While the main focus of academic and professional investment research in the risky corporate credit market has typically been on the performance of high-yield bonds, including, importantly, the default and recovery rates of those bonds that default, there is very little similar work on defaulted bonds and loans. This is understandable, as the high-yield bond market is significantly larger and has a 40-year track record, with coverage and data from many sources.1
To address this issue, we have accumulated data on the monthly performance of defaulted bonds and loans, typically throughout the Chapter 11 reorganization period, for our analysis. The performance data have been documented in the Altman/Kuehne NYU indexes of monthly returns for securities from default to emergence from Chapter 11 (or liquidation from Chapter 7 or 11).2 These indexes are used extensively by distressed debt investment managers as an important benchmark of their performance, as well as by analysts and researchers in this space.
It is appropriate now to analyze, in considerable detail, the performance of a large sample of defaulted bonds (and loans) during the post-default period. The last time such an analysis was done was on a much smaller sample of bonds covering the period 1980-1992.3 That sample involved only 202 bond issues from 91 firm issuers and considerably smaller samples when the analysis was by bond seniority. This current study covers the period 1987-2Q 2016 and involves a sample more than five times that of the prior study, including 1,189 issues with bond prices at default and emergence (1,727 issues overall) from 803 firms, and 730 loan facilities from 398 firms.
We also analyze the monthly performance 1-24 months after default, as well as the performance from before and after the major changes to the U.S. Bankruptcy Code in late 2005. Our analysis also covers the market for defaulted bank loans, which was virtually not analyzable prior to the mid-1990s due to lack of data on this private-debt market and, in our case, was only documented when we created the Index on Defaulted Loans in 1996. For the period 1996-2Q 2016, the number of issuers and the number of issues were 245 and 635, respectively, for corporate bonds, and 201 and 354, respectively, for loans.
Figure 1 shows the monthly, unweighted, average prices on a sample of defaulted bonds from the end of the default month to up to 24 months after default and for the period from default to emergence from the bankruptcy restructuring period, usually Chapter 11, or to liquidation if the reorganization was not successful. Figure 1 covers the period January 1987 to June 30, 2016,4 and includes all defaults except distressed exchanges, which do not have a post-default restructuring period.5 The restructuring period may be longer than the Chapter 11 period, since many defaults (about 50 percent) occur prior to the bankruptcy filing.
For the entire approximately 30-year period, 1987-2Q 2016, there were 1,727 defaulted bond issues in our sample, of which 954 had prices for period 1-12 months after default, 493 issues for the 24-month post-default period, and 1,189 issues for the period of default to emergence.
The default-to-emergence period can last from as little as less than one month to more than 60 months. Figure 2 shows the distribution of monthly periods from default to emergence for the entire 30-year sample period and for the two sub-periods noted. The average emergence period was about 28 months for defaulted bonds for the period 1987-2016, 34 months for 1987-2005, and much less, 16 months, for 2006-2016. The median period was 19, 24, and 11 months, respectively.
It will come as no surprise to most analysts and practitioners in the bankruptcy industry that the 2006-2016 period witnessed a considerably shorter average reorganization period due to a number of changes in the Bankruptcy Code, most clearly the limit of 18 months on plan exclusivity for the debtor. And, note that many Chapter 11, post-default bankruptcies take less than 12 months, and some even less than six months, to conclude due to the increased use of late of prepackaged and prearranged Chapter 11s.
This is most clearly demonstrated in the period 2006-2016, where the most populated six-month reorganization interval was 0-6 months (Figure 2). Of course, many very recent Chapter 11 filings have not concluded their reorganization. However, we included their data in our time series for some of our calculations but not for the default-to-emergence sample.
In terms of return performance, Figure 3 shows the annualized change in average bond prices for four intervals for our entire sample of defaulted bonds: (1) from default to emergence, (2) from default to 12 months post-default, (3) from default to 24 months post-default, and (4) from the 12-24 months period. These results are shown for both the entire 30-year sample period and for the most recent 10 years.
We can observe from Figure 3 that the annualized average rate of return for all bonds (1987-2016) was 11.08 percent per year. This compares with 25.34 percent per year for the more recent 10-year period (2006-2016). Even when excluding the 38 energy company issues that concluded their Chapter 11 restructure periods in 2015-2016, these results change very little.
If an investor purchased our entire sample of defaulted bonds at default and held them for 12 or 24 months, the average annualized returns were 8.49 percent for 12 months and 13.58 percent for 24 months on just those bonds that lasted for those intervals for the entire 30-year sample period. For the most recent 10-year sample period, the annualized returns were 26.20 percent for 12 months and 19.92 percent for 24 months. Many of the more recent samples were prepackaged or prearranged Chapter 11s that lasted less than 12 months, and the results were annualized for the one and two years compilations.
In general, returns appear to be quite impressive for investing in a broad, diversified portfolio of defaulted bonds. However, while average annual returns are high, there is a considerably high “cost” in terms of volatility, as measured by the standard deviation in price change among the issues that comprise our portfolios’ returns. Investors must understand that distressed debt in general and especially defaulted bonds have exceptionally high variability of their returns. We can observe this high volatility in our bond sample for all periods, but see lower volatility for defaulted loans.
In our standard deviation calculations (Figure 3), we truncated the sample by eliminating the top 5 percent of issues that had exceptionally high returns (mainly “penny” bonds), but we did not truncate for negative price change issues; e.g., some bonds’ prices go to zero as certain bondholders, usually subordinated owners, see their bonds get almost completely wiped out. We were concerned with outlier observations, especially on the positive return side.
Indeed, in the case of extremely low prices at default, the change in price, i.e., returns, could be incredibly high, even with a small absolute change. For example, a price movement from $2 to $8 during the reorganization interval would result in a 400 percent return. Even with our truncation of outlier positive returns, the standard errors are quite high, ranging from about 70.8 percent for the default-to-emergence interval for the period 1987-2016 to 106.6 percent for the 2006-2016 period.
Despite reasonably good average return performance for defaulted bonds, especially in the last 10-year period, we recall from our prior study (endnote 3) some stark differences between bonds with different seniorities. Indeed, despite very small samples in our prior study, published in 1994, it appeared that only the two most senior tranches of corporate bonds, i.e., senior secured and senior unsecured, did well, and the junior bonds, i.e., subordinated and discounted bonds, did not do well at all. Our results in the current study had mostly similar results, with a few important exceptions.
Figures 4 and 5 show the changes in price performance for the 1-12, 12-24, and 1-24 months periods, and from default to emergence for the entire 30-year sample period, stratified by the four seniority classes. We combined the two subordinated classes, senior subordinated and junior subordinated, for sample size purposes and because, recently, junior subordinated bonds essentially no longer exist, with companies having only one class of subordinated bonds, if any at all.
Senior unsecured bonds demonstrated excellent average annual returns for all four interval periods, ranging from 17.4 percent for default to emergence, to 28.5 percent for the 12-24 months period, while the senior secured sample showed only modest high-single-digit returns. The latter results are considerably lower than was the case for the period before 1992, when senior secured bonds were clearly the superior performing seniority class.
In our current study, sample sizes were 335 issues for senior secured bonds and 925 issues for senior unsecured bonds at default. The number of months for the interval default to emergence averaged 22.3 for senior secured bonds and 28.2 for senior unsecured bonds for the period 1987-2016. But, as we saw earlier, for the more recent period, 2006-2016, the intervals were much shorter (Figure 5). The major reason for the poorer performance of senior secured bonds in our current study compared to our earlier study is the much lower average price at emergence ($55.26) compared to slightly above par value for the very small number of issues in our earlier study that covered the period 1980-1992.
We were surprised to find that the average prices at emergence for the senior secured sample ($55.26) was only slightly higher than for the senior unsecured sample ($53.96). It is quite likely that the quality of the collateral for senior secured bonds has deteriorated as the senior secured class of bonds has become so commonplace in bond syndications. And, many of the companies had only one class of bonds outstanding, so samples for senior secured and senior unsecured are not identical.
The results for the subordinated class were very disappointing, with the average annual return from default to emergence actually negative, at -2.4 percent, and even lower, at -10.9 percent, during months 1-12. These results are similar to our earlier study, when both the senior subordinated and subordinated classes had lower average prices at emergence than at default. While discounted bonds fared much better than subordinated ones, the size of our discounted sample was only 13 issues.
The defaulted loan performance statistics for all periods, as shown in Figures 6 and 7, indicate much poorer performance than for defaulted bonds. Indeed, all average annual returns are in the low to medium single digits for the interval default to emergence. Excluding energy company defaults in the period 2015-2016 from the calculated returns makes little difference.
The volatility of these returns, as measured by the standard deviation for the 492 issue sample in the 1996-2016 20-year period, was considerably lower than the 30-year period for bonds, but still quite high. This volatility measure was about 38 percent for the 20-year sample period, and slightly higher, at about 41 percent, for the more recent 10-year period. There was not much difference when we did not include the 2015-2016 energy defaults.
For defaulted loans, the average default-to-emergence interval for the 20-year period 1996-2016 was about 20 months and much less, at 13 months, for the more recent 2006-2016 period (Figure 7). It was as high as 27 months for the period 1996-2005 and hence fell significantly for the 2006-2016 period. The median period was 14 months for the period 1996-2016 and 18 months for 1996-2005, and also fell significantly, to only 10 months, for 2006-2016. This again reflects the changes in the 2005 revisions to the Bankruptcy Code and the increased reliance on prepackaged and prearranged Chapter 11s.
We will expand upon our defaulted debt performance analytics in the future when we risk-adjust returns by discounting prices at emergence for opportunity costs over the reorganization period. Likely candidates for an appropriate discount rate to use are the comparable period returns on a diversified portfolio of high-yield bonds and/or distressed bonds, i.e., those bonds yielding more than 1,000 basis points over comparable duration Treasurys. Our subsequent results will also feature returns and volatility measures for bonds stratified by industrial sectors.
In this study, we have used the percentage change in the mean prices over the period from default to emergence as well as from 1-24 months after default as a measure of investment return over the full period since 1987 or 2006. We will be expanding on this analysis to compute returns based first on calculating individual security returns; then comparing them with high-yield bonds, the Standard & Poor’s 500 index, or other indexes over the same calendar months; and finally aggregating them using either equal or nonequal weights. This will allow us to compare the return/risk trade-off between defaulted bonds and loans and these other indexes.
We also feel that comparing the volatility around the annualized return of a sample of numerous individual securities, as shown in Figures 3 and 7 for defaulted bonds and loans, with the volatility of an index like the high-yield bond or the Standard & Poor’s 500 stock index is not consistent or a fair comparison. Index returns, which represent a single summary statistic, will almost assuredly have a much lower variance around some time series mean than will a large number of individual observations, some with outlier properties. So, we plan to construct an index of defaulted bond and loan mean monthly price levels from the default month to emergence and compare the return and variance with alternative indexes for the same calendar month periods, i.e., to compare apples to apples.
To better assess the risk of defaulted bond and loan investments, given their unique return characteristics, we are also looking at other methods to measure volatility rather than relying uniquely on the traditional standard deviation. We will compute other volatility measures, such as the standard deviation of returns below the mean, which is called the semideviation, or below a target return, which is called the downside deviation.6
Because upside volatility is a plus for an investment and perhaps should not be included in the risk calculation, these methods only factor in downside standard deviation, rather than the total standard deviation that includes both downside and upside returns. In fact, high outlier returns, such as with defaulted bonds and loans, as demonstrated earlier, can have the effect of increasing the total standard deviation more than the average return.
Finally, we plan to examine the price behavior of defaulted bonds for at least the six-month period before the default date. Time constraints did not allow these additional tests to be performed for this report. We hope these additional tests and comparisons will allow us to provide a more robust and complete anatomy of investing in defaulted corporate bonds and loans.
In this article, we addressed the performance characteristics of investing in defaulted bonds and loans. Such research work has rarely been done before, given the specialized nature of such investments and the lack of extensive historical data.
For both defaulted bonds and loans, we analyzed the time periods 1987-2016 and 2006-2016, given that the Bankruptcy Code was revised extensively in late 2005. While both time periods showed double-digit average rates of returns for defaulted bonds, the time period 2006-2016 showed superior performance, possibly due to the Bankruptcy Code revisions. However, the volatility of these returns, as measured by the traditional standard deviation, is quite high.
Pertaining to defaulted loans, the average annual returns are much poorer, albeit with lower volatility than for defaulted bonds. Given that investment returns in defaulted bonds and loans can have extreme outliers, we will be considering other types of return and risk measures to compute volatility that will address this issue.
Investment returns for defaulted bonds were also analyzed by seniority. Senior unsecured bonds performed the best, with double-digit average returns, while senior secured bonds showed only modest high-single-digit returns. Results for the subordinated class were disappointing, with negative average returns, while discounted bonds fared much better, albeit with a small sample size.
For the purpose of portfolio allocation, our next work will be expanding on this analysis by calculating returns for defaulted bonds and loans for the same calendar months as the high-yield bond index, the Standard & Poor’s 500 index, and possibly other benchmarks, and then constructing an index of defaulted bond and loan performance and comparing it to that of these indexes.