LONG-MEMORIED PROCESSES, UNIT ROOTS, AND CAUSAL INTERFERENCE IN POLITICAL-SCIENCE

Theory: It has been argued that because researchers have not taken int o account the long-memoried natures of certain political processes-esp ecially the fact that some political time series appear to contain uni t roots-some users of level Vector Autoregressions may have reached er roneous conclusions about the validity of important causal. relationsh ips and model specifications. Hypothesis: For the first time, this arg ument is evaluated. The difficulties associated with modeling long-mem oried political processes are reviewed. Then several approaches to dea ling with them are discussed. One of the most promising approaches, Fu lly-Modified Vector Autoregression (FM-VAR) is studied in detail. Meth od: The usefulness of FM-VAR is evaluated in a stylized Monte Carlo in vestigation and in reanalyses of major existing studies in political s cience-reanalyses that are representative of the ways in which level-V ARs are employed in our discipline. Results: Our experiments indicate that FM-VAR performs well (particularly in terms of size) in small and large samples, in fully and near-integrated systems, and in stationar y systems. Most important, use of FM-VAR calls into question some of t he major causal findings and specification test results in published s tudies. The implication, therefore, is that taking into account the tr end properties of political processes is essential in theory building in political science.