What recovery in Asia could look like

Gavyn Davies on the parallels with past banking crises

The key question for the world economy, and global asset prices, today is how fast the crisis economies of Asia can be expected to recover from their recent debacle. All economic crises have their own particular characteristics, and it is unlikely that the pattern of recovery from the Asian crisis will exactly replicate any particular historical precedent. The nature of any Asian recovery will naturally depend on many things - the policy response, the global backdrop for capital flows in general, the interplay with domestic politics, etc. It is fairly clear that there will be significant differences between countries in these respects, just as there have been wide disparities between individual episodes in past crises.

Nevertheless, it is interesting to ask whether there are any broad patterns in past historical episodes which might be instructive for the Asian case. In the accompanying graph and table, the impact of 11 previous banking crises is traced out, six of which occurred in Latin American economies, and five in developed industrial economies. In each case, the onset of the crisis is defined to occur at time zero, and we trace the development of various economic and market aggregates in the years preceding and following the crisis.

It is interesting to note that some patterns do appear to replicate themselves from one crisis to another. In particular, the pattern of real gross domestic product (GDP) growth relative to trend appears to be fairly consistent, both in the Latin American economies and in the industrial economies. Real GDP tends to rise much more rapidly than trend in the run-up to the crisis, but growth typically drops below trend around 12 to 24 months before the crisis starts. Then, in the year of the crisis itself, growth drops sharply below trend and this continues to worsen for at least 12 months after the crisis breaks.

The total decline in GDP relative to trend in the two years surrounding the crisis is phenomenally large. In the Latin American economies, GDP declines by around 12 per cent relative to trend over the two relevant years, while in the industrial economies GDP declines by about 8 per cent relative to trend in these two years, and then declines by a further 3 per cent in the subsequent year. Hence, there is a fairly consistent pattern that GDP relative to trend declines by some 11 to 12 per cent.

In other aggregates, especially those relating to asset prices, there is rather more disparity between individual episodes, so the conclusions are more dubious. In the case of Latin American stock markets (measured in dollar terms, in order to derive figures which are relevant for international investors), there is typically a massive rise in market prices from the fifth year to the second year prior to the onset of the crisis.

Over this period, market indices in dollar terms typically rise four- fold. However, from that point onwards, stock markets begin to decline. In the two years prior to the crisis, they typically lose around one-quarter of their peak value, while in the year following the crisis they approximately halve again. However, from that point onwards, there is a rebound in stock prices, a rebound which cumulates to well over 50 per cent in the third and fourth year following the crisis.

Turning to real exchange rates, there is once again a somewhat more consistent pattern in Latin American cases than in industrial economies. In the former, there is typically a sharp appreciation in the real exchange rate for several years prior to the onset of the crisis. This appreciation peaks at or just before the crisis breaks, from which point onwards the real effective exchange rate begins to decline. Typically, the decline ends around six months after the crisis has hit, with the cumulative drop in the real exchange rate amounting to at least 25 to 40 per cent. There then follows a period of around two years in which the real exchange rate typically stabilises or even rebounds slightly, but on average it never fully re-attains the high levels that helped trigger the crisis.

Finally, turning to total returns on bonds or short-term deposits (again measured in dollars), there is again huge variability between individual historical precedents. In the Latin American cases, total returns dip sharply in the six months following the crisis, but then they typically return to positive territory for the next two years as the currencies stabilise, and investors begin to benefit from high interest rates on local currency deposits.

What does this imply for the recovery pattern in Asia over the next year or two? Goldman Sachs' recently revised real GDP forecasts for the Asian crisis economies show growth rates in calendar years 1997 to 1999 respectively of 4.8 per cent, 0.7 per cent and 3.5 per cent. Trend growth is around 6.5 per cent per annum, so if we cumulate these growth rates, then the forecasts imply that the level of GDP relative to trend will decline by around 10.5 per cent over the relevant three-year period. This "fat U-shaped" recovery is very close to what the precedent of previous crises would suggest is likely.

During the early stages of the Asian crisis, stock markets and currencies have, of course, totally collapsed. So far, the decline in stock markets (measured in dollars) has been around 70 to 80 per cent in the crisis economies, while the depreciation in real exchange rates has been around 40 per cent. In both cases, these declines are broadly what one would have expected, based on past precedent from Latin America.

Since the declines in stock markets and real exchange rates have already been fairly close to the peak-to-trough declines for our Latin American precedents, there may be some case for arguing that further declines in real exchange rates and stock markets from present levels are not particularly likely to be sustained. In fact, there might be some rebound in both aggregates from current levels, though it is difficult to say definitively that this will start soon, in view of the policy adjustments that still have to be made.

Past experience suggests that, if such a rebound does occur, then stock markets in dollar terms may eventually rally much more than real or nominal exchange rates, though it may take far longer for this to occur. As noted above, a 50 per cent increase in the level of stock markets has typically occurred after the trough of the crisis, whereas real effective exchange rates have generally recovered only slightly.

Of course, when one allows for the high domestic interest rates which typically persist for some time following the crisis, then total excess returns from short-dated currency instruments (eg three-month interest rates contracts) are quite often highly positive in the aftermath of the crisis but, by the second year of recovery, they seem generally to be considerably smaller than stock market returns measured in dollars.

Overall, then, past experience suggests the Asian crisis economies will permanently lose at least 10 per cent of their GDP as a result of the recent debacle, that their real exchange rates will never return to pre- crisis, and that their stock markets may take five years to regain lost ground. However, given the extraordinary extent of recent collapses, there should still be some useful money to be made by those brave investors who are willing to look for a partial rebound in asset prices over the next few years.