Divided Government and the Passage of Partisan Legislation, 1947-1990
Thorson, Gregory R., Political Research Quarterly
Recent literature has provided some evidence that the presence of divided government does not affect the amount of significant legislation passed by Congress and enacted into law (Mayhew 1991). In this article, I argue that although there may not be a difference in the absolute number of bills passed during unified and divided periods, there nevertheless may be an important difference in the formation of coalitions' during divided and unified periods. Specifically, I argue that party unity votes that favor the dominant party2 are more likely to form on final votes of passage during periods of unified government. I use regression analysis and probit analysis to determine if there is empirical support for this hypothesized difference in coalition formation. I find that the presence of unified government significantly increases the likelihood that a bill will pass through Congress with a party unity vote favoring the dominant party. The estimates are statistically significant for both the House and the Senate.
A great deal of literature, both in scholarly journals and in the popular media, suggests that the presence of divided government inevitably leads to a state of policy gridlock (Sundquist 1988). Such a belief is predicated on the assumption that for effective governance to occur in democracies, political parties must be both sufficiently strong and cohesive as well as have enough strength in each stage of the legislative process to allow them to overcome institutional barriers. Other authors have argued that united government was a necessary, but not necessarily sufficient, condition for legislative productivity (Ripley 1983; Key 1964). Yet a leading view among political scientists, based largely on work done by David Mayhew, is that divided government does not have a substantial effect on the legislative process. Referring to Ripley and Key, Mayhew argues "that the above claims are wrong, or at least mostly or probably wrong" (Mayhew 1991: 3).
This research develops and tests a hypothesis that predicts that although divided government may not produce "gridlock" in the traditional sense, it never theless has a substantial impact on the formation of coalitions. Specifically, I argue that partisan coalitions are much more likely to form on final votes of passage for significant legislation that were enacted during unified versus divided periods of government. Thus, although unified periods may produce the same amount of significant legislation as divided periods, the nature of the coalitions, and by implication, the content of the legislation, has been affected. Within this context, although divided government does not produce the expected "gridlock," it nevertheless has a significant effect on the legislative process.
The "problem" of divided government has become increasingly important due to the unusual frequency with which the postwar period has experienced the President and the Congress being controlled by different parties. According to Fiorina (1992a), the elections between 1952 and 1992 generated unified governments at the national level just 35 percent of the time. This era is very unusual when compared to the period between 1900 and 1952, when unified governments were elected 84.6 percent of the time. Fiorina identifies this recent trend toward/divided government as the "single most important feature of contemporary politics" (Fiorina 1989).
There has been a rather substantial amount of work devoted to the origins of the divided government (Jacobson 1990; Petrocik 1991; Fiorina 1992b; Wattenberg 1991; Thurber 1991;Jacobson 1990; Bums 1963). The high level of theoretical development and the systematic testing that characterizes these theories of the causes of divided government can be contrasted with the scarcity of work done on the consequences of divided government. Although some of this material may be helpful theoretically, most attempts at demonstrating the effects of divided government have been anecdotal rather than systematic. …