The global financial crisis of 2008, along with the ensuing global recession and European sovereign debt crisis, is arguably the worst since the Great Depression of the 1930s, resulting in the highly politicised bailout of banks by national governments, prolonged unemployment and substantial declines in consumer wealth.
The aftermath still continues to unfold in 2015, with a US federal judge ruling that Nomura Holdings and Royal Bank of Scotland knowingly sold defective mortgage bonds containing misrepresentations and errors, which contributed to the US Federal National Mortgage Association (so-called Fannie Mae) and the US Federal Home Loan Mortgage Corporation (Freddie Mac) being rescued – by taxpayers – in September 2008. Other banks avoided court and settled to the tune of billions of US dollars.
Thailand is not an exception. Though we were not the origin of the 2008 global financial crisis, our stock market was also adversely affected. Luckily, our problem was not as severe as that in the US and Europe – at least not this time around. We have other exposures, as were painfully revealed in the Asian economic crisis of 1997.
More sophisticated risk management skills (and open ears at the political level) could have gone a long way to mitigating that event. With the now often-used quote of the Stanford economist Paul Romer in mind, “A crisis is a terrible thing to waste”, an arguably positive aspect of its aftermath is the recognition of requiring more realistic models for the risk associated with financial products and portfolios of assets.
However, despite the large academic literature, very few models are suitable for genuine application in modelling and predicting very large sets of assets – which is precisely what is required by large financial institutions, as they have literally thousands of investment positions. This is not for lack of importance: the successful modelling and prediction of comovements of financial asset returns is crucial for successful risk management, and can play a key role in avoiding future financial crises.
Part of the problem stems from the nature of the statistical problem: in such models, there is a massive proliferation of parameters to be calibrated, as the number of assets under study increases.
This is highly challenging, even with modern computing power, and, even given the required computing power, it is a well-known phenomenon in statistical inference that, in general, as the number of parameters to be determined increases, their obtained accuracy decreases. There exist methods to deal with this problem, but they are rather obscure for typical practitioners without extensive quantitative and statistical training.
Another problem is devising a model which not only addresses this estimation problem, but also adequately accounts for the actual features in the data. The major so-called “stylised facts” of asset returns are well-established. The return for a daily traded stock is just today’s price, minus yesterday’s price, divided by yesterday’s price (the so-called relative return) or, when multiplied by 100, the percentage relative return. For individual stock and other asset returns, the features of daily stock returns include “leptokurtosis”, which means that extreme events (such as very large and unexpected moves in stock and other asset prices) occur more frequently than would be expected if the returns were to be described by the ubiquitous “bell curve”, or normal distribution, which is the model for numerous real-life phenomena, such as human height, prices of consumer goods, etc. This idea is often referred to as “fat tails”, because the tails of the distribution are where the extremes occur, and they are “fatter” (riskier) than those exhibited by the normal distribution.
This notion of asset returns not being “normal” (or following the bell curve) is far from just some ivory tower academic observation without practical value for industry. Alan Greenspan, who served as chairman of the US Federal Reserve from 1987 to 2006, stated in 1997: “The biggest problems we now have with the whole evaluation of risk is the fat-tail problem, which is really creating very large conceptual difficulties. Because as we all know, the assumption of normality enables us to drop a huge amount of complexity of our equations very much to the right of the equal sign. Because once you start putting in nonnormality assumptions, which unfortunately is what characterises the real world, then the issues become extremely difficult.”
Another feature of asset return data is asymmetry: a simple inspection of the daily returns on a major stock market index indicates that large negative returns (losses, if you are long in the market) occur more frequently than similar-sized positive gains. A third, and very prominent feature of daily returns not seen as much in monthly, quarterly, or yearly data, is that the volatility (statistically, the standard deviation) is changing markedly over time. Classic models for asset allocation ignore all three stylised facts, and in doing so, the risk associated with any particular position can be vastly underestimated.
These issues are now well known by academics, and numerous effective models have been proposed over the last three decades to deal with them. To some extent, they are also accommodated in industry, in larger financial institutions with quantitatively trained staff (the so-called quants). The recent financial crisis has brought negative publicity to “quants”, blaming them for designing highly complicated products which few people can understand and ultimately leading to crisis.
With regard to stock markets, from these three stylised facts, Thailand is not so different. Our stock returns are also leptokurtic, asymmetric, and also exhibit quite strong changes in volatility over time. The only thing in which we deviate from the Western world (besides us having tastier food) is quants: we have no quants! At least not in the numbers that fill the offices of large investment banks and other financial institutions in the West. Ironically, our stock returns are sensitive to internal politics and rumours. However, the Thai stock market still has much more to explore.
Marc Paolella is a professor of empirical finance at the University of Zurich and Nuttanan Wichitaksorn is a Research Fellow at the Thailand Development Research Institute. This article is part of a project supported by ETH Zrich for research collaboration between Switzerland and Thailand.
First published: ฺBangkok Post, June 10, 2015