Some ballpark math on the US financial sector (i.e. just how big is $700b?)

The Federal Reserve’s flow of funds data indicates that US households have $11.2 trillion in outstanding mortgage debt on non-farm homes, another $2.4 trillion in outstanding "other mortgage debt" (a category that includes corporate farms, who knew ... ) and $2.5 trillion outstanding consumer credit.

Figuring out how much households owe is the easy bit. Figuring out who owns the debt of US households is a bit harder.

For now, I am going to focus on the $13.6 trillion in mortgage debt. The Agencies have $7.9 trillion in debt and guaranteed bonds outstanding, but the GSEs hold around $800b of the debt that they guarantee, so their total holdings of mortgages and mortgage backed securities is more like $7.1 trillion. Like the Agencies or not, no financial institution is going to go bust holding Agency debt. That leaves around $6.5 trillion of outstanding mortgage exposure in the hands of the financial system, give or take. It might be higher because it is possible to create "synthetic exposure" to various kinds of securities -- basically, financial institutions can make a side bet among themselves about the future value of a mortgage backed security. That exposure nets out, but the netting only works if the losing party to the bet can pay up (Note: this paragraph has been edited in light of the comments, which highlighted that my initial post failed to reflect the Agency backed mortgage pools held by the GSEs)

This math is not inconsistent with the Fed’s data on the outstanding stock of asset backed securities. There are $4.365 trillion in asset backed securities outstanding. Not all those contain mortgages, but many do. Table L 126 suggests that ABS issuers hold about $2.1 trillion in residential mortgages, another $640b in commercial real estate and a non-trivial $400b in Agency debt.

This math also seems consistent with data indicating that the commercial banks hold $3.7 trillion in mortgages and another $1 trillion in corporate bonds (a category that should include ABS) -- i.e. up to $4.7 trillion in exposure. The thrifts report about $1.2 trillion in mortgage exposure -- mostly from mortgages, "private" MBS and collateralized mortgage obligations (CMOs) sum up to under $100b -- see table LII4). The broker dealers have $270b in corporate bonds (think ABS) -- not a huge exposure. But they over half ($1.6 trillion of a $2.8 trillion total) of their assets is just labeled other.*

Willem Buiter thinks that the US Treasury will be buying up assets at roughly 33 cents on the dollar, which would broadly speaking move $2 to $2.1 trillion in face value of debt off the balance sheets of major US financial institutions. Buiter:

I assume that $700 bn will allow the purchase by the US Treasury (or its agents) of at least $2 trillion worth of mortgage-related securities at face value, as it would not make sense for the US tax payer to pay much more than 33 cents on the dollar for the mortgage-related rubbish that banks have loaded onto their balance sheets.

That would leave, on first approximation, $3.7 or $3.8 trillion of mortgage exposure in the hands of the financial institutions.

I am not though convinced that the banks can afford to sell at 33 cents on the dollar. If these assets were valued at par before the crisis, I think that implies taking an aggregate loss of $1.4 trillion -- which seems much higher than the losses that the banks have recognized to date. The hit to the banks balance sheet might be too big.

My guess is that the Treasury’s average purchase price will be a bit higher -- which implies that fewer bad assets will be moved off the aggregate balance sheet of large financial institutions.

What is the point of all this?

Well, to me it helps to highlight the challenge the Treasury faces. If it pays a high price for various dud assets, it won’t move nearly as much off the banks’ balance sheet -- which may leave residual questions about the health of key institutions. On the other hand, if the Treasury pays a low price, it may leave a lot of banks in trouble and in desperate need of new equity.

It also helped me try to think through whether $700b is a lot a money or a little bit of money, relative to the enormous challenges the US now faces supporting a financial sector that is gravely ill. I was reminded just how big the balance sheet of the US financial sector is -- and just how much of that balance sheet is tied up in the real estate market.

Finally, I wanted to see if others found this kind of analysis - call it the beginning of a balance sheet analysis of the US -- at all helpful to a broader range of people before I try to go further.

*Since q2 2007, the broker dealers have reduced their holdings of corporate bonds from around $420 to around $270b -- a meaningful fall. From q2 2007 on, their agency holdings soared from $120b to $310b. Total assets though from $3.2 trillion in q2 2007 to $2.9 trillion in q2 2008, so there was a bit of deleveraging, not just a shift toward Agencies. That though likely corresponds with the collapse of Bear. The bigger trend is probably less a brutal deleveraging and more the end of the rapid growth of broker-dealer balance sheets.

From the end of 2003 to q2 2007, the broker-dealers total financial assets (and liabilities) basically doubled. That seems to correspond with a rapid increase in their aggregate leverage. The expansion of their balance sheet was especially rapid in 2006 (up over $600b) and the first half of 2007 (up $400b). That expansion led to a roughly $100b increase in the broker dealers holdings of corporate bonds, but most of the growth came from "other" assets. See table L. 129. By comparison, figuring out what China holds is easy ...