European Economic Perspectives 21

From Russia with Love

Just as in the 1890s, the question of how to predict the financial collapse of nations is high on the international agenda. New CEPR research explores some lessons from a century ago.

The Asian crisis and its aftermath once again illustrate the perils of international finance. But while the policy implications of the crisis, in particular the future role of multilateral agencies like the IMF, remain far from obvious, a number of private bodies are taking action to minimize their exposure to currency risk. For example, several investment banks, including Credit Suisse First Boston, JP Morgan and Lehman Brothers have developed models that seek to predict currency crashes.

From the point of view of policy, it is essential to understand the logic of this research agenda. The potential emergence of a private standard of risk perception must influence future IMF interventions since if all market participants believe that, say, a low level of reserves signals an impending crisis, they will liquidate their positions, thus provoking the crisis. The multilateral agencies are in turn bound to factor in the effects of market opinions, whatever they think of the actual severity of the situation. Alternatively, they may want to propose a different standard. At the very least, it is vital for them to understand what forces shape market sentiment.

Macroeconomists have surprisingly little to say on this issue, although a growing body of literature addresses the related question of what explains the grades given by rating agencies. From a theoretical perspective, there are only models of ‘perfect foresight’ where everything is correctly forecast, or of ‘bubbles’ where everything is self-fulfilling. From an empirical perspective, there is the problem of disentangling market opinion from the actions of the multilateral agencies. For example, the expectation of bailouts by public bodies may make investors complacent, reducing their awareness of risk and so distorting their incentives to acquire information and process it in original ways.

To study the market mechanism ‘in the wild’, it is worth examining the pre-1914 period when bailouts were not, as a rule, to be expected. In a recent CEPR Discussion Paper, Marc Flandreau focuses on the experience of Crédit Lyonnais, then a standard of prudence and planning.

Lyonnais was one of the largest international banks of the time, in a country that was second only to Britain as the largest global creditor. The bank’s management was obsessive about limiting exposure to market risk while at the same time assuming a leading role in syndicated loans. So it became very concerned to provide finance only to those countries that it judged sound and, given its pre-eminence in European finance, Lyonnais opinion became extremely important in influencing the market mechanism. Each financial crisis brought a new wave of caution and analysis, to a point in the 1890s where its research department - 100 members strong at the turn of the century - produced a sovereign debt crisis indicator.

The development of the Lyonnais indicator was a direct consequence of the collapse of Barings in 1890 as a result of excessive exposure to Argentine debt. As the crisis spread through contagion reminiscent of today’s turmoil, Lyonnais began to construct a precise measure of debt sustainability. Its database sought to distinguish unproductive from productive borrowing, as well as measuring the efficiency of past borrowing. This led the bank to estimate the value of government assets - which could be very large since a number of governments, notably in continental Europe, were involved in infrastructure development and heavy industry such as railroads and mines - in order to discount this item from public indebtedness.

The estimates was then applied to past data, and the Lyonnais economists found that problems had only ever occurred in countries where the ratio of interest service (net of revenues from state-owned infrastructure) to total government revenue exceeded 40%. The ratio in Argentina, for example, had exceeded 50% when it collapsed in 1890.

This measure was then combined with other indices to serve as the basis of a formal ‘rating’. There were three categories: first, ‘sound countries’; second, intermediate countries, a group that included Britain because of its high debt though it was treated as a ‘blue chip’ country; and third, bankrupt countries, such as Brazil and Argentina.

Interestingly, the measure was responsible for an indicator that designated Russia as ‘safe’. Because most loans to the Russian government were used for industrial investments and because those investments provided a revenue, Russia’s ‘indebtedness’ remained moderate in the eyes of the Lyonnais economists. This in turn led the French to invest massively into Russia with what, of course, turned out to be disastrous consequences.

Several lessons can be drawn. First and foremost, this episode indicates how late nineteenth century investors reacted to the absence of official data, formal ratings and international bailouts. A second lesson is the considerable importance of investors’ perceptions in shaping the market mechanism itself. In particular, the markets’ preference for investment finance - as opposed to consumption finance - proved decisive in comparing sovereign risks.

Third, and somewhat discouragingly, the experience underlines the problems of basing predictions on past history. Regardless of Lyonnais’ caution in trying to avoid a recurrence of the Argentine problem, it entirely missed a different kind of problem. What went wrong with Russia was not the result of unproductive use of foreign resources. Instead, it came when a new regime, aiming for autarky, decided to renege on its foreign commitments, while retaining the plants, mines and railroads for which foreign capital had paid.

The last lesson concerns the economics of financial opinion. The Lyonnais view of the world was not an individual voice in a broad market of opinions: it was the voice of an international institution. So while the bank’s desire to retain credibility certainly encouraged it to do its homework, this probably led other banks to pay far less attention to risk. And the absence of other serious competing opinion - at least in France - probably induced other banks to join Lyonnais-led syndicates without due caution. But finance is a game where you make money not because you are right but because your view becomes the market view: the externalities of being a market leader cannot be ignored.

These conclusions may help to shed some light on current developments and suggest a few policy implications. First, the aggressive move that today’s investment banks have taken towards research on emerging market risk seems to provide evidence that contrary to a popular view, bailouts are not systematically expected by the market - otherwise, why should they pay so much attention? In a sense, Credit Suisse First Boston, Lehman Brothers and JP Morgan are just doing what Crédit Lyonnais did one century ago - and for the same reasons.

Second, yesterday as today, there is the striking influence of past collapses on the design of current indicators. The Lyonnais method was a formidable tool to track Argentine-like crises but was totally powerless in alerting investors to the risks of Russian default. This problem is still a serious challenge for today’s ‘data miners’, whose predictors are based on past data.

Third, the Lyonnais tale should warn multilateral agencies against the dangers of relying on market sentiment as represented by private indicators when designing their policies. Because opinion is a public good as much as anything else, private endeavours to define a standard of risk are open to manipulations and should thus be taken with a grain of salt.

This article reviews research reported in ‘Caveat Emptor: Coping with Country Risk without the Multilaterals’ by Marc Flandreau, CEPR Discussion Paper No. 2004 (October 1998). Flandreau is at OFCE and CNRS in Paris, and a Research Affiliate in CEPR’s International Macroeconomics programme.

Download PDF

Return to contents

 Browse Archives