The Risks and Implications of External Financial Shocks: Lessons from Mexico

B ofGovernoorfstheFederRaelservSeystem IntematiFoimnlancDeiscussPiaopners Number535 Januar1y996 THE RISKSAND IMPLICATIONOSFEXTERNALFINANCIALSHOCKS: LESSONSFROM MEXICO EdwinM.Truman NOTE: InternatiFoinnaalncDeiscussPiaopnerasrepreliminmaartyericailrsculattosetdimulate discussainodncriticcoamlmentR.eferencienpsublicattiooInnsternatiFoinnaalncDeiscussion Paper(sothetrhananacknowledgmtehnattthewritehrashadaccestsounpublismhaetderisahlo)uld beclearweidththeauthoorrauthors.

ABSTRACT The lessons from the 1994-95Mexican peso crisis are examined from the perspective of creditors and their markets, countries that are recipients of large capital inflows, and the functioningof the internationalsystem as a whole. From each of these perspectives, recent changes in the financial world are sketched, lessons from the Mexican experience are derived, and implicationsfor policies are considered.

The Risks and Implications of External Financial Shocks: Lessons from Mexico Edwin M. Trumanl Introduction In this paper, I focus on the risks and implications of external financial shocks in the context of the lessons to be learned from the Mexican experience from three perspectives: the creditors and their markets, the countries that are recipients of large capital inflows, and the functioningof the international financial system as a whole. With respect to each of these perspectives, 1,consider how the financialworld has changed in recent years, the lessons from the Mexican experience, and the implications, in particular for policies, that should be drawn or at least examined. This three-by-three classificationsystem is somewhat arbitrary and, therefore, not entirely satisfactO~. First, to the extent that creditors are identifiedwith investors in industrial countries, recipients of capital inflows (net or gross) are identified with developing countries, and the system is identified with the governments (andthe central banks) of 1. The author 1s Staff Director, Division of International Finance, Board of Governors of the Federal Resene System. An earlier version of this paper was prepared for and presented to the Aspen Institute Seminar “The Future of the World Economy,“ August, 1995. I benefitted from comments and assistance from my colleagues Tom Connors, Allen Frankel, David Howard, Catherine Mann, Larry Promisel, Margarita Serafini, Charles Sievan, Lois Stekler, Betsey Stevenson, and Henry Terrell. Of course, none of them should be held to be responsible for any errors of fact or interpretation. This paper represents the views of the aut LI and should not be interpreted as reflectingthe views of the Board of Governors of the Federal Reserve System or other members of its staff.

, I -2creditor countries and with the international financial institutionsthat are held responsible for its smooth operation, my classificationis oversimplified. In today’s liberalized financialmarkets, potential creditors include investors in developingcountries, industrialcountries are large scale recipientsof internationalcapital flows, and the authoritiesin developingcountries as well as in industrial countries have a stake in the efficient and effective functioning of the internationalfinancial system. Thus, the notion that it is either appropriate or desirable for developed countries to operate under one set of rules while developing countries operate under another set of rules is increasinglyoff the mark. Second, the origins of the 1995 Mexican crisis can be traced, in part, I believe, to trends in the globalizationof finance over the past decade, trends with respect to the technologyof markets, the liberalizationof financial systems, Whether these and diversificationof investors’portfolios= factors contributed importantlyto what happened in Mexico in late 1994 and early 1995 or whether Mexican economic policy decisions were more decisive does not need to be agreed for the sake of my argument. I would merely stipulate that there is more in common between the behavior of financial markets during the Mexican crisis and their behavior during the ERM crises of 1992 and 1993 and the bond market collapse in 1.994than many observers

-3- 2 may be willing to contemplate or acknowledge. Moreover, the similaritiesbetween the economic policies contributing to the Mexican crisis and the ERM crises attest to the fact that the former was not a unique or unidimensionalevent.3 As a final qualification, I would note that establishing the lessons to be learned from the Mexican experience is complex. There is no consensus on the factors behind the crisis. The InternationalMonetary Fund lists three major views -- adverse domestic political and external economic shocks, an unsustainable external position, and domestic policy slippages -- and notes 4 that these views are not mutually exclusive. Moreover, the IMF’s list of explanations largely omits economic and financial trends and developments originatingoutside of Mexico. In part because there is no consensus on the factorsbehind the Mexican crisis, there is no consensus about what should have been done or not done during the crisis. Therefore, the lessons one person draws from the crisis are likely to be quite different from another person’s lessons. 2. Alan Greenspan testified before Congress on January 26, 1995, ‘althoughthe speed of transmissionof positive economic events has been an important plus for the world in recent years, it is becominq increasinglyobvious -- and Mexico is the first major case --‘that significant mistakes in macroeconomicpolicy also reverberatearound the world at a prodigiouspace.” Federal Reserve Bulletin, March 1995, page-26l.- 3. These parallels are much too interesting,complex or controversialto be explored extensively in this paper, but they are there and important to our understandingof today’s financialworld. 4. InternationalMonetary Fund, world EconomicOutlook, May 1995, pp. 90-97.

. -4- 1. The Creditors and Their Markets A. Recent Trends and Developments Table 1 below provides a summary overview of capital flows to developing countries since the early 1970s. A number of points can be illustratedby the data presented in the table. First, note the absence of lines for official capital inflows. Such flows are buried in the ‘other” lines and their importancein capital inflows to developingcountries in Asia and the Western Hemisphere has declined substantiallyin recent years. Considering total net capital inflows to all developing countries,borrowing from official creditorsdeclined from $2o billion per year from 1987 to 1990 (60 percent of total net flows) to $16 billion per year from 1991 to 1994 (11 percent of 5 total net flows). Second, net flows directly involving foreign commercial banks are also included in the ‘other”lines. This was the principal source of capital inflows from 1973 to 1982 for developing countries in Asia and the Western Hemisphere. During the debt-crisisperiod of the 1980s, countries in the latter group experienceda reversal of such flows, and bank flows played 5. These data are compiled on a differentbasis from those presented in Table 1 and intra alia include ‘exceptional financing”some of which comes importantlyfrom the official sector even for the more advanceddevelopingcountries. The comparable figures for developingcountries in the Western Hemisphere are net borrowin~ from official creditors of $7.9 billibn per year from 1987 ~o 1990 and net repayments of $0.7 billion per year from 1991 to 1994. Finally, for 22 countries classifiedby the IMF as market borrowers,net inflows from of official creditors were $2.6 billion per year from 1987 to 1.920 (14percent of the total) and $3.2 billion per year from 15Y1 to 1994 (3.6percent of total). Source: IMF, World Economic Outlook, May 1995, Tables A33, A34 and A35.

- 4a - Table1 CapitalF1OWS toDevelopinCgountries1,973-94L/ (AnnualAvera~es;inbillionsofU.S.dollars) 1973-76 1977-02 1983-89 1990-94 Alldeveloplnc~ountrie~s/ Totalnetcapitalinflows 14.8 30.5 8.8 104.8 Foreigndirectinvestment plusportfolioinvestment(net) -1.8 19.8 82.7 Netforeigndirectinvestment (3.7) (1::;) (13.3) (39.1) Netportfolioinvestment (-5.5) (-10.5) (6.5) (43.6) Other~/ 16.6 29.8 -11.0 22.2 Asia TOtainetcapitalinflows 6.7 15.8 16.7 52.2 Foreigndirectinvestment plusportfoliion~estrnen(tnet) 35.8 Netforeigndirectinvestment (;::) (::;) (R) (23.4) NetportfoliOinVeStrnent (0.1) (0.6) (1.4) (12.4) Other~/ 5.3 12.5 10.1 16.3 WesternHemisphere Totalnetcapitalinflow 13.0 26.3 -16.6 40.0 Foreigndirectinvestment 38.5 plusportfolioinvestment(nec) Netforeigndirectinvestment (::;) (:::) (;::) (11.9) Netportfoliolnvestrnent (0.2) (1.6) (-1.2) (26.6) Other~/ 10.6 19.4 -19.8 1.5 Otherdevelopinc~ountrie~s/ Totalnetcapitalinflow -4.9 -11.6 0.7 12.7 Foreigndirectinvestment plusportfoliOinvestment(net) -5.6 -9.5 10.0 8.3 Netforeigndirectinvestment (0.2) (3.2) (3.7) (3.8) Netportfolioinvestment (-5.8) (-12.7) (6.3) (4.6) Other~/ 0.7 -2.1 -1.3 4.3 MemoranduImtems: Mexico Totalnetcapitalinflow n.a. 9.7 -2.1 21.2 Foreigndirectinvestment plusportfolioinvestmen(tnet) 2.3 18.7 Netforeigndirectinvestment (::::) (1.6) (R) (4.9) Netportfolioinvestment (n.a.) (0.7) (-1.1) (13.8) Other~/ n.a. 7.4 -2.2 2.5 Source:InternationMaolnetaryFund. WorldEconomiOcutlookdatabase. &/ F1OWS excludeexceptionaflinancing(fromIMFor IBRD,andbilateral officialorprivatesectorreschedulingorarrears.) A numberof countries donotreportassetsandliabilitiesseparately.Forthesecountriesi,tis assumedthatthereareno outflowss,othatliabilitieasresetequaltothe netvalue.To theextentthatthisassumptioinsnotvalid,thedata underestimattehegrossvalue.Adjustmentasrealsomadeto theWEOdatato netouttheeffectsof bondsexchangedforcommercial bank loansindebtand debtservicereductionoperationasndtoprovideadditionadletailon selectedprivatecapitalflows. ~/ Excludescapitalexportingcountriessuchas Kuwaitand Saudi Arabia. ~/ Consistsof-netlendingto theofficialsector(includinggeneralgovernment andthemonetaryauthoritya)ndnetlendingtotheprivatesectorbybanks andnon-bankssuchas insurancceompanieasndpensionfunds.

, -5only a moderate role during the first half of the 1990s. These data, however,do not reveal the extent of the involvementof commercialand investment banks in intermediatinginternational capital flows because they include as banking flows only assets that end up on the balance sheets of those financial institutions;however, these institutionsare heavily involved in the placement and arrangementof portfolio capital flows. For example, when the peso crisis erupted for Mexico in December 1994, Mexican commercial banks had about $4 billion in certificatesof deposit outstandingto nonresidents;an overwhelmingproportion of those deposits had been placed or brokered by foreign financial institutions. Third, the absolute and relative importanceof net foreign direct investment has increasedsubstantially. This trend reflectsa widespread belief that this type of capital inflow has advantages in terms of relative stability and the countercyclicalnature of the associatedservicing requirements, and consequentlya more hospitableattitude in recipient countries toward such inflows has developed over the past decade or so. Fourth, the increase in net portfolio investment has been particularlydramatic. These flows take many forms, including investments in equity markets well as investmentsin marketable debt instruments -- denominatedin domestic as well as foreign currencies. In consideringthe implicationsof the recent Mexican experience, it is importantto distinguish among these subcategoriesof portfolio investmentsbecause the

-6investmentsinvolve a variety of different risks--pricerisk, liquidity risk, and exchange rate risk. Broadly speaking the investorsbehind those flows are of two types: direct holders of the instrumentsin question or indirect holders through investmenttrusts or mutual funds. Whatever the type of investor, they are seeking to maximize their return given their appetite for risk. unlike many direct investors,portfolio investorsoften have relatively short time horizons regardlessof the maturity of the underlying instruments. unlike direct investorsand traditional commercial bank lenders, they assume that they can liquidate their investmentsrelativelyquickly in well-developedtrading markets. Moreover, relative near-term rates of return are important for some instruments,as may be considerationsof capital gains and losses. Fifth, in consideringpatterns of net capital flows to developing countries, not all countries or groups of countries are the same. The Asian countries all along have received a larger proportion of their net inflows through foreign direct investment. The Western Hemisphere countriesexperienced a reversal of inflows from banks in the 1980s, embedded in the ‘other” line, and they were relatively large beneficiariesof net portfolio inflows during the first half of the 1990s. Finally, the pattern of net flows to Mexico shown in the memorandum items has been broadly the same as that shown for the

-7- 6 Hemispherecountries as a group. B. Lessons from the Mexican Experience The lessons that observers draw from the Mexican experience for the creditors and their markets depend importantly on the perceiveduniqueness of the Mexican circumstances.7 As we have seen, portfolio capital flows were important for Mexico, but its situationwas similar to other countries in Latin America. It is true in the Mexican case that the portfolio inflows were concentratedin instrumentswith relatively short maturities that were also readily transferable. It is also true that by the time the crisis hit a large proportion of those instrumentswere Tesobonos whose peso value was linked to the value of the dollar. At the end of 1993, foreignersalso were very large holders of Cetes -- short-termgovernment securities whose value was not linked to the dollar. Indeed, one of the curiositiesof the Mexican experiencewas that over the course of 1994 international (Mexicanas well as foreign) investors as a group got out of Cetes but they willinglygot into Tesobonos, instrumentspaying a much lower interestrate than Cetes but a higher rate than similar U.S. Governmentobligations. This trend continued even during the period after the middle of 1994 when it became clear that the outstandingstock of Tesobonos was larger than Mexico’s foreign exchange holdings. By definition, the 6. In the period 1990-94, net portfolio investmentwas about two-thirds of net capital inflows to Mexico and the developing countries of the Western Hemispherecomparedwith just under a quarter for developing countries in Asia. 7. The same qualificationapplies to the lessons for the recipients of capital inflows (SectionII) and for the internationalfinancial system (SectionIII). ,.

-8return associatedwith Cetes involved both an exchange rate and a credit (transfer)risk, while the return associated with Tesobonos involvedprincipally the latter, but both were 8 substantial. A third ‘fact” about the Mexican situationmay help to explain the relatively large ex ante gap between the rate on Cetes and that on Tesobonos: the internationalfinancial community led by the U.S. authorities had come to the financial assistance of Mexico on numerous occasions over the period dating back to 1976. The size and novelty of these operations may have suggested to investorsthat Mexico was different if not unique. Alternatively,investors did not fully appreciate that Tesobonos were a potential indirect claim on Mexico’s international reserves. Nevertheless,when the crisis broke, the objective situation was one in which a large number of geographically dispersed investorswere caught holding short-termclaims on Mexico that could not be serviced without incurring a massive short-run depreciationof the peso. They realized that their investment strategieshad been based on one or more false premises concerningthe nature of Mexico’s exchange rate regime or the probability that they could liquidatetheir holdings before any crisis hit. While it is difficult to prove, a third 8. For example, at the end of November, 1994, the 91-day Cetes rate was 15.60 percent, a spread of 988 basis points over the U.S. 3-month Treasury bill rate of 5.72 percent, while the 91-day Tesobono rate was 7.49 for a spread of only 177 basis points. Technically, there was some exchange rate risk originally associated with Tesobonos because their principal was only indexed to the dollar, but it was paid in pesos and the holder had to handle or cover the conversion of the pesos received into dollars.

-9premise may have been that in the case of a crisis ‘bondholders~ would not be affectedbecause even during the severe debt crises of the 1980s there were only isolated instancesof failures by countries to meet the original terms of this type of obligation. Finally, it is possible that investorshad excessivelybought into the so-called Washington consensus that the policy regimes in Mexico and similar countries had fundamentallychanged in a direction that would produce sustained,rapid economic 9 expansion. In the event, in 1994-95 holders of some types of portfolio claims on Mexico suffered losses and holders of other types did not. Holders of equity securitiessuffered losses, or at least paper losses; the Mexican stock market dropped by twothirds (68 percent) in dollar terms between December 19, 1994 and its low on March 9, 1995, and as of the end of June was still about 45 percent below its level prior to the peso devaluation. The remaining foreign holders of Cetes as well as domestic holders of those instrumentssuffered losses as well when their instrumentsmatured over the course of 199s. Holders of the Tesobonos have not in the end suffered losses; the Mexican government has been able to honor its obligations,initially out paying out pesos and meeting the resultingdemand for dollars of its reserves and later paying off foreignholders of Tesobonos directly in dollars. However, on a marked-to-marketbasis, holders of Tesobonos suffered non-trivial,but temPorarY,Paper I 9. See Paul Krugman, ‘Dutch Tulips and Emerging Markets,“ Foreiqn Affairs, vol. 74, number 4, July/August1995, pp.28-44.

- 10 losses as well, as did holders on other longer-datedMexican debt instrumentssuch as Brady bonds. Nevertheless,the widespread perception is that portfolio investors in Mexican paper suffered no losses as a consequenceof the peso crisis and on the whole were well-rewardedfor the limited risks they had taken. When the crisis erupted, investorspanicked, not only investors in the Mexican stock market and in Mexico debt instrumentsbut also investors in similar instruments issued by borrowers in other countries, especially countries in the same part of the world or perceived to be in similar circumstances. These contagion sales of assets were induced by at least two types of forces. First, as perceived risks rose and expected returns fell, individualinvestors were induced to disinvest. Second, institutionalholders such as mutual funds faced with actual or threatenedredemptionswere led to liquify their holdings not only of Mexican paper but also of the paper of other countries especially if they could do so while limiting their capital losses. These patterns can be seen in the sympathetic movements in the stripped yields on Brady bonds of various countries in Latin America and elsewhere -- see Charts 1 and 2. Whether they deserved it or not, the wealth of Mexico’s external creditors as a group was only marginally affected by losses followingthe crash of the peso. The principal reason was that the investorswere numerous and Mexican paper was not a large portion of any final, non-Mexican investor’s total portfolio; a secondary reason was that some of the investors benefitted from the actions taken to stave off a larger cris~s.

. Lua — ——.——. —— .—.— —- — =s = — = I I I I I I , I I 8 , 1 *— I I 1I I ! I I 4 I , 1 n i 1 ! — I 1 I u 0 , , I 1- I UI 1 1 I +’, t I 1 I t 1 , I * 1 t 1 , I t I 1 I 1 I 1 I I I t I 1 I 1 u -, LIJ ---,-----_—--. d I m ! 1 I s I I I I I 1 I 1 1 I 1 , I I I 1 1 I —— — .— - —. I 0 0 0 0 0 0 06 R 0 0 0 0 m 0 m % ml ml I —

—..—-—— — .— .— —..— —.- — I I I I I f I , 1 = — s . I I w 1 I , 1 d t I I I I I ln I ‘4 1 1 I ( I 1 ( I ! I I I , I I ‘i ! 1 I t It 1 1 1 I 1 I t1 , I I f I I I 1 , I t t I ! 8 I , ! I I I 1 I A 4 I 1 I I 1 1 t I I I I I I ,- , I 1 I I I i w u I I 1 1I I I I $ & ! I — ._J —. — — .-— — — — --

. -11- However, 1 would submit that the Mexican situationwas not unique. As long as no major institutionor group of institutions is heavily invested in claims on such a country or a group of similarly situated countries,creditors and their markets are likely to suffer limited damage. Consequently,they are unlikely to be motivated to act in concert to limit these losses on their investments;they have every incentive to step back from their investmentsand to seek to dispose of them quickly thereby adding to pressures in financialmarkets. On a global basis, portfolio investmentsin developing countries amounted to about $250 billion as of the end of 1994 (see table 1) . This is a large number, but it represents less than 1/2 of one percent of total 10 portfolio holdings of investors in industrialcountries. While they would not have been happy loosing, say, half the value of their investmentsin developingcountries, the direct aggregate consequencesin terms of lost wealth, welfare and demand were likely to be inconsequential.ll One fact is clear in Mexico’s case and applies with roughly equal force to many other countries: In contrast with the situation in the debt crisis of the early 1980s when a small 10. As of 1992, the total GDP of high-incomecountrieswas three times U.S. GDP (Source:IBRD, World Develo~mentRe?aort,1994) and as of the end of 1994 U.S. households’financialwealth equalled $18 trillion (Source:U.S. flow of funds accounts); if the wealth/GDP ratio for all high-incomecountries is the same as the U.S. ratio, financialwealth of households in high-income countries equalled about $54 trillion when the peso crisis broke out. 11. However, it is possible that in the future portfolio investmentsin these markets may become a larger share in ~l~bal portfolios. Moreover, the indirect consequencesfor globa. growth of the hypothesizedloss in value in early 1995 might have been substantial.

- 12 number of internationalcommercial banks -- roughly 25 in total -held a very dominant share of Mexicois debt, in 1994-95 that was not the case. In 1982 commercialbanks accounted for 70 percent of Mexico’s external debt, and claims on Mexico by the top nine U.S. commercialbanks amounted to 50 percent of their capital. In contrast, at the end of 1994, less than 40 percent of Mexico’s internationaldebt was held by foreign commercial banks, and had Mexico defaulted,the consequencesfor those institutionswould have been painful but not life threatening. In the case of the same group of major U.S. banks, claims on Mexico at the end of 1994 representedonly 15 percent of their capital. C. Implications The Mexican peso crisis of 1994-95 is likely to be unique in at least one respect: in future liquidity crises, holders of large amounts of portfolio claims on the country facing the crisis are much less likely to be made whole as the consequenceof official actions that provide financial assistance 12 to the country in question. However, the fundamentalpoint is that the scale of potential financial assistance needed to stave off a full-blowncrisis in Mexico has proved to be much larger than anyone could have imagineda year ago, and the scale of any similar operation in the future (even after allowance for the special circumstancessurroundingthe Mexican case) is likely to be larger than the official sector will be able or willing to Moreover, like it or not, true or false, as noted 12. This implication1s related, in part, to the discussion in Section III below.

. - 13 above, the widespread perception is that many portfolio investors were inappropriatelyprotected from the consequencesof their investmentdecisions. A further implicationof the Mexican experience is that investorswill be, or at least should be, more careful in the future. At a minimum, they should fine-tune their early-warning systems. Many of them reportedlydid not understanddevelopments in Mexico in 1993-94; if they had, they would not have invested so heavily in Tesobonos. on the other hand, many of the investorsor at least their advisors, in fact, did understandwhat was going on in Mexico in 1994, and those developmentseither w ignoredby the managers of the investmentsor they believed that they could get out before a full-blowncrisis erupted. Just as in the ERM crises of 1992 and 1993, many investorswere mistaken. Thus, a third implicationof the Mexican peso crisis is that institutions should pay more attention to their risk management systems in the broadest sense of that term. High or higher yields on debt instrumentsshould serve as signals to their holders that compensation is being paid in advance for the costs of a possible default or capital loss. Holders of portfolio claims on developingcountries,as well as the financial institutionsthat are involved in placing the instrumentswhether or not they continue to hold any in their own portfolios, should plan on the basis that the next sovereign liquidity crisis will not unfold in the same way as the Mexican

- 14 - 13 crisis unfolded. Their planning should have two dimensions: They should expect to take more extensive losses. In part to improve the risk-rewardtrade off, they may want to consider how they might participate responsibly in the ex post resolutionof crisis situations. II. The Recipientsof Capital Inflows A. Recent Trends and Developments The changing pattern of internationalcapital flows has both a supply side and a demand side. Above we considered primarily the demand for a different mixture of investmentsthan had been characteristicof the 1970s or 1980s, but the supply side is also important. From the standpoint of the recipient countries, the 1990s opened up new opportunitiesto attract foreign capital. In part those new opportunitieswere the consequenceof changes in the governing political and social philosophies and 14 economic policies of the recipient countries. They became more hospitable to foreign direct investmentby relaxing restrictions, rewriting discriminatoryregulations,and reworking the landscape of the public sector through massive privatizationprograms that, in turn, meant that portfolio investments in equity securities became more attractive. These forces of economic reform led to more flexible economies, economies that in principle were better 13. Again, this implicationdraws in part on the discussion in Section III below. 14. As noted above, some have argued that this so-called ~Washingtonconsensus” on policies may have been overblown o~ overinterpretedin terms of its short-term implicationsfor growth.

15 equipped to respond to shocks. However, they involved two-way risks, at least potentially,because the recipient countries to some extent became more exposed to the risk of a sharp change in investor sentiment. Funds that flowed in and easily financed current account deficits could also easily seek to flow out if conditionsor perceptions changed. This proposition about the increased risk to the recipient country is debatable,and it deserves closer scrutiny than it can receive in the paper. However, let me illustrate what I have in mind. To induce foreign investors to hold claims on developing countries in the form of marketable debt instruments,the recipient councrieshad to compensate investors for the potential risks involved. However, by qualifyingto borrow in these markets, even if they should have been considered analogous to high-risk borrowers in domestic markets, borrowers in developing countries began co compete with a broader group of potential borrowers, not just other developing countriesbut issuers of bonds in developed countries as well. Moreover, the competitionwas based on judgmentsconcerning the adequacy of the returns considering the risks involved -- evaluationsthat are relatively easy to make, at least in principle. But the comparisonsare inherentlymulti-sided. Thus, when yields declined on the bonds issued by industrialcountries,yields on instrumentsissued by developingcountries became relativelymore attractive. (Investorsbegan to reach for higher yields, perhaps, as discussed above, not being as fully informed as they might have been about the risks involved.) Similarly,when

- 16 yields in industrialcountries rose in 1994, those offered by developing countries such as Mexico became relatively less attractive. The issue is whether as a consequence of these structural changes, the borrowing countries become more vulnerable to external financial shocks. B. Lessons from the Mexican Experience The principal lesson from the Mexican experience in 1995 for recipients of capital inflows derives from the size, scope, and speed of the crisis once it broke. By the standards of the 1980s, this was a new world. The earlier Mexican crisis took about six months to develop from the peso’s devaluationin February 1982 to the Mexican weekend in mid-August. Once that crisis erupted, it was not until December that mechanismswere more or less fully in place to contain the situation. Meanwhile, about $3-1/2 billion in bridge loans in August sufficed to buy time to put more permanent solutions in place, but it took three months before an agreement with the IMF was completed to form the other end of the bridge. In 1994-95, the pre-crisisperiod lasted about a month, from mid-llovemberto mid-December. An $18 billion package of promised short-term financialassistance was developed within two (holiday)weeks, and by mid-January it was clear that the classical 1982 type of approach had failed to arrest the downward spiral of confidence. The U.S. Administrationsought, and initially received, Congressionalsupport for a $4o billion program of guarantees for Mexican Government borrowings in

. - 17 internationalmarkets to refinance its short-termdollar and dollar-linkeddebt. That approach was abandoned on January 31 in , favor of the approach now being followed. Thus, the crisis phase lasted a mere six weeks it took another six weeks until mid- ; March before confidence began to return to Mexico.15 of course, there were many important differencesbetween Mexico in 1982 and Mexico in 1995 that make comparisonssomewhat problematic. However, there should be no doubt that the 1982 approach quickly proved to be inadequate in 1994-95,whether it deserved that fate or not. A second lesson from the Mexican experience is merely a variation on a familiar, long-standingtheme: If a country is going to run a large current account deficit financed by net private capital inflows, it needs to be careful to ensure that the funds are being wisely invested. This is the first principle for any type of borrowing. In the internationalcontext, it is relatively easy to articulate,but much more difficult to apply. Howeverr it is clear in the Mexican case that its current account deficit was being driven in part by a decline in national savings from more than 18 percent of GDP in 1988-90 to less than 14 percent of GDP in 1994 and there was essentially no change in 16 gross domestic investment. Thus, Mexico’s domestic savings rate was relatively low, and when the Mexican economy increased its reliance on foreign savings very little of it went to 15. See the charts on strippedyields on Brady par bonds. 16. InternationalMonetary Fund, World Economic Outlook, May 1995, page 92.

- 18 increased domestic investment.17 Three other obvious lessons of the Mexican experience deal with other aspects of governmentalpolicy: First, countries should not be tempted to try to sustain overvalued exchange rates too long; this is another principle that is easy to articulate but not so easy to apply. Second, an easier lesson to apply is that in their debt management policies countries should avoid excessive reliance on short-term borrowing; since foreign as well as domestic investors buy internal as well as external Mexican debt, this lesson clearly applies to both areas of debt management. Third, when a country devalues or otherwise is forced to change its exchange-rate regime, there is a strong presumption that there should be compensating and complementary changes in other macroeconomic policies. In the Mexican case, this did not happen immediately either because the authorities were paralyzed by their governmental transition or they did not understand the fundamental issue. Supporting the ‘denial” interpretation is the fact that the Mexicans did not request IMF support until during the first week of January. A final lesson from the Mexican experience I put forward more as a hypothesis than a firm conclusion. Has the changing nature of international capital flows left recipient countries more vulnerable to shocks? On the negative side, one can argue that countries can more easily attract Capital flows and that 17. This is what the aggregate statistics show. Arguably, .ith the increased flexibility and openness of the Mexican economy, the actual investments were more efficient and productive in the 1990s than earlier.

. - 19 they are now more open and can more easily do without the capital inflows and adjust to its 10SS with less (not zerot but less) pain -- in terms of lost output. On the affirmative side, it might appear that the unforgiving nature of capital markets implies that countries are more susceptible to severe punishment (in terms, again, of lost output) for marginal policy errors; on the other hand, increased market discipline contributes to more 18 responsible policies. As a practical matter, whether borrowing countries are more vulnerable to shocks today or not, they are less likely to receive much cooperation from their creditors in . helping to cope with a crisis once it has erupted because individual creditors are more numerous and dispersed with less of a stake in the success of failure of efforts to resolve or contain a financial crisis. C. Implications Regardless of where one comes down on the issue of whether capital-importingdeveloping countries are more vulnerable to shocks in today’s globalized capital markets, one implication for the recipients of large scale of net capital inflows is that the authorities in these countries will need to pay a good deal more attention than they have in the recent past to potential shocks both external and internal. They will need to develop their O- early warning systems. These SYStemS should differ from and be independentof the early warning systems used 18. This M a lesson Involving today’s global financial ma kets that is far from uni~e to Mexico’s situation. It is central to the evaluation of the Ew crises of 1992 and 1993 and the behavior of bond markets in 1994. It is also subject to dispute.

- 20 by investorsor those used by official internationalfinancial organizationsbecause the requirementsand risks are inherently different. one area to which particular attention needs to be paid in shock proofing the economies of the borrowing countries is the domestic banking system. Either because those banking systems lack the managerial or financialstrength to exploit effectively liberalized financialmarkets, because national supervisorysystems are underdeveloped,or because of the discipline that internationalcapital flows exerts over macroeconomicpolicies (withconsequentstrains on the banking systems), such shock proofing is clearly needed. In the Mexican case, all three rationales were present. The newly privatized banks lacked strength and managerial experience, the effective tenor of their foreign currency liabilitieswas much less than that of their correspondingassets, the supervisory system was underdeveloped,and the weaknesses of the banking system contributedto the reluctance of the authorities to take the macroeconomicpolicy steps that would have been necessary to contain the peso crisis once it appeared that a devaluation was inevitable. A second set of implicationsconcerns macroeconomic policies in the recipient countries. Many advocates of the use of exchange rates as nominal anchors have been forced by recent events to retreat somewhat from their advocacy; it would be unfortunate if the pendulum now swung to the other extreme of

-21- 19 The search for a absolutely freely floating exchange rates. workable, happy medium must continue. At the same time, recipient countrieswill need to rethink how they calibrate their monetarypolicies, their debt management policies, and their fiscal policies. It follows from what I wrote above that I believe fiscalpolicy has a role to play in striking the proper balance between savings and investment,that is with respect to judging and achieving a sustainablecurrent account balance, not only among recipientsof net capital inflows among industrialcountries like the United 20 States but also for developing countrieslike Mexico. In the face of unwanted capital inflows, which was Mexico’s situation in 1992 and 1993, countries face difficult choices. Either fiscal policy should be tightened further even if it involves running a substantialfiscal surplus, or the real exchange rate must be allowed to appreciate,or capital inflows must be sterilized and reserves built up which often has negative fiscal consequences since interest receiptson external reserve holdings are less than interestpayments on domestic obligations’ or there should be resort to controls on capital inflows, or some combination. 19. Advocates of permanently fixed exchange rates and currencY boards do not appear to have tempered their advocacy by much. 20. William Cline points out in his retrospectivelook at the debt crises of the 1980s that the flaw in Nigel Lawson’s dictum that current account deficits don’t matter as long as theY are accompanied by balanced budgets or SUrPIUSeSaPPlles ewallY to developing countries and developed countrles~and he correctlY diagnosed this flaw as applying to the Mexican case well b~fcre the crisis broke. William R. Cline, InternationalDebt “ Reexamined, Institute for InternationalEconomics, Washington/ DC, February 1995.

-22- The capital controls ‘solution”has attracted an increasedamount of favorable attention in some quarters in the 21 aftermathof the Mexican crisis. However, I am skeptical about this latest turn in internationalmonetary fashion. In many cases, it is only the countrieswith very sound macroeconomic policies and high domestic savings rates that can afford to limit capital inflows, and even they pay a price by distorting intertemporaldecision making. Even when they do cut themselves off from some kinds of inflows (e.g.,short-termborrowing), they are reluctant to cut themselvesoff from other kinds of flows (e.g., i markets or in the form of trade credits), and once the possibility of allowing some forms of short-term or portfolio capital inflows is opened up, the nature of any ensuing crisis is at most a matter of degree. Moreover, the notion that capital controls are a good idea for developing countries but a bad idea for developed countries runs counter to the truth that at the margin these two groups of countries cannot and should not be distinguished.22 21. See, for example, the 1995 Annual Renort of the Bank for InternationalSettlements, “emergingeconomies should perhaps be..more prudent in dismantlingcontrols on short term capital inflows,” p. Slo. 22. Larry Summers has expressed my bias with his characteristic zing, ‘it is clear that we would all rather live in countries in which capital is trying to get in, rather than in countries from which capital is trying to get out. That suggests that coun’ries should be very cautious about imposing capital controls with the objective of discouraging capital inflows.n Remarks at Symposium on Capital F1OWS, Jerusalem, Israel,April 3, 1995.

- 23 - III. The Functioningof the Svstem A. Recent Trends and Developments The principal change in the functioningof the internationalfinancial system in recent years has been the diminished role of governments. This trend is not only evident with respect to the process of privatizationand market-opening reforms in the non-industrialworld, it also has been manifested in the trend toward deregulationin the industrialworld. The Ministry of Finance no longer has quite the unchallenged power and influence it once had in Japan, and financialmarkets have become increasinglyderegulated in all industrialcountries.23 This trend toward deregulationhas been driven by some of the same forces that are behind the globalizationof financial markets and financial flows: technologicalchange and improvements in global communications. It has facilitatedthe relative rise in the importanceof securitiesmarkets and the relative decline in the role of depository institutionsas direct financial intermediaries;that is, institutionsthat book both assets and liabilitieson their balance sheets. While there has been no observed trend toward increased volatility in those markets for financialassets that have been freely functioning for extended periods of time, for example, the market for U.S. Treasury securitiesand spot markets among the major currencies, recorded volatility has increased in markets 23. One does not need to go so far as to argue that central bankers are like the little Dutch boy with his finger in th~ &ike against the onslaught of stateless money as Steve Solomon dues in his Confidence Game (Simonand Schuster, 1995) to recognize that the internationalfinancial system has changed.

- 24 that previouslywere controlled. Where previously prices were tightly controlledwith the result that sharp movements were ruled out or transactionswere never consummated,now prices are allowed respond to shocks. As noted above, the authoritieshave responded to these developmentswith a mixture of fear and awe. At one extreme they are concernedby the scale of potential disturbances,appearing to be handcuffed in their efforts to implementappropriate macroeconomicpolicies. At the other extreme, they have sought to exploit new opportunities includingnew ways of raising money. Could the Brady bond market have developedwithout the debt crisis of the 1980s and without the financialtechnology availableto support it? Without this market providing valuation benchmarksfor trading in securitiesof developing countries, would it have been easy for borrowers to price and come to market with other securities? These are difficult questions on which to reach firm conclusions. However,my answer to both questions is negative. B. Lessons from the Mexican Experience The principal lesson from the Mexican experience for the functioningof the internationalfinancialsystem is that the authoritieshave been required to rethink how they interact with the market in crisis situations. As noted earlier, gone are the days when the G-1o central banks could assemble a bridge loan in a few days that would serve to stabilizeexpectations about a major borrowing country’s situation. Also gone are the days when the Managing Director of the IMF and the Chairman of the Boa~d of

-25- Governors of the Federal Reserve System could get representatives of 15 major private internationalfinancial institutionsin a room and easily convince them that a systemic crisis is, first and foremost, a crisis for their own institutions. The number of major players is now much larger and each of them perceives that . it has less of a stake in the successfulresolutionof a crisis situation. Thus, when the Mexican authoritiesin December 1994 called upon the commercialbanks to assemble a line of credit to help Mexico cope with what appeared to be a liquidity crisis, the commercial bankers’ principal focus was on the terms of the deal rather than on the rationale for the deal. Whether this was a short-sightedor mistaken judgement is open to debate. A closely related lesson concerns the lack of consensus in the official community about the nature of the Mexican crisis and whether it involved so-called ‘systemicrisk.~ From a broad perspective there were four possible elements of systemic risk in the Mexican situation. First was the risk to banking systems in countries other than Mexico; this narrow definition of systemic risk focuses on depository institutionsthat are the core of monetary and payment systems and that have access to governmentalsafety nets for depository institutions. While bank claims on Mexico in early 1995 were a smaller share of Mexico’s debt than in 1982, a fullblown Mexican crisis, which could have affected a number of other

- 26 major borrowingcountries, could have been a real threat to at least some national banking systems.24 Second was the risk to the internationalfinancial system more broadly defined covering not only depository institutionsbut other types of financial institutionsand extending to stock and bond markets around the globe. I argued above that the 10SS of financialwealth as a consequence of contagion from the Mexican crisis was not likely to have been large enough by itself to have had a major impact on wealth, welfare or demand in the industrialcountries,but adverse knockon financialor psychologicaleffects could not be ruled out. Third was the risk to economic activity around the world, the possibilitynot only that the Mexican economy might go into a deep recessionwith negative spillovereffects but also that the Mexican crisis might spread to other borrowers and impart a global deflationary impetus of considerablesize. From the perspectiveof the end of December 1994, this risk was not seen as either very large or very troublesomecoming off the euphoric growth performance in most of the industrialcountries in 1994; from the perspective of mid-1995 against the background of slowing growth in many industrialcountries (manywith still very high unemploymentrates) and deep fissures in the Japanese financialsystem, this risk might be evaluated differently. Finally there was the risk to the global trend toward market-orientedreforms that had swept the developing world over 24. Based on BIS data, which are not fully comparable for th= two dates, bank claims on non-OPEC developingcountries rose from $247 billion in December 1982 to $489 billion in December 1994.

- 27 the previous decade, drawing into the mainstream not only other countries in Latin America and the economics in transition in Eastern Europe and the area of the former Soviet Union but also countries such as China and India. If the authorities in Mexico, which, rightly or wrongly, was perceived to be a leader in this trend, concluded from their experience that they had chosen the wrong model and had reverted to a model emphasizingnon-market solutions, so the argument went, then what would be the reaction in other formerly like-mindedcountries? Whether this is a relevant considerationunder the heading of ‘systemicrisk~ debatable, but that the authoritiesof most major borrowing countries sat down in January to consider the implications the 25 Mexican situation for their strategies is a fact. My point is not primarily to argue that all these elements of systemic risk were present in the Mexican situation although I think they were, noting that risk means a non-zero probability. My point is that a lesson from the Mexican experience is that there was no consensus about the nature of the systemic risk involved, to say nothing of the size of that risk. Consequently, it is not surprising that there was a lack of consensus in the official community, as well as in private 26 financial markets, about what to do about the problem. 25. Again, see Krugman for a contrarianview: the Mexican peso crisis marked a healthy “beginningof the deflationof the Washington consensus,” op. cit. p.31. 26. This lack of consensus has been further exacerbatedby the success of internationalefforts to stabilizeMexico’s exte]~ai financial situation. Some argue that it proves that the meaicine was necessary while others argue, incorrectlyin my view, the reverse.

-28- A final lesson from the Mexican experience concerns the issue of transparencyand markets. It is a lesson for the internationalfinancial system because transparencyand the role of markets affect how the global financial system functions. In retrospect, it is clear that the Mexican authoritieswere less than fully forthcomingabout their economic and financial situation; they were more transparent than critics in the market have argued, but they were not as transparentas they might have been. It was inappropriatethat until early 1995 an important country like Mexico announced its internationalreserve position only three times a year x when it was otherwise convenient. It was an understandableanomaly that the Mexican authorities felt that it was in their interest to prevent the developmentof forward or futuresmarket contracts in pesos. It was understandablebecause financial authoritiesare often behind the curve in such matters. It was an anomaly because it was inconsistentwith other elements of market-orientedreform in Mexico. Some observers argue about the proper phasing of financial sector reforms in countries like Mexico and would like to slow down such reforms, but I would argue that in the Mexican case the absence of financial market facilities,such as a forward or futures market, to absorb pressures associated with the peso’s devaluationwas one, but only one, of the reasons why the peso crisis of 1994-95 was more virulent than the ERM crisis of 1992.

-. - 29 - C. Implications The first implicationfor the internationalfinancial system from the Mexican experience is that a better consensus needs to be established about the nature of systemic risks in these types of situations. That evaluation should take full account of the moral hazard implicationsof adopting too broad-or explicit a definition. What were the stakes of the international financial system and the world economy as Mexico was forced to devalue the peso in December 1994? What were the potential systemic implications? The U.S. authoritiesdid not see them the same way as did the authorities in some of the other major 27 countries. The second implicationis that efforts to understand the functioningof financialmarkets and to safeguard their integrity should not be confined to markets in the industrialcountries. This, again, is a respect in which the Mexican experience revealed that there is a continuum extending from the most sophisticatedtrading in foreign exchange markets involving the major currencies to domestic financialmarkets in developing 28 countries. Third is the implicationfor preventativeactivities. How best can the internationalfinancial community (private sector as well as public sector, including the international 27. This is not a clean distinctionbecause disagreementsabout nature of the threat were mixed with disagreementsabout whose responsibilityit was to meet any threat. 28. If the reader is not convincedby my assertion,conside I~he debate in early 1995 about how and why the dollar-markand dollar-yen exchange rates may or may not have been affected by the Mexican peso crisis.

- 30 financial institutions)organize itself in advance to increase the probabilitythat Mexican-type situations either do not arise or do not involve such massive shocks either to the economy of the country directly involved or to the world economy and financialsystem? Among the elements of better prevention are increased transparencyand provision of data to markets as well as the three types of early warning systems that were discussed in Sections II--one each for the recipient country, the market participants,and the official internationalfinancial organizations. Fourth, assuming that prevention is only 90 percent of any cure (atbest), what should be the role of international rescue operations in such circumstances? Here there is the beginnings of a consensus in the statement that came out of the 29 Halifax Summit. While there is little objection to the principle that multilateralfinancial support should be 29. The Halifax Communique stated: If prevention fails, financialmarket distress requires that multilateralinstitutionsand major economiesbe able to respond where appropriatein a quick and coordinatedfashion. Financingmechanisms must operate on a scale and with the timelinessrequired to manage shocks effectively. In this context we urge the IMF to: establish a new standingprocedure -- ‘EmergencyFinancingMechanism” -- which would provide faster access to Fund arrangements with strong conditionalityand larger upfront disbursementsin crisis situations. To support this procedure, we ask: the G-10 and other countries with the capacity to support the system to develop financifig arrangementswith the objective of doubling as soon as possible the amount currently available under the GAB [GeneralArrangements to Borrow] to respond to financial emergencies.

. - 31 potentiallyavailable to deal with certain crisis situations, there is considerabledifferenceof view about how to define those crisis situations,whether it is realistic, in light of trends in internationalfinancialmarkets, to think that the multilateralinstitutionscan mobilize enough financial resources to deal with the ‘next MexicoN or the ~fifth Mexico” thereafter, and how to deal with these situationsoptimally (themoral hazard issue). A final implicationof the Mexican experience for the functioningof the internationalfinancial system is whether there is a better way to manage these crises? Such an examinationrealisticallymight proceed on the assumption that all crises will not, and perhaps should not, be preventable. The analysis might also assume that there will be a perceived need to try to manage a crisis so that it does minimal damage to the functioningof the internationalfinancial system and the world economy; in other words, the option of leaving the country to work out its problems with the market will not be attractive in all circumstances. Finally, the analysis might assume that sufficientexternal emergency resourcesmay well not be available to handle all such situations. Put this way, the answer to the question of whether there should be a better way to handle these crises obviously is yes. An obvious answer to a complex question suggests the need to examine the stated assumptions. At the same time there is a need to examine the possible modalities of more orderly workout arrangementsgoverning internationaldebt crises .& than are currently available, for example, whether an offic~ally

- 32 sanctionedstandstillprocedure that potentiallywould govern all external financialrelations of a country in a crisis situation 30 would be either feasibleor desirable. It is premature to concludewhat the results of such an examination should be, but it is not unreasonableto undertake it. 30. The Halifax Summit cautiously endorsed such an examination, “recognizingthe complex legal and other issues posed in debt crisis situationsby the wide variety of sources of internati~nal finance involved,we would encourage further review by G-10 Ministers and Governors of other procedures that might also usefully be considered for their orderly resolution.u

I l -33- InternatioFnianlancte)iscussPiaopners IFDP m Ii .1996 535 Implications 534 533 GeographVyS. JohnH.Rogers 532 of 531 A 530 529 H 528 527 I 526 M 525 524 I 523 A I

-34is 516 M countries D. a S .. 510 Options.

,0 l.