Estimation of a fiscal policy rule for EMU countries (1984–2005)

Abstract

The primary objective of this article is to estimate a fiscal policy rule for each of the EMU member States from 1984 to 2005 in order to know if there has been a systematic response of the cyclically adjusted primary balance to output gap and debt level variations. Also, we aim to discover whether the change in the fiscal framework, which took place after 1992 has had a substantial impact on the fiscal policy applied. The principal novelty is that the estimation is performed simultaneously by means of a seemingly unrelated regression estimator model. We are thus able to obtain different coefficients for each country, while developing possible correlations between national fiscal policies, which would reveal the existence of common factors. The results provide clear evidence of a structural break in the rule after the introduction of the new fiscal regulations and, as the hypothesis of equality in the national coefficients of the rule is clearly rejected, reveal a need to consider specific national factors.

Notes

¹ Taylor (1993).

² Seemingly Unrelated Regression Estimator.

³ This analytical procedure is followed by the European Commission (2006) and in all the previous editions of Public Finances in EMU. Larch and Salto (2003) offer a discussion of the utility of this indicator to measure the sign of discretionary fiscal policy.

⁴ Calmfors (2003) says that ‘the Ricardian equivalence results require very restrictive theoretical assumptions which are not likely to apply in reality’. Blinder (2004) summarizes the most common criticisms of this hypothesis.

⁵ Burnside et al. (1999) and Fatás and Mihov (2000) point out that in the United States fiscal shocks also produce changes in output, consumption, investment and employment. Furthermore, Fatás and Mihov (2002), based on time and cross-section data pertaining to 51 countries, find that the relation between the magnitude of output changes and the use of discretionary fiscal policy is statistically significant, in the sense that countries with larger governments suffer less economic cycle volatility. On the other hand, Perotti (2002) studies the effects of fiscal measures on GDP, price and interest rate growth in 5 OECD countries (United States, Germany, United Kingdom, Canada and Australia) using autoregressive vectors. First, the analysis reveals that the estimated effects of fiscal policy on the GDP tend to be positive, although minor, and that public expenditure multipliers (positive and mostly less than one) are usually larger in absolute values that those obtained as tax multipliers (negative). Hemming et al. (2002) present a selection of this empirical literature, showing that the estimated mean tax multiplier is 0.5.

⁶ A Chow test and other statistics have confirmed us such a structural rupture in 1992. See Evidence of structural change.

⁷ See Green (1999), Chap. 15.

⁸ European Commission (2006).

⁹ The data used in the study are annual and, due to their low frequency, there are unlikely to be important noncontemporaneous correlations. The cross-correlation functions calculated from the GLS residuals of the different countries have confirmed this.

¹⁰ Novales (1993), p. 274.

¹¹ Although there can be heteroscedasticity between the different countries.

¹² m is the number of equation in the model.

¹³ Green (1999), page. 571.

¹⁴ With a value of one from 1992 on and zero beforehand.

¹⁵ The McElroy R2 corresponds to the following expression (see Green, 1999, p. 585):

¹⁶ The fact of this coincidence does not imply that the new rules are the only explanation for the different behaviour observed in the fiscal authorities. This institutional change probably encouraged more budgetary discipline in Europe, reducing the weight of the public sector, which has a broader territorial dimension.

¹⁷ The highest in the period analysed.

¹⁸ See Bohn (1998) and Ballabriga and Martínez-Mongay (2005).

¹⁹ The SGP does not literally establish that discretionary measures cannot be applied to stabilize the economy, but it is implicit in its philosophy when it states that governments must maintain equilibrium or a superavit position in the medium term. This approach is defended, for example, in European Central Bank (2004), and in relation to the relative role of discretionary measures and automatic stabilisers, Buti and Van der Noord (2004) say that: ‘While the potential usefulness of fiscal stabilisation is being re-considered, the “heritage” of the debate in the 1980s casts a strong scepticism over the use of discretionary fiscal action to fine tune the economy. (…) The use of discretionary fiscal policy for stabilising purposes should be confined only to exceptional situations’.

²⁰ These effects will be significantly different from zero whenever the coefficients are. Since the accumulated effect is β _{j i} /(1 − β _{5 i} ), when the test is performed, the null hypothesis (the estimated coefficient is zero) will only be accepted when the same hypothesis, but considered with the numerator coefficient, is not rejected.