Nationally, the median home price fell roughly 22% between 2006 and 2010. Most metro areas experienced similar price declines, but these declines were not spread evenly within metro areas. A simple examination of price data suggests that prices in the upper third of most metro areas were more resilient to price declines than at the lower and middle tiers during the down-swing in home prices.
Each month, S&P/Case-Shiller release a group of supplemental price indexes along with their headline figures. Among these indexes are data series which track same-sale price movements for three tiers of home sales. The tiers are determined by evenly dividing the sales volume into three segments by sold price. These price tiers are calculated separately for each of 16[1] metro areas.[i]
Simple statistics about the 16 metro areas reveal important trends. In every metro area, the high tier experienced the smallest decline over the 5-year period between April of 2006 and April of 2011.[ii] In addition, the low tier experienced a sharp rebound in price growth subsequent to the sharp decline in most metro areas. Both of these trends are depicted in the graph of price trends for the San Francisco market above.
This pattern was shared by metros with a lower median price as well. For example, the same two trends are evident in the prices for Denver. Two factors may account for these trends. First, during the housing boom, neighborhoods coveted for their short commutes, good schools and other features experienced sharp price gains earliest in response to falling mortgage rates. As prices in these areas peaked, prices in more far flung areas or those with less desirable characteristics began to escalate. After the market bust, these submarkets began to re-price, leaving those neighborhoods with more desirable characteristics with less room to decline as they were relatively closer to their “correct” price. In addition, just as in investing, demand for homes with strong pricing characteristics likely rose as buyers sought out homes with characteristics that would make them more resilient to price declines.
The resilience of prices in the S&P’s highest price tier index has important implications for one of the current policy debates in real estate. Lending not backed by the Federal government evaporated in the wake of the credit crisis, so the conforming loan limits were temporarily raised in 2008 to extend access to credit to higher-priced portions of some markets. These temporary limits are set to expire on October 1st. The local median home price as determined by HUD in 2008 was used to create the proposed limits in most cases, but in some cases it was based on data from 2009 and 2010. It is widely assumed that since prices have fallen significantly since 2008 that prices for homes in the temporarily expanded conforming loan ranges have fallen enough to make the temporary loan limits unnecessary. This may be true in some markets, but not for all. For example, the limit at which the FHA can originate loans for homes in Boston is $523,750 (set in 2008). That limit will decline to $465,750 on October 1st, a decline of $58,000 or 11.1%[iii]. However, as depicted in Table 2, the S&P’s top price tier index in Boston only fell 4.1% between April of 2008 and April of 2011. Since this is a same-sale index, it suggests that the mortgages for many of the homes that were in the expanded conforming range back in 2008 would not have fallen enough to be in the lower limit proposed for the fall. A home with a mortgage of $500,000 in April of 2008 would likely need a mortgage of $479,500 (assuming the same downpayment), in April of 2011, but would no longer be eligible for a conforming loan in October. Five other markets followed this pattern of a stronger decline in the proposed conforming limit than prices since 2008 including Denver, San Diego, Los Angeles, the suburbs of New York City, and Washington, DC.
Three points should be made regarding the price declines in the S&P’s upper-tier price indexes. First, the lower limit of the highest S&P price tier index for most metro areas is well below the local conforming loan limit. As stated earlier, the price declines moderate as the price tiers rise. This suggests that prices for homes at the upper end of the 3rd tier may have been more resilient than the growth rate for the 3rd tier as a whole suggests. If this is the case, more metro areas would be negatively impacted by the reduction in the conforming limits. Second, much of the decline in home prices occurred in 2007 and 2008 and several markets have seen subsequent price stability or improvement. Consequently, markets where the limit was based on 2008 median price data may not reflect current price conditions, again pushing some borrowers above the proposed conforming limits. Finally, as prices begin to rise, more homes will fall outside of the new, lower conforming loan limits, regardless of whether the market has stabilized.
How much of an impact will the new limits have? The FHFA estimates that loans above the new limits will experience an increase in mortgage rates of 0.5% to 0.75%, enough to make a home unaffordable. That difference contributed to a significant reduction in jumbo homes sales, those above the conforming limit, between 2008 and 2010.
The trends depicted in the tiered S&P Case-Schiller price indexes suggest that home prices in the highest tiers did not fall as much as prices for metro areas as a whole. This pattern makes sense in light of a scenario where markets re-price around traditional drivers of demand. It also suggests that it is not the case that falling prices at the upper echelon of the market have made the extension of GSE and FHA loans to high-cost markets unnecessary. To the contrary, the data suggests that this portion of the market may still be dependent on access to government backed credit and that lowering the conforming limits would expose these borrowers to significantly higher mortgage rates.
[i]
[ii]
[iii]
[1] There are 17 price series, but the S&P has not published the series for Cleveland since November of 2008 due to issues with the data.