Wednesday, May 20, 2015

Housing Tax Policy, A Series: Part 32 - Macroprudential regulations will continue until morale improves

I was depressed to see this headline today (FT, Reuters): "Banks calls for global coordination of bubble-busting measures".

While the banking executives did include warnings about the need to be careful in implementing these measures, the very first sentence of the Financial Times article kicks off with the typical, if unlikely, interpretation that high demand for low risk assets is a sign of high risk taking:
A group of leading financial executives have urged authorities around the world to bolster their crisis-busting arsenals amid fears that ultra-low interest rates have increased the risks of financial instability. 
One of the proposed regulations would be a limit on mortgage levels relative to incomes.  It is common for banks to use guidelines for debt expenses relative to income.  But, this would limit the size of the mortgage. The FT article says, "Authorities in countries ranging from the UK and Switzerland to Israel and Hong Kong have been making greater use of these regulatory levers to curb rising asset values, especially in the housing market."

As I mentioned on a recent post, while we tell ourselves there isn't generally widespread support anymore for explicit price controls, this sort of thinking really is a softer version of the same thing.  And, this is so wrong-headed.  I have outlined here many times how home prices, more or less, behave like a very long term real financial asset - because that is what homeownership is.  We shouldn't expect anything else.  Yet, everyone seems to simply accept the idea that prices in the 2000s were crazy - "exuberant" - because nominal prices were high.

Let's apply that logic to bonds.  Here is a graph of the price of a 10 year bond, with a $100 face value and a 10% coupon rate.  Why, look at that!  The market price doubled between 1985 and 2010.  This is the price for a certificate with a fixed level of income - $10 per year.  This must be irrational exuberance!  Why would sane investors pay twice as much for the same income?

So, what would be the best policy here?

1) Don't do anything.

2) Correct the irrational exuberance with a tight money policy.

3) Make a rule that banks can't sell these bonds for more than $140.

Unfortunately, #2 is exactly what we did from 2006 to 2008.  Imagine if we had done policy #3 instead.  Bond issuance would collapse if we did that to the bond market.

So, I would expect new home building to dry up if this rule became constraining.  But, housing has a stable existing stock.  I suppose in the housing market, the market reaction on existing homes would be to treat them as if there was a rolling call option at the price level that would tend to trigger the loan limit rule.  If long term interest rates remain low enough to make this policy constraining, then housing supply would continue to be inhibited, and rent inflation would continue to climb.  But, before I think through this more, I want to look at a few graphs to show how misplaced this entire concern is.

First, here is a measure of the mortgage payment required to buy the median new home, as a portion of median household income.  This was not particularly high in the 2000s, and it has been extremely low since then.  Since home prices in the 2000s were mostly a product of low real long term interest rates, they didn't require large mortgage payments, because mortgages also had low rates.  This is basic asset/liability matching.  The funny thing is, we have this whole set of controversial quasi-public institutions set up to create this asset/liability match, and now we ignore the fact that we have it when we interpret activity in the housing market.

Using Federal Reserve and BEA data, we can compare estimated rent and mortgage payments all the way back to at least 1950.  Mortgage debt service was much higher in the 1970s.

Here we can see that the way to lower home prices is through very loose monetary policy.  If housing was a problem that needed to be solved (it didn't), then the solution would have been to create expectations for 5% inflation.  Mortgage payments for new homes would have been higher, so there would have been less pressure on home prices.  And, for households with variable mortgages, their nominal home values and incomes would have gotten a 5% annual boost to help alleviate any difficulties.

But, I want to go back and think about what effect a loan size to income limit would have on the housing market if it was a tighter constraint than debt service to income ratios.

One of my running themes here has been that housing does tend toward a no-arbitrage price level - the aggregate market is relatively efficient.  In the late 1970's and 1980s, real long term interest rates were almost as low as the 2000s, (although very high inflation and inflation uncertainty muddy the picture). This would justify a high Price/Rent ratio.  It appears that home prices still found an efficient level in that period, and that households tended to downsize in order to settle at a debt service level that was manageable.  This makes sense.  There would have been tremendous incentive to own property that would appreciate along with inflation.

But, if a loan size limit becomes the constraint, households buying with leverage won't be able to bid the prices of homes up to the efficient level.  So, when the efficient price moves above the level that regulators would allow a mortgaged purchaser to fund, wealthy households would have an advantage, because they would be able to use cash equity to bid the prices of homes higher.  Households without a large pool of savings would need to downsize in order to buy homes at the market price (because the price would reflect some demand from cash buyers).  Homeowners in general would earn excess returns because limited mortgage availability would limit demand and prevent the efficient price from being reached.

It would like a combination of today's housing market and the early 1980s market.  There would be less new homebuilding because of the artificially low price level, which would cause rents to continue to climb.  That would possibly be mitigated by the downsizing among new owners.  But, just as there now is, returns to ownership would be very high, so owning a home would be lucrative, especially for institutional owners who could obtain outside financing.  So, there would be a tendency for homeownership rates to fall and for institutions and households with high net worth to capture above-market returns.

If the inhibition of supply was strong enough, this would lead to a vicious cycle of rents rising faster than incomes, so that marginal households would qualify to buy smaller and smaller homes.  This would further erode the ownership rate and produce high returns for well-funded owners.

Marginal households that did manage to purchase highly leveraged homes would see especially large returns because they would reap the excess returns on home ownership but their interest expense would not reflect a premium, since there would not be a constraint on the banks themselves to issue mortgages.  But, I think an end result of these trends would be to push back toward a pre-HUD context where ownership was less mortgage-dependent and less owner-occupier based.  Eventually, landlord funding and organizational foundations would be available for landlord owners to earn those leveraged excess returns.

To the extent that the landlord market developed and housing prices were bid to near-efficient levels, there would be lower rent inflation and lower homeownership rates.  To the extent that this didn't happen, marginal households would tend to purchase downsized homes, rents would tend to rise over time, and real estate owners would earn excess returns.

I wonder if these possible trends, themselves, (stagnant housing stock, less credit expansion and more equity ownership) would tend to have a downward influence on real long term interest rates, adding to the vicious cycle of pushing home prices above the loan constraint.


  1. The housing cost/new home payment stability is a shot against supply constraints. If there was a supply issue you would expect that buyers would compete and bid prices up and would be willing to spend a larger, not the same amount, of their disposable income (unless you posit that there was near zero consumer surplus for housing).

    1. Keep in mind that these are as a proportion of rent. (Price/Rent and Mortgage Payment/Rent) The supply shortage would play out in rising rents. It would lead to rising prices and mortgage payments, but generally in proportion to rent. These ratios should be mainly related to real and nominal interest rates.

    2. Are you saying that the graph labled Medium new home mortgage payment/Median household income is really mortgage payment/rent/median income?

    3. Oh. Yes. That graph is mortgage payments / income. But, there, it would depend on elasticity of demand, wouldn't it? Wouldn't we expect elasticity of such a large budget item to be pretty close to unitary? I would expect total nominal household spending on housing (in terms of rent) to be pretty stable. So, with supply constraints, real consumption would decline, prices (in terms of rent) would be inflationary, and nominal spending would be roughly level.

      So, mortgage payments / income reflect inflation expectations more than anything. When high inflation caused mortgage payments to hit cash flow constraints in the late 1970s, it looks to me like Price/Rent still reflected low real interest rates, and households were downsizing in order to purchase. I haven't completely worked out the data from the period, but it appears to me that households, in the aggregate, really do account for the complexities of inflation premiums, expected capital gains, etc. so that rent (and imputed rent) is the primary signal of supply and demand.

    4. I would expect housing demand to be relatively inelastic- it is difficult for the population to consume less housing as a whole because the market is dominated by the stock of housing, not the flow. The primary way for a nation to reduce total housing consumption would be to have the person per square foot of housing ratio rise (iirc we saw the opposite early in the 2000s).

    5. I think those are interesting issues. On the real spending issue, you are right. But, I am talking about spending as a portion of total income, so real spending can fall just because the denominator is rising. In this post:
      there is a graph of real housing spending, according to the BEA, which has fallen from 24% to 18% of PCE since the early 1980s.

      The square footage thing is complicated (isn't it all?) because sq. ft. is only one aspect of housing consumption. Lot size is important. But, probably most important is location. So, I think one of the big issues has been the sharp supply constraints in the high rent metropolises. Building a 4,000 sq. ft. house in Modesto instead of a 1,000 sq. ft. house in San Jose is actually a reduction in real housing consumption.

      Then you have the issue of low long term real interest rates, which really work through the price of land, because the cost of building is fairly stable. My personal observation in the Phoenix area has been that in the last 15 years, the lots have gotten really small compared to homes. Many homes just have a patio. Within their housing budget, households are substituting square footage for lot size, because within "housing" consumption, the lot has become much more relatively expensive. So, counterintuitively, to the extent that home prices were rising because of real interest rates, we would expect square footage to rise.

    6. To avoid the Modesto to SJ issues- let us talk same home sales. If the nominal price of an individual house doubled- where ever its location- due to insufficient supply then we would expect that the winner of the auction would be paying "more" for the same house. The most likely way for a supply shortage to manifest for a good composed primarily of stocks would be for that good to take a higher portion of income. If incomes were rising faster than home costs were rising, then yes you could have that ratio flatten. But real median incomes weren't rising- they peaked in the late 90s in the US. If real rents were rising post 2000 (which you have documented, right?) then how does all this fit together?

    7. There are a couple of issues here, I think. One is that a significant reason why real incomes have been flat is housing inflation. Low income households spend more on housing, so this is also a decent factor in the drag on median incomes.

      Second, the way I treat this is different than some people do. I think my way is the only correct way, though. And that is to be very careful about separating housing consumption from home ownership. Housing consumption is measured by rent. Home ownership is more like a financial security. Rent is fairly stable. So, over long periods of time, paying more for a home can be a function of higher nominal housing consumption. But, when homes double in price while rents rise 5% or 10%, that is just an increase in the nominal price of the security of pre-paid rent, and it has nothing to do with housing consumption.