The Volokh Conspiracy - Bankruptcy Mortgage Modification Update:

UPDATE 8:31 p.m.: I just received an email that this version on Thomas, although the most recent that is there, is the version of legislation as it was introduced yesterday and doesn't contain action on amendments. But from what I can tell reading news stories, the provisions I focus on below (the "void against public policy" and clawback provisions) are still in the legislation.

One of my concerns about allowing modification of mortgages in bankruptcy was that it might refreeze credit markets because of provisions in mortgage-backed securities that allocate bankruptcy losses above a certain threshold pro rata to all tranches of MBS's.

Yesterday the House passed a revised version of the legislation that contains this remarkable clause (section 124):

SEC. 124. UNENFORCEABILITY OF CERTAIN PROVISION AS BEING CONTRARY TO PUBLIC POLICY.

No provision in any investment contract between a servicer and a securitization vehicle or investor in effect as of the date of enactment of this Act that requires excess bankruptcy losses that exceed a certain dollar amount on residential mortgages to be borne by classes of certificates on a pro rata basis that refers to types of bankruptcy losses that could not have been incurred under the law in effect at the time such contract was entered into shall be enforceable, as such provision shall be contrary to public policy. Notwithstanding this section, such reference to types of bankruptcy losses that could have been incurred under the law in effect at the time such contract was entered into shall be enforceable.

So to deal with the problem, Congress will simply declare those loss-allocation provisions in those contracts to be simply "unenforceable... as being contrary to public policy." Well, I have to confess that provision really just leaves me speechless. Has anyone ever seen anything like this--just a blanket declaration that a particular contract term is simply void against public policy, when it is a freely-bargained for, risk-allocation term? Does it solve the potential credit freeze problem? I guess it depends on how those losses will be allocated when there is no longer any contract term by which to allocate them. Especially given that the whole point of that contract term was to be the catch-all for losses that are not otherwise allocated under the MBS contracts. It is hard to see how this provision alone solves the problem of the uncertainty in valuing these securities if we don't know how the losses will otherwise be allocated.

I had also read reports that indicated that there were new protections in the legislation for recoupment by lenders when a borrower sells a home for a profit after stripping it down in bankruptcy. If so, I can't find it. All that I see is the same restrictions, namely that if the debtor sells the house while in bankruptcy there is a sliding scale starting at 80% recoupment in the first year down to 20% in the fourth year, but I don't see anything post-discharge, unless I missed something in reading through the legislation.

While on this topic should also note that I should correct the record on one minor point in my WSJ column on which I made a mistake. It relates to the precise effect of modification eliminating the ability to modify car loans in the 2005 bankruptcy amendments. In the article I referred to a recent NY Fed Staff report that looked at the elimination of the ability to modify car loans ("cramdown") in the 2005 bankruptcy reform legislation. The authors of the study found that eliminating the cramdown power substantially reduced interest rates on auto loans. But this finding was not the centerpiece of their paper and so they didn't report the actual estimate of the reduction in interest rates as a result. They were interested in the effect of exemption laws on car loan interest rates, not this intermediate step. I mis-read the paper as reporting a 265 basis point drop in the spread on car loans as a result of eliminating the cramdown power (if you read the paper, or at least the draft as it was at that time, you will see that it is hard to figure out what the authors are saying). That was actually the overall mean spread on car loans above the cost of funds, not the reduction as a result of eliminating the cramdown power. They stated that it had an effect but didn't report the actual figure for the reduction in the spread so it turns out that you can't tell just by looking at the paper itself how large the effect is. The WSJ asked me for a specific figure that day right before deadline and I supplied them this figure, which was a mistake.

I asked the authors if they would re-run the regressions and estimate the actual value of the reduction in the spread as a result of eliminating the cramdown, which they graciously agreed to do. I understand that they will be posting an updated version of the paper that includes this result. But to summarize, they find that the impact of eliminating cramdown was a reduction in interest rates of 56 or 46 basis points depending on the regression treatment and that this result is highly statistically significant (at .01 level). So they estimate that the reduction of the interest rate spread was about 50 basis points off of the 265 average spread, or about a 19 percent drop in the spread as a result (and an additional 12 points in states with unlimited exemptions (which is marginally significant). Obviously my mistake was inadvertent because it was irrelevant to my argument whether the effect was big and statistically significant or really big and statistically significant.

One thing that I should add here is that the effect of bankruptcy reform is not just 15 basis points and marginally significant, as some readers apparently have concluded. Ironically, I have even been criticized because I did not accept this finding as the conclusion of the study. The reason I did not report that finding is because it has nothing to do with the issue I was looking at. That estimation in the paper relates to a completely different question--it is a difference-in-difference regression that looks at whether state exemption laws have an independent effect on top of the general effect of eliminating the car loan cramdown. It was obvious to me at the time that this was not and could not be the correct figure. I knew that I was looking for a general baseline figure of some sort, and the authors only mention one baseline figure and read in context I thought it was the correct one and it was obvious that the difference-in-difference results was not the correct one. But in fact, there was another baseline number that they did not report. I won't belabor the point except to note that the findings in the difference-in-difference regressions simply are not the correct estimates of the impact of BAPCPA on car loan interest rates.

So the authors found that effect of BAPCPA on car loan spreads was substantial in size and highly statistically significant although I supplied the incorrect point estimate. I should note that the purpose of citing the study in the article was to simply illustrate the point that if you increase the risk of lending (such as by allowing modification of loans) this will lead to higher interest rates and costs for borrowers, which was shown with statistical significance. I was not to try to make a prediction about the exact size of the interest-rate increase that will result.

In fact, whether the effect for cramdown of home mortgages will be bigger or smaller than for car loans will depend on several factors. First is the expected risk that property values will fall in the future, which is different for car loans because cars are depreciating assets so they will almost certainly decline in value. Second, the size of the decline--for car loans it is probably a large number of relatively small decreases in value while for houses the value declines are likely to be much larger. The size of the decline is also relevant to how much value a lender might lose relative to losses from a foreclosure. Third is the risk that bankruptcy judges will tend to set the interest rates on mortgages in bankruptcy at a below-market rate, so that the lenders will get less from the modified mortgage than they would by reinvesting the money. For reasons discussed in my op-ed, this is likely to be a real risk.

Fourth, and perhaps most importantly, is what will be the effect of permitting mortgage modification on leading to increased bankruptcy filings. Cramdown of car loans were different from mortgages in several ways. Car loans were shorter-term loans and would be paid off during the Chapter 13 case. As a result, cramdown of a car loan reduced the payments owed to the secured lender but those savings were instead just made available to unsecured creditors. So there was no incentive for borrowers to file bankruptcy just to cramdown car loans because the debtor would not be able to capture the surplus. With mortgages, however, it is likely that most homes that are underwater today will rise in value in the future. And the cramdown will apply for the entire duration of the loan--up to 30 years. So it extends beyond the time of the chapter 13 plan. This means that an underwater borrower today who believes his home will rise in value in the future will have an incentive to file chapter 13 and capture future appreciation--an incentive that is absent in the context of car loans.

This means that permitting cramdown of home mortgages will almost certainly have an incentive effect of increasing bankruptcy filings at the margin, unlike for car loans. Moreover, there are certainly at least some people out there who would not be willing to permit foreclosure on their homes, but would be willing to file bankruptcy if it meant that they can strip-down their mortgages and capture future appreciation. So while there will be some substitution of foreclosure for bankruptcy, there will be a new pool of people who would be unwilling to permit foreclosure, but would be willing to file bankruptcy.

We don't know how big this group might be. Supporters of the mortgage modification proposal argue that this group is likely to be small because of the difficulties of bankruptcy and the negative impact on their credit scores and so that those who will file bankruptcy will do so only as a last resort. That seems unduly optimistic to me. In fact, a year or two ago there was a general belief that few people would walk away from their homes just because they were underwater, because of the hardship of foreclosure and the negative impact on their credit score. But I think it is abundantly clear at this point that there are a lot of people who are doing exactly that. And I don't see why we'd expect the use of bankruptcy to be any different. In fact, in his recent paper, Eric reports that 48% of homeowners reported in a survey that they would consider walking away from their homes if underwater by over $100,000. Moreover, given the huge surge of filings that are likely to result from allowing mortgage modification, bankruptcy judges are going to by necessity given only passing scrutiny to these cases, so the opportunities for fraud and abuse are likely to increase substantially as well.

So what this means is that we don't know how many people will choose to file chapter 13 and write-down their mortgage who otherwise would have not defaulted and instead paid their mortgages. But if the number of homeowners is nontrivial--and the events of the past two years lead me to believe that it probably is nontrivial--this will significantly increase bankruptcy filings. More importantly, it will increase bankruptcy filings among people who otherwise would not have been in foreclosure but instead would have paid their contracted loans. So for this group of borrowers it is a pure loss to the lenders because they are not saving foreclosure losses.

Update:

Calvin Massey provides an historical analogue to this provision--the Gold Clause cases. An excellent example in many ways. Calvin concludes that the provision likely is constitutional. Whether it is a good idea is a different question.

Update:

See update at top.

	[Todd Zywicki, March 6, 2009 at 4:32pm] Trackbacks Bankruptcy Mortgage Modification Update: UPDATE 8:31 p.m.: I just received an email that this version on Thomas, although the most recent that is there, is the version of legislation as it was introduced yesterday and doesn't contain action on amendments. But from what I can tell reading news stories, the provisions I focus on below (the "void against public policy" and clawback provisions) are still in the legislation. One of my concerns about allowing modification of mortgages in bankruptcy was that it might refreeze credit markets because of provisions in mortgage-backed securities that allocate bankruptcy losses above a certain threshold pro rata to all tranches of MBS's. Yesterday the House passed a revised version of the legislation that contains this remarkable clause (section 124): SEC. 124. UNENFORCEABILITY OF CERTAIN PROVISION AS BEING CONTRARY TO PUBLIC POLICY. No provision in any investment contract between a servicer and a securitization vehicle or investor in effect as of the date of enactment of this Act that requires excess bankruptcy losses that exceed a certain dollar amount on residential mortgages to be borne by classes of certificates on a pro rata basis that refers to types of bankruptcy losses that could not have been incurred under the law in effect at the time such contract was entered into shall be enforceable, as such provision shall be contrary to public policy. Notwithstanding this section, such reference to types of bankruptcy losses that could have been incurred under the law in effect at the time such contract was entered into shall be enforceable. So to deal with the problem, Congress will simply declare those loss-allocation provisions in those contracts to be simply "unenforceable... as being contrary to public policy." Well, I have to confess that provision really just leaves me speechless. Has anyone ever seen anything like this--just a blanket declaration that a particular contract term is simply void against public policy, when it is a freely-bargained for, risk-allocation term? Does it solve the potential credit freeze problem? I guess it depends on how those losses will be allocated when there is no longer any contract term by which to allocate them. Especially given that the whole point of that contract term was to be the catch-all for losses that are not otherwise allocated under the MBS contracts. It is hard to see how this provision alone solves the problem of the uncertainty in valuing these securities if we don't know how the losses will otherwise be allocated. I had also read reports that indicated that there were new protections in the legislation for recoupment by lenders when a borrower sells a home for a profit after stripping it down in bankruptcy. If so, I can't find it. All that I see is the same restrictions, namely that if the debtor sells the house while in bankruptcy there is a sliding scale starting at 80% recoupment in the first year down to 20% in the fourth year, but I don't see anything post-discharge, unless I missed something in reading through the legislation. While on this topic should also note that I should correct the record on one minor point in my WSJ column on which I made a mistake. It relates to the precise effect of modification eliminating the ability to modify car loans in the 2005 bankruptcy amendments. In the article I referred to a recent NY Fed Staff report that looked at the elimination of the ability to modify car loans ("cramdown") in the 2005 bankruptcy reform legislation. The authors of the study found that eliminating the cramdown power substantially reduced interest rates on auto loans. But this finding was not the centerpiece of their paper and so they didn't report the actual estimate of the reduction in interest rates as a result. They were interested in the effect of exemption laws on car loan interest rates, not this intermediate step. I mis-read the paper as reporting a 265 basis point drop in the spread on car loans as a result of eliminating the cramdown power (if you read the paper, or at least the draft as it was at that time, you will see that it is hard to figure out what the authors are saying). That was actually the overall mean spread on car loans above the cost of funds, not the reduction as a result of eliminating the cramdown power. They stated that it had an effect but didn't report the actual figure for the reduction in the spread so it turns out that you can't tell just by looking at the paper itself how large the effect is. The WSJ asked me for a specific figure that day right before deadline and I supplied them this figure, which was a mistake. I asked the authors if they would re-run the regressions and estimate the actual value of the reduction in the spread as a result of eliminating the cramdown, which they graciously agreed to do. I understand that they will be posting an updated version of the paper that includes this result. But to summarize, they find that the impact of eliminating cramdown was a reduction in interest rates of 56 or 46 basis points depending on the regression treatment and that this result is highly statistically significant (at .01 level). So they estimate that the reduction of the interest rate spread was about 50 basis points off of the 265 average spread, or about a 19 percent drop in the spread as a result (and an additional 12 points in states with unlimited exemptions (which is marginally significant). Obviously my mistake was inadvertent because it was irrelevant to my argument whether the effect was big and statistically significant or really big and statistically significant. One thing that I should add here is that the effect of bankruptcy reform is not just 15 basis points and marginally significant, as some readers apparently have concluded. Ironically, I have even been criticized because I did not accept this finding as the conclusion of the study. The reason I did not report that finding is because it has nothing to do with the issue I was looking at. That estimation in the paper relates to a completely different question--it is a difference-in-difference regression that looks at whether state exemption laws have an independent effect on top of the general effect of eliminating the car loan cramdown. It was obvious to me at the time that this was not and could not be the correct figure. I knew that I was looking for a general baseline figure of some sort, and the authors only mention one baseline figure and read in context I thought it was the correct one and it was obvious that the difference-in-difference results was not the correct one. But in fact, there was another baseline number that they did not report. I won't belabor the point except to note that the findings in the difference-in-difference regressions simply are not the correct estimates of the impact of BAPCPA on car loan interest rates. So the authors found that effect of BAPCPA on car loan spreads was substantial in size and highly statistically significant although I supplied the incorrect point estimate. I should note that the purpose of citing the study in the article was to simply illustrate the point that if you increase the risk of lending (such as by allowing modification of loans) this will lead to higher interest rates and costs for borrowers, which was shown with statistical significance. I was not to try to make a prediction about the exact size of the interest-rate increase that will result. In fact, whether the effect for cramdown of home mortgages will be bigger or smaller than for car loans will depend on several factors. First is the expected risk that property values will fall in the future, which is different for car loans because cars are depreciating assets so they will almost certainly decline in value. Second, the size of the decline--for car loans it is probably a large number of relatively small decreases in value while for houses the value declines are likely to be much larger. The size of the decline is also relevant to how much value a lender might lose relative to losses from a foreclosure. Third is the risk that bankruptcy judges will tend to set the interest rates on mortgages in bankruptcy at a below-market rate, so that the lenders will get less from the modified mortgage than they would by reinvesting the money. For reasons discussed in my op-ed, this is likely to be a real risk. Fourth, and perhaps most importantly, is what will be the effect of permitting mortgage modification on leading to increased bankruptcy filings. Cramdown of car loans were different from mortgages in several ways. Car loans were shorter-term loans and would be paid off during the Chapter 13 case. As a result, cramdown of a car loan reduced the payments owed to the secured lender but those savings were instead just made available to unsecured creditors. So there was no incentive for borrowers to file bankruptcy just to cramdown car loans because the debtor would not be able to capture the surplus. With mortgages, however, it is likely that most homes that are underwater today will rise in value in the future. And the cramdown will apply for the entire duration of the loan--up to 30 years. So it extends beyond the time of the chapter 13 plan. This means that an underwater borrower today who believes his home will rise in value in the future will have an incentive to file chapter 13 and capture future appreciation--an incentive that is absent in the context of car loans. This means that permitting cramdown of home mortgages will almost certainly have an incentive effect of increasing bankruptcy filings at the margin, unlike for car loans. Moreover, there are certainly at least some people out there who would not be willing to permit foreclosure on their homes, but would be willing to file bankruptcy if it meant that they can strip-down their mortgages and capture future appreciation. So while there will be some substitution of foreclosure for bankruptcy, there will be a new pool of people who would be unwilling to permit foreclosure, but would be willing to file bankruptcy. We don't know how big this group might be. Supporters of the mortgage modification proposal argue that this group is likely to be small because of the difficulties of bankruptcy and the negative impact on their credit scores and so that those who will file bankruptcy will do so only as a last resort. That seems unduly optimistic to me. In fact, a year or two ago there was a general belief that few people would walk away from their homes just because they were underwater, because of the hardship of foreclosure and the negative impact on their credit score. But I think it is abundantly clear at this point that there are a lot of people who are doing exactly that. And I don't see why we'd expect the use of bankruptcy to be any different. In fact, in his recent paper, Eric reports that 48% of homeowners reported in a survey that they would consider walking away from their homes if underwater by over $100,000. Moreover, given the huge surge of filings that are likely to result from allowing mortgage modification, bankruptcy judges are going to by necessity given only passing scrutiny to these cases, so the opportunities for fraud and abuse are likely to increase substantially as well. So what this means is that we don't know how many people will choose to file chapter 13 and write-down their mortgage who otherwise would have not defaulted and instead paid their mortgages. But if the number of homeowners is nontrivial--and the events of the past two years lead me to believe that it probably is nontrivial--this will significantly increase bankruptcy filings. More importantly, it will increase bankruptcy filings among people who otherwise would not have been in foreclosure but instead would have paid their contracted loans. So for this group of borrowers it is a pure loss to the lenders because they are not saving foreclosure losses. Update: Calvin Massey provides an historical analogue to this provision--the Gold Clause cases. An excellent example in many ways. Calvin concludes that the provision likely is constitutional. Whether it is a good idea is a different question. Update: See update at top.