Using real-time data to test for political budget cycles

We use real-time annual data on the fiscal balance, government current spending, current revenues and net capital outlays as published at a half yearly frequency in the OECD Economic Outlook for 25 OECD countries. For each fiscal year t we have a number of forecasts, a first release, and subsequent revisions. It turns out that revisions in the fiscal balance data are not affected by elections. However, we do find that governments spend more than reported before an election which provides support for moral-hazard type of political budget cycle (PBC) models: through hidden efforts the incumbent tries to enhance his perceived competence. We also find that governments had higher current receipts than reported before an election, which is in line with adverse-selection type of PBC models in which incumbents signal competence through expansionary fiscal policy before the elections.