International Reserves: Precautionary vs. Mercantilist Views, Theory and Evidence

Transcription

1 WP/5/198 International Reserves: Precautionary vs. Mercantilist Views, Theory and Evidence Joshua Aienman and Jaewoo Lee

2

3 25 International Monetary Fund WP/5/198 IMF Working Paper Research Department International Reserves: Precautionary vs. Mercantilist Views, Theory and Evidence Prepared by Joshua Aienman and Jaewoo Lee 1 Authoried or distribution by Gian Maria Milesi-Ferretti October 25 Abstract This Working Paper should not be reported as representing the views o the IMF. The views expressed in this Working Paper are those o the authors and do not necessarily represent those o the IMF or IMF policy. Working Papers describe research in progress by the authors and are published to elicit comments and to urther debate. This paper compares the importance o precautionary and mercantilist motives in the hoarding o international reserves by developing countries. Overall, empirical results support precautionary motives; in particular, a more liberal capital account regime increases international reserves. Theoretically, large precautionary demand or international reserves arises as a sel-insurance to avoid costly liquidation o long-term projects when the economy is susceptible to sudden stops. The welare gain rom the optimal management o international reserves is o a irst-order magnitude, reducing the welare cost o liquidity shocks rom a irst-order to a second-order magnitude. JEL Classiication Numbers: F15; F31;F43 Keywords: International Reserves, Precautionary Demand, mercantilist, inancial crises Authors Address: jaien@ucsc.edu; jlee3@im.org 1 Joshua Aienman, Proessor o Economics at the University o Caliornia, Santa Cru was a Visiting Scholar in the Research Department o the IMF when part o this paper was written. We thank Hali Edison or sharing the data and Aleksandra Markovic or research assistance in the earlier phase o the project. We thank Michael Dooley, Brian Pinto, Ramkishen Rajan, Partha Sen, and the SERC conerence participants, Singapore 25, or their useul comments.

4 - 2 - Contents Page I. Introduction and Summary...3 II. International Reserves: Evidence...6 III. The Model...9 IV. Concluding Remarks...14 Reerences...16 Data Appendix: Deinitions o the Regression Variables...18 Tables 1. Reserves to Broad Money A. Auxiliary Regressions Reserves to GDP Reserves to Broad Money with Capital Account Liberaliations Reserves to GDP with Capital Account Liberaliations Volatility, Reserves, and Expected Surplus...24 Figures 1. Reserves to Broad Money Reserves to GDP Eects O Selected Variables on the Reserves/GDP Ratio Country-Speciic Eects The Time Line Liquidity Shocks, Reserves Deposit Ratio, and Output...3

5 - 3 - I. INTRODUCTION AND SUMMARY This paper has two goals: i quantiying the relative importance o alternative views explaining international reserves accumulation, and ii modeling precautionary demand or international reserves, viewing it as sel-insurance against costly output contractions induced by sudden stops and capital light. This model is used to provide a welare evaluation o the costs and beneits o hoarding reserves and the optimal sie o precautionary demand. The crisis in East Asia led to proound changes in the demand or international reserves, increasing over time the hoarding. Several salient eatures o the crisis may provide clues to the changing attitude towards international reserves. First, the magnitude and speed o the reversal o capital lows throughout the crisis surprised most observers. While the 1994 Tequila crisis induced the market to expect similar crises in Latin America, most viewed East Asian countries as being less vulnerable to the perils associated with hot money. 2 This presumption ollowed rom the prevalent pre-1997 view that East Asian countries were more open to international trade, had sounder overall iscal policies, and had stronger growth perormance than Latin American countries. In retrospect, the crisis exposed hidden vulnerabilities o East Asian countries, orcing the market to update the probability o sudden stops aecting all countries. The crisis also led to sharp output and investment contractions, credit crunches, and in several countries to ull-blown banking crises. 3 Finally, most aected countries went through tough adjustments, reversing the output contraction and resuming growth within several years. While a ew countries lirted with capital controls, within two to three years most countries retained or increased their inancial integration. The above observations suggest that hoarding international reserves can be viewed as a precautionary adjustment, relecting the desire or sel-insurance against exposure to uture sudden stops. This view, however, aces a well-known contender in a modern incarnation o mercantilism: international reserve accumulations triggered by concerns about export competitiveness. This explanation has been advanced by Dooley, Folkerts-Landau, and Garber 23, especially in the context o China. They interpret reserve accumulation as a by-product o promoting exports, which is needed to create better jobs, thereby absorbing abundant labor in traditional sectors, mostly in agriculture. Under this strategy, reserves accumulation may acilitate export growth by preventing or slowing appreciation. Some view the modern mercantilist approach as a valid interpretation or most East Asian countries, arguing that they ollow similar development strategies. This interpretation is intellectually intriguing, especially in the broader context o the Revived Bretton Woods system, yet it remains debatable. Some have pointed out that high export growth is not the new kid on the block it is the story o East Asia during the last ity years. Yet, the large increase in 2 See Calvo 1998, Calvo and Mendoa 2, and Edwards 24 or urther discussion on sudden stops o short-term capital lows. 3 See Kaminsky and Reinhart 1999, and Hutchison and Noy 22 or urther discussion on the output costs associated with sudden stops.

6 - 4 - hoarding reserves has happened mostly ater This issue is o more than academic importance: the precautionary approach links reserves accumulation directly to exposure to sudden stops, capital light, and volatility, whereas the mercantilist approach views reserves accumulation as a residual o an industrial policy, a policy that may impose negative externalities on other trading partners. Our empirical test augments previous econometric speciications o international reserves by adding two sets o variables. The irst set deals with actors associated with mercantilist motives: lagged export growth and deviations rom predicted purchasing power parity PPP. The second set o variables attempts to capture precautionary adjustment in the atermath o unanticipated sudden-stop crises, using dummy variables. Speciically, two crucial events were the 1994 Mexican crisis and the 1997 East Asian crisis. Both happened at times o greater inancial integration, promoted by relaxing capital controls. Our results provide only a limited support or the mercantilist approach. While the variables associated with the mercantilist motive are statistically signiicant, their economic importance in accounting or reserves hoarding is close to ero and is dwared by other variables. Speciically, trade openness, measured by the GDP share o imports, and crises variables are playing a much more important role in accounting or reserves accumulation than lagged export growth and PPP deviations. This result applies to all countries, including China. Indeed, inspecting the magnitude o country-speciic dummies reveals that China is not an outlier in the level o reserves. We also ind strong localied eects o crises: while the 1994 Mexican crisis increased reserves in Mexico, it did not aect reserves in East Asia. Similarly, the 1997 crisis strongly increased the hoarding o reserves in East Asia, but not in Latin America. Across all speciications, a more liberal capital account regime is ound to increase the amount o international reserves. This by itsel constitutes evidence in avor o the precautionary view, or capital account liberaliation will boost the precautionary motive more than the mercantilist motive. Moreover, the inclusion o capital control variables weaken the statistical signiicance o deviations rom PPP, one o the two mercantilist variables, while having little eect on the statistical signiicance o crisis variables. Overall, the empirical results o Section II are in line with the precautionary demand. Yet, the precautionary demand approach has not been endorsed uniormly. Skeptical views point out that the sheer magnitude o reserves accumulated by East Asian countries seems excessive once attention is paid to the opportunity costs o reserves. In order to deal with these concerns, we provide in Section III a simple model characteriing and quantiying the welare gains attributed to hoarding reserves in the presence o exposure to external liquidity shocks. The model extends the literature dealing with the demand or bank reserves in the closed economy to the important, yet less studied open-economy context. 4 Speciically, we consider a country exposed to international liquidity shocks, which in turn can cause liquidation and consolidation o investment. A key postulate o the analysis is that, short o 4 See Bryant 198; Diamond and Dybvig 1983, and Prisman, Solvin, and Sushka 1986 or earlier literature dealing with optimal reserves liquidity policy in a closed economy.

7 - 5 - having a credible international lender o last resort, hoarding international reserves is among the ew options allowing developing countries to reduce the output costs o sudden stops. While hoarding international reserves has its opportunity costs, we identiy circumstances where the welare gain rom hoarding reserves is o a irst-order magnitude, leading to potentially large precautionary demand or reserves. The earlier literature ocused on using international reserves as part o the management o an adjustable-peg or managed-loating exchange rate regime Frenkel, 1983; Edwards, 1983; and see Flood and Marion, 21, or a literature review. To our knowledge, our paper is the irst econometric attempt to evaluate the relevance o the mercantilist approach in the atermath o the 1997 crisis see Aienman and Marion, 23; Edison, 23; and Aienman, Lee, and Rhee, 24, or earlier empirical analysis o related issues. The model advanced in Section III contributes to the growing literature linking international reserves with sovereign risk and limited access to the global capital market. Past literature has considered precautionary motives or hoarding international reserves needed to stabilie iscal expenditure in countries with limited taxing capacity and sovereign risk see Aienman and Marion, Insurance perspectives o international reserves applying the option pricing theory are provided in Lee 24. The model in this paper is more closely related to the literature viewing international reserves as output stabiliers see Ben-Bassat and Gottlieb, 1992; Aienman, Lee, and Rhee, 24, and García and Soto, 24. Our paper adds to this literature by providing an explicit model o inancial intermediation and adjustment subject to liquidity shocks, where hoarding international reserves emerges as part o the optimal inancial intermediation. As our ocus is on developing countries, we assume that all inancial intermediation is done by banks, relying on debt contracts. Speciically, we consider the case where investment in a long-term project should be undertaken prior to the realiation o liquidity shocks. Hence, shocks may orce costly liquidation o earlier investments, thereby reducing output. We solve the optimal demand or deposits and international reserves by a bank that inances investment in long-term projects. The bank s inancing is done using callable oreign deposits, which exposes the bank to liquidity risk. Macro liquidity shocks stemming rom 5 The precautionary demand modeled in this paper supplements the precautionary demand stemming rom iscal considerations. For example, one may argue that the prospect o uniication o two Koreas the Republic o Korea and the Democratic Peoples Republic o Korea may explain part o the hoarding o international reserves by the Republic o Korea. Yet, we may qualiy this argument by noting that one expects the United States and other avanced economies to provide the credit needed to inance the uniication or the conlict. This argument, however, does not extend to the case o a sudden stop and capital light. As the 1997 crisis illustrated, external inance at times o sudden stops is not orthcoming without stringent conditions and is requently limited due to moral haard considerations.

8 - 6 - sudden stops and capital lights cannot be diversiied away. 6 In these circumstances, hoarding reserves saves liquidation costs, potentially leading to large welare gains, and these gains hold even i all agents are risk neutral. In this ramework, deposits and reserves are complements higher volatility o liquidity shocks will increase both the demand or reserves and deposits. The optimal hoarding o reserves to accommodate more volatile liquidity shocks reduces the output cost o these shocks rom irst-order to second-order magnitude. II. INTERNATIONAL RESERVES: EVIDENCE Our empirical analysis adds several new controls to past regressions. The mercantilist view ocuses on hoarding international reserves in order to prevent or mitigate appreciation, with the ultimate goal o increasing export growth. Hence, we expect that hoarding o reserves provoked by mercantilist concerns should be associated with higher export growth rates, and with a deprecated real exchange rate relative to the undamental PPP real exchange rate. In order to control or export growth, we constructed a three-year moving average o the growth rate o real exports denoted MVGX, lagged two years in the regression. 7 Our undamental PPP real exchange rate is deined as the itted value rom the regression o national price levels on the per worker income relative to the United States or nearly 15 countries, motivated by the classic Penn eect see the regression reported in Table 1A. 8 The deviations rom the undamental PPP value, denoted by PLDE, are measured by the residuals o this regression, and are ound to bring about an appreciation in the nominal eective exchange rates in the subsequent year or our sample countries lower panel o Table 1A. I a country whose price level is higher than the level implied by its relative income tends to accumulate international reserves in an eort to slow the pace o appreciation in its currency, the coeicient on PLDE will be positive in the regression o international reserves on usual determinants including PLDE. The second set o controls attempts to capture the eects o two important crises: the 1994 Mexican and the East-Asian crises. This is done by applying a dummy variable to each crisis CRMEXEM: 1 since 1995, beore; CRASIAEM: 1 since 1998, beore. In one o the regressions we apply continental dummies or each crisis see data appendix or deinitions. In addition, we control or log o population LPOP; log o percent import share LIMY; exchange rate volatility VOL_XC; and log o per capita income LYPC in one set o regressions. Various permutations o these regressions are summaried in Tables 1 6 The recent history o Argentina provides a vivid illustration o the limited ability to diversiy away liquidity shocks. In the mid-199s Argentina negotiated contingent commercial credit lines in an attempt to provide external insurance against liquidity shocks. These lines, however, dried up as Argentina approached the crisis. 7 We used lags to deal with possible endogeneity issues. 8 See Kravis 1984 or a classic reerence on PPP, and Samuelson 1994 or the apt expression Penn eect.

9 - 7 - and 2, covering Figures 1 and 2 summarie the contribution o the various variables in regression III to the dependent variable in the 199s, or six countries Argentina, Brail, Chile, China, Korea, and Mexico. The dependent variable in Table 1 is the reserves/broad money ratio. Higher lagged export growth and national price level above the undamental level predicted by relative GDP per capita regression are associated with higher reserves/broad money, and this eect is statistically signiicant. Similarly, the Mexican and the East Asian crises increased the demand or reserves, and this eect is statistically signiicant. Figure 1 allows one to inspect the economic signiicance o each variable in accounting or the observed reserves ratios or six countries. The solid line denotes the dependent variable, the ratio o reserves to broad money. All other lines, which denote the contribution o each variable to the reserves/broad money ratio, are calculated by multiplying each variable and the associated coeicient rom regression. The variables presented separately in the igure are the import share LIMY in the table, post-crisis eect the combined eect o CRMEXEM and CRASIAEM, export growth MVGX, and the relative PPP PLDE. All other variables the population, exchange rate volatility, per capita income, and constant term are combined into one series others, because their eects show little variation over time. Figure 1 indicates a similar pattern or all the countries: trade openness is requently the most important consideration. The variables associated with mercantilist concerns are practically lat, and their economic signiicance in accounting or the observed hoarding o international reserves is close to ero. The crisis variables play an important role in all the six countries, including China. The regional crisis dummy variables used in regression IV reveal an intriguing pattern the Mexican crisis is associated with higher demand or reserves in Latin America, but not in Asia. Similarly, the 1997 East Asian crisis is associated with higher hoarding o reserves in Asia, but lower reserves in Latin America a drop o 5 percentage points in the atermath o the 1997 crisis. Regression V reveals that the sie o the variables associated with mercantile concerns is not impacted by crises, hence it rejects the possibility that crises magniied mercantilist concerns. The dependent variable in Table 2 is the reserves/gdp ratio, and per capita income is excluded rom the regression. Overall, the results are very similar to those associated with reserves/broad money. The main changes are that the impact o the crises is sharper on reserves/gdp than on reserves/broad money. Figure 2 summaries the economic signiicance o each variable in accounting or the observed reserves ratios or six countries. It reveals similar patterns to Figure 1. Note that in the case o China, the reserves/gdp ratio increased mostly ater 1994, rom.1 in 1994 to about.16 in The most important variable explaining this increase in the reserves/gdp ratio is the crises dummies about.5 out o the increase o.6. All the other variables, including the two mercantilist variables, provide practically ero explanation to reserves/gdp ratio, in terms o the level or

10 - 8 - the change. 9 Openness, measured by the import share, did not play a prominent role in this increase o the reserves/gdp ratio, but is an important explanator o the level o reserves. The import share accounts or.11 out o.16, the average o the reserves/gdp ratio. The sie o population also makes a very large contribution to the level o reserves, but varies little over time and thus is combined with other variables and the constant term in the igure. Nor is the mercantilist eect an important actor in accounting or dierences in the level o reserves across dierent countries. Figure 3 compares the relative importance o several regressors by plotting the eect o an increase in the value o each variable by one standard deviation. In this igure, the standard deviation o each variable is calculated across countries using the data in 2, but similar results arise when the standard deviations are calculated or the pooled data over the whole sample period. Among the two mercantilist variables, the deviation o the PPP exchange rate plays a more important role in explaining the reserves/gdp ratio, but its eect pales by the eects o crisis or openness. Population plays quantitatively the most important role in explaining cross-country dierences in the level o reserves, but is not presented in Figure 3. Population moves very little over time unlike other economic variables, making it conceptually more comparable to country-speciic eects rather than the eects o other economic variables. Figure 4 plots the distribution o the country-speciic eects, identiying the names o the six countries evaluated in Figures 1 2 and several others with country-speciic eects that dier rom the average o all country-speciic eects by nearly or more than two standard deviations. Note that China s country-speciic eect is negative and is inconsistent with the notion that China s large reserves make it an outlier in the context o the cross-ountry panel comparison, For both China and India, the clear negative values o countryspeciic eects relect the large sies o their population. In regressions that excluded the population variable rom the regressors, the country-speciic eects on China and India were closer to ero than in the regressions with population. With or without considering the eect o population, China is not an outlier with a large positive country-speciic eect. One such country is Singapore, a country well known or its traditionally very high level o international reserves which oten exceeded 8 percent o its GDP during the sample period, and its country-speciic eect is close to three standard deviations. Two countries with smaller but still large country-speciic eect about two standard deviations away rom the average are Cyprus and Hong Kong SAR, in the latter o which the currency board system necessitates a high level o reserves. In terms o the horse race between the mercantilist and precautionary views o international reserves, our results suggest that the precautionary motive played a more visible role in the accumulation o reserves than the mercantilist motive. At minimum, we could identiy the likely eect o precautionary motive more easily and strongly than the likely eect o the mercantilist motive. 9 See Prasad and Wei 25 or recent skeptical perspectives about the mercantilist interpretation o Chinese reserves accumulation.

11 - 9 - This summary interpretation remains intact when we control or changes in capital account regimes. Tables 3 and 4 repeat the regression o Tables 1 and 2, respectively, but include the variable that captures the degree o capital account liberaliation K liberaliation. This variable, constructed by Edwards 25, measures the degree o capital account liberaliation in iner grids than most existing measures. Across all speciications, a more liberal capital account regime is ound to increase the amount o international reserves. This by itsel constitutes evidence in avor o the precautionary view, or capital account liberaliation will boost the precautionary motive more than the mercantilist motive. Moreover, the inclusion o capital control variables weakens the statistical signiicance o PLDE, one o the two mercantilist variables, while having little eect on the statistical signiicance o crisis variables. III. THE MODEL We construct a minimal model to explain the sel-insurance oered by international reserves in mitigating the output eects o liquidity shocks. The structure o the model is akin to Diamond and Dybvig 1983 investment in a long-term project should be undertaken prior to the realiation o liquidity shocks. Hence, the liquidity shock may orce costly liquidation o the earlier investment, reducing second period output. As our ocus is on developing countries, we assume that all inancial intermediation is done by banks, relying on a debt contract. We simpliy urther by assuming that there is no separation between the bank and the entrepreneur the entrepreneur is the bank owner, using it to inance the investment. The time line is summaried in Figure 5. At the beginning o period 1, risk-neutral agents deposit D in banks, which in turn use D to inance long-term investment, K 1, and hoarding reserves, R. A liquidity shock, with the aggregate value o Z or the borrowing economy, materialies at the end o period 1, ater the commitment o capital. A liquidity shock exceeding reserves induces a premature liquidation o Z - R. Output increases with the capital invested at the beginning o period one, K 1, and declines with liquidation at a rate that depends on the adjustment cost, θ. Assuming a Cobb-Douglas production unction, the second period output is Y { Z, } 2 [ 1 R = K 1 + θ MAX ] ; where θ < 1, and < 1. 1 Recalling that K1 = D R, the net capital ater liquidation is: K 2 D R 1 + θ Z R = D Z θ Z R = D R i i Z > R Z R 2 It is convenient to normalie the liquidity shock by the level o deposits, denoting the normalied shock by : Z = D ; < τ 1, and density. 3

12 - 1 - Depositors are entitled to a real return o D r on the loan that remains deposited or the duration o investment. 1 Assuming agents subjective discount rate is ρ, competitive intermediation implies that: ρ ρ τ τ = + + = D D r d r d Net reserves held until period 2 are assumed to yield a return o r. We denote the marginal liquidity shock associated with liquidation by D R /, =. The expected second period surplus i.e., net income ater paying depositors is: [ ]. 1 1 ] [ 1 ] [ d D d Z R r d R Z Z D d R D E = Π τ τ ρ θ 5 It is the sum o the expected output, plus the income associated with reserves net o liquidation, minus the repayment to depositors who get a return o ρ on the net deposit position, Z D. Applying 3 and the deinition o the, we re-write the expected surplus as: [ ] ] [ = Π d d r D d d D E τ τ ρ θ 5 The FOC determining the optimal demand or international reserves is:. 1 ] [ d r d d D = τ θ θ 6 1 The possibility that the outcome o investment is not large enough to meet the promised rate o return is discussed later. To preview, this possibility does not aect the main conclusion o our analysis, because o the assumption o risk neutrality.

13 This condition is equivalent to: [ MP Z R] [ MPK 1 + r ] Pr[ Z < R] = θ E K >, 7 1 where MPK is the marginal productivity o capital, and Pr[ Z < R] is the probability that the 1 liquidity shock is below the level o reserves. The expected opportunity cost o holding reserves is equalied to the expected precautionary beneit o holding reserves. Figure 6 plots the inal output the solid line as a unction o liquidity shock,, drawn or a given initial investment and reserves hoarding. For liquidity shocks below, output is lat, independent o the realied liquidity shock. A liquidity shock above requires costly downward adjustment o capital, thereby reducing inal output. A marginal increase o the initial reserves position will shit the output line in two dierent directions. First, hoarding extra dollar reserves reduces the initial capital by one dollar, reducing output or liquidity shocks below, shiting the output line downward or < the downward shit equals MP K. Extra dollar reserves implies, however, lower deadweight loss associated with 1 liquidation, thereby shiting the output line to the right or >. The decrease in output associated with extra dollar reserves is depicted in Figure 6 by the shaded area below the old production curve, or <. Similarly, the increase in output associated with the extra dollar reserves correspond to the shaded area to the right o the old production curve, or >. The expected net gain in production rom holding reserves corresponds to the dierence between the two shaded areas, properly weighted by, as well as the expected gross income attributed to extra dollar reserves. Optimal reserves, which satisy equation 7, maximie the overall expected gain. The irst order condition characteriing optimal deposit can be rewritten as: = D 1 1 {1 + r 1 d + d ρ τ 1 θ[ ] τ 1 1 d} 1 [1 + θ ] d 8 We irst consider the case with small shocks to gain the basic insight or the welare gains associated with reserves. In the absence o uncertainty, the optimal level o deposits D, and the resultant surplus Π are: 1/1 D = 1 ρ ; + 1 Π = 1 + ρ D. 8 Suppose that the liquidity shocks are either ero or, with probability hal each, and I reserves are set to ero, and deposits at D, the expected surplus is: ρ = r.

14 [ D ] 1 + ρ D [ D θ ] [ 1 + ρ D 1 E Π ] R = = Applying 8 to 9, the irst order approximation o the expected surplus can be reduced to: 1 + ρ D E[ Π ] R = Π θ. 9 2 Liquidity shocks have a irst-order adverse eect on expected surplus. In the absence o the insurance provided by reserves, liquidation induces a deadweight loss equal to the adjustment cost, θ, times the expected liquidation. This result is not aected i we allow the optimal adjustment o deposits: the envelope theorem implies that such an adjustment would have only second-order eects. 11 In a two-states-o-nature case, perect stabiliation can be achieved by hoarding reserves equal to the liquidity shock: R = D ; adjusting deposits to D = 1+ D, thereby setting the stock o capital at K 1 = D. I the liquidity shock materialies, R would provide the needed liquidity, preventing costly output adjust. I the shock is nil, there would no need to use R. The assumption that ρ = r implies that the cost o this insurance is ero. Consequently: 12 E [ = Π 9 Π ] R= D This simple example suggests that liquidity shocks have irst-order welare eects in the absence o reserves, and that hoarding reserves can reduce the cost o liquidity shocks rom irst-to second-order magnitude. We conirm this conjecture by a detailed simulation o the case where liquidity shocks ollow a uniorm distribution, = 1/ λ; λ = τ < 1. Figure 7 plots the association between volatility and the reserves/deposit ratio or the case where the level o deposit is kept at the level o equation 8. The reserves ratio increases with the 11 This ollows rom the observation that: d E[ ] R= E[ Π] R= d D E[ Π] R= + d D d E[ Π] R= deposits is =. D Π E[ Π] R= = recall that the FOC determining 12 With more than two states o nature, R would be preset at the ex ante eicient level, providing ull insurance or liquidity shocks below and partial insurance above. While there is no way to insure complete stabiliation, one expects large welare gain rom setting R at the ex ante eicient level relative to the case o R =.

15 volatility. Allowing or the optimal adjustment o D according to equation 8, it ollows that dd >. The increase in D is needed to mitigate the costly drop in output induced by dr R = reserves accumulation and is needed to keep the planned capital at the optimal level. 13 Table 5 traces the impact o higher volatility or the case where both reserves and deposits are adjusting optimally, contrasting it to the case where reserves are set to ero the last two columns. Speciically, the irst our columns report the optimal reserves/deposit ratio, deposits, reserves and expected surplus as a unction o volatility, assuming that R and D are adjusted optimally. The last two columns report D and expected surplus or case where R is ero, and only D is adjusted optimally. Discussion: In the absence o reserves, the volatility has irst-order eects on output: increasing volatility rom ero to.6 reduces expected surplus by about 15 percent. Hoarding the optimal level o reserves reduces the cost o volatility into a second order magnitude, about 3 percent. Hence, optimal reserves have a irst-order welare eect, increasing the expected surplus by about 12 percent relative to the case o ero reserves. Accomplishing this gain requires relatively large reserves, about hal o the deposit level or the case where λ =. 6. The eect o volatility with optimal reserves hoarding is to increase both deposits and reserves, while keeping the level o planned capital K 1 almost constant. Our discussion assumed so ar that the limited liability constraint does not bind: that is: D 1 θ[ ] > D1 + ρ1 or all. 1 Indeed, it can be veriied that the limited liability constraint is not binding in the simulation reported in Table 5. We now show that our main results are not dependent on these parametric assumptions. The limited liability constraint would bind i D 1 θ[ ] < D1 + ρ1 in some states o nature, which may hold or large enough volatility and adjustment costs. We denote the contractual interest rate on deposits in the presence o binding liability constraint by ρ d, and by ~ the threshold liquidity shock associated with ero surplus: Recalling 2, higher R reduces the stock o capital in states o nature where Z < R by R, but increases the stock o capital in states o nature where Z > R by θ R θ Note that or =, output is ero, and the bank would deault. Hence, a suicient 1+ θ 1+ θ condition or the limited liability constraint to bind is < λ. Equation 11 implies, 1+ θ 1+ θ however, that ~ < λ, and the limited liability constraint may bind even i > λ. 1+ θ

16 ~ [ ~ D 1 θ ] = D1 + ρ 1 ~ d. 11 For liquidity shocks above this threshold, we assume that depositors are paid a raction φ o the output, φ Note that binding limited liability constraint implies that depositors are exposed to the downside risk associated with large liquidity shock. Hence, depositors would demand a high enough deposit interest rate ρd to compensate or the exposure. For risk-neutral depositors, the equilibrium interest rate is determined by the ollowing breakeven condition: τ % 1 + ρ D 1 d = 1 + ρ D 1 d+ φ D1 θ[ ] d d τ % 12 where the threshold ~ is determined by 11. Consequently, the expected surplus is: % % E [ Π ] = D 1 d + 1 θ[ ] d 1 + ρd D1 d + τ 1 φ D1 θ[ ] d + D1 + r d = % τ D 1 d+ 1 θ[ ] d + τ D 1 + r d 1 + ρ 1 d. 13 Note that 13 is identical to the expected surplus in the base case o the previous section, 5. With risk-neutral agents, binding limited liability constraint changes the deposit interest rate, without changing the entrepreneurs expected surplus and investment patterns. 16 IV. CONCLUDING REMARKS Our study has outlined a procedure that helps to identiy the contributions o precautionary and mercantilist motives to the hoarding o international reserves. Applying it to 198 2, 15 The conventional closed-economy assumption is φ = 1. The case where φ < 1can capture the presence o repatriation risk, where the banks pays oreign creditors only a raction φ o output or > ~, or the eiciency loss associated with debt restructuring. 16 This result holds because we assumed the absence o enorcement and monitoring costs, and that all agents are risk-neutral.

17 we ound that variables associated with trade openness and exposure to inancial crises are both statistically and economically important in explaining reserves. In contrast, variables associated with mercantilist concerns are statistically signiicant, but economically insigniicant in accounting or the patterns o hoarding reserves. These results hold or most countries, including China. We provided a model that shows that precautionary demand is consistent with high levels o reserves. We close the paper with qualiying remarks. As is the case with all empirical studies, more accurate and updated data may modiy the results. Our empirical study does not imply that the hoarding o reserves by countries is optimal or eicient. Making inerences regarding eiciency would require having a detailed model and much more inormation, including an assessment o the probability and output costs o sudden stop and the opportunity costs o reserves. Our study reveals, however, that existing patterns o growing trade openness and greater exposure to inancial shocks by emerging markets go a long way towards accounting or the observed hoarding o international reserves.

18 REFERENCES Aienman, J., and N. P. Marion, 23, The High Demand or International Reserves in the Far East: What's Going On? Journal o the Japanese and International Economies, Vol. 17, No. 3, pp , 24, International Reserves Holdings with Sovereign Risk and Costly Tax Collection, Economic Journal, Vol. 114 July, pp Aienman, J., Y. Lee, and Y. Rhee, 24, International Reserves Management and Capital Mobility in a Volatile World: Policy Considerations and a Case Study o Korea, NBER Working Paper No.1534, Cambridge, Massachusetts: National Bureau o Economic Research. Ben-Bassat A., and D. Gottlieb, 1992, Optimal International Reserves and Sovereign Risk, Journal o International Economics, Vol. 33, pp Bryant, R., 198, A Model o Reserves, Bank Runs, and Deposit Insurance, Journal o Banking and Finance, Vol. 4, pp Calvo, G., 1998, Capital Flows and Capital-Market Crises: The Simple Economics o Sudden Stops, Journal o Applied Economics November., and E. Mendoa, 2, Contagion, Globaliation, and the Volatility o Capital Flows, Capital Flows and the Emerging Economies, ed. by Sebastian Edwards, Chicago: University Chicago Press. Diamond, D., and P. Dybvig, 1983, Bank Runs, Liquidity and Deposit Insurance, Journal o Political Economy, Vol. 91, pp Dooley, M., D., Folkerts-Landau and P. Garber, 23, An Essay on the Revived Bretton Woods System, NBER Working Paper No. 9971, Cambridge, Massachusetts: National Bureau o Economic Research. Edison, H., 23, Are Foreign Exchange Reserves in Asia Too High? in World Economic Outlook, September Washington: International Monetary Fund. Edwards, S., The Demand or International Reserves and Exchange Rate Adjustments: The Case o LDCs, , Economica, Vol. 5, pp , 24, Thirty Years o Current Account Imbalances, Current Account Reversals, and Sudden Stops, Sta Papers, International Monetary Fund, Vol. 51, Special Issue.

19 - 17 -, 25, Capital Controls, Sudden Stops, and Current Account Reversals, NBER Working Paper No. 1117, Cambridge, Massachusetts: National Bureau o Economic Research. Flood, R., and P. N. Marion, 21, Holding International Reserves in An Era o High Capital Mobility, Brookings Trade Forum 21, ed. by Collins S. M. and Rodrik D., Washington: Brookings Institution Press. Frenkel, Jacob, 1983, International Liquidity and Monetary Control, in International Money and Credit: The Policy Roles, ed. by George M. von Furstenberg, Washington; International Monetary Fund, pp García P. S., and C. G. Soto, 24, Large Hoarding o International Reserves: Are They Worth It?, manuscript, Santiago: Central Bank o Chile. Hutchison, M., and I. Noy, 22. How Bad Are Twins? Output Costs o Currency and Banking Crises, orthcoming, Journal o Money, Credit and Banking. Kaminsky, G., and C. Reinhart, 1999, The Twin Crises. The Causes o Banking and Balance-o-Payments Problems, American Economic Review, Vol. 89 June, pp Kravis, I.B., 1984, Comparative Studies o National Incomes and Prices, Journal o Economic Literature, Vol. 22 March, pp Lee, J., 24, Insurance Value o International Reserves, IMF Working Paper 4/175, Washington: International Monetary Fund. Prasad, E., and S-J Wei, 25, The Chinese Approach to Capital Inlows: Patterns and Possible Explanations, IMF Working Paper 5/79 Washington: International Monetary Fund. Prisman, E., M. Solvin, and M. Sushka, 1986, A General Model o The Banking Firm Under Conditions o Monopoly, Uncertainty and Recourse, Journal o Monetary Economics, Vol. 17 2: pp Samuelson, P.A., Facets o Balassa-Samuelson Thirty Years Later, Review o International Economics, Vol. 2, No. 3, pp

20 APPENDIX I Data Appendix: Deinitions o the Regression Variables Reserves: international reserves holdings minus gold, measured in U.S. dollars. R to M: ratio o reserves to the dollar value o broad money. R to Y: ratio o reserves to the dollar value o nominal GDP. LPOP: log o population LYPC: log o per-capita income LIMY: log o percent import share MVGX: three-year moving average o the growth rate o real exports log change, lagged two years in the regression. VOL_XC: exchange rate volatility, calculated rom the monthly exchange rate against the U.S. dollar. PLDE: the residuals rom the regression o national price levels measured in U.S. dollars on the per-worker income relative to the U.S. Table 1A Time dummies or each year were used to control or time-speciic common actors including the unit o denomination. CRMEXEM: dummy variable or the period ater the Mexico crisis, applied to developing and emerging market countries. CRASIAEM: dummy variable or the period ater the Asian crisis, applied to developing and emerging market countries. CRMEXEMLA: dummy variable CRMEXEM, applied only to Latin America CRMEXEMAS: dummy variable CRMEXEM, applied only to Asia CRASIAEMLA: dummy variable CRASIAEM, applied only to Latin America CRASIAEMAS: dummy variable CRASIAEM, applied only to Asia Regressions o Table 1 and Table 2 all include country-speciic constant terms. The sample comprises 53 countries that include advanced and emerging-market economies as well as several major developing economies. They are Australia, Austria, Belgium, Canada, Cyprus, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Japan, Luxembourg, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switerland, United Kingdom, United States, Hong Kong SAR, Israel, Korea, Singapore, Taiwan Province o China, Argentina, Brail, Chile, Colombia, Cech Republic, Hungary, Indonesia, Malaysia, Mexico, Pakistan, Peru, Philippines, Poland, Russia, South Arica, Thailand, Turkey, Veneuela, Algeria, China, Croatia, Egypt, India, and Morocco. Owing to data availability, Greece is excluded rom the regressions or Table 1, and Luxembourg, Egypt, and Taiwan Province o China are excluded rom the regressions that include price level data.

21 Table 1. Reserves to Broad Money I II III IV V LPOP LYPC LIMY VOL_XC MVGX PLDE CRMEXEM CRASIAEM CRMEXEMAS CRMEXEMLA.65.2 CRASIAEMAS CRASIAEMLA MVGXCRASIAEMAS PLDECRASIAEMAS R squared Cross-section Statistically signiicant at 1 percent, 5 percent, and 1 percent. All regressions included country ixed eects.

22 - 2 - Table 1A. Auxiliary Regressions Dependent Variable: National Price Level Variable Coeicient Constant Relative GDP per worker R-squared.439 Sample period Cross-section observations 149 Time dummies were included. Statiscally signiicant at 1 percent Dependent Variable: Log Change in the Nominal Eective Exchange Rate Variable Coeicient PLDE PLDE R-squared Sample period Cross-section observations 5 Country ixed eects were included. PLDE reers to the residuals rom the price-level regression

23 Table 2. Ratio o Reserves to GDP I II III IV V LPOP LIMY VOL_XC MVGX PLDE CRMEXEM CRASIAEM CRMEXEMAS CRMEXEMLA CRASIAEMAS CRASIAEMLA MVGXCRASIAEMAS PLDECRASIAEMAS R squared Cross-section Statistically signiicant at 1 percent, 5 percent, and 1 percent. All regressions included country ixed eects.

24 Table 3. Ratio o Reserves to Broad Money with Capital Account Liberaliations I II III IV V LPOP LYPC LIMY VOL_XC MVGX PLDE K liberaliation CRMEXEM CRASIAEM CRMEXEMAS CRMEXEMLA CRASIAEMAS.67.2 CRASIAEMLA MVGXCRASIAEMAS PLDECRASIAEMAS R squared Cross-section Statistically signiicant at 1 percent, 5 percent, and 1 percent. All regressions included country ixed eects.

25 Table 4. Ratio o Reserves to GDP with Capital Account Liberaliations I II III IV V LPOP LIMY VOL_XC MVGX PLDE K liberaliation CRMEXEM CRASIAEM CRMEXEMAS.17.8 CRMEXEMLA CRASIAEMAS.67.1 CRASIAEMLA MVGXCRASIAEMAS PLDECRASIAEMAS R squared Cross-section Statistically signiicant at 1 percent, 5 percent, and 1 percent. All regressions included country ixed eects.

26 Table 5. Volatility, Reserves, and Expected Surplus λ = R/D D R E[Π] E [ Π] R = D R = The simulation values are.33; θ =.5; ρ =.2; r =. 2. =

27 Figure 1. Ratio o Reserves to Broad Money.4 Argentina.4 Brail Chile China Korea.5 Mexico R to M Export Growth Relative PPP Post Crisis Others Import Share Residual

28 Figure 2. Ratio o Reserves to GDP.12 Argentina.12 Brail Chile China Korea.2 Mexico R to GDP Export Growth Relative PPP Post Crisis Others Import Share Residual

29 Figure 3. Eects o Selected Variables on the Reserves/GDP Ratio Eects o A One-Standard-Deviation Change in Explantory Variables PLDE MVGX VOLXC CRASIA CRMEX LIMY

30 Figure 4. Country-Speciic Eects.8 Reserves to Broad Money Country Speciic eects Singapore.6.4 Argentina.2 Brail Chile Korea Mexico Pakistan China India -.6 Standard Deviation:.21 Reserves to GDP Country Speciic eects 1 Singapore.8.6 Cyprus Israel Hong Kong.4.2 Chile -.2 Argentina Korea Brail Mexico -.4 ChinaIndia -.6 Standard Deviation :.17

31 Figure 5. The Time Line Beginning o Period 1: Savers deposit D, Banks use D to inance investment K1and hoarding reserves, R, D = K 1 + R End o Period 1: Liquidity shock Z materialies, reducing the net capital to K 2 ; K2 = K1 1 + θ MAX{, Z R}. Period 2: Output Y2 materialies, Y 2 = K2 ; depositors are paid D Z 1 + r. D

32 - 3 - Figure 6. Liquidity Shocks, Reserves Deposit Ratio, and Output τ 1 Figure 7. Volatility and R/D Ratio, Constant D. R/D λ The simulation values are =.33; θ =.5; ρ =.2; =.2; D = D. 15 r =