Pretty new!! DefaultoMeter@ updates below
DefaultoMeter@
I am interested in default risk. I have therefore tried to develop a model that can predict bankruptcies or general financial distress trigged by credit events. I call my model DefaultoMeter@. A nice feature of this approach is that it can be applied in a reliable way to portfolios of firms (or industrial sectors). DefaultoMeter@ is particularly well suited for small companies outside the US that are not covered by the large rating agencies or for banks and other financial industries with a complicated capital structure.
The default probabilities produced by DefaultoMeter@ are constructed using an econometric model relying on stock market data, accounting data and if available bond market data. The method is a hybrid method based on a combination of Merton's (1974) structural model, GARCH volatility forecasting models and credit spreads term structures weighted in an optimal way (neural networks?). Extreme Value Theory is used in order to avoid having to rely on the Normal distribution. A description of the most basic ideas behind the technique can be found in my working papers (some can be found above and the others can be sent upon request):
"The Market's View on the Probability of Banking Sector Failure: Cross-Country Comparisons", and
"A simple continuous measure of credit risk" (with OhKang Kwon)
"Merton for Dummies: A Flexible Way of Modelling Default Risk.
"Default Risk, Systematic Risk and Thai Firms Before, During and After the Crisis" (with Srisuda Chongsithipol and Lugkana Worasinchai)
Using a simplified version of DefaultoMeter@ (implemented on an Excel spread sheet) that I call DefaultoMicroMeter@ I have produced some default curves for the four major Swedish banks together with a major Swedish insurer and on three Industrial companies listed on the Stockholm stock exchange. The numbers on the x-axis are probabilities (in %) that a company will default within the next 12 months. A probability of default between 5% and 10% must be considered high. If the probability is above 10% it must be considered very high. The probabilities can be compared to, for instance, Moodys ratings and their associated default probabilities. A typical speculative grade bond (Ba and below, typically considered risky) has a default probability of around 4%. An investment grade bond (Baa and above) has a default probability around 0.1%.
Lately the major Swedish banks have seen a drastic reduction in their default risk. At the moment SEB seems to be the most risky banks (around 0.001%) and Föreningssparbanken and Handelsbanken the least risky. These risks are all very small, however, and the last couple of months all banks have become safer according to the markets. Skandia, the insurer, was riskier than all the banks for a long while during the second half of 2002 (around 10%). Lately, however, it has also become very safe according to the market. It is considerd more risky than the banks though.
Obviously, these are only estimates! "Tolka sannolikheterna med en nypa salt".
Due to data problems I have not updated the graph below this time.
According to the model ABB has been fairly close to some kind of default for quite a long time now. Ericsson has become more risky the last year and is now slightly more risky thanNokia. Default probabilities for these industrial companies are perhaps ten times higher than those for the banks (in summer). Part of this might be caused by some kind of implicit bank guarantee incorporated in market prices (government bail out, too-big-to-fail). The last two months (May-June) all three companies have become less risky according to the market.
Obviously, these are only estimates! "Tolka sannolikheterna med en nypa salt".
I will update this page once a month or so. I was too optimistic before!