Swedroe: Liquidity Risk Not Worth It
A study on the liquidity premium in fixed income suggests it’s not worth chasing.
Economic theory posits that liquidity risk should be an important determinant of security prices, with investors demanding protection (in the form of a premium) for the future challenge to trade in a cost-effective and timely manner.
Diogo Palhares and Scott Richardson of AQR Capital Management contribute to the literature in this area with their March 2018 study, “(Il)liquidity Premium in Credit Markets: A Myth?” Their objective was to test for a liquidity premium in the $14 trillion global corporate bond market.
While the market is large, it is far less liquid than the equity market, with bonds trading far less frequently, and typically with a much higher bid/offer spread relative to underlying volatility.
Prior research has established that illiquid bonds tend to have higher spreads (i.e., greater trading costs) than liquid bonds. While that is interesting, it doesn’t tell us about returns to corporate bond investors. Palhares and Richardson focused their study on the risk-adjusted returns investors actually earned.
Their data set included constituents of the Bank of America Merrill Lynch investment-grade (US Corporate Master Index) and high-yield (US High Yield Master Index) bond indexes, and covers the period January 1997 through December 2016. Their sample included almost 28,000 bonds from 5,310 issuers. In dollar terms, about 80% of the bonds were investment-grade.
To determine liquidity risk, the authors used a variety of measures that reflect bond liquidity, including measures related to bid/ask spread, average daily trading volumes, turnover, issue size, price impact and frequency of zero-trading days. Each month, Palhares and Richardson sorted corporate bonds into quintiles based on each liquidity measure and computed the return of a long/short portfolio that buys the least liquid bonds (i.e., smaller issue sizes, higher bid/ask spreads, lower trading volume, higher price impact or higher frequency of zero-trading days) and sells the most liquid bonds (i.e., larger issue sizes, smaller bid/ask spreads, higher trading volume, lower price impact or lower frequency of zero-trading days).
Results
Following is a summary of their findings:
- The average bid/ask spread was 29 cents (per $100 par value) for both investment-grade and high-yield bonds, and the average daily trading volume was $2.2 million ($2.5 million) for investment-grade (high-yield) corporate bonds.
- The average market impact cost was 29 basis points (39 basis points) per $1 million traded for investment-grade (high-yield) corporate bonds.
- The average investment-grade (high-yield) bond trades on less than 32% (36%) of days over the prior six months—liquidity in corporate bonds was considerably lower than in traditional listed equity markets.
- Spread differences between liquid and illiquid bonds were insignificant within the same issuers.
- Measures of bid/ask spread and price impact (the cost of trading) were positively correlated, reflecting the direct costs of trading both in investment-grade and high-yield bonds.
- Daily trading volume, issue size and frequency of zero-trading days were all strongly positively correlated, reflecting the ability to trade.
- There was a weaker correlation between the ability to trade (daily trading volume, issue size and frequency of zero-trading days) and credit spreads for both investment-grade and high-yield markets. Across multiple measures of liquidity and a variety of methods to control for correlated characteristics of more (less) liquid bonds, there was only limited evidence of a liquidity premium in the cross section of corporate bonds.
- The 20% most illiquid investment-grade bonds had a bid/ask spread 64 basis points larger than the 20% most liquid. That large bid/ask spread is equivalent to almost two years of the spread advantage of illiquid (as measured by bid/ask spread) investment-grade bonds, creating a large hurdle for providing a liquidity premium.
- While illiquid bonds had slightly higher credit spreads and directionally higher average returns, portfolios that tilt toward (away from) less (more) liquid bonds exhibit considerably higher levels of volatility.
- Economically, the low Sharpe ratios of illiquidity-factor-mimicking portfolios were hard to justify for an investor, which is puzzling, as theory suggests investors should demand a risk premium for holding less-liquid assets. This finding was robust to alternative measures of risk-adjusted returns.
The one “positive” result the authors found was that “older” bonds are associated with higher future credit excess returns. However, they noted: “The interpretation of this result is not that it supports the existence of a liquidity risk premium in the cross section of corporate bonds. This is because the more direct measures of liquidity such as trading volume, price impact, turnover, bid-ask spreads and frequency of zero trading days show no relation with future excess credit returns. Instead, we argue that time since issuance reflects [a] characteristic of corporate bonds that investors either neglect or shy away from.”
Palhares and Richardson concluded: “Illiquidity factors would receive close to a zero allocation in typical investor portfolios, and this conclusion is before considering the impact of higher expected transaction costs for less liquid bonds.” They added: “While liquidity risk affects price a little, consistent with theory and past empirical research, the effect is very small and importantly liquidity risk is not a key driver of credit excess returns.”
The authors attempted to provide an explanation for the anomaly. They hypothesize: “The marginal investor in this market has a preference for less liquid bonds. Whilst conjectural, this is possible as the typical investors in the corporate bond market are large institutional investors such as corporate and public pension plans and insurance companies. To the extent that these investors face regulatory capital constraints and/or solvency requirements that can be adversely affected by having to mark to market their asset holdings, investing in less liquid assets provides a degree of discretion/optionality to smooth reported asset values in periods of stress.”
Additional Research
Palhares and Richardson do note that an alternative approach to measuring liquidity risk is to treat liquidity as an exposure that has time-varying risk. For example, Lubos Pastor and Robert Stambaugh, authors of the 2003 study “Liquidity Risk and Expected Stock Returns,” found that stocks whose prices drop when illiquidity increases earn a risk premium above and beyond their standard factor exposures.
And this approach has been extended to corporate bond markets. Hai Lin, Junbo Wang and Chunchi Wu, authors of the 2011 study “Liquidity Risk and Expected Corporate Bond Returns,” found that bonds with high sensitivities to liquidity risk earn returns 4.0% higher than those with low sensitivity.
Thus, while Palhares and Richardson were “unable to find any evidence of associations between multiple measures of liquidity and future credit excess returns, it is possible a liquidity ‘beta’ may resurrect the case for a liquidity premium in the corporate bond market.”
Summary
Palhares and Richardson examined whether liquidity risk in the corporate bond market is sufficiently compensated. Their answer is a puzzle, or an anomaly, for economic theory, as they found little evidence to support the notion that investors were compensated with a risk premium for bearing liquidity risk. They did note, however, that prior research had found a liquidity beta that was compensated.
Larry Swedroe is the director of research for The BAM Alliance, a community of more than 140 independent registered investment advisors throughout the country.