The Office of Federal Housing Enterprise Oversight (OFHEO) announced their quarterly home price indexes for the U.S., the 9 Census divisions, all individual 50 states, and 381 Metropolitan Statistical Areas (MSAs).
The OFHEO analyzes the mortgage records of Fannie Mae and Freddie Mac's conforming mortgage transactions (currently $417,000 maximum loan amounts and "temporarily" up to $729,750 in high-cost areas). They analyze repeat transactions to determine the home price index. Unlike the S&P Case-Shiller index which only uses purchases, the OFHEO analyzes refinances and purchases. They also issue a purchase only index for the U.S., the Census divisions and the individual states.
Before seasonal adjustments, the U.S. index was virtually unchanged versus a year ago with a decline of 0.03%. The U.S. purchases only index showed a decline of 3.07%. In contrast, the S&P Case-Shiller Composite-10 index was down 13.6% versus a year ago in the latest reading. This is quite a disparity.
The first reason for the disparity is the geographic composition. The S&P Case-Shiller Composite-10 is an index of 10 cities: Los Angeles, Miami, Washington D.C., San Diego, Las Vegas, San Francisco, New York, Boston, Chicago, and Denver. There are a lot of cities in that index that had a huge runup in home prices followed by a large decline. The composite-10 covers 30.2% of U.S. real estate. The rest of the country did not have as large of a swing. The S&P Case-Shiller Composite-20 adds Tampa, Phoenix, Seattle, Portland, Minneapolis, Atlanta, Charlotte, Dallas, Cleveland, and Detroit to the list. The spike up and down in these cities was not as dramatic. The composite-20 covers 42.5% of U.S. real estate. Finally, S&P Case-Shiller put out a National index that covers all 9 Census divisions and covers about 70.8% of the U.S. Real estate. Fittingly, the National index did not move up and down as the other two indexes. The OFHEO, having a national composition, had the flattest curve of all the indexes.
The OFHEO issued a research paper detailing the major factors for the disparity between the S&P Case Shiller index and the OFHEO index. They found that besides the geographic makeup, there are 3 major differences accounting for the other variance. The first major difference is the fact that the OFHEO's index looks at both purchases and refinances whereas the S&P Case-Shiller index only looks at purchases. Purchase transactions are considered to be more accurate. For one, it is the price at which the buyer is valuing the property. The buyer does not want to pay more than a fair price. In a refinance transaction, there may be pressure to overinflate the value (to maximize cash out, or a certain value may be needed for the transaction to work). The refinance values are also staledated. An appraisal can usually be up to 120 days old by the time the property is completed. The appraisal uses comparable sales to value the house. The comparables that the appraiser uses are sometimes up to 6 months old. A typical appraisal may have comparables that are 4 to 6 months old by the time the loan closes but could be up to 10 months old (120 days plus 6 months). In an update to the study, the OFHEO found that on average their figures were 6.87% higher than the S&P Case-Shiller year over year declines for the 10 cities in the Composite-10. By eliminating refinances from the data, the discrepancy was reduced by 2.38%.
The second big factor in the difference between the two indexes is the weight that is given to homes that have lengthy intervals between valuations. OFHEO discounts homes with long intervals between valuations more than the S&P Case-Shiller index does. This led to a bigger variance by 1.35%. The biggest difference between the two indexes was the omission of homes that sold with financing other than Fannie Mae and Freddie Mac loans (subprime loans, jumbo mortgages, VA, FHA and other types of financing arrangements). This led to a variance of 2.85%. The majority of ARMs, and interest only loans were financed outside of Fannie Mae and Freddie Mac. Borrowers with this type of financing may have overpaid on their purchases. For example if a borrower wanted $2000 a month payments and the interest rate was 6%, then loan amount for an interest only loan would be $400,000; a fully amortized loan would be $333,583. An interest only loan would give more buying power. Moreover, ARMs usually had a lower rate inflating the possible loan amount at those payments even more. Another reason for the variance in non agency loans was the lack of skin in the game, or 100% financing. A borrower without a vested interest in a property is apt to take more risks. If a borrower overpays on a house and puts 10-20% down, then they risk losing their money. A borrower that puts 0% wins if the property shoots up in value. If it drops in value they lose, or in today's world they walk away. 100% financing also led to overinflated purchase prices in some cases. For example, if a borrower wanted to buy a $200,000 house, but had no money for a downpayment or closing costs, they might structure the purchase at $205,000 with a seller credit to the buyer for $5,000 to cover closing costs. They would get a loan for $205,000 and basically have the bank finance the closing costs. A borrower putting 10-20% down might not want to do this. Surprisingly the loans that were not eligible for sale to Fannie Mae and Freddie Mac due to loan amount (over $417,000 loan amount) did not contribute to the larger declines in the S&P Case-Shiller index (they lessened the decline by 0.19%). It was the lower and mid priced homes with non-agency financing that led to the variance of 2.85%. This is contrary to the common assumption that the reason for the variance is the loan size constraints.
Here are graphs of the 10 cities in the composite-10 comparing the S&P Case-Shiller index to the OFHEO index. The OFHEO index (includes refinances) lags the S&P Case-Shiller index by around 6 months. Amazingly the peaks reached were very similar across the board. The lag between the indexes was also fairly consistent.
Sphere: Related Content
No comments:
Post a Comment