Valuing companies whose business is based on OPM(Other People's money)

All entities use their owner's(equity holders) money initially. Some rely on heavy debt financing(for example infrastructure companies have D/E of around 4x). But for some(mainly banks, insurance companies), the whole business model revolves around using other people's money(depositors, policy holders) to whom they have a fiduciary responsibility(and not merely borrower-lender relationship). It is this class of companies which are the subject of discussion in this post.

Banks have historically created value for their shareholders. Barring the subprime blip(post which they were the first to post super profits), banking stocks have a secular trend of going up. For insurance companies, this is even more pronounced where the most legendary value investors alive today(USA's Warren Buffet and Canada's Prem Watsa) have made fortunes for themselves and their shareholders. Which begs the question, to what can we attribute the high valuations of those businesses.

Leverage is one argument. If the banks/insurance cos apply conservative standards of capital acceptance(tight underwriting norms, low deposit rates etc), they can safely leverage(sometimes called 'lazy banking') OPM to make outsized profits for their narrow equity base.

1. So using a ROE metric would be flawed because though banks/insurance companies need to maintain an equity base for regulatory comfort/keeping their license, their actual profits are made from income on that OPM float(net interest income for banks, investment income for insurance companies). If that float is persistent in the form of loyal long term depositors or policyholders, one could theoretically do a proforma ROE calculation considering float as equity. But since the present disclosure standards do not permit this, we need to get other metrics. 
2.  Where OPM outstrips equity,  ROA could be a good proxy for return on float. Banks with high ROA are given higher valuations(better P/BV; higher P/E). Insurers with higher investment income also get a premium abroad(no insurer is listed in India).
3.  Using the old workhorse DCF may pose a difficulty because investors just do not have the information/expertise to value the returns generated by float(or indeed the possible MTM in the float itself). DCF may still be possible for banks but is difficult for insurers. 
4. Barring these metrics, we can use hybrid metrics like P/BV, multiples of premium etc. But these would at best serve as triangulation measures and not for doing the primary calculations. 

One should remember that OPM dies to a trickle at the sniff of reputation risk. Bank runs start and policyholders get concerned. So which doing a valuation, it would be a good thing to embedd a discount to keep that crucial margin of safety for such events.