5.6.1 Hypotheses tested

Below, the hypotheses as formulated under section 5.3.3 are tested and analysed, starting with the time hypothesis. It is of importance that, in their formulation, the hypotheses use the levels of compliance whilst the tables in the bivariate results below show the numbers of deviations within a certain level of compliance. Taking time hypothesis H1 into account, it is expected that the table will show negative relations (the longer the principle has been applicable, the fewer deviations), whilst the hypothesis mentions a positive relation between the level of compliance and the number of years the comply or explain principle has been applicable to the company.

A distinction is made between the period of time the comply or explain principle has been applicable in a country (varying from 1 to 15 years) (AgePrinciple) and the financial year the corporate governance statement dates from (2004 until 2007) (YearAccount). Table 5.6 below shows negative significant (at the 0.01 level) relations between the number of deviations for the levels of compliance I.a, I.b, II.a, II.b, III.a and III.b (CompI.a, CompI.b, etcetera) for all 711 corporate governance statements and time (AgePrinciple): the longer the comply or explain principle has been applicable, the lower the number of deviations is, hence for this part the time hypothesis is consistent. For YearAccount (the financial year the corporate governance statements date from (2005-2007)) the correlations show no significance.

To conclude, with respect to the period of time the comply or explain principle has been applicable in a country (varying from 1 to 15 years) (AgePrinciple) the time hypothesis is consistent: a positive relation between the level of compliance (I to III a and b) with corporate governance codes and the number of years the comply or explain principle has been applicable to the company. This is not the case for the financial year the corporate governance statement dates from (2005 until 2007) (YearAccount).

The judicial corporate governance arrangement involves the variable judicial corporate governance arrangement (pure self-regulation, supported by non-statutory norms, facilitation by statutory rules and meta-regulation), which is a nominal variable. And since the levels of compliance (I to III a and b) (CompI.a, CompI.b, etcetera) are a ratio variable, an ANOVA-test is performed (see table 5.6a below) (Van Dalen and De Leede 2002, p. 205). The ANOVA has a significance of 0.001 or lower and the F-value is between 5.831 and 21.707, implying that the overall ANOVA-test is significant (Field 2005, p. 351). Hence, it can be concluded that the population means are not all equal: differences in the level of compliance between the corporate governance arrangements occur as can also be seen under the descriptives showing the means at a 95% confidence interval for mean. Since the one-way ANOVA test is significant, post hoc multiple comparisons by means of the least-significant difference tests (LSD) are performed and various significant mean differences can be seen between the corporate governance arrangements for the different levels of compliance (Van Dalen and De Leede 2002, p. 327) (Field 2005, p. 341). For the purposes of better understanding, only the significant results from the test are shown in the conclusions table below. Annex IV shows the more specific ANOVA-output, descriptives and the least-significant difference tests.

The ANOVA-tests above show that differences in the level of compliance between the corporate governance arrangements occur. The corporate governance arrangement hypothesis is not rejected, the manner in which the comply or explain principle is embedded in a national corporate governance system does matter. For all the different levels of compliance, the companies for which the comply or explain principle has been laid down in non-statutory norms score best (lowest means thus fewest deviations); they have the highest compliance rate compared to the other corporate governance arrangements. Meta-regulation or statutory rules alternately score lowest for the different levels of code compliance (highest means thus most deviations) (with the exception of self-regulation for level of compliance Ill.b). When the quality of the explanations is taken into account, the principle laid down in statutory rules scores significantly better. It was expected (see section 5.3.3) that the stronger a comply or explain principle is embedded in a country the higher the level of compliance and/or quality of explanations is, since a strong embedding shows faith in and consensus about the principle and likely more compliance. Therefore, one would expect corporate governance arrangement meta-regulation and the comply or explain principle laid down in statutory rules to achieve the highest compliance rates. Apparently the comply or explain principle laid down in non-statutory norms (e.g. listing rules) suffices for the highest compliance rates.

As shown in table 5.3.3, extensive research is performed on code compliance related to firm size which is considered a common predictor for code compliance (Dedman 2000) (Pollock, Fischer et al. 2002) and a positive relation is confirmed by several studies. Levels of compliance I to III a and b (CompI.a, CompII.b, etcetera) are measured for the number of deviations. With respect to the market capitalisation (Compartment) a distinction is made in three compartments (A, B and C), i.e. the large caps, mid caps and small caps.

Levels of compliance I.a and I.b show no significant relation with market capitalisation. These levels involve application, as stated by the companies, of all the provisions (I.a) and of the comparable code provisions (I.b). For the remaining levels of compliance (II.a, II.b, III.a and III.b) a positive significant (at 0.01 or 0.05 level) relation with market capitalisation exists. Hence, the smaller the company, the lower the level of compliance (the more deviations). With respect to code compliance as stated by the companies themselves (I.a and I.b), no significant relation can be seen, unlike for the levels of compliance in which the quality of explanations is taken into account. So with respect to code compliance in general size does not matter, but when taking the quality of the explanations into account the larger companies score better than the smaller ones.

Next to market capitalisation the stock exchange index is also a commonly used variable in the size hypotheses (Andres and Theissen 2008) (Von Werder, Talaulicar et al. 2005). Since per country the indexes and their admission criteria differ, a country's main stock exchange index is coded as 1, first to main stock exchange as 2 and second to main stock exchange as 3 (IndexType). For all levels of compliance, a positive relation (at 0.01 or 0.05 level) with the IndexType can be seen, hence implying that the smaller the company (the higher the code for the stock exchange), the lower the code compliance (the more deviations) and thus confirming the stock exchange index hypothesis.

For hypothesis 4, per level of compliance thirteen variables are used for the thirteen different categories in which code provisions (hence deviations) can be subdivided (see table 5.5.3a above). For the b-levels of compliance - the comparable code provisions of the five countries under research - the categories compliance, management board conflict of interests, management board independence, supervisory board conflict of interests, one-tier board structure and financial reporting are excluded from the calculation since they are not part of the comparable code provisions (see section 5.5.3 and table 5.5.3a). For all the levels of compliance it is counted how many times a deviation within a certain category occurs within the measured level of compliance. By that means ratio variables are construed to perform a correlation analysis (Pearson) with the level of compliance.

Obviously positive relations can be seen; the higher the number of deviations, the more scores occur in the different categories. Therefore, the categories of code reasons that have the highest relation with the level of compliance are of interest, since they typically occur in the case of low compliance. Hence, for level of compliance I.a the deviations mostly concern the board's remuneration, the supervisory in general and the shareholders. For level of compliance II.a the categories the board in general, the board's remuneration and the supervisory board's remuneration score high. In level of compliance III.a the categories the board in general, the board's remuneration and the key committees show high correlations. With respect to level III.a it must be taken into account that this concerns the deviations with insufficient explanations. Apparently the explanations in the case of deviations from code provisions on the key committees are really insufficient. The scores for the b-levels of compliance (the comparable code provisions for the countries under research) are similar, implying that the provisions that are more culturally and nationally specific are complied with properly. Although in section 5.5.5 and 5.5.6 the most explained code provisions per se are reviewed, the same topics do occur; these are mainly management board issues.

For hypothesis 5 (on the expected relation between the level of compliance CompI.a, I.b, Ill.a and III.b),1per level of compliance sixteen variables are used for the sixteen different categories of reasons for deviations (Reasonl-CompI.a etcetera) (see Annex II and section 5.5.7). It is counted how many times a certain reason occurs within the measured level of compliance. By that means ratio variables are construed to perform a correlation analysis (Pearson) with the level of compliance. Obviously positive relations can be seen; the higher the number of deviations, the more reasons for deviations occur. The reasons provided that have the highest relation with the level of compliance are of interest, since they typically occur in the case of low compliance. Hence, for level of compliance I.a, reason 13 (having own regulation), reason 10 (non-compliance because of sort/structure/character/interests of the company) and reason 6 (respect existing agreements/contracts) show the highest correlation rate. Forlevel ofcompliance III.a (the deviations withinsufficientexplanations: no explanation, general and inline), reason 1 (no explanation/reason), reason 10 (non-compliance because of sort/structure/character/interests of the company) and reason 13 (having own regulation) show the highest rate. Since this level III.a shows the deviations insufficiently explained, reason 1 (no explanation/ reason) obviously scores high.

With respect to level of compliance I.b (applied for comparable code provisions of countries under research), again reason 13 (having own regulation), reason 10 (non-compliance because of sort/structure/character/interests of the company) and reason 6 (respect existing agreements/contracts) show the highest correlation rate. The same applies to level of compliance III.b as to level III.a: reason 1 (no explanation/reason), reason 10 (non-compliance because of sort/ structure/character/interests of the company) and reason 13 (having own regulation) show the strongest relation.

To sum up and taking the 16 categories of reasons provided for deviations into account, the explanations given that correlate best involve the more general categories (1, 6, 10 and 13) of the ones distinguished. This implies that the lower the compliance, the more general the explanations, hence indeed improvements in the quality of explanations are possible.