Hi-Stat Vox No. 1 (October 27, 2008)

Towards an Evidence-Based Policy Design

Yukinobu Kitamura

Professor, Institute of Economic Research, Hitotsubashi University

Photo : Yukinobu Kitamura

The global COE program, “Research Unit for Statistical and Empirical Analysis in Social Sciences (G-COE Hi-Stat),” launched this year at Hitotsubashi University, adopts a three-part research organization for building statistical databases and resource archives, for integrating empirical research with theoretical studies, and for statistically verifying forecasts obtained through the program in order to translate the forecasts into solid and practical policy proposals. The basic idea for this program is that carefully thought-out theoretical models become economically meaningful and verified for the first time after they are verified using appropriate statistical data and the right statistical methods.

Naturally, this approach is the same as acknowledging the validity of forecasts made by theoretical physicists for the first time after empirical researchers build huge experimental facilities and use the facilities to verify their forecasts. To begin with, a theory that has no relationship with any existing phenomenon is unthinkable in physics. To give an example, even theories that offer no hitherto known method for verifying their existence, such as the superstring theory, must find a path for verification from a clue of some kind and confirm its existence. That is how the theory is proved right for the first time.

Economists have constructed many theories and produced many empirical results over the years. However, there have been relatively few cases in which theories were verified and their validity was proved with empirical research. In fact, I suspect the reality is that theoretical economists keep constructing theories without assuming their verification and empirical economists continue to engage in empirical research without questioning how universal the outcome of their research may be.

This global COE program deals with efforts to remove barriers that exist between the two groups and actions for paving the way for useful policy proposals and economic reforms. I would like to share my thoughts on several points regarding these initiatives through the program.

There have been a succession of cases in which economic policies fail to function or require reviews because of the emergence of problems after their implementation. For example, the health care system for the very elderly clearly created chaos for people in the field after the lack of preparation prior to the policy implementation and flaws in the system design became evident. Initially, the Ministry of Health, Labour and Welfare tried to force through the policy. However, the Ministry then made minor adjustments to the system on a patchwork basis, and then produced a political volte-face with the minister of Health, Labour and Welfare stating that the Ministry would review the system itself. This policy confusion is not limited to health care. Only recently, the amendments to the Building Standards Law imposed a heavy burden on builders and caused a sharp and unexpected drop in the volume of construction work.

The problems identified in these cases can be summarized as follows. Designing policies based on empirical data and known behavior models prior to their implementation is essential. Additionally, carefully considered policies and designs that require minimum changes should be adopted for systems with the potential to influence such matters as health care and pensions on a long-term basis, because policy revisions in the course of implementation may entail a swelling of adjustment costs. Economists and econometricians have conducted many studies on ex-post policy evaluation. However, they have not sufficiently addressed the issue of designing policies based on the forecasts to a certain degree of predicted responses to implemented policies, leaving that effort entirely to bureaucrats. As the cases stated above clearly show, and taking into account the significance of policy design and ex-ante policy assessment, I believe economists should be actively involved in the forums of policy planning and policy design going forward. Researchers at universities have a relative advantage over bureaucrats in charge of policies when it comes to predicting responses to enforced policies with empirical data. In this respect, I think the government should disclose statistics and other information and work with universities to create a framework for more sophisticated policy designs.

Economic policies and systems do not always produce the anticipated results because, fundamentally, the policies are not compatible with incentives offered by economic agents. In other words, a system does not function unless individuals who fall under the scope of policies understand that taking part in the system and enjoying its benefits is in their best interests. Policies and systems cannot achieve the desired effects, even if they are forced, if the intended beneficiaries feel uneasy about them, distrust them, or think other policies are better. So what can we do? Again, the important rule is to comprehensively analyze the reactions of economic agents using empirical data before executing the policies. Officials in charge of policies at government offices may respond to this opinion with the view that policies must be enforced within limited periods and no such leeway is allowed. If that is the case, I believe researchers at universities should continue their basic analytical research of the responses of economic agents under realistic scenarios. It is too late to carry out empirical research at the stage of planning actual economic policies. Accumulating empirical research findings that can be used for policy planning is essential for researchers at universities.

It is said that the bureaucracy in Kasumigaseki has traditionally functioned as a Japanese policy think-tank. With civil service reforms and a shift to a legislative approach led by Diet members, I believe groups of researchers, centering on those at universities, will be asked to shoulder an increasing share of the basic research that is then made available for policy planning. It is my sincere wish that the global COE program is introduced with an awareness that this new situation represents an opportunity for those of us involved in the program.

Original text in Japanese