Cost of the 2007--9 financial crisis

[Kimura]

The cost of the financial crisis was $50.

That's a typo.

 

[oldfart]

The figures I started with are from the Dallas Fed research paper and I added some commentary. You may well be right that the worst is yet to come; I hope not, but it's possible. I think very few people have realized how much damage has been done and underestimate what it will take to engineer a true recovery.

The most depressing thing to me is that Americans seem to have accepted the "new normal". I don't see an economic or political mechanism that will reliably put us on a recovery path. Most forcasting models have a five-year horizon. We are nearer to the worst case scenario of the late-2008 model runs than the average projection, and the models looking forward to 2016-7 don't look good (except for CBO which always is too optimistic two or more years out). And we still have not made the changes in financial markets which would prevent another financial collapse. I just have a feeling that we are running out of time to fix financial markets.

Why do you think we are nearer to the worst case scenario?

OK, I have to back up some here. In the fall of 2008 I was fortunate to have some friends from grad school who work for the two of the big forecasting shops. I'm just on the fringe, but when things look like they are blowing up, people want to talk. Some of it was what they were doing, and some was what other shops were doing. When the government was pitching a stimulus plan "to keep the unemployment rate under 8%", my public prediction was for it to top off at 10.5%. I chose the number mainly because it was a rough average of models I respected. My prediction for the low on the Dow (I was 350 points and two months off) was sheer luck.

Anyway, one of the models was popping out probabilities for where the economy would be in five years. Absent the Depression, no downturn has lasted much over a year and a half, so the five-year period to achieve pre-slump real per capita GDP was considered a given. The models said it wasn't. In fact, in January 2009, that model had a 40% probability of no recovery at all. This kind of result has only happened in the Great Depression and the Long Depression of the late nineteenth century.

If you remember late 2008 and early 2009, most every economist and policy-maker assumed that automatic stabilizers and reasonable monetary and fiscal policy would make it a severe by short recession. No one could envision the economy being fucked up enough to stall that. In early 2009, the middle probability scenario for 2013-4 was unemployment at 7.5%, growth at 2.0%, slight deflation, and real per capita GDP even with 2007, plus or minus a percent or two. At the time those seemed to be horrible numbers and I like everyone else had a hard time getting my brain around it.

Check out the graphs in the study I linked to in the first post. They pretty well tell a story that almost everyone in 2008 would have said was unthinkably bad. But we are now used to it.

My pessimism is rooted in two observations. First, we have seemed to have accepted the "new normal" and there is little political will to actually return the economy to robust growth. Second, in 2008 it was a given as the crisis developed that the collapse of the global economy was a real possibility and that significant structural changes were needed to avoid another meltdown. Those ideas are off the table and the potential to repeat is much too high.

I apologize for this sounding too "touchy-feelly". If anyone wants to pick any of the components here to discuss, I'd be glad to join in.

Peace all. Jamie

 

[Toro]

I think the nature of how central banks approach the markets has changed. They are now more focused on asset prices. One can argue that this began in 1987 after the crash but I think it really started in 1997-98 during the Asian Contagion and the collapse of LTCM. I believe the Fed facilitated the tech bubble and was the primary reason behind the housing bubble.

From WWII to 2000, stocks were always higher 12 months after the Fed cut interest rates 3 times (IIRC). However, stocks fell ~40% after the Fed cut three times in 2001 and 2008. That is a fundamentally different response than all the prior recessions, which tells me that the nature of the economy has changed.

 

[william the wie]

The figures I started with are from the Dallas Fed research paper and I added some commentary. You may well be right that the worst is yet to come; I hope not, but it's possible. I think very few people have realized how much damage has been done and underestimate what it will take to engineer a true recovery.

The most depressing thing to me is that Americans seem to have accepted the "new normal". I don't see an economic or political mechanism that will reliably put us on a recovery path. Most forcasting models have a five-year horizon. We are nearer to the worst case scenario of the late-2008 model runs than the average projection, and the models looking forward to 2016-7 don't look good (except for CBO which always is too optimistic two or more years out). And we still have not made the changes in financial markets which would prevent another financial collapse. I just have a feeling that we are running out of time to fix financial markets.

Why do you think we are nearer to the worst case scenario?

OK, I have to back up some here. In the fall of 2008 I was fortunate to have some friends from grad school who work for the two of the big forecasting shops. I'm just on the fringe, but when things look like they are blowing up, people want to talk. Some of it was what they were doing, and some was what other shops were doing. When the government was pitching a stimulus plan "to keep the unemployment rate under 8%", my public prediction was for it to top off at 10.5%. I chose the number mainly because it was a rough average of models I respected. My prediction for the low on the Dow (I was 350 points and two months off) was sheer luck.

Anyway, one of the models was popping out probabilities for where the economy would be in five years. Absent the Depression, no downturn has lasted much over a year and a half, so the five-year period to achieve pre-slump real per capita GDP was considered a given. The models said it wasn't. In fact, in January 2009, that model had a 40% probability of no recovery at all. This kind of result has only happened in the Great Depression and the Long Depression of the late nineteenth century.

If you remember late 2008 and early 2009, most every economist and policy-maker assumed that automatic stabilizers and reasonable monetary and fiscal policy would make it a severe by short recession. No one could envision the economy being fucked up enough to stall that. In early 2009, the middle probability scenario for 2013-4 was unemployment at 7.5%, growth at 2.0%, slight deflation, and real per capita GDP even with 2007, plus or minus a percent or two. At the time those seemed to be horrible numbers and I like everyone else had a hard time getting my brain around it.

Check out the graphs in the study I linked to in the first post. They pretty well tell a story that almost everyone in 2008 would have said was unthinkably bad. But we are now used to it.

My pessimism is rooted in two observations. First, we have seemed to have accepted the "new normal" and there is little political will to actually return the economy to robust growth. Second, in 2008 it was a given as the crisis developed that the collapse of the global economy was a real possibility and that significant structural changes were needed to avoid another meltdown. Those ideas are off the table and the potential to repeat is much too high.

I apologize for this sounding too "touchy-feelly". If anyone wants to pick any of the components here to discuss, I'd be glad to join in.

Peace all. Jamie

I think the problems you are identifying are based on a lack of qualitative analysis.

Take for example Mandelbrot's analysis of the power law distribution in capital markets. He never asked why this should be. But there are solid reasons why this should be true.

Stop losses multiply downturns and the only investor who regularly used stop gains was Benjamin Graham who cut his gains when he reached Fair Market Value. When a common market practice amplifies downturns and its logical complement even if rarely used reduces gains that creates a downward bias in outcomes.

One additional reason for downward bias is margin calls.

So, what are the rational behaviors that lead to amplification of upturns? I can't think of any. So why do people and economists in particular assume a random walk?

Yet how much of econometrics is based on normal curve distribution of asset prices?

 

[oldfart]

I think the problems you are identifying are based on a lack of qualitative analysis....

So, what are the rational behaviors that lead to amplification of upturns? I can't think of any. So why do people and economists in particular assume a random walk?

Yet how much of econometrics is based on normal curve distribution of asset prices?

A large part of the problem is that the amplifiers of upturns are "irrational exuberance". Until the Minsky moment, every market participant has a vague feeling that the market is oversold, but they can't get off for fear of making the call too early and being left out of the next round of profits.

Random walks, normal distributions, self-reinforcing cycles, and logarithmic growth rates are not assumptions in econometrics. They are a combination of observed behavior and the mechanics of stochastic processes. Any distribution (even a Poisson) begins to look like a normal distribution as the number of observations increases. The trick is to not look at the distribution as being "normal" but to analysize it characteristics: central tendency, variation, skewedness, and kurtosis.

I don't work much with financial market statistics, so I may be missing a lot there. A lot of my work as an expert witness required statistical methods that had little to do with normal distributions. Typically we started out with ordinal data and Chi-square Goodness of Fit tests. Surprise! When you have a 500 employee database, earnings turn out to be a normal curve. The question is which variables are most significant in predicting where a given person will be on that distribution. It turned out that education, job performance, and experience were not significant predictors of income or advancement.

Anyway, if you can give me a better idea of the misuse of statistics you see, I can give you a less general explanation of what I don't know!

 

[william the wie]

I see the collision in use of terms. In the equities market returns that are 2 sd below the mean occur 8% of the time, there are effectively no returns 2sd above the mean as Paul Samuelson pointed out when the current version of the Efficient Market Hypothesis was launching in the 1960s. 0.92/0.95 = 0.97 so EMH is taken as proven even though one day moves greater than -10% are more common than one day moves of +5% for the market as a whole.

There are very effective accelerants of downward moves but a lack of effective accelerants to the upside, puts when available are about the only case of accelerants to the upside. An example of this is that SPLV, the low volatility S&P fraction ETF has an active options market, SPHB the high volatility S&P fraction ETF does not have options. The SPLV has a positive Y intercept, SPHB a negative intercept when regressed against the market as a whole.

By the way thank you very much for the above post and this thread. I am writing a sequel to WWWEP about the process of dumbing down economics to bumper sticker policies and this is most helpful. Unhappily I can't yet green you again.