Comments and Discussion
[The Global Market for Reinsurance: Consolidation, Capacity, and Efficiency]
[Figures]
Comment by Ross Davidson: This paper makes a good contribution
to our understanding of the factors that drive the consolidation within
the industry and the effects of that consolidation.
One of my favorite commercials these days is an automobile commercial showing a certain scene, and then the camera backs up and shows that the scene is part of a larger context, which continues throughout several phases. And, all of a sudden, the viewer realizes that the context of the first scene was very micro in comparison to the context of the overall scene. This is a very clever commercial, and it has application to this paper.
It is very interesting to see the factors that drive consolidation of the reinsurance market and the effects on efficiency of the use of capital. However, there is a broader context within which all of this is happening. This broader context influences the pace at which firms will be acquired, who will acquire them, and how those acquisitions will affect capital. This issue can be viewed beneficially in a more global context.
Let me offer a few comments. These are not criticisms. They simply emphasize points that the authors have noted in their paper, but in a more limited context.
Ultimately we would all like to see the reinsurance market employ its capital to its highest use. But there are some very practical constraints that limit efficiency. For example, many believe that the maximum amount of reinsurance available in any given region within the United States is about $20 billion. That is just a fraction of the $100 billion or so of total reinsurance capital in the market. And, of course, disasters happen locally. Given this context, I was surprised that the authors' efficiency ratios were [End Page 210] so high. We probably cannot squeeze $100 billion of capacity into a particular local market, because of this practical constraint.
While the authors make the assumption that capital will be dissipated to the last dollar, in reality they will expose themselves to the extent that they will pay out losses large enough to bankrupt the industry. The reinsurance market thrives on what is insured initially in the primary market, and we are seeing some lumpiness and some discontinuities in the private market. I recently looked at every company that serves disaster-prone areas in the United States and accumulated all of the net worth of those companies in a particular region just to see the kind of outer-bound theoretical capacity that may be present in the private market. The result was that primary insurers do not put all their eggs in one basket either. Markets like Florida tend to attract a lot of companies, and some, but not all, of that capital is at risk. But most regions do not have a lot of industry capital at risk. One reason is that, recently, there has been a significant reengineering of primary policies, such that the risk of capital loss is a lot less than it was in 1992 when hurricane Andrew took place. Higher deductibles, narrower coverage, limited top-side exposure, and the transference of some risk to government mechanisms have all had the effect of limiting exposures of the primary market.
All of this affects how the reinsurance market operates within the broader context. Efficiency within the reinsurance market is important, but efficiency in covering the risk that individuals and businesses have is another broader context. I encourage the authors to look at that context as well, perhaps in a follow-up to this very fine paper.
I was stricken by the fact that the United States has about 40 percent and Western Europe, 35 percent of the market. After Japan, no other market provides material opportunity for global diversification. Although the concept of global diversification is important to the efficiency of the reinsurance market, the opportunities for true global diversification seem to be fairly limited. You can diversify between Europe and the United States, but how much can you effectively spread your risks to markets that represent only 4 or 5 percent of the global market. That probably will change over time as other economies evolve, because insurance tends to be used by more advanced economies. Of course, the ability to achieve efficiencies of diversification should increase as that evolution unfolds.
There is another interesting development that will have an impact on the global context of risk management. As primary firms get larger, those firms [End Page 211] realize that they can take on risk and retain and manage more efficiently themselves. The idea brought out at the beginning of this paper was that insurers at risk are willing to pay a premium over the pure premium to have somebody else cover that risk. As the capacity of industry and firms to absorb small- to medium-size hits increases, the primary industry will be more sophisticated and more willing to retain risk, because they will realize that it is beneficial and profitable to do that.
So as these institutions grow, the role of the reinsurance market may shift to dealing with smaller, emerging institutions that have less capacity to bear risk. Dominant forms of ownership also will have a major impact on this trend.
If you step back and look at the relationship between primary insurers and reinsurers, reinsurers are really not risk-transfer mechanisms. They are time diversifiers, but not necessarily true risk-transfer mechanisms. In the global context, reinsurance is part of the capital market. Investors expect return on and return of capital in the long term. Investors might make and lose individual bets, but they do not expect the portfolio to have a long-term negative return. Viewing reinsurance as a time diversifier, as opposed to a geographic diversifier, is a helpful context. The authors point this out to some extent in the paper. It is helpful because mechanisms are emerging that might be more efficient for time diversification than reinsurance can be because of regulation.
Insurance regulation forces risk capital into relatively inefficient structures. But capital markets tend to be very efficient. And if, as is emerging, the capital markets can be applied to risk management in such a way that capital does not have to be forced through inefficient mechanisms, then the efficiencies addressed in the paper can be achieved in ways that reinsurers cannot facilitate. That is a bit scary for some reinsurers. It elicits a bipolar reaction that manifests itself in one of two attitudes--either "let's join them" or "let's resist to the very end and maintain barriers to entry." Reinsurers that use the emerging tools as part of their risk management portfolio ultimately will survive and win. Trying to preserve some kind of a regulatory franchise in the reinsurance market is a long-term strategy doomed to failure. So it is helpful to think of the issue as trying to get to the most efficient approach to time diversification.
Circling back to a thought alluded to earlier, another contextual concern is that large blocks of risk are not being covered by either primary insurers or reinsurers. And the trend over the time period studied in this paper has been for the primary market to stop covering some risks. [End Page 212]
Take a California earthquake as an example. Basically, the California mini-policy was engineered because insurance rates allowed by regula-tion did not support the risk that was being borne by the primary market. So the strategy of the mini-policy was to reduce the risk to conform with the rates that could be charged in the marketplace. What happened as a practical matter is that a large segment of risk was avoided--put onto the policyholders.
I asked the director of the California Earthquake Authority (CEA) about the impact on the CEA of an $80 billion quake in California. The CEA provides about 70 percent of earthquake coverage in the state. His response was that the impact would be $4 billion to $6 billion. My follow-on question was, "What will happen to the rest of it?" There really was not a good answer. The CEA's mini-policy is basically a way of saying that there is only so much risk that primary insurers are willing to take at allowed rates. And reinsurers only get to cover so much of that risk, based on what primary insurers are willing to take. So there is a massive movement of insurance capacity out of this market. The efficiency of the employment of capital in the reinsurance market is interesting in the global context. If less risk is being covered but there is more capital to cover it, the supply and demand relationship may get out of line and drive a lot of the consolidation we see.
The other issue is how to deal with cataclysmic, essentially noninsurable risks. I used to think that earthquakes were the biggest risk--the $115 billion gorilla risk. But the other night the Discovery Channel was reviewing the impact on global weather that the Indonesian island had when the volcano, Krakatau, blew up in the late 1800s. That disaster caused a yearlong winter. There was massive famine because crops failed. There was no growing season. That kind of cataclysmic event is more probable than a comet hitting the earth. Again, much depends on the breadth of the time frame and the context.
Another potential cataclysmic event was identified recently by satellites capable of peering through clouds at Antarctica's ice shelf. There is a greater chance of part of the ice shelf breaking off and raising the level of the ocean than we previously thought. It would raise global ocean levels about seventeen feet. Think of the existing shorelines. I was fishing the other day near Padre Island, Texas. A mile offshore, the guide said that we could get out of the boat, wade in the water, and fish. We could not see land, but we were standing only in about three or four feet of water. If [End Page 213] the ocean rose seventeen feet, much of Texas would be under water. That would have quite an economic impact. I do not know how we should deal with that, but it is a broader context for considering these issues.
The evolution of the securities markets will be very important in the consolidation of the insurance market because the securities market is a more efficient risk-management mechanism than the reinsurance market. I agree with the authors that the effects will be quite interesting over time.
I congratulate the authors on a great paper. It is a good foundation for additional work. I encourage them to go to the next level in future research.
Comment by Thomas Holzheu: Let me briefly go through the main hypotheses and offer some comments on the findings of the various models presented in the paper and some related data issues.
The first hypothesis assumes an improvement in claims-paying efficiency, which is accompanied by an improvement in the response function. The hypothesis that there is more capacity in the reinsurance market is a significant result of the empirical testing.
As a counterargument, Ross Davidson asks why there is only a $20 billion capacity at one point in time in one region. This seems to contradict the findings of the response function. The caveat of this model is the assumption that every company shares the same portfolio. That necessary simplification misses the fact that the majority of reinsurance is not written on a proportional basis and the risks are individually segmented or structured. The reinsurers assume clearly defined contingency liabilities from their clients' portfolios, resulting in each reinsurer holding quite different portfolios. If there is a big catastrophe, you discover that some reinsurers are heavily involved, and others are not. The response function in reality will be lower on average than the model would suggest.
The second hypothesis assumes an improvement in mean variance efficiency as a result of consolidation. The hypothesis is confirmed in the empirical testing. The underlying mechanism is the law of large numbers, which implies that combining two more or less independent portfolios will reduce the volatility of the combined portfolios. Insurers and reinsurers react to that diversification effect by holding less capital. They need less capital to share each risk. Getting the same profits on a lower capital basis, of course, raises the return on equity. Both effects together result in efficiency gains.
The next hypothesis states that merger targets are less efficient. This hypothesis is confirmed in the context of the mean variance analysis. But [End Page 214] merger targets also have lower combined ratios, which normally indicate higher technical profitability. Merger targets also tend to have higher assets in relation to their capital base and to the volume of premiums, and they tend to have a higher return on investment. All these are usual measures of the efficiency of insurance companies. The results do not seem to be entirely consistent. Some of these inconsistencies might be related to data issues, which I address later.
The next hypothesis is that acquiring firms are more efficient. There is weak empirical evidence confirming that hypothesis.
The final hypothesis claims that financially vulnerable firms are more likely to be takeover targets. Here empirical testing provides mixed results.
All of these hypotheses are plausible, and some of the problems in proving them empirically in the right model setting might have to do with data issues. One of those data issues is that all available databases provide company data on the individual company level but not on a consolidated basis. As we are talking about consolidation here, some differences between targets and consolidators do not come out as strongly as they would on a consolidated basis.
The next data issue arises from cross-country accounting differences. If one compares U.S. insurers with continental European companies, one faces totally different accounting for capital. Capital is of particular interest for the analysis here, because it influences both the measures of capacity and the return on equity.
Another data issue relates to the Asian figures. Almost all of the companies covered might be Japanese companies that are, with only one or two exceptions, primary insurance companies. Japan practices a very special system in which a fraction of motor insurance risks must be pooled and reinsured back to the primary insurers on the basis of quota shares. These companies appear in the reinsurance market statistics without really being reinsurance companies. Including these companies in an analysis of reinsurance companies results in distortions, as they have huge assets and a huge surplus but no real reinsurance risk.
Now I move on to the model framework and add some thoughts supporting the practical relevance. In the theoretical framework, different levels of the diversification of portfolios are separated. The first level of risk consolidation due to the law of large numbers is referred to as locally insurable risks. The next level consists of globally insurable risks, which combine geographic regions. This is where the reinsurance market comes [End Page 215] into play. Next are the globally diversifiable risks. For example, the loss potential of a large earthquake in Japan is way beyond the region's capacity and probably even exceeds the capacity of the global reinsurance market. Such extreme scenarios could even describe globally undiversifiable risks.
The current consolidation of the reinsurance market affects the category of globally insurable risks. I would like to share some findings, showing that there really are advantages to combining portfolios globally. Figure 1 shows the results from reinsurance operations for different countries since 1980. The results of the European reinsurance markets are pretty much in line, but the U.S. market shows more fluctuation and much different performance than the European markets.
The correlation coefficients between the technical results confirm this evidence. The correlation coefficients among European markets are similar and rather high. The U.S. reinsurance market is much less correlated to the European markets. The European reinsurance markets, however, look quite alike. European reinsurers are already holding quite diversified portfolios of the European primary markets. This explains why there is not a big difference in the results for reinsurers, because they all somehow share the same European regional risk portfolio.
More support for that hypothesis comes from looking at the correlation among European primary markets, which are much lower than the correlations among the European reinsurance markets. Also the correlations between European primary markets and the U.S. and Japanese primary markets are not so different from the correlations of the European markets with each other. All these primary markets are independent to some degree. The European reinsurance markets are much more interdependent, which is the result of sharing a similar portfolio.
What is the implication for the consolidation story? Apparently, there is big value in combining a U.S. and a European reinsurance portfolio, because this can reduce volatility. At the same time, there seems to be less value in combining one European reinsurer with another one, because they are already quite alike. And that is exactly what happened in the takeover wave that started somewhere in 1994.
The improvement in portfolio diversification can be seen in the geographic split of the business of the major reinsurers, all of which were involved in this consolidation wave. For all players, the mix of U.S. versus non-U.S. business shifted between 1994 and 1997 toward a more balanced [End Page 216] portfolio. That was a main goal of the mergers and acquisitions and practically confirms the theoretical work on risk consolidation.
In closing, I wish to return to the theoretical framework of locally insurable risks, globally insurable risks, and diversified risks. Figure 2 shows the development of the corresponding markets over time. In addition, it shows a layer representing the self-retention of corporate insurance clients. The development of captives and self-insurance, which describes conscious retention of risks, started somewhere in the 1950s. Insurance and reinsurance have always been the traditional carriers for risk transfer. Beyond the globally insurable risks lies the market for globally diversifiable risks, which can be addressed with securitization. Somewhere in 1994 and 1995, insurance risks began to be securitized.
What happens is that the whole risk portfolio is stripped into several layers with several specialized players. The increasing retention is a result of companies getting bigger and developing more sophisticated risk management [End Page 217] techniques. The next layers are insurance and reinsurance companies. The reinsurers moved closer to the efficiency frontier as a result of the merger wave. The potential of global diversification is now used more efficiently. It is harder now to expand the capacity limits within the reinsurance sector. The next step is to expand to other markets. Securitization of insurance risks enters the much larger financial markets directly.
Overall, the changes describe specialized players and specialized layers of risk transfer resulting in a more efficient use of risk capital. The consolidation wave contributed to that development.
General Discussion: Allen Berger suggested that because of differences in prudential regulation across countries, it may be important to take into account the country where the firms are insured and not only where the firms are located, as was done in the study. He also wondered whether the study's result that acquirers are more efficient than acquired reinsurers holds true once one controls for possible country-specific effects. J. David [End Page 218] Cummins responded that the effect of differences in regulation may not be large because the reinsurance market has been largely unregulated both in the United States and in Europe for a long time.
Anthony Santomero remarked on the interesting asymmetry between the banking sector and the reinsurance sector. In banking, the presumption seems to be that small is good unless proven otherwise, whereas in the reinsurance business, the opposite appears to be the case, at least based on the findings of this study. He also posed the question of whether, over time, insurers and reinsurers will be less able to diversify their catastrophic risks, since large catastrophes such as hurricanes are becoming global in scale.
Robert Grohowski argued that the demand for reinsurance may be affected by the organizational forms of primary insurers. For example, mutual companies tend to be less responsive to their shareholder base and less concerned with quarterly earnings. For this reason, one would expect mutual insurers to be more willing to self-insure, all other factors being equal.
Charles Calomiris posed the question of whether greater reinsurance concentration and deregulation in the primary market will address the current market failures by properly internalizing catastrophic risks. One participant added that in order to address the issue of large catastrophes, analysts must consider whether the individuals and businesses that purchase insurance against these risks are adequately taking into account those risks involved when they make economic decisions. It is not simply the governments or private insurance industries that must address these issues.
