Wednesday, June 10, 2015

Housing Tax Policy, A Series: Part 38 - The World Record for Reasoning from a Price Change

As I was reading the 1000th account of our supposed pre-crisis over-investment in housing this morning, I realized that this whole chapter must be the most gargantuan example of reasoning from a price change, ever.  It wasn't the level of residential investment that triggered the panic about the housing market.  It was the change that we saw in prices.

I have argued that rising Price/Rent ratios were generally exogenous to the housing market - related to broader trends in interest rates and global capital flows.  Given the change in Price/Rent that we have seen, how would the reaction to the housing boom have changed if there were fewer constraints on housing expansion?  What if, instead of seeing higher rents on a more constrained housing supply we had seen falling rents on a less constrained housing supply?  Residential investment would have been higher, because there would have been more building.  Nominal price increases on homes would have been lower.  And, as a result of all of that extra residential investment, there would be much less moaning today about overinvestment in housing.  Reasoning from a price change leads everyone 180 degrees to the wrong conclusion.

Home Prices were high because Rent Inflation was high

In a previous post, I presented the case that the rise in home prices can be entirely accounted for with (1) the effect of long term discount rates on intrinsic values and (2) the full effect of rising rents on home values.  This is one of several graphs included in that post.

But, I neglected to take the next step in that analysis.  What would the counterfactual be if rent inflation had matched the inflation rate over this time of all the other non-rent elements of core CPI?

Next is a graph of home prices, nationally and in the Case Shiller 10 city index, along with a model of home prices based on long term discount rates, rent levels, and rent inflation rates.  And I have added the modeled national home price level, based on a counterfactual where rent inflation had equaled Core inflation (excluding rent).

How much concern would there have been about overinvestment in housing if home prices had only doubled in 20 years, instead of quadrupling (basically tracking inflation)?

Even the high real estate values that have persisted since the crisis would not have materialized, because the high intrinsic values now are a product of very low long term interest rates that are themselves the product of the collapse of real estate capital markets.

Rent Inflation was high because Residential Fixed Investment was Too Low

The amount of additional residential investment that would have been required in order to reduce rent inflation to levels to this counterfactual level would be a product of the elasticity of housing demand.  The next graph compares real and nominal housing expenditures.  Nominal expenditures have been level since the early 1960s.  (Homeownership rose after the Great Depression until it reached about 63% in the mid-1960s, which is roughly where it has remained since, except for the temporary push up to about 69% in the 2000s.)  Since 1995, when the latest period of rent inflation began, centered around the major coastal metro areas, real housing expenditures have fallen sharply while nominal expenditures remained level, as a proportion of personal consumption expenditures.  The long stability of nominal housing expenditures suggests a unitary elasticity of demand.  This is apparently slightly higher than the typical finding of slightly less than 1.  Households in the problem cities spend about 5% more of their incomes on housing than households in the rest of the country, which also suggests somewhat inelastic demand.  Possibly, national nominal spending on housing over this period increased slightly as a result of added tax benefits to home ownership.  This could explain why slightly inelastic demand hasn't led to a slight decline in nominal spending on housing as real housing expenditures have declined.

Long term elasticity of housing supply is generally found to be very high, which should make persistent rent inflation implausible.  This mystery fits well into the narrative that I am developing that regulatory limits to housing in the large metro areas is the driving factor in persistent rent inflation, and that rent inflation reflects the relative substitutability of real estate outside these areas for the preferred locations where building has been limited.

The average is mis-labeled.  It is for 1962-1995.
The next graph shows residential fixed investment over time.  When we eliminate the cyclical movements, I think this suggests that expenditures are very sensitive to small changes in residential fixed investment.  From 1962 to 1995, residential fixed investment (shown in the graph) averaged about 7.5% of GDP, and both nominal and real housing expenditures were at about the same level at the end of that period as they had been at the beginning.  From 1947 to 1962, nominal housing expenditures as a portion of PCE skyrocketed from about 11% to about 18%.  During this period, residential fixed investment was only about 1% above the 1962-1995 average.  And, from 1982 to 2007, when real housing expenditures fell, residential investment as a proportion of GDP averaged 7.3%, just 0.2% below the 1962 to 1995 average.  The secular variations in residential investment are smaller than the variations we see through the business cycle.

I have adjusted residential investment to create a counterfactual that would have led to stable real housing expenditures (and stable nominal expenditures, assuming unitary demand elasticity).  For instance, if real housing expenditures decline 0.25% in a given year, the counterfactural residential investment is adjusted upward by 0.25% of the value of residential real estate owned by households.  As the cyclically adjusted stability of residential investment suggest, it would appear to have taken a very small change in total residential investment to counter rent inflation.

Because this problem has been so persistent, it is easy to assume that rent inflation is a natural part of city dynamics.  The many growing metro areas that don't share this characteristic should serve as evidence against this notion.  And, if we look at rent inflation over time, we see behavior that appears to relate to the level of residential investment.  When building was allowed to expand, we did not have persistent rent inflation.  From the beginning of the core inflation series in the late 1950s until the late 1970s, shelter and core inflation generally moved together.  Shelter inflation first moved significantly above core inflation during the building contraction associated with the 1980-1982 recession.  Residential investment failed to move back above the long term average in the ensuing recovery, and so shelter inflation failed to recede back to core inflation levels.  Then, after the 1991 recession, residential fixed investment failed to even reach the long term average until 2003.  And rent inflation has been well above core inflation for most of this period.

The reason that this is even a topic of discussion is because nominal home prices rose to such heights.  Because practically everyone, from populist pundits to skilled economists, is reasoning from a price change, they associate the high home prices with high demand and high residential investment.  But, if only we could have had just a small amount of additional residential investment (focused in the large coastal metro areas, where the natural market supply response has been constrained), home prices would have been tame.  If only residential investment had been higher, nobody would be complaining that it had been too high!

The Substitution of Fixed Investment for Location Value

I made my own error in a recent post where I suggested that urban real estate owners were working against their long term interests by fighting against urban progressives who want to control the housing stock.  I was actually the one being short-sighted in my analysis.  Urban real estate owners do earn excess rents on the artificially high values of their actual properties.  But, they would clearly be better off in absolute dollar terms if they were earning normal returns on their potential properties.

In other words, they would earn much higher profits on their land by financing a 40 story condo building that earned 4% net returns than they earn on the same lot with a couple of duplexes that are currently earning 6% net returns plus 1% annual capital gains from rent inflation.  They are earning excess profit on the land they own, but that land is being prevented from providing the housing stock that it is capable of providing.  It is clearly in their self interest (and our collective interest) for them to develop that land further.

And, I think this adds a subtle twist to the picture of residential fixed investment in the 2000s.  Rent inflation moderated from 2003 to 2005, when residential fixed investment rose.  But, it was still running slightly higher than core inflation.  Why didn't this high level of investment cause rent inflation to fall below core inflation, reverting to lower levels?

The effect of limited building in the cities is that landlords are capturing higher rent for limited housing stock.  If housing wasn't limited, they would be capturing lower rents on expanded housing stock.  So, if housing was allowed to expand in the cities, capital income would be accruing to urban land owners, but it would be accruing to them for providing housing, not for owning an artificially scarce resource.

Now, high rents and limited housing in the cities is pushing households out to the suburbs and exurbs.  One effect of substituting lower value locations for the high value urban locations is that households purchase larger homes.  So, in the constrained context that we have created, fixed residential investment has been inflated.

Residential fixed investment never really rose above the historical range.  But, even the levels it did reach were inflated by our anti-housing metropolitan areas.

If urban development had expanded, households would have favored urban housing where more of their housing expenditures would go to rents on the land.  This is a scenario where capital income would have increased, but as a result of broad improvements in real living standards.  Lower rent on housing would mean that real incomes would be much higher.

No comments:

Post a Comment