An Analysis of The Harvard Center’s Case Against the Housing Bubble
By Dean Baker1
October 27, 2003
In response to the surge in home prices in the last eight
years, and fears that this could lead to a future collapse in housing prices,
the Joint Center for Housing Studies of Harvard University has published a study
arguing that the housing market is sound, and that there is little basis for
concern about a housing bubble.2
The study makes four points to support its case that there
is no housing bubble:
- over the years 1991 to 2001, household income rose almost as rapidly as
the home price index;
- it is rare for home prices to actually decline in nominal terms, more
typically after a sharp price run-up homes experience modest declines in real
prices, as their price does not rise as rapidly as the overall rate of
there continues to be a rapid rate of household formation (estimated in
the paper at 1.2 million annually) due in part to high levels of immigration;
the country is experiencing a low interest rate environment, which is
likely to persist for the indefinite future.
It is easy to show that none of these four points should
provide any reassurance to homeowners about the future value of their property.
Each point is addressed in turn below.
Family Income and
Figure 1 in the Harvard study compares the growth in median
family income with the growth in the Home Price Index (HPI) over the years 1991
It shows that median income increased at only a slightly slower pace than the
HPI over this period, approximately a 4.3 percent annual rate compared to 4.5
percent for the HPI. It infers that income has therefore kept pace with housing
prices and therefore there is no housing bubble.
This logic is flawed at a very basic level. The HPI tracks
the resale prices of the same homes. In other words, if the HPI rises by 5
percent in a year, it means that, on average, every individual home has
increased in price by 5 percent compared with the price it sold for last year.
There is absolutely no reason whatsoever why it should be expected that
individual home prices would rise in step with family income – and it has
never happened before during a period when family income was rising.
Unfortunately, the HPI only goes back to 1975, but prior to
1981, the homeownership component of the consumer price index was constructed in
a manner that is similar to the current HPI, tracking the resale prices of the
same house. In the years from 1955 to 1973, the shelter index of the CPI (which
includes the homeownership component in addition to a rent component) rose by
77.9 percent, or an average of 3.3 percent annually. By contrast, median family
income rose by 173 percent over this period, or at an average annual rate of 5.7
The study appears to confuse two distinct propositions. It
is reasonable to believe that as a first approximation expenditure shares do not
change much as income rises. This means that if a family’s income rises by 10
percent, we might expect that it will spend 10 percent more on clothes, cars,
and housing. In this sense, if we found that spending on housing was rising in
step with family income, then we may conclude that this is perfectly normal and
should provide no basis for concern. Presumably people are moving into better
homes as their income rises.
However, the HPI is not tracking spending on housing – it
is measuring the change in the resale price of the same homes. Making an analogy
to the auto market, if a family’s income goes up by 10 percent, then it may
spend 10 percent more on cars, due to the fact that it will buy a better car.
However, it would not spend 10 percent more to buy the same car.
This is exactly what the HPI is telling us – people are
paying far more money to buy the same house. In the eight years from the first
quarter of 1995 to the first quarter of 2003, the rise in the HPI has exceeded
the overall rate of inflation by more than 33 percentage points. There is no
historical precedent for this sort of run-up in home prices. Over the whole
post-war period prior to 1995, home prices had moved roughly at the same pace as
the prices of other consumer goods. There is no theoretical reason why the
increase in home prices should exceed the increase in the price of other goods,
nor is there any reason why it should keep pace with family income.
Past Patterns in Home Prices
Assuming that nominal home prices will not decline in the
future, simply because they have not declined in the past, is simply a case of
faulty logic. There has never been a nationwide run-up in housing prices (at
least for which we have data) comparable to what has taken place over the last
eight years. The extent to which housing prices can fall depends on the extent
to which they have become over-valued.
The Harvard Center’s logic on this point is exactly the
same as that of people who argued that the stock market was a sound investment
even at the peak of the bubble. Extrapolating from prior experience (there had
not been a prolonged downturn in the stock market since the Great Depression) is
inappropriate in the face of a historically unique event. The run-up in housing
prices since 1995 is historically unique in the same way that the run-up in
stock prices prior to 2000 was a historically unique event. The extent to which
either the housing market or stock market can be expected to fall depends on the
extent to which these markets have become over-valued. Prior experience is
helpful in this respect only to the extent in which comparable periods of
over-valuation can be identified.4
The Rate of Household Formation
The Harvard Center expects demand for new homes to remain
strong because it anticipates an annual rate of new household formation of 1.2
million annually, driven in part by the natural increase in the population and
in part by immigration. While this 1.2 million figure may prove high
(immigration has probably lagged in a weak economy and post 9-11 world), the
important factor in determining home prices is the relative growth in demand and
supply. The economy is currently adding more than 1.8 million housing units
annually, an amount that exceeds the Center’s estimate of the rate of
household formation by 600,000 a year.
Furthermore, with the housing sector proving extremely
profitable for builders, there is no reason why this pace will not be maintained
and even increased if housing prices remain at their current levels. The fact
that supply growth is currently outstripping demand growth is showing up now in
the record vacancy rates for rental housing (see table A-2 in the study). Supply
will continue to exceed demand, and probably by a growing margin, unless prices
fall back to more sustainable levels.
The Center’s study was written at a time when interest
rates were headed down to a forty year low. Since that time, interest rates have
rebounded, although they still are low compared to their levels over most of the
last three decades. However, almost no economists anticipate that interest rates
(and most importantly real interest rates) will still be low long into the
future. The prospect of large budget deficits for the indefinite future, coupled
with the likelihood of rising inflation due to a falling dollar virtually
guarantee that interest rates will be substantially higher in two or three years
than they are today. The Congressional Budget Office projects that the interest
rate on ten-year government bonds will average 5.7 percent from 2005-2008,
approximately 1.4 percentage points above their current level. Most other
forecasters have made similar projections.
Therefore, if low interest rates are the explanation for
the run-up in housing prices, then it should be expected that housing prices
will plummet when interest rates rise. This would be an argument for the
existence of a housing bubble – not an argument against it.
There is little data in the Harvard study that can be taken
as evidence against a housing bubble. The study relies primarily on a basic
error in logic to make its case – noting that family income has kept pace with
home prices – failing to recognize that there is no reason to expect that the
price of individual homes will rise in step with family income, even if total
spending on housing keeps pace with family income. It should be expected that
families will spend more money to buy better homes, not pay more money for the
same house. Apart from this error in logic, the Center’s study offers little
evidence that undermines the case for a housing bubble.
 Dean Baker is the
co-director of the Center for Economic and Policy Research
 “The State of the
Nation’s Housing: 2003.”
 While the 2002 data may
not have been available when the study was written, it is worth noting that
the HPI increased by 6.9 percent in 2002, median household income increased
by 0.5 percent.
 It is also important to
note that the distinction between a fall in real verses nominal prices is
not nearly as important as the Harvard study appears to suggest. If the
nominal price of a house does not change at all over a ten-year period, but
the overall price level has doubled, then the real value of this home has
declined by 50 percent. If a homeowner had been planning to sell the home in
order to support his or her retirement, then they will find that they have
half as much wealth as they may have expected when they purchased their
home. The fact that this decline in the real value of their home came about
as a result of an increase in other prices, as opposed to a decline in home
prices, makes little difference.