POLITICAL OPINION AND THE TRENDS IN ROMANIAN GOVERNMENT EXPENDITURES.

• Author: Marian, Cosmin Gabriel

1. Introduction

   Democratic representation entails a two-directional exchange--from mass to elite through the process of electing and replacing important officials, and from elite to mass through government policies that, to some considerable degree, should respond to citizen needs and wants. This paper examines Romanian post-revolution national government expenditure trends, compared to the preferences expressed by Romanian citizens as best we can determine from available public opinion polls. The finding, despite limited data, is that government spending and popular preferences show little correspondence. As reflected in high levels of expressed discontent, it is doubtful whether the Romanian public is receiving an appropriate return for its vote.

   A critical question for citizen deliberation is whether the country expends the right amount of public funds, allocated in the right proportions across the various substantive tasks of government. Romania shows clear linear expenditure trends over time, largely independent of the government or political party in power. The Median Voter Theorem suggests that this might be an anticipated and legitimate democratic outcome given the incentive for politicians to follow movements in majority public opinion or else risk electoral defeat. Yet the theorem, we find, does not succeed in explaining the direction of Romanian budget trends, indicating potential problems regarding democratic responsiveness.

2. Theoretical problem and literature review

   The most striking fact about Romanian national expenditures is the strength of linear trends across the time series and the relative insignificance of the particular government or political party in power in deflecting them. Nor is there an effect resulting from the election cycle or from reforms in electoral rules (King and Marian, 2016). The central finding is presented in Table 1.

   The regression model used, simplified for current purposes, is:

   Expenditure/GDP = Intercept + Year + Right-Left Party

   The regression dependent variable is Expenditure by the national government as a share of GDP, between 1991 and 2012, overall and by substantive budget function. Data are yet unavailable for the period after 2012. Total expenditures have been calculated from the General consolidated budget (issued by the Ministry of Public Finances --Ministerul Finantelor Publice, undated), which includes the State Budget, Local Administrative Budgets, and those Special Budgets for earmarked funds that (in whole or in part) generate their own revenues. Of the substantive expenditure categories, pensions are found in a Special Budget, the only one for which there is consistent data for the entire period under consideration. The remaining substantive categories are component parts from the State Budget, as reported in the Annual Statistical Yearbooks (issued by the National Institute of Statistics--Institutul National de Statistica, undated).

   The regression independent variable Year is operationalized such that 1991 = 1 and 2012 = 22. The variable is intended to capture any linear trend in spending, upward or downward from the estimated intercept.

   Right-Left Party, the other main regression independent variable, is a dummy variable coded 0 when the party of the left-center (Social Democrats) dominates the government and selects the Prime Minister, and 1 when a party of the right-center (Democratic Convention, Liberals, Democratic-Liberals) dominates the government and selects the Prime Minister. Because the annual Romanian national budget is approved in late December, around the same time as periodic Romanian parliamentary elections, any incoming government exercises minimum budgetary control during its first year in office. Thus the Right-Left party variable, as is common in studies of this sort, is lagged t+1, such that the first Nastase budget year is 2002 and the first Tariceanu budget year is 2006.

   As can be seen in Table 1, there is no long-term linear trend in total Romanian expenditures as a share of GDP, nor for the Pensions fund or for the small category for Culture, Religion, Sport and Youth Activities. There are statistically significant downward Expenditure/GDP trends for the State Budget (implying a reciprocally increasing share over time for Local Budgets), and for the substantive categories Education, Health, Economic Development and Energy, Agriculture, and Defense. There are statistically significant upward Expenditure/GDP trends for Public Assistance, Transportation and Communication, Public Order, and General Public Administration. The only category for which political party matters is Education, where governments of the right-center, while maintaining the general downward trend, do so to a somewhat lesser degree than governments of the left-center.

   The conclusion is that linear trends predominate across the entire post-revolution period, far more than political variables. As seen in Table 1 (and supported by our wider study of expenditure patterns), there are but minimal observed politically-inspired deviations from the linear trends, and even these are often inconsistent. It is a result at considerable variance from popular impression, anticipated from the intense partisan controversies that beset Romanian politics. There is, however, a major proposition in the social science literature that predicts just this sort of outcome: the Median Voter Theorem (Downs, 1957; Black, 1958).

   The theorem is quite simple. Rational, utility-maximizing individuals with single-peaked preferences will always select the policy option closest to their own position over any policy option further away. Then assume two competing political actors, for example the government and its opposition, each of which wishes to retain/obtain power. Each therefore will select the position anticipated to bring it the most support, which is always the position of the median voter, M; any other position would be defeated by majority rule. Imagine, for instance, that some actor has a preference for location L, minimally to the left of the median; all else being held equal, L would be viewed as inferior to M by all individuals to the right of the median, plus by those sitting exactly at the median. The implication is that both competing actors, despite their initial differences, will adopt the same policy position and that the opinion of the median voter always prevails, regardless of the spread of individual opinions. Furthermore, over time, if the position of the median voter shifts, political actors have incentive to respond rapidly, and outcomes thus will move in the direction desired by the median voter independent of which of the actors happens to be in power.

   Normatively, the theorem predicts policy outcomes that are highly responsive to aggregate citizen opinion as the consequence of free democratic competition. The empirical results, however, are mixed. The traditional approach from economists is to address the problem objectively, estimating the median voter's demand function for public goods in terms of the marginal tax price at the 50th percentile income level, comparing across localities (Barr and Davis, 1966; Bergstrom and Goodman, 1973). Despite the increasing sophistication of such studies, they have been criticized methodologically for faulty inferences and conceptually for...