Cost disease versus Leviathan explanations of rising government cost: An empirical investigation

In this paper we reexamine the apparently conflicting empirics of Borcherdi ng et al. (1977) versus those of Barry and Lowery (1984). The latter, desig ned to test the cost disease versus bureau voting power hypotheses on US Ci tibase annual data between 1947 to 1979, was retested for the longer period available through 1989. Second, and more importantly, we isolate and test for the presence of a second channel for the exercise of bureaucratic power . That channel is the bureau's ability to use its information advantage to capture a portion of newly generated government rents through higher person al benefits (such as higher salaries). Such an analysis (following West, 19 91) requires first that those factors generating new rents for government a ctually result in successful bureaucratic rent-seeking in the form of highe r compensation levels. In addition, the analysis requires that these "artif icial" increases in bureaucratic wages show up as significant determinants of the higher cost of providing government services. Incorporating a constr ucted Kau/Rubin variable into the Barry and Lowery database is then shown t o improve the predictive power of both the cost disease and bureaucratic po wer hypotheses for US annual data between 1948 and 1989.