The sky must surely be falling on the financial sector. Reported or estimated subprime related losses have, since last summer, gone from $50bn, to $100bn, $200bn, $400bn, even $800bn. Let's call it $1 trillion, or even $2 trillion, just to be sure we catch most of the likely eventual losses. What has not been reported is the matching subprime-related gains, which without a shadow of a doubt also follow the sequence $50bn, $100bn, $200bn, $400bn, $800bn, $1 trillion and $2 trillion. Why this failure to report the subprime-related gains?
One reason, no doubt, is that there is a lot of ignorance and stupidity around - the distinction between inside and outside assets appears to be a difficult one for economists, especially financial specialists, brought up in a partial equilibrium tradition. I am lucky in having had Jim Tobin as my PhD adviser and mentor. Balance sheet constraints, budget constraints, Walras' Law, adding up constraints - it was the bread and butter of what he taught. A little general equilibrium does go a long way.
The second reason is that the losses are highly concentrated among a few hundred commercial banks, investment banks, hedge funds and similar shadow banking sector institutions, while the matching gains are widely dispersed among the many millions of homeowners who owed the mortgages that have been written down or written off. Mancur Olson's Logic of Collective Action strikes again. In addition, many of the winners may not wish to advertise the fact that, given the amount by which the value of their property fell, they are better off now because they were able to force the bank that held their mortgage to eat their negative equity.
Inside and outside assets
For every financial asset there is a matching financial liability. That is, financial assets are inside assets. Inside assets are assets owned by a natural or legal person that are the liability of some other natural or legal person(s). Outside assets are assets of a natural or legal person that are not a liability of some other natural or legal person(s). When you ‘net out' all inside assets against the corresponding liabilities, you are left with just the outside assets, or the net wealth of the system. In a closed economy (foreign assets and liabilities present no conceptual problems but clutter up the argument), the outside assets are the stocks of natural resources (including land) and physical capital (residential housing, other structures, equipment, infrastructure), the human capital (the current and future labour endowments of the economy, that is, the resources embodied in current and future natural persons) and the productive resources (goodwill, synergy, monopoly power) embodied in legal persons such as incorporated firms.
There is an interesting argument as to whether the labour endowments of the unborn should be included among a society's outside assets. In a society without hereditary slavery, future endowments of labour embodied in natural persons yet to be born are not owned by anyone alive today, and therefore don't constitute private wealth. They can, however, be viewed as part of the tax base, because the institution of the state (and the associated power to tax) is likely to endure as long as mankind. That issue will have to wait till some future occasion to be treated in earnest.
So residential property is an outside asset and constitutes net wealth. A mortgage is a liability of the homeowner and an asset of the mortgage lender (bank). The mortgage held by the bank is an inside asset and does not constitute net wealth.
Assume the bank securitises the mortgages by selling them to an SPV that pools them and issues mortgage-backed securities against them (RMBS). Securities backed by residential mortgages are a liability of the SPV that issued them and an (inside) asset of whoever holds them, say an SIV owned by another bank. The SPV has as (inside) assets the mortgages it bought from the originator. The mortgages are still liabilities of the homeowner borrower. All CDOs backed by subprime mortgages (or by Alt-A or prime mortgages), by credit card receivables or by car loans are inside assets for which there is a matching liability. They are not net wealth. The cars themselves are net wealth.
Even a fall in outside residential housing wealth doesn't make you worse off
The US residential housing stock at the beginning of 2007 was worth around $23bn. Let's assume that its value has declined by 10 percent. There has therefore been a reduction in the value of this outside asset, of $2.3 trillion. I have argued elsewhere ("Housing wealth isn't wealth";”OK then, housing wealth is wealth, but not NET wealth!”; and “The coming decline in UK house prices: how large and how helpful?”) that, because this outside asset yields its future income stream in kind, in the form of consumable housing services, and because on average, home owners expect to consume (over their life time) the housing services yielded by the stock of housing they own, a change in the value of residential property on average does not make anyone better off.
A fall in house prices redistributes wealth from those long housing (for whom the value of the house they own - the present discounted value of the future actual or imputed rental income of the property - exceeds the present discounted value of the future housing services they plan to consume) to those short housing (for whom the value of the house they own is lower than the present discounted value of the future housing services they plan to consume). Simply put, a decline in house prices redistributes wealth from landlords to tenants. On average, an American household is a tenant in its own home. Changes in house prices do not make the average American better or worse off, unless there is a lot of ownership in US housing by non-resident foreigners, in which case a decline in house prices would make the average US resident better off. The same point has been made by many others, including Mike Buchanan and Themistoklis Fiotakis.1 It is also a viewpoint that, subject to all the aforementioned qualifications and further qualifications to be mentioned below) is shared by Mervyn King, the Governor of the Bank of England, who first explained the issue to me in 1997.
This argument is false if the decline is house prices reflects the bursting of a bubble rather than a reduction in its fundamental value (the present value of future rentals). In that case the homeowners lose the bubble value, without a corresponding gain for the tenants through lower present value of future rents. Other necessary qualifications come from the fact that the average expected remaining lifetime of housing consumers is likely to be less than the remaining lifespan of the existing stock of residential property. This is certainly true if the durability of the land is taken into account. In that case a fall in house prices can hurt homeowners more than it helps renters. But with reasonable discount rates, this effect is probably not very large.
Even if there is no net wealth effect from a change in home prices, this does not mean it will not have any behavioural effect. Unlike human capital, housing wealth can be collateralised. A lower value of residential housing, even if it does not make you worse off, may lower the amount you can borrow against the security of your property. Mortgage equity withdrawal becomes more restricted. This means that, through this credit or liquidity channel, falling house prices will have a temporary depressing effect on consumer demand (approximately, the level of consumer spending goes up with the change in house prices).
What banks lose on mortgages, mortgage borrowers gain
What follows is independent of whether you buy the argument that a change in house prices does not make the average American household worse off or better off. Mortgages, like any other IOU, secured or unsecured, are inside assets. If the value of the asset goes down for the investor (the bank holding the mortgage), the value of the liability goes down for the borrower (the homeowner who took out the mortgage with the bank). There is no change in net wealth, no economy-wide net wealth effect.
There has been $800bn worth of redistribution from banks and other mortgage lenders (and/or from those who invested in securities backed by the mortgages) to those who took out the mortgages (and/or from those who issued the mortgage-backed securities). The same is true for changes (up or down) in the value of any financial claim, bonds, options, CDS, complex financial structures like ABS, CDOs, CBOs or any of the other alphabet soup financial instruments. Changes in the value of inside assets, like RMBS, represents pure redistribution between those who hold them and those who issue them; the point is most easily seen for options and other derivatives. All financial claims can, of course, be viewed as derivatives that are in zero net supply.
Redistribution can matter for aggregate demand. It will not, in general be neutral. But the non-neutralities have to be documented and substantiated carefully. The size of the losses on inside assets by themselves (multiple trillions no doubt before this crisis is over) bears no necessary relationship to the size of the aggregate demand effects.
(1) The person owing a debt (a mortgage, in the subprime case) may not value it in the same way as the person owning it. In other areas there have been spectacular examples of this. Most workers enrolled in defined benefit company pension plans probably put a positive present discounted value on their expected future stream of pension benefits. For a long time, the companies that owed the matching liabilities kept them off-balance sheet. Out of sight, out of mind, and before long these future pension liabilities were not viewed as liabilities at all. The realisation that they were indeed unsecured liabilities has crippled much of the US domestic steel and automobile industry.
(2) When default risk increases but default has not (yet) occurred, the marked-to-market value of the bank's asset (the mortgage) goes down, but the borrower is still servicing the debt in full. While the homeowner owing the mortgage should also mentally mark it to market, that is, allow for the prospect that (s)he will service the mortgage in full in the future, the continuing full debt service in the present may, because of liquidity and cash-flow constraints, restrain household spending.
(3) Consider a household that purchases a home worth $400,000 with $100,000 of its own money and a mortgage of $300,000 secured against the property. Assume the price of the home halves as soon as the purchase is completed. With negative equity of $100,000 the homeowner chooses to default. The mortgage now is worth nothing. The bank forecloses, repossesses the house and sells it for $200,000, spending $50,000 in the process.
The loss of net wealth as a result of the price collapse and the subsequent default and repossession is $250,000: the $200,000 reduction in the value of the house and the $50,000 repossession costs (lawyers, bailiffs etc). The homeowner loses $100,000, his original, pre-price collapse equity in the house - the difference between what he paid for the house and the value of the mortgage he took out. The bank loses $150,000, the sum of the $100,000 excess of the value of the mortgage over the post-collapse low price of the house and the $50,000 real foreclosure costs. The $300,000 mortgage is an inside asset - an asset to the bank and a liability to the homeowner-borrower. When it gets wiped out, the borrower gains (by no longer having to service the debt) what the lender loses.
The legal event of default and foreclosure, however, is certainly not neutral. In this case it triggers the repossession procedure that uses up $50,000 of real resources. This waste of real resources would, however, constitute aggregate demand in a Keynesian-digging-holes-and-filling-them-again sense, a form of private provision of pointless public works.
(4) Continuing the previous example, how does the redistribution, following the default, of $100,000 from the bank to the defaulting borrower - the write-off of the excess of the face value of the mortgage over the new low value of the house - affect aggregate demand?
There is one transmission channel that suggests it is likely, had this redistribution not taken place, that demand would have fallen more than it does following the default. The homeowner-borrower is likely to have a higher marginal propensity to spend out of current resources than the owners of the bank - residential mortgage borrowers are more likely to be liquidity-constrained than the shareholders of the mortgage lender.
(5) Finally, we have to allow for the effect of the mortgage default on the willingness and ability of the bank to make new loans and to roll over existing loans. Clearly, the write off or write-down of the mortgage will put pressure on the bank's capital adequacy. The bank can respond by reducing its dividends, by issuing additional equity or by curtailing lending. The greatest threat to economic activity presumably comes from new lending.
The magnitude of the effect on demand of a cut in bank lending depends of course on who the banks are lending to and what the borrower uses the funds for. If they are lending to other financial intermediaries who are, directly or indirectly, lending back to our banks, then there can be a graceful contraction of the credit pyramid, a multi-layered de-leveraging without much effect on the real economy. If bank A lends $1 trillion to bank B, which then lends the same $1 trillion back to bank A again, there could be a lot of gross de-leveraging without any substantive impact on anything that matters.
With a few more non-bank intermediaries tossed in between banks A and B, such intra-financial sector lending and borrowing (often involving complex structured products) has represented a growing share of bank and financial sector business this past decade.
A group of people cannot get richer by shining each other's shoes/taking in each other's laundry. Similarly, financial institutions (‘intermediaries') cannot get richer by lending to each other. They can only get richer by intermediating, that is, by lending to the real economy. Of course, a more efficient structure of intermediation adds to the productive potential of the economy (by better matching savers with profitable investment opportunities), but the degree of efficiency of the structure of intermediation (markets and institutions) need bear no relation to the gross volumes of inside assets issued by the financial intermediaries.
Somehow, the financial markets and those buying shares in financial intermediaries forgot about the Mutual Shining of Shoes Theorem. A bubble or Ponzi finance scheme developed that caused the gross value of intermediation and leverage in the financial sector to rise massively. When the bubble burst, there was a loss of net wealth equal to the bubble component in the valuation of the financial sector. The subsequent de-leveraging and contraction of balance sheets does not, however, destroy net wealth.
Some of the lending of the financial sector went to the real economy - households and non-financial corporations. There will undoubtedly be an increase in the cost and a reduction in the availability of such lending beyond what we have seen already. The effect of this on spending by households and non-financial firms (consumption and investment) is not, of course, equal to the reduction in bank lending to these sectors.
There are other outside sources of funds for non-financial corporates, and both households and firms can maintain spending by reducing household saving and corporate retained profits respectively. So there is many a slip between the cup of the massive de-leveraging and inside asset blow-out in the banking and non-bank financial sector on the one hand and, on the other hand, the lip of private consumption and investment. I consider the estimate of David Greenlaw, Jan Hatzius, Anil K Kashyap, Hyun Song Shin in their paper "Leveraged Losses: Lessons from the Mortgage Market Meltdown", that a one dollar loss in bank assets reduces spending on goods and services in the long run by just under 44 cents, to be an order of magnitude too large; it also is bound to be far from a ‘structural' effect, that is, an effect invariant under plausible changes in the economic environment driving these two endogenous variables.
A little statistical rant (don't read unless you are interested in identification, endogeneity & simultaneity)
The authors calculate/calibrate a value for the ratio of total credit to end-users (either the non-leveraged sector or just households and non-financial corporates) to the total assets of the leveraged sector (banks, the brokerage sector, hedge funds, Fannie May and Freddie Mac and savings institutions and credit unions). They then treat this ratio as a constant, which means that once they have the change in the value of the total assets of the leveraged sector, they know the change in credit to the end-users.
The next step is the empirical estimation of a correlation between the growth rate in (real) credit to end users and the growth rate of real GDP.
There are just too many ways to poke holes in the empirical argument. To start with, as noted by the authors) the credit variable used domestic non-financial debt, includes financing from non-leveraged entities and therefore does not correspond to the credit variable of the theoretical story.
More painfully, the authors seem blithely unaware of the difference between causation and correlation, or prediction and causation. What they perform is, effectively, half of what statistically minded economists call a Granger causality test but should be called a test of incremental predictive content. They run a regression of real GDP growth on its own past values and on past values of real credit growth and find that past real credit growth has some predictive power over future GDP growth, over and above the predictive power contained in the history of real GDP growth itself: past real credit growth helps predict, that is, Granger causes, real GDP growth. Lagged real credit growth is (barely) statistically significant at the usual significance level (5%).
When you do this kind of regression for dividends or corporate earnings and stock values, you find that stock values Granger-cause (help predict) future dividends. Of course, anticipated future dividends determine (cause) equity prices, so causation is the opposite from Granger-causation.
The authors are undeterred and treat the estimate of GPD growth on credit growth as a deep structural parameter.
Even if the equation is taken as structural (which it almost surely is not), the demand for credit vs. supply of credit interpretation of the equation has to be settled; after all, credit is not only not an exogenous variable, it is not even a predetermined variable (given by history at a point in time but becoming endogenous as time passes). Unless there is no default risk, the value of the marked-to-market stock of credit is endogenous at a point of time, because it depends on forward-looking expectations. Even if there were no problem of endogeneity through forward-looking expectations, the credit supply vs. credit demand issue can only be resolved by brute force, that is, by imposing a particular interpretation (identification assumption).
The authors recognise the issue but completely fail to address it. They use the TED spread (the price difference between three-month futures contracts for U.S. Treasuries and three-month contracts for Eurodollars having identical expiration months - an measure of bank default risk), and a survey-based measure of banks' willingness to lend as statistical instruments for credit growth.
Instruments are variables that are highly correlated with the variable that you are trying to purge of endogeneity and simultaneity problems, but independent of the random disturbance in the equation you are estimating.
It is a well-known but ruthlessly suppressed fact in the econometrics profession, that there are no instruments - there is just implicit theorising. The correlation between the instruments and the variable to be instrumented (credit growth in this case) can of course be tested and reported, but the second key assumption - independence of the instruments from the disturbance term in the GDP growth equation - is untestable and simply has to be maintained.
Without boring the readers (if I still have any) with further details of why the empirical work is, at best, utterly unconvincing, let me report that the 3.0 percent contraction in credit growth ($ 910 billion) to the end-users the authors assume will result from the decline in the assets of the leveraged sector), will according to their instrumented equation, reduce real GDP growth by 1.3 percentage point over the following year - the 44 cents mentioned earlier.
The authors could be right about the effect of de-leveraging in the leveraged sector on real GDP growth, but the paper presents no evidence to support that view.
How do we value the outside assets?
In the case of residential property, house prices (the sum of the value of land and structures) provide all the relevant information. For physical capital, there is the problem that part of it (publicly owned infrastructure) is not priced anywhere. For privately owned capital, the asset should be valued at the present discounted value of its future earnings. Where the capital is held by unincorporated businesses or by unlisted companies, it is very hard to get an estimate of their value. When capital equipment is owned by listed corporations, it will contribute to the market value of the corporation, but only in conjunction with the goodwill and other going concern value of this legal person. The stock market value of the firm won't do either, unless the firm is 100% equity financed. Otherwise we have to add the value of the company's net financial debt to its equity. Valuing human capital (the present value of current and future labour earnings, either of those currently alive or of current and future generations) is a bit of a nightmare.
There can be little doubt, however, that net wealth in the US (and to a lesser extent in the rest of the North Atlantic region) has taken a beating. The value of the residential housing stock and of commercial property is down. The value of corporate debt plus equity is down. With employment falling and subdued wage growth, the value of human capital is also likely to be down, unless the appropriate stochastic discount factors act very strangely.
So let's quantify these net wealth effects of changes in the value of outside assets. Let's also study the distributional effects of the massive changes in the values of inside assets. But let's not forget that for every loser in the valuation game for inside assets there is a matching winner, and that the asymmetries don't all point to a stronger negative effect on demand. Defaulting mortgage borrowers, in particular, are likely to have high marginal propensities to spend out of current resources. Not having to service their mortgage debt any longer could give a major boost to consumer spending.
Things are tough enough without us exaggerating the problems through egregious double, triple, quadruple & higher multiple counting. Economic prospects for the US are poor, but nowhere near as bad as the growing crescendo of the moans emitted by the losers in the inside asset revaluation game would have us believe.
1 M. Buchanan and T. Fiotakis, House Prices:A Threat to Global Recovery or Part of the Necessary Rebalancing?, Goldman Sachs Global Economics Paper No 114, 15th July 2004