Notes
1 Furthermore, the Cukierman data refer to a small sample of countries only.
2 Their results on openness confirm earlier findings of Romer (1993), who argues that monetary surprises
cause the real exchange rate to depreciate, and since real depreciations has the most harmful effects in
more open economies, the benefits of unexpected inflation are a decreasing function of the degree of
openness. In the absence of binding pre-commitments, monetary authorities in more open economies are
therefore expected to expand less on average.
3 See Beck et al. (1999). The variable we use is labeled STABNS2 that is defined as the percent of
veto players who drop from the government in any given year. Veto players are the president, the
largest government party, and the largest party in the Senate; for parliamentary systems, veto players
are defined as the Prime Minister and the biggest three coalition members. If there is no legislature, an
unelected legislature, only 1 candidate, or 1 party to choose from during elections this index is based
only on changes in the chief executive.
4 Given the limited variation over time in the TOR variable, a panel data approach is not helpful in the
present context.
5 Part of this data set (referring to 1980-89 only) has already been used in De Haan and Kooi (2000).
6 The underlying data (on an annual basis) are available upon request.
7 Due to the possibly large degree of multicollinearity in these powers, Lin, White and Granger (1993)
suggest using reconstructed powers on principal components, excluding the first and largest one.
8 The test statistic for row 1 in table is e.g. 1.947 (significance level: 0.15).
9 See table A2 in the Appendix for further details.
17