# TMFAleph1 (95.22)

## TMFAleph1's CAPS Blog

Recs

### 2

August 28, 2008 – Comments (7) | RELATED TICKERS: WFC , C , KBE

You can read my article, 'Is it Time to Buy the Banks' here. As promised, here are the assumptions I used in calculating the justified P/BV multiple for the KBW Bank Index:

CALCULATING A JUSTIFIED P/BV MULTIPLE FOR THE KBW BANK INDEX (Part I)

ROE = 10.216%   [Calculated as a weighted average of the ROEs of the index components]

Earnings growth : 8.34%  [Historical 3-year growth rate. I selected the historical growth figure because it lay almost squarely in the middle of a published 3-5 year future earnings growth rate for the index (8.44%[1]) and the weighted average growth rate across all index components (8.27%).]

#1) On August 28, 2008 at 9:28 PM, rd80 (97.75) wrote:

I don't think Justified P/BV metric is valid in these cases.  If the cost of equity and earnings growth rate are nearly equal, you start dividing by very small numbers and the J P/BV blows up.  In cases where COE and EGR are very close, even tiny errors in estimating or rounding those values will make big changes in the metric.

Similarly, if the earnings growth rate is between the cost of equity and return on equity, the metric goes negative.

To get the .84 J P/BV from your linked article and the values provided here for ROE and earnings growth, the COE would need to be 10.57%.  If I apply that same COE to WFC, I come up with a Justified P/BV of 2.8 vs the (3.86) in your article.  Could you elaborate on how you came up with the Cost of Equity both for the KBW and for the individual banks?

I also don't think your conclusion that the J P/BV for the KBW is valid because it's an average across a bunch of banks is correct.  The metric clearly has problems as shown by the wide range of values it produces for the six banks you checked. Averaging across a large sample will take care of statistical flyers, but that's not what you've got here.  This metric is saying BBT should trade at over 16 times book and that you should have to pay someone to take Well Fargo stock - both extremes are non-sensical.  In the six banks you consider, only two, Wamu and Wachovia come out with J P/BV numbers that are anywhere close to reasonable.  A metric that gives you nonsense two-thirds of the time clearly has big problems, at least for what you're trying to do here.

Clearly I'm missing something because I don't see how this metric makes any sense.

Report this comment
#2) On August 28, 2008 at 10:45 PM, anchak (99.87) wrote:

Holy mother of god! I read your second blog first and I was wondering where you got the 10.216 and 8.34 numbers.

I have never ever sort of said anything like this in CAPS - I do really apologize upfront, do not want to offend you - but really you believe that banks historical ROE and Earnings growth are representative?????!!!!!!

If you do, I honestly suggest you take a hard look at this backtesting with Single Variable Linear Regression formula you are propounding ( And yes I know its based om William Sharpe's seminal work - he wanted to explain something in simplistic terms to Wall St/Mutua Fund Managers) - linear regressors are theoretically unbounded and monotonic - ie they never turn - ie THEY MAKE HORRENDOUS EXTRAPOLATORS -which is what you are trying to do here.

If we applied this technique to House Prices - by 2030 , US Home prices would reach infinity or something!

Report this comment
#3) On August 28, 2008 at 10:50 PM, TMFAleph1 (95.22) wrote:

rd80 -- Thanks for your comment. Here are a few points worth considering:

First, you're quite right -- the cost of equity for the KBW Bank index is 10.572%. The calculation is outlined in the second post.

Two, I calculated the cost of equity seperately for each stock in the KBW Bank index. Wells Fargo has a very low cost of equity, at 5.90%.

Third, I only included six stocks in the table contained in the article,  but the KBW Bank index contains 24 stocks.

Fourth, if you think of the index as a holding company with interests in these 24 different banks, it's easier to "visualize" why the P/BV multiple for the index may be valid, despite the fact that some of individual stocks display P/BV values that are nonsensical.

With all that said, the P/BV multiple should be treated with caution -- even that of the index. I hope this is helpful.

Alex Dumortier (XMFMarathonMan)

Report this comment
#4) On August 28, 2008 at 10:51 PM, TMFAleph1 (95.22) wrote:

rd80 -- Thanks for your comment. Here are a few points worth considering:

First, you're quite right -- the cost of equity for the KBW Bank index is 10.572%. The calculation is outlined in the second post.

Two, I calculated the cost of equity seperately for each stock in the KBW Bank index. Wells Fargo has a very low cost of equity, at 5.90%.

Third, I only included six stocks in the table contained in the article,  but the KBW Bank index contains 24 stocks.

Fourth, if you think of the index as a holding company with interests in these 24 different banks, it's easier to "visualize" why the P/BV multiple for the index may be valid, despite the fact that some of individual stocks display P/BV values that are nonsensical.

With all that said, the P/BV multiple should be treated with caution -- even that of the index. I hope this is helpful.

Alex Dumortier (XMFMarathonMan)

Report this comment
#5) On August 28, 2008 at 11:22 PM, TMFAleph1 (95.22) wrote:

Anchak,

Thanks for your comment. You're quite right -- using straight historical ROE figures would be heretical. That's not exactly what I did, though. The secret sauce is a bit more complicated than that.

I did start with TTM ROE figures. When that figure was negative, I used instead an average ROE figure over the period for which the data was available in Capital IQ (typically going back to 1990 or 1991, if I recall correctly -- more than a full credit cycle, in other words).

When the figure was positive, I penalized it by 150bps -- the high end of the estimated range for bank profitability directly attributable to real estate securitization in the period Q2 2001 - Q3 2006 (see Banking and Securitization, Jiangli, Pritsker & Raupach, August 16, 2007)

As far as earnings growth is concerned, I used estimates of future growth, not historical data.

Alex Dumortier (XMFMarathonMan)

Report this comment
#6) On August 29, 2008 at 12:02 AM, TMFAleph1 (95.22) wrote:

Anchak,

Thanks for your comment. You're quite right -- using straight historical ROE figures would be heretical. That's not exactly what I did, though. The secret sauce is a bit more complicated than that.

I did start with TTM ROE figures. When that figure was negative, I used instead an average ROE figure over the period for which the data was available in Capital IQ (typically going back to 1990 or 1991, if I recall correctly -- more than a full credit cycle, in other words).

When the figure was positive, I penalized it by 150bps -- the high end of the estimated range for bank profitability directly attributable to real estate securitization in the period Q2 2001 - Q3 2006 (see Banking and Securitization, Jiangli, Pritsker & Raupach, August 16, 2007)

As far as earnings growth is concerned, I used estimates of future growth, not historical data.

Alex Dumortier (XMFMarathonMan)

Report this comment
#7) On August 29, 2008 at 12:13 AM, TMFAleph1 (95.22) wrote:

Anchak,

Thanks for your comment. You're quite right -- using straight historical ROE figures would be heretical. That's not exactly what I did, though. The secret sauce is a bit more complicated than that.

I did start with TTM ROE figures. When that figure was negative, I used instead an average ROE figure over the period for which the data was available in Capital IQ (typically going back to 1990 or 1991, if I recall correctly -- more than a full credit cycle, in other words).

When the figure was positive, I penalized it by 150bps -- the high end of the estimated range for bank profitability directly attributable to real estate securitization in the period Q2 2001 - Q3 2006 (see Banking and Securitization, Jiangli, Pritsker & Raupach, August 16, 2007)

As far as earnings growth is concerned, I used estimates of future growth, not historical data.

Alex Dumortier (XMFMarathonMan)

Report this comment