Time-varying illiquidity premium in the Nordic stock markets : Empirical evidence from Finland, Sweden, Norway and Denmark
Mellas, Kasper (2021-06-02)
Time-varying illiquidity premium in the Nordic stock markets : Empirical evidence from Finland, Sweden, Norway and Denmark
(02.06.2021)
Julkaisu on tekijänoikeussäännösten alainen. Teosta voi lukea ja tulostaa henkilökohtaista käyttöä varten. Käyttö kaupallisiin tarkoituksiin on kielletty.
suljettu
Julkaisun pysyvä osoite on:
https://urn.fi/URN:NBN:fi-fe2021060434074
https://urn.fi/URN:NBN:fi-fe2021060434074
Tiivistelmä
Illiquidity premium has been studied extensively in the previous financial literature. The findings suggest that illiquidity is priced in various stock markets and that the relation between any asset’s liquidity and return is negative; lower liquidity results in higher average returns. Most recent research on asset pricing has increasingly focused on examining the time-variability of both the betas and risk premiums with conditional models. The main object of this thesis is to study whether illiquidity risk is priced in the Nordic stock markets and whether this premium is higher in an economic state characterized by high liquidity-betas – the stock returns’ sensitivities to shocks in market-wide liquidity.
The empirical part of the thesis compares unconditional and conditional specifications of the asset pricing model with varying risk factors and includes data from Finland, Sweden, Norway and Denmark. The data sample covers years between 2000–2020. The estimation is conducted using Fama-MacBeth two-step regression procedure with traded-factors for market excess return, size, value and illiquidity as explanatory variables.
The results indicate that illiquidity risk is not priced in any of the tested four markets, since most of the estimated coefficients for both unconditional and conditional illiquidity premiums were statistically insignificant. This can be a result of misspecification of the regression model that used traded factors as explanatory variables and assumed no intercept in accordance with the selected theoretical framework. Additionally, the estimation results did not support the hypothesis that the conditional version of the asset pricing model provides better explanatory power for stock return behavior.
However, the estimated monthly IML – the return difference of portfolios including the least liquid stocks over portfolios including the most liquid stocks – was higher in high liquidity-beta state in all of the tested markets implying that a conditional structure driving the illiquidity premium could exist. These results were robust to changes in key parameters of the estimation process. Overall, even though the illiquidity premium was found insignificant, the estimation results did not provide sufficiently strong evidence so that the existence of the conditional illiquidity risk premium could be rejected with full confidence.
The empirical part of the thesis compares unconditional and conditional specifications of the asset pricing model with varying risk factors and includes data from Finland, Sweden, Norway and Denmark. The data sample covers years between 2000–2020. The estimation is conducted using Fama-MacBeth two-step regression procedure with traded-factors for market excess return, size, value and illiquidity as explanatory variables.
The results indicate that illiquidity risk is not priced in any of the tested four markets, since most of the estimated coefficients for both unconditional and conditional illiquidity premiums were statistically insignificant. This can be a result of misspecification of the regression model that used traded factors as explanatory variables and assumed no intercept in accordance with the selected theoretical framework. Additionally, the estimation results did not support the hypothesis that the conditional version of the asset pricing model provides better explanatory power for stock return behavior.
However, the estimated monthly IML – the return difference of portfolios including the least liquid stocks over portfolios including the most liquid stocks – was higher in high liquidity-beta state in all of the tested markets implying that a conditional structure driving the illiquidity premium could exist. These results were robust to changes in key parameters of the estimation process. Overall, even though the illiquidity premium was found insignificant, the estimation results did not provide sufficiently strong evidence so that the existence of the conditional illiquidity risk premium could be rejected with full confidence.