State-Dependent Illiquidity Premium in the Korean Stock Market

ABSTRACT

We study the relation between the illiquidity premium and economic states in the Korean stock market. We find that aggregate market liquidity improves following real economic expansions and expansive monetary states and worsens after economic recessions and restrictive monetary states. The improved liquidity in the expansion–expansive state generates a huge illiquidity premium, while an illiquidity premium does not exist in the recession–restrictive state. As a result, the observed illiquidity premium displays strong state-dependent variations. Our empirical results indicate that a significant unconditional illiquidity premium in the Korean stock market arises due to a substantial illiquidity premium in the expansion–expansive state.

KEY WORDS:

• illiquidity premium
• monetary conditions
• real economic conditions
• stock market liquidity

Notes

1. Although several researchers have performed empirical work on the liquidity-return relation in emerging equity markets (e.g., Batten and Vo 2014; Jun et al. 2003; Lou et al. 2014), none of them investigates its time variation across economic states.

2. Throughout this article, the macroeconomic conditions include both real economic condition and monetary condition.

3. While the two articles use changes in the Fed discount rate to define monetary policy regimes, we employ a more direct measure of monetary conditions: unexpected innovation in the growth of the monetary base. For changes in monetary policy to indicate future monetary conditions correctly, clear evidence on the transmission of monetary policy from the target variable to monetary aggregates should be provided, which is not covered in the current article. For example, Jensen and Moorman (2010) use the Fed discount rate after confirming that it is related to monetary aggregates. However, since such a relation has not been confirmed in the Korean economy, we use a more direct measure of monetary aggregates instead.

4. The number of lags included is determined by Aikaike’s information criterion, and our empirical results are not sensitive to the choice of lags.

5. We replicate Figure 2 and Table 3 incorporating all liquidity measures used in this study. Specifically, we first construct time series of aggregate illiquidity innovations based on each of four liquidity measures; we then compute the principal components of four different series of illiquidity innovations. We define a new variable of aggregate illiquidity innovations as the first principal component of them. We find that our empirical findings are qualitatively similar.

6. Since the responses of aggregate illiquidity to shocks in monetary conditions peak after two months, as shown in Figure 2, we report aggregate illiquidity innovations in month t + 2 across economic states in month t, allowing for a two-month time lag.

7. When we examine the illiquidity premium solely based on ILLIQ-A, the results are not similar with the results using TO, ILLIQ-B, and LM12. However, when we construct size-controlled portfolios based on ILLIQ-A, we find that the state-dependent nature of the illiquidity premium becomes consistent with the results based on other measures. The result based on the size-controlled portfolios indicates that the result based on ILLIQ-A alone could be significantly affected by the size effect as we expect.

Additional information

Funding

This work was supported by the Hankuk University of Foreign Studies Research Fund of 2015.