by Calculated Risk on 12/06/2011 05:49:00 PM

CR Note: The following is a long discussion by Tom Lawler of the recent Fed paper looking at the role of real estate "investors" (really speculators) during the housing bubble. Lawler cautions about drawing quick conclusions from the graphs.

First, a quote from Lawler in 2005:

“The investor share of the purchase market has increased dramatically, to levels not seen since at least the late 1980’s. The investor share increase has been concentrated in MSAs with exceptionally rapid home price growth”

“(C)onditions in many parts of the country mirror past conditions that preceded regional housing ‘busts’.”
Tom Lawler, NAHB Spring Construction Conference, May 25, 2005

People were warned! (In April 2005, I wrote Housing: Speculation is the Key)

From economist Tom Lawler:

In a relatively recent Federal Reserve Bank of New York “Staff Report” entitled “Real Estate Investors, the Leverage Cycle, and the Housing Crisis,” FRB of NY economists Andrew Haughwout, Donghoon Lee, Joseph Tracy, and Wilbert van der Klaauw utilize data from the FRBNY’s Consumer Credit Panel (CCP) data set of US individuals with credit files to attempt to document the role of mortgage-financed purchases of real estate by “investors” in last decade’s housing boom and bust. Here is a summary. Here is the abstract:

“We explore a mostly undocumented but important dimension of the housing market crisis: the role played by real estate investors. Using unique credit-report data, we document large increases in the share of purchases, and subsequently delinquencies, by real estate investors. In states that experienced the largest housing booms and busts, at the peak of the market almost half of purchase mortgage originations were associated with investors. In part by apparently misreporting their intentions to occupy the property, investors took on more leverage, contributing to higher rates of default. Our findings have important implications for policies designed to address the consequences and recurrence of housing market bubbles.”

In re the data, here’s an excerpt from the report.

“We can use this information (in the datset) to separate mortgage borrowers based on how many distinct first-lien mortgage accounts appear on their credit reports. Since each property can secure at most a single first-lien mortgage, the number of such mortgages on a borrower’s credit report is a reliable, non-self reported, indicator of the minimum number of properties a given individual has borrowed against.24 This kind of information about individual borrowers is not available in loan-level data sets and thus the FRBNY CCP data provide a unique perspective into important questions about who is originating new mortgages at any point in time, as well as their subsequent behavior.”

Folks who read the paper should read it carefully, as a “quick read” can lead to a bit of confusion on what some of the data and charts mean. E.g., there are charts on investor shares of new purchase mortgage borrowing that a “quick” reader might conclude represent estimates of the shares of mortgage-financed purchases FOR investment by non-owners as opposed to estimates of the share of mortgage-financed purchases BY folks who own more than one property with a mortgage (i.e., BY investors), even if the a GIVEN purchase by an “investor” happens to be for his/her primary residence. (By and For mean very different things!)

Similarly, there are charts showing the share of first-lien mortgages outstanding owed BY “investors,” which the casual reader might incorrectly conclude represent the share of mortgages outstanding on non-owner-occupied properties owned FOR investment, as opposed to the share of first-lien mortgages owed by folks who happen to own more than one mortgage-financed property, one of which is likely to the be the real estate investor’s primary residence! (Many real-estate “investors” actually live in a home with a mortgage!

E.g., here is a chart from the report showing the “investor share” of new purchase-mortgage borrowing.

Click on graph for larger image.

(Shares are, I believe, shares of purchases by folks with “x” number of first-lien mortgages on their credit reports)

A casual reader might look at this chart and say, “holy crap,” in 2006 the share of mortgage-financed purchases FOR investment (or vacation homes) was almost 35%!

That would be an incorrect interpretation. Rather, the above chart represents an estimate of the share of mortgage-financed purchases BY individuals who had a mortgage on more than one property. That is a non-trivial distinction.

E.g., if a person who had a mortgage on his/her primary residence AND who had a mortgage on a different property (such as a vacation home) used a mortgage to finance the purchase of his/her new primary residence, that transaction would count as an “investor purchase” by the authors’ definition (a purchase BY an investor, not necessarily FOR investment).

Ah! Now inquiring minds probably want to know how the authors deal with folks who moved, but either (1) who moved before selling their previous primary residence and who carried two mortgages for a period of time; or (2) who, because of “reporting delays” on credit reports, were “shown” as carrying two mortgages when in fact they were not. Here’s how the authors attempt to deal with that.

“We observe borrowers’ credit reports on the final day of each quarter. Because there can be delays in credit reporting such that a mortgage that has been paid off may stay on the credit report for a period of time, we use the data’s panel structure to correct for these delays. Throughout this section of our analysis investor status is determined based on the maximum number of first mortgages that appear in both of the two most recent quarters. Thus we can be more confident that each first-lien we consider is in fact associated with a unique property.”

I did confirm with the authors that if I financed the purchase of our current residence in July 2007, but did not sell my previous residence until a year later (and if it had a mortgage), that I would have been classified as a “2 first mortgages” holder for the intervening period. Similarly, if I had a mortgage on our beach house as well, for a brief period I would have been classified as a “3 first mortgages” holder, and after the sale of our previous primary residence I’d be back down to a “2 first mortgages” holder. (In real life I don’t have any mortgages).

Similarly, consider another chart from the report.

Investor share of number of first mortgages outstanding

Again, a “casual” reader might look at this chart and conclude that in the first quarter of 2008 the “share” of non-owner-occupied first mortgages outstanding was almost 25%. That would be incorrect.

The authors, by the way, are not meaning to confuse. One just needs to read the report carefully.

E.g., in the above chart it seems reasonable to conclude that for most borrowers with 2 first mortgages, one is a mortgage on their primary residence. The same is probably true for the 3- and 4+- holders. That would lead one to conclude (Q&D) that in the first quarter of 2008 the % of non-owner-occupied first mortgages outstanding was more like half what is shown above¹.

¹ Hope Now estimates that non-owner-occupied share of residential first-lien mortgages outstanding at the end of September was 9.4%, but that figure is likely understated.

A final chart I’m going to show from the paper is one that estimates the “investor” share of seriously-delinquent mortgages.

Investor share of 90% DPD first mortgage balances

One of the intriguing and important issues raised in the “conclusions” section of the paper relates to a question many folks, including myself, would like answered: what % of properties backing seriously delinquent loans are actually occupied by the owner of the property? The authors believe that the CCP data suggest that the % may be significantly greater than many folks believe, though, as noted above, the data aren’t “conclusive.” E.g., Hope Now estimates that at the end of September the % of residential first-lien loans 60+ days delinquent that were non-owner-occupied properties was 11.5%,, though its 60+ numbers don’t include loans in the foreclosure process. Unfortunately, I can’t for sure tell from the above chart what the RIGHT % is – but it makes a difference both in terms of how effective mod programs targeted at owner-occupied homes will be in terms of addressing the TOTAL delinquency/foreclosure issue, AND in terms of the wisdom of foreclosure moratoria/delays/etc. E.g., for vacant properties with delinquent loans, or properties with delinquent loans that are being rented out but not adequately maintained, it probably makes sense from a policy perspective to ACCELERATE the foreclosure process!

All in all, the paper is quite interesting and does shed some new light about the role real estate “investors” played in the housing boom/bust, and it is worth reading. The paper also shows some “investor” shares for “bubble” states that are quite illuminating.

Amusingly, the paper cites the amazingly bad paper by the clueless economists Himelberg, Mayer, and Sinai from 2005 (FRBNY Staff Report) entitled “Assessing High Home Prices: Bubbles, Fundamentals, and Misperceptions,” where the authors not only displayed a complete lack of knowledge of mortgage data sources, but who also DID NOT MENTION subprime, alt-A, low/no documentation mortgages, real-estate investor behavior. In a shocking but refreshing “mea culpa” piece on the FRBNY’s blog site entitled “Failure to Forecast the Great Recession,” Simon Potter not only documents the poor performance of the FRBNY’s economic projections, but also notes that FRBNY research (1) “misunderstood” the housing boom (citing the embarrassing HMS paper, as well as the silly 2004 McCarthy and Peach paper), and (2) there was “a lack of analysis of the rapid growth of new mortgage finance.” (See The Failure to Forecast the Great Recession)

While it now appears that “occupancy fraud” was distressingly common during the real-estate boom (one wonders, by the way, why mortgage originators didn’t check credit reports to see if a buyer claiming that he/she was going to purchase a home for his/her primary residence had more than one first mortgage outstanding!), there was still evidence that the non-owner-occupied share of mortgage-financed purchases was rising during the housing boom based on “accurately” reported occupancy purposes. E.g., here are some data from HMDA.

[This data was] taken from various annual articles from the Federal Reserve Bulletin on the HMDA data. I don’t know for sure what the 2000 and beyond data were “revised.”

Data from the old LoanPerformance “prime conventional conforming” database, which was not loan level but which had aggregate stats for PCC loans, also showed rising investor shares, especially in “bubble” markets. [The next] chart [is] from a presentation I made at the NAHB Fall Construction Forecast Conference in October, 2007.

“Second” refers to second homes such as vacation homes, while “investor” refers to investment (rental income) properties.

At the NAHB Spring Construction Conference on May 25th, 2005 I noted (using this data source) to support my statement that “(t)he investor share of the purchase market has increased dramatically, to levels not seen since at least the late 1980’s. The investor share increase has been concentrated in MSAs with exceptionally rapid home price growth.”

The National Association of Realtors “Investment and Vacation Home Buyers also suggested a big increase in buying of properties “for investment” in the middle of last decade. Unfortunately I only have data back to 2003, and I should note that I’ve never been able to “reconcile” this data to either the NAR’s Profile of Home Buyers and Sellers (which is based almost but not completely on a survey of principal-residence buyers) or to other sources).

According to this survey, 68% of buyers of “investment” properties used a mortgage to finance the purchase in 2006, compared to just 39% in 2010. While I can’t find the shares for earlier years, the NAR’s “Profile of Second Home Owners” published in 2006 suggested that about 76% of folks who purchased an investment property from 2003 to 2005 used a mortgage to finance the purchase.

This is all a long-winded way of saying that despite what appears to have been a decent amount of “lying” on mortgage applications on a borrower’s intended use of his/her home purchase, there was still a significant amount of evidence available DURING the housing bubble that the share of mortgage-financed purchases by investors was rising SIGNIFICANTLY, and was rising exceptionally rapidly in the “most bubbly” of areas.

There was also, of course, LOTS of data on the rise in the subprime share of the mortgage market, the rise in the low/no doc share of the mortgage market, the rise in piggyback financing and low/no down payment purchases, the rise in interest-only/pay-option mortgages, etc.

CR Note: This was from economist Tom Lawler.