Friday, March 29, 2019

Housing: Part 347 - Price/Rent Ratios over time

Here are some graphs of price/rent ratios, by zip code, for several metro areas.  (All x-axes have the same 2013 rental value, while the y-axes represent the estimated price/rent ratio for each year.)

The data is from the inestimable Zillow.com.  Price/rent ratios before 2010 are based on my estimates, using price and affordability data. (Except the first graph, where the x-axis is price, and it changes over time.)

This relates to one of the points I made in Shut Out, that the positive relationship between Price/Rent and a range of other measures (price, rent, income) means that you have to be careful attributing rising prices in low-tier zip codes to easy credit.  In the few metro areas where low tier prices did rise dramatically, I contend that this was really the result of high tier prices levelling off because there is a maximum price/rent level within each city.

Comparing LA and Seattle is useful here.  Low tier prices didn't rise at a different rate than high tier prices during the boom in Seattle.  But they did in LA.  Here, comparing LA to Seattle, we can see that the crescent shape of the Price/rent pattern was the same in both cities throughout boom and bust.  The difference is that homes across LA had gotten so expensive by 2006 - more than $400,000, even in the lowest quintile of zip codes - that by 2006, the P/R on low end homes was close to high tier P/Rs.  In 1998, Quintile 1 P/Rs in LA were about 7x (!), less than half Quintile 5.  In 2013 they were a little more than half, at 12.7x.  In 2006, they were within 17% of the peak P/R at 25x.

In Seattle, the peak P/R was 28x, not much lower than LA.  But, rents, and thus prices, are much lower in Seattle, so Quintile 1 P/Rs only went from about 70% to 75% of Quintile 5 P/Rs.  The ratio started much higher than LAs but didn't move much.  That is because prices in Seattle never went high enough for more than a small portion of the market to hit peak P/R levels.

Following are scatterplots of P/R against annual rent value.  In the updated 2019 numbers, P/R ratios across Seattle have moved higher than they are in LA for any given rental value.  But, since rents are lower in Seattle than they are in LA, the metropolitan median P/R ratio is still a little lower in Seattle.


Phoenix and Miami both seem to have seen some recovery at the low end.  They seem to contradict my claim that low end markets are underpriced.


Dallas and Atlanta are interesting.  Atlanta has partially recovered and Dallas has fully recovered to peak P/R levels, generally across the metro.  This also contradicts my claim that low end markets are underpriced.  I might have been able to say that in 2013 about both the cities and the low end within the cities.  But, not now.

This is a great example of how the premises determine the conclusion in a way that I am not sure can be resolved.  First, note that by this measure, in both Atlanta and Dallas, there was absolutely no sign of excess prices in 2006 and the collapse to 2013 was outrageous.  That would be the case even if you believe that home prices should not be sensitive to real long term interest rates.  Long term inflation protected treasury rates dropped from 4% in the late 1990s to about 2% in 2006 and are now near 1%.  Yet, in Dallas, at all three points, price/rent ratios were similar.

Now, if home prices aren't sensitive to rates, you might conclude that prices look reasonable in Dallas today.  But, you would also have to conclude that prices were reasonable in 2006 and that they were outrageously low in 2013.

On the other hand, you could argue that home prices should be somewhat sensitive to interest rates, and so prices in Dallas have been undervalued both during the bubble and after the bubble.  Or, taking all cities into account, it seems reasonable to conclude that home prices are sensitive to real long term interest rates, and where supply is elastic, supply increases rather than price.  Where supply is inelastic, price increases rather than supply.

If that is true, though, then cities with moderate or low price/rent ratios today should be building like crazy and rents should be declining.  Instead, building is tepid and rents are climbing.  I blame that on tight lending.  But, what if it's really because home prices really aren't sensitive to long term rates?  If that was true, then when rates are low, high land costs would cause building to be tepid and rents would rise.  Price/income would rise too.  That matches today's environment, but it doesn't match the boom period where low rates were associated with building and with moderating rents in the cities with elastic supply.

It seems to me that we have some sort of control over how sensitive home prices are to interest rates.  If they are not sensitive, it seems like an unstable equilibrium.  It is where we are now, with rising rents, rising price/income, but very low mortgage expenses for a leveraged buyer, compared to 1998 when mortgages cost 7%.

In the unstable environment, potential buyers are locked out because mortgage payments may take a large portion of their incomes.  But, in the meantime, so does rent.  And in that equilibrium, it will only get worse.  In the stable equilibrium, the important comparison would be real mortgage expenses vs. renting expenses, which would lead to a negative feedback loop where buyers could bring on new supply, and low rates would naturally help moderate housing costs.

1 comment: