Register now for a free trial to gain access to
this piece and see how
you can benefit from an RGE subscription.
A Fresh Look at the Housing Market: What Has Changed?
By Prajakta Bhide and Christian Menegatti
Back in September 2009—when U.S. housing market indicators were beginning to look up—analysts began heralding the data as a sign of stabilization and subsequent sustainable recovery in the housing market. Indeed, housing starts and new and existing home sales were improving, and home prices appeared to be finally bottoming out after a long decline that left home values about 30% below the peak reached in Q1 2006, according to the S&P/Case-Shiller national home price index. However, examining the underlying forces at work in the housing market showed that the stabilization was of a temporary nature and held together by extensive government support.
Based on data for a range of housing market indicators, we generated detailed forecasts for U.S. home prices across major cities and on a nationwide basis. Our quantitative results turned out to be very much in line with our expectations: While there was considerable divergence in outcomes across U.S. cities, on an average, nationwide basis, home prices were expected to correct downward through 2010 following the withdrawal of the first-time homebuyer tax credit in November 2009. Indeed, to a certain extent, our view was confirmed in the early months of 2010, as housing data softened and home price indexes showed month over month declines. However, the downward correction was limited, as the distortion in the market from the homebuyer tax credit was pushed further ahead through April 2010.
With the extended April 30 deadline of the first-time homebuyer tax credit now behind us (and another extension of this “expensive” tax credit unlikely) we decided to take another look at the housing market and the implications of demand and supply fundamentals for home prices in the coming years. Our results continue to bolster our view that the first-time homebuyer tax credit has merely delayed correction in the housing market, where clearly deflation has yet to run its full course.
Forecasting Framework: Unsnarling the Tangles
To fully capture the distortion of the homebuyer tax credit on the housing market, we first constructed a "pseudo" forecast series of single-family existing home sales—a forecast of existing home sales under the assumption that the first-time homebuyer tax credit expired as scheduled in November 2009. (The existing home sales data are published on a monthly basis by the National Association of Realtors.) This involved detailed examination of the trend in sales through 2009, the boost to sales from the original tax credit and the payback in sales in the early months following the canceled expiration. Estimating this pseudo series of home sales allowed us to factor out the boost effect of the extended tax credit from recent data and also allowed us to project the payback in sales after April 30—the final tax credit deadline. The estimated boost effect was smaller than that witnessed during the buying rush of 2009—a predictable phenomenon, given that a large number of sales in 2009 (when the extension of the tax credit was not yet embedded in the expectations) had likely been pulled forward from 2010.
In our home forecasting framework, we considered the movement of existing home sales (accounting for the estimated tax credit boom-bust effect in 2010), the owner’s equivalent rent as published by the U.S. Bureau of Labor Statistics, homeowner vacancy rates published by the U.S. Census Bureau and home prices as per the S&P/Case-Shiller national home price index. Unlike the earlier home price forecasts, this time we chose to work with seasonally unadjusted data. This stems from concerns surrounding the effectiveness of seasonal adjustment in home price data in the current stage of the housing cycle. An April 22 statement by S&P observed, "In some recent reports the two series [seasonally and non-seasonally adjusted home prices] have given conflicting signals, with the seasonally adjusted series rising month over month and the unadjusted series declining. After reviewing the data, the S&P/Case-Shiller Home Price Index Committee believes that, for the present, the unadjusted series is a more reliable indicator and, thus, reports should focus on the year over year changes where seasonal shifts are not a factor.”
Forecasts generated from a vector error correction model showed that by Q4 2010, non-seasonally adjusted home prices are projected to correct downward by an additional 6% on a year over year basis, (lower than the estimate of a 9% downward correction implied by our previous forecast estimates). Home prices were projected to remain sluggish through 2011 before picking up about 2% in 2012.
Figure 1: U.S. National Home Prices
Source: S&P, RGE Calculations
We Don’t See a Rebound Just Yet
The latest available S&P/Case-Shiller national home price index showed a 1.3% q/q seasonally adjusted decline in Q1 2010 after three consecutive quarters of gains. While Q2 2010 may show the effects of the extended home buyer tax credit, our forecasts continue to suggest further downward correction in home prices. Incoming data on the housing market after the tax credit deadline show a predictable collapse. In May, mortgage applications collapsed to 1997 levels and builder confidence regarding current and future sales expectations moved sharply lower in June. While a collapse in new home sales was largely unexpected, the estimated 32.7% m/m decline in May was unprecedented in magnitude. Also, single family existing home sales fell 1.6% m/m in May, when most analysts expected sales to at least post positive readings through June, as contracts signed before the tax credit deadline were gradually closed. These signs indicate the underlying weakness in demand. The collapse in housing starts and building permits in May is also telling; current and estimated ”shadow” housing inventories remain far too bloated to boost a recovery in construction, given weak demand-side indicators. In the medium term, the moderation in new supply and absorption of existing housing inventories will clear the path to a gradual recovery in the housing market. While long-term forecasts must always be taken with a grain of salt, we’re led to believe that the recovery in housing will only gather pace in earnest in 2012.
Technical Note: All variables are used at a quarterly frequency in natural logarithm form. Variables are tested for their order of integration using the Augmented Dickey Fuller (ADF) test. All series are found to be I(1). A Vector Error Correction Model is estimated with the appropriate lag length as selected by the Akaike Information Criterion from an unrestricted VAR estimated with the variables. Since the data is non-seasonally adjusted, the model includes centered seasonal dummies. The Trace test was used to establish the number of cointegrating relationships among the variables, using critical values tabulated in MacKinnon, Haug and Michelis (1999). Residuals were tested for remaining serial correlation and normality.