The Behavior of Non-Oil Commodity Prices€¦ · Volume and Real Value of Non-Oil Commodity Exports 2 3. Export Volumes by Region and Commodity Grouping 2 II. 4. Real Non-Oil Commodity

The Behavior of Non-OilCommodity Prices

Eduardo Borensztein, Mohsin S. Khan,Carmen M. Reinhart, and Peter Wickham

INTERNATIONAL MONETARY FUND

Washington DC

August 1994

OCCASIONAL PAPER 112

©International Monetary Fund. Not for Redistribution

© 1994 International Monetary Fund

Library of Congress Cataloging-in-Publication Data

The behavior of non-oil commodity prices / Eduardo Borensztein . . .[et al.].

p. cm. — (Occasional Papers, ISSN 0251-6365; 112)Includes bibliographical references.ISBN 1-55775-412-81. Prices. 2. Prices — Government policy. I. Borensztein, Eduardo.

II. Series: Occasional paper (International Monetary Fund) ; no. 112.HB231.B44 1994338.5'2—dc20 94-33965

CIP

Price: US$15.00(US$12.00 to full-time faculty members and

students at universities and colleges)

Please send orders to:International Monetary Fund, Publication Services

700 19th Street, N.W., Washington, D.C. 20431, U.S.A.Tel: (202) 623-7430 Telefax: (202) 623-7201

recycled paper


Contents

Preface

Page

vii

I. Introduction

II. Characteristics of Commodity Price Behavior

Trends and Cycles in Commodity PricesSize of the Permanent Component of Price ShocksIncreasing Volatility of Commodity Prices

678

III. Determinants of Commodity Price Movements

Economic Growth and the Demand for CommoditiesExpansion in SupplyInternational Commodity AgreementsDevelopments in Economies in TransitionEconometric Model Results

10

1012141416

IV. Policy Issues

Temporary Price ShocksRole of Government in Smoothing IncomeIntervention Strategies

Stabilization FundsAgricultural BoardsInternational Commodity AgreementsExternal Compensatory Finance

Market-Related InstrumentsGovernment Policies for the Longer Term

17

171718191921212122

V. Conclusion 25

Tables

Section

I. 1. Structure of Merchandise Exports:Selected Developing Countries, 1965 and 1990

2. Real Export Earnings, 1970-92II. 3. Share of the Variance of Changes in Real Non-Oil

Commodity Prices Due to Permanent Shocks, 1957-92III. 4. Intensity of Metals Use in Major Industrial Countries

34

811

iii

1

6


5. Demand for Selected Commodities of the Former SovietUnion

6. Supply of Selected Commodities from the Former SovietUnion

15

15

Charts

Section

I. 1. Real Non-Oil Commodity Prices: Long-Term Developments,1900-92 1

2. Volume and Real Value of Non-Oil Commodity Exports 23. Export Volumes by Region and Commodity Grouping 2

II. 4. Real Non-Oil Commodity Prices: Trends and Cycles,1962:I-1993:II 7

5. Rising Volatility in Real Non-Oil Commodity Prices,1972:I-1993:II 9

III. 6. Factors Affecting Commodity Markets 127. Cereal Yields by Region 148. Real Non-Oil Commodity Prices: Actual and Forecasts 16

AppendixTables

A1. Country Groupings for Charts 2 and 3A2. Country Groupings for Chart 7

2627

References 28

iv

The following symbols have been used throughout this paper:

to indicate that data are not available;

to indicate that the figure is zero or less than half the final digit shown, or that theitem does not exist;

between years or months (e.g., 1991-92 or January-June) to indicate the years ormonths covered, including the beginning and ending years or months;

between years (e.g., 1991/92) to indicate a crop or fiscal (financial) year.

"Billion" means a thousand million.

Minor discrepancies between constituent figures and totals are due to rounding.

The term "country," as used in this paper, does not in all cases refer to a territorial entitythat is a state as understood by international law and practice; the term also covers someterritorial entities that are not states, but for which statistical data are maintained andprovided internationally on a separate and independent basis.

CONTENTS


This paper is based on a study carried out in late 1993 in the Research Depart-ment of the Fund. The authors received useful comments from staff in otherdepartments. The authors wish to thank Ximena Cheetham, Brian Casabianca, andJared Romey for research assistance and Maria Orihuela and Elizabeth Whitely forsecretarial assistance. Thanks are also due to Simon Willson of the External Rela-tions Department who edited the manuscript and coordinated the publicationprocess.

The opinions expressed in the paper are those of the authors and do not neces-sarily reflect the views of the IMF or its Executive Directors.

V

Preace


This page intentionally left blank


I Introduction

he need to understand the factors that influ-ence the behavior of commodity prices has

taken on a special urgency in recent years, as non-oil real commodity prices have been decliningalmost continuously since the early 1980s. Sincetheir short-lived recovery in 1984, real non-oil com-modity prices have fallen by about 45 percent,translating into a sharp deterioration in the termsof trade for most commodity-dependent exporters.As Chart 1 illustrates, in 1992 the price of non-oilcommodities relative to that of manufacturesreached its lowest level in over 90 years.1 The long-term behavior of real commodity prices would thusappear to lend some support to the well-knownPrebisch-Singer hypothesis.2 This pattern in com-modity prices has important practical implicationsfor policymakers. For example, the presence of anegative trend in commodity prices implies contin-uously worsening terms of trade for manycommodity-dependent countries, and further thatefforts to stabilize the incomes of producers for anextended period of time may not be financiallysustainable.

While the decline in prices has affected allcommodity-producing countries in some measure,for some countries the growth in the volume ofexports has been of such magnitude that they havenot experienced a decline in total real earningsfrom commodity exports. As can be seen fromChart 2, industrial countries and Asian countrieshave achieved sustained increases in the real valueof their commodity export earnings. Latin Ameri-can countries have generally maintained the real

1Thc source of the data plotted in Chart 1 is Grilli and Yang(1988) for the period 1900-86 and updated by Reinhart andWickham (1994) for 1987-92. Throughout this paper, referencesto commodity prices and exports exclude energy products.

2Basically, the Prebisch-Singer hypothesis posits that thelong-term trend of the price of primary commodities relative tothe price of manufactures is negative—see Prebisch (1950) andSinger (1950). Empirical evidence for the Prebisch-Singerhypothesis is, however, not conclusive (see, for example,Sapsford (1985), Cuddington and Urzua (1989), and Ardeni andWright (1992)).

value of their commodity exports, while for thegroup of African countries the commodity pricedecline of the last decade has caused a sharpdecrease in the real value of exports, as the volumeof their exports has declined as well. As shown inChart 3, the regional pattern of the evolution of

1

Chart 1. Real Non-Oil Commodity Prices:Long-Term Developments, 1900-92

Sources: Grilli and Yang (1988); and Reinhart and Wickham(1994).

Note: Commodity prices deflated by export unit values ofmanufactured goods.

T


commodity exports is suggestive of similaritiesacross commodity groups.3

It is to be expected that the decline in com-modity prices will have its largest impact on coun-tries with the least diversified production structure.Primary commodities still account for the bulk ofexports in many developing countries. Moreover,this group of countries, which encompasses manyof the lowest-income countries in the world, tendsto have less flexible economic systems, making sub-stitution away from commodity production moredifficult or costly. In effect, the reliance on primarycommodities as the main source of export earnings

3Note that the huge percentage increase in exports of min-erals and metals by the group of Middle Eastern countriesarguably did not have a major impact on markets since theyrepresent a very small fraction of supply, namely, less than 2percent by the end of 1990.

has only slightly diminished for many countries,particularly in Africa, where manufactures oftenaccount for less than 15 percent of merchandiseexports (see Table 1).

In this context, it is also useful to assess thedependence of countries on a limited range ofexport commodities, and how primary export vol-umes (and market shares) have evolved overrecent years. While a more diversified export struc-ture is not necessarily an economic objective in andof itself, export diversification is an important wayof reducing vulnerability to volatility and sustaineddeclines in commodity prices.

Not surprisingly, the performance of real exportearnings for many developing countries during the

2

1 INTRODUCTION

Chart 2. Volume and Real Value of Non-OilCommodity Exports(Index Numbers: 1970= 100)

Source: United Nations Conference on Trade andDevelopment, Handbook of International Trade and DevelopmentStatistics, 1992 (Geneva: UNCTAD, 1993).

Note: For country groupings see Table A1.

Source: UNCTAD, Handbook of International Trade andDevelopment Statistics, 1992 (Geneva: UNCTAD, 1993).

Note: For country groupings see Table A1.

Chart 3. Export Volumes by Region and commodity Grouping(Index numbers: 1970 = 100)


1980s and 1990s has been closely linked to thecountries' success in diversifying their export base,as well as their ability to expand their primary com-modity export volumes. As Table 2 highlights for aselected group of countries, the Asian experience ischaracterized by a marked shift toward exports ofmanufactures and strong increases in real exportearnings; at the other extreme, the situation inAfrica can be generally described as one wherecontinued reliance on primary commodity exportshas resulted in a marked and persistent deteriora-tion in real export earnings. But it is noteworthythat the group of countries in Asia and LatinAmerica that has achieved the largest increase inexport diversification has not done so at theexpense of the production of commodities; thesecountries are also among those that have increasedthe volume of commodity exports the most.

This paper reviews the major recent develop-ments in international non-oil commodity marketsand policy issues that have arisen in connection

with these developments.4 Its purpose is essentiallythreefold. First, the paper reviews some of the keycharacteristics of commodity prices that may buseful in formulating the policy response to com-modity price shocks and in assessing the outlookfor commodity prices over the near term.5 Forinstance, the usefulness of a stabilization fund willdepend crucially on whether the shocks to com-modity prices are primarily of a temporary or of amore permanent nature.6 On a similar note, the

4The technical work that provides the background for theanalysis is included in two recent studies on the subject byBorensztein and Reinhart (1994) and Reinhart and Wickham(1994). The technical background to the discussion of themacroeconomic response to uncertainty created by commodityprice shocks is provided in Ghosh and Ostry (1994).

5Unless otherwise noted, real commodity prices refer to theIMF all nonfuel commodity price index deflated by the IMF indexof manufacturing export unit values (MEUV) of industrial coun-tries. Both indices are in U.S. dollars. This is the traditional defla-tor used in analysis of international commodity prices.

6See Cuddington and Urzua (1989) and Deaton (1992).

Country

AfricaBurundiCote d'lvoireKenyaMauritaniaSenegalTanzania

AsiaMalaysiaPakistanPhilippinesSri LankaThailand

Latin AmericaArgentinaBoliviaBrazilColombiaMexicoUruguay

Fuels,minerals,

and metals

02

1394

91

342

110

11

193

91822

0

1965

Otherprimary

commodities

949377

58886

6062849986

934

83756295

Manufactures

65

1013

13

636

513

6487

165

Fuels,minerals,

and metals

010198122

5

191

1262

669163243

0

1990

Otherprimary

commodities

988070135684

3729264734

592731421360

Manufactures

210116

2211

4470624764

351453265440

Source: World Bank, World Development Report, 1992 (Washington, 1992).

3

Table 1. Structure of Merchandise Exports:Selected Developing Countries, 1965 and 1990(As a percent of merchandise exports)

Introduction


potential benefits that can be obtained from stabili-zation funds, or more generally from precautionarysavings, will be greater the more volatile and uncer-tain the environment.7 Hedging strategies, as well,acquire greater importance as volatility and uncer-tainty increase.

Second, the study provides an assessment of theeconomic fundamentals determining the decline incommodity prices. Analyses of commodity marketdevelopments have usually focused on macro-economic conditions in industrial countries as theprincipal factor affecting commodity prices.8 Tounderstand commodity price developments in thelast decade, however, it is necessary to consider theimpact of the expansion in commodity supplies andof economic developments in transition economies.The large increases in the volume of commodityproduction have apparently been linked to a num-ber of factors, including increasing openness andexport incentives in developing countries (in some

7See, for example, Ghosh and Ostry (1994).8See, for instance, Dornbusch (1985) and Chu and Morrison

(1986).

cases motivated by the debt crisis of the 1980s),productivity-enhancing technological improve-ments, agricultural policies in the largest industrialcountries, and, more recently, the surge in thexports of various metals from the former SovietUnion. In addition, this paper examines how recentmacroeconomic developments in transition econo-mies, such as the sharp contraction in aggregatdemand in the former Soviet Union, have affectedthe behavior of real commodity prices. The roleplayed by international commodity agreements ialso discussed.

Third, the paper reviews the main policy issueraised by the recent behavior of commodity prices.It re-examines to what extent there is a role forgovernment intervention and assesses the relativemerits of alternative intervention strategies. Animportant distinction is drawn between two sets ofmotives for government intervention. First, there isthe issue of how to react to temporary fluctuationsin commodity prices and to their volatility. The useof price and income stabilization funds, agriculturaltrading boards, and hedging strategies are some ofthe policies designed to deal with this type of prob-lem. Second, there are the issues raised by the sus-tained decline in the real price of commodities.This decline poses the question of the desirabilityof structural policies that foster productivity increases in the primary commodity-producing sectorand that encourage export diversification.

Concerning the characteristics of the evolutionof commodity prices, three main empirical reg-ularities emerge from the analysis. First, the recentweakness in real commodity prices appears to bprimarily of a secular, persistent nature, and notthe product of a large, temporary deviation fromtrend. Second, the relative importance of noncycli-cal or "permanent" shocks varies considerablyacross commodity groupings. Thus, while perma-nent shocks are estimated to account for only30 percent of the variance of metals prices, theyaccount for about 85 percent of the variance ofbeverage prices. And third, the volatility of com-modity prices has risen steadily and considerablysince the early 1970s, particularly in the once rela-tively stable food grouping.

The analysis of the major factors behind theobserved behavior of commodity prices revealsthat conventional factors, such as the prevailinmacroeconomic conditions in industrial countries,are estimated to have contributed only in a limitedway to the recent weakness in real commodityprices and that the expansion in the supply of com-modities played a fundamental role. In addition,while output changes in Eastern Europe and theformer Soviet Union traditionally played a rela-tively minor role in price developments, they

Table 2. Real Export Earnings, 1970-921

(Average annual percent change)

Country

AfricaBurundiCote d'IvoireKenyaMauritaniaSenegalTanzania

AsiaMalaysiaPakistanPhilippinesSri LankaThailand

Latin AmericaArgentinaBoliviaBrazilColombiaMexicoUruguay

1970-79

3.76.1

-3.0-6.6-0.9-2.7

8.92.63.81.29.9

2.59.17.47.4

11.30.5

1980-92

-3.7-8.2-5.413.23.6

-5.6

8.16.13.55.9

10.7

2.2-0.74.68.55.64.3

Source: International Monetary Fund, World Economic Outlook,various issues.

1Real export earnings defined as value of exports deflated byimport unit values.

I INTRODUCTION

4


acquired an increasingly important role in thepost-1988 period.

Regarding the policy response to temporaryprice fluctuations, a case has traditionally beenmade for governments to influence prices receivedby small commodity producers in order to smooththeir income. However, there is some evidenceindicating that private agents in developing coun-tries may be more resourceful than usuallybelieved in finding ways to smooth consumption. Incontrast, governments have shown a tendency todisplay less foresight, for example, by overspendingduring temporary booms. Thus, institutionalarrangements, such as stabilization funds, thatestablish a rule for government behavior concern-ing revenue accruing from commodity exports canbe, in principle, useful tools. However, tradingboards and other devices for governments tosmooth incomes of producers can also easilybecome distortional. In addition, the developmentof financial markets for derivative securities

(futures and options) and other kinds of commodity-related securities offers ways for exportingcountries to trade away the risk associated withcommodity price volatility at relevant time hori-zons and would diminish the need for income- orconsumption-smoothing policies.

With respect to possible policy responses tolonger-term declines in commodity prices, anenvironment conducive to export diversification iimportant, but it is not desirable to introduce dis-tortions in order to move resources away from pri-mary commodity production. Indeed, the ex-perience of the high-performing Asian countriesindicates that establishing a dynamic agriculturalsector is an important phase in the diversificatioprocess. There may, however, be a role for govern-ment in this process to the extent that informationproblems or imperfections in financial markets pre-vent prices from transmitting the correct signals, oragents from responding to them.

Introduction

5


II Characteristics of Commodity PriceBehavior

he principal aim of this section is to assess thenature of commodity price shocks and to deter-

mine to what extent the recent weakness in com-modity prices is associated with reversible cyclicalforces, or in what measure it is part of a longer-termsecular decline. In addition, some of the more gen-eral stylized facts about the behavior of commodityprices that may be useful in formulating a policyresponse to commodity price shocks are presented.These include: (a) an analysis of the persistence,duration and other features of the commodity pricecycle; (b) a quantitative assessment of the relativeimportance of "permanent" shocks in explaining thevariation in real commodity prices; and (c) an assess-ment of whether the nature of the shocks to com-modity prices has changed over recent years andwhether the volatility of commodity prices hasincreased over time.9

Trends and Cycles in CommodityPrices

As noted earlier, it is critical, when formulating apolicy response to a commodity price shock, to dis-entangle the trend from the cycle; the former isassociated with permanent (or at least very per-sistent) changes while the latter is temporary innature. Similarly, when assessing the near-to-medium-term outlook for commodity prices it isimportant to have an idea to what extent the recentweakness is cyclical, and therefore potentiallyreversible, or in what measure it appears to beassociated with more persistent secular forces thathave a lower probability of being reversed.

There are a number of statistical methodologiesfor disentangling the trend from the cycle in realcommodity prices, and the detailed results are pre-sented in Reinhart and Wickham (1994).10 For

9Quarterly data for the period 1957:1 to 1993:11 constitutedthe sample. The analysis covers the all non-oil commodity, bev-erages, foods, and metals indices.

10 A full description of the various methodologies used andthe full results can be found in Reinhart and Wickham (1994).The illustration provided here is based on the Beveridge-Nelsondecomposition technique. The shorter sample reflects the loss ofobservations associated with this procedure.

illustrative purposes, Chart 4 reproduces fromthat paper the estimated permanent or trend com-ponent for real commodity prices using one of theavailable statistical methodologies. The differencebetween the estimated permanent or trend com-ponent and the actual price series constitutes theestimated cyclical or temporary component.11 Theessence of the results, which is confirmed in majorpart by alternative estimates of the trend andcycle components, is as follows. Large deviationsfrom trend were evident in the all non-oil com-modity, food, and metals indices during 1973, atthe time of the first oil shock. For the beveragesindex, the largest deviation from trend during thesample took place during 1977 and is associatedwith a supply shock.12 With the possible exceptionof the metals basket, for which the actual price isbelow the estimated trend since late 1990, all theother indices are currently estimated to be quiteclose to their permanent or trend component.This observation would seem to suggest that themajor part of the price decline during recent yearsis due to more persistent causes than, say, a cycli-cal downturn in economic activity in the majorindustrial countries.

The evolution of the permanent component alsochanges considerably during the sample period.While during the 1960s and up to the first oil shockall prices were relatively stable, the trend becomesmarkedly negative from the mid-1970s onwards (forbeverages somewhat later). In addition, the cyclesare correlated across commodity groupings. Thetemporary shocks that drive the cycle can exhibitdiffering degrees of persistence. From a policy per-spective, the difference between permanent changes

11It should be emphasized that the available methodologiesattempt to determine ex post the nature of commodity priceshocks; this does not imply that recent trends can simply beextrapolated into the future. The trend is not necessarily asimple linear function, but can itself vary over time. SeeChart 4.

12 Since coffee has a large weight in the beverage index, thisprice spike is associated with a severe frost in Brazil that didconsiderable damage to the trees.

6

T


Size of the Permanent Component of Price Shocks

Chart 4. Real Non-Oil Commodity Prices: Trends and Cycles,1962:1-1993:11

Sources: IMF, International Financial Statistics; and Reinhart and Wickham (1994).

and temporary, but highly persistent ones may notbe that significant as attempts to smooth out theirconsequences would run into similar difficulties.A n analysis of the duration and persistence of thecycle indicates that it may take quite some time fora temporary shock to fade away almost completely.Shocks to the metals index are the least persistent(becoming arbitrarily small in 6 to 7 quarters) andmost persistent for beverages (12 to 13 quarters).13The temporariness of the cycle also seems to varyconsiderably across individual commodities (Cud-dington (1992)).14

In sum, there is no compelling evidence to sug-gest that the price weakness that prevailed through

13See Reinhart and Wickham (1994).14It is worth noting that the duration of the cycles for the all-

commodity index corresponds rather well with the duration ofthe business cycle (as defined and measured by the NationalBureau of Economic Research) in the United States.

mid-1993 was part of an anomalous cycle. Thismeans that the decline in commodity prices of thepast decade should be interpreted as being largelypermanent, in the sense that there are no forces inplay to reverse that decline automatically, as wouldbe the case with purely cyclical fluctuations. Itshould be noted, however, that the permanent ortrend component of commodity price movementscan itself change over time and be partly random.While there is some evidence that the downwardtrend has steepened in the recent past, it cannot, ofcourse, be ruled out that factors influencing thetrend will be more positive in the future.

Size of the Permanent Component ofPrice Shocks

While Chart 4 provides a useful visualization ofcyclical and trend movements in real commodity

7


prices, it does not provide a quantitative measureof the relative importance of permanent shocks inexplaining the variance of changes in real com-modity prices. Table 3 summarizes the main find-ings on the extent to which price changes areattributable to permanent price shocks. For each ofthe four commodity indices, permanent shocks playa role in explaining the variance of commodityprices. However, their relative importance variesconsiderably across commodity groupings. Perma-nent shocks are least important for metals(29 percent of the variance of yearly changes) andmost important (85 percent) for beverages. The allnon-oil commodity index, reflecting the differingcharacteristics of its components, falls in the middleof the range, with permanent shocks accounting forabout 45 percent of the variance of yearlychanges.15 For the two price indices with thelargest permanent component—beverages andfood—there is evidence that even the temporarycomponent of these series exhibits a high degree ofpersistence. The latter observation would indicatethat, even when the shock is temporary, the costsassociated with a price (or incomes) stabilizationpolicy could be quite large.

Increasing Volatility of CommodityPrices

In addition to assessing whether a commodityprice is driven primarily by permanent or transitoryshocks, it is important to consider other aspects ofthe nature of the shocks when designing policy. Forinstance, even if all shocks are temporary, there arerelatively few gains from setting up a stabilizationfund and/or hedging if the variance of prices issmall and large shocks are rare. The benefits thatcan be obtained from stabilization funds or, moregenerally, from precautionary savings will increasein a more volatile and uncertain environment, asgreater uncertainty in a country's export revenuestream increases the value to that country ofaccumulating assets as a means of protecting itselfagainst future shocks.16 Similarly, hedging strat-

15This result is in line with the findings of Cuddington andUrzua (1989), who, using annual data for an all non-oil com-modity index for the period 1900-83, find that about 39 percentof the shocks to real commodity prices are permanent.

16See Ghosh and Ostry (1994).

egies acquire greater importance as volatilityincreases and the probability of large, destabilizingshocks becomes higher.

Statistical analysis indicates that shocks to com-modity prices have been larger in the latter part ofthe sample than in the comparatively stable 1950sand 1960s. Several results of this exercise are worthnoting. First, the average price is markedly lowerduring the most recent sample, consistent with thepresence of a negative trend, or a series of down-ward jumps in average prices. Second, there is a sus-tained and sharp increase in the variance ofcommodity prices.17 This is evident in all the indicesbut is most pronounced in the all non-oil com-modities and food groupings; for the all non-oil com-modities index volatility increases fivefold, while forfood the increase in the volatility measure is sixfold.These results can be observed in Chart 5, whichplots the volatility indicator over the 1972-93period. Following the relative tranquillity of com-modity prices during the 1950s and 1960s, the sharpincrease in volatility in the early 1970s followed aperiod of expansive monetary policy in some indus-trial countries, the breakdown of the Bretton Woodsexchange rate system, and came on the heels of thefirst oil shock. However, volatility has apparentlyremained high during the 1980s and 1990s.18

17This feature has also been discussed in Deaton (1992) andDeaton and Laroque (1992).

18It is interesting to note that from a long-term perspectivethe relative tranquillity of the 1950s and 1960s contrasts quitesharply with commodity price behavior prior to World War IIand after 1974. For a discussion of volatility over the longerterm, see Boughton (1991).

1Annual data were used. All commodity price indices are de-flated by the export unit values of manufactures. A fuller descrip-tion of the methodology used and more detailed results arepresented in Reinhart and Wickham (1994).

II CHARACTERISTICS OF COMMODITY PRICE BEHAVIOR

8

Table 3. Share of the Variance of Changesin Real Non-Oil Commodity Prices Due toPermanent Shocks, 1957-921

All Commodities

0.448

Beverages

0.846

Food

0.731

Metals

0.287


Increasing Volatility of Commodity Prices

Chart 5. Rising Volatility in Real Non-Oil Commodity Prices,1972:1-1993:11(Coefficients of Variation)

Sources: IMF, International Financial Statistics; and Reinhart and Wickham (1994).Note: The coefficient of variation is based on 15-year moving averages and 15-year standard deviations that are backward looking, hence the

observation for 1972:1, for instance, is based on data from 1957:11-1972:1.

9


Ill Determinants of Commodity PriceMovements

ssessing the evolution of commodity prices byapplying time-series methods has shed some

light on the persistence and increased volatility ofshocks in commodity markets. This section comple-ments those findings by examining the economicfundamentals that determined the decline in com-modity prices and their relative importance. To theextent that such economic fundamentals explain asubstantial part of the documented decline in com-modity prices, this analysis may assist in the assess-ment of the near-term prospects in commoditymarkets.

A number of reasons have been put forward toexplain the weakness in commodity prices. Amongthose explanations, the response in developingcountries to the effects of the debt crisis of the1980s and, more recently, the economic develop-ments in the economies in transition in EasternEurope and the former Soviet Union stand out asmajor shocks that are likely to have exerted consid-erable influence on commodity markets. To evalu-ate those factors in a more systematic way, thissection will also make use of the results from asimple aggregate econometric model of commodityprice determination (Borensztein and Reinhart(1994)). Within the framework set out in thatpaper, the distinct importance of supply-side fac-tors, in addition to demand factors, in the recentdecline in commodity prices becomes evident.

Economic Growth and the Demandfor Commodities

Analyses of commodity market developmentshave usually focused on the macroeconomic condi-tions in industrial countries as the principal factoraffecting commodity prices. Given the role of manycommodities as inputs for production by manufac-turing industries, their demand is closely related tothe level of industrial economic activity, the majorpart of which takes place in industrial countries.19

Therefore, the trend decline in commodity pricesis, at least partially, related to the secular slowdownin the growth of real output in the industrial coun-tries that has been observed since the early 1970s,with 1973 marking the end of the strong postwarexpansion phase for most countries.20

In addition to the slowdown in growth, a declin-ing intensity of resource use for some commodities,particularly in industrial countries, has contributedto reduced rates of growth in commodities demand.Since 1974, for example, the growth in consump-tion of six major metals by the Group of Sevenindustrial countries has decelerated sharply and, insome cases, consumption has declined in absoluteterms (Table 4). While there are some differencesbetween countries and between particular metals,the consumption slowdown is quite general and islinked to the slowdown in real economic growth.However, Table 4 clearly shows the decline in theintensity of metals use as indicated by averageannual changes in consumption per unit of real out-put (real GDP). Thus, in addition to the slowdownor decline in demand growth for metals engen-dered by lower real output growth, these indicatorsof intensity of metals use provide summary infor-mation suggesting the effects of longer-term trendsin technological innovation or in structural factorsthat govern metal resource usage in the industrialcountries.21

Various explanations have been put forward forthe trends observed in the intensity of metals use inthe industrial countries. Some of these explana-tions stress the changing composition of output: theshift away from manufacturing toward services; thedeclining share in GDP of gross domestic invest-ment, including structures, which is more metalintensive than other components of expenditure;and, more recently, declining output in the

19Even with an increasing fraction of industrial activity takingplace outside the group of industrial countries, indicators ofeconomic activity in that group are still relatively good proxiesfor world demand because of international linkages.

20Perron (1989), among others, dates the start of the slow-down in gross domestic product (GDP) growth at the first quar-ter of 1973. For a discussion of the factors behind the slowdownin growth, see, for example, Adams, Fenton, and Larsen (1987).

21Tilton (1990) has shown that this pattern holds morebroadly for member countries of the Organization for EconomicCooperation and Development.

10

A


Economic Growth and the Demand for Commodities

Consump-tion

1962-73

RealG D P

Intensityof use1

Consump-tion

1974-81

RealG D P

Intensityof use1

Consump-tion

1982-91

RealGDP

Intensityof use1

AluminumUnited StatesJapanFranceGermanyItalyUnited KingdomCanada

9.320.3

7.49.6

10.55.28.6

3.59.54.94.25.33.15.3

4.113.03.16.37.71.71.4

-0.61.13.03.24.1

-4.11.3

2.13.62.51.93.40.83.8

-2.9-2.6

0.41.10.3

-4.9-2.5

0.34.93.23.05.02.54.5

2.34.22.32.22.32.52.4

-2.10.70.90.72.6—2.0

CopperUnited StatesJapanFranceGermanyItalyUnited KingdomCanada

4.711.24.62.73.40.46.1

3.59.54.94.25.33.15.3

-0.64.2

-0.20.20.3

-5.1-2.5

0.41.81.20.72.9

-5.40.8

2.13.62.51.93.40.83.8

-1.9-2.0-1.3-1.3-0.5-6.2-3.0

0.62.81.33.02.7

-2.0-1.4

2.34.22.32.22.32.52.4

-1.8-1.4-1.1

0.80.3

-4.4-3.9

LeadUnited StatesJapanFranceGermanyItalyUnited KingdomCanada

3.56.93.13.59.00.33.8

3.59.54.94.25.33.15.3

-0.7-1.50.8

-0.76.9

-4.2-1.7

-2.06.1

-0.3-0.3

2.2-0.410.7

2.13.62.51.93.40.83.8

-4.12.2

-2.8-2.3-1.4-1.1

6.6

1.11.12.02.40.20.1

-3.8

2.34.22.32.22.32.52.4

-1.2-3.0-0.3

0.1-2.1-2.4-6.2

ZincUnited StatesJapanFranceGermanyItalyUnited KingdomCanada

7.111.44.03.37.91.57.6

3.59.54.94.25.33.15.3

2.02.31.1

-0.33.0

-2.00.2

-5.9-1.00.4

-1.21.1

-4.80.9

2.13.62.51.93.40.83.8

-8.0-4.5-2.2-3.3-2.5-5.7-2.8

0.32.00.83.92.8

-0.3-0.5

2.34.22.32.22.32.52.4

-2.1-2.1-1.5

1.70.5

-2.7-2.9

TinUnited StatesJapanFranceGermanyItalyUnited KingdomCanada

1.38.61.7

-0.24.2

-1.44.2

3.59.54.94.25.33.15.3

-3.51.6

-2.3-0.2-1.3-4.6-2.0

-4.2-2.1-3.3-1.9-6.8-5.5-4.0

2.13.62.51.93.40.83.8

-6.3-5.5-5.7-3.9

-10.2-6.3-7.5

-0.41.3

-0.84.03.3

-0.5-2.5

2.34.22.32.22.32.52.4

-2.8-2.7-3.0

1.81.0

-2.9-4.8

NickelUnited StatesJapanFranceGermanyItalyUnited KingdomCanada

5.017.36.88.8

10.32.38.3

3.59.54.94.25.33.15.3

-1.09.20.24.27.8

-4.30.6

-2.11.42.83.1

-0.4-3.2-0.4

2.13.62.51.93.40.83.8

-4.2-2.2

0.20.9

-3.9-4.1-4.0

-0.56.01.71.45.43.78.2

2.34.22.32.22.32.52.4

-2.81.7

-0.6-0.9

3.01.15.4

Table 4. Intensity of Metals Use in Major Industrial Countries(Average percentage changes)

Source: World Bank.1 Intensity of use measured as the average annual change in the ratio of consumption to real GDP.

11


III DETERMINANTS OF COMMODITY PRICE MOVEMENTS

defense-related industries. Other explanationsfocus on the role of resource-saving technologiesand the process of materials substitution. Highrates of growth in metals consumption during the1960s were by the early 1970s generating wide-spread fears about the future availability and priceof depletable resources (such as nonferrous met-als), the so-called "Limits to Growth" hypothesis.22Rising metals prices in the late 1960s, reinforced bythe price spike in 1973 that was one of the eventssurrounding the first major oil shock, are believedto have set in motion a surge in resource-savinginnovation and material substitution that is notlikely to be reversed.23

Much less information is available on intensity ofmetals use outside the industrial countries. For theformer Council for Mutual Economic Assistance(CMEA) countries, Dobozi (1990) shows that theperiod of growing metals consumption and risingintensity of use also came to an end in themid-1970s as output growth in the former C M E Acountries also slowed. However, rather than fallingas in the industrial countries, the intensity of metalsuse stagnated or showed little change beforedeclining quite sharply in the 1980s. With regard todeveloping countries, the rapid pace of industrial-ization in the Asian region and in parts of LatinAmerica in the 1960s through to the early 1980s hasbeen linked with rising intensity of metals use, andappears to be related to the relatively high levels ofgross domestic investment being maintained duringthis period.24 The subsequent picture is moremixed as external financing difficulties loweredinvestment and growth rates in many developingcountries, particularly in Africa and Latin Amer-ica. Still, the observed trend toward less metal-intensive technologies in industrial countriesimplies that overall metals demand has not grownas rapidly, contributing to the overall weakness incommodity prices.

Expansion in Supply

A remarkable feature of developments in com-modity markets in the 1980s has been the vigorousgrowth in the volume of imports of commodities(Chart 6). Since 1983, the volume of imports of

22Meadows and others (1972).23There is evidence consistent with the hypotheses. Tech-

nological progress, for example, has over time permitted metalbeverage and food containers (chiefly aluminum and tinplate) tobe made much thinner. In the area of materials substitution,plastics, ceramics, and composites have increasingly been used,for example, in automobile manufacturing as automakersendeavored to make lighter and more fuel-efficient cars.

24See Radetzki (1990).

12

Chart 6. Factors Affecting CommodityMarkets

Sources: IMF, International Financial Statistics and World EconomicOutlook; and Borensztein and Reinhart (1994).

non-oil commodities by industrial countries hasalmost doubled even though, during the sameperiod, GDP of the industrial nations grew by lessthan 30 percent. Imports of non-oil commoditiesalso grew faster than those of other goods, as worldimports of all types of goods increased by approx-imately 70 percent in real terms during the sameperiod. This large increase in the volume of com-modity production and trade suggests the impor-tance of supply-side factors in explaining pricedevelopments. These data suggest that the declinein prices cannot be explained by demand factorsalone, and has followed from the rapid expansionin supply outstripping the growth of demand fornon-oil commodities.

A number of factors contributing to this expan-sion in supply have been identified. For example, inthe early 1980s, many developing countries facedconsiderably more restricted borrowing oppor-


tunities in international financial markets as thedebt crisis unfolded. This situation required bal-ance of payments adjustment, which brought aboutpolicies geared to encouraging exports and resultedin an expansion in commodity supplies in manydeveloping countries.25 It could be argued that thedebt crisis may have only accelerated a necessaryshift from import substitution toward export-oriented development strategies. Moreover, theexpansion of supply has been promoted by the pro-cess of structural economic reform in manydeveloping countries (particularly in Asia andLatin America). In particular, as countries opentheir economies to international trade and adjusttheir economic policies in a more market-orienteddirection, resources tend to flow towards produc-tive sectors with comparative advantage, whichinclude exportable goods and, in the case of manydeveloping countries, primary products.

In addition, a continuous process of technologi-cal innovation and productivity enhancement hascontributed to a persistent expansion of supplies ofnumerous commodities, both in developing andindustrial countries. While there are marked dif-ferences between commodities and quantificationis difficult, there are a number of important exam-ples of where these technology-driven shift factorscan be shown to be important.

The diffusion of information about prices,techniques, and marketing opportunities can bringabout significant increases in overall commoditysupplies through new entrants into internationalmarkets. The emergence during the 1980s of, first,Malaysia and then Indonesia as major cocoa pro-ducers is one of the prime examples of this phe-nomenon.26 Whereas producers other than thesetwo countries on aggregate increased productionby about 15 percent between 1980-81 and 1991-92,Malaysia and Indonesia increased their combinedraw cocoa output by 500 percent in the sameperiod; their output share rose from just under4 percent to 17.5 percent.27

The most remarkable examples of informationdiffusion and productivity growth are found in foodcrops. Despite demand growth, particularly due tohigh rates of population growth in developingcountries, the rise in production has been such thatreal food prices have, with the exception of a briefepisode in the mid-1970s, declined significantly in

25See, for example, Aizenman and Borensztein (1988) andReinhart (1991).

26For details, see Reinhart and Wickham (1994).27Cote d'Ivoire also expanded cocoa production in the 1970s

and 1980s and increased its market share. However, bothGhana's and Nigeria's share declined significantly.

the postwar period.28 While land under cultivationhas increased in many parts of the developingworld, a sustained increase in yields over time hasplayed a major role in the agricultural supply-demand equation. Advances in agroscience thathave led to the development of higher-yield andmore disease-resistant varieties, the increased useof fertilizers and pesticides, and the spread of irri-gation, are the principal factors cited for theincreases in cereal yields (see Chart 7). Theimprovement in yields was a more or less contin-uous process in the agricultural sectors of theindustrial countries, but in the developing countriessharp increases occurred more recently. High-yielding wheat and rice varieties were introducedin a number of developing countries and the rate atwhich high-yielding varieties were adopted wasoften rapid.

The improvements in technology are likely to be"irreversible" in nature and are not limited to a fewcommodities. These developments have had wide-spread effects and have led to improving yields formany types of crops, including the beverages(cocoa, coffee, tea), oilseeds, vegetables, and agri-cultural raw materials, such as cotton. It seemslikely, particularly in light of recent major advancesin biotechnology, that in the future as in the pastinnovation and information diffusion will continueto play a significant role in increasing agriculturalyields still further.

National (or transnational in the case of theEuropean Union (EU)) agricultural policies in theindustrial countries typically have also acted tostimulate output and discourage consumption,thereby reducing import needs or increasing sup-plies available for export. Such government inter-vention has been a common feature in industrialcountries in the postwar period; however, a markeddeterioration in the agricultural trade environmenttook place in the 1980s in part owing to increasedsubsidies to agriculture. Prices of many agriculturalcommodities fell considerably, the budgetary andother burdens associated with price supportschemes grew, and trade disputes becameincreasingly acrimonious between agriculturalexporters, both developed and developing. Thecase of wheat provides an instructive example.Price support under the EU's Common Agri-cultural Policy has been so effective that wheatsupplies surplus to European requirements havegrown over time and, assisted by the provision ofexport subsidies, have had to be sold on world mar-kets. Between 1980-81 and 1991-92, wheat exports

28The price of services such as transportation, packaging, pro-cessing, and retailing has generally been rising over time, tend-ing to offset the impact of lower "raw" food prices on retail foodprices.

Expansion in Supply

13



Chart 7. Cereal Yields by Region(In thousand kilograms/hectare)

Source: Food and Agriculture Organization, ProductionYearbook, 1992 (Rome: FAO, 1993).

Note: For country groupings refer to Table A2.

from the E U rose by 55 percent to nearly 22 milliontons (an increase in market share from 14 percentto 20 percent). In order to counter E U penetrationin its traditional markets, the United States intro-duced the Export Enhancement Program in 1985.The so-called "subsidy war" contributed to lowerinternational wheat prices to the detriment of otherwheat exporters.

While trade in wheat has been perhaps the mostnotable source of contention, other grains andcommodities (e.g., meat, sugar, oilseeds) have beenaffected by price support schemes. There is littledoubt that the agricultural policies of industrialcountries have been a factor in the rapid expansionin world commodity supplies during the 1980s,which has contributed to the weakness in prices forcertain agricultural products. Liberalization mea-sures, including a phased reduction in export sub-sidies to be implemented under the recently

14

concluded Uruguay Round, should improve priceprospects for a number of agricultural products.29

International Commodity Agreements

The weakening of commodity prices during the1980s and 1990s has often been attributed in part tothe breakdown of international commodity agree-ments (ICAs) designed to help stabilize prices andensure a reasonable rate of return to commodityproducers.30 Several such agreements were ineffect during the 1970s and early 1980s, but in June1984 the agreement covering sugar collapsed.Thereafter, other agreements followed suit orbecame economically moribund—tin (October1985), cocoa (February 1988), coffee (July 1989),and natural rubber (April 1993). The immediateimpact on price of a breakdown in a commodityagreement may have a large "temporary" compo-nent as, for instance, when beverage prices fellbelow trend from mid-1987 to mid-1989(Chart 4).31 However, the effects of the breakdownare likely to have an important "permanent" com-ponent, as the behavior of exporters who partici-pated in the ICA adjusts to the more competitiveregime. In part, however, the breakdown of thenumerous agreements reflects the difficulties of try-ing to influence prices by managing output or byother means in the face of developments (such asproductivity increases) acting to foster an expan-sion in supplies.

Developments in Economies inTransition

Since 1990, another major shock affecting com-modity markets has been the aftermath of eco-nomic developments in Eastern Europe and,particularly, the former Soviet Union. These coun-tries are large participants in international trade incommodities, both as importers (especially ofgrains) and as exporters (particularly of metals).Their demand for imported commodities fell con-comitantly with the fall in output and aggregatedemand that followed the collapse of the centrallyplanned economic systems. Some examples of the

29Most partial and general equilibrium models suggest thatreal prices for most agricultural products would rise if reformswere implemented by the industrial countries. See Goldin andKnudsen (1990).

30See, for example, Maizels (1992). Conversely, it can beargued that it was the large expansion in supplies by countriesoutside the agreements that contributed to the ICAs' demise.

31Unless the breakdown of the ICA was fully anticipated, inwhich case, prices would have already reacted.


decline in imports of commodities by the formerSoviet Union are shown in Table 5. But the impactof the Eastern European shock was probably morepronounced in the metals markets, where the for-mer Soviet Union is an important supplier, andwhere dramatic increases in exports were observedin some cases. For example, zinc exports from theformer Soviet Union rose by nearly 700 percentduring 1989-92.

As can be seen in Table 6, the increase in exportsof metals was to a large extent a reflection of thelarge decline in domestic demand for metals in thecountries of Eastern Europe and the former SovietUnion, which responded to the declines in the levelof activity in the defense industry and in otherpoorly competitive manufactures, and to disrup-tions in C M E A and interrepublican trade. Otherfactors may have also contributed to the increase inthe volume of exports of metals, principally theincreased profitability in energy-intensive metalsproduction and exports (due to the still very lowdomestic price of energy), a reduction in stock lev-els that are no longer justified from nationalsecurity or economic standpoints, and some exportactivity conducted principally to build up holdingsof foreign assets.

In some cases, discrepancies between domesticand international prices in the context of only par-tial price and trade liberalization have createdarbitrage opportunities and contributed to thesharp increase in exports to western markets.32

Nickel prices in late 1992, for example, wereUS$1,800 per ton for domestic deliveries to Rus-sian state enterprises, US$3,000 per ton on theMoscow Commodity Exchange, and nearlyUS$6,000 per ton on the London Metal Exchange.Not surprisingly, smuggling and bypassing of regu-lar domestic and export trade channels reportedlyflourished during the first phase of the reform pro-cess, aided by the growth in unmarked andunprotected international borders between thenewly independent states. The Baltic countries arereported to have become leading exporters of alu-minum, copper, and nickel from the former SovietUnion, which suggests that much of the irregulartrade took place via Russia's traditional conduits tothe West.33 Overall, developments in EasternEurope and the former Soviet Union have gener-ated increases in exports of metals and drops inimports of grains and other commodities, contrib-uting to the observed weakness in the aggregateprices of primary products.

In general terms, it does not appear thatdevelopments in Eastern Europe and the formerSoviet Union will affect commodity prices in a per-manent way. While the restructuring of industrialproduction into a more marketable compositionmay take some time and gains in the efficiency ofresource use can be expected, it will eventuallyinduce a recovery in domestic demand; unnecess-ary stocks of metals will be run down, the eventualadjustment of domestic energy prices will reduceexcess profitability, and a more complete price andtrade liberalization—along with the privatizationof industry—will eliminate incentives for arbitrage

32Effective early in 1993, domestic prices for metals were fur-ther liberalized and other steps were also taken to reduce thescope for illegal trading.

33It seems quite likely that the customs data reported inTable 6 underestimate former Soviet Union exports to westernmarkets. Some industry sources estimate exports of aluminumfrom the former Soviet Union in 1992 at about 1.2 million tons,of which it is estimated that 0.4 million tons were smuggled out.

Developments in Economies in Transition

Table 6. Supply of Selected Commoditiesfrom the Former Soviet Union

Table 5. Demand for SelectedCommodities of the Former Soviet Union

15

ImportVolumes

Cocoa1

CornTeaWheat2

Percent Change1989-92

-48.1-62.7-55.7-17.0

Former Soviet UnionShare of WorldImports, 1989

(In percent)

4.826.026.921.3

Sources: E.D. and F. Man Cocoa Ltd., Cocoa Market Report(London, various issues); International Tea Committee, AnnualBulletin of Statistics (London, various issues); and U.S. Departmentof Agriculture, World Grain Situation and Outlook (Washington,various issues).

1Grindings of raw cocoa.2Percent change is through November 1993.

ExportVolumes

AluminumCopperZinc

Percent Change1989-92

219.471.2

686.0

Former Soviet UnionShare of WorldExports, 1992

(In percent)

8.35.42.2

Source: World Bureau of Metal Statistics, World Metal Statistics(London, various issues).



or capital flight-motivated exports. However, it isdifficult to predict the level of metals exports thatwill be sustained once the transformation of theeconomies of the former Soviet Union republics iscompleted.

Econometric Model Results

The main conclusions from a recent econometricstudy (Borensztein and Reinhart (1994)) are thatthe weakness in commodity prices since 1984 can-not be explained by the two factors that have beenconventionally linked to the determination of com-modity prices: the level of economic activity inindustrial countries and the real exchange rate ofthe U.S. dollar. Instead, a more accurate picture isobtained by including proxy variables that wouldtake into account two main developments: thebooming supply of exports of primary products,and the change in the net international demand forcommodities of the former Soviet Union.

The improvement in predictive performance canbe observed in Chart 8, which displays the threeestimated specifications. Model 1 includes conven-tional demand-side determinants only; model 2adds a supply proxy to the estimated equation; andmodel 3 extends model 2 by incorporating the tran-sition economies into the measure of worlddemand. The poor forecasting performance ofmodel 1 after 1984 and the much more adequateperformance of models 2 and 3 are evident fromthis chart.

The econometric results also permit an estimateto be made of the relative importance of the dif-ferent factors that have determined the decline inthe prices of commodities during the last decade. Itsuggests, for example, that industrial production inindustrial countries explained 25 percent of thevariation of commodity prices in the period 1971-84, but just over 5 percent for 1985-88. By contrast,

16

the supply expansion explained about 40 percent ofthe variation for the period 1971-84, and over60 percent for the period 1985-88.

The current trend toward recovery in growth inindustrial countries will help to relieve downwardpressures in commodity markets, but it appearsunlikely that it will reverse the declines of the pastdecade. On the basis of the econometric results,5 percent growth in industrial countries over thenext two years (consistent with current IMF projec-tions) would generate an increase of between 6 1and 9 percent in real commodity prices over thesame period. For that increase to materialize, how-ever, it would be necessary that other factors, inparticular the trend increase in supplies and theeconomic developments in the former SovietUnion, exert a neutral influence on commoditymarkets over the next two years.

Chart 8. Real Non-Oil Commodity Prices:Actual and Forecasts

Source: Borensztein and Reinhart (1994).


IV Policy Issues

he previous sections of the paper have exam-ined how commodity prices have behaved

over recent years and discussed possible explana-tions for the observed behavior. Both the volatilityin real commodity prices and the downward trendexperienced over recent years present serious chal-lenges for many developing countries because ofthe impact on export earnings, domestic incomes(and hence private savings and investment), andgovernment budgetary positions. The response tovolatility in commodity prices, however, gives riseto issues that are quite different from those ema-nating from secular changes in the average level ofprices. It is therefore crucial to interpret correctlymarket developments in each case in order to elab-orate the appropriate policy response or, in fact, toconsider whether a policy response is warranted atall.

Temporary Price ShocksTemporary shocks to commodity prices affect

both the private and public sectors in thecommodity-producing countries. In addition, com-modity price shocks can have large effects on themacroeconomic and balance of payments positionof a commodity exporter, and also affect a coun-try's access to private international financial mar-kets because of sovereign risk considerations.When facing temporary changes in income, theoptimal response by individual consumers is to usesaving and borrowing in an attempt to smooth thepath of consumption. Similarly, governments facingtemporary changes in revenue should adjust theirspending to be consistent with sustainable levels.

Traditionally the argument has been made thatgovernments should assume most of the con-sequences of temporary commodity price changesand take an active role in helping to smoothincome for private producers. These policy actionsare often implemented through special institutionalarrangements such as agricultural boards, withprice stabilization as a major objective. Further-more, international efforts to influence worldprices through buffer stock operations and/or

export restraints have been pursued through inter-national commodity agreements. Another reasonfor intervention is that in some cases, particularlyfor countries with a heavy dependence on mineralsand metals, there is a concern with how best tohandle fluctuations in government revenues.34

Role of Government in Smoothing Income

Before examining some of the policy prescrip-tions that have been advanced to help deal withinternational commodity price variability and itsmacroeconomic effects, it is worth consideringwhether it is desirable to split the effects of tempo-rary commodity price movements on private pro-ducers and on government.

A traditional argument for government toabsorb the effects of commodity price fluctuationsis based on microeconomic considerations relatingto the behavior of private agents.35 If privateagents bear all the commodity price risk (and sup-ply is inelastic in the short run), they will face vari-able and unpredictable income streams over time,with possibly very limited opportunity to smoothconsumption. Therefore, it may be optimal for thegovernment to intervene to stabilize the pricesreceived by private agents and thus enhance thewelfare of risk-averse individuals by permittingsmoother income and consumption streams. How-ever, some additional assumptions are necessary tosupport this argument. For example, it can beargued that producers face liquidity constraints,because of underdeveloped financial systems orowing to information problems, so that it is notpossible for them to borrow when commodityprices are temporarily low.36 The implicit assump-

34An additional argument for government intervention that issometimes put forward is that large temporary shocks if fullypassed on to producers may induce persistent (or permanent)effects on supply. In the case of tree crops, for example, replace-ment plantings may cease, leaving productive capacity reducedfor a lengthy period of time and resulting in loss of marketshare.

35See Newbery and Stiglitz (1981) and Deaton (1992).36But note that governments may also face liquidity con-

straints in international financial markets because of marketimperfections or sovereign risk considerations.

17

T


tion may be that individual agents lack the fore-sight or the means to engage in suitable savingsstrategies, or they lack the information to form rea-sonable expectations about the nature of com-modity price shocks. Or alternatively, theassumption may be that income levels in manycountries are so low that consumption spending byagricultural smallholders, for example, is close tosubsistence levels with the result that the ability ofproducers to engage in precautionary saving isminimal.

Empirical evidence on savings behavior indeveloping countries is far from conclusive, but itdoes suggest that consumption smoothing takesplace to a larger extent than traditionally assumed.At the macroeconomic level, one strand of litera-ture has found that consumption behavior in somedeveloping countries has often been close to whatwould be predicted on the basis of a standard per-manent income model.37 The implication of thesestudies is that, in the face of fluctuations in incomestreams due to commodity price shocks and otherdisturbances, consumption paths in a number ofdeveloping countries have still been relativelysmooth. As might be expected, however, this seemsto be more the case for countries with higherincome levels than for low-income countries inAfrica and elsewhere.

The issue of whether developing countries havebuilt up precautionary financial balances to act as abuffer or insurance policy in the face of changes inthe underlying volatility of their export earningshas also been examined.38 According to the pre-cautionary savings hypothesis, a consumer wouldbuild up a larger stock of financial assets the morevolatile or unpredictable is his or her income. For acountry as a whole, the volatility of export reve-nues creates an aggregate need for precautionarysavings, which can only be accumulated through amore positive current account balance (savingsminus investment). Looking at aggregate data forthe current account for a large panel of developingcountries, it has been found that the volatility ofexport earnings had a significant effect on the cur-rent account outcome, in conformity with the pre-cautionary savings hypothesis. These resultssuggest that a considerable number of developingcountries do indeed attempt to build up precau-tionary savings balances when they perceive anincrease in the volatility of their export earnings.Note that, since the current account is the result ofdecisions by both private agents and the govern-ment sector, these test results apply to the aggre-

gate behavior of the private and public sectorstoward the volatility of export revenues.

At the microeconomic level, although the avail-ability of data seriously limits the number of casesthat can be studied, the evidence that agents reactto commodity price instability in a manner consis-tent with theory is perhaps even more suggestive.This is because microeconomic survey data providea better description of individual saving behaviorthan aggregate macroeconomic data. It has beenfound that rice farmers in Thailand, for example,smooth their consumption quite successfully bothwithin and between harvest years.39 Evidence fromGhana further suggests that smallholders are capa-ble of making sensible intertemporal choicesbetween consumption and savings.40 Also, an anal-ysis of the coffee boom in Kenya of 1976-79 foundthat as much as 60 percent of boom income wentinto private savings; farmers were aware that theboom was due to frost damage in Brazil and thattheir income gains were largely of a windfallnature. In contrast, it was estimated that the Ken-yan government increased its spending by morethan the increase in revenue generated by the cof-fee boom.41 At a minimum, the evidence at boththe macroeconomic and microeconomic levels sug-gests that governments may be well advised toreassess their intervention strategies and theresponsibility they assume for managing com-modity price risk at the country level.

All in all, while in principle a case can be madefor a government role to help smooth the incomesof private commodity producers because of finan-cial market imperfections and information prob-lems, there is empirical evidence to support theview that the private sector often tends to find waysof smoothing consumption in the face of com-modity price shocks. Moreover, the experiencewith commodity price booms indicates that govern-ments can find it difficult to manage properly theproceeds of temporary revenue windfalls. Thus, tothe extent feasible, it seems advisable to adopt thepolicy of not interfering with price signals andusing other means to deal with the side effects ofprice shocks. For example, the effects of com-modity price fluctuations on low-income producerscould be better addressed by resorting to socialsafety net measures.

Intervention Strategies

In part for the reasons discussed above, manygovernments in commodity-producing countries

37See Ostry and Reinhart (1992) and Ghosh and Ostry(1993).

38See Ghosh and Ostry (1994).

39See Paxson (1992).40See Ingham (1973).41See Bevan, Collier, and Gunning (1989). Cuddington (1989)

examines the fiscal difficulties experienced by Kenya, Nigeria,and Jamaica associated with commodity export booms.

IV POLICY ISSUES

18


have chosen to assume a role in attempting to dealwith the effects of fluctuations in internationalcommodity prices on the domestic economy. Insome metal- or mineral-exporting countries, themining and refining companies are often partiallyor wholly owned by the state, which in addition toreturns from holding equity may also levy exportand income taxes. In countries exporting primaryagricultural commodities, many governments levysignificant export taxes that take on various forms(specific, ad valorem, adjustable levies) so thatcommodity price fluctuations may generate reve-nue instability. In other cases, state interventiontakes the form of operating parastatal marketingboards. These boards typically set producer pricesconsiderably below international prices and do notchange them for lengthy periods of time inresponse to changes or trends in internationalprices. These kinds of arrangements tend to stabil-ize (at a lower level) the incomes of private agentswhile the fluctuations in international commodityprices are transmitted directly to the general bud-get via export taxes and/or the operating surplus ofthe parastatal marketing board. In some cases,however, the impact of commodity price fluctua-tions impinges in the first instance on the incomesof commodity-producing agents in the economy,and only indirectly on the fiscal position.42

The institutional structure can thus have animportant influence in determining the aggregateimpact on the economy of changes in internationalcommodity prices and how macroeconomic stabili-zation and adjustment are effected. The nature ofthe intervention performed by these institutionshas been varied, but the most often-used strategiesare outlined below.

Stabilization Funds

Stabilization funds have been designed to dealwith the impact of commodity price volatility ongovernment revenues, especially in countries with aheavy dependence on metals and minerals. Thesefunds act as a buffer for government expendituresand financing and have at least two appealing fea-tures. First, they impose a rule on governments,designed to use resources optimally from a long-term perspective, even if the effective planninghorizon of the government is much shorter. Therule should lessen the risk that governments maybe tempted to increase spending excessively during

temporary booms. And second, they constitute aninstitutional device by which governments can evenout their expenditure, given that countries, as wellas individuals, may be limited in their ability tosmooth spending by imperfect access to interna-tional financial markets.

One scheme of this type has been put in place inPapua New Guinea whereby government revenuesfrom mining are placed in the Mineral ResourceStabilization Fund (MRSF) whose assets, afteraccounting for drawdowns by the Government, areinterest-bearing securities and deposits with theBank of Papua New Guinea, the counterpart ofwhich consists mainly of foreign deposits andsecurities. The Copper Stabilization Fund in Chileis another example of a formal institutional struc-ture intended to save at least part of the gains real-ized by state-owned mining considered likely to betemporary, for use when copper prices drop.

The implementation of such schemes, however,is not without difficulties. One major problem is toset an appropriate benchmark price in order todetermine at what rate disbursements should bemade from the fund. For the MRSF in Papua NewGuinea, for example, mining revenue projectionswere made over a five-year planning horizon basedon price and output forecasts. In practice, year-to-year revisions in the rolling horizon revenue fore-casts led to considerable variations in the govern-ment drawdowns recommended by the MRSF eachyear.43

42In Kenya, for example, export taxes are relatively minorand booms and slumps in world coffee and tea prices arereflected to a large extent in domestic producer prices; the gov-ernment's budgetary position is largely affected by how privateagents react in terms of consumption and investment decisions.These indirect effects can, nevertheless, be substantial.

Agricultural Boards

The other principal type of institutional structurefor government intervention has been the state-owned agricultural board or fund. These are usu-ally charged with a domestic price stabilizationfunction, the transfer of explicit export taxes tocentral government, and a role in determining theextent to which operating surpluses are channeledinto a formal stabilization fund or are transferreddirectly to central government (or to other uses). InCote d'Ivoire, for example, the first objective wasachieved by declaring a fixed producer price innominal terms for coffee and cocoa. The board wasable to stabilize producer prices (and thus pro-ducers' income) until international prices begandropping in 1986, which exerted increasing pres-sure on the organization's financial position. In1989, the Government suspended the explicitexport taxes, but even so was obliged to cut pro-ducer prices in half in 1990 in an attempt to limit

43See Claessens and Coleman (1991), who found that thevariation in government drawdowns over the 1980-88 periodwas only modestly lower than that of government mining reve-nue accruals.

19

Temporary Price Shocks


the operational deficit of the parastatal. Duringmuch of the period prior to 1986, fluctuations ininternational prices showed up primarily in the rev-enues of general government via the operating sur-pluses of the board. At times, particularly duringthe extraordinary beverages boom of 1975-78 andthe smaller boom in the mid-1980s, these revenuesbecame very large and were principally used tofinance spending on urban infrastructure and otherdevelopment projects.

While similar schemes operated elsewhere inAfrica, some other countries operated agriculturalstabilization schemes quite differently. Papua NewGuinea, for example, had schemes for coffee,cocoa, copra, and palm oil that used a ten-yearmoving average of inflation-adjusted commodityprices as the basis for paying bounties and mount-ing levies on actual free-on-board prices. As inter-national prices declined in the mid- to late 1980s,however, many of these programs experienced dif-ficulties as drawdowns began to exhaust the funds.

The experience with stabilization schemes raisesa number of issues about their desirability. First,stabilization schemes that transfer most of theeffects of commodity price fluctuations to govern-ment may in fact be prone to exacerbating a coun-try's fiscal management problems.44 For somecountries the exercise of sufficient control overgovernment expenditure is likely to pose seriousdifficulties, and there is some evidence to suggestthat government responses to revenue gains andlosses from commodity price fluctuations are asym-metric.45 That is, it has proved much easier toadjust public expenditures upwards in boom peri-ods than to adjust them downwards in slump peri-ods. This reinforces the point made earlier that itmay well be advisable for governments to sharemore of the commodity price risk with the privatesector, particularly if reforms can also be under-taken to remove certain impediments (such asstrict capital controls) that hinder the private sec-tor's ability to handle fluctuations in commodityincomes.46

The situation may be further complicated by thefact that, to achieve a significant amount of incomesmoothing, it may be necessary for a fund to hold arelatively high level of foreign reserves or be in aposition to borrow not insubstantial funds frominternational capital markets for stabilization pur-poses during downturns in commodity prices. Theneed for a relatively high level of foreign reserves(on average) arises from the fact that temporary

44See Cuddington (1989).45See Deaton (1992).46This does not mean that the agricultural sector should not

be taxed; rather it means that the form of taxation needs to bereformulated.

commodity price shocks show considerable per-sistence and revert relatively slowly to trend. More-over, holding such levels of foreign reserves willhave a high opportunity cost as the rate of returnon investment in most capital-short developingcountries will be considered to exceed the rate ofinterest paid on foreign reserve holdings. This fac-tor and the political pressures that may arise makesuch a reserves management strategy difficult toimplement. In addition, during boom periods itmay prove more difficult to enforce discipline onfollowing appropriate project selection criteria,with the result that low-quality projects may beundertaken. Regarding international borrowing,many developing countries do not enjoy a creditrating that would enable them to obtain significantresources from private sources. Moreover, even forcountries with some degree of access to interna-tional financial markets, financing may be more dif-ficult to obtain, and more expensive, at times whencommodity export prices fall, as this would create aless favorable outlook for the balance of payments.

Second, even if improvements to the institutionalframework for stabilization schemes were to beimplemented, they are still vulnerable to the diffi-culty of ascertaining ex ante the appropriate pricelevel around which incomes should be stabilized.This requires an evaluation of the temporariness ofprice shocks and whether and how the underlyingtrend is changing. In particular, the difficulties thatmany stabilization schemes experienced from themid-1980s onward may be traced to a failure totake into account the downward trend in com-modity prices. Forecasting the future path of com-modity prices is a notoriously difficult exercise, butthe analysis presented earlier in this paper suggeststhat it might be possible to improve the working ofstabilization schemes by paying greater attention tohow particular commodity prices are evolving overtime. The analysis of the beverages price index, forexample, indicated that quite a large percentage ofthe shocks over recent years have been of a secular,persistent nature requiring adjustment in exportingcountries. Temporary shocks do occur in the bev-erages markets, and when they occur they appearas price spikes generally associated with eventssuch as the Brazilian frost in 1975 that led to atemporary boom in coffee prices. Thus, it makeslittle sense to put in place a domestic price stabili-zation scheme that does not take account of suchfactors and tries to stabilize prices around a con-stant mean in nominal or real terms. What mightmake more sense for designers of such schemes isto make estimates of the recent trend for the com-modity price in question (which should be updatedcontinuously), set up an intervention band basedon the evolving properties of the price series, and

IV POLICY ISSUES

20


intervene only when shocks push prices outside theband.

International Commodity Agreements

As volatility in international commodity pricesaffects the economies of all primary productexporters, there is an obvious basis for the idea ofcooperative commodity agreements of some kindwith the aim of attempting to achieve greater sta-bility in international commodity prices. There mayalso be similar interest in importing countries, butit is likely to be less acute as their import structuresgenerally tend to be much more diversified. Experi-ence with ICAs has not in general shown them tohave been successful.

Two aspects of ICAs should be differentiated.First, ICAs may attempt to increase the mean priceof a commodity by restricting its supply by somemethod. And second, ICAs may simply limit pricevariability, for example, by recourse to bufferstocks. The first aspect is not considered favorablyby importing countries and has been a source ofcontention, while the second one is perhaps lesscontroversial. However, it is hard to pin down howmuch price stability is desirable; presumably, somedegree of price flexibility must be allowed so thatprices can provide the necessary signaling functionfor producers and consumers alike. More impor-tantly, an ICA (or a producer-only scheme) runsthe risk of trying to stabilize prices around the"wrong" (non-market-clearing) price, which canquickly lead to financial nonviability, the sameproblem discussed in the case of agriculturalboards above, or to the collapse of output-sharingarrangements. On both political and economicgrounds the prospects for ICAs playing a role inaddressing the issues raised by recent internationalcommodity price behavior appear somewhat dim.

External Compensatory Finance

Another means of trying to smooth out theffects of negative, temporary commodity priceshocks for credit-rationed developing countries hasbeen external compensatory finance from the IMFunder the compensatory financing facility (CFF)and subsequently the compensatory and cotingency financing facility (CCFF) and from theEuropean Community under the STABEXscheme.47 The IMF's CCFF is designed to work

47It should be noted that such compensatory finance is a loanor a grant to the monetary authority or government of the coun-try concerned. It need not, therefore, necessarily find its way tothose who may be most directly affected by an export shortfall.

best when temporary price shocks hit export earn-ings over a short period (the so-called "shortfallyear"); then the destabilizing impact of the tempo-rary shock can be offset in part by drawing underthe facility, subject to early identification of thshortfall, access limits, and an appropriate policyresponse. In a similar vein, the contingency ele-ment of the CCFF is designed through a mix ofinancing and adjustment to help protect the programs of member countries being supported byIMF resources from being thrown off course byunanticipated shocks to key exogenous variables,such as commodity prices. Clearly, the design ofpolicy responses and the form of IMF assistance tomember countries must take into account thenature of the commodity price shock, namelywhether it can be expected to be a short-lived, tem-porary shock or one that is likely to be longerlasting.

Market-Related Instruments

In addition to the possibility of reacting ex postto commodity price changes, it is possible for coun-tries to trade away much of the commodity pricerisk, and the accompanying macroeconomic distur-bances, by using financial securities such as futuresand options. Producer countries can then limit theirexposure to unanticipated price changes whileinvestors in other countries assume it. In this way,risk is shifted internationally. But financial transac-tions of this sort at best insulate the economy fromthe problem of uncertainty, and at a cost. They donot avoid the effects of anticipated temporary orsecular declines in prices.

The rise in short-run volatility of commodityprices since the early 1970s has increased theattractiveness of hedging instruments. In additionto well-developed futures and options markets withshort-dated instruments for commodities thataccount for a significant part of total commoditytrade, financial innovation has led to developmentsin other instruments (commodity swaps, commodity options, commodity-linked bonds, for-wards) that have increased the scope for hedgingcommodity risk. Such financial instruments havethe advantage over other means of trying to handlerisk in that they transfer much of the risk to partiesoutside the country concerned. Nevertheless,establishing a coherent hedging strategy is not astraightforward proposition and the use bydeveloping countries of futures and derivativemarkets has only increased relatively slowly. Itwas for these reasons that both the World Bankand the United Nations Conference on Trade andDevelopment created technical assistance pro-grams focused on increasing awareness of external

Temporary Price Shocks

21


exposure in developing countries and providingassistance in risk management implementation.48

Here again it is necessary to take account of theinstitutional, marketing, and financial frameworkas it affects commodity price risk and who bears it.In countries where exporters face much of the com-modity price risk, exchange controls may well ruleout commodity hedging by the private sector andthus preclude optimal borrowing and lendingbehavior; changing this state of affairs may wellhelp in reducing uncertainty and smoothing exportearnings.49 Physical and price regulation of com-modity exports constitutes an additional barrier toprivate sector hedging. In other cases, marketinginstitutions, marketing boards, or private exportersmay act as simple intermediaries buying from smallproducers and arranging sales and shipmentsabroad. Some advance payments to producers maybe made, but final settlement may await the end ofthe crop year; the intermediaries may have littleincentive to hedge so that most of the commodityprice risk is transferred to the small producer whois least able to protect himself. By marketingreform and amending marketing regulations it maybe possible to reapportion the price risk andencourage those who are better placed to hedge todo so through the use of short-dated instruments.

Using exchange-traded futures contracts andoptions purchases can allow countries to offsetsome of the exposure to risk due to price volatilityat the higher frequencies. Maturities tend to be lim-ited to one to two years and market thinness canbecome a problem for the longer-dated maturities.There are also the costs of hedging in the form ofbrokerage fees and the need for agents to fulfillmargin requirements (and prospective margincalls) or pay options premia in order to gain accessto the market. Creditworthiness considerationsmay, however, make it considerably more difficultfor developing countries to access other financialmarkets, particularly for nonstandard contracts andlonger-dated instruments which would allow risk tobe hedged more fully over time. Forward, swap,and option sales contracts all require considerationof the counterparty's creditworthiness, and as thelength of a contract and the period of potentialprice movement extends out over time, the greaterthe risk that an intermediary in over-the-counter(OTC) markets assumes. In particular, sovereignrisk for contracts with private agents and govern-

48Among countries assisted by the World Bank have beenChile, Colombia, Costa Rica, Indonesia, Papua New Guinea,Poland, Tunisia, and Uganda.

49In Colombia, for example, use of market-based risk man-agement instruments by the private sector was banned until1992.

ments in developing countries constitutes a seriousobstacle to greater use of OTC markets. Overcom-ing high levels of sovereign risk by such means ascollateralization and guarantee mechanisms isbeing investigated by OTC intermediaries andinternational organizations.50

Government Policies for the LongerTerm

The previous sub-sections have dealt with pol-icies aimed at mitigating the impact of interna-tional commodity price variability. But commodity-exporting countries have also faced a downwardtrend in real prices. The earlier analysis suggeststhat it would be unwise to base policies on theassumption that the weakness in real commodityprices will be quickly overcome. A stronger eco-nomic recovery (and better growth performanceover the medium term) in the industrial countrieswould certainly help, as would continuing highrates of growth in the newly industrialized coun-tries. Some supply-side factors may ease over time(e.g., recent shocks to metals markets from the for-mer Soviet Union), but other factors (e.g., struc-tural change in exporting countries andproductivity gains due to technical progress) seemunlikely to be reversed. If such is the case, thosecountries most seriously affected by permanent orhighly persistent commodity price shocks must facethe necessity of adjusting to the shocks and orient-ing their macroeconomic and structural policiestoward this end.

Therefore, the secular decline in commodityprices poses the question of what—if any—are theappropriate structural policies for exporting coun-tries, in particular those concerning resource real-location away from primary products

Documents

The Behavior of Non-Oil Commodity Prices€¦ · Volume and Real Value of Non-Oil Commodity Exports 2 3. Export Volumes by Region and Commodity Grouping 2 II. 4. Real Non-Oil Commodity