The Role of Leverage in the Asset Bubble in Housing in the USA
The hypothesis is that in the major metropolitain areas studied in this paper, Atlanta, Boston, Los Angeles, and Miami, the size of each bubble as measured by the Case-Schiller (CS) index would be larger for cities that experienced the largest increase in leverage, as measured by the US census’ American Housing Survey (AHS) in Mortgage Characteristics of Owner Occupied Units and their median outstanding principle. The results were different than expected; in the cities with the smallest housing bubble, Atlanta and Boston, the growth in the CS index was similar to the growth in total outstanding principle and the current total loan as a percent of value was essentially unchanged, whereas in the cities with the largest housing bubble, Los Angeles and Miami, the growth in total principle outsanding was significantly less than the growth in the index and the current total loan as a percent of value fell.
The expectation that the rise in asset prices was related to an increase in leverage is due to the behavioural aspect of finance where people will buy an asset with the expectation that they can sell it in the future at a higher price, known as ‘the castle-in-the-air theory’. (Malkiel, pg. 30)[1] If this belief is paired with greater access to liquidity (mortgage financing) people who would otherwise not be able to enter the market will be able to purchase the asset by putting little or no of their money as a down payment. Prices will rise due to the increase in demand for the asset likely out pacing the increase in supply spurred on from higher prices.
Leverage also increases investors risk because with a certain percentage rise (or fall) of the asset there is a greater total rise (or fall) in the investors’ return even after deducting interest costs. In the housing market this means that the higher the loan to value ratio (current total loan as percent of value) the more risk there is of homeowners defaulting if the value of housing prices drops. (MacGee, pg. 1)[2]
The asset bubble in housing was noticeably different in the 20 major metropolitan areas of the US, as shown on page 2.[3] The differences in the four cities compared to the 20 city composite index are illustrated on page 6. This is interesting considering that all US cities had the same monetary policy of low interest rates, and federal regulations allowed for easier access to mortgage financing across the US. (MacGee)
In Atlanta, the median total principal amount grew at a rate of 6.10% from 1996 to 2004, whereas the CS index in the same period grew at 5% (page 7), the current total loan as percent of value was flat at -0.02%.[4]
In Boston, the median total outstanding principal amount grew at a rate of 9.66% from 1998 to 2007, while the CS index grew in that period a similar 8.10% (page 7), and the current total loan as percent of value decreased at a rate of 1.36%.[5]
In Los Angeles, median total outstanding principal grew 5.60% from 1999 to 2003, the CS index grew 13.18% (page 8), and current total loan as percent of value decreased at a rate of 13.18%.[6]
In Miami, the median total outstanding principal grew 10.26% from 2002 to 2007, the CS index grew an amazing 14.44% (page 8), and current total loan as a percent of value decreased by 4.66%.[7]
The values from the CS index and the AHS are summarized in Table 1.
Table 1
| Medians | Atlanta | Boston | Los Angeles | Miami | ||||
| Total Outstanding Principal Amount | 1996 | 70,341 | 1998 | 79,667 | 1999 | 117,295 | 2002 | 75,342 |
| 2004 | 113,004 | 2007 | 182,740 | 2003 | 145,884 | 2007 | 122,764 | |
| Growth rate | 6.10% | Growth rate | 9.66% | Growth rate | 5.60% | Growth rate | 10.26% | |
| Case-Schiller Index Growth | 1996 | 82.17 | 1998 | 83.97 | 1999 | 96.12 | 2002 | 132.58 |
| 2004 | 121.45 | 2007 | 169.28 | 2003 | 157.18 | 2007 | 260.25 | |
| Growth rate | 5.00% | Growth rate | 8.10% | Growth rate | 13.18% | Growth rate | 14.44% | |
| Current Total Loan as Percent of Value | 1996 | 67.5 | 1998 | 45.8 | 1999 | 61.1 | 2002 | 56.5 |
| 2004 | 67.4 | 2007 | 40.5 | 2003 | 47.2 | 2007 | 44.5 | |
| Growth rate | -0.02% | Growth rate | -1.36% | Growth rate | -6.25 | Growth rate | -4.66% | |
These bizarre results may be caused by a few different factors. The US Census Bureau conducts the American Housing Survey (AHS) and they measures mortgage characteristics. For the four cities the median term of the primary mortgage at origination was 30 years[8]. There could be problems in the AHS data used to measure leverage, because some people who responded simply may not have known how much outstanding principle they had. Especially because the AHS includes the total number of households in each surveyed area that have 1, 2, and 3 or more mortgages, and there was a considerable number of households with 2 mortgages or 3 or more mortgages. They might have only reported information for the first mortgage. The AHS survey is conducted in different years (with no regularity) for different metropolitan areas. This makes comparisons difficult both between cities and to observe changes between years. For the 4 cities compared the survey was conducted since 1974 at intervals of between 3 to 9 years for Boston, 3 to 8 years for Atlanta, 3 to 6 years for LA, and 3 to 7 years for Miami.[9] Due to this irregularity growth in mortgages outstanding could not be compared between the cities because the time periods were different, but trends were still noticeable in the time period leading to the peak in the housing bubble (peak in July 2006 in CS index). High quality yearly mortgage data separated by metropolitan areas would be preferable to improve this research. The CS index is based on census and Fiserv data[10] and is likely very accurate.
The magnitude of the housing bubble in the four cities does not seem to be linked to the increase in leverage according to the limited data provided by the AHS. An alternative that would improve this research is to collect mortgage information from all the lending institutions in a metropolitan area, monthly or quarterly, which would provide better more accurate information than the AHS which relies on people knowing their principle outstanding on all their mortgages published at varying intervals.
References
“American Housing Survey for the Atlanta Metropolitan Area in 1996”. U.S. Department of Commerce. Bureau of the Census. Nov 1997. Pg 59-60.
“American Housing Survey for the Atlanta Metropolitan Area: 2004”. U.S. Department of Commerce. Bureau of the Census. Oct 2005. Pg. 76-77.
“American Housing Survey for the Boston Metropolitan Area 1998”. U.S. Department of Commerce. Bureau of the Census. Nov 2000. Pg. 70.
“American Housing Survey for the Boston Metropolitan Area: 2007”. U.S. Department of Commerce. Bureau of the Census. Feb 2009. Pg. 80-1.
“American Housing Survey for the Los Angeles-Long Beach Metropolitan Area 1999”. U.S. Department of Commerce. Bureau of the Census. March 2001. Pg. 71.
“American Housing Survey for the Los Angeles-Long Beach Metropolitan Area: 2003”. U.S. Department of Commerce. Bureau of the Census. Dec 2004. Pg. 76-7.
“American Housing Survey for the Miami-Ft. Lauderdale Metropolitan Area: 2002”. U.S. Department of Commerce. Bureau of the Census. July 2003. Pg. 74-5.
“American Housing Survey for the Miami-Ft. Lauderdale Metropolitan Area: 2007”. U.S. Department of Commerce. Bureau of the Census. Feb 2009. Pg. 80-1.
Gjerstad, Steven and Smith, Vernon L. “Monetary Policy, Credit Extension, and Housing Bubbles: 2008 and 1929”. Critical Review. 21:2, Pg. 269-300.
“Index Methodology”. S&P/Case-Shiller Home Price Indices. Nov 2009.
MacGee, James. “Why Didn’t Canada’s Housing Market Go Bust?”. Federal Reserve Bank of Cleveland. Sept. 2009
Malkiel, Burton. A Random Walk Down Wall Street. New York: W. W. Norton & Company, 2007.
[1] Malkiel, Burton. A Random Walk Down Wall Street. New York: W. W. Norton & Company, 2007. Pg. 30
[2] http://www.clevelandfed.org/research/commentary/2009/0909.pdf
[3] Source for all Case-Schiller data used in this report: http://www2.standardandpoors.com/spf/pdf/index/CSHomePrice_History_022445.xls
[4] American Housing Survey for the Atlanta Metropolitan Area in 1996, pg. 60; American Housing Survey for the Atlanta Metropolitan Area: 2004, pg. 75-6
[5] American Housing Survey for the Boston Metropolitan Area 1998, pg. 70; American Housing Survey for the Boston Metropolitan Area: 2007, pg 80-1
[6] American Housing Survey for the Los Angeles-Long Beach Metropolitan Area 1999, pg 71; American Housing Survey for the Los Angeles-Long Beach Metropolitan Area: 2003, pg 76-7
[7] American Housing Survey for the Miami-Ft. Lauderdale Metropolitan Area: 2002, pg 74-5; American Housing Survey for the Miami-Ft. Lauderdale Metropolitan Area: 2007, pg 80-1
[8] AHS Atlanta 1996, 2004; Boston 1998, 2007; Los Angeles 1999, 2003; Miami 2002, 2007
[9] http://www.census.gov/hhes/www/housing/ahs/metrodates.html
[10] Fiserv uses census data for the number of single family housing units in each metropolitan region. Fiserv calculates the average and aggregate value of single family homes. The value of each unit is based on the sale date and price of a single family home, which is compared to historical records of the same home (if this data is available) which gives two points to compare for a yield rate. The yield rates are aggregated using their proprietary algorithm. Fiserv uses a number of techniques to maintain data integrity. To avoid upward bias data is excluded if there have been improvements or additions to the home since the previous sale. If there it appears that the transaction was not arms length (same family name, or the name of a property developer) it is excluded. The index is normalized to January 2000.
Source: “Index Methodology” S&P/Case-Shiller Home Price Indeces. Pg 6-18
No comments:
Post a Comment