CPI Housing Gap Quantified in New Study

http://quote.bloomberg.com/apps/news?pid=10000039&refer=columnist_baum&sid=aa49I53YXUPw

Understate Housing Costs, Understate Inflation: Caroline Baum

Aug. 30 (Bloomberg) -- If there had been any pressure on Federal Reserve officials to normalize interest rates more quickly early in the year, the last two inflation reports alleviated it.

The consumer price index excluding food and energy rose 0.1 percent in both June and July following increases of 0.2 percent to 0.4 percent from January through May.

Maybe the Fed was right, and the acceleration in inflation was transitory, the result of a spike in energy prices that filtered through to core prices.

That determination is of vital interest to policy makers, who seem less concerned about the reliability of the inflation data.

A new study by Alliance Capital Management's economic research department finds that a chronic underestimation of housing costs, the largest component of the CPI, has led to a 0.6 percentage point annual understatement of inflation since 1998 -- a gap that has grown to 1 percentage point annually in the last two years, with a somewhat greater impact on the core.

Understating inflation has ramifications for monetary policy, interest rates and cost-of-living adjustments, according to Alliance's Joe Carson, who wrote the study.

Two Markets Diverged

The source of the measurement error is the imputation of a rental value for owner-occupied homes. Conceptually, rental equivalence is a good way to measure the cost of housing consumption, which is what the CPI measures. The problem arises when a single survey of rental units is used to impute owner- occupied rents at a time when the characteristics of the two markets are grossly different.

The owner-occupied market is vibrant, with strong demand, fast absorption of new supply and low vacancy rates. The rental market is plagued by weak demand, slow absorption, and high and rising vacancy rates, which hit a 50-year high of 10.4 percent in the first quarter of 2004, according to the Census Bureau's American Housing Survey. That compares with a 1.7 percent vacancy rate in the owner-occupied market in the first quarter, little changed since 1997.

A little history is in order. Until the early 1980s, the Bureau of Labor Statistics calculated housing costs in the CPI from home prices, mortgage interest rates, property taxes, and insurance and maintenance costs.

Because a house is not just a home but also an investment, and because the CPI is a measure of consumption costs, the BLS abandoned the asset-price approach to housing costs in favor of an imputed rental value in 1983.

Surveying Tape

Rental equivalence measures the change in the implicit rent, or what a homeowner would earn from renting his home in a competitive market. Initially the BLS re-weighted the survey of rental units to compute owner-occupied rents. In 1987, the sample survey was expanded to include both rental and owner-occupied units.

Because of the small number of single-family detached homes for rent and cost constraints, the BLS dropped the owner-occupied survey in 1999 and went back to reweighting the rental survey to impute owner-occupied rents.

At about that time, the two sectors of the housing market parted ways. Low interest rates encouraged renters to become owners, pushing up home prices as the rental market stagnated.

``The theoretical measure of owner-occupied housing costs is moving further and further from actual experience,'' Carson says in his study.

Short Run/Long Run

The year-over-year increase in owners equivalent rent (OER) fell to 1.9 percent in January and February, the smallest rise since the BLS went to a rental equivalency measure in 1983. Home prices, as measured by the Office of Federal Housing Enterprise Oversight's Home Price Index, rose 7.7 percent in the first quarter from a year earlier, in line with the gains for the last four years.

It makes sense that the rent on a hot home in a hot area would reflect the price and desirability. It doesn't seem to work that way in the world of government statistics.

``The real problem is that people look at the house price and think the rental equivalence is highly correlated,'' says Pat Jackman, a BLS economist with the CPI. ``In the short run, rental equivalency may not be a good measure of consumer costs (of housing). In the long run, it is.''

While finding rental units to survey in owner-dominated areas is difficult, these units get a heavy weighting in computing OER, Jackman says.

Model Homes

Carson created a statistical model of rental costs (vacancy rates are the big driver) that tracks the BLS measure of rents closely over the last decade. A similar model for rents on owner- occupied homes tracks OER until 1997, after which the two series diverge. Carson's model for owners' rents suggests the BLS series has been ``understated by more than 3 percentage points per year for the last three years,'' he says.

The accuracy of imputing owner-occupied rents wouldn't be an issue if it had a small weighting in the CPI. When the component in question is the single largest in the index, accounting for 23.4 percent of the CPI and almost 30 percent of the core CPI, accuracy is a big deal.

When you consider that the CPI's big mover is an imputed price -- no transactions ever take place at that price -- it puts any single month's reading in perspective. It also calls into question the intense market reaction when the CPI comes in as little as 0.1 percentage point above or below expectations.

What about Fed policy makers, who rely on these same inflation measures for feedback on their effort to achieve and maintain price stability?

``Inflation should be measured by recording the prices of goods and services actually bought and sold,'' Carson says. ``Hypothetical imputed prices should be ignored by policy makers if for no other reason than that they can be misleading about inflation impulses.''

To contact the writer of this column:
Caroline Baum in New York at [e-mail].

To contact the editor responsible for this column:
Bill Ahearn at [e-mail].