Uncategorized
A liquidity score for leveraged loans
admin | September 13, 2019
This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors.
Bid depth and facility size are longstanding benchmarks for the liquidity of loans backing CLOs, but the link between these measures and liquidity is subtle. Liquidity does not rise in a straight line with bid depth, and the liquidity lift from bid depth also depends on facility size. Earlier work by APS captures some of that subtlety, but new work allows scoring liquidity at the loan level. For investors concerned about their managers’ ability to adapt to changing market conditions, knowing the full distribution of liquidity behind a CLO could be valuable.
Extracting illiquidity signals from unchanged prices
Asset prices in efficient markets should change almost continually to reflect new information. Economic announcements and earnings come and go, for instance, and prices change. But in illiquid markets, high transaction costs may discourage trading and leave observed prices more stable than the true, underlying view of value. In a set of assets affected by the same information, the ones with the most price volatility could end up being the most liquid.
With leveraged loans, reliable measures of liquidity are hard to find. Loans may have quoted bid-offer spreads, but the depth of the market often is uncertain. Trading volume is hard to come by. And the price impact of large transactions is hard if not impossible to calculate. A measure that gauges liquidity from the persistence of asset returns—an asset’s return autocorrelation—works only on an aggregated basis but not on the individual loan level. Individual loan prices can sometimes stay constant for long periods, translating into a steady string of 0% returns. Without continual variation, it becomes difficult to calculate a reliable return autocorrelation. However, loans with prices that rarely change signal illiquidity. In this way, price volatility itself can serve as a measure for liquidity.
Calculating a liquidity score for each leveraged loan
This analysis calculates a liquidity score for each leveraged loan in the market since October 2016 with an initial facility size of at least $250 million. The score simply reflects the percentage of weeks with a change in loan price of any amount. Loans with no price changes would show a liquidity score of 0% at one extreme, and loans with a change each week would show a liquidity score of 100% at the other extreme.
Measured by weekly price volatility, leveraged loans show a wide dispersion. A notable 10% of loans in the sample show no changes in price over their entire history. Around 50% show a liquidity score of 50% or less and the remaining 50% show liquidity score of 50% or more.
Exhibit 1: Liquidity Score is widely dispersed between 0% and 100%
Source: Markit, Amherst Pierpont Securities; Note: Only U.S. loans with initial facility sizes greater than $250 million and weekly data between October 2016 and August 2019 are included in this analysis.
Liquidity depends on both bid depth and facility size
Investors can calculate bid depth and facility size as proxies for the liquidity of the loans backing CLOs, so a comparison between these two measures and our liquidity score can validate the new measure and help explore its additional value. The analysis puts every loan into a group defined by bid depth and facility size (Exhibit 2). As facility size rises, bid depth tends to rise, too. The weighted average bid depth rises from 1.93 in loans with Bid Depth 1 to 6.22 in loans with Bid Depth $2.5 billion.
Exhibit 2: Number of price observations on leveraged loans in each group
Source: Markit, Amherst Pierpont Securities; Note: Only U.S. loans with initial facility sizes greater than $250 million and weekly data between October 2016 and August 2019 are included in this analysis.
The data reveals nuances in the liquidity of the loans that neither bid depth nor facility size alone captures (Exhibit 3). It summarizes the liquidity of all loans in a group defined by bid depth and facility size as an average of their liquidity scores weighted by facility sizes.
Exhibit 3: Liquidity score rises faster in lower bid depth groups
Source: Markit, Amherst Pierpont Securities; Note: Only U.S. loans with initial facility sizes greater than $250 million and weekly data between October 2016 and August 2019 are included in this analysis. For loans falling into each group, data shows percentage of weeks where prices changed for each loan weighted by facility size.
A walk down the columns shows an almost uniform pattern where liquidity score rises fast in the lower bid depths and more slowly in the higher bid depths. For example, for loans with Bid Depth 1 and Facility Size $250 million, week-to-week prices change only 18.3% of the time. For loans with Bid Depth 2 and Facility Size $250 million, the liquidity score more than doubles to 45.6%. And for Bid Depth 3 and Facility Size $250 million, the liquidity score rises more slowly to 52.4%. This pattern repeats for all other categories of Facility Size, too. It is a clear sign that liquidity rises with bid depth.
The pace of rising liquidity depends on facility size, however. For larger facilities, liquidity accelerates much faster as bid depth rises. For Facility Size $250 million, liquidity doubles from Bid Depth 1 to 2. For Facility Size $2.5 billion, liquidity goes up more than four times from Bid Depth 1 to 2. The nonlinear relationship between bid depth, facility size and liquidity score is clear in a three-dimensional display (Exhibit 4). The liquidity curve is relatively flat at smaller facility sizes and steepens as facility size grows. Surprisingly, the most illiquid loans are ones with Bid Depth 1 and the largest facility size.
Exhibit 4: Liquidity score rises faster – the liquidity curve is steeper – at higher facility sizes
Source: Markit, Amherst Pierpont Securities; Note: Only U.S. loans with initial facility sizes greater than $250 million and weekly data between October 2016 and August 2019 are included in this analysis. The plot is made using a Generalized Additive Model to estimate the contribution of bid depth, facility size, and their interaction to predicting the liquidity score. This model fits a nonlinear relationship for each of the three features and adds them together, allowing for easier interpretation.
The uses of loan-level liquidity scores
Most investors have good intuition for the relative liquidity of loans in a CLO portfolio, but it often does involve combining intuition about bid depth with intuition about facility size. The nonlinear relationship between these common attributes and liquidity makes loan-level scores useful for identifying the distribution of liquidity in a CLO portfolio. It should also help identify differences between CLOs that have similar averages but wider dispersions. For CLO investors interested in whether their managers have enough liquidity to adjust to changing markets, a picture at the loan level could prove valuable.