Does Really Lower Growth?

MICHAEL BRUNO

of the last bastions of very high inflation 20-25 percent. But growth rates decline more IN THE 1950s and 1960s, in Latin America, has recently joined the steeply as inflation rates approach 25-30 per- region's other successful stabilizers (Bolivia, cent, and growth rates become increasingly the growth costs of Chile, Mexico, and, more recently, negative at higher rates of inflation. inflation were not consid- and ); even in Eastern and Central The picture changes when we look at differ- Europe, the number of countries stabilizing ent subperiods. During 1960-72, per capita nI ered a serious issue. The from the post-collapse, explo- growth on average increased as inflation rose surge in inflation in the 1970s sion is gradually increasing. from negative or low rates to the 15-20 per- Why is it that the prevailing view in the cent range, suggesting that, at inflation rates and 1980s changed views on this 1950s and 1960s, that the growth costs of below 20 percent, inflation and growth may issue radically. Large growth inflation are not serious and, indeed, that some even have been positively related (Chart 2). costs surface only at higher rates inflation is good for growth, has today been In the 1950s and 1960s, low-to-moderate infla- discarded? It could be simply that after two tion went hand in hand with very rapid of inflation, but the upward bias decades of rampant and often persistent infla- growth because of investment demand pres- of inflationary persistence tion and growth crises, and the painful lessons sures in an expanding economy. (There were, of stabilization, we are much wiser than however, few cases of inflation in excess of 25 argues for keeping the inflation before. Or, it may be that the growth costs of percent per annum.) In contrast, the 1970s and genie tightly in the bottle. inflation and the importance of keeping it in 1980s were marked by supply shocks. check only become significant beyond a cer- Inflation and growth were negatively related tain inflation threshold. If so, it may be that during the 1973-92 subperiod, at least in part the underlying mechanism for low inflation because supply shocks affected both variables The last quarter of this century has witnessed is inherently different from that for much simultaneously. But for higher inflation fre- a prolonged and rather unique spurt of world- higher inflation. quencies, causality also ran from inflation to wide inflation amid lower growth. After a tur- Why worry about low or moderate inflation lower growth. bulent two decades, this period now seems to if the growth costs seem to be less pronounced A striking picture emerges when we look at be coming to an end. Very low inflation is at those rates? The answer lies in the dynam- the transition from the 1970s into the 1980s again becoming the norm, not only in the ics of inflation itself. If not held in check, (Chart 3). The industrial countries that had industrial world but also in developing a little inflation can lead to higher inflation, maximum inflation rates below the 20 percent regions that had long been plagued by chronic and eventually also to the inevitable costs of range during the 1970s all stayed below that inflation rates of three digits and more. In the low growth. level during the 1980s. In contrast, those coun- industrial countries, growth seems to have tries with inflation rates in the 20-40 percent permanently dropped. However, in some Studying the evidence range during the 1970s experienced higher developing countries that suffered from very A recent study of inflation and inflation in the 1980s, although they still did high inflation, the reform process may be growth in 127 countries, from 1960 to 1992 not go into extreme inflation. The 40 percent yielding post-crisis growth rates that are shows a close association between rising infla- inflation rate appears to be a threshold: those higher than before the crisis. tion and diminishing growth rates across a countries that crossed it in the 1970s were The transition from high inflation to rela- variety of inflation ranges (see Chart 1). Over almost certainly on their way to extreme infla- tive price stability, while not yet universal, has this period, average growth rates seem to fall tion rates in the 1980s. This experience sug- been widespread and dramatic. Brazil, one only slightly as inflation rates move up to gests that relatively low-to-moderate inflation

Michael Bruno, an Israeli national, is Vice President, Development Economics, and Chief Economist of the World Bank.

Finance & Development / September 1995 35

©International Monetary Fund. Not for Redistribution Chart 1 Chart 2 Inflation and per capita growth, 1960-921 Inflation and per capita growth, 1960-722

Source: Michael Bruno and William Easterly, "Inflation Crises and Long-Run Growth," unpublished, World Bank, 1995. 1 Inflation and growth rates for a sample of 127 countries over the period 1960-92. For example, inflation rates of 5-10 percent are associated with growth rates of just over 2 percent. 2 Inflation and growth rates for a sample of 127 countries over the period 1960-72.

(20-40 percent) may not lead to immediate average for the period and below the country's the sub-period 1950-65, all of these countries growth reductions, but it is very likely to lead growth prior to the outbreak of the crisis. The had an average per capita growth rate that to much higher inflation. The truly dangerous magnitudes of the growth depressions during was higher than any of a comparator group of occur at 40 percent and above. the crisis were large. In some countries with 26 countries that did not have serious inflation Does high inflation cause low growth, or repeated inflation crises, like Brazil, Somalia, after World War II. low growth cause high inflation—or is most and Zaire, the growth crises matched the infla- Finally, the dramatic output collapse in of what we observe due to other factors (bud- tion crises exactly. By contrast, when all infla- Eastern Europe and the Baltic countries, get deficits, debt rescheduling, and supply tion crises (above 40 percent) were excluded Russia, and the other countries of the former shocks) that affect inflation and growth simul- from the international data set, there was no Soviet Union (BRO) has occurred simultane- taneously? One relevant bit of evidence is that evidence of a growth-inflation relationship. ously with high rates of inflation in many there is a significant effect from past inflation The study found surprising uniformity in countries. There is a correlation between the to current growth but no evidence for a feed- the response of growth after the crisis for two. Output has been recovering mainly in back from past growth to current inflation. those countries that had brought inflation those economies that have stabilized inflation, Once the supply shocks of the 1970s set in, the down by the end of the 1980s. The post-crisis while output continues to drop in economies negative relationship between inflation and growth rate was also above the pre-crisis where inflation remains high. Also, the most growth was at least partly a simultaneous growth rate and above the world average in recent successful Latin American stabiliza- response to these shocks. But at higher rates most cases. This growth acceleration suggests tions, in Argentina and Peru, have been of inflation (certainly at those experienced by a strong return to trend after the collapse accompanied by vigorous and immediate out- many non-industrial countries in the 1980s), of output during the crisis; it could indicate put recoveries. the evidence is consistent with causality also a change for the better in long-run growth Both the various historical episodes and running from high inflation to low growth. after the crisis. the recent data on crisis and recovery sug- The experience of countries that suffered gest—though they are far from conclu- Inflation crises the great in the period after sive—that something more than reversion to The World Bank study examined what hap- World Wars I and II is also illustrative. All of trend could be occurring. A high-inflation cri- pened to per capita growth (compared with the European countries that experienced sis may policymakers into instituting the world average) before, during, and after a hyperinflation in the 1920s (Austria, Germany, productivity-enhancing reforms that would high-inflation crisis. The threshold for defin- Hungary, Poland, and Russia) had higher- not have taken place if the country had just ing an inflation "crisis" was set at an annual than-average growth from the ends of their muddled along without severe crises. Inflation rate of consumer-price inflation of 40 percent inflation crises to 1938, the last year before may be but a barometer of a more general gal- or higher, prevailing for two consecutive years World War II broke out. Likewise, the recovery vanizing crisis. or more. The study found no uniform pattern from the post-World War II inflation crises is for countries' growth rates before the inflation extremely strong for the affected countries, Roots of high inflation crisis—some were above average and some, which include the East Asian tigers Japan, High inflation is always associated with below average. That is, there is no evidence Korea, and Taiwan Province of China), the large fiscal deficits. Detailed country compar- that countries that have inflation crises are fastest-growing industrial countries (Greece isons show that in all cases of inflation crisis, those that have subnormal growth before- and Italy), and three of the fastest-growing countries had large fiscal deficits before the hand. In most cases, however, growth during socialist economies (China, Hungary, and crisis; even larger ones during the crisis; the inflation crisis was both below the world Romania). For the period 1950-87, as well as and, finally, significantly below-average bud-

36 Finance & Development / September 1995

©International Monetary Fund. Not for Redistribution Charts Inflation transitions, 1970s to 1980s (logarithmic scales)

Source: Michael Bruno and William Easterly, "Inflation Crises and Long-Run Growth," unpublished, world Bank, 1995.

get deficits during the post-crisis recovery. pegging or an exchange-rate crawl, the fre- ronment, prices will be stickier on the way Moreover, there is no evidence that terms of quency of adjustment of the nominal ex- down than on the way up. trade shocks or political turning points played change rate to rising prices, to avoid erosion of Second, one could argue that in an inflation- a role in these inflation crises. competitiveness, will likewise increase with ary environment where government follows In contrast, there is no evidence that varia- the average (expected) inflation rate. Similar an expansionary fiscal and/or accommodative tions in rates of inflation below 40 percent considerations apply to other nominal vari- , individuals and firms will (annual rate) are related to fiscal deficits. ables which tend to be linked to inflation. For tend to discount negative price shocks as Terms of trade shocks, on the other hand, example, there will be an increase in the share being temporary. seem to matter in these low-to-moderate infla- of indexed assets in financial portfolios, more Finally, perhaps the most important reason tion cases. The fact that, under chronic infla- frequent adjustments of prices that are con- for this asymmetry is that the source of tion, price-level shocks lead to inflation-rate trolled by the government, increased indexa- shocks in a high-inflation environment is usu- jumps has been known for a long time. It is tion of contracts, etc. The upshot of all these ally an upward adjustment in those prices that only recently, however, that the theory tying reactions is that the lag between inflation are kept temporarily constant (exchange rates, price-level shocks and monetary accommoda- adjustments becomes shorter and shorter. controlled food and utility prices, etc.). Thus, tion with the medium- and long-term dynam- This mechanism is an essential element in the shocks themselves are partly endogenous ics of inflation has been spelled out in more the explanation of the increasing step-like to the inflationary process. All of this leads precise terms. A price-level shock (which is a function of the inflationary process under one to believe that even though chronic infla- one-time inflation-rate shock) will cause an dynamic inflation. Thus, for a given shock, the tion may exhibit temporary periods of stable immediate increase in inflation with subse- shorter the lag of adjustment to changes in rates, price-level shocks are very likely to lead quent fluctuations, at a decreasing amplitude, prices, the larger the rise in the rate of infla- to a step-like pattern of inflation that, sooner until inflation eventually settles on a new, tion. On top of this, as the rate of inflation or later, ends in hyperinflationary explosions. higher steady state. But if inflation becomes accelerates, the adjustment lag decreases, fuel- Inflation of all nominal magnitudes, with no endemic, as expected inflation goes up, the ing still larger rises in the rate of inflation. sizable relative price changes, seems at times speed of adjustment of the various nominal Positive price-level shocks translate into to have a life of its own, divorced from the real magnitudes to lagged inflation will go up also ever-increasing upward shifts in the inflation economy. Obviously this divorce is to some so as to avoid the erosion of real variables that rate. But, why isn't this process symmetric, so extent an optical illusion: a sustained real fis- would otherwise take place. that negative shocks translate into ever- cal deficit is needed to keep the nominal Consider, for example, the frequency of decreasing inflation rates? First, there may be process alive. More important, high inflation nominal-wage adjustments. As inflation rises, inherent factors in the nature of microeco- hurts the real economy's output performance workers will obviously demand faster adjust- nomic price adjustment that make downward by discouraging investment and hindering ment of nominal wages to protect their real shocks lead to a fall in output rather than productivity growth. incomes. This will show in the shortening of reductions in inflation. When there is a fixed wage contracts. Likewise, when there are for- cost to price adjustment, firms will adjust Stabilization strategy mal backward-looking indexation arrange- their prices only when the real price of their The high-inflation dynamics just described ments, higher inflation will make workers output (relative to the general price level) falls have important implications for the design of demand a shortening of periods over which ex below a threshold. Lags in adjustment will be an effective stabilization strategy. Just as there post adjustment is made. longer when shocks are in the opposite direc- are substantial differences between lower and In an open economy with exchange-rate tion to the trend. Thus, in an inflationary envi- higher inflation rates in the nature of the infla-

Finance & Development / September 1995 37

©International Monetary Fund. Not for Redistribution tion process and the inflation-growth tendency of inflation to ratchet upward. trade-offs, so, too, there should be a dif- "Relatively low-to-moderate Thus, getting inflation down to single ference in the nature and mix of stabi- digits is important even for longer-term lization policies. Cumulative evidence inflation (20-40 percent) growth reasons. points to the overriding role of a large may not lead to immediate fiscal deficit as a root cause of high Conclusion inflation (with monetary accommoda- growth reductions, but it is Chronic inflation tends to resemble tion playing an important role in sus- very likely to lead to much smoking: once you get the habit, it is taining and expanding its momentum). very difficult to escape a worsening Thus, behind every successful stabi- higher inflation. The truly addiction. On the other hand, if you do lization lies a corresponding fiscal cor- dangerous inflations occur kick the habit, your health suffers no rection. Bubble theories of inflation that long-run damage. Monetary control by contend that some instances of inflation at 40percent and above." an independent can play could be driven purely by expectations, an important role in achieving disinfla- and thus be cured with wage and price tion at low rates of inflation, under rea- controls alone, have been proven sonable fiscal behavior, before inflation plain wrong. Can inflation be brought down all at once to moves to the moderate, let alone the high, In cases of high inflation, fiscal retrench- zero? The answer is that it may be possible in range. When this is done, the worst scenarios ment alone may not be sufficient to bring an extreme case of hyperinflation. Both in need never come to pass. inflation under control. Instead, governments Germany in the 1920s and, more recently, in The past quarter century of unusual infla- may need to intervene directly to signal credi- Argentina in 1991, inflation was brought tionary developments started with the world- bly that inflation will, in fact, be contained down to virtually zero almost overnight. In wide monetary expansion of the late 1960s and reduced. This can be done by establishing several recent cases of stabilization from high and early 1970s, followed by sizable oil and one or more nominal anchors—economic vari- inflation, there has been a sharp commodity price shocks. Such price-level ables that serve as anchors for the rest of the from three digits to rates that were either shocks were themselves largely endogenous, price system. The establishment or re-estab- within or on the borderline of the low-to-mod- or were at least facilitated by an expansionary- lishment of a credible monetary anchor—usu- erate inflation range. For example, during monetary environment. Shocks of such magni- ally in the form of an initial exchange rate several years, until around 1990, the four tude may not happen again soon—but there is peg, together with fiscal retrenchment—has most successful stabilizers from high in- no guarantee that they will not. Moreover, as also become a centerpiece of almost all suc- flation—Bolivia, Chile, Israel, and Mexico we move into the next century, the globaliza- cessful, fast stabilizations. —were running inflation rates of close to 20 tion of trade and payments is making individ- There are several ways to establish a nomi- percent, and only subsequently was inflation ual countries increasingly susceptible to even nal anchor to treat high or hyperinflation. An brought down more gradually. The renewed small shocks. In such a situation, recent eco- extreme response, one that completely ties the macroeconomic crisis (and the temporary nomic history provides a useful lesson: even government's hands, is to establish an inde- resurgence in inflation) in Mexico in early in the face of very large shocks, the extent of pendent currency board. Several countries, 1995 showed again how hard it is to make sta- their spillover into high inflation depends on among them Argentina and Estonia, have bilization permanent. the resolution of governments and central recently done so. In Mexico, continued infla- Two points are worth mentioning in this banks. For those that lack such resolution, the tion after the fiscal crunch of 1982-83 finally- context. In a country that had suffered from crisis that follows can offer an opportunity for resulted in the 1988 agreement involving gov- high inflation for a long time, a reduction to a far-reaching reform that may eventually put ernment, business, and labor in which the 20 percent annual rate—which, for all intents the country on a better footing. However, pro- exchange rate and wages were simultaneously and purposes, would be regarded as runaway tracted crises also tend to be very costly, both frozen. In contrast, the costs of not establish- inflation in a typical industrial country—is socially and politically. The only way to avoid ing an unambiguous monetary anchor, of regarded as an achievement of relative price such extremes is to bolster the weaker parts of using only fiscal means to treat high inflation, stability. But, there is more to it than that. As the economy's structure and institutions can be seen in Chile's 1973-74 stabilization we have seen, the costs to the economy in ahead of time. ^B program, which reduced inflation only gradu- terms of lost growth are incurred at rates that ally, in an unsynchronized manner, and at a lie much above 20 percent. Likewise, the main huge cost. growth dividends seem to be reaped in mov- Some cases may call for an ex ante multiple ing from three-digit inflation rates to those of anchor—particularly in systems with wide- about 20 percent. There is no obvious empiri- spread indexation. Within the group of coun- cal evidence for significant long-run growth tries that are successful stabilizers, cases can costs of inflation at rates that lie below 20 per- be identified in which, in addition to a fiscal cent or so, while there is a significant negative This article is partly based on joint icork with retrenchment and an exchange rate anchor, short-term trade-off with the reduction of in- William Easterly of the World Hank. For more complementary nominal anchors, particularly flation at those rates. A synchronized reduc- information on the empirical study, see Michael freezing wages, became part of the initial tion from a 20 percent inflation rate to zero is Bruno and William I'Msterly, "Inflation Crises anil Long-Run Grow//!," unpublished. World Bank. thus much harder to engineer, both economi- stabilization package. This was the case in 1995. Part of this discussion is also based on the Israel (1985), Mexico (1988), and in several cally and politically, than a reduction from 500 lecture by the author (to be published by the Central Eastern and Central European countries percent to 20 percent. The point is, however, Bank of Italy, in memory of Paolo Baffi), "Inflation, (Poland and Yugoslavia in 1990, and the for- that resting on one's laurels after stabilization (iroietli and Monetary Control: Nonlinear Lessons mer Czechoslovakia in 1991). to 20 percent can be dangerous because of the from Crisis and Recovery," unpublished, 1995.

38 I-'inance it- Decelopment ' Sep/cinher 199:i

©International Monetary Fund. Not for Redistribution