PRELIMINARY: Corporate Bond Spreads and the Pandemic III: Variance Across Sectors and Firms
We saw a marked increase in corporate bond spreads during early March, and Federal Reserve announcements starting March 23 contributed to stabilizing the disruptions in credit markets.1 We documented, however, that the evolution of credit spreads is quite different across economic sectors. For example, sectors that are more vulnerable to social distancing measures, such as arts and entertainment or accommodation and food services, performed worse than sectors in which containment and mitigation measures should have a smaller impact, such as manufacturing or utilities.
In this post, we continue exploring the heterogeneity in credit spread movements across sectors and dig deeper by looking at variation within sectors, firms, and bond issuances. We find that there is substantial heterogeneity both across firms and across bonds for the same firm, particularly during the period of credit market disruption. Moreover, we find a strong correlation between the increase in the sectoral credit spread and the increase in within-firm dispersion. These results imply that, to understand the evolution of credit markets, it might be important to look at and understand not only the average spreads but also higher-order moments.
First and Second Moments
We divide 2020 into three periods. First, the “pre-COVID-19” period, before financial market instability started around February 28. Second, the “pre-intervention” period, before financial market volatility and instability were curbed by Federal Reserve interventions in these markets on March 23. Finally, period three is after the March 23 interventions. Figure 1 compares the evolution of first and second moments of credit spreads, specifically their increase between the first and second periods. The figure shows the change in the average spread in each sector (on the x-axis) versus the change in the standard deviation of credit spreads in each sector (on the y-axis) between these two periods.2
There is a tight correlation between the change in first and second moments across sectors: Sectors that experienced larger increases in average credit spreads also tended to display larger increases in the variance of those spreads. Next, we try to understand where this relationship is coming from.
Decomposing the Second Moment
We decompose the variance of credit spreads into variation at three different levels: (1) variation across sectors; (2) variation across firms, within sector; and (3) variation across bonds, within each firm and sector.
Table 1 shows the results of an orthogonal variance decomposition between these three components in the three different periods of 2020. We report the relative contribution of each component to the total variance in credit spreads within a period. First, the variation across sectors has a small and relatively constant contribution, explaining between 11% and 13% of the overall dispersion. Second, the contribution of variation across bonds of the same firm increased considerably during the pre-intervention period (from 18% to 42%), at the expense of a reduction in the variation across firms (from 69% to 47%).
This decomposition suggests that most of the increase in variance during the second period happened across bonds issued by the same firm and not so much across firms. In other words, it is not so much that some firms’ or sectors’ credit spreads are rising considerably more than those of other firms or sectors. Instead, secondary market prices of bonds issued by the same firms are becoming more disperse.
The Second Moment Across Sectors
Sectors with larger increases in average spreads also registered larger increases in the standard deviation of those spreads. Is this correlation associated with increases in dispersion across or within firms?
Figures 2 and 3 study how credit spreads changed in period 2 across sectors. In the horizontal axis we show the increase in the average spread of the sector (as in Figure 1). In the vertical axis of Figure 2 we show the change in the variation across firms, while in Figure 3 we show the change in the variation of the within-firm component. The figures also report the R-squared of a simple linear regression of the two variables. While both components are positively related to the average spread, the correlation is much stronger for the variation within firms, across issuances. This means that sectors with larger increases in credit spreads tended to register larger increases in dispersion across spreads of bonds issued by the same firm and not so much larger increases in dispersion of average spreads for different firms.
How to interpret these results? Pure credit risk should be the same for all bonds issued by a single firm (as we are focusing on bonds of equal seniority). For this reason, an increase in the dispersion of credit risk should be reflected in the across-firm component. We find, instead, that the increase in average spreads is related to the within-firm component that may be related to other factors, such as the decrease in liquidity documented by Kargar, Lester, Lindsay, Liu, Weill, and Zuniga (2020).
1 See previous blog posts: https://www.stlouisfed.org/on-the-economy/2020/april/effects-covid-19-monetary-policy-response-corporate-bond-market and https://research.stlouisfed.org/resources/covid-19/preliminary/corporate-bond-spreads-and-pandemic-heterogeneity-across-sectors.
2 More concretely, we start from a daily panel of corporate bond spreads as in our previous blog posts and compute the standard deviation of those corporate bond spreads over the January 1–February 28 and March 1–March 22 periods. The change is then the difference between the standard deviation in the latter and the former periods. The daily corporate bond spreads correspond to the average credit spread over all transactions in a single day. Thus, our analysis abstracts from intraday variation in bond spreads across transactions. For an analysis of intraday corporate bond liquidity during the same period that complements ours, see Kargar, Lester, Lindsay, Liu, Weill, and Zuniga (2020): http://www.econ.ucla.edu/cbml/corporate-bond-liquidity.html.
Preliminary, incomplete. To cite, please request author’s permission.
© 2020, Federal Reserve Bank of St. Louis. These views do not reflect the opinion of the Federal Reserve Bank of St. Louis or the Federal Reserve System.