IN THE 1950s and 1960s, the growth costs of inflation were not considered a serious issue. The surge in inflation in the 1970s and 1980s changed views on this issue radically. Large growth costs surface only at higher rates of inflation, but the upward bias of inflationary persistence argues for keeping the inflation genie tightly in the bottle.
The last quarter of this century has witnessed a prolonged and rather unique spurt of worldwide inflation amid lower growth. After a turbulent two decades, this period now seems to be coming to an end. Very low inflation is again becoming the norm, not only in the industrial world but also in developing regions that had long been plagued by chronic inflation rates of three digits and more. In the industrial countries, growth seems to have permanently dropped. However, in some developing countries that suffered from very high inflation, the reform process may be yielding post-crisis growth rates that are higher than before the crisis.
The transition from high inflation to relative price stability, while not yet universal, has been widespread and dramatic. Brazil, one of the last bastions of very high inflation in Latin America, has recently joined the region’s other successful stabilizers (Bolivia, Chile, Mexico, and, more recently, Argentina and Peru); even in Eastern and Central Europe, the number of countries stabilizing from the post-collapse, hyperinflation explosion is gradually increasing.
Why is it that the prevailing view in the 1950s and 1960s, that the growth costs of inflation are not serious and, indeed, that some inflation is good for growth, has today been discarded? It could be simply that after two decades of rampant and often persistent inflation and growth crises, and the painful lessons of stabilization, we are much wiser than before. Or, it may be that the growth costs of inflation and the importance of keeping it in check only become significant beyond a certain inflation threshold. If so, it may be that the underlying mechanism for low inflation is inherently different from that for much higher inflation.
Why worry about low or moderate inflation if the growth costs seem to be less pronounced at those rates? The answer lies in the dynamics of inflation itself. If not held in check, a little inflation can lead to higher inflation, and eventually also to the inevitable costs of low growth.
Studying the evidence
A recent World Bank study of inflation and growth in 127 countries, from 1960 to 1992 shows a close association between rising inflation and diminishing growth rates across a variety of inflation ranges (see Chart 1). Over this period, average growth rates seem to fall only slightly as inflation rates move up to 20-25 percent. But growth rates decline more steeply as inflation rates approach 25-30 percent, and growth rates become increasingly negative at higher rates of inflation.
The picture changes when we look at different subperiods. During 1960-72, per capita growth on average increased as inflation rose from negative or low rates to the 15-20 percent range, suggesting that, at inflation rates below 20 percent, inflation and growth may even have been positively related (Chart 2). In the 1950s and 1960s, low-to-moderate inflation went hand in hand with very rapid growth because of investment demand pressures in an expanding economy. (There were, however, few cases of inflation in excess of 25 percent per annum.) In contrast, the 1970s and 1980s were marked by supply shocks. Inflation and growth were negatively related during the 1973-92 subperiod, at least in part because supply shocks affected both variables simultaneously. But for higher inflation frequencies, causality also ran from inflation to lower growth.
A striking picture emerges when we look at the transition from the 1970s into the 1980s (Chart 3). The industrial countries that had maximum inflation rates below the 20 percent range during the 1970s all stayed below that level during the 1980s. In contrast, those countries with inflation rates in the 20-40 percent range during the 1970s experienced higher inflation in the 1980s, although they still did not go into extreme inflation. The 40 percent inflation rate appears to be a threshold: those countries that crossed it in the 1970s were almost certainly on their way to extreme inflation rates in the 1980s. This experience suggests that relatively low-to-moderate inflation (20-40 percent) may not lead to immediate growth reductions, but it is very likely to lead to much higher inflation. The truly dangerous inflations occur at 40 percent and above.
Chart 3Inflation transitions. 1970s to 1980s
Source: Michael Bruno and William Easterly, “Inflation Crises and Long-Run Growth,” unpublished, World Bank, 1995.
Does high inflation cause low growth, or low growth cause high inflation—or is most of what we observe due to other factors (budget deficits, debt rescheduling, and supply shocks) that affect inflation and growth simultaneously? One relevant bit of evidence is that there is a significant effect from past inflation to current growth but no evidence for a feedback from past growth to current inflation. Once the supply shocks of the 1970s set in, the negative relationship between inflation and growth was at least partly a simultaneous response to these shocks. But at higher rates of inflation (certainly at those experienced by many non-industrial countries in the 1980s), the evidence is consistent with causality also running from high inflation to low growth.
The World Bank study examined what happened to per capita growth (compared with the world average) before, during, and after a high-inflation crisis. The threshold for defining an inflation “crisis” was set at an annual rate of consumer-price inflation of 40 percent or higher, prevailing for two consecutive years or more. The study found no uniform pattern for countries’ growth rates before the inflation crisis—some were above average and some, below average. That is, there is no evidence that countries that have inflation crises are those that have subnormal growth beforehand. In most cases, however, growth during the inflation crisis was both below the world average for the period and below the country’s growth prior to the outbreak of the crisis. The magnitudes of the growth depressions during the crisis were large. In some countries with repeated inflation crises, like Brazil, Somalia, and Zaire, the growth crises matched the inflation crises exactly. By contrast, when all inflation crises (above 40 percent) were excluded from the international data set, there was no evidence of a growth-inflation relationship.
The study found surprising uniformity in the response of growth after the crisis for those countries that had brought inflation down by the end of the 1980s. The post-crisis growth rate was also above the pre-crisis growth rate and above the world average in most cases. This growth acceleration suggests a strong return to trend after the collapse of output during the crisis; it could indicate a change for the better in long-run growth after the crisis.
The experience of countries that suffered the great hyperinflations in the period after World Wars I and II is also illustrative. All of the European countries that experienced hyperinflation in the 1920s (Austria, Germany, Hungary, Poland, and Russia) had higher-than-average growth from the ends of their inflation crises to 1938, the last year before World War II broke out. Likewise, the recovery from the post-World War II inflation crises is extremely strong for the affected countries, which include the East Asian tigers (Japan, Korea, and Taiwan Province of China), the fastest-growing industrial countries (Greece and Italy), and three of the fastest-growing socialist economies (China, Hungary, and Romania). For the period 1950-87, as well as the sub-period 1950-65, all of these countries had an average per capita growth rate that was higher than any of a comparator group of 26 countries that did not have serious inflation after World War II.
Finally, the dramatic output collapse in Eastern Europe and the Baltic countries, Russia, and the other countries of the former Soviet Union (BRO) has occurred simultaneously with high rates of inflation in many countries. There is a correlation between the two. Output has been recovering mainly in those economies that have stabilized inflation, while output continues to drop in economies where inflation remains high. Also, the most recent successful Latin American stabilizations, in Argentina and Peru, have been accompanied by vigorous and immediate output recoveries.
Both the various historical episodes and the recent data on crisis and recovery suggest—though they are far from conclusive—that something more than reversion to trend could be occurring. A high-inflation crisis may shock policymakers into instituting productivity-enhancing reforms that would not have taken place if the country had just muddled along without severe crises. Inflation may be but a barometer of a more general galvanizing crisis.
Roots of high inflation
High inflation is always associated with large fiscal deficits. Detailed country comparisons show that in all cases of inflation crisis, countries had large fiscal deficits before the crisis; even larger ones during the crisis; and, finally, significantly below-average budget deficits during the post-crisis recovery. Moreover, there is no evidence that terms of trade shocks or political turning points played a role in these inflation crises.
In contrast, there is no evidence that variations in rates of inflation below 40 percent (annual rate) are related to fiscal deficits. Terms of trade shocks, on the other hand, seem to matter in these low-to-moderate inflation cases. The fact that, under chronic inflation, price-level shocks lead to inflation-rate jumps has been known for a long time. It is only recently, however, that the theory tying price-level shocks and monetary accommodation with the medium- and long-term dynamics of inflation has been spelled out in more precise terms. A price-level shock (which is a one-time inflation-rate shock) will cause an immediate increase in inflation with subsequent fluctuations, at a decreasing amplitude, until inflation eventually settles on a new, higher steady state. But if inflation becomes endemic, as expected inflation goes up, the speed of adjustment of the various nominal magnitudes to lagged inflation will go up also so as to avoid the erosion of real variables that would otherwise take place.
Consider, for example, the frequency of nominal-wage adjustments. As inflation rises, workers will obviously demand faster adjustment of nominal wages to protect their real incomes. This will show in the shortening of wage contracts. Likewise, when there are formal backward-looking indexation arrangements, higher inflation will make workers demand a shortening of periods over which ex post adjustment is made.
In an open economy with exchange-rate pegging or an exchange-rate crawl, the frequency of adjustment of the nominal exchange rate to rising prices, to avoid erosion of competitiveness, will likewise increase with the average (expected) inflation rate. Similar considerations apply to other nominal variables which tend to be linked to inflation. For example, there will be an increase in the share of indexed assets in financial portfolios, more frequent adjustments of prices that are controlled by the government, increased indexation of contracts, etc. The upshot of all these reactions is that the lag between inflation adjustments becomes shorter and shorter.
This mechanism is an essential element in the explanation of the increasing step-like function of the inflationary process under dynamic inflation. Thus, for a given shock, the shorter the lag of adjustment to changes in prices, the larger the rise in the rate of inflation. On top of this, as the rate of inflation accelerates, the adjustment lag decreases, fueling still larger rises in the rate of inflation.
Positive price-level shocks translate into ever-increasing upward shifts in the inflation rate. But, why isn’t this process symmetric, so that negative shocks translate into ever-decreasing inflation rates? First, there may be inherent factors in the nature of microeconomic price adjustment that make downward shocks lead to a fall in output rather than reductions in inflation. When there is a fixed cost to price adjustment, firms will adjust their prices only when the real price of their output (relative to the general price level) falls below a threshold. Lags in adjustment will be longer when shocks are in the opposite direction to the trend. Thus, in an inflationary environment, prices will be stickier on the way down than on the way up.
Second, one could argue that in an inflationary environment where government follows an expansionary fiscal and/or accommodative monetary policy, individuals and firms will tend to discount negative price shocks as being temporary.
Finally, perhaps the most important reason for this asymmetry is that the source of shocks in a high-inflation environment is usually an upward adjustment in those prices that are kept temporarily constant (exchange rates, controlled food and utility prices, etc.). Thus, the shocks themselves are partly endogenous to the inflationary process. All of this leads one to believe that even though chronic inflation may exhibit temporary periods of stable rates, price-level shocks are very likely to lead to a step-like pattern of inflation that, sooner or later, ends in hyperinflationary explosions.
Inflation of all nominal magnitudes, with no sizable relative price changes, seems at times to have a life of its own, divorced from the real economy. Obviously this divorce is to some extent an optical illusion: a sustained real fiscal deficit is needed to keep the nominal process alive. More important, high inflation hurts the real economy’s output performance by discouraging investment and hindering productivity growth.
The high-inflation dynamics just described have important implications for the design of an effective stabilization strategy. Just as there are substantial differences between lower and higher inflation rates in the nature of the inflation process and the inflation-growth trade-offs, so, too, there should be a difference in the nature and mix of stabilization policies. Cumulative evidence points to the overriding role of a large fiscal deficit as a root cause of high inflation (with monetary accommodation playing an important role in sustaining and expanding its momentum). Thus, behind every successful stabilization lies a corresponding fiscal correction. Bubble theories of inflation that contend that some instances of inflation could be driven purely by expectations, and thus be cured with wage and price controls alone, have been proven plain wrong.
In cases of high inflation, fiscal retrenchment alone may not be sufficient to bring inflation under control. Instead, governments may need to intervene directly to signal credibly that inflation will, in fact, be contained and reduced. This can be done by establishing one or more nominal anchors—economic variables that serve as anchors for the rest of the price system. The establishment or re-establishment of a credible monetary anchor—usually in the form of an initial exchange rate peg, together with fiscal retrenchment—has also become a centerpiece of almost all successful, fast stabilizations.
There are several ways to establish a nominal anchor to treat high or hyperinflation. An extreme response, one that completely ties the government’s hands, is to establish an independent currency board. Several countries, among them Argentina and Estonia, have recently done so. In Mexico, continued inflation after the fiscal crunch of 1982–83 finally resulted in the 1988 agreement involving government, business, and labor in which the exchange rate and wages were simultaneously frozen. In contrast, the costs of not establishing an unambiguous monetary anchor, of using only fiscal means to treat high inflation, can be seen in Chile’s 1973-74 stabilization program, which reduced inflation only gradually, in an unsynchronized manner, and at a huge unemployment cost.
Some cases may call for an ex ante multiple anchor—particularly in systems with widespread indexation. Within the group of countries that are successful stabilizers, cases can be identified in which, in addition to a fiscal retrenchment and an exchange rate anchor, complementary nominal anchors, particularly freezing wages, became part of the initial stabilization package. This was the case in Israel (1985), Mexico (1988), and in several Eastern and Central European countries (Poland and Yugoslavia in 1990, and the former Czechoslovakia in 1991).
“Relatively low-to-moderate inflation (20–40 percent) may not lead to immediate growth reductions, but it is very likely to lead to much higher inflation. The truly dangerous inflations occur at 40 percent and above.”
Can inflation be brought down all at once to zero? The answer is that it may be possible in an extreme case of hyperinflation. Both in Germany in the 1920s and, more recently, in Argentina in 1991, inflation was brought down to virtually zero almost overnight. In several recent cases of stabilization from high inflation, there has been a sharp disinflation from three digits to rates that were either within or on the borderline of the low-to-moderate inflation range. For example, during several years, until around 1990, the four most successful stabilizers from high inflation—Bolivia, Chile, Israel, and Mexico—were running inflation rates of close to 20 percent, and only subsequently was inflation brought down more gradually. The renewed macroeconomic crisis (and the temporary resurgence in inflation) in Mexico in early 1995 showed again how hard it is to make stabilization permanent.
Two points are worth mentioning in this context. In a country that had suffered from high inflation for a long time, a reduction to a 20 percent annual rate—which, for all intents and purposes, would be regarded as runaway inflation in a typical industrial country—is regarded as an achievement of relative price stability. But, there is more to it than that. As we have seen, the costs to the economy in terms of lost growth are incurred at rates that lie much above 20 percent. Likewise, the main growth dividends seem to be reaped in moving from three-digit inflation rates to those of about 20 percent. There is no obvious empirical evidence for significant long-run growth costs of inflation at rates that lie below 20 percent or so, while there is a significant negative short-term trade-off with the reduction of inflation at those rates. A synchronized reduction from a 20 percent inflation rate to zero is thus much harder to engineer, both economically and politically, than a reduction from 500 percent to 20 percent. The point is, however, that resting on one’s laurels after stabilization to 20 percent can be dangerous because of the tendency of inflation to ratchet upward. Thus, getting inflation down to single digits is important even for longer-term growth reasons.
Chronic inflation tends to resemble smoking: once you get the habit, it is very difficult to escape a worsening addiction. On the other hand, if you do kick the habit, your health suffers no long-run damage. Monetary control by an independent central bank can play an important role in achieving disinflation at low rates of inflation, under reasonable fiscal behavior, before inflation moves to the moderate, let alone the high, range. When this is done, the worst scenarios need never come to pass.
The past quarter century of unusual inflationary developments started with the worldwide monetary expansion of the late 1960s and early 1970s, followed by sizable oil and commodity price shocks. Such price-level shocks were themselves largely endogenous, or were at least facilitated by an expansionary monetary environment. Shocks of such magnitude may not happen again soon—but there is no guarantee that they will not. Moreover, as we move into the next century, the globalization of trade and payments is making individual countries increasingly susceptible to even small shocks. In such a situation, recent economic history provides a useful lesson: even in the face of very large shocks, the extent of their spillover into high inflation depends on the resolution of governments and central banks. For those that lack such resolution, the crisis that follows can offer an opportunity for far-reaching reform that may eventually put the country on a better footing. However, protracted crises also tend to be very costly, both socially and politically. The only way to avoid such extremes is to bolster the weaker parts of the economy’s structure and institutions ahead of time.
This article is partly based on joint work with William Easterly of the World Bank. For more information on the empirical study, see Michael Bruno and William Easterly, “Inflation Crises and Long-Run Growth,” unpublished, World Bank, 1995. Part of this discussion is also based on the lecture by the author (to be published by the Central Bank of Italy, in memory of Paolo Baffi), “Inflation, Growth and Monetary Control: Nonlinear Lessons from Crisis and Recovery,” unpublished, 1995.