The Macroeconomic Determinants of the Term Structure of Illiquidity Premia

• Applicants: Professor Dr. Philipp Schuster; Professorin Dr. Marliese Uhrig-Homburg
• Subject Area: Accounting and Finance ; Statistics and Econometrics
• Term: from 2012 to 2020
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 226975202

Illiquid bonds pay a premium to compensate bond investors for the lack of liquidity. Our results from the first part of the project, however, suggest that a clear connection between illiquidity premiums and typical liquidity measures like the bid-ask spread is only significant during economic crises. Therefore, we ask the natural follow-up question, why existing liquidity measures seem to work only well in stress periods. Building on this question, we analyze how the measurement of liquidity in general should be adapted to the economic conditions. From a conceptual point of view, at least two reasons point towards a measurement of liquidity that depends on the economic context. First, typical liquidity measures calculated from completed transactions provide only an incomplete picture. The reason is that transactions that have not been executed due to extreme illiquidity do not show up in the data. Second, for bond illiquidity premiums, expected transaction costs at the future trading date are the relevant costs and not today's transaction costs. Since transaction costs increase strongly in times of economic stress, it remains unclear whether current liquidity in calm periods is a good proxy for expected future transaction costs. In the proposed second part of the project, we first want to develop a forecasting model for expected transaction costs. For that, the remaining maturity of the bond as a natural ceiling on the holding horizon plays a central role, leading to a term structure of liquidity measurement. In the second step, we use the liquidity measures derived from the forecasting model to re-assess the impact of illiquidity on bond prices. The explanatory power of the newly developed liquidity measure in this exercise also serves as a criterion to evaluate and compare different measures. Since we are not aware of an approach to measure liquidity dependent on the economic environment in any security market, we plan to extend our analysis to the stock market in the last step. We expect that our results have practical influence in at least two ways. First, our forecast model provides expected trading costs for bonds, for which there are few or no data. From a practical perspective, information on trading costs is very important for these illiquid securities. Second, our new measures allow establishing a connection between the liquidity of a security and the associated price impact at an arbitrary point during the business cycle. Financial institutions could then apply, for example, scenario analyses, to predict the impact of a deterioration of liquidity in times of crisis, which would improve risk management.

• DFG Programme: Research Grants