This case study analyzes the credit risk of a sample portfolio of corporate bonds and private debt holdings. The sample portfolio was intended to mimic a largely safe lender or asset manager that has recently expanded into private debt. Scroll down to learn about the methodology and findings.
Introduction
The growth in private debt markets has implications for credit quality and portfolio risk that have not yet been tested in an economic downturn. Private debt offers attractive returns, can provide a hedge against rising inflation, and may help diversify a portfolio. It has recently seen aggressive expansion by new market entrants, such as insurers and pension funds. These benefits, however, must be weighed against the downside, which includes illiquidity and often higher credit risk.
Today, in the fourth quarter of 2022, the global economy is slowing and may be headed for recession. Rising interest rates have boosted returns but place additional pressure on a borrower’s debt servicing. It is critical for lenders and portfolio managers to understand the performance of their portfolio under a variety of conditions, and to ensure that they hold adequate capital against these outcomes. A well-constructed portfolio analysis provides an estimate of portfolio losses under various downside scenarios and can reveal hidden pockets of risk that may not be readily apparent when viewing an exposure or portfolio segment in isolation. Diligent risk analysis can also justify expansion into a new segment by identifying better investment or hedging opportunities to increase profits while managing risk.
This report shows how to analyze and manage a portfolio of public and private corporate credit exposures. For example: Is there a segment that contributes excessively to portfolio risk? What is the risk contribution of the private debt portfolio? Is the return on a particular segment worthwhile given the risk? The key finding in this study is that private debt contributes materially to tail losses but that, when managed within a robust risk framework, provides opportunity for growth and can improve the portfolio’s overall risk-adjusted return.
Industry-level analysis of risk vs return
Overview of analysis + key findings
In this analysis, we set up a portfolio of public and private debt exposures and ran it through PortfolioStudio to assess the portfolio’s credit performance. The key finding is that the private debt segment, overall, contributes materially to portfolio risk, despite accounting for a relatively small portion of exposures. We also examined concentrations at the industry level and uncovered opportunities to improve the portfolio’s risk-adjusted return by limiting and expanding exposure in different segments. Finally, we ran a hypothetical “What-if” scenario to assess whether this lender could safely expand its footprint in private debt. This exercise showed a modest increase in tail losses, but a material increase in returns, suggesting that the portfolio’s risk-adjusted returns could be improved by further expansion into private debt.
Portfolio and data description
The portfolio used in this analysis consisted primarily of public bonds, with a smaller portion of loans to private companies. It was intended to mimic a largely-safe lender or asset manager that has recently expanded into private debt. The overall portfolio was 6,515 instruments across 1,031 entities, with a total exposure of $20.6 billion.
The portfolio consisted of three segments as follows:
1. Corporate bonds from public companies: This segment largely mimicked the composition of an investment-grade bond index and contained 6115 bonds issued by 631 publicly-listed companies. Total exposure was $18.6 billion, or 78% of the sample portfolio. Probability of default measures (PD – needed as an input to PortfolioStudio) were drawn from the Moody’s CreditEdge database for publicly-listed firms, which calculates PDs on a daily basis, for all publicly-listed firms worldwide (over 40k daily). Our loss given default assumption (LGD – also needed to run PortfolioStudio) was a flat rate of 50%. Correlations and R-squareds (i.e. company-specific correlation sensitivity) were drawn directly from the PortfolioStudio database.
2. Loans to private companies, low credit risk: This was a segment of loans issued to private companies, comprised of 200 loans, issued to 200 companies, of $10 million each, with a total exposure of $2 billion, or 11% of the portfolio. These companies were drawn from the Moody’s Credit Research Database (CRD), which is a data consortium of more than 92 million financial statements representing more than 18 million global private firms, built in partnership with over 90 financial institutions worldwide. The 1,000 firms chosen were all US-based and were, on average, high quality credits. These financial statements were run through the Moody’s RiskCalc model to derive PDs. We used a flat LGD assumption of 75%. For the interest rate, we mapped the PD to an implied rating, then used the St Louis Fed’s corporate bond spreads by broad rating category, plus an illiquidity premium of 2% for private debt. The average interest rate spread (over US Libor) was 4.5% across the sample. Correlations and R-squareds were proxied using size, industry, and country characteristics of each exposure. We refer to this segment as “Private credit: Investment Grade” as the average PD corresponds with a Baa3 rating.
3. Loans to private companies, high credit risk: This was a segment of loans issued to private companies, but with very limited information on each exposure. To generate PDs, we used the EDF-X benchmark model, which generates a proxy PD based on limited input data. These exposures were, on average, lower quality credits. Each entity was assigned at random an industry, size (assets), and debt leverage. All firms were US-based. We used an LGD of 75%. Correlations and R-squareds were proxied using size, industry, and country characteristics of each exposure. Interest rates were derived using the same logic as the Private debt: Investment Grade loans – the average interest rate spread was 5.5%. We refer to this segment as “Private credit: High Yield” as the average PD corresponds with a Ba2 rating. This segment was also comprised of 200 loans, issued to 200 companies, of $10 million each, with a total exposure of $2 billion, or 11% of the portfolio.
The following section summarizes the portfolio characteristics along key dimensions, broken down by these three segments.
Probability of default
The average 1-year PD in this sample was 0.31% for Corporate bonds, compared to 0.49% for Private credit: Investment Grade, and 1.4% for Private credit: High Yield. This ordering reflects that the Corporate bonds were, by design, investment grade, indicating high credit quality; Private credit: Investment Grade were smaller and riskier; while Private credit: High Yield were riskier still. The density of each cohort is provided in Figure 1, below.
Figure 1: Distribution of 1-year PDs
Firm size
The parent companies of Corporate bonds in the portfolio were larger, with most reporting (book) assets between $10 billion and $100 billion. Private credit: Investment Grade exposures drawn from the CRD were notably smaller, with the largest cohort between $100 million and $1 billion in assets. The segment of Private credit: High Yield showed an even distribution from under $100 million through to above $10 billion.
Figure 2: Distribution of firm size, Book Assets (USD)
Industry
The four industry spikes in Corporate bond counts below are in REITS, Banks and S&Ls, Finance NEC, and Electric Utilities. Among Private credit: Investment Grade, the four largest industries were Business Services, Business Products Wholesale, Consumer Services, and Transportation. The portfolio of Private credit: High Yield was randomly assigned and shows an even distribution among non-financial industries.
Figure 3: Distribution of industry
R-squared
PortfolioStudio requires an R-squared, which is a measure of how correlated an entity is with an economy’s various “underlying factors”. There are over 900 factors in the GCorr framework covering different countries/regions, assets classes, industries, and macroeconomic variables – this depth and granularity allows the model to accurately capture the relationship of each firm to the factors. The correlation of each exposure, both to the underlying factors and to other assets in the portfolio, is critical for understanding concentration risks in the tail of the distribution of outcomes. A higher R-squared indicates that a company will be more closely tied to the ups and downs of the broader economy.
Corporate bonds issued by public firms, which tend to be larger and more interconnected, showed a broad range of R-squareds, but were higher on average compared to private credit. The R-squared of private credit were relatively lower, reflecting that the credit risk of these exposures is driven more by firm-specific factors unrelated to the broader economy. The lower R-squared for Private credit: Investment Grade reflects the cohort’s smaller firm size compared to Corporate bonds and Private credit: High Yield.
Figure 4: Distribution of R-squared
Overall risk
Economic capital in this exercise was $1.58 billion, or a loss rate of 7.0% for this 1 in 10,000 event. The median loss rate in the distribution was 0.06%, while the expected loss rate (mean) was 0.18%. Figure 5 shows the portfolio loss rates, ranked from lowest to highest, confirming that portfolio losses are highly asymmetric.
Figure 5: Portfolio loss rate per trial (%), 1mil. simulations
Is this tail loss rate of 7.0% too high? The answer is “it depends”. A firm with a more aggressive mandate to pursue riskier assets, perhaps to expand in a new market or to generate higher returns, is likely to hold more capital reserves and may be comfortable with this loss rate. A more conservative firm managing a supposedly safe portfolio – possibly with lower capital holdings – may be more alarmed by this number.
Slicing the data by the three entity types above, tail losses were driven by $1.2 billion in losses on the Corporate bond segment (6.6% loss rate), $132 million in losses on Private credit: Investment Grade (6.6% loss rate), and $220 million on Private credit: High Yield (11% loss rate). The Corporate bond segment accounted for the largest share of tail risk – this is expected, given that it accounted for the largest share by exposures. One slightly unintuitive result is that the tail loss rate on Corporate bonds was the same as Private credit: Investment Grade, even though the average 1-year PD (and LGD) was lower in Corporate bonds (0.31% vs 0.53%). This is because of the correlation component. The R-Squareds on Corporate bonds were notably higher, which results in higher losses in the tail, all else being equal.
The analysis also revealed a sharp difference in risk between the two private credit segments, with the High Yield segment accounting for almost twice the losses in the tail. Overall, the private debt segments accounted for a material share of tail losses (22% of the total). A more important question, which we address below, is whether the returns on these assets adequately compensate for the risk taken.
Table 1: Economic capital by portfolio segment
We can continue to slice the data along various dimensions to gain insight into losses in the tail of the distribution and which portfolio segments are driving this loss. This analysis is critical for understanding a portfolio’s credit risk concentrations. Table 2, below, shows the breakdown of economic capital by industry. Similar to the analysis by instrument type, this breakdown can reveal hidden pockets of risk beyond a simple view of expected loss in the middle of the distribution.
Table 2: Economic capital by industry
Banks and S&Ls accounted for $311 million or 20% of total tail losses. This is unsurprising as it had the largest exposure (15% of the total), although it suggests some concentration risk over and above its simple exposure. Security Brokers & Dealers also contributed an outsize share of tail losses, accounting for 9% of economic capital, compared to 5% of total exposure. Conversely, Utilities, Electric was the second largest industry by exposure (6% of the total), but accounted for just 3% of economic capital.
Risk vs return
Examining the tail loss in isolation can provide some insight into the riskiness of the portfolio, however portfolio managers are also interested in the returns of each segment. Analyzing the risk-return payoff can yield important insights about over or under-performing segments.
Figure 6 plots segment-level tail losses against revenue (bubble size indicates exposure) – a data point in the top-left offers a risk-return payoff better than the overall portfolio, while a data point in the bottom-right offers a worse risk-return payoff. In this analysis, the portfolio-level return on capital (=Revenue/Economic capital) was 58%, comprised of 54% for Corporate bonds, 87% for Private credit: Investment Grade, and 60% for High Yield. By this measure, private credit exposures, particular the Investment Grade segment, provide an attractive risk-return payoff compared to the rest of the portfolio, suggesting that the excess returns adequately compensate for the additional risk taken.
Figure 6: Expected revenue vs Economic capital, by portfolio segment
We can slice the portfolio along different dimensions to identify segments which provide a better or worse risk-return payoff. Looking at the industry breakdown, below, we see considerable variation, Security Brokers & Dealers provide a risk-adjusted return of 29%, while Electric Utilities is more attractive at 105%.
Figure 7: Expected revenue vs Economic capital, by industry
For more information about active portfolio management solutions and the benefits it brings, please visit our Portfolio Studio page.
What can I do with this information?
Clients can use the information above to optimize the risk-adjusted return of their portfolio. For example, a portfolio manager may increase their holdings of instruments or segments with a high risk-adjusted return, while reducing exposure to instruments or segments with a low risk-adjusted return. Similarly, users may implement a risk-adjusted-return hurdle rate to screen new exposures. Figure 8, below, shows instrument level risk-adjusted returns to illustrate how this portfolio management could work in practice.
“What-if” analysis
Analysts can gain deeper insight into portfolio risk by adjusting various model inputs. In this section, we simulate an expansion in the private loan segment of the portfolio by doubling the holdings of every private company loan. We reran the portfolio to assess the impact of this shift on the entire portfolio.
Figure 9: Expected revenue vs Economic capital, by portfolio segment (private loan exposures doubled)
Conclusion
An understanding of portfolio credit risk is an essential part of a prudent risk manager’s toolkit. The exercise in this report shows how a portfolio analysis can help determine the optimal holdings of private debt in a broader portfolio, and inform decisions about asset allocation, credit limits, and concentration risk measures.
Appendix: Portfolio results
For more information about active portfolio management solutions and the benefits it brings, please visit our Portfolio Studio page.
