Amihud (2002)

Results for average betas for decile portfolios, correlation matrix for Amihud (2002) and panel regressions are discussed as follows:

5.2.1 Average betas for Decile Portfolios

Table 7 presents the times series average betas estimated for ten portfolios based on the Amihud (2002).

Table 7. Average betas for Amihud (2002)

Illiquidity ratio β1 β 2 β 3 β 4 β 5 β 6

(Lowest)1 0.8633 0.0037 -0.0040 -0.8088 -0.8085 0.0548 2 1.1265 0.0047 -0.0052 -0.9901 -0.9896 0.1370 3 1.1964 0.0046 -0.0048 -1.0413 -1.0411 0.1553 4 1.2205 0.0047 -0.0049 -1.0725 -1.0723 0.1482 5 1.1004 0.0043 -0.0045 -0.9243 -0.9241 0.1763 6 1.1383 0.0052 -0.0052 -1.0180 -1.0181 0.1203 7 1.0647 0.0044 -0.0045 -0.9175 -0.9175 0.1472 8 1.0878 0.0043 -0.0044 -0.9598 -0.9597 0.1282 9 1.0639 0.0043 -0.0045 -0.9078 -0.9076 0.1562 (Highest)10 1.1991 0.0045 -0.0047 -1.0303 -1.0300 0.1691

Beta 1 is depicting an increasing trend for decile portfolios sorted on the basis of Amihud (2002) illiquidity measure. It can be observed the values of beta 1 are high and positive. The results are in line with Acharya and Pedersen (2005). Vu et al. (2015) also reported increasing trend in the beta 1 but the values of the beta in their study under this measure are small. Beta 2 is also showing an increasing trend and has positive values through portfolio 1 to portfolio 10.

The values of the beta 2 are small and a sudden jump in the value of beta 2 is observed for portfolio 6. These values of beta 2 contrast with the results reported by Vu et al. (2015) for Amihud (2005) measure, as the values get very large from portfolio 1 to portfolio 10.

Beta 3 values are negative and depict an increasing trend, however, the values are of small magnitude. These results for beta 3 are similar to Lee (2011) and Vu et al. (2015), however, contrast with Acharya and Pedersen (2005) results. Beta 4 which presents depressed wealth effect has values which are negative and of high magnitude .Moreover, the values of beta 4 are depicting an increasing trend i.e. with the level of illiquidity the beta 4 increases. These results

are in line with Vu et al. (2015).Beta 5 and Beta 6 also show an increasing trend however beta 5 values are negative. The results of beta 6 are in line with Vu et al. (2015), however beta 5 are contradicting.

5.2.2 Correlation Matrix

Table 8 presents the correlation matrix between the estimated betas for Amihud (2002), E_c is the residual of the autoregressive process 2 of the Amihud (2002). SIZE and MOM (momentum) are the control variables.

Table 8. Correlation matrix for Amihud (2002)

Beta 1 and beta 6 are highly and negatively correlated. Beta1, beta 2, beta 3 and control variable SIZE have moderate negative correlation between them. Beta1 and control variable Momentum have very weak correlation. Beta 1 and beta 5 have significant positive correlation between them. The correlation between residuals of autoregressive process 2 for Amihud (2002) and beta 1 is weak. Beta 2 and beta 3 have a significantly high correlation between them. Beta 2 and beta 5 have significant but negative correlation between them. Whereas, correlation between beta 2 and beta 4, beta 2 and controls variables is very weak. The correlation between residuals of autoregressive process 2 for Amihud (2002) and beta 2 is almost zero.

Beta 3 and beta 5 exhibit significant negative correlation. Correlation between beta 3 and beta 6 is negative but very weak, whereas, correlation between beta 3 and beta 4 is very weak.

Looking at the correlations between beta 3 and control variables Momentum and SIZE, very weak correlation is found. Correlation between beta 3 and between residuals of autoregressive

process 2 for Amihud (2002) is almost zero. Beta 4 and beta 5 have weak negative correlation.

The correlation between residuals of autoregressive process 2 for Amihud (2002) and beta 4 is near to zero. Beta 5 and beta 6 have moderate correlation. However, Beta 6 and control variable SIZE have moderate correlation and beta 6 and control variable Momentum (MOM) have almost zero correlation between them. Correlation between the control variables SIZE and Momentum is significantly weak.

5.2.3 Panel Regression Results

Table 9 presents the Panel regression results for the Amihud (2002) according to equation (22) to (28).The columns (1) to (7) under the heading models of Table 9 correspond to estimations for equations (22) to (28). The results of Hausman test can be found in APPENDIX 1.

Individual liquidity betas (beta1 to beta 4) after controlling for market risk, firm size past returns are all significant according to regressions 1 to 4. Beta 1 is also significant after controlling for firm size and momentum. Now, hypotheses of the study shall be discussed for betas estimated under Amihud (2002). Beta 1 the level of liquidity is positively priced in the cross section of returns for UK market. Hence, we shall accept hypothesis 1 of the study. This result is in line with Acharya and Pedersen (2005) and Vu et al. (2015). Second hypothesis deals with co-movement between individual stock illiquidity and market illiquidity, also termed as the commonality in liquidity. Having a look it at the significance level of the coefficient of beta 2, it can be concluded that commonality in liquidity is significantly priced for UK market. Which is line with results of studies carried out by Foran et al. (2015), Galariotis and Giouvris (2007, 2009) for UK market.

Beta 3 which captures the flight to liquidity phenomenon and is hypothesized that covariance between stock return and market illiquidity negatively effects stock returns for UK market.

After controlling for factors such firm size and momentum the beta 3 has negative sign is significant at 1% level. From, these findings we are able to conclude that beta 3 is negatively priced at UK market. Beta 4 is negative and significant and therefore we accept the hypothesis 4 estimated under Amihud (2002). Now, aggregate betas shall be discussed for Amihud (2002).

The results of regression (5) and regression (6) presented in Table 9 suggests that beta 5 (combined liquidity risk) and beta 6 (aggregate systematic risk) are positive and significant.

Therefore, hypothesis 5 is accepted and it can be concluded that aggregate systematic risk is positively related to the stocks.

The results of model (7) in table 9 show that value of coefficient of beta 1 has decreased compared with model (1).

Table 9. Panel Regression Results for Amihud (2002)

Models (1) to (7) represent equations (22) to (28). E_c presents residuals of AR (2), MOM and SIZE are control variables momentum and size. Values in the parenthesis indicate t-statistics for each coefficient.

All the coefficients of regression and F-stat values are significant at 1% level Variable