ANALYSIS technical

Choosing a skilled CLO manager Roberto Liebscher and Thomas Mählmann argue that some managers are more skilled than others – and that skilled managers deliver higher returns from CLO equity tranches

I

s there heterogeneity in CLO management skill? And if so, what is it that distinguishes the best from the worst? In our study we tackle these questions using proprietary data about the performance of equity tranches during reinvestment periods from CLO-i and Bloomberg. This data spans the period from 1998 through to the middle of 2012 and covers 817 deals. Hence, our sample represents roughly 70% of the market, according to figures from the Securities Industry and Financial Markets Association (Sifma). To find out if some managers are more skilled, we use a well-known argument: if a manager is indeed more skilled (or better informed) than its peer group, the returns it delivers should be persistently higher, conditional on the risk taken. On the other hand, if a manager delivers an occasional great performance we would presume it was just lucky.

Is performance a matter of luck? To test this hypothesis we first calculate for each deal a time series of half-year equity tranche cash-on-cash returns by summing up all cash distributions to equity tranche holders within each half-year and dividing this sum by the nominal value of the tranche. In a second step, we sort managers into quintiles according to their deal-size weighted-average half-year equity return in excess of the corresponding average half-year equity return of all deals in our sample and follow quintile performance in subsequent half-years. We find striking results consistent with the notion that some managers are skilled while others are not. For instance, at the sorting date managers in the top quintile generate a 25% higher annualised excess cashon-cash return than managers in the worst quintile. More importantly, we find that differences in performance diminish only slowly. The best managers significantly outperform the worst for more than four years. The chart (above right) shows this difference in manager average excess cash-on-cash returns along with its 95% confidence interval. It is also interesting to know whether persistence in skill is evident between various deals from a manager. Such a finding would guard against a potential misinterpretation of return differences as differences in skill. For instance, it might be the case that, by pure chance the 16

16-17 1502 analysis technical.indd 2

portfolio composition of one of a given managers’ deals is of higher quality (higher spreads for the same risk) ex ante. Then return differences would evolve mechanically without any difference in skill. However, our results on a deal-by-deal basis point in the same direction as our previous analysis: if a manager outperforms its peers with one deal, it is likely to repeat this success with follow-on deals. As can be seen from the table (see right), going with the best managers results in a 19% higher annualised excess equity tranche return for the first CLO of a manager but yields also a significantly higher performance for all subsequent deals. You can judge a manager by its track record Is this because some managers enjoy laxer trading restrictions or covenants? Probably not. We analyse the auto-correlation of deal performance controlling for deal covenants. The results remain the same: equity tranche performance is highly persistent. Previous performance is predictive of future performance, so investors may judge a manager by its track record. The importance of this finding is not restricted to equity tranche investors. Since high equity tranche returns represent a big cushion against any debt tranche losses, debt tranche investors can learn from a manager’s equity tranche return record, too. Of course, this only works out as long as the high equity tranche returns we find are not the result of risk-shifting, that is, managers do not increase the volatility of returns in order to increase expected returns at the cost of debt tranche investors – a point we will address later. A natural question is how do the best managers come to be at the top of the league? We consider three potential answers. First, we analyse whether CLOs of superior managers show lower primary market spreads on debt tranches. In a hypothetical income statement of a CLO, lower interest expenses raise profits, ceteris paribus, implying higher equity tranche returns. Second, we look at CLO portfolio summary statistics and ask: conditional on a certain level of default risk, do managers trade in higher yielding (potentially mispriced or mis-rated) assets? Recycling our income statement example this is equivalent to an analysis of the revenues of a CLO. Finally, we consider the potential risk-shifting explanation already mentioned. February 2015 Creditflux

28/01/2015 19:09

ANALYSIS technical Excess return over average cash-on-cash CLO equity return 30

%

25

20

15

10

5

0

-5 -12

Half-years after sorting -9

-6

-3

Manager performance across deals Manager Deal 1 Deal 2 quintile Q5 9.18% 6.90% Q4 3.09% 4.05% Q3 -0.18% 0.18% Q2 -3.51% -2.52% Q1 -9.50% -4.75% Q5 minus Q1 18.68% 11.66% p-value (0.000) (0.000)

0

3

6

9

12

Deal 3

Deal 4

Deal 5

5.98% 5.03% 0.42% -2.60% -4.15% 10.13% (0.000)

5.74% 3.61% 0.52% -2.08% -2.53% 8.27% (0.001)

3.14% 4.51% 0.81% -1.92% 0.70% 2.44% (0.037)

Looking at our income statement again we can rule out that lower interest expenses drive the results. The name of the manager does not play a role in determining debt tranche spreads. In fact, spreads can be almost perfectly explained by overall market conditions and the tranche’s rating. A simple regression of the logarithm of spreads on rating notch and half-year fixed effects explains 95% of the variation in log spreads. Where do excess returns come from? In order to understand the origin of excess returns we look at the other side of the income statement. Controlling for several key variables like the weighted average rating factor of the portfolio, the diversity test level and the issuance year of the deal (among others), we find that managers that generate higher equity tranche returns purchase or hold loans with higher spreads. This is in line with loan spreads not only capturing rating information but also other characteristics that skilled (or better informed) managers are able to exploit. Another explanation might be that those at the top of the league anticipate rating changes. For instance, they might invest in loans rated at the edge in expectation of an upgrade. Once the anticipated rating change becomes effective, spreads tighten as prices rise and the weighted average rating of the portfolio improves in favour of the rating threshold. Creditflux February 2015

16-17 1502 analysis technical.indd 3

Ratio CPU time 10,000 Interestingly, high equity tranche returns do not come at a cost to debt tranche investors. In our analysis we AAD 1,000 neither find higher collateral default rates nor higher bumping debt tranche rating downgrade probabilities for deals 100 managed by the most skilled. Rather, we see that top managers serve both investor types equally well. Debt tranche investors benefi10t from lower downgrade probabilities and higher overcollateralisation ratios. Equity Line label style 1 investorsAxis benefi t from higher cash flows. label style This result is indicative of successful deal structures Chart head style in which deal covenants 0like overcollateralisation and 10 interest coverage ratios as well as minimum spread and diversity requirements are effective in mitigating problems of information asymmetries and risk-shifting. It is in the interest of the manager to maximise the return on the equity tranche even if it does not hold a stake of the equity portion of the deal because it will benefit from high equity tranche performance through incentive fees.

Good reasons for an investment in CLOs Overall, our results make a case for CLO investments. While performance might differ a lot from one deal to another, defaults have been rare. Deal covenants seem to be effective in aligning different noteholders’ interests. This makes CLOs an interesting investment alternative especially in times of low interest rates. Our analysis shows that even the worst managers outperform the S&P/LSTA Leveraged Loan Index, in line with a higher risk compensation for CLO equity tranche investments. In spite of these findings, and arguments from other authors that a liquid secondary market for syndicated loans reduces financing costs for companies, regulation has not been favourable for CLOs. But regulators should consider the speciality of the CLO market rather than lump all securitised products together. Ideally, secondary market trading of loans through CLOs should be encouraged in order to increase efficiency and liquidity in the market. This would reveal whether top performers can stay ahead even under increased competition. However, theory suggests that if pricing becomes more efficient the chances of consistent outperformance will decrease. This article summarises results from the working paper “Are Professional Investment Managers Skilled?” See ssrn.com/abstract=2417082. We thank Michael Brehm and Raphael Ruess for research assistance and Steven Henny for helpful comments on the data.

Roberto Liebscher is a research associate at the Catholic University of Eichstätt-Ingolstadt.

Thomas Mählmann is Professor of Finance and Banking at the Catholic University of Eichstätt-Ingolstadt. 17

28/01/2015 19:14