Goldman Sachs Economist on the Bubble

Peter Coy

Just got off the phone with Jan Hatzius, a Goldman Sachs economist who has written extensively on the housing market. He gave me permission to post a research note that he wrote today about the same academic study that I questioned in a post yesterday.

Hatzius says he's not sure I'm correct that the authors of the study assumed their conclusion. His criticisms are different. The biggest one is that the authors ended their analysis too soon--last year--missing the further inflation of housing prices since then. Here's what he wrote:

Bubble Trouble? Probably Yes

Monday's Wall Street Journal featured an op-ed by Christopher
Mayer and Todd Sinai, two academic real estate analysts, who
argue that worries about a housing bubble are overblown. Their
view is that the increase in metropolitan housing prices relative
to rents and incomes has been fully offset by the decline in real
interest rates and is therefore appropriate.

The op-ed is based on a new research paper issued by the Federal
Reserve Bank of New York (see Charles Himmelberg, Christopher
Mayer, and Todd Sinai, 'Assessing High House Prices: Bubbles,
Fundamentals, and Misperceptions,' September 2005, In
this paper, Himmelberg, Mayer, and Sinai (HMS) calculate the
total cost of owner occupation as the sum of interest,
depreciation, property taxes, and a risk premium for taking on
house price risk, and then adjust these costs for the tax
deductibility of mortgage interest and property taxes as well as
a term for expected capital gains. They call the resulting
measure 'imputed rent,' divide it by an index of actual rents,
and set the resulting ratio equal to 1 for the average of the
1980-2004 period. They then argue that a metropolitan housing
market is overvalued relative to its own history when the ratio
is above 1 and undervalued when the ratio is below 1.

Their main result is that 31 out of their 46 metropolitan housing
markets had values below 1 as of 2004. Of the markets usually
considered 'hot,' New York, San Francisco, and Phoenix had values
below 1, while Boston, Los Angeles, and Washington DC, had values
just marginally above 1; only Portland, San Diego, and Miami
showed some cause for concern. HMS conclude that there is no
evidence for a general housing bubble, and not even much evidence
for a localized bubble.

What do we make of this analysis? Of course, HMS are right that
interest rates matter for valuing capital assets such as
residential homes. It is important to be clear, however, that
interest rates can go up as well as down. Thus, if interest
rates rise from their unusually low current level, house prices
would decline, perhaps sharply. An interest-rate-induced decline
in house prices might have fewer macro implications because the
interest rate increase itself might be the result of strength
elsewhere in the economy. But, the house price decline would
nonetheless be painful.

But even on the narrower point, namely whether house prices are
out of line with rents and interest rates, we are somewhat
skeptical of the analysis. First, the results are sensitive to
minor changes in the assumptions, and there is no way of knowing
which assumption is the correct one. For example, HMS assume
that households require a constant risk premium of 2 percentage
points for owning instead of renting, and that they expect the
rate of capital gains to be equal to the 1940-2004 average for
their metropolitan area. Of course, it is impossible to know
whether these or any other assumptions about unobservable
concepts such as risk premia or expectations are correct. The
problem is that given the way their model is specified, the
precise choice of numbers can make a qualitative difference to
the results -- that is, it can change the assessment of whether a
metropolitan housing market is more or less highly valued than
the historical average.

Second, the analysis uses annual data that end in 2004. This is
unfortunate, because the case for an outright housing bubble was
still quite weak as of 2004 but has grown much stronger since
then. For example, our May 14, 2004, US Economics Analyst noted
that the housing market run-up "...does not look like a 'bubble'
but is easily explained by the decline in mortgage rates." Since
then, however, prices have not only continued to rise very
rapidly but have in fact accelerated further, at a time when
rents have grown slowly and interest rates have been basically
flat. In such an environment, a verdict of 'no bubble' can grow
stale quickly.

As a concrete example, take the Washington DC market, whose
valuation ratio for 2004 HMS report as 1.02. While this is 2%
above the historical average, it is far below the 1980-2005 peak,
which HMS calculate as 1.24. However, as of the second quarter
of 2005, house prices in Washington DC were up 26% year-on-year
according to the Office of Federal Housing Enterprise Oversight
(OFHEO), the house price measure used by HMS. Over the same
period, rents only rose 4%, while interest rates were roughly
stable. A rough calculation using the formulas provided by HMS
shows that these price and rent changes should have boosted the
valuation ratio to somewhere between 1.2 and 1.3, i.e., far above
the long-term average and very close to the prior historical
peak. Thus, an analysis along the lines of the HMS paper that
used more up-to-date inputs would probably come to a considerably
more cautionary conclusion.

Of course, comparing the economic costs of owning with the
economic costs of renting is not the only way of adjusting house
prices for changes in the fundamentals, including interest rates.
Our own preference remains with an 'affordability' concept that
asks what percentage of their disposable income households must
expend to cover mortgage payments on the median-priced home.
This approach not only relates house prices to incomes --
compared with rents, probably a more meaningful comparison for
most US households residing in the suburbs -- but it also
recognizes that the vast majority of households are unable to
borrow as much as they want and therefore cannot engage in the
theoretically 'pure' arbitrage considerations assumed by HMS.

As we described in detail in the May-June Pocket Chartroom,
housing affordability is deteriorating quickly. In hot markets
on the coasts, such as Los Angeles, the income share required for
mortgage payments on a newly purchased home already matches the
previous two peaks seen in the early 1980s and the late 1980s.
More recently, even affordability in the country as a whole has
started to deteriorate quickly. For example, the National
Association of Realtors -- not an organization known for
excessive bearishness on the housing market -- reports that their
US affordability index now stands at the lowest level since 1991.
Thus, housing valuations are stretched, and are becoming more
stretched the longer the current boom continues.

Jan Hatzius

Thanks to Jan Hatzius for allowing us to reproduce the above note.

Before it's here, it's on the Bloomberg Terminal.