Wednesday, March 18, 2015

Housing Tax Policy, A Series: Part 20 - Never reason from an asset allocation change.

I was reading this conversation between William Sharpe and Robert Litterman, and I came across this exchange, which considers the problem of market-matching asset allocation:
Sharpe:   ...Let’s say you believe that you’re an average person and that you want to take as much risk and get as much expected return as the average investor. The theory says to hold a market portfolio. To keep it simple, assume that the current market portfolio consists of stocks and bonds with weights of 60% and 40%, respectively— the fabled 60/40 portfolio. Now, imagine that down the road, you look at the values of stocks and bonds in the market as a whole and, for whatever reason, it’s 70% weighted in stocks and 30% in bonds. If you want to be like the market because you’re an average investor, then you should be 70/30 as well. However, if you look at the investment policy statement of, say, a typical defined benefit pension fund, it will state that the policy is to be 60/40 or some other combination of relative values. Should the pension fund follow that policy? Most of them don’t, but imagine if they were to. If stocks were to go up and thereby make the proportions 70/30, pension funds would have to sell stocks and buy bonds. Almost all policy statements that I’ve seen are stated in terms of percentage by value for each asset class. Those policies are all contrarian. They all recommend a decision to sell winners and buy losers, and not everybody can do that. In any event, for the investor who wants to be average in risk and return, the adaptive asset allocation formula says that if the market is 70/30, you’re 70/30, too, and that’s how you write your policy statement. It goes on to say that if you want to be x% riskier than the market, then you use a formula that tells you how to adjust your desired proportions when the market moves in a particular way, in terms of stocks and bonds. Even then, you don’t do a lot of trading, which is a good thing. 
Litterman: Why did the market go from 60/40 to 70/30?  
Sharpe: That’s not in my range of expertise. Predicting things and making probability assessments of what the future outcomes may be is your job, not the job of theoreticians. 
Litterman: Presumably, either the expected returns have gone up to justify this or the risk aversion has changed. 
Sharpe: Or risk has changed.
There are several problems to consider here.  First, is the equilibrium vs. disequilibrium problem.  If firms or home owners are temporarily over-leveraged because of a demand shock, then an optimizing, tactical investor with access to credit should utilize that credit to capture excess gains in corporate or real estate equity.

But, assuming equilibrium, Litterman and Sharpe seem to be assuming that changes in allocation are products of investor demand.  But, what if the move to 70/30 is due to corporate deleveraging?  What if it is due to rising treasury debt and real estate allocations, which serve to meet the demand for fixed income securities?  There are a lot of moving parts here.  The conceptual question of how to match the market portfolio does not have an easy answer.

Thinking about the topic, I decided to look at asset values in the US over time, from the Flow of Funds report.  This is difficult, because different types of agency and ownership methods are difficult to categorize.  I have used the equities and bond levels from Tables L212, L213, and L228 and household real estate levels from Table S.3.  Please let me know if you believe that this is an incomplete or incoherent treatment.

The first graph is as a proportion of assets.  The second graph is as a percentage of GDP.  I was surprised to see that, in the first graph, real estate levels were not unusual in the 2000s.  They were mostly rising from very low levels in the 1990s.

In the second graph, we can see that these levels have risen as a percentage of GDP, from about 250% in the mid 20th century, to about 400% now.  This 150% increase is roughly described in thirds of roughly 50% each:

1) Foreign profits for firms, which add to firm valuations, but not to GDP.  As I have argued, this, in effect, pays for the trade deficit.  I think subtracting the trade deficit from GDP probably understates our national production because this is consumption that has been paid for with our foreign capital gains.  More precisely, we reinvest those gains in high return foreign corporate assets, and foreign savers must export goods and services to us in order to fund an even larger basket of low risk/low return capital in the US.

2) Additional real estate value.  This is essentially deferred consumption.  Values have probably been bid up because of the high demand for deferred consumption among baby boomers.  As I work through my housing series, I will also treat this as a sort of tax arbitrage, and I hope to estimate how much of the added real estate value since the mid-1980's is simply the added value of the tax advantages given to home ownership.

3) Additional firm value, mostly in the growth of financial corporation bonds.  I think I am double counting here.  Many of the mortgages held against household real estate would be corporate assets related to the corporate debt and equity shown.  So, some of the increase probably comes from an increase in mortgage levels, which also have increased corporate capital levels.  I have not made the complicated adjustments required to avoid double counting that.  And, some of the added real estate value ends up in proprietors' balance sheets.

In these last two graphs, I show capital incomes as a proportion of GDI.  These are from BEA tables 1.11 and 7.12, so they don't match up exactly to the previous charts, but I think they basically give us the profit side of the story.  These numbers are for domestic income, so the foreign source of corporate valuations has been removed here (point number 1, above).

The first graph shows profit to proprietors and firms and interest income.  This has been fairly flat over the entire period.  The second graph shows returns to homeownership.  This has risen over time.

As above, the interest income in the first graph includes mortgage interest, so there is some double counting.  This may be adding confusion, but since it reflects debt holder interest in real estate, it sort of fits in both categories.  So, I haven't done the detailed work it would take to carefully cull it out of corporate asset and income measures.

Keep in mind that part of the income to bond and mortgage holders is an inflation premium.  In real terms, this is more accurately described as a purchase of capital by equity holders.  So, the treatment of interest as income is somewhat incoherent.  In this regard, some of the interest income is really equity income and some of the mortgage income is homeowner income.  But the effect on total corporate or homeowner returns would not be affected by this transfer.

I think, except for the growth in foreign profits, the gains in capital as a percentage of GDP are mostly coming from the issue of interest rates.  Since homes are a relatively fixed income type of asset, and since rents tend to be a relatively stable portion of GDI, home prices increase as a percentage of GDP when real interest rates are low.

This effect has recently been inflated by higher housing consumption (which has pushed returns to housing capital to new highs).  Owner-occupied housing consumption has increased because (1) it is a useful form of deferred consumption for baby boomers, (2) tax advantages implemented in the 1980s and 1990s increase owner-occupier housing demand, and (3) low real and nominal interest rates have increased demand for owner-occupier housing.

Long term real interest rates were low in the late 1970's, but at the time, rental income was only 3% of GDI, so the rise in home values was hidden.  If rental income had been 5% of GDI, total real estate would have been worth another 75% or so of GDP.  Total capital, shown above, would have been around 325% of GDP.

In the late 1970's, high inflation made mortgage payments very high.  I had originally assumed that this led to decreased housing demand because households were blocked from ownership by the high hurdle for mortgage access because of those high payments.  But, curiously, homeownership rates bumped higher in the late 1970's, and low real interest rates, which have a strong effect on home values, seemed to be pushing home price/rent ratios up at the time.  How could all of these things be true?  I think the graphs here may give us a clue.

Low real interest rates were making homes more valuable.  In order to capture the benefits of homeownership, which included significant tax savings due to the high nominal capital gains homes were earning as inflation drove their prices higher each year, households were reducing their housing consumption (rent) in order to match with a house in which they could afford the mortgage payments in nominal terms.  So, there was a dampening of demand because of the high inflation premium, but the lower demand wasn't manifest through fewer homeowners; it was manifest through smaller homes (or at least, homes with lower rents).  So, the prices of homes as a security (Price/Rent) were still relatively efficient.  The adjustment in demand was made through reduced housing consumption more than through a decline in home buyers or in the Price/Rent relative to intrinsic value of forward cash flows.

In the 2000's, the inflation premium wasn't a binding issue, so the households incentivized into home ownership by the low long term real interest rates didn't reduce their housing consumption.  Rents remained stable, and the full effect of low interest rates on the nominal value of real estate expanded the value of capital, both in terms of real estate, and in terms of corporate holdings of mortgages.

Ironically, this nominal increase in capital values, which I expect to see again as the housing market continues to recover, would be reduced by looser monetary policy (which would produce higher inflation, and probably somewhat higher real long term interest rates).  I don't personally think the high capital values, in and of themselves, are a problem.  But, those who do think it's a problem, and who blame accommodative monetary policy, have the story exactly backwards.  There will be great pressure for tight monetary policy as these valuations expand, and they will be brought back down in the only other way possible - by creating a demand shock.  Money will be tightened until it is catastrophic.  This will only serve to keep real long term rates low, because savers will be reasonably risk averse and growth expectations will be muted, which will mean that these capital valuations will return to high levels as the economy recovers again, and the Fed, after once again kneecapping capital will be painted as the capitalist's handmaiden as nominal capital valuations move higher again.

I sure don't want that to happen, but I don't see how we avoid it, given the current zeitgeist.


  1. > I sure don't want that to happen, but I don't see
    > how we avoid it, given the current zeitgeist.

    At the risk of sounding like a broken record... how about telling the monetary authority to quit paying any attention to interest rates, inflation, unemployment or asset values? They should be worrying about only one thing: the level-path of total nominal spending. NGDPLT would end the dilemma between kneecapping capital versus appearing as capital's handmaiden.


    1. I'm with you. But from 2001 to 2007, NGDP growth averaged 5.2%, and this supposedly caused a devastating economic bubble, where we tricked ourselves into thinking we were better off than we really were. The great recession was our unavoidable comeuppance. The payment for our excess. The problem with the Fed is that they are still trying to fake our way out of it, instead of letting us take our medicine. If the Fed doesn't randomly inflict 40% drops in corporate profits, that's a false economy, and it will lead to more bubbles. Profit stability is a corporate bailout, really.

      That's the story I hear from, basically, everyone. We've been infected by the inequality meme. Success is failure. We have a national case of Munchausen syndrome.

      Your mistake isn't that you've mis-prescribed the solution. Your mistake is thinking that a solution is welcome.

    2. Okay, thank you Kevin, that was serious depressing :-)

      But you haven't convinced me to stop trying. I'm going to DC end of March, meeting with a few well-known economists, we'll see what we can do.


    3. Sorry. That last line wasn't nice, in hindsight. For the love of Pete, keep trying. Things that are worth it aren't easy....

  2. Slightly off topic...
    In Ironman's latest blog post, he talks about the "financialization" of the economy. This seems to play a part in the housing decisions people make. Decades ago, banks made mortgages which they kept in their own portfolios. Now, they are seldom held; sold off in big chunks of MBS's. Could this securitization of loans be the jumping off point in the changes you have detailed in this series?
    Chuck E.

    1. Banks still hold a lot of mortgages. I think the idea that they sell them all off in MBS's is highly overstated.
      I think a lot of the decline in consumption Ironman is finding might come from non-cash benefits. It looks like his measure just covers cash outlays for healthcare.
      I have a post in the to-do list about the financialization issue. There are many papers that purport to make that connection. Usually the idea is that the finance sector gets big, which leads to a recession. I think the causation is that risk aversion climbs, possibly due to demographics or reduced expectations, this leads to falling interest rates, which cause the nominal value of homes to rise, which causes the financial sector to rise. Some of the confusion comes from the fact that homes are a store of value and a consumption good. When rates fall, home values rise as a store of value, but their value as consumption goods remains stable. Since the land under the home is the source of the store of value, the part of the home that is a security changing in value is a perpetuity, which makes the accounting non-intuitive, because that part of the home doesn't decline in value as rental income is realized.
      There is basically a trade off between investing in homes and investing in other production. When investment moves to homes, finance looks larger and production falls.

  3. TravisV here.

    I thought this was a great observation by Michael Darda: "Corporate bond spreads are actually somewhat above historical norms despite a pervasive but errant perception to the contrary."

    1. Yes. The idea that it's the Fed's job to keep risk premiums HIGH (!) is &$#*%ing nuts. Yet, it seems to be the consensus position. Their job is to create instability? How else are they supposed to keep risk premiums up?