Dit artikel is alleen beschikbaar in het Engels.

Political Forecast Cycles – door Frank Bohn

In 2008, Boylan showed in an empirical study based on 49 US states and the period from 1988 to 2004 that state revenue growth forecasts are substantially larger in election years. This allows the state government to expand the budget before an election in order to appear more potent which may be taken as more competent by some voters; hence re-election chances of the incumbent can be increased. Overly optimistic revenue forecasts allow the government to circumvent the constitutional balanced budget constraints which is in place in most US states; after the election, the government feigns surprise of the “lower than expected” revenue.

Against this backdrop, Francisco Veiga (Minho University, Portugal) and I wrote a paper on “Political Forecast Cycles” which was published in the European Journal of Political Economy in 2021. We wanted to see whether Boylan’s results could be replicated under other circumstances and whether they follow a general logic. This means that we wanted to answer two questions.

First, we wanted to see if we can capture the aforementioned mechanism analytically. Can the idea of a forecast manipulation be included in an existing political budget cycle model? Such models are used to explain why governments use fiscal manipulation in election years to increase their re-election chances. Basically, this works because uninformed voters mistake the fiscal activity of the government for competence. As a result, the budget is expanded in election years, but cut in off-election years. Hence the political process produces a budget cycle.

Second, we wanted to see, if the aforementioned empirical results are also relevant in other countries and on different levels of government. We could use a dataset of Portuguese municipalities we had already employed for answering other research questions. Municipalities are the highest subnational government level in mainland Portugal, and the second in the autonomous regions of Azores and Madeira, which have regional governments. The dataset is ideal because all municipalities are subject to the same laws and regulations, have the same responsibilities for public service provision, and have the same institutional structure.

Here are the results of our paper. We can analytically show the aforementioned mechanism in a political budget cycle model; it is optimal for an incumbent government to try to improve re-election chances by using manipulated revenue forecasts, thereby producing political forecast cycles. In our regression analysis we can show that forecasts are indeed increased in election years. We then ask under which conditions the forecast manipulation is more or less prevalent. We derive three additional theoretical results for which we find support in our empirical analysis. The extent of manipulations is reduced when (i) the winning margin is expected to widen, i.e. when the incumbent is sure to win; (ii) the incumbent is not re-running; and/or (iii) the share of informed voters (proxied by education) goes up.

Given that we can show a theoretical mechanism and complement findings for US States by similar findings for Portuguese municipalities, there is a good chance that political forecast manipulation is a much more widespread phenomenon.

By Frank Bohn, January 2022