Financial Crises
World Capital Markets to Blame?

A conference on ‘World Capital Markets and Financial Crises’ was held at the University of Warwick, Coventry, on 24/25 July 1998. The conference was organized by CEPR, in conjunction with the Economic and Social Research Council’s Global Economic Institutions (GEI) programme. The organizers were Marcus Miller (University of Warwick and CEPR), William Perraudin (Birkbeck College, London, and CEPR) and Jonathan Thomas (University of Warwick). There were ten papers, together with a roundtable discussion on crisis prevention and management in global capital markets.

The aim of the conference was to discuss the origins and implications of the recent East Asian financial crisis. The crisis involved neither the fiscal profligacy, nor the political temptations to devalue (to reduce unemployment), which had lain at the root of ‘first- and second-generation’ models of speculative attack. Instead, the source of the turmoil seemed to lie in excessive short-term dollar liabilities, and over-optimistic bank lending in an under-regulated environment.

This theme was developed in ‘Paper Tigers?’ presented by Giancarlo Corsetti (Università di Roma Tre and Yale University) – joint work with Paolo Pesenti and Nouriel Roubini. The authors offered a ‘preliminary’ assessment of the Asian crisis in two stages. First, they analysed the effects of promised government bail-outs; and second, they provided empirical evidence on the roles played by fundamentals in the build-up to the crisis. Although fiscal deficits were low in East Asia, the authors argued that government bail-outs could be expected to help translate current external private debt into future public liabilities. Assuming perfect foresight, therefore, speculative attacks could be expected to take place when the stock of external debt was sufficiently high to induce expectations of a sustained money expansion. Using data from 24 emerging economies, including those in East Asia, the authors concluded that both the magnitude of current account deficits (which were a measure of external imbalance), and the share of non-performing loans (which were a measure of the fiscal costs of financial bail-out), would be crucial in determining the likelihood and severity of a crisis.

But do people really look ahead to the future cost of government bail-outs prior to a crisis? Richard Portes (London Business School and CEPR) was sceptical about the degree of foresight implied in this behaviour. He also questioned the robustness of the empirical support claimed by the authors, and wondered why there was no explicit role in their model for the real exchange rate, which was one of the key elements in determining the domestic burden of external debt.

In ‘Asian Currency and Financial Crises: Lessons from Vulnerability, Crisis, and Collapse’, David Vines (Institute of Economics and Statistics, Oxford, and CEPR) and Jenny Corbett (Nissan Institute of Japanese Studies, Oxford, and CEPR) explored the interactions between currency and financial crises, and argued that it was these interactions that led to collapse. The critical feature was the prior build-up of massive unhedged foreign borrowing, which meant that, when the currency initially devalued, foreign depositors no longer trusted the government’s guarantees. Fears of sovereign insolvency thus triggered capital flight and currency collapse. The paper ended with a discussion of the implications for crisis management and the lessons for global economic institutions. Christopher Bliss (Nuffield College, Oxford) noted that a slowdown of rapid economic growth by itself could easily expose financial weaknesses. The existence of implicit bail-out promises, so much emphasized in the Asian context, was too widespread to be the principal explanation. Whether East Asia would recover as quickly as Mexico had done in 1995 was uncertain. Vinod Aggarwal (University of California, Berkeley) pointed out that the United States’s role had had very important signalling effects in terms of debt renegotiation, which had helped Mexico to regain access to private capital markets more quickly.

Two papers focused on the role of credit-market imperfections in causing instability. First, Philippe Aghion (University College London, EBRD and CEPR) presented ‘Capital Markets and the Instability of Open Economies’, written jointly with Philippe Bacchetta and Abhijit Banerjee. The authors constructed a model of a small open economy in which a firm’s creditworthiness depends on its current cash flow. With credit-constrained firms, a rise in the price of non-traded inputs (such as land) squeezes profits and brings the boom to an end; conversely, a price fall helps recovery. The authors argued that premature liberalization could set in train a repeated cycle of this nature. They also suggested that a policy of never bailing out insolvent banks (or of closing down a large number of banks) might be counterproductive, because it would run the risk of inhibiting firms from borrowing and, hence, of prolonging the slump. By contrast, FDI during a slump could help limit output volatility.

Noting that, in emerging economies, land’s role as collateral for loans was more relevant than the cash flow costs of paying rent, Gianluca Femminis (Università Cattolica del Sacro Cuore, Milano, and CEPR) observed that land prices in East Asia had started falling some time before the crisis, and that this was contrary to the sequence suggested by the authors. David Vines noted that unlimited foreign investment might not stabilize the economy, if the markets for non-traded goods were not allowed to clear. Manmohan Kumar (Credit Suisse First Boston) wondered about the timing: why not allow FDI before the onset of the crisis? Giancarlo Corsetti contrasted the authors’ approach with Krugman’s model in which the moral hazard problems arose in the banking sector. In response, Aghion warned that a no bail-out policy could induce banks to hide their bad loans by refinancing non-performing customers.

The second paper on credit-market imperfections was ‘Asset Bubbles, Domino Effects and Lifeboats’, presented by Marcus Miller (University of Warwick and CEPR) and co-authored by Hali Edison and Pongsak Luangaram. Following the ‘credit cycles’ approach developed by Kiyotaki and Moore, the authors argued that rising land prices added to a boom by enhancing the collateral that firms can pledge against loans. It is a fall in land prices that leads to a slump, with negative shocks being greatly magnified as credit relations collapse. The authors suggested that these credit-market effects had played an important role in East Asia, especially in Thailand. They gave a brief account of the policy interventions proposed in the literature – such as temporary financing, lifeboats (i.e. mergers with healthy banks) and nationalization – before analysing the measures adopted by the Thai government to prevent financial collapse.

Jagjit Chadha (Bank of England) suggested that the transversality condition for the existence of an asset bubble be examined, and that its bursting be endogenized. He also suggested incorporating bank behaviour. The paper had suggested a financial stabilization policy whereby finance houses would provide temporary financing to property companies. Manmohan Kumar, however, stressed that changes in market sentiment can have powerful effects, as happened when the operation of finance houses in Thailand was suspended. Had these finance houses continued to operate, the contingent government liabilities would have been even greater, and stabilization of market sentiment could have been rendered more difficult.

In ‘Financial Crises and Financial Intermediation with Reference to the Asian Crisis’, Jorge A Chan-Lau and Zhaohui Chen (both IMF) presented a theoretical model of emerging markets with imperfect financial systems. The model was designed to capture recent events in Asia, where a period of relatively low capital flows was followed by a fast build-up of capital inflow, and ended in crisis with a large capital outflow. The authors suggested that, in circumstances of rapid and sustained capital inflows, policy should try to ensure that domestic fundamentals were sound, while credit-market interest rates needed to be monitored (and influenced). Kenneth Kletzer (University of California, Santa Cruz) noted that the results of the paper depended crucially on the assumed structure of monitoring cost. He suggested that the model be extended by using imperfect information to derive the cost structure of the intermediary. Kletzer also considered that endogenizing interest rates would be useful – for example, in studying the effects of financial liberalization. Chen replied that the reason for assuming fixed prices had been that interest rates in East Asian economies were typically either controlled by the government or fixed by monopolistic banks. John Driffill (University of Southampton and CEPR) thought it would be interesting to aggregate the micro model for macro analysis.

Although multiple equilibrium models of speculative attacks have gained acceptance among many commentators, in their paper, ‘A Theory of the Onset of Currency Attacks’, Hyun Song Shin (Nuffield College, Oxford) and Stephen Morris (University of Pennsylvania) criticized this approach for leaving unexplained the shift in beliefs which moves the system from one equilibrium to another. The authors presented a model in which, given full knowledge, there may be multiple equilibria. Consequently, for any given set of fundamentals, there are two equilibria – either crisis or no-crisis – around which speculators might coordinate. They then demonstrated how, if each speculator has a small amount of uncertainty regarding others’ information sets, the possibility of such multiple equilibria can be removed by virtue of the fundamentals acting as a coordinating device. They also concluded that, in contrast with other second-generation models of speculative attacks, the exact timing of the shift in expectations could be found ‘inside’ the model. Jonathan Thomas (University of Warwick) suggested that the authors examine the robustness of the unique equilibrium, for example, when the scale of depreciation is dependent upon the magnitude of the attacks. He doubted the relevance of the uniqueness result in explaining the crisis in East Asia, since this was perceived to have come as a surprise. Kevin Chang (University of Southern California and CSFB) suggested incorporating contagion effects.

In ‘Korean Financial Crisis: A Price for a Miracle’, In-Ho Lee (University of Southampton) argued that, although the Asian financial crisis involved several countries, a unique background existed for each country. He chronicled the Korean economy’s moves from a ‘miraculous’ trajectory of fast growth to financial-market turmoil in the real-sector disruption, massive lay-offs and wholesale economic restructuring. Lee’s explanation of the Korean crisis relied on self-fulfilling expectations and attributed a key role to short-term debt. The author noted that, during Korea’s period of rapid economic growth, the similar foreign-account problems had arisen in the early 1980s, but without generating the crisis that had followed in 1997. The difference lay in the proportion of short-term debt, which was higher in 1997, causing reluctance among foreign lenders to roll over short-term debt and leading, in turn, to further depreciation of the Korean won. (On the reasons for firms’ greater reliance on short-term debt, Lee noted that the allocation of funds was mostly decided by political, rather than economic, considerations.) One policy implication of Lee’s ‘self-fulfilling’ approach was that alarm signals and the release of economic data played an important role during a crisis.

Olli Castren (Bank of Finland) remarked that the short-term Korea-US interest rate differential had been quite stable up until 1997, despite the Korean economy’s large borrowing needs. It was not surprising, therefore, that Korean firms and banks had found it attractive to borrow abroad. Richard Portes noted that the short-term debt position of Korean firms had deteriorated rapidly, suggesting that the devaluation had been unexpected. Lee’s opinion was that the possibility of devaluation had been considered, but that Korean banks had expected the government to guarantee their loans. Philippe Aghion argued that Korea’s boom-and-bust cycle was typical of the growth process: during booms, bank competition increases and banks try to pre-empt one another; this leads to less monitoring and thence to a deterioration of the loan portfolio.

In an empirical paper on ‘Predicting Emerging Market Currency Crises’, William Perraudin (Birkbeck College, London, and CEPR) and his co-authors, Manmohan Kumar (CSFB) and Uma Moorthy (Birkbeck College, London), looked at the value of lagged information on financial and macroeconomic variables in the forecasting of crises. The authors levelled two criticisms against the utility for macroeconomic policy-makers, private investors and financial regulators of the existing empirical literature. First, these studies use contemporaneous variables in mainly descriptive frameworks. Second, in the few studies where forecasts exist, these are generally ‘in sample’. The authors set out to remedy both these drawbacks by using a large dataset for 32 emerging countries, comprising over 40 variables at monthly frequencies from 1985 to 1998. They concluded that, with the use of lagged monthly data, simple logit models possessed significant explanatory power. In particular, major crisis episodes were correctly forecast, and out-of-sample trading strategies yielded profits.

Giancarlo Corsetti expressed surprise that, given the authors’ aim of analysing trading strategies, they focused only on depreciations of the exchange rate and ignored periods of appreciation. Analysis of both periods would reduce the profitability of the reported trading strategy. Paul Mizen (University of Nottingham, Bank of England and CEPR) doubted whether a country’s vulnerability to a crisis could really be quantified. Mizen was also unsure whether the reported information about crises could be timely enough for policy-makers to be able to react. Richard Portes noted that the authors’ definition of a crisis (which did not include foreign-exchange reserves, for example) left out many relevant events, and that they also ignored successful defences of the currency. Another limitation was that their data overlooked the importance of forward positions. Kumar replied that the authors were seeking to determine whether, in spite of the forward positions, the available data could still be used to predict crises.

Alessandro Missale (IGIER, Università Degli Studi di Brescia, and Bank of England) presented ‘High Public Debt and Currency Crises: Fundamentals versus Signalling Effects’, written with Benigno Pierpaolo. In a complex analysis, the authors sought to determine how public debt, policy-makers’ credibility and external circumstances all affected the probability of exchange-rate devaluations. Their fully specified model enabled them inter alia to measure the credibility of the foreign-exchange regime following a devaluation, and to assess the probabilities of both imminent and long-run devaluations (using the short-term interest rate and the forward rate respectively). Key features of their model were the ‘debt burden’ and ‘signalling’ effects.

Berthold Herrendorf (University of Warwick and CEPR) thought that several of the authors’ main results could well hold in a more parsimonious version of their model. As the paper offered many precise predictions, it would be worthwhile to test them with data. In this connection, Herrendorf also questioned the robustness of the results with respect to a change in the time horizon, and particularly with an infinite horizon (Missale and Pierpaolo’s model had three periods). Marcus Miller noted that, whereas the authors’ model had been developed with the Italian experience in mind, its analysis of the interest rate as a signalling device might prove useful also in relation to the Asian crisis. Manmohan Kumar added that the model yielded interesting propositions about the currency-composition issue (i.e. local- versus foreign-currency denominated debt). Kevin Chang thought the model could also be quite useful with regard to the convergence criteria in the European Monetary Union.

Jim Rollo (Foreign and Commonwealth Office) chaired a roundtable discussion of the policy implications of financial crises. Richard Portes opened the discussion by addressing the issues of early-warning indicators and lender-of-last-resort functions. He argued that, via its rescue packages, the IMF was creating a considerable moral-hazard problem and that it had played the role of international lender-of-last-resort without having the necessary supporting structures. He therefore favoured a market-based solution. Portes also expressed strong doubts about the validity and usefulness of the empirical literature on early-warning indicators. Philip Turner (Bank for International Settlements) spoke about risk in financial markets and the public sector’s role in that context. Although the recent crises had led to more quantitative analyses of risk, such studies still needed improving. If countries wanted to benefit from capital flows, it was necessary to price risk correctly.

The conference ended with a presentation by Joshua Aizenman (Dartmouth College) of his paper on ‘Capital Mobility and Crises in a Second Best World’. The aim of the paper was to study the welfare effects of financial integration in the presence of moral hazard. In Aizenman’s model, entrepreneurs faced a trade-off between risk and return, while banks could mitigate the resultant excessive risk by costly monitoring. The author showed that a drop in banks’ cost of funds increased the risk they tolerated. (Likewise, a less efficient intermediation technology, higher macroeconomic volatility, and a more generous deposit insurance scheme all raised the riskiness of projects.) A key result was that, with relative scarcity of funds, financial integration was welfare-reducing if financial intermediation was relatively inefficient. This led to an important policy recommendation: if a country started with a highly inefficient banking system, then a precondition for beneficial financial integration was reform and improvement of the banking system.

John Driffill was puzzled by the connection made with Bhagwati’s ‘immiserizing growth’, given that the Asian economies were still growing incredibly fast, even if there was too much investment and it was too risky. Manmohan Kumar, while agreeing with Aizenman’s main point, asked whether the introduction of political considerations into the process of financial liberalization could affect the analysis. Aizenman responded by saying that his paper questioned, in an important way, the orthodox view that the opening of financial markets was a good thing. A key practical policy implication was that, instead of opening financial markets, linking the inflow of capital to proper capital-adequacy ratios that were contingent on the sophistication of the domestic financial system might be considered. Thus his analysis was especially relevant for the international financial institutions and the policies they advocated.