Solvency 2: a real advance?

Summary The future solvency standards for the insurance industry, Solvency 2, aim to improve risk management by identifying different risk classes and modules, and by allowing companies to use internal models to estimate their regulatory capital. The standard formula defines this capital as being equal to a 99.5% VaR over a one-year horizon for each risk module. Then, at each intermediate consolidation level, the different VaRs are aggregated through a correlation matrix. Several problems appear with this method: - The regulator uses the term "VaR" without communicating marginal or global distributions. This multi-variate risk measure is only relevant if each risk follows a normal distribution. - The time horizon of one year does not correspond to that of an insurance company's commitments, and poses problems when it comes to determining the frequency of updates of internal models. - The dependency structure proposed by the standard formula does not correspond to the one usually implemented by companies and is difficult to use in an internal model. The first part of this paper will present in detail the key points of the reform and will give some thoughts on its application in risk management. In the second part, it will be shown that this multi-variate risk measure does not satisfy the main axioms of a risk measure. Moreover, it does not allow for the comparison of capital requirements between companies, since it is not universal. The third part will show that in order to value capital at an intermediate point before maturity, a risk measure must be able to adjust to different periods, and thus be multi-period. Finally, the fourth part will focus on an alternative to the correlation matrix to model dependence, namely copulas.