Bridge over ocean
1 December 2013 CFA Institute Journal Review

Noise as Information for Illiquidity (Digest Summary)

  1. Yueping Liu

Using the noise that exists in the US Treasury bond market, the authors create a measure that they establish as a priced liquidity risk factor. The measure extends beyond existing liquidity proxies and reflects the overall market conditions. This noise measure captures liquidity crises and helps explain cross-sectional returns on hedge funds and currency trades.

What’s Inside?

Hedge fund and currency carry trade returns are known to be associated with the availability of arbitrage capital. The thesis is that the shortage of arbitrage capital allows yields to deviate from the model-implied yield curve, resulting in more noise in prices. The authors use the noise measure as a factor in cross-sectional pricing models and confirm that the beta premiums are significant on hedge fund reported returns and hypothetical currency trade returns. The constructed noise measure spikes up during major liquidity events (e.g., the Lehman Brothers bankruptcy) and also provides a relative severity measure.

How Is This Research Useful to Practitioners?

The noise measure that the authors create contains more information than that explained by existing measures, such as the VIX, on-the-run premiums, or default spreads. It is also unrelated to the volatility of the US Treasury market, which makes this constructed noise measure unique.

The authors claim that the noise measure reflects the liquidity supply in the overall Treasury market rather than shocks to the liquidity demand of individual Treasury securities, which are mostly averaged away as capital moves fluidly across the yield curve.

Past analysis tended to focus on the impact of liquidity on a specific market (e.g., corporate bonds). Through empirical analysis, the authors argue that the noise measure reflects overall market liquidity conditions beyond the Treasury market, as evidenced by the explanatory power of the noise measure for asset returns in hedge funds and currency trades. But the model results are better for currencies in developed countries than for currencies in developing countries. The hedge funds used in the test have a fair representation of various styles and sizes. Furthermore, by replacing the noise measure with one of the other proxies for liquidity risk in the factor pricing model, the authors find that the other measures are not priced into hedge fund returns, which further proves the uniqueness of the noise measure.

The authors highlight the risk and return characteristics of an investment strategy’s liquidity needs, which has pricing and asset allocation implications. In addition to stock market beta, portfolio managers should evaluate whether the expected return is justified by the liquidity risk being assumed. The results may be more directly relevant to funds that invest in the broad market, as well as in hedge funds and currencies.

How Did the Authors Conduct This Research?

The authors back out zero-coupon yield curves from daily Treasury bond prices from 1987 to 2011 and define the noise measure to be the root mean squared distance between the market yields and the model-implied yields. Bonds with maturities of between 1 and 10 years are included in the model to construct the noise measure. Bonds with maturities between 1 month and 10 years are used for the hypothetical yield curve.

Hedge fund data are from the Lipper TASS database, with information on both “live” and “graveyard” funds from 1994 to 2011. Factor beta on the noise measure is estimated by using a two-factor model (the other being the stock market). Then the authors sort the data into 10 portfolios according to beta, with a large negative beta indicating high exposure to liquidity risk. Distribution of various fund styles is even across different liquidity exposures, except for emerging market funds, which tend to be more aggressive. On average, the more conservative portfolios tend to have lower returns, and the most conservative ones exit more often than the most aggressive ones. But during 2008, the exit rate was higher for funds in the most aggressive category, which reflects the severity of the liquidity crisis. The authors also add a lagged factor to capture return smoothing and obtain a more accurate estimate of liquidity exposure.

A Fama–MacBeth cross-sectional regression is conducted on the noise beta, the market beta, hedge fund age, and assets under management. The noise factor premium is indeed significant and has the correct sign, which indicates that the corresponding risk has been priced in. By putting the risk premium back into the two-factor model and pricing in the liquidity risk, the alphas are no longer statistically significant.

For currency trades, the authors consider a maximum of 34 currencies from 1987 to 2011. They construct hypothetical currency portfolios and sort them according to interest rates (or forward discounts) from the perspective of a US investor. Factor exposure is again estimated on the two-factor model. High interest rate (asset currencies) portfolios tend to have more negative noise beta, which implies worsening performance during a liquidity crisis. This result is also consistent when using carry trade indices data.

Abstractor’s Viewpoint

As the authors imply, liquidity risk is priced into returns and pricing anomalies may be difficult to exploit precisely because of the lack of capital to implement the trades. In addition, certain “surprise” liquidity events and the severity of capital shortages are not easy to predict in advance. Without a reasonable forecast of the noise measure itself, the pricing model has limited predictive power. The authors’ results may also be partly driven by a few extreme values (e.g., those during and after the Lehman Brothers bankruptcy), which may reduce its explanatory power during “normal” times. Furthermore, additional risk factors may be needed for specific asset classes when it is not feasible to invest in a broad range of different markets.

We’re using cookies, but you can turn them off in Privacy Settings.  Otherwise, you are agreeing to our use of cookies.  Accepting cookies does not mean that we are collecting personal data. Learn more in our Privacy Policy.