Contagion is one of the more elusive concepts in the current debate about the financial crisis. Indeed, the logic behind it is often unclear.
Take, for example, the suggestion of pricing externalities across markets following a default on Greek sovereign debt. It implies that the financing costs of other “vulnerable” sovereign borrowers may also increase. Yet if we consider the insolvency of one private household, it doesn’t follow that his neighbour is denied credit as well. Why, then, should this be the case for countries with very different economic fundamentals and access to a wide range of investors?
Sometimes, theoretical arguments based on coordination failure in a multi-equilibrium setting are invoked to provide at least some logical backup. But these platonic constructs leave all the important questions unanswered, namely if and how investors or banks coordinate in their withdrawal of credit and how far such credit withdrawal goes. This loose “contagion” concept resembles a magic spell invoked chiefly by those who try to scare politicians into ever increasing financial guarantees without apparent policy alternatives.
A new research agenda
The best financial research has realised these shortcomings in the vast literature on contagion and called for a research agenda that provides a clear identification of contagion channels accessible to empirical testing (see Forbes and Rigobon 2002). The 2007/2008 financial crisis provides a new case study that can improve our understanding of various contagion channels.
Our own work (Chua et al. 2011) focuses on the role of mutual funds as a source of contagions from financial to non-financial stocks. We hypothesise that funds that experienced large value losses due to holdings in bank stocks faced large asset redemptions from their retail investors and therefore were forced to sell off other non-financial stocks in their portfolio. Such asset fire sales by distressed funds could transmit the crisis from bank stocks to non-financial stocks.1
Disaggregate fund holding data
To obtain a clear identification of the fund-based contagion channel, we use a new comprehensive sample of the stock positions of 20,477 equity funds around the world. For each fund, we calculate fund exposure to financial stocks as the losses induced by financial sector positions in the initial phase of the financial crisis. For example, a fund holding 20% of its value in financial stocks with average value losses of 40% over the second semester of 2007 and the first semester of 2008 has a fund exposure of -8%. Such high losses are likely to trigger investor outflows from the fund, requiring fire sales of other non-financial stocks owned by the same fund. To capture this selling pressure on non-financial stocks, we define stock exposure as the average fund exposure of all funds owning the stock weighted by the fund ownership shares in the stock. Thus, non-financial stocks held mostly by funds with large fund exposure are considered highly exposed stocks. The contagion effect (through distressed equity funds) should be concentrated in these exposed stocks.
Our empirical analysis focuses on the relative return performance of the 15% non-financial stocks worldwide with the highest stocks exposure. Exposed stocks are found particularly in the US stock market where they represent 27% of all US stocks and cover all industries. This allows us to control for industry specific asset sensitivities to the crisis using industry fixed effects. The contagion effect is measured as the (cumulative) stock return difference between exposed and non-exposed stocks in the same industry and country.
Surprisingly strong contagion effects
Exposed stocks are indeed those that suffer the largest stock price drawdown during the financial crisis. The stock price for the 15% most exposed stocks worldwide underperformed relative to non-exposed peers in the same country and industry by 17% at the peak of the stock market downturn. The corresponding figure for the 27% most exposed US stocks, relative to their US non-exposed industry peers is even more startling at 36% (as of 27 February 2009).
Figure 1 shows the cumulative underperformance of exposed stocks worldwide (left) and in the US (right) relative to stocks in the same country and industry and after accounting for risk premia from a model with four local and four international risk factors. The vertical bars provide robust standard errors (±1 SE) around the point estimate of the average cumulative underperformance.
Figure 1. Fire sales effects for exposed stocks
We find that an incremental return shortfall of 36% for the 27% most exposed US stocks implies an aggregate 10% value decline for an equally weighted US stock index. Considering the fact that exposed sample stocks are on average larger than non-exposed stocks, the contribution of this effect to the decline of the overall US stocks index (which would be value weighted) is likely to be at least as large. Therefore, our analysis suggests that at least some 10% out of the 52% crisis-related decline in the US stock market can be attributed to distress selling by mutual funds.
Additional insights on contagion
Our research also uncovers two additional insights about the 2008 stock market crash. First, the fire sale discount is most pronounced for stocks which performed best during the crisis. Figure 2 illustrates this aspect by depicting the relative performance of non-exposed and exposed stocks as a function of the stocks overall return from July 2007 to July 2008. Most of the return shortfall of exposed stocks is concentrated among the stocks with the highest returns. This somewhat counterintuitive result can be explained by fund discretion about which asset positions to liquidate. Faced with funding constraints and investor redemption requests, distressed equity funds liquidated the best performing stocks rather than stocks with recent large capital losses. Thus, fire sales were more pronounced for stocks among the 10% best performing stocks. For these stocks we find average fire sale discounts above 75%.
Figure 2. Fire sales effects, exposed and non-exposed and by returns
Note: The graph on the left shows the relative performance of exposed and non-exposed US stocks by stock return quantiles, controlling for industry fixed-effects. The y-axis denotes the cumulative (weekly) returns from 29 June 2007 to 27 February 2009, adjusting for risk premia from a model with four local and four international risk factors. The x-axis denotes the quantiles of the cumulative stock returns. The right graph plots the performance difference between the exposed and non-exposed US stocks. The robust standard errors (±1 SE) around the point estimate of the average cumulative underperformance are also plotted.
A second insight about contagions concerns the role of fund ownership itself. While ownership by distressed funds adversely affected the performance of a stock during the crisis, the opposite holds for overall fund ownership. Stocks in the top 15% quantile of the highest fund ownership share suffered considerably lower capital depreciation than otherwise similar stocks. This suggests that investors who delegate investment decisions might have a lower propensity for equity sales or “flight to quality” than direct investors. The implication is that during bad times (i.e. when the overall index is strongly declining), stocks mostly held by funds generally experience less selling pressure than those primarily held directly. The exceptions here are stocks held mostly by exposed funds and in industries with good crisis performance as argued above. Retail investors and their behaviour therefore deserve particular attention in studies of contagion.
What does this mean for the European bond market?
Overall, we find strong evidence that the fund ownership structure at the outset of the crisis in June 2007 had surprisingly large effects on the crisis performance of individual stocks and stock groups. This suggests that common ownership of different assets can be a very potent vector for transmitting price declines in one asset to another asset, over and above the actual deterioration of the fundamental valuation of the assets. Our results therefore provide valuable insights for analysing the current fear of contagion in Europe's sovereign bond market.2
Europe's asset management industry oversees approximately €5.9 trillion in bonds investments, of which €1.2 trillion is held by bond funds. Could a sovereign default of Greece or Portugal trigger large fund redemption calls by retail investors and consecutive fire sales of other sovereign bonds? The following observations are relevant in this context:
- Greece and Portugal account for only 4.7% and 2.1% of all Eurozone sovereign debt, respectively. By contrast, financial stocks accounted for some 15% of US stock capitalisation in 2007. Moreover, the effect of sovereign default on bond funds could be mitigated by guarantees about a partial repayment in a debt restructuring plan. In the 2007/2008 banking crisis, equity funds did not benefit from a comparable lower cap on value losses in bank stocks. Both aspects suggest a much weaker contagion effects in the bond market.
- Asset management mandates from institutional investors like insurance companies and pension funds account for 50% more bond positions than investment funds, which are primarily held by retail investors (European Fund and Asset Management Association 2010). These institutionally-held assets are less likely to suffer from severe redemption pressures in the event of a (partial) default.
- Managing expectations about the maximal bond holder losses is important in limiting capital withdrawals from fixed income funds and therefore the price externalities on other sovereign bonds. Any disorganised default (without a debt restructuring plan) should be avoided.
- Large discounts on Greek or Portuguese debt show that the market already anticipates a possible default. Such anticipation will moderate the additional value loss at the moment of default. Artificial bond price support by the European central bank is therefore particularly problematic as it inflates the value shock in case of default.
- Evidence from the stock market shows that more liquid stocks suffer fewer price contagions. More information about global bond ownership can help speculators to identify and arbitrage fire sale by distressed bond funds, increase asset liquidity and thus limit the corresponding liquidity externalities. Speculative trading of bonds should be encouraged rather than discouraged as it helps to share sovereign default risk among a wider group of investors.
Overall, the case for a strong and persistent contagion effect via fire sales by bond funds is rather weak as far as the default by small countries like Greece and Portugal is concerned. The massive bond market intervention by the European central bank is therefore hard to justify. Rather than trying to prevent sovereign default at all costs, a better policy consists of preparing the recapitalisation of those banks most affected by it. Without such contingency plans, Europe's politicians risk being taken hostage by the fear of asset contagion.
Chua, C.T., H. Hau, and S. Lai, (2011), “The Role of Equity Funds in the Financial Crisis Propagation”
European Fund and Asset Management Association (2010), Asset Management in Europe, EFAMA's Third Annual Review, April, page 3.
Forbes, K and R Rigobon (2002), “No Contagion, Only Interdependence: Measuring Stock Market Comovements”, Journal of Finance, 57(5): 2223-2261.
Pulvino, T (1998), “Do asset fire sales exist? An empirical investigation of commercial aircraft transactions”, Journal of Finance, 53:939-978.
1 See also Pulvino (1998) for related evidence that fire sales by distressed firms (airlines) also produce lower asset values (for used airplanes).
2 Europe's sovereign debt crisis poses additional problems of bank solvency. In the 2007/2008 crisis, bank solvency issues have been addressed at the national level and the policy tools to do so are still in place. Contagions through bank default are a separate issue not addressed here.