For both models, combining the effect of investment efficiency and corporate governance (INVEFF1*CG and INVEFF2*CG) has a wider and higher effect. Merging both variables gives a more negative mean reaching −33.65 and −125.77 respectively. A boost in the maximum results reaches 444.35 and 1117.16, recording the highest maximum results among other variables of the sample. It happens with the minimum, which shows the extreme decline to be −392.53 and −859.67.
Finally, the table shows that the mean size for UK firms is 1,799,360,000 (3.26) billion GBP as the value of firm assets, with a maximum of 50.781 (4.71) billion GBP, and a minimum assets’ value of 97.96 (1.99) billion. Financial leverage (LEV) has a mean of 19.6% indicating that the sample firms are not highly leveraged. The maximum mean of the leverage is 57% and the minimum is 0% show a large dispersion in firms’ debt ratios (
Habbash et al. 2016). Profitability (PROF) has a mean of 0.059, with a maximum value of 0.18, minimum −0.09 and a standard deviation of 0.043, and skewness of 0.287. Firm liquidity (LIQ) has a mean of 1.342, with a maximum value of 4.09, minimum of 0.40, a standard deviation of 0.634, and skewness of 1.465. Sales growth (SG) has a mean of 5.379 similar to
Hassanein and Hussainey’s (
2015) findings, while a dispersion shows the maximum growth of firms by 49.61 and least (minimum) growing firms of −31.9 showing a deterioration. The average firm age is 66 years (as the table shows the natural logarithm of the age figures), with a maximum age of 255 years (2.41) and minimum age of less than a year, and a standard deviation of 0.36 and skewness of −0.66.
VOLDIS is the performance-related disclosure score; DISTONE is the disclosure tone score; INVEFF1 is the first investment efficiency proxy (Equation (5)); INVEFF2 is the second investment efficiency proxy (Equation (6)); CG is the corporate governance score; INVEFF1*CG is the interaction between corporate governance and the first proxy of investment efficiency; INVEFF2*CG is the interaction between corporate governance and the second proxy of investment efficiency; FS is the log firm size measured by total assets; LEV is firm leverage; PROF is firm profitability; LIQ is firm liquidity; SG is sales growth; AGE is the log of firm age; LOSS is a dummy variable equals 1 for loss-making companies and 0 otherwise; MTB is market-to-book ratio; DIV is a dummy variable equals to 1 for firms pay dividends in year t and 0 otherwise.
Table 3 and
Table 4 show the correlation analyses.
Table 3 shows the correlations for the variables included in Models 1 and 2; while
Table 4 shows correlations for the variables included in Models 3 and 4. The tables show that investment efficiency (
INVEFF1 and
INVEFF2) have an insignificant correlations with voluntary disclosure (VOLDIS) and disclosure tone (DISTONE). The corporate governance (CG) indicated a positive highly significant correlation with VOLDIS, while a negative insignificant correlation with DISTONE. The interacted variables of
CG and investment efficiency (
INVEFF1*
CG and
INVEFF 2*CG) have an insignificant correlation with VOLDIS and DISTONE. The tables also show that FS shows a positive significant correlation with VOLDIS having a 99% confidence level while an insignificant correlation with DISTONE. LEV and AGE have an insignificant correlation with both VOLDIS and DISTONE. LIQ also has a highly significant but negative correlation with VOLDIS with a confidence level of 99%. However, PROF and SG have a negative and insignificant correlation with VOLDIS and a positive highly significant correlation (99% confidence level) with DISTONE. MTB has a positive and highly significant correlation (99% confidence level) with DISTONE, and LOSS correlation with DISTONE has the same strength but with a negative direction. DIV has insignificant correlation with DISTONE.
Regression Analysis
The joint effect on corporate governance and investment efficiency on voluntary disclosure.
We introduce an interaction variables (INVEFF*CG) in our regression models to examine the joint impact of governance and investment efficiency on disclosure. This study complements that of
Elberry and Hussainey (
2020).
Table 5 shows that after adding
CG, INVEFF1)*CG, and INVEFF2*CG, the ANOVA test still reflects that both models are significant. The R-squared and adjusted R-squared values increased slightly by on average 2%, which indicates that
CG and
INVEFF*
CG do contribute in explaining the change in the dependent variable VOLDIS.
The results show that the first model’s independent variables INVEFF1, CG, and INVEFF1*CG have a significant impact on VOLDIS, while the second model’s independent variables INVEFF2, CG, and INVEFF2*CG have an insignificant impact on VOLDIS.
According to the hypotheses developed for estimating the scenarios which can explain the impact of
CG and the joint effect of
INVEFF*
CG on VOLDIS (
Hussainey and Walker 2009), the first model’s variables can justify which of these hypotheses are accepted and which are not.
Table 5 shows that the values of the coefficients for
INVEFF1,
CG, and
INVEFF1*
CG are −1.073, 0.121, and 1.014 respectively, which consistent with the third scenario. The sum of the coefficients on
INVEFF1,
CG and
INVEFF1*
CG is significantly greater than the sum of the coefficients on
INVEFF1 and
CG, shown in the following equation:
Therefore, the hypothesis is rejected and the third scenario applies which states that investment efficiency and corporate governance are complements. This complementary effect shows that investment efficiency and corporate governance provide related information and complete each other’s’ roles.
Considering the direction of the relationships,
INVEFF has a negative and significant impact on VOLDIS while
CG and
INVEFF*
CG have a positive and significant impact. The negative association between
INVEFF and VOLDIS could be explained by the proprietary costs theory as when firms’ investment efficiency rises, the cost of disclosing information increases by threatening their competitive advantage and position. And so, these firms tend to disclose less (
Verrecchia 1983;
McKinnon 1984;
Feltham and Xie 1992;
Newman and Sansing 1993;
Darrough 1993;
Gigler et al. 1994). The positive impact of corporate governance on VOLDIS is expected and supported by previous studies such as
Samaha et al. (
2015);
Habbash et al. (
2016); and
El-Diftar et al. (
2017). So, we can conclude that introducing the
CG positive impact had reduced the negative impact of
INVEFF and this leads to a positive impact of
INVEFF*
CG on VOLDIS, indicating that when firms having strong corporate governance and investing efficiently, they tend to disclose more.
As for the control variables, concerning both models, FS shows a positive and significant relation with VOLDIS supporting the findings of
Hassan et al. (
2006) and
Al-Hadi et al. (
2016), while the remaining control variables show an insignificant association with VOLDIS.
VOLDIS is the performance related disclosure score; INVEFF is investment efficiency; CG is the corporate governance score; INVEFF*CG is the interaction between corporate governance and investment efficiency; FS is the log of firm size measured by total assets; LEV is firm leverage; PROF is firm profitability; LIQ is firm liquidity; SG is sales growth; AGE is the log of firm age; LOSS is a dummy variable equals 1 for loss-making companies and 0 otherwise; MTB is market-to-book ratio; DIV is a dummy equals to 1 for firms pay dividend in year i and 0 otherwise.
The joint effect on corporate governance and investment efficiency on disclosure tone.
This section explains the joint impact of corporate governance and investment efficiency on disclosure tone, showing how the good and bad news information is affected by investment efficiency and corporate governance. This relationship was tested by introducing corporate governance (CG) and the combined variable (INVEFF*CG) to the regression models.
Table 6 shows that after adding
CG, INVEFF1*CGand INVEFF2*CG, the ANOVA test still reflects that both models as a whole are significant. The R-squared and adjusted R-squared values increased slightly by on average 1.5%, which indicates that
CG and
INVEFF*
CG do participate in explaining the change in the dependent variable DISTONE.
The results show that the second model’s independent variables INVEFF2, CG, and INVEFF2*CG have a significant impact on DISTONE, while the first model’s independent variables INVEFF1 and INVEFF1*CG have an insignificant impact on DISTONE and CG has a negative significant impact.
According to the hypotheses developed for estimating the scenarios which can explain the impact of
CG and the joint effect of
INVEFF*
CG on DISTONE (
Hussainey and Walker 2009), the second model’s variables can justify which of these hypotheses is accepted and which is not.
The values of the coefficients for
INVEFF2,
CG, and
INVEFF2*
CG are 0.991, −0.189 and −1.081 respectively, which applies to the fourth scenario explained before in the joint effect section. The sum of the coefficients on
INVEFF2,
CG, and
INVEFF2*
CG is significantly less than the sum of the coefficients on
INVEFF2 and
CG, shown in the following equation:
Therefore, the hypothesis is rejected and the fourth scenario applies which states that investment efficiency and corporate governance are substitutes. This substitution effect shows that investment efficiency and corporate governance provide related information and can replace each other’s roles (similar role).
Considering the direction of the relationships,
INVEFF2 has a positive significant impact on DISTONE while
CG and
INVEFF2*
CG have a negative significant one. The investment efficiency positive relation with disclosure tone is supported by the signaling theory, as
Foster (
1986) and
Inchausti (
1997) state that managers of profitable companies are encouraged to disclose more of this good news. Disclosure of such type of news is considered as a mechanism to attract investments, improve firms’ reputation and justify directors’ compensation (
Ross 1977;
Verrecchia 1983;
Campbell et al. 2002). Investment efficiently is a sign of great performance and so disclosing such information would achieve the mentioned benefits. Contrary to the results of corporate governance (CG) and VOLDIS stated above,
CG shows a negative impact on DISTONE. Firms having strong governance can be more inclined to voluntarily disclose more information (
Samaha et al. 2015;
Habbash et al. 2016) but not to disclose more good news in specific. Managers may prefer not to disclose good news as their firms’ competitive advantage might be threatened. The proprietary costs theory can explain this relationship showing that the costs of disclosing such good news may exceed its potential benefits (
Verrecchia 1983;
McKinnon 1984;
Feltham and Xie 1992;
Newman and Sansing 1993;
Darrough 1993;
Gigler et al. 1994). So, it can be concluded that introducing
CG negative impact had reduced
INVEFF2 positive impact reflected in the joint effect of
INVEFF2*
CG which has a negative impact as well on DISTONE, indicating that even if firms have improved investment efficiency, strong corporate governance effect still has a say in the selection of the type of voluntarily disclosed information.
As for the control variables, LEV, SG and MTB show a positive and significant impact on DISTONE with confidence levels of 90% and 99%, respectively, supporting the findings of (
Hussainey and Aal-Eisa 2009;
Ressas and Hussainey 2014). LOSS and AGE have a negative and significant impact on DISTONE with confidence levels of 99% and 90% respectively supporting the findings of
Rogers et al. (
2011), while contradicting the findings of
Aly et al. (
2018). The remaining variables FS, PROF, and DIV all show insignificant relationships with DISTONE.
DISTONE is the disclosure tone score; INVEFF is the investment efficiency; CG is the corporate governance score; INVEFF*CG is the interaction between corporate governance and investment efficiency; FS is the log of firm size measured by total assets; LEV is firm leverage; PROF is firm profitability; LIQ is firm liquidity; SG is sales growth; AGE is the log of firm age; LOSS is a dummy variable equals 1 for loss-making companies and 0 otherwise; MTB is market-to-book ratio; DIV is a dummy variable equals to 1 for firms paying dividends and 0 otherwise.