2.3.3 Results for Different Groups of Countries

We now break down the results to obtain a more granular picture of how ownership disclosure rules have evolved over time. Specifically, we are interested in the question of whether ownership disclosure rules have evolved differently in countries that belong to the same group, such as developed countries, EU countries or common law countries.

First, we examine the difference between developed countries and developing and transition economies. The category of developed countries is based on the World Bank's definition of High Income Economies.1 This covers 13 countries in our sample (Canada, Czech Republic, France, Germany, Italy, Japan, the Netherlands, Slovenia, Spain, Sweden, Switzerland, the United Kingdom and the United States). Table 4 contrasts the evolution of ownership disclosure rules in these developed countries against the 12 other countries in our sample.

Table 4: Developed Countries vs. Rest of the World

Table 4 shows that on average, developed countries have had more stringent ownership disclosure rules. The results are statistically significant. Notably, the difference between the two groups of countries decreased over time, as developing countries gradually lowered their disclosure thresholds towards 5%, the level at which most developed countries had their thresholds set throughout the entire sample period.2 This finding is consistent with Armour et al.'s finding of "a degree of overall convergence in the legai protection of shareholders", based on time series analysis of the development of shareholder protection as proxied by all 10 variables of the CBR Index.3

It is also interesting to see whether countries that form part of the EU display stronger convergence of ownership disclosure rules than the other countries in the sample. Table 5 contrasts the evolution of ownership disclosure rules in the 10 current EU Member States in our sample against the 15 other countries in the sample.

Table 5: EU Member States vs. Rest of the World

The decrease in standard deviation for EU Member States (-0.22) is considerably sharper than the decrease in standard deviation for non-EU Member States (-0.10). This confirrns that countries forming part of the EU display stronger convergence of ownership disclosure during the sample period than the other countries in the sample. Does this mean that changes in the ownership disclosure mies of EU Member States during the sample period were not exogenous, but merely reflected prevailing EU standards? Not necessarily. The 1988 Large Holdings Directive required EU Member States to set the disclosure threshold at 10%, while allowing Member States to set the threshold even lower (the Directive aimed at minimum harmonisation).4 Yet, table 5 shows that the mean score for EU Member States was 0.73. Coupled with the low mean standard deviation (0.14), this suggests that most EU Member States had their disclosure threshold set at 5% for most of the sample period. This can also be seen from Table 2. Thus, even though minimum EU standards applied during the sample period, many EU Member States exercised their discretionary power to set their threshold at a different (lower) level. From this perspective, changes in the ownership disclosure mies of EU Member States during the sample period were exogenous.

To be sure, the 2004 Transparency Directive, which superseded the Large Holdings Directive, required EU Member States to set the disclosure threshold at 5%. But the Transparency Directive needed to be implemented only in 2007. While it is possible that EU Member States anticipated the Transparency Directive by setting their disclosure threshold at 5% before 2007, the data indicates that this is unlikely. Table 2 shows that the bulk of EU Member States had their disclosure threshold set at 5% as early as 1995, nearly a decade before the Transparency Directive was adopted. This is different for the three EU countries (Latvia, Slovenia and the Czech Republic) that joined the EU only in 2004. In 1995, the beginring of the sample period, none of these had their disclosure threshold set at 5%. It is plausible that the subsequent lowering of the disclosure threshold by these countries was driven by a desire to conform to EU standards in preparation of their accession to the EU. This would be consistent with a study by Pistor et al. on the protection of shareholder rights in transition economies (including these three countries). They find that that "substantial efforts have been made to strengthen shareholder rights since the inception of economie reforms" and note that "European harmonization guidelines have unleashed what some commentators have called a tornado of legislative activities in the countries wishing to join the EU".5

Finally, it is interesting to see whether countries with a common law origin have, on average, more stringent ownership disclosure mies than countries with a civil law origin, as the legal origins theory proposed by La Porta et al. predicts they would.6 A major problem of the common law/civil law divide is that in reality, most legal systems are hybrids. For instance, South African law derives from both civil law and common law traditions; Japanese company law used to be based on the German model, but sine the 1950s has been heavily influenced by US law; Swiss company law is influenced by UK company law and, due to the influence of the EU, UK law itself has become more "continental".7 Therefore, more precise, but related, criteria need to be used than the mere distinction between common law and civil law countries. The most obvious one is "language", because most English-speaking countries are common law countries.8

Table 6 shows the evolution of ownership disclosure mies for countries that have English as at least one of its official languages (Canada, India, Malaysia, Pakistan, South Africa, UK, US) and the rest of the world.

Table 6: English-Speaking Countries vs. Rest of the World

Table 6 shows that on average, English-speaking countries had slightly more stringent ownership disclosure mies. But this was not consistently the case throughout the sample period, and over time the difference between the two groups appears to have become smaller. Be this as it may, the results are not statistically significant. This suggests that the common law/civil law divide is not helpful in explaining differences between countries' ownership disclosure tules.