6. Empirical Results
OLS regressions results obtained for the 13 countries with three sectoral aggregates each show
that lagged investment, barriers to entry and the interaction term including regulatory
independence and incentive regulation are significantly correlated with contemporaneous
investment (Table 1).23 Barriers to entry are found to influence investment negatively. The
coefficient estimate of -0.046 indicates that a one step reduction in the barriers to entry indicator
would be associated with a 4.6% increase in the investment ratio.24 The combination of regulatory
independence and incentive price regulation has a significant positive effect on investment,
though when taken separately the two policy variables do not have any significant effect on
investment. This suggests that a right policy mix is important in determining investment.
Table 1 reports the main results for the different regulatory variables using Bayesian model
averaging. If both country and industry fixed effects are used, the most robust findings are that for
the whole sample and all sub-samples (where one country at a time is dropped) posterior
inclusion probabilities are always higher than 0.50 in the case of entry barriers and the interaction
term combining incentive regulation with regulatory independence. Moreover, the size of the
estimated coefficient is similar to that of the OLS estimate. Table 1 also indicates that in the cross
section under consideration public ownership is not an important driver of investment rates and
that the absence of price regulation tends to lower investment. Yet this latter result is not robust to
changes in country coverage, since it vanishes once the United States is dropped from the sample.
The size of the coefficient estimates for the different subsamples suggests that on average a 1 step
change in the interaction term (e.g. a change from 1 to 2) would induce an average increase of the
investment ratio by 4.9 percentage points. The lowest and highest coefficient estimates give a
lower and upper bound of the increase in the investment ratio of respectively 1.6 percentage
points and 6.5 percentage points. A move from the bottom to the top in the observed distribution
of the interaction term -- from 0 to 4.5 -- would on average increase the investment ratio by
22.2 percentage points, with the lower and upper bounds being 7.2 percentage points and
29.3 percentage points.
23. The estimations use both country-fixed effects and then country- and industry-fixed effects as
well as the explanatory variables lagged one period. An exception is the variable capturing
incentive regulation because observations are only available for the most recent period.
24. This finding is broadly in line with results reported in Alesina et al. (2005). Using panel data
estimation methods, Alesina et al. (2005) find that the coefficient estimate for the barriers to
entry variable is around -0.01.
17