If you’ve ever found yourself wondering why the financial industry doesn’t learn from its mistakes, today’s Capital & Conflict is for you.

Financial crises are bizarrely common given how painful they can be. Bankers seem to learn slower than toddlers who’ve touched a stove.

But why? Why do we have to learn the same lessons over and over again? Why haven’t we managed to abolish bubbles and busts? Why do banks still fail and leverage ratios reach dangerous heights?

The answer to the mystery is surprisingly simple. Loopy feedback loops.

Consider for example the recent whistle-blower awards. Three senior anonymous Bank of America staff tipped off regulators about misconduct at the investment bank Merrill Lynch. They were awarded $83 million between them for their tip off, which brought in $415 million in fines for the government.

The problem is, Bank of America owns Merrill Lynch. The bankers effectively got a bonus from their own company for its misbehaviour. Not only was such misbehaviour profitable for bankers at the time it was done, it’s proving to be in hindsight too!

It only gets worse if you take a look at who actually paid for the misconduct of banks going into the financial crisis. It wasn’t the wrongdoers. It was the shareholders of the banks who purchased the failed banks.

Bank of America’s shareholders paid for Merrill Lynch’s misbehaviour, for example. The $415 million came from their company in the end. Meanwhile, those who actually performed the misconduct are long gone with their bonuses.

And now the remaining bankers are getting bonuses they could only dream of – $85 million between three – for exposing the misconduct of the bank owned by their employer!

This is the most bizarre set of twisted feedback loops possible. Imagine if your toddler touching the stove made his older brother yelp in pain, while their sister got a $50 million bonus for telling you what happened. Chaos would ensue because the incentives are completely misaligned.

The actual misconduct uncovered by the whistle-blowers in the Merrill Lynch case is fascinating.

Merrill Lynch would keep its clients’ funds pending in the purgatory of a clearing account for an unnecessarily long time. That’s because the bank can use them for its own trading activities while they are there. If Merrill Lynch had failed, the money would’ve accrued to the bank’s creditors, not the customers whose money it was.

The Financial Times has the figures, which make the fine look pretty measly:

Merrill held as much as $58bn per day of customer securities in a clearing account that could have been exposed to claims from the institution’s creditors in the event of bankruptcy. By failing to deposit customer cash in a reserve account, it also freed up $5bn that it used to finance its own trading activities.

The question is, how widespread is this sort of conduct? That’s sort of the general point here too. We still don’t really know how bad things were going into the global financial crisis of 2008. And if we don’t know that, we can’t reach the right conclusions about what happened.

That problem of misaligned incentives applies to our entire understanding of the financial system too. Without an understanding of what happened, our interpretation is likely to be wrong. And that means our response is too.

And that’s how crises keep repeating.

What really happened in 2008?

We all know about the US sub-prime crisis. At least we think we do.

For most of us, it was a harrowing experience as contagion spread around the world. A British bank was nationalised. European governments wobbled. Stocks crashed.

All from a small problem in an obscure corner of the US mortgage market. At least, that’s what they told us at first.

The episode put on show, once again, how a small problem can snowball into a very big one in the financial system. But what caused the crash to become so disastrous?

Academics, journalists, politicians and historians will tell you the wrong answer. They believe the crisis was all about complex financial instruments. Collateralised debt obligations (CDOs), synthetic CDOs, swaps, and tri-party swaps are their favourite topics to discuss. Some add the disastrous miscalculations of ratings agencies like Moody’s, Fitch and S&P.

But if you avoid getting caught up in the complex explanations of all these instruments and risk calculations, you’ll notice something. No never explains how they actually failed. How did all those whizz-kid engineers who designed these complex financial products get it so wrong?

The answer is simple. Derivatives are not to blame. Nor are ratings agencies. As anyone who builds mathematical and risk models will tell you, garbage in, garbage out. It doesn’t matter how good or correct your financial engineering is if it’s based on incorrect data.

Now, I don’t know if the financial system would still have gone through the 2008 crisis if financial engineering like CDOs had been fed good data. But I’m sure they weren’t fed good data. What really happened is all about the information those derivatives and ratings relied upon.

In 2015 a study was published which compared tax records with home loan application records for several geographic regions. If the two diverge for a given suburb, you know people are lying on their mortgage applications. After all, who overstates their income to the tax authorities?

In the study, the researchers looked at the difference between income reported on US home purchase mortgage applications and tax reported income from 2002 to 2005, broken down by zip code.

They found a sudden divergence between the two income figures for neighbourhoods with poorer credit scores – those that would need to falsify income to get a loan. They also found a strong correlation between the divergences and indicators of mortgage stress in subsequent years.

In an interview with journalist Matt Taibbi, an employee of JPMorgan Chase claimed “around 40 percent of [mortgages] were based on overstated incomes”.

The Mortgage Asset Research Institute decided to do an in-depth review of 100 loans. After comparing the income on the loan documents with tax documents, research found almost 60% of the income amounts were inflated by more than 50%.

Other academic studies found that 30% of the loans they examined featured a variety of different misrepresentations in both low and full documentation loans, with incomes for loans in 2005 and 2006 inflated on average between 15% and 20%.

In testimony to the Federal Reserve, the president of the Mortgage Brokers Association for Responsible Lending, Steven Krystofiak, claimed 60% of 100 low and no-doc loans examined overstated income by at least 50%.

One researcher summarise the two key findings of this review:

What are the important lessons from the existing research? First, fraud was endemic to mortgage markets during the mortgage credit boom and income on mortgage applications was routinely falsified. Second, the estimates of fraud from this literature are by their nature lower bound estimates. The researchers in these studies have made a conscious effort to only report fraud when they can explicitly detect it.

The Financial Crisis Inquiry Commission estimated a trillion dollars of loans between just 2005 and 2007 were fraudulent.

In short, manipulation of loan applications was rife in the US. Even now, we don’t really know how rife. It’s just not possible to go through every mortgage in the US and establish the application’s accuracy.

The problem is that this uncertainty allows us to completely misdiagnose the problem. It comes back to the “garbage in, garbage out” discussion above.

Wall Street and its many affiliates relied on the information on loan documentation. Ratings agencies did too. They classified high-risk mortgages as low-risk, and went on to securitise them as such. This is what led to such devastating and unexpected failure, not the financial engineering itself.

Perhaps the academics are correct and CDOs, swaps and other financial innovations are dangerous. But blaming them for what happened is simply lazy. It’s determining your conclusions based on the research that is possible to conduct instead of what really happened.

Even after the attempts of some researchers to uncover that it was liar loans which caused the crisis, this view has not been widely adopted.

Instead, all manner of other things continue to be blamed in academia and the media. The problem is not only that this view is incorrect. After all, it is impossible to analyse the risk of a mortgage if the information you rely on is wrong. So blaming the quality of financial analysis is missing the point.

The deeper problem is that academic and industry analysis with the benefit of hindsight is making the same mistake as debt ratings agencies and investment banks did before the crisis. They’re using that same incorrect data. And that’s why their research into what went wrong continues to be misguided too.

In the end, we can expect many more crises. Just as investments failed due to the wrong information, so too will the policy response. Mistreatments because of a misdiagnosis.

Until next time,

Nick Hubble
Capital & Conflict