XREAP2007-14
alleviate the omitted variables problem. Fiscal rules have indeed been used gauge the sensitivity of
some fiscal policy indicator to the cycle yt. There are cyclical variations in the surplus because of the
workings of automatic stabilisers. In economic booms, tax revenues rise as tax bases grow. In contrast,
spending on transfers and unemployment benefits will fall. Wibbels and Rodden (2006) examine the
cyclicality of central versus regional fiscal policies in the US and Germany. There are some alternative
‘political’ explanations for growing regional public debt, such as the political affinity of regional
governments and the federal government, the political party in power, coalition formation or the size
of the region, among others. Berger and Holler (2007) find that common economic factors and state-
specific economic conditions are more important determinants of state fiscal performance than these
political factors in Germany. There are also few robust findings for the effect of political factors in US
states (Wibbels and Rodden, 2006). Hence, as political factors do not seem of prime importance, we
rewrite the fiscal rule (1) accounting for cyclical developments only:
st = ρbt* +αyt +εt . (6)
We specify all fiscal variables in ratios to GDP in (6). We will take as the initial debt stock bt* the
lagged debt ratio to GDP. As for the cyclical indicator, we take the growth rate of GDP, as proxied by
the first log difference of real GDP.
In first instance, we simply compare the debt sustainability response for the different levels of
government. We test sustainability of fiscal policy on a baseline fiscal rule as (1) and (6) for the
general government with OLS. We then disaggregate this consolidated budget response into the
contribution of either the central government or regional fiscal policies. We provide a fiscal rule
estimate for the central government and for each regional government separately. If each government
tier controls its public debt, no problems of soft budget constraints arise. Alternatively, we may
compare the average debt response of all regional governments to the one of the central government.
We apply panel OLS estimates of the fiscal rule for all regions jointly. There are important cross-
dependencies in state budgets due to economic and institutional links (Case et al., 1993). All regions
share a common monetary and federal fiscal policy. Also, changes in federally mandated expenditures
influence state budgets. Moreover, the mobility of tax bases imposes some implicit constraints on
revenues. Apart from these common economic factors, there could be other idiosyncratic
characteristics of each region that are controlled for by the introduction of region fixed effects.