Regular readers are all too familiar with the saga of Goldman Sachs, which back in December 2010 [7]called for a new American golden age, only to crash and burn as the economy not only slid right back into its depressionary glidepath but had to be bailed out by the Fed yet again. Sure enough, back in December of last year, the same firm made a surprising forecast, being the first of many (as others naturally jumped on the Goldman bandwagon), calling for an imminent housing bottom. Naturally, we scoffed at said proclamation. Two months later, which have seen two months of deteriorating conditions and declining prices, Goldman is out, saying that it may have just been kidding. From Goldman's Hui Shan: "In December 2011 we published a new house price model for 147 metro areas that pointed to a decline of around 3% from mid-2011 through mid-2012 before stabilizing in the year thereafter. Excess supply and negative house price momentum were the main drivers of the projected decline over the subsequent four quarters. In the year thereafter, the model suggested that house prices would stabilize as the negative momentum faded. Our model also pointed to substantial variation in house price appreciation across metro areas. Although city-by-city house price dynamics are particularly difficult to model, we projected increases in Detroit, Miami and Cleveland, but significant declines in Portland, New York and Atlanta during the next two years. Since publication of this forecast--which was based on Case-Shiller house price data up to 2011Q2--house prices have weakened anew....The implications of these changes are threefold: First, we now see a somewhat weaker near-term house price outlook. Specifically, we forecast that house prices will decline by 3.3% from 2011Q3 until 2012Q3, and by an additional 1.1% between 2012Q3 and 2013Q3. Second, the expected bottom in house prices is pushed out from end-2012 to mid-2013. Third, the long-run outlook for house prices is not significantly affected by our update." So for anyone basing their housing recovery call on Goldman, sorry - Goldman was only kidding. Again.
Full note
House Prices Bottom: A Bit More Distant, But Still in Sight
- In December 2011 we published a new house price model for 147 metro areas that pointed to a decline of around 3% from mid-2011 through mid-2012 before stabilizing in the year thereafter. Since publication of the model--which was based on Case-Shiller house price data up to 2011Q2--the decline in house prices has reaccelerated slightly. In today's (February 29) comment we update our forecast in light of this and also use the opportunity to make a couple of technical changes to the model.
- We now project that house prices will decline by around 3% from 2011Q3 until 2012Q3, and by an additional 1% in the year thereafter. As a result, the expected bottom in house prices is pushed out from end-2012 to mid-2013. Although the house price outlook has weakened very slightly, we believe that the house price bottom remains in sight.
In December 2011 we published a new house price model for 147 metro areas that pointed to a decline of around 3% from mid-2011 through mid-2012 before stabilizing in the year thereafter. (For details see Hui Shan and Sven Jari Stehn, "US House Price Bottom in Sight," Global Economics Paper, No. 209.) Excess supply and negative house price momentum were the main drivers of the projected decline over the subsequent four quarters. In the year thereafter, the model suggested that house prices would stabilize as the negative momentum faded. Our model also pointed to substantial variation in house price appreciation across metro areas. Although city-by-city house price dynamics are particularly difficult to model, we projected increases in Detroit, Miami and Cleveland, but significant declines in Portland, New York and Atlanta during the next two years.
Since publication of this forecast--which was based on Case-Shiller house price data up to 2011Q2--house prices have weakened anew: the national Case-Shiller house price index declined by 1.4% in 2011Q3 and 1.7% in 2011Q4. In today's comment we update our forecast in light of this and also use the opportunity to make a couple of technical changes to the model.
First, we update our model with the 2011Q3 Case-Shiller house price indexes, which are now available for all of the 147 metro areas in our sample. Unfortunately, we do not yet have 2011Q4 Case-Shiller data available for all metro areas.
In a second step we incorporate the 2011Q4 20-city Case-Shiller house price data, which was just released yesterday (February 28). To take into account this new information, we adjust our forecasts for these 20 cities (but keep the forecasts for the remaining 127 metro areas unchanged). For example, if our model projects a city’s house price to increase by 4% from 2011Q3 to 2012Q3 but the Case-Shiller index showed a decline of 2% in 2011Q4, then our adjusted forecast would be 1% ( = -2% + 0.75*4%). In this way, we ensure our forecasts incorporate all the information that we have at the time of forecasting.
Third, we make a technical change to our model. We previously used a four-quarter-ahead specification for the first-year forecast and an eight-quarter-ahead specification for the second-year forecast. Moreover, we used the four-quarter model with dynamic updating to forecast house prices in the next 10 years (i.e. the first-year forecast was fed into the model to generate the second-year forecast, and so on). We decided not to use the eight-quarter-ahead model any longer because it fits the data less well compared with the four-quarter specification and it may generate forecasts that are inconsistent with the four-quarter model with dynamic updating. In addition, our original model used the contemporaneous homeowner vacancy rate as one of the control variables. To reduce the measurement error in the quarterly MSA-level vacancy rate measure and minimize potential attenuation bias, we now use its four-quarter moving average. These technical changes have relatively small effects on our results.
Finally, we incorporate updated CBO long-run economic projections for inflation, wages and salaries, and 10-year Treasury yield which we use to calculate our long-run house price forecasts.
The implications of these changes are threefold:
First, we now see a somewhat weaker near-term house price outlook (Exhibit 1). Specifically, we forecast that house prices will decline by 3.3% from 2011Q3 until 2012Q3, and by an additional 1.1% between 2012Q3 and 2013Q3. The key reason for this downgrade in the forecast is the more negative trend in house prices since mid-2011 (rather than the technical changes). The forecast still displays substantial regional variation. We now expect (small) increases during the next two years in Detroit, Cleveland and Phoenix, and significant declines in Atlanta, Los Angeles and New York. (For a detailed list with our new forecasts for all metro areas, see US House Price Projections by Metro Area.)
Exhibit 1: Our Updated House Price Model Shows More Near-Term Weakness…
Second, the expected bottom in house prices is pushed out from end-2012 to mid-2013 (Exhibit 2). This change looks fairly large when compared with the relatively small shift in the predicted percent change in house prices. However, the reason is simply that the predicted pattern of house prices is quite "flat" around the bottom.
Exhibit 2: … And A Later House Price Bottom
Third, the long-run outlook for house prices is not significantly affected by our update (Exhibit 2). Although our updated model points to a slightly lower path for equilibrium house prices--as a result of the CBO's new economic forecasts--we continue to expect nominal house price appreciation of around 30% over the next ten years.
The bottom line from our updated model is that the near-term house price outlook looks softer in light of recent weakness in house prices. But our core view--that the house price bottom is in sight--remains intact.