mechanism is not available in a country that pursues a fixed exchange-rate regime or participates
in a monetary union such as EMU, and therefore has little or no monetary autonomy.
In order to test whether the impact of the change in the cyclically-adjusted primary fiscal
balance is greater in those countries that have no monetary policy autonomy, a dummy variable is
constructed which takes value 1 when the country is not engaged in any kind of fixed
exchange-rate arrangement (e.g. a peg, a monetary union or the former exchange-rate mechanism
of the European Monetary System) and zero otherwise. Information on official exchange-rate
arrangements is drawn from various issues of the annual IMF publication Exchange Arrangements
and Exchange Restrictions.104 This definition obviously ignores that fixed exchange-rate
commitments come in different shapes and forms. For example, the ERM contained countries that
pursued a “hard” peg to the Deutsche Mark (e.g. Netherlands) as well as countries that underwent
frequent depreciation. The lack of sophistication in the definition of monetary autonomy may
create a bias in the estimations, probably in the direction of not finding any effects. 105
All regressions in Tables 2 and 3 are then re-run including an interaction between the
fiscal adjustment variable and the dummy variable for monetary policy autonomy.106 The results
obtained for the variables of interest -namely the interaction term and the fiscal policy variables-
are reported in Table 4. The interaction term is correctly signed in all equations and statistically
significant in most of them. Taken at face value, the coefficient estimates imply that monetary
policy autonomy partially offsets the impact of the change in the (cyclically-adjusted primary)
fiscal balance on the propensity to undertake reforms.
104
Note that de jure exchange-rate regimes are not necessarily applied in practice (Levy Yeyati and Sturzenegger, 2001), even though this
issue is generally far more relevant for developing countries than for the developed ones considered here. As a result, some adjustments were
made. In particular, Austria was classified as having a fixed exchange rate regime over the full sample period even when it was not a formal
member of the exchange rate mechanism of the European Monetary System. This is because this country maintained de facto a fixed exchange
rate with the Deutsche Mark.
105. Another potential concern could be the asymmetry of the former European Monetary System. Indeed, it has been argued that German
monetary policy acted as a “Stackelberg leader” under this system, while other central banks had to follow suit in order to maintain fixed
exchange rates. In order to account for this possibility, the regressions presented in Tables 4 below were re-run assuming that Germany retained
its monetary policy autonomy under the former EMS. The findings were qualitatively unaffected, with the interaction term even being
marginally more significant in all equations.
106
These estimates also include the monetary policy autonomy dummy variable itself. Following a standard procedure, interactions
between variables Xit and Yit -here, the autonomous monetary policy dummy and the fiscal adjustment variable, respectively- are specified as
(Xit - X. )(γr-t - Y ), where X and Y are the sample averages of Xit and Yit , respectively. With this specification, the
coefficients of Xit and Yit -and therefore the estimated effect of the fiscal adjustment variable- have a straightforward interpretation, i.e. they
measure the direct effects of these variables for an average, hypothetical country. This is because the interaction term is equal to zero at the
sample average.
199