Irrational Economist_ Making Decisions in a Dangerous World - Erwann Michel-Kerjan [110]
• If we assume that the market for insurance continues to be one with short-term insurance contracts, then stage 1 risk implies that future insurance premiums will be random.
• The second major form of stage 1 risk lies in asset price risk. Assets that are exposed to a changing risk will be re-priced. For example, coastal homes will tend to devalue as the risk is perceived to increase.
• There are other forms that the stage 1 risk could assume. For example, if future governments bail out uninsured property owners, then climate change implies a random tax liability to fund these bailouts.
There may be a demand to insure these risks. I will focus on only two of the stage 1 risks: the insurance of risky premiums (i.e., premiums may be volatile going forward) and risky asset prices. However, to explore the potential for insuring these risks, I need to be more specific about the generic properties of the stage 1 (and stage 2) risks, and what these properties imply for the feasibility and design of insurance contracts.
While relatively little is known about the properties of stage 1 risk, some conservative conjectures can be offered. First, there seems to be a general (perhaps a better word is “vague”) consensus that the trend is upward. It is difficult to predict the impact of climate change on future tropical storm risk in a specific area for a given year. Nevertheless, the general view is that global warming will lead to more intense (though not necessarily more frequent) tropical storms in some regions.4 Note that, insofar as an upward trend is expected, wealth transfers associated with this trend are not amenable to normal risk transfer mechanisms. Rationally, the expected increase in risk should already be impounded in house prices. Moreover, the expected increase in hazard risk in a given region translates into an expected increase in insurance premiums.
A safer conclusion one can draw is that while the mean predictions of future regional storm activity are unclear, the uncertainty surrounding those predictions is likely to increase. This uncertainty or ambiguity could be transferred under a hedge or insurance contract; clearly recognizing that the risk to be hedged is the unexpected change in the level of hazard, one might contemplate an insurance against unanticipated premium increases. However, the availability and pricing of such insurance might depend on whether the ambiguity is independent in different places (e.g., we could underestimate risk in one place and overestimate it in another) and whether the ambiguity relates not just to the year-by-year estimates of loss but also to the long-term trend. Trending or momentum in the hazard path implies uncertainty about the long-term hazard and diminishes the possibility of inter-temporal diversification. 5 With such momentum, we could find that, if we underestimated the hazard risk for year 2020, then we would likely also have underestimated for 2021, 2022, and so on. In addition, there might be contemporaneous correlation. Suppose that nature chooses a general path for climate change (e.g., a 2° increase in mean global temperature over a fifty-year horizon). With such a path, how correlated will the impacts on different hazards in different regions be? If storms will be more severe/frequent in some regions and less severe/frequent in others, then the uncertainty about which regions will have more or fewer storms is diversifiable. However, if there is more correlation across