Behaviors of Policy Analysts in Public Investment Decisions: How Policy Analysts Make Judgments

Policy analysis emphasizes analytical methods to get better information. Better information, however, plays a limited role in improving the quality of policy making if it is not appropriately interpreted. Analytical information measures the different aspects of a policy problem using different methods: analyses result in information that is created different forms with varying qualities and relative importance. In order to be more appropriate for policy making, analysts have to integrate and to interpret the information using contextual and expert knowledge. However, few studies have paid attention to analysts' judgment behaviors. This study examined the judgment structures of analysts who perform actual investment analysis. I analyzed why politicians and bureaucrats rely on policy analysis, and how a growing demand for policy analysis leads to an increase in analysts being more actively involved in investment decision process. Especially, I note that it is not realistic or desirable to restrict the role of policy analysts as a technical information provider. As analysts are required to consider multi-dimensional aspects of investment problems, they have to do more integrative analysis with high level of judgment to respond to the needs of their clients. Analyses of policy analysts' judgments show that policy analysts are not obsessed with economic efficiency when evaluating investment projects. The analysts gave a similar weight to economic efficiency (51%) as they did to policy factors. Also, the large variation of judgments in weighting and scoring that was observed can be explained by several factors: the project fields, analysts' role in analysis, and their affiliation. Most importantly, we can find strong evidence that analysts' judgments are highly related to their self-interests. I showed that analysts' self-interests are more problematic in the judgments than the cost underestimation. With the judgment analyses, I suggest developing management techniques using the statistical distribution, which allows us to infer the possible range of variation of weighting and scoring