There is a lot of well-deserved YIMBY dunking going on about comments Joseph Stiglitz made as apologia for American urban zoning. I want to hone in on another comment he made, because the comment is a perfect example of the complete lack of empirical curiosity that infects economics about the 2000s boom and bust, which I touched on in the
I'm a new admirer of your work, Kevin, so please forgive me if this question has been address prior. But as a resident of the American southwest, I noticed that the location of a lot of the housing built in the 2000's was quite remote. For instance, the Lake Las Vegas community is still sub-optimally populated. Since the FC, though I've seen a lot more infill building (there is a lot of room for it in LV), which seems more practical to me. I've seen similar trends in Phoenix and elsewhere. In your opinion, does this relative location play into the boom/bust?
I'm sympathetic to your main point that although there was a credit boom, supply constraints led the housing boom prior to 2006 and the credit crunch following was largely behind the housing price collapse and supply constraints lead the ensuing price surge outside of poor urban areas. I think, though, you overstate your case and will try and stick to the empirical data to refute some of your claims or my reading of your claims.
1) There was a supply glut in Nevada post 2007, not in new buildings but in REO/Foreclosures. FNMA alone reported REO acquisitions increasing form 530 properties in 2007 to 9,418 in 2010. AZ, 751 to 20,691. Freddie was likely no different (happy to share the FNMA presentation via email as well as subsequent comments.
2) Easing credit conditions thru 2006 fueled demand and subsequent supply. To quote a John Maudlin newletter from 7/1/2005, https://www.mauldineconomics.com/frontlinethoughts/thoughts-on-the-housing-bubble-mwo070105 "Private market share went from around 30% to over 50% from early 2000’s to 2006. The growth was disportionally [sic] in the lower fico neighborhoods." True, FICO's underwritten by the agencies didn't go up much, but private label mortgages went from a estimated 30% in 2000 to the quoted 50%.
But the agencies were very active in both Alt A and subprime and amassed huge AAA portfolios of both and both made up the majority of new mortgages in places like Nevada. Over a third of homes purchased in 2005 was estimated by the National Association of Realtors as investment properties and second homes.
3) The collapse in credit was almost all private label led. Bear Stearn's and Lehman's respective collapse came as they built up positions in subordinated bonds they couldn't move and certainly weren't financing them for servicers. Banks couldn't securitize as investors went on strike. Both Freddie and Fannie tried creative fixes like simply wrapping subprime, but that proved very costly with hindsight.
4) There was support for the agencies to step in. To quote from a mid-2007 Credit Suisse research piece: "With a dysfunctional non-agency securitization market, all eyes turned towards the GSEs to provide some relief. Lenders are tightening offerings that they cannot sell to the agencies and raising rates in a dramatic fashion on products including even jumbo mortgage loans that are offered to creditworthy borrowers.
To maintain liquidity in the mortgage market, lawmakers like Senator Dodd, are requesting OFHEO, the regulator of Fannie Mae and Freddie Mac, to increase the limit on the amount of mortgages and related MBS the two GSEs can buy.
On a related note, on July 30, Freddie Mac issued a $105.6 million subprime deal, Structured Pass-Through Certificates (SPCs) Series T-074, backed by loans originated by Wells Fargo. This deal is actually a wrapped Wells Fargo Home Mortgage ABS 2007-M09 deal, to which Freddie Mac provide guarantees on four Class A Certificates. It is part of Freddie Mac's $20 billion commitment to Congress to provide liquidity to the subprime market."
I'm not sure I'm arguing against what you've written; rather, I hope to give some context. As a structured product salesman at the time, I witnessed first hand both the excess and the near disappearance of credit and its affect. As one of my previous libertarian traders told me at the time "Thank God for the agencies".
1) This is a description of excess inventory from a demand shock, not a supply glut.
2) I agree that increasingly homes were purchased by investors rather than owner-occupiers. There was not an aggregate change in credit scores on the average mortgage and the Survey of Consumer Finance shows that generally from the 90s to 2007 homeownership was increasingly among college educated professionals with higher incomes.
3) Yes. Private label collapsed in 2007. Then the GSEs took up the slack, but with much tighter standards than the GSEs had used before. The rise and fall of the private markets was associated with about 8% rise and fall in home values from 2006 to 2007. Then, the tightening at the agencies was associated with highly regressive price collapse in 2008 and after that averaged more than 20%.
4) Definitely the agencies were the backstop, and I suspect if they had not tightened standards, it would not have devolved into a generation-defining crisis.
On this side of the GFC houses were still relatively cheap in Summerlin as late as 2016. So $250k bought you a single family home in a 9/10 school district which was probably the cheapest nice neighborhood in America. Boca Raton single family homes were last $250k in 2012. This is a big iirc but I’m pretty confident I remember this correctly.
1) yes, demand collapsed as loose credit otherwise provided by private markets disappeared, but that led to foreclosures by those in homes they couldn't afford or were hoping to flip and a subsequent "glut" in inventory. That initial credit fueled demand wasn't sustainable. Not sure if we really differ other than our emphasis.
2) I'm suspect that there was not a decrease in aggregate credit scores given subprime by 2006 made up a growing % of overall mortgages, but I don't have data to support this and it never got much higher than 10% so maybe.
Still, "Pay Option Arms" (with negative amortization), 100% LTV or seconds mortgages, "Alternative A" like no income or asset verification, all had decent FICOs but reflected large drops in underwriting standards.
3 and 4) I think you underestimate the cascading effect -- an 8% fall in housing on a very high LTV mortgage led some to walk away which reduced the shame in walking away for others which led to more REO hitting the market place which led to further drops. I have a hard time believing that wiping out ~ 22% of the lending market (best estimates are that by 2006 subprime represented ~ 10% of mortgage debt, Alt A 9%, Prime IO and Option Arms 3%) didn't have more of an impact. The Prime Jumbo private market (5% market share) also collapsed but banks (like First Republic!) and REITs ( collectively ~ 25% of market) picked up much of that slack but with crazy tight lending standards.
That leaves the agencies which undoubtedly tightened their standards. But wouldn't you? Don't they have a fiduciary responsibility to limit losses to the tax payers? The market certainly thought the risk for baring mortgage credit risk was very high. Even in 2013 when Freddie sold their first credit risk transfer deal (STACR), the AA with 1.65% credit enhancement (and 1.35% thick) sold at a spread of 2.35% and that had an average FICO of 766 and a LTV of 73!
I just don't think it was politically plausible or economically sound for them to actually loosen their credit standards to take some of the slack from the private label collapse. Certainly with hindsight they should have. They did create a largely successful HAMP program to allow for loan modifications to stem the foreclosure tied. Loan forgiveness went nowhere. All this seems like a political argument of the role of government/agencies in our housing market.
I appreciate the engagement, too. I really appreciate your work but also want to give a different perspective of the agencies role in the collapse.
I think you’re right that we generally agree on these things.
One area I would nitpick is to be clear about the difference between risky terms and loose standards. The Urban Institute has a great tracker that helps highlight this issue. The subprime boom was largely about more reckless terms. The GsEs only engaged in a small amount of that. Then, after subprime fell, there was an extreme tightening of borrower standards. The GSEs never changed their customer base during the boom and then massively cut it off during the bust.
There was no cascading. That’s one of the things I’ve discovered in my work. Every conventional narrative has everything backwards. Owners defaulted because prices collapsed. Prices collapsed because prime lending that had been available for decades was pulled away.
These aren’t just vibes. I ran the numbers and was surprised that they told a new story and so now that’s what my work is.
The intuition to accept or encourage the extreme tightening of credit was the cause of the crisis. If the average credit score in September 2008 on GSE mortgages had been the same as it was in, say, 1999, none of the worst things would have happened, and even the GSEs wouldn’t have been taken over.
I think our primary disagreement is in your statement "Prices collapsed because prime lending that had ben available for decades was pulled away" and my observation that prices collapsed as DQ's started to escalate in 2006 causing private mortgage credit to tighten and escalate in 2007. That tightening in private credit was behind the stall in house prices in the hottest markets cascading into a shut-down in private markets by early 2008 and a further increase in DQs then REO in these markets.
I don't have data by month and I'll concede your data might be better than mine, but from the FHGA.gov website in "Date on the Risk Characteristics...from 2001 through 2008..." pdf they published in September of 2010, "...in years 2004 through 2006, such loans [above 660 FICOs] account for 70 percent to 72 percent of the data [total mortgage originations], whereas in other years they account for 78% [roughly pre-2004] to 92% [by 2008]." The GSEs $ volume went up of below 660 FICO originations went up in 2007 while private label orinations collapsed. I don't think you can't refute that the housing crisis started in 2006 and gained pace in 2007, seemingly before the GSE's tightened their guidelines.
To your point, by 2008, private label origination had collapsed and GSE underwriting tightened. But the bulk of the 90+ day DQ bucket was with private label originations, not the GSEs' (e.g., 2006 vintage where GSE's market share (excluding bank retained mortgages) originations share was 67% whereas the DQ shares was only 37%). The ensuring "supply glut" on REO hitting the market as adding to my view of a cascade was also primarily stemming from the private markets (unclear to me how much the agency led HEMP vs the private label servicer led modifications (including loan forgiveness) stemmed that tide but certainly not enough. It seems clear that prior to 2007, the marginal buyer (as defined by subsequently delinquent) was primarily financed largely with private label mortgages and the evaporation of that source of financing was behind the initial collapse in demand. The tightening of standards in 2008 exasperated the problem but the crisis would have happened regardless.
The conclusion that "none of the worst things would have happened' without the GSE tightening also doesn't seem consistent with what happened overseas, especially Spain and Ireland who experienced only a collapse in private funding.
Perhaps we agree to disagree on the origins and eventual extent of the private mortgage market's role relative to the GSEs and agree that the GSEs exasperated and extended the problem by not absorbing some of that lost demand stemming from the marginal buyer.
If you're interested in these questions, I would recommend poking around under the "Research" tab. You may not be satisfied with my answers to these questions, but certainly they are central to the inquiry.
Atlanta is an example of a city where there was basically no price reaction to the rise and fall of subprime from late 2003 to mid 2007, but there was a huge price reaction that started at the beginning of 2008 when credit was tightened at the GSEs. Figure 2 here shows the price/income trends in Atlanta.
The US needs 10 million additional housing units, maybe 20 million. Imagine how much credit that would take.
Stiglitz may, indirectly, have one concern sorta-kinda right: There is so little new construction, that the average age of housing units in the US is aging by nearly one year in every year.
At some point, this will begin to take a toll. Plumbing goes bad, wiring, new roofs are needed, termites etc. Windows do not keep out rain and so on.
I'm a new admirer of your work, Kevin, so please forgive me if this question has been address prior. But as a resident of the American southwest, I noticed that the location of a lot of the housing built in the 2000's was quite remote. For instance, the Lake Las Vegas community is still sub-optimally populated. Since the FC, though I've seen a lot more infill building (there is a lot of room for it in LV), which seems more practical to me. I've seen similar trends in Phoenix and elsewhere. In your opinion, does this relative location play into the boom/bust?
CPI hit 5.5% YoY in 2008…everything you’ve been told about 2008 is wrong.
I'm sympathetic to your main point that although there was a credit boom, supply constraints led the housing boom prior to 2006 and the credit crunch following was largely behind the housing price collapse and supply constraints lead the ensuing price surge outside of poor urban areas. I think, though, you overstate your case and will try and stick to the empirical data to refute some of your claims or my reading of your claims.
1) There was a supply glut in Nevada post 2007, not in new buildings but in REO/Foreclosures. FNMA alone reported REO acquisitions increasing form 530 properties in 2007 to 9,418 in 2010. AZ, 751 to 20,691. Freddie was likely no different (happy to share the FNMA presentation via email as well as subsequent comments.
2) Easing credit conditions thru 2006 fueled demand and subsequent supply. To quote a John Maudlin newletter from 7/1/2005, https://www.mauldineconomics.com/frontlinethoughts/thoughts-on-the-housing-bubble-mwo070105 "Private market share went from around 30% to over 50% from early 2000’s to 2006. The growth was disportionally [sic] in the lower fico neighborhoods." True, FICO's underwritten by the agencies didn't go up much, but private label mortgages went from a estimated 30% in 2000 to the quoted 50%.
But the agencies were very active in both Alt A and subprime and amassed huge AAA portfolios of both and both made up the majority of new mortgages in places like Nevada. Over a third of homes purchased in 2005 was estimated by the National Association of Realtors as investment properties and second homes.
3) The collapse in credit was almost all private label led. Bear Stearn's and Lehman's respective collapse came as they built up positions in subordinated bonds they couldn't move and certainly weren't financing them for servicers. Banks couldn't securitize as investors went on strike. Both Freddie and Fannie tried creative fixes like simply wrapping subprime, but that proved very costly with hindsight.
4) There was support for the agencies to step in. To quote from a mid-2007 Credit Suisse research piece: "With a dysfunctional non-agency securitization market, all eyes turned towards the GSEs to provide some relief. Lenders are tightening offerings that they cannot sell to the agencies and raising rates in a dramatic fashion on products including even jumbo mortgage loans that are offered to creditworthy borrowers.
To maintain liquidity in the mortgage market, lawmakers like Senator Dodd, are requesting OFHEO, the regulator of Fannie Mae and Freddie Mac, to increase the limit on the amount of mortgages and related MBS the two GSEs can buy.
On a related note, on July 30, Freddie Mac issued a $105.6 million subprime deal, Structured Pass-Through Certificates (SPCs) Series T-074, backed by loans originated by Wells Fargo. This deal is actually a wrapped Wells Fargo Home Mortgage ABS 2007-M09 deal, to which Freddie Mac provide guarantees on four Class A Certificates. It is part of Freddie Mac's $20 billion commitment to Congress to provide liquidity to the subprime market."
I'm not sure I'm arguing against what you've written; rather, I hope to give some context. As a structured product salesman at the time, I witnessed first hand both the excess and the near disappearance of credit and its affect. As one of my previous libertarian traders told me at the time "Thank God for the agencies".
1) This is a description of excess inventory from a demand shock, not a supply glut.
2) I agree that increasingly homes were purchased by investors rather than owner-occupiers. There was not an aggregate change in credit scores on the average mortgage and the Survey of Consumer Finance shows that generally from the 90s to 2007 homeownership was increasingly among college educated professionals with higher incomes.
3) Yes. Private label collapsed in 2007. Then the GSEs took up the slack, but with much tighter standards than the GSEs had used before. The rise and fall of the private markets was associated with about 8% rise and fall in home values from 2006 to 2007. Then, the tightening at the agencies was associated with highly regressive price collapse in 2008 and after that averaged more than 20%.
4) Definitely the agencies were the backstop, and I suspect if they had not tightened standards, it would not have devolved into a generation-defining crisis.
Thanks for engaging!
On this side of the GFC houses were still relatively cheap in Summerlin as late as 2016. So $250k bought you a single family home in a 9/10 school district which was probably the cheapest nice neighborhood in America. Boca Raton single family homes were last $250k in 2012. This is a big iirc but I’m pretty confident I remember this correctly.
1) yes, demand collapsed as loose credit otherwise provided by private markets disappeared, but that led to foreclosures by those in homes they couldn't afford or were hoping to flip and a subsequent "glut" in inventory. That initial credit fueled demand wasn't sustainable. Not sure if we really differ other than our emphasis.
2) I'm suspect that there was not a decrease in aggregate credit scores given subprime by 2006 made up a growing % of overall mortgages, but I don't have data to support this and it never got much higher than 10% so maybe.
Still, "Pay Option Arms" (with negative amortization), 100% LTV or seconds mortgages, "Alternative A" like no income or asset verification, all had decent FICOs but reflected large drops in underwriting standards.
3 and 4) I think you underestimate the cascading effect -- an 8% fall in housing on a very high LTV mortgage led some to walk away which reduced the shame in walking away for others which led to more REO hitting the market place which led to further drops. I have a hard time believing that wiping out ~ 22% of the lending market (best estimates are that by 2006 subprime represented ~ 10% of mortgage debt, Alt A 9%, Prime IO and Option Arms 3%) didn't have more of an impact. The Prime Jumbo private market (5% market share) also collapsed but banks (like First Republic!) and REITs ( collectively ~ 25% of market) picked up much of that slack but with crazy tight lending standards.
That leaves the agencies which undoubtedly tightened their standards. But wouldn't you? Don't they have a fiduciary responsibility to limit losses to the tax payers? The market certainly thought the risk for baring mortgage credit risk was very high. Even in 2013 when Freddie sold their first credit risk transfer deal (STACR), the AA with 1.65% credit enhancement (and 1.35% thick) sold at a spread of 2.35% and that had an average FICO of 766 and a LTV of 73!
I just don't think it was politically plausible or economically sound for them to actually loosen their credit standards to take some of the slack from the private label collapse. Certainly with hindsight they should have. They did create a largely successful HAMP program to allow for loan modifications to stem the foreclosure tied. Loan forgiveness went nowhere. All this seems like a political argument of the role of government/agencies in our housing market.
I appreciate the engagement, too. I really appreciate your work but also want to give a different perspective of the agencies role in the collapse.
I think you’re right that we generally agree on these things.
One area I would nitpick is to be clear about the difference between risky terms and loose standards. The Urban Institute has a great tracker that helps highlight this issue. The subprime boom was largely about more reckless terms. The GsEs only engaged in a small amount of that. Then, after subprime fell, there was an extreme tightening of borrower standards. The GSEs never changed their customer base during the boom and then massively cut it off during the bust.
There was no cascading. That’s one of the things I’ve discovered in my work. Every conventional narrative has everything backwards. Owners defaulted because prices collapsed. Prices collapsed because prime lending that had been available for decades was pulled away.
These aren’t just vibes. I ran the numbers and was surprised that they told a new story and so now that’s what my work is.
The intuition to accept or encourage the extreme tightening of credit was the cause of the crisis. If the average credit score in September 2008 on GSE mortgages had been the same as it was in, say, 1999, none of the worst things would have happened, and even the GSEs wouldn’t have been taken over.
I think our primary disagreement is in your statement "Prices collapsed because prime lending that had ben available for decades was pulled away" and my observation that prices collapsed as DQ's started to escalate in 2006 causing private mortgage credit to tighten and escalate in 2007. That tightening in private credit was behind the stall in house prices in the hottest markets cascading into a shut-down in private markets by early 2008 and a further increase in DQs then REO in these markets.
I don't have data by month and I'll concede your data might be better than mine, but from the FHGA.gov website in "Date on the Risk Characteristics...from 2001 through 2008..." pdf they published in September of 2010, "...in years 2004 through 2006, such loans [above 660 FICOs] account for 70 percent to 72 percent of the data [total mortgage originations], whereas in other years they account for 78% [roughly pre-2004] to 92% [by 2008]." The GSEs $ volume went up of below 660 FICO originations went up in 2007 while private label orinations collapsed. I don't think you can't refute that the housing crisis started in 2006 and gained pace in 2007, seemingly before the GSE's tightened their guidelines.
To your point, by 2008, private label origination had collapsed and GSE underwriting tightened. But the bulk of the 90+ day DQ bucket was with private label originations, not the GSEs' (e.g., 2006 vintage where GSE's market share (excluding bank retained mortgages) originations share was 67% whereas the DQ shares was only 37%). The ensuring "supply glut" on REO hitting the market as adding to my view of a cascade was also primarily stemming from the private markets (unclear to me how much the agency led HEMP vs the private label servicer led modifications (including loan forgiveness) stemmed that tide but certainly not enough. It seems clear that prior to 2007, the marginal buyer (as defined by subsequently delinquent) was primarily financed largely with private label mortgages and the evaporation of that source of financing was behind the initial collapse in demand. The tightening of standards in 2008 exasperated the problem but the crisis would have happened regardless.
The conclusion that "none of the worst things would have happened' without the GSE tightening also doesn't seem consistent with what happened overseas, especially Spain and Ireland who experienced only a collapse in private funding.
Perhaps we agree to disagree on the origins and eventual extent of the private mortgage market's role relative to the GSEs and agree that the GSEs exasperated and extended the problem by not absorbing some of that lost demand stemming from the marginal buyer.
If you're interested in these questions, I would recommend poking around under the "Research" tab. You may not be satisfied with my answers to these questions, but certainly they are central to the inquiry.
Atlanta is an example of a city where there was basically no price reaction to the rise and fall of subprime from late 2003 to mid 2007, but there was a huge price reaction that started at the beginning of 2008 when credit was tightened at the GSEs. Figure 2 here shows the price/income trends in Atlanta.
https://kevinerdmann.substack.com/p/the-great-recession-was-caused-by
I discuss Ireland and Spain here:
https://kevinerdmann.substack.com/p/housing-in-ireland-the-us-and-australia
https://kevinerdmann.substack.com/p/housing-in-ireland-the-us-and-australia-217
Egads.
Too much credit going to housing construction?
The US needs 10 million additional housing units, maybe 20 million. Imagine how much credit that would take.
Stiglitz may, indirectly, have one concern sorta-kinda right: There is so little new construction, that the average age of housing units in the US is aging by nearly one year in every year.
At some point, this will begin to take a toll. Plumbing goes bad, wiring, new roofs are needed, termites etc. Windows do not keep out rain and so on.