Thursday, April 23, 2015

The Housing Half of the Treasury/Housing Trade

I mentioned the other day that it looks like things are turning up for a position that is long housing and short bonds.  On the bond side, I expect long term rates to move up as housing begins to attract capital again, and short term rate expectations should move up as GDP growth accelerates.  Short Eurodollar contracts in the early 2018 time frame seem like the best spot to target those movements.  That time frame should capture most of those rate movements while avoiding unrelated Fed discretionary moves regarding the date of the first Fed Funds rate hike or the cyclical peak in the Fed Funds rate.  On the other hand, if long term rates on housing and treasuries converge, generally, then there are probably more gains to be captured farther out on the yield curve as the entire curve moves higher.  My main concern there would be that aggressive hawkish postures by the Fed could hold those rates down while leaving rates in the 2016-2018 time frame relatively unchanged or even higher.  If the Fed somehow moves into a dovish position, long term rates and mid-term rates would rise, due to positive market conditions and inflation expectations, and this would also seem to favor that 2018 time frame.

Here's yesterday's graph of new home sales and home prices over time.  We can see here how home sales have suddenly moved up sharply, but still have a long way to go (although housing starts have not seen this same sharp move up in the last couple of months).  I have already noted that mortgages have begun to expand at the banks.  I expect home price growth to begin to turn upward again, also.  Note in this graph how home prices tend to lag home sales.  This supports my general thesis that home prices are sticky, and that much of the speculative activity in the 2000s boom revolved around that.  I had originally expected to be able to value homebuilders, in part, as land speculators.  But, their business and their valuations seem to be much more closely tied to sales volume than price.  I think this is partly because of this price stickiness issue.  The outlet valve for housing when the equilibrium price is moving sharply is the new home builders.  The quantity response moves through the homebuilders.

On the housing side, one way to take a position would be to take long exposure in a homebuilder that would be expected to gain the most from positive surprises in new home sales.  Here are two earlier posts on the idea.  Here, we really want to aim for a highly leveraged, high beta firm.  In addition, homebuilders tend to be sitting on large amounts of tax assets, many of which are still off the balance sheets at the firms that have struggled the most through the housing bust.  I had hoped to find a clever valuation method for these firms, specific to the situation.  But, generally, it looks to me like the homebuilders tend to have a pretty stable ratio for Enterprise Value / Revenues of 1 or slightly higher.  It tends to move up somewhat when growth expectations or margins are running high, but it appears to be a stable valuation metric for the industry as well as for individual firms.

It is running high now, partly because of those tax assets, although the tax assets don't generally amount to a large sum for the largest firms at this point.  Mostly, this valuation is running high because of growth expectations.  Analyst 2 year revenue growth for the industry is averaging about 33%, but I suspect that the market price reflects growth at more like 50%.  This isn't as crazy as it sounds.  Looking back over the past 20 years or so, 2 year growth levels during recovery times have commonly run in the 50-60% range.

In the next graph, homebuilder enterprise values are compared to current revenues and prospective revenues.  My "bullish" 2016 revenue level here equates to 2 year growth of about 66%.  That is similar to growth rates since 2012, and would put home sales in 2016 only back up to about the levels of the mid-1990s, so I don't think that is excessive in the current climate.  And, that puts industry-wide valuations, with some modest annual gains, at the range of baseline valuation levels when growth might be expected to moderate in a few years.

Because growth rates seem to be somewhat priced in, I am not sure how much gain is available, industry-wide, for a bullish forecast.  There might be 25% or more in industry-wide excess gains if my bullish forecast comes to fruition over the next 2 years, but I think the gains for the entire industry may be dependent on the length of time the recovery is allowed to run, and there is too much political/monetary uncertainty there for valuations to reflect hopeful growth more than a few years in advance.  So, what industry-wide gains there may be may happen in real time as the recovery progresses.

For this first phase, I think the gains will come mostly from the more distressed firms.  In these charts, the firms are arranged by leverage (red bars).  Debt here is shorthand for Enterprise Value minus Market Capitalization.  The blue bars reflect growth forecasts.  Dark blue is revenue, and light blue is how that growth would flow to equity holders, given current leverage levels.  Growth expectations are fairly tight across the industry, so the inferred equity growth is strongly related to leverage.

In the next chart, the measures are the expected market cap with an Ent.Value/Revenue ratio of 1 at 2014 revenue levels, 2016 forecasted levels, and my bullish 2016 levels.  All values are as a percentage of the current market capitalization.  All enterprise values have tax assets deducted (including off balance sheet allowances).  We can see that the EV/Rev. value is somewhat lower than most of the market capitalizations of the more healthy builders, and for the industry as a whole.  The builders who are highly leveraged are currently priced below that level.  (For instance, Hovnanian (HOV) would need to nearly double in price for its Enterprise Value to equal its current revenues.)  This makes sense, as the distress caused by the over-leveraged balance sheets creates added risk, so if a recovery does not come, these firms will likely suffer valuation losses.  But, these are the firms which offer the most upside from a bullish market.  And Hovnanian really is the firm most aligned to this proposal.

Betas for the firms also tend to follow the same pattern as the leverage levels, with Beazer, Hovnanian, and KB tending to have higher betas than the other equities.  I think this is a case where extremes in potential outcomes and betas can make it difficult for theoretical models to apply to actual financial performance.  Even Hovnanian tends to have a beta less than 2.  I think part of what happens is that things like operating and financial leverage do create a multiplier effect as revenue ebbs and flows.  But, additionally, the leverage also serves as a sort of optimization of a certain set of expectations about revenue growth.  So, some of what is actually beta will be measured as alpha, depending on how optimized that firm's leverage is to actual revenue growth.

Here is a graph of equity returns, normalized to current share prices.  Notice that since the crisis, Hovnanian has exhibited a beta that looks like something around 3 (compared to the industry), which is more in the range of the estimate for expected returns above.  KB and Beazer have also occasionally shown similar behavior.  But, what we see are separate periods.  Its measured beta (compared to the industry) has ranged around the mid 1's during this time, depending on the specifications you use.  But, depending on how conditions differed from expectations, the beta embedded in its current business model came through as very high levels of alpha (positive or negative).  Then, over the past two years, as housing has progressed more or less as expected with few positive growth surprises, along with a pause in home price appreciation, Hovnanian, has acted kind of like a call option, with a time decay, because by necessity their business model is optimized now for higher growth rates, until they move back toward their organizational capacity and optimal capital structure.  It also probably reflects less faith that tax assets will be claimed.

Statistically measured betas simply can't capture these nuances.  Good timing in these matters can capture gains that market-wide statistical analysis won't find.  Good timing ... that should be easy.  I mean, what could be so hard about that?


  1. Kevin, looking again at the new home sales in units, it's amazing how fast and how far it falls. The peak is actually in August 2005. New home sales (fred2:HSN1F) drop precipitously, from 1.3M (annualized) in December 2005 to under 400k by the end of 2008, a factor of 3. Existing homes show a similar pattern (fred2:EXHOSLUSM495S), though dropping by more like a factor of 2. As you point out, prices are sticky, so we don't see the price collapse until mid-2008! More than two years of stickiness.

    The situation begs the question in my mind, why did volume collapse? Did prices rise slightly above market clearing rates, and then due to extreme stickiness, it took a massive volume collapse to build up enough inventory to force prices down, and then a feedback loop in the market (falling prices lead to people walking away from mortgages, uncertainty in pricing leads to a tightening of credit, etc) drove prices down far below what would have been market clearing prices in normal times? This system seems startlingly unstable. Seems like someone with access to capital, courage, and patience could make a lot of money whenever housing prices fall too much, say, by using their ability to print money to buy lots of mortgage-backed securities...

    You have probably already addressed this question (why did sales volume collapse around the end of 2005), but my understanding of the housing market wasn't up to following you at the time, apologies.


    1. This is so complicated. I hope to eventually air all of this out in the housing series. I think part of the problem is that it is just so hard not to "reason from a price change" and not to think of asset prices in terms of supply and demand. In stocks, for instance, if a stock goes up 10% after an earnings report, we naturally think of it in terms of new buyers coming into the market. But that is misleading. Apple isn't selling for 50x what it was 10 years ago because there are more buyers. It is selling for 50x more because it has 50x more earnings potential. It's worth 50x more.

      With housing, rent is determined by supply and demand, because that's consumption. It's less complicated. If there are not enough houses, rents will tend to go up and if there are too many houses, rents will go down. The problem with thinking about home prices is that rents tend to be very stable, deviating from core inflation by maybe 1 or 2% per year, at most, usually. So thinking about homes as assets (as securities with future vash flows), the change in value comes almost entirely from changes in the denominator of the sum of the present value of those cash flows. It's very hard to think about intuitively.

      What I think is happening is that when the Fed pushes short term rates past the neutral rate, usually late in the recovery, it creates a drag in liquidity, which is usually associated with lower expectations about economic growth and, thus, lower long term interest rates. In bonds, that means that basically the same amount of cash still goes into bonds, but it fetches a lower income (lower yields). But, in homes, there is no way to reduce yields, because rents will be pretty stable, so the only way for home prices to remain at non-arbitrage prices is for the prices to rise. But, that means buyers need cash or credit.

      Existing homes still sell at prices in equilibrium with bonds. If they remain too low, cash will move from bonds to homes, raising yields on bonds and lowering yields on homes (which means raising the price). Since there is limited availability of cash and credit, there is no way for those yields to get pushed down further. Any cash that is available will be used to bid up existing homes. So, the lack of cash is expressed mostly through the falling quantity of new homes built.

      But at that point, it's not even price stickiness. It is the intrinsic value of the homes that is still rising. I'm sure there was price stickiness in the home market after 2007, but the price homes were moving to was a disequilibrium price caused by the credit breakdown. Any price stickiness was actually keeping home prices closer to intrinsic value. But, now I'm just confusing matters more.

      It's probably no different than, say, the IPO market, which is more volatile than the stock market. Each home is like a real estate IPO.

      Here is a Fred graph that shows the co-movement of rent inflation with the Fed Funds rate:
      As the Fed Funds rate reaches the cyclical peak, rent inflation tends to rise as we enter the recession, because overtightening by the Fed has cut off home building.

      Here is a chart with housing starts compared to the yield curve spread (Fed Funds minus 10 year):
      It's a typical pattern that the yield curve inverts, then home building falls and rents rise.

    2. Kevin, thank you, as always, for the detailed response.

      > In bonds, that means that basically the same amount of cash still goes
      > into bonds, but it fetches a lower income (lower yields). But, in homes,
      > there is no way to reduce yields, because rents will be pretty stable,
      > so the only way for home prices to remain at non-arbitrage prices is for
      > the prices to rise. But, that means buyers need cash or credit.

      Wow, okay, got it, now I am starting to understand this. But it just means I have another question for you. Right now, a 30-year treasury yields about 2.6%. But (from a very brief look at it looks like you can buy a modest single-family home for around $150,000 and rent it at around $1500/month. I realize that renting a home has expenses too but I imagine they are less than the 10-point spread here, meaning an investment property is a better deal than bonds even if your expenses are 7% a year. If inflation picks up, which seems more likely than inflation falling further, the home will look even better, as a meaningful increase in inflation expectations would kill 30-year bond prices but leave home prices more or less unaffected, and in fact homes would presumably appreciate wiht inflation as inflation increases their replacement cost.

      How can this be? Why doesn't every rich bond-holder sell their bonds and buy houses and rent them out?


    3. No problem, Ken. Your questions tend to help me think through the issue and to notice where I wasn't clear. Sometimes, as I write posts, I feel like I am starting to repeat myself. But the topic is so complex, I think to an extent this is helpful.

      The BEA has numbers on gross and net rent which I have been using for aggregate time series comparisons. You're right. This is a killer investment. And, there have been investors flooding into the market. Much of the buying since the crisis has been all cash or investment buyers. There are many reports of institutional investors coming into the real estate market in new ways.

      There are frictions that keep this correction from happening instantaneously. Since tax rules favor owner-occupiers so much, single family homes have been dominated by owner-occupiers to the tune of something like 85% of the market. So, the single family home rental market is growing from a very small base. There is a lot of organizational groundwork that would need to be done for institutions to counter the stagnation of the owner-occupier market. But, that is probably what has helped home prices recover as much as they have. Of course the people that are wrong about all of this complain that speculators and foreigners are pushing up prices again and pricing middle class families out of the market.

      I am hoping that we are seeing a recovery in the mortgage market now and that will put some juice into home values. To be clear, I don't care if home prices are high or low. I care that the market is efficient and relatively accessible. It just so happens that right now that means higher prices. I'd also love to see risk premiums decline (which would raise interest rates) and housing tax subsidies die. Both of those things would cause intrinsic home values to fall. But, if those things aren't going to happen, then I at least want to see markets functioning, which would mean higher prices.

      Ironically, if we untethered the housing market and let building expand to meet demand, I think it would lower rents, lower rent expectations, and would meet the demand for safe assets, so that after that we would see lower relative home prices, lower rents, higher bond yields, and more growth oriented corporate investing. I'm starting to see the limitations on homebuilding as a possible significant cause of low interest rates.

    4. One last comment. Repeating your point, as interest rates fall, if rents are stable, then house prices "must" rise. However, aspiring homeowners intend to pay their mortgage out of a relatively stable income. This climate appears stressful for the mortgage market in at least three ways. First, the quantity of dollars required to buy a home is rising --- loans have to be bigger. Second, assuming the borrower needs a fixed interest rate, the lender is facing a larger inflation risk. Third, the lender (who "learned" from 2008-2010) is facing a larger principal risk, because if interest rates rise sharply and home prices (as assets) fall enough, the owner may walk away, leaving the lender with principal loss. Thus the spread between fixed mortgage rates and the risk-free rate should grow as the risk-free rate drops to zero. (I can't reach Fred2 right now, so not sure if this actually happened). This may be one key "friction" you are referring to preventing home prices from correcting to non-arbitrage levels. The other key friction is that investors don't step in because buying a bunch of homes and renting them out sounds like too much work. I can't prove that a lot of investors would think that way, but I have at least one very solid example ... :-)


    5. (1) This is one reason I think everyone has the causation backwards. Low down payments didn't lead to a price bubble. The high nominal prices led buyers to demand low down payments. People are kind of playing mad lib with the available facts and their narrative of what happened here. There were some avenues for lower down payments and prices were rising, ergo low down payments led to high prices. But, really, if anybody would think about it for a second, it's absurd to think that a few tweaks in mortgage terms would cause nominal home prices to more than double.

      In theory, low down payments could create systemic risk, but while there were more mortgages than usual that started out with higher LTVs, aggregate loan to value levels were pretty level through the period. Also, from 1994 to 2004, homeownership rates went from 64% to 69%. The last couple of years of the boom did not coincide with growth in marginal new homeowners. Only the last 1% of those ever saw home prices below their purchase price. In gross terms, some of those later buyers were new homeowners, and some of them had high LTVs, but as today's post notes, there was not a correlation between those factors and delinquency rates.

      (2) Those premiums are already factored into the prevailing rates. The type of mortgage, or no mortgage at all, shouldn't change the net expected value of the transaction, assuming marginal market expectations.

      (3) I agree with your intuition. The spread may be slightly higher than it was when rates were higher - maybe a 1/4% or so. But, I'm not sure if these things add up to as much as it seems they should. If banks think there is a 20% chance that another crisis will happen that causes 10% of their mortgages to take a 10% write-off, that's only a 0.2% risk. I think it would be smart for a bank to acquire a private real estate firm and transfer the properties to them as rentals, if another crisis happens. But, I suspect that FDIC rules or some other regulation prevent this.

      I'm with you. Being a landlord does not sound fun.