Liquidity Discount Model and Spillover Effects
52 Pages Posted: 5 May 2017
Date Written: April 13, 2017
The recent global financial crisis reignited concerns over systemic risk in the financial industry as a new type of systemic risk emerged – the severe loss of asset value due to illiquidity. The crisis has sparked a large body of research and has led to a number of new quantitative indicators of systemic risk. Despite the successful empirical results, the proposed indicators are primarily empirically motivated and are not capable of differentiating liquidity risk from other types of risk such as market and credit. In this paper, we develop a liquidity model to measure the risk caused by liquidity distress in the financial market. The model is parsimonious and has a semi-closed form to compute a discount due to a lack of liquidity. The liquidity discount measure may be applied to particular firms or may be aggregated into a liquidity index. A key feature of the model is its ability to distinguish between and compute the probability of liquidity default and economic default (insolvency).
The model is applied to the study of liquidity in the financial sector leading to new empirical results. First, during the financial crisis, the probability of liquidity default (20%) was more than two times greater than the probability of economic default (8%). Second, standard measures of expected shortfall greatly underestimated shortfalls by 10% during the financial crisis by ignoring the loss of asset value due to illiquidity. Third, shocks to systematic liquidity risk are long lasting and can persist for as long as nine months depending on the financial subsector, and fourth, the level of bank interconnectedness peaks during periods immediately preceding crises creating the perfect environment for an unexpected liquidity shock to rapidly permeate the financial sector leading to a liquidity-induced crisis.
Keywords: liquidity discounts, liquidity index, asset liquidity, systemic risk
JEL Classification: G, G01, G02, G12, G21, G24, G28
Suggested Citation: Suggested Citation