Figure 2 shows that regulators are overwhelmingly independent in the electricity, gas and
telecommunications sectors, while regulators tend to be part of government agencies or ministries
in transportation and in the water industry. Two measures of regulatory independence were
constructed. The first was whether the regulator was independent of the government. The second
measure was a gauge of de facto independence assessing the extent to which the executive can
influence the regulator’s decisions and operation (using responses from the OECD infrastructure
questionnaire).19 The two variables are not significantly different. The type of price regulation
regime was interacted with the regulator’s independence variable to see whether
complementarities between them matter for sectoral investment.
As questionnaire responses only provide a snapshot of regulation in late 2007 and early 2008, the
data on the type of regulatory regime do not have a time dimension. For this reason, only cross-
sectional regressions are carried out in the analysis (with sectoral investment being the dependent
variable). The regulatory variables include variables capturing the presence of incentive
regulation (inc), the absence of any price regulation (dereg) (unregulated or deregulated), the
presence of an independent sector regulator (ri), and an interaction term of incentive regulation
and regulatory independence (inc*ri). The other explanatory variables are lagged investment
( It_1), long-term real interest rate ( r ) deflated with sectoral deflators, and variables capturing the
general regulatory environment: entry barriers ( entry )20 and public ownership ( po ).21,22 As a
result, a linearised version of the following investment function was estimated:
I = f ( It 1, r, entry, po, inc, dereg, ri, inc * ri )
(1a)
Investment data are obtained from the OECD’s STAN and SNA (Sectoral National Accounts)
databases. These two databases have two serious shortcomings. Firstly, the level of
19. The de jure measure of independence was adjusted for the possibility of the executive overruling
the regulator’s decisions. The de facto measure of independence was calculated as the average of
the de jure independence and the possibility of overruling.
20. The definition of entry barriers differs across sectors to reflect sector-specific characteristics. For
more details, see Conway and Nicoletti (2006). The data can be downloaded from:
http://www.oecd.org/document/32/0,3343,en 2649 34323 35791136 1 1 1 1,00.html
21. Entry barriers and public ownership were used in Alesina et al (2005) who performed a panel
data analysis of sectoral investment.
22. The dependent and explanatory variables are averages for 2001-2006, except the two variables
capturing the regulatory regime (inc and dereg). In addition, the explanatory variables are
included with a lag of one period (using averages for 1995 to 2000), except inc and dereg.
14