Tuesday, April 14, 2015

Housing Tax Policy, A Series: Part 26 - Rent Inflation as a Signal of Supply and Demand

There is the problem in macro data of considering the effect of inflation on different types of households.  We tend to discount nominal income numbers with national inflation measures, but over long periods of time, we really don't have a good idea of how cost of living has changed for different households.  Depending on geography, socio-economic status, etc. the effect of individualized cost of living adjustments might even dwarf the kinds of trends in incomes that dominate public discourse.  I was reminded of this recently when I was looking at rent inflation statistics.  Rent for housing amounts to about 30% of the component weights in the CPI.  And, this category is specifically divided between home owners (Owner Equivalent Rent for Primary Residence - about 23%) and renters (Rent for Primary Residence - about 7%).

So, for renters, 23% of the CPI is, by definition, a cost that is entirely missing from their actual basket of goods and services.  Now, over time, housing supply for owners and rents does have some marginal substitutability, so they have some relationship.  But, they can diverge quite a bit.  Owner rent inflation cumulatively moved higher than renter inflation by nearly 10% in the late 1990s, but renter inflation has caught up with it since then.

I think the difference between these inflation categories can be meaningful in my eventual analysis of tax policy's effects on home prices.  There are types of housing that are more amenable to renting, usually multi-unit structures.  These generally have more community areas and less privacy, which act as a natural mitigation to the principal-agent problems that arise from having a tenant who is not the owner.  The close neighbors and property managers help regulate the tenant/owner relationship.

For home owners transitioning from renter to owner, ownership itself adds value to the property by eliminating these principal-agent problems, and this added value is maximized by owning a separate, private residence that isn't regulated by close neighbors and managers.

Obviously, on the margin, there are counterexamples.  A small portion of the real estate market consists of condominiums.  And, many neighborhoods try to capture some benefits of communal regulation through Homeowners' Associations.  But, for our purposes in this analysis, the fact that the owner and renter market are highly segmented is useful.

Looking at the difference between owner and renter inflation, we see a sustained rise in owner inflation from the mid 1980s to the mid 1990s.  This coincides with the new importance of the mortgage interest deduction in 1986.  So, we can think of this as a signal of added demand in the homeowner market, as households moved to capture the added after-tax value of owning a leveraged home.  We would look to see how much of an increase in homeownership there was, and estimate the added consumer surplus captured by homeowners.

But, surprisingly, during this period, there was no rise in the rate of homeownership.  We need to think about the effects of the mortgage tax deduction carefully.  As this paper notes, and I discussed here, the mortgage interest tax deduction is overwhelmingly captured by high income households.  In fact, real estate leverage tends to increase with income.  So, this tax benefit went overwhelmingly to households who already had pretty universal homeownership rates among households who wanted to be homeowners.  So, the effect of the mortgage interest deduction is pretty simple.  The entire benefit went to owners.

If there had been an increase in ownership, we would need to estimate what marginally extra dollar of after tax value enticed each marginal renter into ownership.  If the tax benefit added 5% to the nominal value of a property, it may be that renting was worth 4.9% more to the individual, and so a 5% increase in owner-occupied housing consumption would only represent a 0.1% increase in utility for the household.  But since ownership was flat, we can estimate the gain to homeowners with the rise in owner-occupied housing consumption.

The benefit had two effects.  The demand for homes by owners shifted to the right, because home consumption was now relatively less expensive, after taxes, than other forms of consumption.  This increased real housing consumption.  But, to the extent that there are limits and costs to the availability of housing, because of the scarcity of land and materials, regulatory limits, etc. some of this extra demand will show up as inflation, specific to owned properties.  That shift upward in owner consumption would also be associated with a decline in renter consumption to the extent that there is competition for supply between the two markets That is what we see in the graph above for the decade after the mid 1980s.

This chart suggests that amounts to more than 1% of GDP.  The Wharton paper mentioned above puts it at about 1% of GDP in 2003.

The value of the non-taxability of imputed rent is a much larger value.  (The Wharton paper estimates it to be nearly twice the value of the mortgage subsidy.)  This would explain why the mortgage interest deduction didn't move homeownership higher.  There were already such large tax advantages to ownership that any household that valued ownership at all (and even some who would have preferred renting in a tax neutral context) were already homeowners.

So, while this increases the possible effect of the mortgage tax deduction, compared to the estimates I used at the beginning of this series, the effect of the capital gains exemption on housing might be less than I thought.  As we can see in the first graph, this inflation signal goes away after the mid 1990s.  I think what has happened is that the capital gains tax rate on non-housing assets was reduced at about the same time that the exemption was strengthened in housing.  These may have had offsetting effects.

So, while there may be less of a capital gains exemption baked into current home values, this could suggest danger for future capital gains tax rate increases.  If capital gains tax rates are increased, it could lead to a new inflow of capital into housing which we avoided in the 1990s because of the tax rate reductions.

So, what did happen in the 1990s and 2000s?  One of the surprising details of Matthew Rognlie's recent work that I think has gone unnoticed is that, while it is true that the transfer of income from labor to capital has gone to housing capital, not to corporate capital, none of the gains to housing capital came during the housing boom.

As we can see in Rognlie's graph here, and in the following graph from one of my previous posts, from the mid-1990s to 2007 total capital income to housing was level, at best.  The gains to housing capital came before 1995 and after 2007.

I have explained how the pre-1995 gains might have come from the inducement from the mortgage tax deduction for homeowners to increase their nominal consumption.  The gains after 2007 are clearly excess gains from limited access.  Homeowners are earning excess returns because supply is limited by the hobbled mortgage market.

So, the boom of the late 1990s and 2000s wasn't a product of new demand from homeowners.  Nominal home values were being pushed up by outside forces (manifest in low real long term interest rates, mainly). This is corroborated by the fact that owner inflation was not high during that time, and owner consumption (gross rent) leveled off after 2000.

As I have argued previously, the sharp rise in nominal home values, and frictions in the housing market made house prices sticky and kept supply from rising enough to meet demand. So, interest rates were pushing up Price/Rent ratios, and the lack of supply was pushing up rents for both renters and owners.  The flippers and speculators weren't creating a bubble, they were just capturing the arbitrage profits of sticky prices (which is a more reasonable description of typical speculator behavior across markets to begin with).

This is mistitled.  Home prices are not included in the graph.
Here is a comparison of shelter inflation with Core CPI, which removes some of the food and energy noise out of the comparison.  Notice that after a period of shelter inflation from 1985 to 1988, there isn't any net excess shelter inflation until the mid 1990s.  This suggests that, even though households weren't switching from renting to owning, there was a gradual shift among homeowners to consume more housing while renters shifted to consume less, because of the unbalanced effect of the new tax subsidy on demand that favored owners and caused prices to rise in properties held by owner-occupiers relative to rented properties.

Homeownership started climbing in 1995, which seems to be related to policies like the evolving Community Reinvestment Act.  This might have led to some of the higher owner inflation that persisted until about 1997 (in the first graph above).  But, after that, shelter inflation in general continued to run high, but not particularly for owners.

I have come to the opinion that the CRA was a largely beneficial policy.  It looks to me like it moved several million middle-to-upper-middle class households into homeownership who were previously held out of the market for home ownership for reasons other than price.  That is all for the good.  The fact that much of the increase in the homeownership rate increase happened before nominal home prices began to rise and much of it happened without leading to owner-specific rent inflation suggests to me that these marginal new owners weren't pushing up the Price/Rent ratio on homes.  They were consuming homes as owners instead of renters, so there was some decline in renter expenditures and an increase in owner expenditures, in terms of Gross Rents / GDP.  And, there was some total increase in Gross Rents/GDP because access to home ownership was allowing these households to capture the gains from eliminating the principal-agent problem that exists in rented residence.  But, the intrinsic Price/Rent value of the homes these households were buying wasn't any higher than it had been for existing homeowners, so this movement into homeownership didn't, itself, lead to rising implicit rents and home prices.

All of this suggests that, outside of the immediate aftermath of the GDP collapse in 2008, there has generally been a shortage of housing, first from the inability of prices to adjust in the face of falling long term real interest rates, and second from the collapse of mortgage funding.

I have previously been somewhat muted in my expectations for housing starts going forward, because of widely known demographic trends, but there could be a large amount of unnoticed housing demand that has been waiting for 20 years to be supplied.  The recovery for homebuilders, if the banks are allowed to fund it without everyone crapping themselves about it, could be much higher than the current marginal expectation.

Further, if the 10% cumulative rise of owner inflation over renter inflation in the late 1980s represents the tax arbitrage value of the mortgage tax deduction, then we should expect a further 10% recovery in owner-occupier home prices above the prices of renter-occupied homes, if mortgage markets are allowed to begin to fund growing levels of home purchases again.

PS.  I suspect that there are many errors here, as this is quite complicated.  If you know of a technical correction that should be made, please note it in the comments.  That is your fee for having read it, accepted by the author on the honor system, as a free will offering.  If you gained negative utility from reading this, however, please do me a favor and refrain from giving bad advice in the comments in retaliation.  I am resigned to your goodwill.


  1. Always so much to think about in your posts.

    I do owner if housing demand is flexible, even beyond the obvious ways. For example, it has alway made sense o me that elderly should share apartments etc. As the baby boom gets old, maybe they will--it will become socially acceptable (I am not talking about sex here, just about living arrangements). If people per household rises, you get less housing demand.

    I like your comments on the home mortgage interest tax deduction.

    I also think an ownership society is better than a renter society, for a lot of reasons.

    But then I also think there is a percent of the population that belongs in organizations like the Army, in which housing, medical, eating, pensions, everything is provided for. A lot of people cannot take care of themselves in a modern society.

    1. Thanks Benjamin.

      On your last paragraph, I think we all spend part of our lives in that category, whether because of maturity (or lack of it) or just life circumstances, we wouldn't benefit as an owner. That could be in college or in the Army, or just being a bachelor in his 20s. Homeownership in the mid 60s range has been the norm for quite a while. I wonder how much of the remaining 35% would not choose ownership even in a perfectly liquid market? It must be most of it.

      On the baby boomers, I think this is interesting, because we are seeing the effects of their movement out of the labor force now. But, I think their movement out of single family home ownership will tend to lag that by a couple of decades. Even if homebuilders come back like I suspect they might, I suspect that 20 years from now, regardless of how well the economy is doing, homebuilders might see some pretty lean years.