Thursday, September 26, 2013

How Helpful is the Price-to-Income Ratio in Flagging Bubbles?

by Rocio Sanchez-Moyano
Research Assistant
With the continued growth of house prices across the country, talk of a housing bubble is beginning to reappear in the headlines. House price-to-income ratios are often used to indicate a bubble, as prices have historically had a relatively stable relationship with incomes (both mean and median).  In the US, nationally, the price-to-income ratio remained relatively stable throughout the 1990s.  It began to increase around 2000 and surpassed its long-run average of 3.65 by 2002 (Figure 1).  The national price-to-income ratio continued to increase in the mid-2000s, reaching a high of 4.63 in 2006 before rapidly declining in 2007 and 2008 and eventually hitting a low of 3.26 in 2011.  The classic bubble shape is clearly visible in this trend.  Recent price gains, viewed in this context, do not seem to indicate the return of a bubble; price-to-income ratios today match their early 1990s rates and still have some room for growth before reaching their long-run average.  (Click chart to enlarge.)


Notes: Prices are 1991 National Association of Realtors® Median Existing Single-Family Home Prices, indexed by the FHFA Expanded-Data House Price Index.  Incomes are median household incomes.
Sources: JCHS tabulations of FHFA Expanded-Data House Price Index;  US Census Bureau, Moody’s Analytics Estimates.

However, despite the seemingly straightforward relationship between house prices and incomes in Figure 1, this indicator can be difficult to interpret.  To start, many data sources are available for measuring prices.  One used frequently is the National Association of Realtors ® (NAR) Single-Family Median Home Price as it is widely available for many metros and provides an actual house price (rather than an index showing change in values) that can be compared to income levels.  The disadvantage to this measure is that NAR house prices also capture changes in the types of units that are being sold over time and so does not reflect how the value of the same home changes.  Repeat sales indices, like the Federal Housing Finance Authority’s (FHFA) Expanded-Data House Price Index, which was used to produce the figures in this post, are designed to take into account changes in the values of homes themselves by tracking sales of the same homes over time.  However, the FHFA index can be more difficult to interpret since, as an index, it does not provide information about current prices.  Price-to-income ratios using this data must peg the index to a starting or ending house price.

Furthermore, identifying bubbles or other price anomalies from price-to-income ratios can be difficult because it is not clear what is an appropriate baseline value of the measure for comparison.  Even in the aggregate US case, where the ratio did not fluctuate more than one percent in either direction for much of the 1990s, the linear trend is not flat and the long run average is above the 1990s levels.  This becomes even murkier when observing trends at the metro level.  Some metros, like Dallas, had stable price-to-income ratios over the last two decades (Figure 2).  Dallas did not experience a significant bubble in the mid-2000s and its long-run average mirrors the linear trend.  In other metros, like Phoenix, the boom-bust period led to significant fluctuations in the price-to-income ratio after having been relatively stable in the 1990s.  If the 1990s levels are to be considered normal for Phoenix, then current price-to-income ratios remain below average and recent growth in prices can be considered a return to normal after an overcorrection.

For other metros, the price-to-income trend is more difficult to interpret.  Ratios in Cleveland are well below their long-run average, but the historical trend has been drifting downwards, so ratios in recent years could be indicating a reset of the ratio in Cleveland to lower levels.  At another end, San Francisco has experienced a wide range of price-to-income ratios in recent history.  Price-to-income ratios boomed in the late 1980s, decreased throughout much of the 1990s, and then surged through the mid-2000s.  Compared to its long-run average, ratios in San Francisco are above historical norms, but, when the historical trend is considered, prices can continue to increase before they appear “too high.”  Finally, if a fundamental relationship exists between prices and incomes, it is unclear why the ratio can vary significantly from metro to metro.  The national average is around 3.6.  In the metros observed here, Cleveland and Dallas both have historic averages below 3.0 while San Francisco’s is double the national average. (Click chart to enlarge.)


Notes: Prices are 1991 National Association of Realtors® Median Existing Single-Family Home Prices, indexed by the FHFA Expanded-Data House Price Index.  Incomes are median household incomes.
Sources: JCHS tabulations of FHFA Expanded-Data House Price Index;  US Census Bureau, Moody’s Analytics Estimates

Given this variation, what can we make of the price-to-income ratio?  On a national level, this ratio does a relatively good job of identifying substantive shifts in the market.  In the aggregate, there appears to be a “normal” price-to-income ratio and prolonged deviation from this trend can signal an underlying shift.  However, on a metro-by-metro level, where it can be difficult to identify an appropriate baseline value, long-run historical context is necessary to interpret point-in-time estimates.  In markets like Dallas and Phoenix, historical trends are consistent enough that it can be useful to compare the current ratio to past ones.  In others, like Cleveland and San Francisco, the price-to-income ratio on its own is not especially helpful since there is no clear way to identify a “normal” price-to-income ratio.

1 comment:

  1. Dear Rocio. Thanks for your sharing. may I ask, if you , or your colleagues, have gather info about the house price-to-income ratio for diff metro in diff years that I can conveniently download? thanks. charles

    ReplyDelete