Home Price Growth Is Strong

Despite a recent interest rate-related slowdown between the third and fourth quarters of 2022, home price appreciation (HPA) has remained strong in most regions across the U.S. for the past decade, particularly during the period immediately following the onset of the COVID-19 pandemic. The Federal Housing Finance Agency (FHFA) All Transactions Home Price Index for the U.S. (see chart 1) has recently surpassed the record-high level of 2022 as a supply-constrained housing market copes with demand from the millennial generation. Chart 1 also shows the quarter-over-quarter percentage change in the index. These data are not seasonally adjusted, so there is evident cyclicality due to elevated home buying demand in the spring and summer months.

Chart 1

The trend that is apparent at the national level masks the nuance of the price dynamics that play out in regional markets. For example, Boise, Idaho, has enjoyed home price gains for years (although it has recently cooled somewhat), presumably due to population migration out of the overpriced West Coast. Meanwhile, the Chicago housing market continues to struggle, due partly to rising inner city crime and the popularity of remote work that allows people to move out of the city core.

Portfolio Management Theory Can Help Account For Volatility

Idiosyncratic economic forces play a large role in how certain regions sustain "hot" markets while others are more "tepid" or even "cold". It is likely, however, that some portion of home price gains derives from the volatility inherent to specific local markets. To account for this effect, we draw from portfolio management theory, which aims to measure returns normalized for volatility (as measured by the standard deviation).

The most popular and well-known metric for this purpose is the Sharpe ratio, defined as the excess expected portfolio return (relative to a risk-free rate) divided by the volatility of the portfolio. This methodology can easily be applied to regional home price indices as measured by the FHFA; however, the choice of a risk-free rate is not obvious because the FHFA Home Price Index is not a direct investment option. In other words, it is not clear what the concept of "home price gains above a risk-free rate" means in practice.

While we could use the 10-year Treasury note yield as a benchmark risk-free rate, another sensible choice might be the aggregate national return of U.S. home prices, which over the long term is about 4% to 5% per year--coincidentally, on par with the current yield of the 10-year Treasury note. If we consider the time series of quarterly U.S. home price changes starting in 1990, the Sharpe ratio can be calculated for each of the metropolitan statistical areas (MSAs) in the U.S.

The results (see charts 2 and 3) indicate that home price gains in the eastern U.S. and the Great Lakes region are more likely to be below the national average. Moreover, the Pacific Northwest appears to have the largest concentration of MSAs with HPA consistently above the national average. Meanwhile, despite outpacing national price appreciation, several MSAs in California, Nevada, and Florida appear tepid due to higher volatility in those regions.

Chart 2

Chart 3

Some of the home price volatility across the U.S. was driven by price changes that occurred from 2007 to 2009 during the global financial crisis (GFC). Table 1 lists the top 10 MSAs ranked by the standard deviation of HPA from 1990 to second-quarter 2023, as well as our estimates of volatility for the same MSAs when home price changes from the GFC era are excluded. Among the 10 most volatile MSAs, the annualized standard deviation of HPA fell by 160 basis points on average when GFC period returns were removed from the analysis. It turns out, however, that while some Sharpe ratios will be impacted by the inclusion or exclusion of this relatively short but volatile period in the calculation (as the denominator in the equation will be altered), many regions of interest are insensitive to the impact of the GFC upon volatility.

Instances in which HPA volatility is markedly reduced when the GFC price movements are excluded from the calculation tend to be in MSAs that were among the hardest hit by foreclosures during the GFC. Meanwhile, the negligible volatility reduction in other areas (such as the Homosassa Springs, Fla., MSA) suggests HPA is naturally volatile there, such that withholding the GFC period has minimal impact on the calculation.

Table 1

Examining Regional Sharpe Ratios Can Suggest Causes Of HPA

The degree to which houses are over- or undervalued is an important feature to consider when assessing potential loss severity should there be defaults and foreclosures in RMBS collateral pools, and it is an important input for our Loan Evaluation and Estimate of Loss System (LEVELS) model. We calculate over- or undervaluation as the affordability (i.e., the average ratio of home price to income) of a region (state, MSA, etc.) relative to its 20-year average. Tables 2 and 3 list the top and bottom 10 MSAs ranked by Sharpe ratio (based on the full historical time series), as well as our estimates of over- or undervaluation for the same MSAs (see "U.S. Home Price Overvaluation Continues To Ease But Remains High," published Aug. 7, 2023). The top and bottom 10 Sharpe ratios are largely insensitive to whether or not the GFC period is incorporated into the calculation of volatility, suggesting that, at least for MSAs with more extreme price deviations from the national average, the Sharpe ratio is a reasonably robust estimate of normalized excess home price gains. In other words, severe events that produce short-term spikes in home price volatility don't affect our long-term assessment of best or worst performing MSAs as measured by the Sharpe ratio.

Table 2

The top three Sharpe ratios belong to Colorado MSAs, suggesting that much of the home price growth in this state is driven by supply and demand factors and that regional volatility in Colorado is a secondary factor. Springfield, Ill., has the most negative Sharpe ratio by a wide margin, despite having a less negative excess return than New Haven-Milford, Conn. Again, this is likely due to fundamental economic factors rather than volatility, as houses in Springfield have registered price growth below the national average more consistently than houses in the New Haven-Milford area. From 1990 to second-quarter 2023, about 60% of MSAs achieved price gains below the national average (i.e., a negative Sharpe ratio), with 40% seeing gains above the national average (i.e., a positive Sharpe ratio). These figures are little changed regardless of whether one applies a population-weighting in the analysis.

Table 3

Another notable feature is the lack of a strong correlation between the Sharpe ratio and over- or undervaluation (the correlation coefficient is 0.58). It would be reasonable to expect, generally, overvalued MSAs to have positive Sharpe ratios and undervalued MSAs to have negative Sharpe ratios. This is because persistent HPA reduces affordability (barring proportionate increases in income) and generates greater excess returns. However, our assessment of an MSA's over- or undervaluation does not incorporate intrinsic price volatility, while the Sharpe ratio does not directly account for household income. An MSA's Sharpe ratio and its over- or undervaluation should be regarded as only weakly codependent indicators of regional housing market dynamics, and one might complement the other in the analysis of regional price dynamics and relative affordability.

The Treynor Ratio Can Also Show Normalized Price Gains

A second measure that may be useful is the Treynor ratio, which is defined as the expected excess return normalized by the portfolio beta (i.e., the estimated coefficient obtained when linearly regressing the portfolio return against a benchmark index). Applying the Treynor ratio to home prices, using the aggregate U.S. index series for both the risk-free return and the benchmark, leads to heat maps that appear similar to those for the Sharpe ratio (see charts 4 and 5). Many MSAs ranked in the top and bottom 10 by Sharpe ratio are ranked in similar positions by Treynor ratio (see tables 4 and 5).

Chart 4

Chart 5

Table 4

Table 5

The Ratios' Differences Can Reveal Local Economic Forces

Despite the clear relationship between the Sharpe and Treynor ratios, there are instances in which the two metrics paint a different picture of an MSA's normalized excess appreciation. To identify examples of MSAs for which this is the case, we computed the difference between ranking for each MSA's Sharpe and Treynor ratio. For instance, the Houston-Woodlands-Sugar Land, Texas, MSA has the 77th-highest Sharpe ratio and the 77th-highest Treynor ratio; therefore, the difference is zero.

Roughly 4% of MSAs have rankings of Sharpe and Treynor ratios that are equal, and 62% are within 10 rank positions. Two MSAs in the West Texas Permian Basin--Midland and Odessa--have the 12th- and 29th-highest Treynor ratios, respectively. However, Midland has the 70th-highest Sharpe ratio, while Odessa has the 105th-highest Sharpe ratio. The source of the wide differential lies in the denominator of each calculation, as the numerators of both ratios are the same.

Home prices in the Midland-Odessa region appreciated greatly and suddenly because of the 2017-2022 shale boom that brought oil companies and their employees into the area, spurring demand for housing. While home prices in Odessa grew at an annual rate of only 2% between fourth-quarter 1989 and fourth-quarter 2000, HPA jumped to an annual rate of 5% during the fourth-quarter 2010 though fourth-quarter 2021 period. It's possible that the resulting price volatility inflated the denominator of the Odessa and Midland Sharpe calculations. Alternatively, it's possible that because aggregate U.S. home price returns did not undergo similar volatility during the same period, the Midland and Odessa beta values (the denominator of Treynor ratio calculations) were deflated. While further research would be required to identify the precise causes and local economic forces driving home price changes in specific MSAs, a close inspection of outliers of the Sharpe-Treynor ratio rank differences is a good place to start.

Conclusion

This analysis is preliminary, and further research is needed to determine how it might be used to better understand drivers of HPA in different MSAs across the country. The Sharpe ratio, the Treynor ratio, or a related approach could complement the use of regional price-to-income ratios to measure affordability trends across the country. However, this analysis can also be appreciated as a stand-alone framework for learning about certain drivers of regional housing markets.