Now that we have a manageable set of legal variables, with a proxy for the basic foundational law (factor), corporate law (adr), and securities law
(priv), it is possible to try to separate their unique effects on financial market development. We can examine whether the basic foundational law is all that is required or whether securities law might be counterproductive, given a nation’s functioning basic and corporate law. The first step is a mul- tiple regression that uses these three legal values as independent variables.
The dependent variables are seven of the financial variables used above, with a focus on measures for equity markets and attention diversification among the World Bank financial measures (mcap/gdp and smv/gdp), the La Porta, et al.financial measures (mcap/s), the WCR survey measures (coc and adeqcap) and other financial measures (cash/price and mvol). In Table 5.9, we report the results with correlation coefficient and statistical significance level in parentheses. As our analysis becomes more refined, we also report the R2level and the number of countries in the analysis. In the following tables, we boldface all correlations significant at the 0.10 level.
Bear in mind that for the final two measures, a negative sign is preferable (lower ratio of cash flow to share price and lower market volatility).
Each of the full equations was statistically significant at the 0.01 level, and the R2levels are relatively high for such cross-country analyses, which inevitably have a high degree of random noise. The financial variables, coming from different sources and slightly different samples, give some confidence in the robustness of the consistent results.
The results in Table 5.9 are not entirely consistent and do not tell a per- fectly clear story. In general, the factor measure for basic foundational law is statistically significant and shows a typically positive effect, though it is significant and negative for the La Porta, et al. measure of mcap/s, a quite curious outcome. The corporate law measure of adr, by contrast is typically insignificant and frequently shows a negative sign, except for mcap/s. The securities law measure of priv, by contrast, shows the expected positive sign for the variables generally and demonstrates statistical significance for a majority of the independent financial variables.
Table 5.9 Regression of groups of legal variables on stock markets
mcap/gdp smv/gdp mcap/s coc adeqcap cash/price mvol
factor 0.453 0.394 0.352 0.842 0.663 0.237 0.592
(0.002) (0.009) (0.018) (0.000) (0.000) (0.193) (0.000) adr 0.066 0.028 0.409 0.107 0.094 0.168 0.086
(0.698) (0.875) (0.026) (0.349) (0.521) (0.482) (0.603)
priv 0.380 0.357 0.234 0.163 0.325 0.560 0.236
(0.034) (0.056) (0.185) (0.161) (0.032) (0.021) (0.159)
R2 0.409 0.348 0.419 0.754 0.577 0.434 0.431
n 38 38 33 34 35 24 37
Before drawing our preliminary conclusions, it is important to introduce a caveat on statistical significance, which should not be worshipped as a stan- dard for validity. Every single securities law measure was significant at the 0.2 level or better, which means that there is a better than 80 percent probability that there exists an authentic relationship between priv and the dependent variable of interest. Moreover, every full equation was significant at the 0.01 level, so one should not place undue emphasis on the statistical significance for each individual independent variable in each individual equation.
At this point, the data suggest that there is a strong positive relationship between the factor variable for basic foundational law and equity market development and efficiency, save for the single anomalous significant nega- tive association. There is a comparably strong positive relationship between securities law and equity markets, with more consistency than even that of basic foundational law. The evidence for the positive value of the antidi- rector rights measure of corporate law, though, is quite frail. The only pos- itive association of any strength was with mcap/s, a regression that seems a little dubious given the significant negative association with the factor vari- able for basic foundational law.
Table 5.9 above included no control variables. Given the multitude of differences among the nations and their potential effects on securities markets, consideration of some third factor control variables is necessary.
The choice of appropriate control variables, though, is a difficult one. For example, GDP is a commonly used control variable in many cross-country analyses but would be inappropriate here. Ample empirical research demonstrates that GDP is significantly affected by our independent vari- able of basic foundational law and significantly affected by the dependent financial market variables as well. With this high level of endogeneity, GDP is not helpful as a control variable, although it has sometimes been used.
Another potential control variable, the degree of corruption in the nation, suffers the same problem of endogeneity to the law.
Because the very nature of the law, especially the basic foundational law, affects all of society, endogeneity is pervasive and some studies have used relatively few or no control variables. Some of the control variables that have been used are inapplicable to our research. For example, settler mor- tality has been used as a control variable when studies have sought to measure the significance of a nation’s legal origin, but the origin variable has not been an important one in our analyses.
Given the relatively limited size of the sample, the number of control variables should be limited as well. One good control variable is ethno- linguistic fractionalization (ethn), which measures the ethnic and linguis- tic heterogeneity of a nation’s people.31This variable is largely exogenous of the legal and financial variables, yet has a plausible effect on the study;
as noted in Chapter 2, greater ethnic homogeneity tends to yield greater trust, which may yield greater investment in equity markets. Another vari- able that could have an important economic effect on markets is educa- tion, or human capital. Such human capital is commonly used in cross-country analyses and produces positive economic outcomes for a nation. Human capital is commonly operationalized as the percentage of the eligible populace enrolled in secondary education (secon).32For both these variables, data is available for the relevant nations. Table 5.10 reports the same multiple regressions as those of the prior table, but with these two control variables added to the equations. Once again, each of the full equations was statistically significant at the 0.01 level, and the R2 levels are good.
The introduction of control variables do not dramatically alter our results though they tend to reduce the associations with the legal variables.
The statistical significance of our legal variables in some equations slips somewhat, typically from the 0.05 level to the 0.10 level. The latter findings are still salient ones, because of the statistical noise associated with cross- country comparisons and the high significance of the full model. The intro- duction of the control variables does not substantially alter or change the directionality of any of the coefficients or patterns of results for the legal variables, and these variables have a very modest effect on the resulting R2 measures. The control variables were generally not significant (or even close to significant) independently.
While there was strong theoretical reason to believe that these control vari- ables would be significant determinants of our financial variables, they Table 5.10 Regression of groups of legal variables and control variables on
stock markets
mcap/gdp smv/gdp mcap/s coc adeqcap cash/price mvol
factor 0.450 0.423 0.202 0.700 0.528 0.773 0.681
(0.053) (0.084) (0.351) (0.000) (0.006) (0.449) (0.501)
adr 0.053 0.021 0.363 0.087 0.080 0.179 0.032
(0.762) (0.911) (0.050) (0.446) (0.593) (0.475) (0.827)
priv 0.341 0.334 0.323 0.136 0.301 0.564 0.113
(0.069) (0.091) (0.089) (0.252) (0.057) (0.031) (0.459)
ethn 0.133 0.086 0.216 0.002 0.016 0.084 0.201
(0.436) (0.634) (0.233) (0.985) (0.914) (0.671) (0.168)
secon 0.096 0.023 0.311 0.199 0.194 0.778 0.720
(0.693) (0.928) (0.208) (0.173) (0.335) (0.440) (0.002)
R2 0.420 0.353 0.460 0.773 0.592 0.452 0.586
n 38 38 33 34 35 24 37
proved significant in only one of the 14 possible instances. This illustrates the great statistical noise associated with any cross-country comparisons and the difficulty in finding significant associations at the more rigorous levels of sta- tistical significance. It is noteworthy that securities law (priv) was a relatively more important determinant offinancial outcomes than was secondary edu- cation, in five of the seven equations. Securities law was significant at the relaxed 0.10 level for five of the financial dependent variables (more than even for the basic foundational law). This finding of the relative importance of securities law is consistent with the results of a study of transition economies, which found that of different indexes for shareholder protection, the securities regulation scale was the only one to show significance.33
The association for securities law, and only securities law, on the cash/price variable is important. The prior studies showing that the basic foundational law increased the valuation of firms and equity markets may be due to the effect of legal systems on cash flow. For example, the greater valuation could be due to reduced risk of expropriation or greater oppor- tunities for sales. This effect is independent of that of the risk premium demanded by investors, to account for agency problems. Our results for this variable suggest that the latter concern is directly addressed by securities law, rather than basic or corporate law.
At this point in our analyses, it appears clear that the basic foundational law and securities law have an important, if not entirely consistent, effect in strengthening equity markets. The association for corporate law, at least the antidirector rights measure, is much less clear. The lack of positive effect for corporate law is somewhat surprising, given the theory and some of the past empirical findings on antidirector rights measures of corporate law.
The prior empirical research has not combined corporate law and securi- ties law, though. It may be that extensive securities law envelops the effects of corporate law, that the positive effect of corporate law disappears when broader antifraud protections of securities law are adopted. To consider this possibility, the next analysis considers only those nations whose secu- rities laws are below the median for the private enforcement variable of significance, to test for the effect of corporate law when securities laws are relatively weak. Table 5.11 reports the results of the multiple regression with control variables and absent the effect of securities laws.
The antidirector rights measure does relatively worse as a determinant of stock market size and efficiency in nations with weaker private enforcement of securities laws. Only the continued anomalous results for mcap/s are significantly positive, as they were for the full sample of nations.
Table 5.11 contains one other suggestive, potentially very significant finding. The correlations for our factor variable of basic foundational law lose much of their significance in nations with weak securities laws. This at
least hints at the prospect that it is the substance of securities laws that is most important for securities markets, rather than the substance of basic property and contract law. However, with substantive securities laws in place, it appears that the procedural attributes of the basic law (for example, judicial independence and efficiency), show a greater effect. Those legal pro- cedural variables would be of relatively little value to markets, absent some effective substantive law to implement. However, the significance of the foundational variables assumes importance in the presence of securities reg- ulation, which demonstrates their contribution under those circumstances.
With the limitation to nations with below average securities law, the sample size in this regression grows small, so one should be cautious about drawing any conclusions from the results. The findings certainly do not support the hypothesis that corporate law is more important in the absence of strong securities law. Indeed, our results show that it is securities law that has par- ticular value, comparable to that of basic foundational law.
The next analysis focuses on public enforcement of the securities laws. As noted above, the strong associations for securities law are associated with private, not public, enforcement of these laws. Theoretically, public enforcement should show some positive effect, though, if the securities laws themselves contribute to efficient capital markets. To consider this effect in a multiple regression with control variables, we repeat in Table 5.12 the analyses reported in Table 5.10 above, but use public enforcement (pub) rather than private enforcement for our securities law variable.
Public enforcement of securities laws is not a significant determinant of any of our financial dependent variables. However, it is noteworthy that the direction of the association is beneficial in every one of the measures of the public enforcement variable, which hints at some positive effect.
Table 5.11 Regression of groups of legal variables and control variables on stock markets in nations with weak securities laws
mcap/gdp smv/gdp mcap/s coc adeqcap cash/price mvol factor 0.005 0.029 0.309 0.560 0.227 0.681 0.350 (0.989) (0.936) (0.457) (0.001) (0.500) (0.140) (0.336)
adr 0.060 0.164 0.891 0.057 0.136 1.029 0.024
(0.806) (0.497) (0.028) (0.519) (0.557) (0.050) (0.910) ethn 0.155 0.689 0.430 0.012 0.356 0.608 0.260
(0.664) (0.065) (0.354) (0.925) (0.302) (0.243) (0.404)
secon 0.507 0.028 0.369 0.478 0.309 0.407 1.192
(0.212) (0.943) (0.507) (0.004) (0.379) (0.395) (0.005)
R2 0.400 0.416 0.541 0.945 0.620 0.659 0.610
n 18 18 12 14 14 9 16
Moreover, it is important to recognize that the public enforcement vari- able does not truly capture the value of the regulatory actions of the Securities and Exchange Commission. A central part of the SEC’s role is the establishment of regulations for disclosure and otherwise, which are enforceable through private actions. These regulations considerably facili- tate private enforcement and their value would be expressed through the private enforcement variable as well. The public enforcement variable cap- tures just that, enforcement, and not the rule creation function of the SEC.
It does seem fair to conclude that in the securities law enforcement context, it is the private right of action and not government prosecution which has the greatest financial value. This is unsurprising, because the remedy for private violations can be many billions of dollars of damages to the harmed investors, while public enforcement actions typically result in lesser sanc- tions, and because private enforcers may be more efficient in their choices of the actions to be pursued.