What is a kelong? How to build a kelong?

What are the best books to read on how to build a financial model for an insurance company?

• Some great value investors have stakes in AIG (Fairholme, GMO (i think), Chieftain, Perry Capital), and I wanted to build a thorough model. Through some preliminary research I understand valuing a financial institution is quite different from your average DCF for a bunch of reasons (changes in WC can be volatile, debt is more like raw material than a source of capital, capex much less significant for a FIG). Wondering if anyone knew of any books that could teach me how to do it in a thorough and correct way. Specifically insurance companies if possible. Thanks in advance.

• Answer:

  Honestly, I do not think such a book exists. For Life insurance company, the best start would be to dig (and dive) into a MCEV-like valuation technique.  See https://www.soa.org/Files/Pd/2014/val-act/2014-valact-session-47.pdf and http://publications.milliman.com/research/life-rr/pdfs/mcev-reporting-mg-alfa.pdf as a next step. MCEV is Market Consistent Embedded Value -- as the name suggests, it uses market-consistent (ie, parameters of the models are derived from observed market data -- eg, implicit volatility in options) data to estimate the value of a Life insurance portfolio.  Akin to a DCF, really. For a P&C insurance company (and this should also work for Health products and most Protection products, maybe excluding Long-Term Care), you could try a DCF with the following drivers : Pick a stable claims ratio. This a big assumption.  This will give you the level of technical provisions as a % of GWP (*) Take a simple rule for the capital to set aside to cover the risks based on the level of guarantees.  Before Solvency II, the core rule in French GAAP (**) requires max of 16% of GWP or 23% of Claims.  Pretty straightforward. But it gets way more complicated when going into the details by product / level of guarantee / re-insurance, etc. Invest the money and assume a return close to the 3-yr "risk-free" or swap rate.  Unless you want to do a full ALM analysis, that's reasonable. Substract maybe 0.5% for asset management costs. That's your financial income. Model your expenses as for any company (contract management costs / per contract, general expenses as % of total sales, IT, commissions, etc.). Do make an explicit assumption about customer retention rates.  This is important in order to accurately model the commissions, the customer acquisition costs, and the contract processing costs. From the level of premiums you earn minus the expenses minus the claims, you get your underwriting income. From underwriting income + financial income, your get total OP. From there, deduct you cost of capital, based on the total technical provisions + capital to set aside to cover risks. Hope it helps -- (*) Careful if you assume a growing company -- the claims do not all come in one year, and modeling the claims schedule can turn into a nightmare. (**) That's what I am familiar with... sorry...