Thursday, April 26, 2012

Are Large Cap Bank Stocks Undervalued?

As will be explored in this post, a major determinant of large capitalization banking stocks is the direction of housing prices.

US centered large cap banking stocks appear undervalued by many measures in late April, 2012. Citigroup is selling at 0.53 net book value and approximately 0.66 net tangible book value, Bank of America is selling at 0.41 net book and approximately 0.50 net tangible book value, and even the generally well respected JP Morgan is selling at 0.91 net book value.

In comparison, in southern Europe, banking stocks are currently in a declining trend, but appear even more compelling to from a valuation standpoint, at first glace: Italy's largest bank, Unicredit is selling at 0.11 net book value (falling over 30% in one day in January 2012), Spain's Banco Santander is selling at 0.59 net book value (falling approximately 25% from march 2012 to April 2012) - even as Banco Santander has a compelling market position in one of the hottest emerging markets, Brazil.

The first step to analyzing large capitalization banking stocks is to note that the earnings power of the banking franchises - potentially massive due to the incredible size of the bank's balance sheets -- can be completely diluted to the current shareholder by loan losses. This is to say, if there are loan losses of sufficient size, then the loss taken by the bank will result low equity to asset ratios (as the loss flows to equity) which necessitates the bank to raise more capital. As the share price tends to plummet with high realized loan losses, the bank will have to sell very high numbers of shares to bring capital ratios back to acceptable levels. So one could say that there are three main factors should be discussed in order to analyze current banking valuation for the shareholder:

1. Earnings power of the banking franchises,

2. Current required write downs from loan losses and

3. Potential future increases for loan losses (or, alternatively, reversal of estimates for loan losses) in future time periods.

On point 3 above, the major determining factor of additional losses beyond current estimates are the direction of housing prices. This is to say, if housing prices go up, then it is possible that current estimates of loan losses on the books will be lower, significantly adding to shareholder value.

The previous post in this blog has argued that housing prices are a larger contributor to GDP than commonly calculated by economists, and therefore will impact not just real estate but other commercial activity and loans, without government and central bank stimulus. Note that, as Warren Buffett stated in bis assessment of investment in banking, realized write downs may be "gained back" if assets recover in value (assuming banks have already taken the full loss, then any gain above that loss will add to earnings). That is to say, if the asset prices of problem loans increases, a firm may "write up" its problem assets, in the most optimistic scenario. In the third part of this post, housing prices are theorized to be driven in the first stage up by "a large scale injection of money" into the economy by one or more of the following forces: large scale investment from positive views of future growth of the economy, additional earnings from exports, and/or monetary expansion.

It is assessed that in the US only large scale investment is moderately possible over the next few years based on computer technology or some other potentially now-unrealized technology (always possible in an innovative economy such as the US, that a new technology can appear over a short time span, such as the Internet in the mid 1990's, or the automobile in the 1910 to 1920's, etc), therefore the recovery in housing prices and the recover in banking stocks is tepid at best.

Bank of America Analyzed as an Example of Value in the Large Cap US Banking Stock Universe:
Bank of America is one of the largest mega-banks in the world, ranked as the number #3 banking institution worldwide in terms of total assets at the end of 2010 with approximately $2.1 Trillion of assets. The simplifying assumption here is that the earnings power of BofA can be expressed as a net interest margin plus fee income as a percentage of total assets. Many analysts will estimate the earnings power of a bank by its divisions -- depending on how the bank divides its reportable segments into real estate lending, commercial lending, trading, credit card services, trading, investment banking etc -- then add up this earnings power on a segment by segment basis. A difficulty of using a segment approach is that such an approach takes emphasis off total loan losses, and treats the divisions as fully separable. I have read one analysis that stated card services, for example, would be worth by itself "more that the entire market capitalization of Bank of America." (and therefore BofA is a buy since the other divisions must be worth at least zero). However, in reality the Bank cannot spin off the credit card division as an entirely separate company: credit card holders will be deposit holders of the bank, deposits are used for funding of all divisions of the bank, if the bank losses depositors then credit card market share will go down.

So one could say, credit cards will report through the entire banking operation which must absorb credit losses. One could say, that each segment will be linked to the overall operations and asset levels of the Bank as a whole, although in different ways for example the trading division will be linked by the borrowing credit of the bank through the acquisition of trading assets, etc for the other segments of the bank. (with the possible exception of the investment bank, which could be potentially separated out from the overall bank). At the end of 2011, a full $311Bn of BofA's total assets were invested in debt securities (secondary market, a low interest rate) and $120Bn were in cash and equivalents -- the rest were in commercial and real estate loans of $900Bn, trading assets ($176Bn, perhaps a majority from the trading division at the acquired Merrill Lynch) and reserves deposited at the Federal Reserve, which earns a small interest margin, and other assets such as property and equipment. The vast majority of assets are therefore accruing income (one could say, of course, why otherwise would these be "assets"?) the tricky part is assigning fee income to the balance sheet.

In 2011, the total net interest income of BofA was approximately $45Bn - representing a slight net interest margin of only approximately 50 bps on interest bearing assets. Fee Service and fee income was higher at approximately $58Bn before loan losses, with the majority of fees coming from credit card, deposit account, brokerage and investment banking fees. One could say , again that the majority of fee and service income is linked to total assets (with the possible exception of investment banking fee income). All personal and other expenses were approximately $60Bn in 2011 for BofA, and tax rates are assumed to remain constant at 38%. If one assumes that interest rate conditions remain similar going forward for at least the next couple of years then net interest and fee income appears to be approximately 50 bps total, then the revenue power of Bank of America is approximately $120Bn ($2Trillion times 0.60%), minus $60Bn in expenses to yield $40Bn in pre-tax income, and $24.8Bn in net income.

Note! This net interest spread of only 50 bps total on both interest and fee income is very, very low -- most banks report net interest margins in normal times in excess of 200 bps and certain banks, such as WestAmerica in Northern California have net interest margins of over 400 bps. If, conceivably, BofA could get their overall retun on assets around 200 bps then the earnings power would be massive, around $210Billion of net income per year (before loan losses). This is the power of having $2Trillion of assets, potentially -- and this is the hope that many analysts point to when they state BofA is undervalued -- however there are several complicating factors.

Potential Loan Losses at Bank of America: BofA's real estate portfolio consisted of approximately $83Bn of home residential and equity loans with Loan to Values above 100% at December 31, 2011, and $47Bn (not indicated if included in the above 100% LTV category above) of home loans with FICO scores of less than 620, according to page 186 of their 2011 10-K. An additional $11.5Bn of credit card balances are outstanding to consumers with FICO scores of less than 620. At total of approximately $94BN-$130Bn of loans in the real estate and credit card category appear at risk if housing prices and the economy continue to deteriorate.

Of course, not the entire amount of the potentially problem loans would be written off, but some percentage, and my point is that this percentage would increase with continued declining house prices and a (however likely) potentiality of a double dip recession. BofA has a combined approximately $24.5Bn of "receivable criticized" commercial loans according to page 186 of their 2011 10-k. Again, the full amount will likely not be written off in future years, but some percentage depending on the strength of the economy. Some percentage of the approximately $70Bn of goodwill from the acquisitions of Merrill Lynch and Countrywide should be written off, if the economy and house prices deteriorates further.

What have other analysts estimated as the potential credit losses at Bank of America? Current CEO of the publication the BusinessInsider, Henry Blodget estimated loan and goodwill loans of between $100Bn and $200Bn, citing certain percentages of of home equity, commercial and goodwill to be written down, citing various sources, including bloggers and bank analysts, including a hedge fund which was short BofA stock. The analysis seems to have gotten under the skin of the Bank, as the Bank issued a statement denying Blodget's article within 24 hours of it being published in late August. I am not going to speculate whether or not Blodget's analysis is correct or not, but rather that the ultimate credit loss by Bank of America, and by extension many of the large cap banking stocks, will be driven by the direction of house prices and, as house prices impact GDP to a larger extent than commonly supposed by economists and analysts. It should be noted that Warren Buffett, in stating his reasons for his purchase of $5Bn worth of preferred shares in Bank of America in late August, stated that he thought that housing prices were going to improve thereby giving a boost to the economy on Buffett's appearance on Charlie Rose. It is to housing prices that we now turn.

Thesis: Housing Prices Cycles Theorized to be Driven by "Macro-Scale Injections of Money into the Economy
A huge number of studies have looked at housing prices, mainly from the standpoint of supply and demand. Factors that impact demand are household income, and by extension employment rates and population growth, availability and interest charged on mortgage financing, cultural dimensions of housing demand (whether the population prefers houses to apartments, number of occupants per unit, etc). Supply is calculated by the existing housing stock and additions to the housing stock provided by surveys of homebuilders, economists and government officials.

One observation -- that does not appear to be in most of the existing studies on house prices -- is that household income levels and growth will be impacted, in part, by a relation with housing prices, in a self-reinforcing manner. As housing appreciates in price, the economy will strengthen and more people will be employed and earn income, which can then be used to purchase housing. Economists disagree on the exact magnitude of the relations between housing prices and GDP and household income, but mainly agree that a relation does exist. A self reinforcing relation can also occur with financing and housing, in so far that financing will be more available and lower cost in a period of steadily rising house prices, all other factors equal, which supports house prices and vis-versa. The opposite can occur in a period of declining house prices -- financing is more difficult to get and at higher cost driving down demand for housing and therefore housing prices.

The "Chicken and Egg" Problem in Determining Housing Prices and Demand Factors: As such, as the forecast for household income will depend in part on housing prices, we have a "chicken and egg" problem in forecasting house prices. This is to ask: what first causes housing prices to appreciate, in the first stage, is it disposable income rising first or housing prices rising first? Once housing prices establish an upward trend (for some reason) then they will impact household income, but in the first time period, what drives up disposable income or does disposable income go up first or housing prices go up first? Well, it seems using common sense, in the first stage disposable, income will go up, which then goes towards supporting housing prices then house prices can "self-reinforce" disposable income. The author of this post would propose that in order to get the housing price cycle started upwards -- and, critically to stop house price declines from continuing -- that some large "macro-injection" of money into the economy is needed, to increase disposable income and therefore begin the housing price-reinforcement cycle.

Three Types of Macro-Injections to Begin Housing Price Appreciation Cycles: An excellent chart of the long term (since 1890) prices of houses in both the United States and Australia is presented here. Data is from Nigel Stapledon in Australia and Robert Shiller in the US, and is inflation adjusted. One can see a few interesting trends, house prices in both countries were relatively flat from 1890 to 1945, then increased dramatically from 1946 to 1950 as the solders came back from WW2, and a wave of optimism swept both countries (the Japanese defeated for Australia, the Germans and Japanese defeated for the US). Investment in the United States took off from 1950 to the late 1960's, (data to be provided) driven by business optimism in the US. An interesting divergence in the housing prices of both countries appears in the 1970's -- the US's house prices kept up with (high) inflation, while Australia's house prices increased at a significantly higher rate -- an increase in Australia of approximately 50% from 1970 to 1975 while the US housing prices remained flat.

It is theorized that Australia benefited from the increase in metal and coal prices during this period while the United States did not to the same degree, which drove housing prices higher in Australia verses the US. House prices remained relatively flat in both countries indexed to inflation through 1980 to 2000 then took off way above trend from 2000 to 2006 in the US and from 2000 to 2010 for Australia. It is proposed here that the driver of house prices was different in the two countries -- in the US it was "an injection of money" from loose monetary policy (interest rates declining from 6% levels in 1999 to 1% levels in 2002) while in Australia the injection of money stemmed from export earnings. Next steps in this post (as it is becoming very lengthy) is to determine the amount of additional capital needed as a percentage of GDP and/or the housing stock to supply continuation of housing price increases.

Monday, April 16, 2012

How do House Prices Impact GDP?

Why are the US economy and stock markets are doing reasonably well, while housing prices are doing poorly? Something here feels "amiss."

Housing prices are currently declining in most areas of the United States, in Australia and New Zealand, in select Chinese cities and in southern Europe. Spain's house prices are forecasted to decline a further 20% after falling by 20% from their peak in 2007. This has lead Citigroup chief economist Willem Buiter to label in early April 2012 the economic situation in Spain as "dire." The forecast for housing prices is uncertain, with Goldman Sachs revising in January 2012 their estimate for the return of appreciating house prices from mid 2012 to the end of 2013.

Review of Existing Approaches to Modeling House Prices and GDP:

Economists have re answered the question of house prices and the economy by correlating house prices with consumer spending, by regression analysis through estimates of the wealth effect. In these models, increases in housing prices make the home owner feel more wealthy, thereby impelling the home owner will spend more of this wealth per year. The increase in consumption per year per dollar of increases in home prices is estimated by several economists (from Cambell and Cocco, Federal Reserve, 2011 to Case, Quigly and Shiller, 2003, among others) to between 3.5 cents to up to 18 cents per year per $1 of house price appreciation\, based on regression analysis of housing wealth to consumption over varying time periods.

Can one say, that with a 30% decline in house prices in the United States from their peak in 2006, consumption would be approximately 8% lower due to a 10% tendency per dollar of house price change? (peak housing wealth of approximately $28Trillion, loss of approximately $9Trillion of housing wealth, 10% lower per dollar of consumption). The results of the economists cited above would appear to say yes, with several caveats:
--> The data only shows past data in which house prices went up unabated for 60 or more years, therefore the models may break down in periods of lower year on year house prices.
--> Further the exact mechanism of house prices on consumption and the economy is not stated in detail: housing is treated as a normal asset without unusual attributes, even as housing comprises approximately two-thirds of the median American's net worth.
--> Lastly, the additional consumption due to housing is treated as the excess of consumption year to year over net disposable income in several models, even as increased spending from housing would result in higher income so would be difficult to separate out: as people spend more due to higher housing prices, this higher spending becomes other businesses' net income, leading to higher disposable income through a multiplier effect, making the cause and effect of higher disposable income due to higher consumer spending from higher house prices difficult to determine.

Home Ownership as a For-Profit Business:

A way to model house price declines on the economy and stock market is proposed here: home ownership in the economy allows individuals to become, in effect, "business owners to themselves." This is to say, appreciating house prices allow home-owning individuals to become successful profit owners (or "capitalists" in a traditional economic sense). Instead of rent paid to another property owner, rent and debt repayment are effectively repaid for a homeowner to himself or herself, with the home price appreciation as a "profit." An unusual situation indeed, but consider: an appreciating home allows individuals to realize gains on their home owning "business" over time, while a declining home market reverses this situation, and in effect produces a loss-making enterprise on home owning individuals.

For further background on this idea, consider that Net National Income -- which according to GDP accounting must approximately equal Gross Domestic Product -- without government and net exports consists of salaries and wages plus corporate profits. For money spend on goods or services, the company or individual selling the service or product must pay for materials, energy, transport, etc - which goes as costs to other companies if the costs are materials -- or to workers if the costs are salaries. This series of payments goes on until all income from sales can ultimately be divided into either wages and salaries or profits. Note that corporate profits have traditionally stayed around one-third of net national income.

Housing ownership moves most home owning wages and salaried individuals from being fully in the category of a) deriving income from mainly salaries and wages to ---> b) deriving income from in part from salaries and wages and in part from "corporate" (sole-proprietorship home ownership) profits.

Appreciating house prices mean that home-owning individuals have a substantial, profitable business supplementing their salaried income, while declining house prices means that individuals have a loss-making enterprise, subtracting from their income.

From 1992 to 2006, aggregate housing wealth in constant 2005 dollars increased from approximately $12.5 Trillion to $28.5 Trillion, accordi data from the federal reserve. The average yearly increase was therefore approximately $1.14 Trillion per year. The question is, how much of this increase contributed to consumption?

Table 1: Household Wealth, Consumption and Housing Price Appreciation in the United States from 1950 to 2008:

According to Haurin and Rosenthal of Ohio State University (2006), a survey of 4,000 home owners found that households took on an additional 15 cents of debt per dollar of home price appreciation to finance consumer expenditures. Assuming here a range of 10 to 25 cents per dollar of home price appreciation/depreciation leading to initial increases/declines in consumer spending, creased spending from housing contributed through a ripple-through effect (a spending multiplier of .10 based on a marginal propensity to consume of .90 - a savings rate of 10%) 11% to 29% of GDP.

If the spending multiplier says consistent through a housing downturn, a similar decline in GDP should have been experienced. The reasons why such a downturn did not occur is likely due to the unprecedented actions of the US Government through stimulus spending and actions on the part of the Federal Reserve to support asset markets.

Note that the loss' or gain's impact from home price appreciation/decline on GDP can be treated with the spending multiplier, if the income or income decline ripples through the economy. Other economists do not touch upon the multiplier, but as consumer spending from house price gains ends up as sales of products and services, which ends up as income for other business to be spent, and so on and so forth, likely there will be a ripple through effect from additional consumer spending.

So one could say, yes housing does impact the economy through consumer spending (as economists have traditionally modeled), but now we can make an estimate utilizing a profit/loss estimate from published data and a multiplier estimate (not solely relying on regressions of past data which could have flaws due to the fact that housing has appreciated for the past 70 years before 2007).

Secondary Effects of Home Price Declines:

Further, secondary effects of declining home prices means that non-real estate related businesses are impacted due to lower effective real income. In any community, a percentage of GDP is represented by local stores, grocery, automotive, convenience, clothing, food retail etc which depends on the consumer spending of the individuals living in that community. One could say that declining house prices can one of the following: produce either 1) a situation of a multiple equilibrium (where economic activity can stabilize at a significantly lower level that less income is available for purchase of goods and services). Oil markets, for example, are considered to have multiple equilibrium by several economists -- at $20 in the 1990's or near $100 per barrel currently -- due to he relative power of the oil producers with spare capacity (OPEC).

2) Declining home prices could produce a "vicious spiral" in which lower home prices continuously reinforce lower income (lower home prices --> lower consumer spending --> lower sales for community business --> lower income for workers in the community --> lower amount of money available for home mortgage payments --> lower home prices ---> etc) At one one economist (Mark Zandi) as expressed concern that declining home prices may be a downward spiral for economies.


If one models home ownership as a business with a certain level of profits or losses "marked to market" per year in an accounting sense, then with any level of spending multiplier it becomes clear that housing is a significant contributor to the US economy. It is estimated in this post that the contribution to GDP from housing is between 11% and 28% per annum of total GDP. This intuitively makes sense, in so far that housing represents approximately two thirds of the median household's net worth, and represents over $23 Trillion in aggregate value, in an economy with a GDP of approximately $14.5Trillion.

The impact of the decline of house prices - 30% from their peak over a 5 year period (with continued declines in the near future at least) -- could modeled as an approximate $1.8 Trillion loss of housing wealth per year in aggregate, which could be realized by home owners at between 10%-25% per year, with a multiplier of 10 would have resulted in a reduction of GDP of between 12% and 31%. So far, this reduction in GDP has not occurred, due to the unprecedented actions of the Federal Government and the Federal Reserve. However, if housing prices do not begin to recover before the end of current stimulus programs, it is possible GDP could once again decline due to the reduction of consumer spending and total income due to lower housing wealth.