Risk-sharing mortgages: Mian and Sufi

Last week I wrote

In their recent book House of Debt, the US academics Atif Mian and Amir Sufi, argued that equity-sharing contracts should become the norm for housing finance.  They argue that such contracts would materially reduce the risk of financial crises, and that the main reason such contracts aren’t common is because of the tax system and the role of US government agencies.  I’m very sceptical of both claims

And here is why.

In the final chapter of their book (so it isn’t the focus of their analysis) Mian and Sufi advocate the introduction of what they label “shared-responsibility mortgages”.  Under these contracts, when house prices in the borrowers’ locality fall the borrower gets an automatic reduction in the principal amount of the mortgage.   The cost of this (what is, in effect, a) put option is covered by providing that five per cent of any nominal capital gain would go the lender (either when the house is sold, or when the loan is refinanced).  In some cases, that payment would go to the lender in just a year or two, but in other cases it might take many decades.

If such residential mortgages were widespread, no mortgage borrower would ever have negative equity in their house as a result of movements in the general level of houses  (severe neglect of maintenance, or specific issues in, say, the street your house was located in could still result in a small number of cases of negative equity).  Whenever general house prices (in your part of the city) fall, the loss is shared with the mortgage lender, and your equity share in the house does not change.

Mian and Sufi argue that this feature would have greatly reduced the severity of the 2008/09 Great Recession in the United States.  In their story –  the thrust of their book –  the severity of the recession was mostly due to the negative equity so many borrowers had once house prices fell, and the impact of those wealth losses on consumption.  I find that story unconvincing.  I’ll skip the detail here, but the paths taken by the New Zealand and the United States economies have been so similar since the mid 2000s, and yet New Zealand had very little sustained fall in nominal house prices, and few cases of material negative equity.  Given that, it is difficult to be confident that falling house prices, and associated wealth losses, are a key causal factor explaining why economies are still lagging so far behind pre-recession trend GDP levels.   For example, fluctuations in house-building activity –  booms and busts –  are much more important than Mian and Sufi recognise.

But I wanted to focus on the suggestions that shared-responsibility mortgages (SRMs) would materially reduce the risk of financial crises, and that the main reason they don’t exist is the role of various government interventions.

Why might the risk of future financial crises be reduced?   They argue that

the downside protection in SRMs would lead lenders to worry about future movements in house prices.  If house prices plummet in the near future, then more recently issued mortgages would generate the greatest loss for the lender.  The lender would have to be very mindful about potential “froth” in local housing markets, especially for newly-originated mortgages.  If lenders fear that the market might be in a bubble, they would raise interest rates for new mortgages in order to cover the cost of the increased likelihood of loss.  SRMs would therefore provide an automatic market-based “lean against the wind”.

To which I would make a few points in response:

  • Newly-issued mortgages have always been the ones at most risk of loss to the lender (borrowers borrow to the hilt, especially to get into a first house, and then typically see their economic position improve over time, through rising nominal wages and house prices, and gradual principal repayments).
  • The nature of very frothy markets is that no one pays much attention to downside risk (or at least, they pay rather less attention to it than in more normal times).  Recall Ben Bernanke’s pre-recession scepticism about the idea of any widespread house price overshooting in the United States.   In periods of extreme optimism and persistent rises in house prices, there may be at least as much lender focus on booking the 5 per cent capital gain as on the possibility of a future loss.
  • Risk isn’t changed by the introduction of an SRM, it is just reallocated.  In principal, lenders should be somewhat more wary about the downside risk, but borrowers have less reason to be concerned about things going wrong.  If house prices fall, their equity will be impaired but- by design – by nowhere near as much as under a conventional mortgage.  There is at least an arguable case that lenders are better placed to bear risk than borrowers, since they can diversify across individual borrowers and geographically.  Mian and Sufi offer no reason to think that the net effect of the reallocation of risk would be to reduce overall risk-taking in the house finance market in boom times.

Mian and Sufi outline one other argument that might act as a modest dragging anchor.  Any  refinancing would trigger the need to make the 5 per cent capital gain payment to the lender, so any cash-out refinancing (drawing more on the mortgages to buy a boat or finance a fancy holiday) would involve meeting the 5 per cent payment.  I suspect that (a) any benefits would be small, and (b) that if such SRMs ever came to market the option premium would probably end up built into the initial value of the loan (you need $300000 to purchase your house, but the loan is booked as $304200[1], and serviced over time on that basis).

The second element of the Mian-Sufi argument is that such  products have not emerged mainly because of mortgage interest deductibility for owner-occupied houses, and because of the dominant role of the agencies in influencing the structure of US mortgage contracts.  The authors report that interest deductibility is available in the US only when the home owner bears the first losses when house prices fall.

These may well be factors that impede the emergence of such contracts in the United States, and no doubt the tax laws could be revised in ways that would bear less heavily on the prospects for SRMs.  But Mian and Sufi’s argument is a very US-centric perspective.  We do not see such contracts having emerged in other advanced economies in which the state has little or no direct role in the housing finance market, and where interest on mortgages on owner-occupied properties is not tax deductible.  New Zealand (in particular) and Australia and the United Kingdom spring to mind.  Bank regulators might not like such products greatly, but capital adequacy frameworks cope with options in other markets, and products like SRMs did not exist either in the decades before risk-weighting and capital adequacy frameworks assumed a key place in the bank regulatory framework.

Robert Shiller has long argued for the emergence of a fuller range economic derivative contracts (house price futures, nominal GDP indexed bonds, and so on), but few of them have emerged.  Revealed preference is a powerful insight.

The Mian and Sufi  SRM is in this tradition – an interesting idea, which seems to have some appeal for some borrowers, few or no regulatory/tax obstacles in many countries, and yet they just have not emerged.  It is interesting to think about why?  Personally, I suspect SRMs have not emerged because conventional mortgages are not “horrible instruments” (Mian’s and Sufi’s term) at all but very attractive and effective instruments.     They have proved to be only moderately risky over many decades, at least  in systems (unlike the US in the 90s and 00s) where government mandates don’t try to override market judgements on credit quality.  In market system the risk around conventional mortgages is managed through lender decisions around initial LVRs and servicing capacity (and, of course, overall capital holdings.  Conventional mortgages also require limited amounts of ongoing monitoring.    Borrowers typically have the most to lose: on any individual loan a new borrower may have only 10% equity in the house so the lender can lose more dollars, but the potential loss for a borrower (in New Zealand or other with-recourse markets) is everything (all their assets).  The system has worked well for many decades, so what would be the impetus for change?  In some ways this comes back to Calomiris and Haber, and the suggestion that the US financial system was made fragile by design.  As they document, the situation in other countries is rather different.

Revealed preference is a powerful insight, but….as an analyst I’m still a bit puzzled why inflation-indexed mortgages haven’t emerged (in countries like New Zealand or Australia).   That might be a topic for another day.

Finally, and harking back to Islamic banking, I presume that SRMs would not be sharia-compliant.  They add some additional uncertainty regarding the future value of the loan but an interest rate is still a key feature of the structure.   Proper equity-finance structures for residential properties still look expensive to establish and monitor, and unlikely to emerge on any scale.

[1] Mian and Sufi estimate that the upfront option premium for the protection the SRM offers would cost 1.4 per cent of the initial value of the loan, on historical US house price performance

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s