Preparing for Major Disasters: What should be the desirable policy evaluation criteria?

NAKATA Hiroyuki
Fellow, RIETI

In recent years, huge and devastating catastrophes have occurred in many countries throughout the world including the Great East Japan Earthquake that hit Japan. It is important to take measures in advance for such calamities, which occur very infrequently but cause extremely severe damage, both physically (such as seismic reinforcement) and financially (such as property and casualty insurance). (Note 1) In the case of Japan, its advance measures appear to be fairly effective as far as the physical aspect is concerned. For example, last year's earthquake, given its scale, would have taken a much heavier toll on human lives had it occurred in a developing country or in Japan in the past. On the other hand, however, regarding the tsunami and the accident at the Fukushima Daiichi Nuclear Power Plant, there are assessments that prior anticipation was not adequate. Also, on the financial side, the earthquake insurance subscription rate in Japan still remains low, although it has been on the increase lately. (Note 2)

Ex-ante criterion and the principle of self-responsibility

In economics, it is generally anticipated that people take sufficient measures in advance as they are typically averse to uncertainty or risk. In addition, a concept called Pareto efficiency is often used in policy evaluation. This concept roughly refers to a situation where no one can be better off without making someone else worse off. However, in what sense precisely does someone become "better off" or "worse off"? When comparing the merits of advance measures and policies based on the concept of Pareto efficiency, it is standard to use the criterion at the time of the decision making—the ex-ante criterion.

In this sense, the ex-ante criterion is closely related to the principle of self-responsibility. For example, if someone decides not to buy insurance and then incurs a loss, the consequence is deemed to be his/her self-responsibility as the individual decided not to do so despite the possibility of loss. In this situation, however, wouldn't the individual often regret his/her own decision? Is it reasonable to ignore such ex-post regret? (Note 3)

Infallibility, diversity, and unawareness

Many economic models assume that each individual knows the true probability, and even when asymmetric information exists, it is assumed that (s)he knows exactly what information (s)he is missing. (Note 4) Thus, it is assumed that each individual neither makes any mistake in decision making nor has any regrets. In the above insurance case, even after incurring a loss with no insurance coverage, it is assumed that the individual will consider his/her decision a right one ex-post. In other words, there is no room for discrepancy between his/her ex-ante and ex-post judgments; rationality in a very strong sense or some sort of infallibility is assumed for each individual.

Nevertheless, rare large catastrophic disasters with no past records do take place from time to time. In such cases, the true probability cannot be considered as already known. Instead, one would estimate the probability by using the past data, experience, or various scientific knowledge, but it would be extremely difficult to reach a consensus on the probability. In such a case, it can be rationally (in a weak sense) said that the probability of disaster is either once in 1,000 years or 10,000 years; it is impossible to reject objectively all of these diverse views/perceptions. Furthermore, once such a disaster occurs, one tends to subjectively estimate the probability much higher than in the past, but it is also difficult to objectively reject such a view as irrational. (Note 5) Therefore, it is fully conceivable that a person assesses ex-post that his/her judgment was wrong, and it is no wonder that discrepancies emerge between ex-ante and ex-post perspectives.

Additionally, a completely unforeseen event could actually occur. In such a case, estimating the probability in itself is impossible, and it is inevitable that discrepancies emerge between ex-ante and ex-post perspectives. (Note 6)

What is the desirable policy evaluation criterion?

If, as described above, rationality in a strong sense or infallibility cannot be assumed, what criterion is desirable for policy evaluation? First of all, the emergence of discrepancies between ex-ante and ex-post perspectives should be taken as given. Especially, a policy that can prevent people, even in the worst case, from being thrown into extremely catastrophic situations is desirable. For example, a mandatory attachment of earthquake insurance to mortgages will lead to the elimination of double mortgages at the time of rebuilding the house. Thus, this policy is expected to decrease the level of losses incurred ex-post by individuals, who underestimate the possibility of losses from earthquakes and do not voluntarily purchase insurance.

The above argument may sound as if a policy that provides ex-post public assistance is desirable, but that is not so. This is because such a policy is likely to create moral hazard. Ex-post public assistance should focus on issues that are difficult for the private sector to deal with—in particular, those that are difficult to anticipate in advance. Instead, public policy should focus on advance measures and their promotion, which are designed to prevent people from being thrown into extremely catastrophic situations ex-post, such as the mandatory purchase of earthquake insurance mentioned above. Furthermore, discussions from similar perspectives are desired on the future reconstruction of the disaster-affected areas, various measures against disasters, and, more broadly, development of mechanisms that are designed to prevent extreme economic crises.

September 25, 2012
  1. ^ On advance measures against large disasters in different countries, OECD (2008) provides a detailed review.
  2. ^ Participation rate of all households in Japan as of the end of fiscal year 2011 was 26.0% (Source: Non-Life Insurance Rating Organization of Japan).
  3. ^ There is a long history of an academic debate on the discrepancy between the ex-ante and the ex-post criteria; see, for example, Starr (1973) and Hammond (1981).
  4. ^ Common prior assumption in game theory; rational expectations in macroeconomics as defined by Lucas and Sargent.
  5. ^ Nakata, Sawada, and Tanaka (2010) explain the demand for insurance focusing on the instability of estimated probabilities of infrequent events.
  6. ^ See Samuelson (2004), a somewhat old but related survey paper on decision theory.
  1. Hammond, P.J. (1981): "Ex-ante and Ex-post Welfare Optimality under Uncertainty," Economica, 48, 235-250.
  2. Nakata, H., Y. Sawada, and M. Tanaka (2010): "Entropy Characterisation of Insurance Demand: Theory and Evidence," RIETI Discussion Paper 10-E-009.
  3. OECD (2008): Financial Management of Large-Scale Catastrophes, OECD Publishing.
  4. Samuelson, L. (2004): "Modeling Knowledge in Economic Analysis," Journal of Economic Literature, 42, 367-402.
  5. Starr, R.M. (1973): "Optimal Production and Allocation under Uncertainty," Quarterly Journal of Economics, 87, 81-95.

September 25, 2012