Using DDM model to value a bank?

Multiples-based valuation always uses implicit assumptions, e.g. if you have a relatively low price multiple on earnings there must be 'something wrong' with the earnings, or you expect a low growth rate, or whatever.

Detailed modelling approaches, of which DDM is one, make these assumptions explicit, i.e. you create detailed forecasts on where income is going, what happens to the cost base and capital, etc. Therefore, if you want to get some degree of comfort with the drivers of your valuation (which granted often is based on bs assumptions...), DDM is 'better'.

I hardly think of the DDM as useful... I may need more guidance on this, but isn't the DDM just d / r - g or am I thinking of something else?

If we want to 'get comfortable' with drivers you might as well create a full operating model with with multiple scenarios that can be toggled as opposed to an abbreviated DCF that only forecasts the IS. But I don't think thats what the OP wanted, I think he wanted a down and dirty quick way to value a bank.

In this light I think the FMEV approach is used to avoid exactly the BS that goes in DCFs and the DDM, since we don't know everything and get only a 'degree of comfort' instead of pretending like we know, why don't we just arrive at an appropriate range and move on to other things i.e. financial position / BS strength? Your valuation can get bogged down in trying to understand an unpredictable income account even before other perhaps more important things should be considered...

I, without knowing anything about banking valuation, would go with dcf APV, dcf WACC and then RIM before DDM.

DCF is fine to use, but you need to do it directly at the equity level (i.e. use fcf-equity and discount at cost of equity) since interest and debt is part of a bank's operating activities. Residual Income models are also popular, since the starting point is 1.0x book value, and banks are a much more focused on their book value of equity than operating companies.

The other twist to bank valuation is that banks have to uphold specific capital ratios - tier 1 capital, tangible common equity, etc - so they are only able to pay dividends out to the extent equity remains above those thresholds.

Multiples don't work that well in today's environment. Book value multiples are used most frequently, but so many banks trade at a discount to book value because of expected future losses on their loans from 2005-2008 that it's hard to apply a group average to a single bank.

Hope that helps.

Multiples - Use P/B Valuation model - Use free cash flow to equity (or adjusted DDM): 1) Project loans and deposits growth 2)Estimate future interest rates 3)compute risk weigthed assets and equity to cover them 4) compute tier ratio and estimate your "dividend" payout possibilities based on the tier ratio you want to mantain 5) discount using CoE, don't forget that you don't unlever comprable betas

Noted. Thanks.

.

apart from the above (i prefer DDM and P/B), you can also do a P/B - ROE regression - maybe look at the average roe for the last 3 years for the comp set and their latest p/b. widely used.