Cluster modelling – better than replicating portfolios?

The increasing demands of modelling insurance portfolios has seen associated computing costs rocket – but cluster modelling, according to its proponents, offers the potential for insurers to achieve almost the same level of accuracy as a full-blown model, but more quickly and at a lower cost. Clive Davidson reports

The move to stochastic modelling for financial and risk analysis presents a challenge to insurers. The size of liability portfolios and the need to project their value in complex ways under many scenarios results in enormous computational demands. Faced with the challenge, insurers can just accept the long modelling runs that leave little time for deeper analysis, or throw expensive hardware at