Extending the Mack Bootstrap

Document description

Winner of the 2011 Brian Hey Prize - GIRO Paper 2011

Foreword

Many general insurance actuaries have a love-hate relationship with the Mack bootstrap. The methodology has usefully provided them with a tool to assess reserve uncertainty that is compatible with the traditional chain ladder method and as such has become a market standard for reserving risk assessment. However, they are confronted regularly with the way that the methodology affords them with little control over the resulting distributions. Moreover, the independence assumptions in the method are not always met, potentially distorting and likely understating risk assessments. The lack of practitioner consideration in this area has been a concern echoed by various regulators recently. In light of this, we judged it important to find other methods that are more suitable for Aspen’s reserving risk modelling. With his responsibility in actuarial research and development, Jo was asked to lead the internal Actuarial and Risk Management effort towards this goal. A collection of methods were subsequently tested and further developed, enriching our internal reserving risk modelling toolbox. Those methods that are aimed at the independence assumptions of the Mack bootstrap are presented in this paper, helpfully supported by step-by-step demonstrations and impact studies on actual data. I commend this paper to practitioners for furtherance of the profession’s debate in this important modelling area.

Stephen Postlewhite, MA,FIA