« Shipping News of the World: Round Two for Argentina | Main | Undocumented Debtors »

CRA and the Housing Bubble

posted by Adam Levitin

There's an interesting new paper out on the role of the Community Reinvestment Act and the housing bubble. The paper, called "Did the Community Reinvestment Act (CRA) Lead to Risky Lending?" is by Sumit Agrawal, Efraim Benmelech, Nittai Bergman, and Amit Seru (ABBS). It is a serious economic analysis, which is a major departure from much of the post-2008 grumbling about the CRA. By exploiting the differences in lending behavior within census tracts between banks that are undergoing CRA exams and those that aren't, ABBS find that undergoing a CRA exam is correlated with a rise in mortgage lending and that those loans perform more poorly than those made in the same census tract by institutions not undergoing CRA exams. In other words, the CRA encouraged more lending and as a result it resulted in less prudent lending. 

There's already some smart commentary on the paper from Mike Konczal. I would add this. There are two separate issues with the CRA. The first is whether CRA caused the bubble, and the second is whether CRA is a good idea generally. My take from ABBS is that the answer to the first question is clearly no--indeed, it seems to provide further evidence of the key role of private-label securitization--while the second question is unanswered. 

Assuming ABBS's analysis is correct, the paper shows pretty clearly that the CRA did not play a significant role in fomenting the housing bubble. While the CRA may have lead to more risky lending, what is most striking about the paper's findings are how small in magnitude the CRA's effects are. While ABBS find statistically significant impacts, the magnitudes are really small:  5% more lending in the six quarters surrounding a CRA exam and 15% higher default rate.  That's not a 15% default rate.  That means a 1.15% default rate instead of a 1% default rate or a 6.9% default rate instead of a 6% default rate. This sort of change is a drop in the bucket relative to what happened during the housing bubble.   

To this small magnitude we can run a cross-check against the "other bubble"--commercial real estate.  The CRA has little application to commercial lending; low-to-moderate income individuals aren't taking out commercial loans. Yet there was a near identical bubble in commercial real estate. The implication of the CRA's non-involvement in the commercial real estate bubble was that it probably did not play much of a role in the parallel residential real estate bubble. Instead, ABBS find that the effects of the CRA in both lending volume and defaults were greatly elevated during the 2004-2006 period, when the private-label securitization market was in full bloom. That suggests that the CRA by itself had a much more minor effect, but the ability to shift risk to MBS investors changed the effect of the CRA. The role of private-label securitization in the bubble also tracks with the commercial real estate market, where there is only private label securitization and where securitization really took off at the same time as the start of the bubble. 

Still, the CRA had its critics well before the housing bubble. These critics argued that the CRA encouraged banks to make riskier loans. ABBS would seem to provide some support for this argument. But this argument hardly resolves the debate on the CRA. Instead, it points to a cost from the CRA that must be weighed against the CRA's benefits, such as access to credit for low-to-moderate income households and decreased redlining. Regardless of how one resolves this analysis, what's clear is that the CRA wasn't driving the housing bubble.  


I was a mortgage originator for an arm of a large national bank. CRA loans were nearly impossible to get off the ground. The arithmetic of juxtaposing low income borrowers and their credit profile with markets where prices were escalating (nearly everywhere in the mid to late 2000's?). It seems to me it was a concession program that was mandated to be offered... so that lenders could keep on working the gravy train side of the business. Whether it was effective, or helped folks get into houses, was an afterthought. Good work in calling out the red herring, Adam!

Just look at the counterfactual Adam. In commercial real estate there is no equivalent to CRA and they had the same boom and bust. CRA doesn't apply to manufactured housing, but it had a huge boom/bust in the 1990s. The primary reason is the same, excess supply of credit via ABS drove underwriting to the ground. If you look at MH lending it's stunning how fast after the introduction of ABS that underwriting went down hill and the boom began. Haven't read the paper, but this isn't hard stuff to analyze and you don't need fancy models to come to a reasonable conclusion.

One statement in this post is a little misleading. Most of the pre-bubble criticism of the CRA was not that it would cause wild and crazy lending that would lead to disaster, but that it could be evaded and had little effect. See http://www.cato.org/sites/cato.org/files/serials/files/regulation/2000/10/gunther.pdf Cato's take on the CRA in 2000. The increased lending in low income neighborhoods, they said, was all private sector innovation and had nothing to do with CRA. Around that time I was a discussant on a paper on CRA effects written by someone connected to a large lender, and they spoke of minor effects of the CRA and how easy it was to game compliance. Complaints about the CRA have done a 180 from pre-bubble to post-bubble.

It would seem an over-simplification to put the blame for the lending crisis on the shoulders of one organization. In my opinion, it was caused by a domino effect due to the destabilization of several economic factors - with flawed lending policy being just one of the factors.

The real concern here should be not on who to pass the blame but on creating stable lending practices that protects the economy and the interests of both lenders and borrowers.

Simon Campbell - http://www.bankforeclosuressale.com

I'm a mortgage originator and CRA wasn't a driver of the bubble, at least not directly. Many of the FDIC insured lenders offered bonuses to brokers for originating loans in CRA tracts, but it didn't really drive originations imho. The bubble was driven by speculation and fraud. When the bubble first imploded, it was mostly what I call specuvestors who were hoping to flip condos and houses using fraudulent stated/stated products/options arms, etc. Unfortunately, they brought down the marginally qualified buyer with them as well.

It really isn't rocket science if you get out of the ivory tower and board rooms and talk to the guys on the front lines about what was really going on.

Housing Bubble / Financial Crisis Timeline:

Pre-2001 Reality: CRA, mortgage interest deduction, government guarantees and other support for the housing sector has long been in place. Both major political parties desire to promote home ownership. These generally support housing, and create a moral hazard, but had not prior to the 2000s produced an “unsustainable boom,” though they were in a position to amplify one should it begin.

Developing countries’ and their citizens’ savings are invested in long-term Treasurys, bringing down long-term rates including traditional mortgage rates. This also supports housing values from the mid-1990s onward, even after the bursting of the dot.com bubble.

Post 9/11, Americans culturally become focused on their own homes rather than travel or external recreation – improvements such as granite countertops, additions, home entertainment centers, become sought-after.

Lending standards typically include credit (primarily, repayment history) check, income and debt service coverage check, and loan-to-value (loan amount versus value of the home being financed). This has not changed in years.

The way you go about buying a home is to first become pre-qualified for a mortgage up to $X. This is based primarily on your ability to cover the loan payments. The lower the rates, the higher your pre-qualification, and vice versa. You’ll be approved for that amount subject to loan-to-value limit on the home you are purchasing – but generally the purchase price is the deemed value. This assumes that nobody wants to overpay and assumes a slow and steady increase in values with population growth.

Low delinquency and default rates on mortgages have been the norm for decades. For decades, historical delinquency and default rates were effective ways to predict future delinquency and default rates . Most people fix for a long period of time, and when rates spike, fewer people take out mortgages and not many mortgages “re-set” at the higher rates. Short rates have never plummeted and then sprung back up, leaving a relatively flat yield curve, thus there has been no incentive to finance with short-term money. ARMs and HELOCs exist but are not as popular, as the curve has been relatively flat and there has not been a period characterized by a massive, sudden creation of new home equity out of thin air.

Because of the historically low delinquency and default rates, and government guarantees, and because a pool of loans diversifies risk, CDOs look like a rational way for banks to go a little further out on the risk-return curve to get a little more yield. The pressure to do this increases as bond yields fall, due to the increase in foreign savings.

Refinance-with-cash-out and HELOCs are approved based on the above factors and an appraisal, which generally is based on comparable sales. The same comps can be used to appraise an unlimited number of homes. If you have a 100-home development built in the 1950s in a stable neighborhood with limited turnover, you might have three homes that were bought in the 1990s and sold between 2000-2002. Because to-date, values have not changed for the better in any rapid fashion, this has not yet been a recipe for disaster.

Banks finance builders with interest-carry at floating rates – thus if short-term rates were to plummet, the same rates that fuel the repayment source also lower the debt service. But this has not happened yet. Subdivision-development loans to most private developers of 10-100 unit sub-divisions, are based on a project-finance model, with cash draw to build 1-2 speculative homes and 1-2 pre-sold homes at a time. They are reviewed annually. Again, in a stable market which has never seen a massive temporary influx of credit, this is not a bad model, the only model possible unless you want all builders to also maintain apartment complexes with income streams.

Mortgage closing attorneys, home sales closing attorneys, real estate agents and mortgage brokers are paid per transaction, thus have no incentive other than possibly reputation to think of the transaction beyond immediate terms. But-for bubble periods, this is not a significant risk – the risk of rapid changes in value is lower than the risk that the home will turn out to have termites, etc…

The internet makes all markets including real estate much faster-paced – you can take a virtual tour of a home, search listings in a neighborhood, etc…

In Sum, the government provides important supports to the housing market and housing and credit systems assume some small amount of monetary accommodation but are not built to deal with dramatic shifts in monetary policy. The market for real estate and mortgages has evolved, but not in a necessarily bad fashion, as long as values and rates remain stable. “The system” has worked very well for decades and are built for the environment that has slowly evolved over those decades. The secondary market for mortgages has grown, and is seen as relatively safe, precisely because the system has worked so well. There is unforeseen exposure to volatility in short-term rates, particularly on the down-side – but short-term rates are set by fiat and have not previously been volatile, at least not on the downside.

2001: In response to the collapse of the dot.com bubble and the 9/11 shock, the Fed cuts the discount rate (and as a result, short term Treasury yields) to record lows, where they stay for almost three years. Some economists cheer this on – whether tongue in cheek or not – precisely as a way to inject cash into the economy through increasing home values. There is no other specific stated reason for the policy other than generally to “support” the economy.

2001-2003: More people buy new homes with ARMs. The payments are incredibly low. The low pricing results in lower debt service for the same debt – meaning the borrowers can qualify for higher debt levels. The expanded credit soon translates into higher amounts paid for the same homes.

2002-2004: People notice the values going up. Would-be renters start to scramble to buy “before we’re priced out of our dream of one day owning a home.” Would-be laborers in non-housing-related fields enter construction, and over the next few years make six-figure incomes, not through the value-add of the houses they build but through the appreciation in value of the land between their purchase of the lot and their sale of the finished home. Would-be professionals in non-housing-related fields become real estate closing attorneys, agents, appraisers.

2002-2004: A refinancing boom begins. The 3-4 homes sold in the subdivision for $100M-$200M more than their purchase price 4-5 years earlier translates into a $100M-$200M increase in appraised values for the other 96-97 homes in the subdivision. People refinance with ARMs, with cash out, increasing their mortgages by $150M, and the payments actually decrease. People take on HELOCs – second position, floating rate mortgages. The money is then spent, temporarily boosting the local and broader retail economy. Banks and other institutions invest in huge mortgage-lending infrastructure to meet the demand for housing credit.

2003-2004: Values really start to shoot up. Some banks, seeing a decrease in loan-to-value, believe that this will enable them to take on higher risk in other factors of the transaction. Lenders understand that rates will one day increase, but the focus in underwriting is on the transaction, and assumes the rest of the environment will remain. “This borrower might have trouble making the payments if rates go up by 300 bps before the ARM re-prices, but by that time the home will be worth another $150M and he can easily sell or refinance.” Mortgage lenders also consider that they will be selling the loan long before it re-prices. Nobody is thinking that the macro picture is just the aggregate of all of the similar transactions.

2002-2005: Retirees who have some savings built up see second homes in the Sun Belt as a win-win - - buy that second home for recreation purposes but also hold it as an investment, thus not having to choose between spending and investment. A building boom occurs in Florida, Nevada and California. People who get in early enough finance a new home in 2002 and then refi or take out a HELOC in 2004.

McMansions become the norm (and in the exurbs and Midwest, ranches), because people can afford them. Home Depot, Tractor Supply Company, Lowe’s, are the chief retailers, because people have to maintain their McMansions.

2001-2004: With rates continually low, delinquency and default rates stay low. If you lower the bar, it makes it easier to clear the bar, thus fewer people will trip on the bar. Institutions continue to use historical delinquency and default rates to predict future delinquency and default rates because their predictive capacity appears to have, if anything, increased. There has never been a period in which millions of people could get 3/1 ARMS in the low 3s and had them re-price in the high 6s, because the rates had never gone that low – so the inability of predictive models to pick this up has never previously been a problem.

2001-2005: With rates remaining so low, the savings rate goes negative. 0% credit cards proliferate. Many people take them on and max them out. There is no incentive to save and every incentive to borrow – if you can get what you want now, with negative real financing cost, why would you save?

2004: The Fed starts to slowly raise rates. There is even more of a frenzy as the former renters who were already thinking “I’d better buy before I am priced out” also think “I’d better buy now while I can still afford the payments, before the rates go up.” Few take the analysis to the next step: “if the rates go up, the values will stop going up.” The “mania” is a follow-on – it is a characteristic of a bubble in the mid-late throes, but it is not the cause.

2005: The short rates are not low enough for HELOCs and ARMs to function as conduits to build up the values and inject new-found cash into homeowners’ pockets. Values peak. Refinancing and HELOC volume slows.

2005-2006: Mortgage lenders, principally those that do not hold on to the mortgages, have spent the last 3 years building out their infrastructure. They need to maintain volume. To do so, with values no longer rising, they need to lower credit standards. This is when the proliferation of no-doc loans, 0% down loans, etc…, occurs.

2007: Those subdivision builders with floating rate loans see their payments going up, and see a slack in demand for the homes they are building. The loans come up for review and the banks stop financing new speculative homes, and pressure the builders to sell faster, even at a discount, to stop the bleeding (interest accruing at the new higher rate). Values start to come down. When a new house sells for $50M less than the same model did a year earlier, this has an effect on comps/appraisals for HELOCs and ARMs – as well as people attempting to refinance their HELOCs and ARMs at fixed rates – that is opposite of the effect of the rising prices 4-5 years earlier.

2001-2007: Prices of non-housing margined assets rise and then plummet as well, in inverse proportion to interest rates. When rates are again cut, commodity prices rise again. Such assets include oil and gold. Oil-rich states such as Texas and Alaska do not suffer the same housing bubble – the values go up but do not come down, as the oil-related revenue continues to fuel a local boom. Gasoline and heating oil prices shoot up, eating up more of the budget of people who bought the McMansions.

2007: HELOCs and ARMs originally priced in the 3s re-price in the 6s, 7s and even 8s, and the payments increase by 50-60%. They cannot all continue to make the payments, particularly with energy and food prices so high. Delinquency and default rates start to spike up. Homeowners with short-term financing are no longer maintaining their consumption levels, and they are not being replaced with new cash-out consumption, and so local and retail spending falls, affecting the economy as a whole. People with non-housing jobs and fixed-rate mortgages lose their jobs and fall behind on the payments.

Most of the second-home mortgages are non-recourse. Because of rising fuel prices, it also costs a lot more money to fly to Florida, Nevada and California. If you’re 70 and under-water on a non-recourse mortgage on a vacation home that costs more to visit, your best bet is to mail the keys to the bank, and that is precisely what happens.

2008: With the sudden spike in delinquency and default rates, the CDO market collapses. The construction workers, the mortgage lenders, the real estate brokers, are all laid off. Unemployment also increases among the ranks of those who marketed, distributed and sold the items that these people used to buy. The bubble in margined-commodities temporarily crashes as well, at the same time, but recovers when the Fed cuts the rates again, thus preventing price deflation which could have provided some relief to the newly out-of-work.

2008-Present: Fed’s response is to cut rates even lower and flood the economy with money. Federal government’s response is bailouts and “stimulus” spending. Fed ends up buying 60% of the Treasurys at auction in 2011. China, Japan, banks and other institutions buy up much of the remainder. Banks got burned with housing and won’t lend there, but commodities are right back up. Prices are not coming down despite the recessionary conditions, because of the new money. Fuel and food prices do not factor into CPI, thus reported inflation is lower than people’s actual experience. Large businesses have locked-in and used the low rates to de-leverage. They deposit the cash in banks, which is mistaken for “cash on the sidelines” – banks use it to buy Treasurys, since the federal regulators do not make them reserve against sovereign debt but make them reserve up to 11% against private debt. With meager individual savings, a result of the previously-discussed prior decade of spend-spend-spend, there is little left over to invest. The cycle generally starts with emerging firms raising equity from these savings, and cyclical firms getting bank loans to finance equipment production to meet demand from these emerging firms – demand was pulled forward in the 2000s at the expense of savings, and so now this cannot happen. 0% rates do not help. This is the classic Hayekian Triangle.

Conclusion: The but-for cause, the first domino, the factor without which most of the other factors would not have occurred and NONE of them would have resulted in a boom-crash cycle, was the Fed’s rate policy from 2001-2005.

Alternative Theories:

“Mania” – as discussed above, while this was part of the vicious cycle, it was a follow-on event, and would not have happened but for the initial boom, which resulted from the rates.

“Exotic mortgage products” – what was exotic about them? The exposure to volatile short-term rates, which previously had not been volatile.

“Risk” – risk of what? Risk that the ARMs and HELOCs originated at low rates would re-price at high rates. Risk that the rise in values that resulted from this influx of credit would stop when the influx of credit ended. No-doc and low-doc? Yes, bad risk policies – but again this was a follow-on event – but-for the initial boom, there would have been no massive increase in overhead in the mortgage and housing industry, and the decision “do we lay these people off or do we get more creative about how to maintain volume” would not exist – there would be no infrastructure and no volume to maintain, and these people would have done what they have to do now, which is find work outside of housing and mortgages.

And again, most of the damage was done not via this late-cycle detour from historical lending practices but via the fact that the historical lending practices did not account for such a volatile shift in rates. Predicting default rates for a portfolio based on historical rates works as long as the environment does not shift; predicting default rates for an individual loan based on debt service coverage and loan-to-value is rational assuming stable rates thus stable loan payments.

“Decline in aggregate demand / not everyone laid off was in construction or mortgage lending” – as noted above, people stopped saving and spent beyond their real means, because the immediate, credit-infused means were suddenly available. This was by-definition unsustainable, which meant that the jobs temporarily created by virtue of that credit-infused demand had to go away when the credit went away.

“Savings glut” – the increase in foreign savings did not reverse itself; the low short-term rates did, and that is when the housing market shifted and the bubble burst.

“CRA” – amplified the bubble and directed some of the new credit into places like Detroit, but cannot explain the Sun Belt or anywhere else. Bad policies but not the policies that caused the bubble.

“Big banks marketing high-risk CDOs as low-risk CDOs” – bad actions, primarily in 2007-2008. When the bubble was bursting, those who saw it first took advantage. That wasn’t the cause of the bubble. You really can’t put the CDOs sold in 2004-2006 on which a higher than expected proportion of the underlying mortgages would go into default three years later into the same category as the last handful of CDOs that they got out the door just before it all fell apart. The actions in the latter category may have been nefarious but the actions in the former were the result of the rapid influx and then pullback of credit. The malpractices that people like Elizabeth Warren complain about are genuinely bad acts that should be punished, but they are not, and should not be confused as being, the cause of the housing bubble and bust, or the prolonged recession that has ensued. The smart money might ride the bubble and jump off in time, sometimes illegally and/or immorally – but it doesn’t create the bubble.

This was a credit-bubble, plain and simple.

Other factors amplified it, and some bad actors rode the wave and got off before it crashed, but the bubble-bust itself was a classic example of the ABCT “unsustainable boom.”

Inflated money and credit translates, in the long run, into higher consumer prices, unless it is sucked right back up before it hits those prices. Delay in flowing into prices occurs because it flows into margined assets. Presently this would be housing and commodities. In prior generations it was business investment. There is a chicken-egg question of whether to make these assets non-margined, or less-margined, or simply maintain a stable money and credit supply, or allow the market to effect a stable money and credit supply, by limiting credit to the actual supply of loanable funds, i.e., savings. I don't really care that much - I just want people to get the analysis right. It's just a puzzle to me - just like figuring out the biblical symbolism in David Lynch movies.

PS - I'm a commercial credit manager for a regional bank, saw this happen first-hand in developer-finance, saw this happen second-hand via the mortgage-finance groups, I know the lending standards. There were - primarily late in the game - "loans made to people who couldn't afford them" from day one. But for each dollar so lent, there were many dollars lent to a combination of people who (1) would no longer be able to afford them at the higher rates, (2) would no longer have the viable exit of selling into a market that was rising because of the flow of credit, and (3) would lose a job in construction or another housing-related field or (4) would lose a job that temporarily existed because of the increased purchases made with borrowed money. It was the rates. The but-for cause was the rates. Without the rates, but with everything else, you might have had a few wrinkles, but not the 21st century version of tulip bulbs. With the rates, and without many of the other factors, you still would have had a bubble.

The math:

A couple applies to prequalify for a mortgage in 2001. They have gross income of $67K. A 3 year ARM, 30 year amortization would be priced at 5.75%. So gross income of 3.5 times the debt service gives you a pre-qualification for $275K with debt service of $1,605/month.

By 2003 the rate has come down to 3.75%. The same couple now qualifies for a $350K mortgage based on the same income to debt service ratio. Payments are $1,621/month. Free cash flow to use for mortgage payments is about $25K, about 1.3x the annual debt service of $19,451.

Notice that underwriting standards have not changed.

In 2006 the loan re-prices at 6.50%. The balance is $335K and the monthly payment is now $2,122. The couple might have received a few raises but gas and food prices, and property taxes, have eaten that up. If free cash flow is still about $25K, the couple can no longer afford the payments.

Patrick T--I don't think the interest rate story quite holds up. On its own, it makes a lot of sense, but it doesn't work with the timing of the bubble. Rates were crazy low after the bursting of the Internet bubble in 2001, and there was an orgy of refinancing in 2001-2003. Pretty much everyone with more than a few payments left on their mortgage refinanced. But these were all prime, fixed-rate loans. Things started to change in 2004, when rates were still low, but rates started to go up in 2005 and continued to do so even as the bubble grew into 2007. On an absolute level, rates were still low, but they were moving upward even as the bubble grew, which makes it hard to point to mortgage interest rates as the key cause of the bubble.

To be honest, this conclusion is due to a whirlwind of deception and fraud; in addition to finding the difficulty to make money from communities...

The comments to this entry are closed.


Current Guests

Follow Us On Twitter

Like Us on Facebook

  • Like Us on Facebook

    By "Liking" us on Facebook, you will receive excerpts of our posts in your Facebook news feed. (If you change your mind, you can undo it later.) Note that this is different than "Liking" our Facebook page, although a "Like" in either place will get you Credit Slips post on your Facebook news feed.

News Feed



  • As a public service, the University of Illinois College of Law operates Bankr-L, an e-mail list on which bankruptcy professionals can exchange information. Bankr-L is administered by one of the Credit Slips bloggers, Professor Robert M. Lawless of the University of Illinois. Although Bankr-L is a free service, membership is limited only to persons with a professional connection to the bankruptcy field (e.g., lawyer, accountant, academic, judge). To request a subscription on Bankr-L, click here to visit the page for the list and then click on the link for "Subscribe." After completing the information there, please also send an e-mail to Professor Lawless (rlawless@illinois.edu) with a short description of your professional connection to bankruptcy. A link to a URL with a professional bio or other identifying information would be great.