Tuesday, May 20, 2014

Home Price Forecasts

I have been forecasting a continuing recovery in home prices, arising out of my point of view that home prices in the 2000's weren't actually out of line with other assets, even though, at the time, I considered them to be crazy.

A Digression

I live in a neighborhood with many young families, and we would laugh back then about how none of us would have been able to afford the homes that we had all purchased a few years before.  That seemed like an obvious sign at the time that the homes were overpriced.  Now, I believe that it was perfectly reasonable for our homes to be too expensive for us to buy.  Long term real risk-free interest rates were about 2% at the time.  Baby boomers and investors who needed long-term fixed income should have owned our homes, and we should have been renting the homes from them.

This whole issue is similar to the difference between owning a AAA-rated bond and buying a bond-holding mutual fund.  The bond fund is marked to market each day, so the volatility of the present value of the investment is made very clear.  Individual bonds are sometimes marketed to individual investors as a source of safe, fixed income.  But, the difference is mainly a matter of framing.  If the investor marked that bond to market each day, it would be just as volatile as the fund.  But, if the investor doesn't mark to market, then they simply see a $100 investment that pays them $4 per year every year until maturity.  The cash flows are very stable.

There is a similar situation going on in real estate, except that real estate prices aren't only a mathematical product of a stated yield, so the risk factors are stated differently.  I imagine my statement about real estate investments in the 2000's eliciting a reaction of incredulity - that even though I might argue that they could have been good investments, that theory clearly didn't pan out.  But, that theory only didn't pan out for homeowners who were marking to market.  Rent levels are fairly stable.  If an investor bought a home with cash in 2005 and continued renting it throughout this period, their cash flows would have been, more or less, in line with their original expectations - much like the owner of that AAA-rated bond.  And, if they had originally planned to sell that house in 2030, there is no reason to think that its eventual value will be such that their investment won't have earned some small real return over that time.  (Homes don't promise a set maturity value like bonds, and so they tend to provide higher returns to compensate for that risk.  In fact, I have been modeling homes as a low-risk bond holding, but maybe I should envision them more as a high dividend equity.)  In any case, my point is that, eventually, in hindsight, even after all that has happened, if in 2005 you had needed a long-term inflation-hedged asset, a house will probably have suited that purpose just fine.


Home Price Forecasts

I have been looking at price to rent ratios from the various home price indexes, and it is interesting how much variation there is.  Here is a graph of some price indexes, indexed to the top of the market at the end of 2005.  In addition to those shown here, there is the CoreLogic series, which I believe shows similar behavior to the Case-Shiller series shown here.  These series seem like the outliers, although they do seem, anecdotally, like they reflected existing home prices in major cities.  To the extent that Case-Shiller is accurate, then homes seem undervalued by 40% in a functioning credit market, given expected long-term real interest rates.  This might be relevant to a new speculative position that involves straightforward exposure to real estate assets.

But, if a position is going to be established by exposure to homebuilders, then the Median Sales Price of New Houses series is probably more relevant.  This series was much less volatile than Case-Shiller during the boom, and was also slightly less volatile than the All-Transaction and Purchase Only home indexes from FHFA.

Oddly, the New Houses series has been much more positive since the crisis than the other series.  I believe that some of this difference comes from two factors:

1) All of the home-tracking indexes (Case-Shiller and the two FHFA series) are affected by short sales and foreclosure sales that tend to be priced below the normal market level.  This probably accounts for a measurable decline in stated prices throughout the crisis, currently probably amounting to less than 5% of the level that would be reported if foreclosures were not unusual.

2) The median price of new homes is not based on tracking of individual properties, so my understanding of it is that it would be affected by changing home characteristics.  I have attempted to account for that by controlling for square footage, which is tracked, along with prices, by the Census Bureau.  Below is the measure for square footage of new homes, and in the Price Index graph above, the light blue line is the relative Price to Rent level for Median Sales Prices of New Houses Sold, after adjusting for changing square footage.  Since the end of the crisis, square footage has climbed significantly.  Adjusting for this pulls the relative New Home Price To Rent measure down somewhat, but not all the way to the price measures for existing homes.

But, these adjustments put all three of the less volatile price series at around 85% of the high Price To Rent levels of the 2000's.  That leaves about a 20% gain from today's prices to reach those levels again, according to each of these indexes.

In addition, the square footage adjustment to the new home price series makes that series even less volatile in the 2000's than the unadjusted measure.  If this is the product of greater price-stickiness among new homes, relative to existing homes, then the high point of prices in the 2000's might still have been below the equilibrium prices.  That could mean that new homes have more room to rise, relative to the other indexes.  The FHFA indexes include a lot of homes in parts of the country with stagnant real estate markets.  Homebuilders would tend to be in growing markets that should reflect more of the price growth that Case-Shiller is picking up.  Therefore, it might be reasonable to expect the New Home price series to come in somewhere between the FHFA series and the Case-Shiller series.

Of course, if prices in the 2000's were too high, then my targets here are wrong, but accepting my basic premise, it seems as though homes are capable of a 15-20% gain to meet pre-crisis levels, at a minimum, and potentially up to more of a 30-40% gain, if the Case-Shiller index is a more accurate portrayal of the position one takes.  However, treating homes as inflation-hedged cash flow instruments might only justify about a 50% gain from the mid-90's to the 2000's, which can account for the FHFA series and the new homes series, but doesn't fully justify the 100% gains in the Case-Shiller index, so it might be a bridge too far to use my premise to call for a return to the Case-Shiller highs.  On the other hand, if the gains take 2 or 3 years to play out, inflation would add another 10% or so, and interest rates could rise at least another 1% while still justifying the 2000's Price to Rent ratios.

At the other extreme, new home price-to-rents, adjusted for square footage, are back at 1990's levels, so by this measure, homes could possibly call for the full 50% increase in value coming from the decreased real interest rates, but this series appears to be much less volatile than the other series, so this probably isn't a reasonable expectation.  Additionally, my Price to Rent ratio for Case-Shiller seems more volatile than the ratio shown at calculatedrisk.com, although it looks he is using the same ratio.  If my measures aren't capturing changes in rents and prices accurately, then there could be drift over time in the expected price level.

It still looks to me like more home price appreciation should be coming, but I don't know if I can say how much with any precision.

No comments:

Post a Comment