The recent sovereign risk crisis in developed countries has reminded equity investors around the globe that, even when not directly holding government bonds, they may be heavily affected by a worsening of sovereign credit risk conditions. For example, it became clear that stocks of companies which benefit from implicit guarantees provided by the government, such as banks, may be affected by news on sovereign risk conditions. Likewise, stocks of companies that benefit from public spending or from tax incentives may suffer in times when public finances worsen. The academic literature actually provides ample evidence that stock returns are indeed sensitive to sovereign risk (see e.g. Belo et al. (2011), Gandhi and Lustig (2011) Cutler (1988), Ang and Longstaff (2011), Longstaff et al. (2011), Jeanneret (2010) and Hume and Kim (2008)).
An investor might be interested in avoiding such sovereign risk exposure in his equity portfolio for a variety of reasons. For example, a public pension fund has a motivation to avoid sovereign risk exposure in its equity portfolio as the contributions to the fund depend on government funding and beneficiaries’ incomes are also sensitive to public finances. Moreover, an investor who is exposed to sovereign credit risk through government bonds may wish to avoid taking too much of the same risk exposure when investing in an equity portfolio.
In this article, we discuss results from recent EDHEC-Risk Institute research on whether it is possible to reliably categorise stocks by their exposure to sovereign risk in order to create equity portfolios with low sovereign risk exposure1. We use robust estimation techniques to measure stock return sensitivities to changes in sovereign credit default swap spreads. Our main finding is that such a measurement of sovereign risk exposure of stocks is reliable out-of-sample: in bad times where negative news occurs on sovereign risk conditions, our low sovereign beta portfolios indeed outperform high sovereign beta portfolios. Our approach thus provides a way of identifying which stock portfolios will allow an investor who is already exposed to sovereign risk to avoid loading up on exposure to the same risk factor in his equity investments.
The remainder of the article first explains why and how we proxy for sovereign risk, then provides details on how we measure exposure to this factor and finally presents out-of-sample performance and characteristics of the portfolios with different levels of sovereign risk beta.
Measuring sovereign risk: a market-based proxy
In order to measure exposure to sovereign risk conditions, one needs to define a relevant proxy for sovereign risk. Natural candidates one may consider include Credit Rating Agency (CRA) ratings, sovereign bond spreads, or CDS spreads. An important consideration is that CRA ratings use accounting and fundamental data and hence are backward-looking, static, and contain lagged information. In fact, it has been documented that innovation in sovereign and corporate CDS spreads leads changes in CRA ratings (see e.g. Hull and White, 2004, Blanco et al., 2005).
While there is no clear consensus in the academic literature on which measure best represents sovereign default risk, the fact that CDSs reflect current market expectations on the strength of the creditworthiness of sovereign economies is important for our empirical exercise as it will help us better understand the cross-sectional differences in individual stock exposure to sovereign risk under contemporary market conditions. For this reason, we utilise CDSs as a sovereign default risk measure, although further research is warranted to investigate the effect of other credit measures (e.g. bond yield spreads) on stock returns. Moreover, our use of CDS spreads means that we do not need to select a risk-free reference, which would be required for the determination of bond spreads.
We look at changes in CDS spreads rather than levels (following Ang et al. (2006), Campbell (1996) and Petkova (2006)) because we want to analyse the effect of “news” about sovereign risk on stock returns. We define a global sovereign risk factor based on the percentage change in the 5-year sovereign CDS spreads of more than 40 countries in our dataset. The risk factor is appropriately signed i.e. a positive value represents good news on the sovereign credit risk front (which corresponds to a reduction in spreads).
Figure 1 plots the World CDS factor from December 2000 to December 2011. The figure clearly shows that, in the second half of the analysis, which witnessed the financial crisis and European sovereign debt crisis, the changes in CDS spreads were quite volatile. Weekly values as extreme as +27.36% and -60.42% can be observed in recent times.Figure 1: World CDS factor returns
The figure plots the weekly World CDS factor from 22nd December 2000 to 16th December 2011.
Constructing low sovereign risk equity portfolios
Global equity portfolios are constructed using all large and mid-cap stocks from developed and emerging economies. An important consideration in our analysis is to obtain robustness. In fact, it has been shown that estimating exposures of individual stocks to risk factors is prone to estimation error and may end up providing unreliable classifications of stocks by their risk exposure (see Ang et al., 2012). We improve the robustness of our approach by mixing data frequencies and by using Bayesian estimation techniques. More precisely, using both daily and weekly frequency2, we regress stock returns on the World CDS factor while controlling for the market factor.3Using a standard OLS regression of stock returns on risk factor could result in substantial estimation error. A more suitable approach is to use a Bayesian estimation approach (Vasicek, 1973) which optimally combines the stock-level sovereign beta with the sector-level sovereign beta in order to find the best compromise between exploiting stock-level information and avoiding over-fitting the data.
We look at the differences between low and high sovereign risk stocks at the portfolio level. We form CDS beta sorted decile portfolios with a control for market beta. We focus on the properties of the first decile (low sovereign beta) portfolio and the tenth decile (high sovereign beta) portfolio. We report the results for only equal weighted portfolios; cap-weighing results in qualitatively similar results.
First, we perform risk-return analysis conditional on sovereign credit risk level by dividing realised portfolio weekly returns over the entire period into five quintiles sorted by World CDS factor. The first quintile is termed “Good News Weeks” and is characterised by high values of the World CDS factor (see the factor construction method). Table 2 shows that the high sovereign beta portfolio out-performs the low sovereign beta portfolio in the “Good News Weeks” regime by 96 bps weekly. In the period of “Bad News Weeks”, the low sovereign beta portfolio out-performs the high sovereign beta by 94 bps. All the statistics are statistically significant.Table 2: World High/Low Sovereign Beta Portfolio returns during good/bad news weeks
The table shows the differences between the mean weekly returns of the High Sovereign Beta and Low Sovereign Beta portfolios and their statistical significance over two periods characterised by good news and bad news weeks respectively. All statistics are based on weekly returns from 20th December 2002 to 16th December 2011.
Table 3 summarises risk and return statistics of the two decile portfolios and compares them with the benchmark (equal-weighted MSCI ACWI). Over the sample period, the returns of both high and low sovereign beta portfolios are not statistically significantly different from MSCI ACWI returns. Also, the low sovereign beta portfolio has a higher Sharpe Ratio than the market index. The low sovereign beta portfolio however displays lower skewness and higher kurtosis than the high sovereign beta portfolio showing that the avoidance of exposure to a specific risk factor did not reduce extreme risks in the distribution of returns.Table 3: Risk and Return Statistics for the World Portfolios sorted on World CDS factor
The table shows basic performance statistics of high and low sovereign beta portfolios, and MSCI ACWI. All statistics are based on weekly returns from 20th December 2002 to 16th December 2011.
Lastly, we test if the decile portfolios have hidden biases to risk factors other than sovereign credit risk, and if our results from the conditional analysis could be attributed to any of those biases? We assess, ex-post, some standard characteristics of decile portfolios that account for factor tilts. These are market cap share, average trading volume per share, M/B ratio and dividend yield. Average values of these measures across 36 quarters are reported in table 4 below. The results confirm that controlling for sovereign risk does not lead to any size, liquidity, or value/growth bias.
Table 4: Characteristics of World Decile Portfolios
The table shows the stock characteristics of decile portfolios of stocks sorted on sovereign beta (while controlling for market beta). All statistics are average values across 36 quarters and are based on beginning of the quarter values from 20th December 2002 to 16th September 2011.
The recent sovereign crisis has made clear that sovereign credit risk conditions have a profound impact on equities. Investors can avoid such risk exposure through a simple stock selection approach that picks low sovereign beta stocks. Our results suggest that a stock’s exposure to sovereign risk can be measured reliably by using a suitable proxy for sovereign risk exposure combined with a robust estimation method. We find that portfolios constructed ex-ante with stocks sorted on their past sovereign betas exhibit strong differences in sovereign credit risk hedging properties going forward. During periods of sovereign stress, a low sovereign beta portfolio exhibits better performance than the corresponding market index. At the same time, it does not lead to significantly lower returns overall for the sample period that we studied. In addition, no biases in terms of stock characteristics are found in the sovereign beta sorted portfolios.
- 1See Goltz F., A. Lodh and F. Rachidy, “Managing Sovereign Credit Risk Exposure in a Global Equity Portfolio,” Bankers, Markets & Investors, May-June 2013.
- 2We average the betas obtained at different frequencies as this has been argued to improve robustness (see e.g. Baesel, 1974, Daves et al., 2000).
- 3Our complete stock universe is a global large and mid-cap universe of more than 40 countries and contains more than 2,500 stocks. Returns are obtained from Datastream for the period December 2000 to December 2011.
- Ang, A., M. Brière and O. Signori, 2012. "Inflation and Individual Equities," NBER Working Papers 17798, National Bureau of Economic Research, Inc.
- Ang, A. and F. Longstaff. (April 2011). Systemic Sovereign Credit Risk: Lessons from the U.S. and Europe.
- Ang, A., R. J. Hodrick, Y. Xing and X. Zhang (2006): “The Cross-Section of Volatility and Expected Returns”. Journal of Finance, Vol. 61, No. 1, 259-299.
- Baesel, J. B., “On the Assessment of Risk: Some Further Considerations,” Journal of Finance, Vol. 29, 1974, pp. 1491-1494.
- Belo, F., V. D. Gala, and J. Li, 2011, Government Spending, Political Cycles and the Cross Section of Stock Returns.
- Blanco, R., Brennan, S., Marsh, I., 2005. An Empirical Analysis of the Dynamic Relation between Investment-Grade Bonds and Credit Default Swaps. Journal of Finance 60, 2255–2281.
- Campbell, J. Y. (1996): “Understanding Risk and Return”. Journal of Political Economy, Vol. 104, No. 2, 298-345.
- Cutler, D. M., 1988. "Tax Reform and the Stock Market: An Asset Price Approach," American Economic Review, American Economic Association, vol. 78(5), pages 1107-17, December.
- Daves, P.R., Ehrhardt, M.C., and Kunkel, R.A., (2000), “Estimating Systematic Risk: The Choice of Return Interval and Estimation Period, Journal of Financial and Strategic Decisions, Vol. 13, No. 1, pp. 7-13.
- Gandhi, P., Lustig, H. (April 2011). Size Anomalies in U.S. Bank Stock Returns: A Fiscal Explanation.
- Hooper, V. J., Hume, T. P. and S-J Kim, Sovereign Rating Changes - Do They Provide New Information for Stock Markets? (2008). Economic Systems, Vol. 32/2, pp. 142-166, 2008.
- Hull, J., M. Predescu, and A. White, 2004, The Relationship between Credit Default Swap Spreads, Bond Yields, and Credit Rating Announcements, Journal of Banking and Finance 28, 2789–2811.
- Jeanneret, A., 2010, “Sovereign Default Risk and the U.S. Equity Market”, working paper, HEC Montreal.
- Longstaff, F. A., J. Pan, L. H. Pedersen and K. J. Singleton, 2011, "How Sovereign is Sovereign Credit Risk?", American Economic Journal: Macroeconomics, forthcoming.
- Petkova, R. (2006): “Do the Fama-French Factors Proxy for Innovations in Predictive Variables?” Journal of Finance, Vol. 61, No. 2, 581-612.
- Vasicek, O. 1973. “A Note on Using Cross-Sectional Information in Bayesian Estimation of Security Betas,” Journal of Finance, 28, 1233-1239.