I think another issue related to complexity is the approach regulators are likely to pursue. Financial regulators, in particular, will try very hard to avoid another collapse like the 2008 market meltdown. They will come up with various regulatory approaches, and evaluate each against different potential scenarios, especially the situation that led up to the 2008 problems. They will design regulations that seem most likely to prevent these conditions from coming again. Then, they will require all institutions to follow this regulatory scheme.
Unfortunately, it is impossible to design a regulatory scheme that will work in every condition. So, when the condition comes that the regulatory regime doesn't cover, all the institutions will fail together. At the same time. I fear that the cost of bailing out the system will be something approximating "more than anything has ever cost before."
Avoiding this scenario would require a certain humility from regulators; they would have to acknowledge that there is more than one approach to risk management - some are better in some circumstances, some may be better in most circumstances, but none is always superior. This would mean allowing different approaches, which would require the regulators to understand different approaches.
I apologize if I have shared this Substack previously, but I think it is important because it reveals the true cause of the 2008 financial crisis: the Basel Capital Standards. Nothing mattered but Basel. (I mention the recourse rule, but that was small potatoes.)
Goodhart’s Law also becomes a factor. Focusing on a specific measure/metric, undermines its value and leads to unintended consequences. All metrics can be gamed.
Was that true for individual credit scores? Was there widespread gaming by people who should have gotten low scores to get higher scores, undermining the predictive utility of the score? My impression is the opposite, that, despite there being a ton of money at stake, people didn't or couldn't game their scores much, and the scores remained valid, but there was instead a loosening of the standards in terms of the threshold for minimum acceptable scores.
That was my impression as well. The nob was turned to less restrictive as they tried to see what they could get away with (not necessarily in a bad way) but the reaction lag time was longer than the turn time, so they didn't realize they had gone too far until it was far too late.
Goodhart's Law works in cases where you use legible measures/metrics to optimize some less legible process or unquantifiable outcome. E.g. if you want to maximize sales and measure sales volume, there is not that much scope for gaming. If you want to maximize citizen wellbeing or health and measure GDP or health spending, though...
These insights point to simplicity is an end in and of itself when considering government reform.
A simple, durable system that people can comprehend is likely preferable to a slightly more "efficient" but incomprehensible and inflexible system. The complexity and inflexibility impose costs at the worst possible times, and encourage rent seeking that far outweighs the efficiency gains.
Indeed. The more layers and rules one applies to try and squeeze a little more efficiency out, the more points of failure and unintended results one creates, and not in a linear fashion. I love my chainsaw, but my ax always works, so one sees wildly more use than the other.
I mean by “the problem” the decline in GDP, fall in the price level unemployment, less than target inflation that went on for years. If the Fed had started fire hosing liquidity into the system from October on, said clearly that it was not going to allow prices to fall below trend, things woud have been very different.
But the Fed didn’t directly cause the problem loans.
And didn’t cause the packaging of the loans.
And didn’t cause the AAA ratings placed on the loans (and were not the ones responsible for regulating the ratings agencies).
The Fed did not create the CDOs and CDSs nor fail to regulate those in the first place (though it was one of many who failed to realize the significance or say something about it earlier).
And the Fed were not the ones who decided to steal from bondholders to give to unions as part of the GM bailout.
All that said the Fed *shared* in the responsibility in causing the crisis and making it worse with bailouts, and I agree with you and Scott Sumner that subsequent monetary policy made the crisis worse than it needed to be.
I also disagree with the auto bailout and similar things done during COVID.
I think I disagree that the bailouts themselves made things worse.
Obviously regulation of CDO/CDS was not right, but there is nothing wrong per se in packaging debt to spread the risk of the individual components and I sure do not know what would have been. The basic problem, however, is that entire financial system cannot hedge itself (although regulators can make sure it tries). Ultimately that depends on the Fed willing to “do what it takes.” It flailed to do that in 1929 and 2008 (not as badly).
“I think I disagree that the bailouts themselves made things worse.”
I don’t have a per se problem that some amount of bailing out was done, as I agree we can’t know that things wouldn’t have been worse absent all bailouts. So I even agree that doing some amount of bailouts was probably prudent.
I have a HUGE problem that the bailouts were 100 cents on the dollar. This was an unnecessary giveaway to Goldman Sachs, rewarded those who were more highly leveraged at the expense of the prudent and the taxpayer, and causes more moral hazard going forward. It was “heads I win, tails you lose” crony capitalism. I don’t accept at all the premise that it had to be a full bailout.
Where between 90 cents and 99 cents on the dollar would have been appropriate and avoided systemic collapse / unduly risky strain, idk. But I absolutely refuse to believe that 99 cents on the dollar would have been bad, and doing that would have at least sent SOME proper signal.
Now the Great Depression is fine to lay almost totally at the feet of the Fed. But 2008 they only deserve partial blame IMO.
I could not agree more and I don’t oppose better prudential regulation [I think it was failure of regulators that SVB was as heavily LT fixed rate bonds as it was] so long as we bear in mind that there can be Type II errors, not just Type I errors.
Even bigger failure of management and board. This one wasn’t complex at all.
Somewhat similar to 2008-2009 bailouts, I’m ok that the depositors lacking FDIC insurance got money back, and I particularly agree it was useful they got their money back quickly, but it was ridiculous that they got back 100 cents on the dollar.
Massive moral hazard, AND massive advantage for big banks over small and midsize ones. Why risk depositing your money in a small or even midsize bank when in a big one you will get bailed out if there are any problems?
Same question I had though I think our chances of modeling complexity are doomed. I’d be interested in thinking about how to muddle through better. This would be a great Arnold book. I understand the libertarian approach. It’s definitely the best if there weren’t humans involved. Politicians will always respond to crises with new regulations. How can we do the best we can given that fact?
Interesting that “complexity economics” (CE) is mentioned below, by W Sylva Bryan. I have not followed AK’s column/blog that long. I was introduced to his writing when I read, Specialization and Trade: a Re-introduction to Economics last year. I was glad to see that early-on in the book he stated, “The economy is a complex, evolving system.” But I found no reference to CE in the book. I have been interested in ecological and economic systems as complex, evolving systems for some time. I discovered the Santa Fe Institute was founded in the 1980’s to study complex systems. One of their early publications (The Economy as an Evolving Complex System) was of a workshop organized in part by Nobel Prize Economist, Kenneth J. Arrow. “Ten theoretical economists and ten people from physics, biology, and computer scientists were invited to this workshop. The purpose of the workshop was to explore the potential usefulness of a broadly transdisciplinary research program on the dynamics of global economic systems by bringing together a group of economists and natural scientists who have developed techniques for studying nonlinear, dynamical systems and adaptive paths in evolutionary systems,” (Foreword, TEECS). Recently, one of those physicists, J. Doyne Farmer, published a book (Making Sense of Chaos: A Better Economics for a Better World) that I daresay represents a current presentation of studies of CE that have been going on by some of that group since then. When interviewed about the book recently by Russ Roberts on EconTalk, Farmer mentioned that although they had tried, they had been unable to publish these studies in top-tier economics journals: they had to publish in second tier economic journals. Russ sounded somewhat surprised: but he admitted that although he knows Robert L. Axtell, a Farmer collaborator (agent-based modeling) and the Chair of the Department of Computational Social Sciences at George Mason University, he never had talked to him about his work. I would be curious to learn if AK has an opinion on CE, and whether he considers that approach to the study of economic systems worthwhile?
“The attempt to separate commercial banking from investment banking was abandoned.”
I agree with AK’s overall take here. And I’m am generally pro-deregulation, even in the financial services area where some regulation clearly and obviously is necessary.
But this particular bit of deregulation *was*, at least in hindsight, bad. Among other reasons, it was a major cause of “too big to fail” bailouts.
Banks that benefit from FDIC deposit insurance shouldn’t be allowed to make highly leveraged risk bets. especially when we are gonna have the idea of “systemically important” banks.
Is it fair to summarize your credit scoring example as follows? Making credit scoring more efficient reduces overall performance of the system. I have long been thinking that such examples should not be treated as exceptions. Is there a systematic theory of when more (typically short term) efficiency leads to worse (typically long term) performance? Or are you saying that such a theory cannot exist for complex system (basically because that is what it means to be complex)?
I think it is important to notice that the advent of credit scoring did not by itself lead to bad results. There was later a requirement by regulators to basically give mortgages to people you wouldn't give them to based on the credit score.
"With credit scoring, Wall Street was able to securitize loans from low-income borrowers. This caused political embarrassment for Freddie Mac, which enjoyed government backing in part because politicians wanted to expand access to mortgage lending. Congress and the President set goals for Freddie Mac to demonstrate that it was providing mortgages to low-income and minority borrowers. In its effort to meet those goals, Freddie Mac allowed the credit quality of its loan portfolio to erode."
That is where the Recourse Rule comes in. It said that banks had to hold only minimum capital against highly-rated mortgage securities. That made the financial engineering effort all about manufacturing high ratings. The process of rating the mortgage securities was gamed, allowing the repackaged bad mortgages to be sold as highly rated mortgage securities.
The rating process was gamed—so as to allow banks to hold less capital, and be more highly leveraged/bigger gamblers, and be more profitable as long as their assumption remains true that “house prices always go up”.
House building peaked in 2006.
The rating agencies were allowed to rate bad mortgages, when combined into a big security, as good. I think the rating agencies were following Fed/ govt guidelines.
The banks should have been allowed to fail, and successfully sue the rating agencies, and all be rebuilt with higher capital requirements, which makes banks safer but less profitable.
Yes, I did. That brings up an interesting question. What if there had been the packaging but no government pressure to make risky loans to politically favored borrowers?
Arnold – what do you think about Russ Robert’s’ discussion the other day on feedback? That was very much focused on education, but I kept feeling that the concepts discussed were related to areas well beyond education.
I think another issue related to complexity is the approach regulators are likely to pursue. Financial regulators, in particular, will try very hard to avoid another collapse like the 2008 market meltdown. They will come up with various regulatory approaches, and evaluate each against different potential scenarios, especially the situation that led up to the 2008 problems. They will design regulations that seem most likely to prevent these conditions from coming again. Then, they will require all institutions to follow this regulatory scheme.
Unfortunately, it is impossible to design a regulatory scheme that will work in every condition. So, when the condition comes that the regulatory regime doesn't cover, all the institutions will fail together. At the same time. I fear that the cost of bailing out the system will be something approximating "more than anything has ever cost before."
Avoiding this scenario would require a certain humility from regulators; they would have to acknowledge that there is more than one approach to risk management - some are better in some circumstances, some may be better in most circumstances, but none is always superior. This would mean allowing different approaches, which would require the regulators to understand different approaches.
I apologize if I have shared this Substack previously, but I think it is important because it reveals the true cause of the 2008 financial crisis: the Basel Capital Standards. Nothing mattered but Basel. (I mention the recourse rule, but that was small potatoes.)
https://charles72f.substack.com/p/basel-faulty-the-financial-crisis
Goodhart’s Law also becomes a factor. Focusing on a specific measure/metric, undermines its value and leads to unintended consequences. All metrics can be gamed.
Was that true for individual credit scores? Was there widespread gaming by people who should have gotten low scores to get higher scores, undermining the predictive utility of the score? My impression is the opposite, that, despite there being a ton of money at stake, people didn't or couldn't game their scores much, and the scores remained valid, but there was instead a loosening of the standards in terms of the threshold for minimum acceptable scores.
That was my impression as well. The nob was turned to less restrictive as they tried to see what they could get away with (not necessarily in a bad way) but the reaction lag time was longer than the turn time, so they didn't realize they had gone too far until it was far too late.
Goodhart's Law works in cases where you use legible measures/metrics to optimize some less legible process or unquantifiable outcome. E.g. if you want to maximize sales and measure sales volume, there is not that much scope for gaming. If you want to maximize citizen wellbeing or health and measure GDP or health spending, though...
These insights point to simplicity is an end in and of itself when considering government reform.
A simple, durable system that people can comprehend is likely preferable to a slightly more "efficient" but incomprehensible and inflexible system. The complexity and inflexibility impose costs at the worst possible times, and encourage rent seeking that far outweighs the efficiency gains.
Indeed. The more layers and rules one applies to try and squeeze a little more efficiency out, the more points of failure and unintended results one creates, and not in a linear fashion. I love my chainsaw, but my ax always works, so one sees wildly more use than the other.
I agree. Reformers ought to understand that reform is a continuing process, not an event.
But I disagree that deregulation brought down the 2008 financial system. It was the Fed's tardy and inadequate response.
The Fed’s tardy and inadequate response made the problem worse. But harder to say that it caused “the problem” in the first place.
I mean by “the problem” the decline in GDP, fall in the price level unemployment, less than target inflation that went on for years. If the Fed had started fire hosing liquidity into the system from October on, said clearly that it was not going to allow prices to fall below trend, things woud have been very different.
Ok. Then we probably 75% agree.
But the Fed didn’t directly cause the problem loans.
And didn’t cause the packaging of the loans.
And didn’t cause the AAA ratings placed on the loans (and were not the ones responsible for regulating the ratings agencies).
The Fed did not create the CDOs and CDSs nor fail to regulate those in the first place (though it was one of many who failed to realize the significance or say something about it earlier).
And the Fed were not the ones who decided to steal from bondholders to give to unions as part of the GM bailout.
All that said the Fed *shared* in the responsibility in causing the crisis and making it worse with bailouts, and I agree with you and Scott Sumner that subsequent monetary policy made the crisis worse than it needed to be.
I also disagree with the auto bailout and similar things done during COVID.
I think I disagree that the bailouts themselves made things worse.
Obviously regulation of CDO/CDS was not right, but there is nothing wrong per se in packaging debt to spread the risk of the individual components and I sure do not know what would have been. The basic problem, however, is that entire financial system cannot hedge itself (although regulators can make sure it tries). Ultimately that depends on the Fed willing to “do what it takes.” It flailed to do that in 1929 and 2008 (not as badly).
“I think I disagree that the bailouts themselves made things worse.”
I don’t have a per se problem that some amount of bailing out was done, as I agree we can’t know that things wouldn’t have been worse absent all bailouts. So I even agree that doing some amount of bailouts was probably prudent.
I have a HUGE problem that the bailouts were 100 cents on the dollar. This was an unnecessary giveaway to Goldman Sachs, rewarded those who were more highly leveraged at the expense of the prudent and the taxpayer, and causes more moral hazard going forward. It was “heads I win, tails you lose” crony capitalism. I don’t accept at all the premise that it had to be a full bailout.
Where between 90 cents and 99 cents on the dollar would have been appropriate and avoided systemic collapse / unduly risky strain, idk. But I absolutely refuse to believe that 99 cents on the dollar would have been bad, and doing that would have at least sent SOME proper signal.
Now the Great Depression is fine to lay almost totally at the feet of the Fed. But 2008 they only deserve partial blame IMO.
I could not agree more and I don’t oppose better prudential regulation [I think it was failure of regulators that SVB was as heavily LT fixed rate bonds as it was] so long as we bear in mind that there can be Type II errors, not just Type I errors.
Yes failure of regulators.
Even bigger failure of management and board. This one wasn’t complex at all.
Somewhat similar to 2008-2009 bailouts, I’m ok that the depositors lacking FDIC insurance got money back, and I particularly agree it was useful they got their money back quickly, but it was ridiculous that they got back 100 cents on the dollar.
Massive moral hazard, AND massive advantage for big banks over small and midsize ones. Why risk depositing your money in a small or even midsize bank when in a big one you will get bailed out if there are any problems?
Agree on the big/small bank bias.
So can we ever hope to model complexity? Or are we fated to “muddling through” rather haphazardly forever?
Same question I had though I think our chances of modeling complexity are doomed. I’d be interested in thinking about how to muddle through better. This would be a great Arnold book. I understand the libertarian approach. It’s definitely the best if there weren’t humans involved. Politicians will always respond to crises with new regulations. How can we do the best we can given that fact?
Interesting that “complexity economics” (CE) is mentioned below, by W Sylva Bryan. I have not followed AK’s column/blog that long. I was introduced to his writing when I read, Specialization and Trade: a Re-introduction to Economics last year. I was glad to see that early-on in the book he stated, “The economy is a complex, evolving system.” But I found no reference to CE in the book. I have been interested in ecological and economic systems as complex, evolving systems for some time. I discovered the Santa Fe Institute was founded in the 1980’s to study complex systems. One of their early publications (The Economy as an Evolving Complex System) was of a workshop organized in part by Nobel Prize Economist, Kenneth J. Arrow. “Ten theoretical economists and ten people from physics, biology, and computer scientists were invited to this workshop. The purpose of the workshop was to explore the potential usefulness of a broadly transdisciplinary research program on the dynamics of global economic systems by bringing together a group of economists and natural scientists who have developed techniques for studying nonlinear, dynamical systems and adaptive paths in evolutionary systems,” (Foreword, TEECS). Recently, one of those physicists, J. Doyne Farmer, published a book (Making Sense of Chaos: A Better Economics for a Better World) that I daresay represents a current presentation of studies of CE that have been going on by some of that group since then. When interviewed about the book recently by Russ Roberts on EconTalk, Farmer mentioned that although they had tried, they had been unable to publish these studies in top-tier economics journals: they had to publish in second tier economic journals. Russ sounded somewhat surprised: but he admitted that although he knows Robert L. Axtell, a Farmer collaborator (agent-based modeling) and the Chair of the Department of Computational Social Sciences at George Mason University, he never had talked to him about his work. I would be curious to learn if AK has an opinion on CE, and whether he considers that approach to the study of economic systems worthwhile?
“The attempt to separate commercial banking from investment banking was abandoned.”
I agree with AK’s overall take here. And I’m am generally pro-deregulation, even in the financial services area where some regulation clearly and obviously is necessary.
But this particular bit of deregulation *was*, at least in hindsight, bad. Among other reasons, it was a major cause of “too big to fail” bailouts.
Banks that benefit from FDIC deposit insurance shouldn’t be allowed to make highly leveraged risk bets. especially when we are gonna have the idea of “systemically important” banks.
Is it fair to summarize your credit scoring example as follows? Making credit scoring more efficient reduces overall performance of the system. I have long been thinking that such examples should not be treated as exceptions. Is there a systematic theory of when more (typically short term) efficiency leads to worse (typically long term) performance? Or are you saying that such a theory cannot exist for complex system (basically because that is what it means to be complex)?
I think it is important to notice that the advent of credit scoring did not by itself lead to bad results. There was later a requirement by regulators to basically give mortgages to people you wouldn't give them to based on the credit score.
"With credit scoring, Wall Street was able to securitize loans from low-income borrowers. This caused political embarrassment for Freddie Mac, which enjoyed government backing in part because politicians wanted to expand access to mortgage lending. Congress and the President set goals for Freddie Mac to demonstrate that it was providing mortgages to low-income and minority borrowers. In its effort to meet those goals, Freddie Mac allowed the credit quality of its loan portfolio to erode."
Dont you leave out the part of the stories were bad mortgages were repackaged and the package was sold as good?
That is where the Recourse Rule comes in. It said that banks had to hold only minimum capital against highly-rated mortgage securities. That made the financial engineering effort all about manufacturing high ratings. The process of rating the mortgage securities was gamed, allowing the repackaged bad mortgages to be sold as highly rated mortgage securities.
The rating process was gamed—so as to allow banks to hold less capital, and be more highly leveraged/bigger gamblers, and be more profitable as long as their assumption remains true that “house prices always go up”.
House building peaked in 2006.
The rating agencies were allowed to rate bad mortgages, when combined into a big security, as good. I think the rating agencies were following Fed/ govt guidelines.
The banks should have been allowed to fail, and successfully sue the rating agencies, and all be rebuilt with higher capital requirements, which makes banks safer but less profitable.
Yes, I did. That brings up an interesting question. What if there had been the packaging but no government pressure to make risky loans to politically favored borrowers?
Arnold – what do you think about Russ Robert’s’ discussion the other day on feedback? That was very much focused on education, but I kept feeling that the concepts discussed were related to areas well beyond education.