overcome the “status quo” bias which in normal times can prevent welfare-enhancing
reforms from being implemented.91
- Trade openness:
Due to “crowding-in” effects already discussed, it may be politically easier to carry out
structural reforms in small open economies than in larger ones. Furthermore, high trade
openness and the resulting high labour demand elasticity in small open economies may
limit the ability of insiders to set wages above market-clearing levels, resulting in lower
rents and thus lower public support for existing institutions (Saint-Paul, 2004).
- Reforms undertaken in other fields:
Some of the specifications estimated below also control for the fact that implementing
reforms in certain fields may pave the way for reforms in others. As argued above,
reducing rents in one area may undermine the support for structural policies in other
areas aimed at capturing such rent. Learning effects may also play a role, i.e. public
support for structural reform may build up as the benefits of past reforms become more
visible. Moreover, policy packages may be easier to implement than isolated reforms
because they spread gains and losses more evenly across population groups and/or
facilitate the compensation of losers.
4.2 Data construction
The econometric analysis is based on an annual database of major structural reforms in
21 OECD countries over 1985-2003, constructed for the purpose of this paper. Sources and
methodology are presented in detail in Annex 1. Compared with the descriptive evidence
presented in Section 3, the focus here is as previously mentioned on major reforms only, and the
coverage of policies is restricted to those areas that allow for quantitative description over an
extended period. Hence, only five fields are considered for which available quantitative indicators
of the policy stance make it possible to spot major reforms: unemployment benefit systems,
labour taxes, employment protection legislation, product market regulation and retirement
schemes.
In each of the five policy fields, a major reform is assumed to be undertaken when the
change in the corresponding quantitative policy indicator (e.g. the OECD summary measure of
replacement rates in the case of unemployment benefit systems) exceeds two standard deviations
of the indicator’s sample average.92 This approach yields one reform indicator for each policy
area, which takes the value 1 when observations - i.e. pairs (country, year) - correspond to a
major policy reform and 0 otherwise. Cross-country patterns of major reforms thereby identified
appear to be roughly consistent with those presented in Section 3 for different types of reforms
91.See for instance Drazen (2000).
92. In the field of product market regulation, there has been a strong general trend toward liberalisation. To distinguish particular moves
toward liberalisation from this general trend the indicator of product market regulation was initially de-trended. Without this procedure, which
admittedly is somewhat ad hoc, product market reforms would have weighed very heavily in the data sample (see Annex 1).
190