempt to support oil prices and stabilize government revenues (see further below). With a powerful need to generate revenues for governments, there may at times, particularly when real oil prices are low, be a conflict with the need to retain earnings to finance oil field maintenance and development expenditures efficiently. Data on the operations of many state-owned oil companies are sparse, but it appears that a number have sought to emulate private oil companies through vertical integration and the use of enhanced recovery techniques, although some have been held back by lack of access to the latest technologies and information.

Fiscal regimes have a significant impact on oil development and production in a number of countries. High fiscal levies can act as a disincentive to the development of new oil fields and to further investment in fields already being exploited. Recently, reforms to the tax regimes facing North Sea oil producers have been a factor aiding the increase in supplies from this region. Norway has reformed its “sliding scale” investment policy that allowed the government to take a portion of the equity in any oil project in its zone. And two years ago, Norway eliminated the “sliding scale” on small oil fields, on future discoveries in the Norwegian Sea, and on the development of reserves in the Barents Sea. In the United Kingdom, the rate of tax on corporate income (33 percent) is relatively low by international standards, while the oil revenue tax was abolished for new fields in 1993, when its rate on existing fields was lowered from 75 percent to 50 percent. These fiscal reforms have encouraged significant oil development and production gains in the North Sea, and in 1995 it was one of the largest oil producing regions outside the Middle East.

OPEC Production

Despite having very large reserves and considerable potential to develop large known oil fields, the share of OPEC in the world oil market has changed little since 1990. The volume of crude oil supplied has increased as consumption has grown, from 23.0 mbd in 1990 to 25.5 mbd in 1995 and an estimated 27.0 mbd in 1996 (Table 39).

Table 39.

Crude Oil Production

(In millions of barrels a day)

article image
Sources: International Energy Agency (IEA), Monthly Oil Market Report; and IMF staff estimates for 1996.

D. Knapp, “Non-OPEC Supply Countinues to Grow,” Oil & Gas Journal. December 15, 1995: IEA, World Energy Outlook 1996: and IMF staff estimates.

2 FSU = the Baltic states, Russia, and other countries of the former Soviet Union.

OPEC production ceilings have been changed several times since 1990 as members have sought to support oil prices. In the first half of 1990, the ceiling was 22.1 mbd; it was adjusted upward to 22.5 mbd in the second half of the year. With the start of the Iraq-Kuwait hostilities in August, however, about 4.5 mbd of crude oil was taken off the market and the production ceiling and quotas were suspended. Other member countries with excess capacity, notably Saudi Arabia, increased production quickly. The mechanism was restored in February 1992 as oil prices showed signs of softness due to slackening oil demand in industrial countries. The ceiling was raised from 21.5 mbd to 24.6 mbd in November 1992 to accommodate Kuwait’s output as its oil sector recovered from damage incurred during Iraq’s occupation. Continued price softness led to a lowering of the ceiling to 23.6 mbd in March 1993, but the ceiling was raised again to 24.5 mbd in October 1993 where it remained unchanged until June 1996. At that time, the ceiling was adjusted upward to 25.03 mbd, a change made to include Iraq’s possible return to the market under the “oil for food and humanitarian supplies” exemption to UN sanctions. But, although OPEC has managed so far in the 1990s to achieve its broad goal of influencing oil prices through supply management, there has been a tendency for actual output to exceed target output, which together with increasing supplies from nonmember countries has somewhat diluted the group’s market power. In addition, two smaller producing countries have left the organization, Ecuador in 1993 and Gabon in 1996.

There have been several other important changes in policies adopted by OPEC members that have influenced the role their hydrocarbon sectors play in their economies. Since 1992, several member countries of OPEC, led by Algeria, have reversed long-standing policies by actively seeking foreign partners in the upstream part of their oil operations. Production-sharing agreements have provided the main incentive for foreign oil companies to engage in upstream participation and more recently, in several cases, to encourage the transfer of technology for the further development of replacement reserves. Another change is that oil companies in several OPEC countries have sought to become more vertically integrated and have acquired international refinery, distribution, and marketing assets. In addition, some member countries of OPEC have increased condensate output, at least in part because condensates are exempt from production quotas and yield needed government revenues.5 Condensate production by OPEC members is estimated to have increased from 2.0 mbd in 1990 to 2.6 mbd in 1996. Currently, several OPEC members have made significant investments in natural gas production and are set to produce higher volumes of condensates over the medium term.

Non-OPEC Production

Crude oil supply from outside OPEC has increased from 41.9 mbd in 1990 to an estimated 43.8 mbd in 1996. Behind this seemingly modest performance, however, lie declines in production in two major areas, the Baltic states, Russia, and other countries of the former Soviet Union and the United States. Crude oil output in countries of the former Soviet Union declined dramatically from 11.5 mbd in 1990 to 7.5 mbd in 1995, a drop of 37 percent. In the same period, crude oil output in the United States fell by 4 percent to 8.6 mbd due to lack of major discoveries and the maturation of the existing North Slope fields in Alaska. Crude oil supplies from other non-OPEC sources, however, particularly from the North Sea and Latin American regions, increased and more than offset the declines in the countries of the former Soviet Union and the United States; production in those regions rose from 30.4 mbd in 1990 to an estimated 36.6 mbd in 1996. The principal factors behind this surge are improved technology, which has lowered development and extraction costs; fiscal reforms in the United Kingdom and Norway; and privatization and upstream foreign investment in Latin America.

Medium-Term Outlook

World consumption of oil in the period 1995–2000 is expected to grow by about 2 percent a year compared with the growth of 1 percent a year in 1990–95 (Table 40). The higher growth anticipated in the late 1990s is mainly the consequence of economic recovery in the transition countries together with the continuing rapid growth of consumption in the Asian countries. At the same time, other sources of energy, especially natural gas, are expected to continue replacing oil in certain uses while improvements in energy efficiency should also continue.6

Table 40.

Average Growth Rates in Consumption of Petroleum, by Groups of Countries

article image
Source: Table 38.

Projections based on Shane S. Streifel, Review and Outlook for the World Oil Market, World Bank Discussion Paper 301 (Washington: World Bank, 1995), as revised in 1996.

FSU = the Baltic states, Russia, and other countries of the former Soviet Union.

Not including Japan, transition economies in central Asia, and countries in the Middle East.

In the industrial countries, the rate of consumption growth in the period 1995–2000 is expected to be broadly the same as in the period 1990–95—about 1 percent a year. Oil’s share in total energy usage is expected to erode as the share of natural gas continues to rise. Growth in demand for motor gasoline is expected to wane as a result of slower growth of “driving age” populations, particularly in Europe and Japan, and improvements in energy efficiency. The quantitative impact of the spread in the use of alternative fuels and zero emission vehicles is likely to be small in the period immediately ahead but could become much more pronounced in the first decade of the next century.7 In the transition economies of central and eastern Europe, oil consumption related to transportation is expected to grow rapidly in the late 1990s while most of the renewed demand for home heating is expected to be met by natural gas. Movement in the same direction should occur in the Baltic states, Russia, and other countries of the former Soviet Union, but the magnitude of the changes is at this stage very uncertain. Growth in oil consumption in China and other developing countries in Asia is expected to be high—about 5 percent a year. This rate of growth is projected to be lower than that recorded in recent years because of improvements in energy efficiency and the maturation of the demand pattern in some countries such as Korea. Oil consumption in other developing countries is likely to be held in check by higher end-use prices, efficiency improvements, and increased shares for natural gas.

On the production side, the medium-term outlook is promising as crude oil reserves are presently adequate, and the potential to develop substantial proven reserves from known oil fields clearly exists. Outside OPEC, output in the Baltic states, Russia, and other countries of the former Soviet Union is expected to recover as much needed investment takes place in the oil sector, while in other non-OPEC countries further favorable developments are likely in the medium term as foreign private oil companies participate in equity and other partnership agreements in countries where exploration and development have increasingly been encouraged. Within OPEC itself, there is also considerable potential to increase production capacity, especially in some of the countries where abundant undeveloped reserves are known to exist. Nevertheless, with an unchanged policy stance to support oil prices in a range close to those prevailing over the last several years and given projected growth in global demand, OPEC may not match the expected production performance of non-OPEC producers. The situation is clouded, however, by the uncertainties surrounding oil production by Iraq, which has been absent from the market since 1990. If Iraq were to return to full production status over the medium term, OPEC output would increase by about 3 mbd. Whether other OPEC members would adjust their production to offset Iraqi supply to the market is an important uncertainty.

Thus, while the oil market is prone to shocks, particularly on the supply side due to political developments, on balance, the medium-term outlook for oil prices is that there is little reason to expect a significant shift in real terms. A high degree of price volatility is likely to continue, however, especially if oil inventories remain at low levels.

Annex III World Current Account Discrepancy

The sum of current account balances (covering countries and international organizations) should in principle be zero for the world as a whole. But this is not true in practice for a variety of reasons relating to differences in the recording of current account transactions among countries with respect to valuation, coverage, timing, and classification. The size of the statistical discrepancy in the global current account has, at times, been a cause of concern to the Fund in connection with its analytical implications for the World Economic Outlook project and the Fund’s multilateral surveillance activities. Following the emergence of large statistical discrepancies in the world current account statistics in the early 1980s, the Fund organized a Working Party to investigate the principal sources of discrepancy in these statistics and to recommend procedures to improve statistical practices.1 The global current accounts display a persistent excess of deficits over surpluses, which exceeded $100 billion in 1982, part of the period studied by the Working Party. The discrepancy narrowed in the 1980s but again reached the $100 billion mark in the early 1990s before declining to about $80 billion in 1993–94 (Chart 54). In 1982, the discrepancy was equivalent to 2 percent of the sum of world current account credits and debits; in 1994, it represented about of 1 percent. The Fund’s Executive Board authorized a second Working Party to do a follow-up study of global balance of payments statistics in December 1989, but this time to evaluate statistical practices relating to the measurement of international capital flows.2

Chart 54.
Chart 54.

Global Discrepancies on World Current Account1 and on World Financial Flows2

(In billions of U.S. dollars)

Source: Compiled from selected IMF publications.1 Excludes capital transfers from 1988.2 Adjusted to exclude reported reinvested earnings. Includes IMF staff estimates for earlier years.

The principal focus of the 1987 current account study was on investment income transactions, where the imbalances between recorded credits and debits had shown an appreciable increase in the early 1980s. Since that time, the discrepancy with respect to investment income other than from direct investment has steadily increased and has become the principal contributor to the overall discrepancy in the world current accounts. It reached $126 billion in 1994, equivalent to about 8 percent of recorded income credits and debits. As the developing countries accounted for only 10 percent of the gross income flows (excluding income from direct investment) recorded in 1994, the large discrepancy on the income account is presumably related to recording difficulties in the industrial countries. The discrepancies in the statistics on other components of the world current account, which are discussed later in this note, are presented in Table 41 for the 1990–94 period.3

Table 41.

Discrepancies on World Current Account

(In billions of U.S. dollars)

article image
Source: IMF, Balance of Payments Statistics Yearbook, 1995.

Excludes income from direct investment.

The current account study concluded that the most important factor in the growth of the discrepancies in the global investment income account in the early 1980s was the emergence of a large body of cross-border assets recognized by the debtor countries but not by the creditors, coupled with higher interest rates after 1979. The study found that the reported data on capital flows for the 1977–83 period showed a cumulative net inflow of nearly $300 billion (apart from reinvested earnings), indicating that countries receiving capital were in a better position to measure the flows than the countries where the creditors resided and were thus better able to record the related investment income flows. This bias in the recording of financial flows continues today; at the world level, recorded inflows (apart from reinvested earnings) now exceed outflows by an average of $150 billion a year (Chart 54), which is largely related to the recording of portfolio and other investment transactions other than direct investment. The magnitude of the net discrepancies suggested by these figures may understate the extent of the problem in the recorded financial flow figures because some positive errors cancel out negative ones and some transactions are likely to have been missed on both the credit and debit sides of the accounts. The cumulated surplus or excess of recorded financial inflows registered over the 1977–94 period totaled $1.5 trillion. This excess of recorded inflows (liabilities) is consistent with the sign of the discrepancy on income flows (excess debits). Although it is possible that inflows may be overstated, the real problem is believed to be difficulties in accurately measuring outflows.

By using the international banking statistics compiled by the Bank for International Settlements (BIS),4 the two IMF Working Parties determined that a substantial part of the global discrepancies in the data on investment income and nonbank capital flows was attributable to cross-border activities with nonresident banks. The Working Parties recommended that national compilers should systematically compare their national data on nonbank capital flows (and associated stocks of foreign assets and liabilities) with BIS banking statistics, and where appropriate, use these statistics in compiling balance of payments estimates.

At present, a small number of countries use the BIS international banking statistics to compile balance of payments estimates. To improve data coverage, the U.S. Department of Commerce began in the early 1990s to substitute BIS counterpart banking data for U.S. source data on nonbank financial claims and liabilities to nonresident banks. These data substitutions led to an upward revision of some $200 billion in U.S. international investment position statistics for nonbank claims on and liabilities to nonresident banks at the end of 1993, with net assets increasing by approximately $20 billion. Revisions to the U.S. investment income accounts arising from these new data amounted to as much as $11 billion in 1993. The revisions did not have a significant impact on the global discrepancy on income transactions, as both income credits and debits were revised upward. For many other countries, the measurement of foreign assets (and hence income credits) may be more problematic than foreign liabilities (and income debits).

The Fund’s Statistics Department and its Committee on Balance of Payments Statistics are working with the BIS to refine the international banking statistics for use in balance of payments compilation.5 Efforts are also being made to disseminate the international banking statistics more widely to national compilers and to provide them with methodological information to assist in the use of these statistics for compiling or checking balance of payments estimates, or both. A second initiative that may yield useful information to improve national balance of payments data is the 1997 Coordinated Portfolio Investment Survey, which, like the international banking statistics exercise, will generate extensive stock data on cross-border claims with a comprehensive geographical breakdown. The Coordinated Survey, which is being conducted under the auspices of the Fund, will entail the collection of information on cross-border holdings of foreign stocks and bonds as at the end of 1997 by some forty countries that have indicated to the Fund their intent to participate in this international undertaking. In the early 1990s, a large positive discrepancy (i.e., credits exceed debits) emerged on portfolio investment transactions, which averaged $100 billion a year.6 As the vast majority of cross-border security transactions are conducted by the industrial countries, a large part of this imbalance likely emanates from this group of countries. With the globalization of financial markets, balance of payments compilers have encountered difficulties in measuring the growing volume of cross-border activity in securities, which may bypass domestic financial intermediaries, the usual source of data for these statistics. The development of comprehensive stock estimates on banking and portfolio investment may assist national compilers to improve their current account estimates for investment income.

In recent years, the merchandise trade component emerged with the second-largest discrepancy in the global current account, rising from an average of less than $20 billion a year in the 1980s to $76 billion in 1994, about 1 percent of the gross trade measures. The global trade data show an excess of recorded credits, unlike the other main components of the global current account in Table 41, which show higher debits. The introduction of a new trade reporting system (INTRASTAT) in the European Union countries in January 1993 as a consequence of the development of the internal market in Europe may be a factor contributing to the recent rise in imbalances on trade flows. The published data on intra-European Union trade show an excess of credits of some $40 billion a year in 1993–94, which is believed to reflect an underestimation of imports.

Historically, recorded debits from international services transactions have exceeded corresponding credits in global aggregations, although, in recent years, the imbalances have been relatively small. Within the components that comprise the services account, transportation transactions (mainly freight and other distributive services associated with the carriage of international trade) have consistently shown the largest imbalances, with debits exceeding credits by some $50 billion a year in the 1990–94 period. The 1987 current account study determined that while all countries are able to compile fairly readily the amount paid to foreign-operated carriers to cover the freight on imports, several economies with large maritime interests (notably those of Greece, Hong Kong, and eastern Europe) do not report the corresponding freight earnings of their fleets. Some important countries with fleets under national registry do not view the fleets as part of the domestic economy and therefore exclude them from the balance of payments.

Another factor that may contribute to discrepancies in the transportation account is the ability of compilers to correctly estimate the value of freight (and insurance) included in merchandise import figures compiled on a c.i.f. basis, which in the balance of payments needs to be reallocated to the services account. The global data for all other international services transactions (e.g., travel, government, communications, and so forth) taken as a group show an excess of recorded credits, which has increased considerably in recent years. As more countries implement the classification of the fifth edition of the IMF’s Balance of Payments Manual, which contains a detailed classification of international services, the Fund will be better able to isolate the nature of these discrepancies. The Fund’s Balance of Payments Statistics Committee has also undertaken some preliminary analysis of bilateral discrepancies in international travel transactions and in selected categories of business services.

Lastly, the global balance of payments data on current transfers also show a persistent excess of recorded debits over credits, which has increased considerably in the 1990s. The net discrepancy amounted to $50 billion in 1994, which was more than 8 percent of the gross value of the recorded debits and credits. The imbalance likely represents an excess of recorded debits over credits for government transfers, such as official aid disbursements and pensions. The reason for this discrepancy is not apparent but may relate, in part, to problems in the valuation of aid received by the recipient countries in the developing world.

Annex IV Capital Inflows to Developing and Transition Countries—Identifying Causes and Formulating Appropriate Policy Responses

While large-scale capital inflows experienced in recent years by a number of developing and transition countries have been clearly beneficial from the standpoint of easing external financing constraints, they have also created concern about their potential effects on macroeconomic stability, the competitiveness of the external sector, and external viability. Key among the concerns are the risks of an acceleration of inflation and of an unsustainable rise in the real effective exchange rate, whether in the form of nominal appreciation or of a rise in inflation relative to trading partners. Policymakers faced with the threat of overheating in the wake of large capital inflows have had to decide on the magnitude, sequencing, and timing of policy actions. The appropriate policy response will depend on the objectives of the authorities, the exchange rate regime, the domestic institutional constraints and, importantly, on what factors are driving the inflows. In practice, however, it is difficult to identify early on the underlying causes of inflows, or to distinguish between temporary and more durable inflows, so that judgments have to be made on the basis of limited information.

This annex sets out a stylized framework aimed at addressing two questions: First, what indicators help to shed light on the causes of capital inflows? And second, what would be the appropriate policy responses in the face of these inflows? Capital inflows in any specific case are likely to be associated with a combination of causes, complicating the task of identification as well as that of formulating an optimal policy response. Any simple rules-based categorization of causes and policy responses can therefore be of only limited practicality, and individual country circumstances will necessarily have to be taken into account. Nevertheless, the framework outlined here provides some general principles and guidelines.

Identifying the Causes of Capital Inflows

The causes of capital inflows can be usefully grouped into three major categories: autonomous increases in the domestic demand for money; increases in the domestic productivity of capital; and external factors, such as a fall in international interest rates.1 The first two causes are examples of what in the literature are called “pull” factors; the last is an example of a “push” factor.

The dominant cause of the inflows and the prevalent exchange rate regime will determine the likely economic impact and the need, if any, for policy responses. Under a fully flexible exchange rate system, capital inflows (regardless of their cause) will lead to an appreciation of the exchange rate, a fall in the relative price of imported goods, and a shift of consumption away from nontradables—all of which will tend to alleviate inflationary pressures. Therefore, all other things being equal, the higher the degree of exchange rate flexibility, the less likely it is that capital inflows will have an inflationary effect.

Under a managed float, or a fixed exchange rate system, the inflationary pressures created by capital inflows will depend on whether the inflows reflect a shift in the domestic money demand function or are caused by other factors, for example, a fall in international interest rates or an increase in the domestic productivity of capital. If capital inflows reflect primarily an increase in the domestic demand for money, they will not have an inflationary effect. In contrast, if capital inflows increase for any other reason, the accumulation of foreign exchange reserves will lead, in the absence of sterilization, to an expansion of the monetary base that will tend to increase inflationary pressures and cause a deterioration in the external position.

Table 42 lists a set of indicators that may help to differentiate between inflows associated with a money demand shift and those associated with other exogenous factors. The indicators are divided into four categories: asset prices, monetary and credit aggregates, balance of payments data, and key international variables.2 In terms of timeliness, data on asset prices, both domestic and international, are likely to be superior to data on monetary and credit aggregates and the external accounts. The usefulness of different domestic financial indicators will depend on the economy’s institutional structure and on the sophistication of the data gathering and statistical reporting systems in place.

Table 42.

Potential Indicators for Analyzing Causes of Capital Inflows in Developing and Transition Countries

article image

In countries with established financial and equity markets, relative asset-price movements may be particularly useful for identifying causes (Table 43).3 An upward shift in the money demand function is likely to lead to downward pressure on domestic bond, equity, and real estate prices as asset holders reallocate their portfolios, while inflows resulting from lower international interest rates will lead to a bidding up of prices on real and financial assets. The effect on asset prices of an increase in domestic productivity is likely to be the same as that of a fall in international interest rates—upward pressure on prices of equities and financial assets.

Table 43.

Symptoms of Causes of Capital Inflows

article image

The analysis here considers an upward shift in the domestic demand-for-money function.