Are Paulson and Bernanke Fear-Mongering?
I wrote a post called “Irresponsible Folly” on Monday that harshly criticized House Republicans who voted against the bailout. Within about two hours, I received over 250 emails, almost all critical of my views (topping the previous record for negative emails received after my critical “Wilma Jennings Bryan” post about Sarah Palin’s acceptance speech). The vast majority of these emails were civil, intelligent and principled. Almost half of them asked some version of what I consider to be an excellent question: “How do we know that we really are at risk for a financial catastrophe if we don’t pass a bailout?”
Contagion – a cascading series of defaults resulting in a loss of confidence in financial institutions, and therefore in a severe economic contraction that hurts just about everybody, whether they were responsible or irresponsible about debt – is the catastrophic scenario that Paulson and Bernanke are asserting is a real risk for us. There are different levels of potential pain. It looks like we’ve moved past the “only the foolish get burned”, and “only the foolish plus rich speculators get burned” levels. The next level is a serious recession in which normal policy tools and the normal operation of the market can correct things. The next level is something like the Great Depression, or full-scale contagion. The true nightmare scenario (yes, even worse) is a loss of confidence in the ability of the U.S government to make good on its obligations. This would result in a run on the dollar, which is to say a run on the big bank called the U.S. government based on a loss of confidence by investors, both foreign and domestic.
This is all easy to say, but what are the odds of any of these negative scenarios occurring?
Let’s start with the observation that we had an almost perfect natural experiment on Monday to measure equity market reactions to a reduction in the probability of a bailout. The widely-tracked Dow (a stock market index that is a very imperfect proxy for the U.S. stock market) lost about 780 points yesterday. That is a really bad day, but on a percentage basis, not even in the top 10 worst days in the history of the Dow. Further, if you look at the Dow minute-by-minute through the day, it opened down, then traded very steadily until 1:30PM when vote results started to come in, then fell off a cliff, and was down about 540 points from 1:30PM to the close. It’s impossible to isolate reliably the impact of any single piece of information in one of these exercises, but I think it’s a fair guess to say the market (or, more precisely, the Dow proxy) lost about 500 points when it integrated the information that the bailout bill had not passed. Reasonable confirmation of this estimate is that as confidence rose during the day Tuesday that a bailout will, in fact, be passed, the Dow went up by about 500 points. To compare this to a random big number, this represents a loss Monday and then a gain back Tuesday of a little more than $700BB of market value.
Now, the probability that the bailout will be made into law sometime soon was not 100% Monday morning, and was not 0% Monday night. In fact, I think most observers believe that some plan somewhat like the modified Paulson plan is more likely than not to pass sometime soon. To pick rough numbers out of the air, let’s assume that the odds of passage dropped from 9-in-10 to 3-in-4 on Monday afternoon when the bill failed. The reduction in odds of passage would have gone from 90% to 75%, or been reduced by 15%. So the market went down about 500 points in response to a 15% reduction in odds of passing. (In addition to being wild guesses for numbers, this is a gross simplification of the real market calculus of the degradation of the probability cloud of possible outcomes, e.g., odds of passage at all, odds of specific worse kinds of bills passing, increased time-at-risk while the debate continues, etc.) So a crude estimate of the reduction in equity value attributable to certain failure of the bill is something like 500 / 15%, or a 3,300 point decline in the Dow. Apparently Paulson and Bernanke have put forth an estimate to congressional leaders that the equity market impact of not passing legislation to address the debt crisis would be “3,000 to 4,000 points”. So far, then, they seem to be doing OK in the prediction department.
How big would a 3,300 point drop in the Dow be? That’s about a 30% decline versus the Dow at 1:30PM yesterday – from 10,907 to 7,607. It would also be about a 50% decline vs. the peak of last July. How bad would that be? That’s a really severe bear market, but not necessarily disaster. The Dow dropped 22.6% in one day in 1987, and there weren’t exactly armies of Joads wandering America in 1988. It’s possible that even if this occurred, we would muddle through with a typical recession, or in an extraordinarily sunny case, no recession at all.
So what’s the big deal then – on what basis could we justify such a huge market intervention if even the bad case just isn’t that bad? Remember that this decline would not be the catastrophe; it would be the bet of the stock market about the odds-adjusted loss in expected future collection of money (from the perspective of equity holders) from businesses. Part of this decline would reflect the loss in expected value from the odds-adjusted impact of the normal recession scenario, and part would reflect the odds of a true Great Depression-style scenario.
How can we estimate a way to break-out these odds? Bear markets come and go, but there were four huge bear markets for the Dow in the 20th century: 1906-1907, 1912-1914, 1928-1932, and 1972-1974. In each of them the Dow declined around 40% in the first two years (a little more in the 1928-1932 case). Three of the four then stopped declining. In the 1928 – 1932 case, it kept going for two more years, ultimately declining more than 80% from its peak, and signaling the Great Depression. 25% of these mega-bear markets didn’t find a floor, and just kept going down. Given the amount of poorly understood debt permeating the U.S. economy right now in combination with ongoing collapse of the real estate bubble, the mechanism by which a 50% decline from July 2007 to October 2008 could continue and expand into full-scale contagion is quite clear. Once this gets going, it would be extraordinarily hard to stop.
So a very rough guess is that if we do absolutely nothing about the credit crisis, we are running something like a 25% chance of a true catastrophe at the level of the Great Depression. I’ve gone through the assumptions of my analysis; you can play with them at leisure, but I think you’d be very hard put to end up with an estimate that is less than 10% or greater than 50%.
Each day they leave this problem unaddressed, Congress is basically flicking a lit match at a lake of gasoline. But, hey, it hasn’t caught fire yet, and maybe we’ll stay lucky.
Is this what you would have believed last week? Would the Dow have moved the same had Paulson announced no such bailout plan? It seems that at least part of the market effect comes from Paulson’s own announcement that a failure to pass the bailout plan would be catastrophic. While your analysis may be a correct parsing of market expectations, those expectations were formed as a part of the bailout announcement.
— Thorfinn · Oct 1, 08:49 PM · #
I believe that is, at least in part, true, and in fact made that argument in a post last week. Undoing the announcement of the Paulson plan can not be done lightly.
The counter-factual of what would have happened in the absence of his announcement a week ago Thursday can not be known, of course. There was clearly a run happening on money market mutual funds between 10AM and noon on that day, and I think there is a reasonable assumption that had something dramatic not been done immediately, there is a realistic risk that the wheels could have come off.
— Jim Manzi · Oct 1, 09:20 PM · #
An interesting post. One thing that I don’t think it addresses is the time-scale. Given that the Paulson plan was going to put money into the markets over the course of several months, not all at once, should we assume that the markets were reacting to the probability of a bailout getting passed, regardless of whether it’s late september or late october?
— Justin · Oct 1, 09:36 PM · #
Jim,
While I usually find your analysis quite persuasive, I am not sure I understand this. Why is the performance of the Dow the relevant indicator of the chances of contagion? It seems to me that the Dow jumps and dives on all sorts of information all the time. Over the very long-term it may indeed reflect the fundamentals of the economy, but it seems unlikely that single day jumps or troughs have any relevant information in them—they are just people placing bets on bad feelings. Indeed, the very premise of the bailout is that accurately predicting the probability of contagion is impossible. Such a prediction would require being able to value toxic MSB assets in the short term, so that we can tell how many firms are insolvent. The primary point of the bailout, it seems to me, is that this is currently impossible, and that the treasury must make the market so that it can happen (in addition to recapitalizing a lot of these firms along the way).
I share your fears about the possibility of a severe contraction. But given that there seem to be no way of knowing its odds, and given that the paulson plan seems awful, difficult to execute, as well as potentially insufficient to stop such a contraction, I think it is fair to ask for a better plan. Other ways have been suggested of achieving the two primary goals of the bailout: keeping credit flowing to commercial businesses and figuring out which big financials are insolvent. Among them are lower capital controls for most commercial lenders and temporarily legalizing insider trading. I am not advocating either of these necessarily, but it seems worth putting a lot of ideas on the table. The problem with the house republicans and democrats is not that they killed the bailout, but that they did so without principled alternatives.
— brendan · Oct 1, 09:40 PM · #
Justin:
Thanks. As per the post, it was surely a chnage in a cloud of probabilities around the timing, scope, design, effectiveness and so on of the plan, in addition to whether or not it passed.
— Jim Manzi · Oct 1, 09:40 PM · #
previous comment should read capital requirements, sorry.
— brendan · Oct 1, 09:42 PM · #
Brendan:
Separate two steps: (1) What would be the Dow reaction to failure to pass the bailout, and (2) what predictive information would this reaction give us as to the likelihood of a massive dislocation.
1. The market reaction tells us about the first. Subject to a million caveats, Monday’s action supports Paulson and Bernanke’s forecast that we would see a 3,000 – 4,000 point drop if the bailout definitively failed.
2. In combination with the decline since last summer, this would represent about a 50% decline in the Dow over about 18 months. 1 in 4 such declines in the 20th century were followed by the Great Depression. Subject to post hoc ergo propter hoc caveats, this in combination with the obvious mechanism for contagion, supports a crude estimate of a 25% chance of such an event if no action is taken.
I purposely did not focus in this post on whether or not the specific proposal is the best possible course of action, just on what are the risks facing us in the case of no action vs. this plan.
— Jim Manzi · Oct 1, 09:56 PM · #
Jim, as someone who manages money for a living (at one of those infernal hedge funds), I think you’re underestimating the pessimism after the failed House vote on Monday. My sense from the various people I dealt with over the last 48 hours is that the market’s perception of the chances of a bailout passing (using your very simplified framework) went from 90% (which is about right) to something closer to 60%-70% after the vote. People were really surprised by the vote outcome and some of my counterparts and friends were convinced that it meant the Paulson plan was dead.
I know that seems to be splitting hairs, but it changes the implied impact of total legislative failure to 15-25% further down from Monday’s close. (Which just goes to show how sensitive the math is to our initial, finger-in-the-air assumptions.)
Also, as you state, you’ve simplified away the possibility of passage of a less favorable bill, which is what nearly everyone saw (and sees) as the most likely alternative outcome. Certainly everyone I know thinks that something will get legislated — it’s a question of how soon and how helpful will it be.
Most importantly, though, the Dow is already down roughly 27% peak to trough. A 20% decline from Monday’s close would be enough to get us to a total of 40% down. But the fact is that we’re most of the way there and the economy is only now tipping into a recession. There are a lot of strong arguments why looking at the economic consequences of a bear market under a gold standard are not especially relevant, but I’ve already gone on long enough; I’ll just say that I think 25% is too high a number for conditional probability of contagion. My number would be much closer to your lower bound.
P.S., for Matt Frost: “Henry Rollins.”
— Rover · Oct 1, 10:31 PM · #
Good post. It’s the second in a row, following the Avent video blog, to think of tail risk versus means in terms of cost-benefit analysis and economic modeling. Interesting to see the parallels.
I know you qualify everything a million times, but both 25% and 10-50% for a Great Depression seems incredibly high given our current economic models, though – plugging away Black-Scholes gives us a 10% and a .3% chance of losing 50% and 80% of the stock market’s value (respectively) given current high volatility over the next year.
I don’t see the contagion freakout in the options market. Backing out current probabilities from S&P and DJIA index options for March 09, I’m still seeing the market say that we are << 10% of losing 60% of the earnings potential from today.
I agree with you though on action – and it’s reasons like these that I tend to distrust relying too heavily on economic models to estimate tail risks for large-scale events with feedback loops.
— rortybomb · Oct 1, 10:35 PM · #
pfft
and you wonder why you antiques are losing the youth vote.
Matt won’t even let us have a little fun.
>:(
but….this is not my domain, but it seems to me that driver is not the market…but liquidity and flow in the credit markets.
i think we can sustain market swings…..not frozen credit.
— matoko_chan · Oct 1, 10:39 PM · #
Rover:
I agree that your assumptions are defensible. As I said, I think it would be hard to have reasonable assumptions that put p < 0.1. Let’s say this is correct. Don’t you think a 10% chance of a another Depression is worht a lot to avoid? (The question of whether this plan is the best feasible way to try to avoid it being a separate issue.)
rortybomb:
Yes, this analogy has occured to me as well.
I guess if I thought there was a 25% of AGW causing a 1/3 reduction of US economic output over the next 10 years, rather than a <1% chance 100 years from now, I would be the biggest carbon tax advocate on earth.
Think of your B-S numbers this way. A 10% chance of a 50% loss (from Monday, I assume), given the losses form last July’s peak, means about a 60% bear market. The only example of this that I know of in the 20th century was the 1928 – 1932 bear market. In order to place it in the company of the four big 40% decline over two years bear markets I described, you’d have to have only about another 20% decline from here. I think that the B-S number for that would be way, way higher than 10%. Further, I think there is a huge problem to applying B-S to this situation post ante, if we believe there is a regime change if there is no bailout. So I don’t necessarily see this as contradictory evidence.
— Jim Manzi · Oct 2, 01:25 PM · #
Jim, you said that your analysis was independent of the type of bill passed, so I have to wonder: If Rover is right, and everyone assumes that some sort of bill will pass, and indeed, that such a bill is in fact a near-certainty, then what are we worried about?
I would guess—caveat, caveat—that most of your correspondents were not objecting to any sort of bill per se, but the sense of urgency behind the bill’s advocation. “We’ve got to do something!” It seems your analysis is contingent on the proposition that success of failure of an immediate passage of some sort of bill is what will make or break the economy. That Times article you linked to claims a 3000 point drop in the DOW could happen in days, but I see nothing in your analysis to support that claim.
So if we have more time than the “Act now or die” folks are claiming, doesn’t taking the time to deliberate make sense, to avoid the fallout of passing a potentially disastrous bill too hastily? It seems like there are some competing risks here.
— Blar · Oct 2, 02:04 PM · #
Blar:
There is clearly a balance to be struck between time for deliberation to improve the bill and the increase in time at risk and decrease in estimated chances of getting anything done. I don’t think there is analysis that can realistically quantify this trade-off (unless you had really good data on the propagation of credit squeezes through the financial system, which it is possible Paulson has). I described this in a post about 10 days ago as the most critical prudential judgment that faces policymakers right now. I think if things get passed this week, it’s fine. If it looked like it was going to take until after the elction (remember only a little more than a month from now), the time and uncertainty (given that we don’t even know who would win what elections, who would be a lame duck, etc.) would be so great, that I think it would be functionally equivalent to “no bill”. Where between this Friday and Novemebr 5th the break-point is, I don’t know.
— Jim Manzi · Oct 2, 03:50 PM · #
Jim, I love your writing, but this post strikes me as pretty pure guesswork. For the reasons Rover says, and also for the many lack of parallels between the 1929 and today. The government will eventually pump both liquidity into the banking system and money into the larger economy, that is certain. By January at the latest, right after the inauguaration. The question is whether we must pass this package this week to avoid Great Depression II. I don’t see how that case has been made.
The real evidence people are pointing to is not the market drop, but the credit market problems. Yet although the interest rate bump for private debt due to the crisis is clear enough, lending is continuing. I’d prefer to see stats on how much money is going out through credit channels rather than guesswork about Dow drops and their connection to the economy.
— MQ · Oct 2, 07:03 PM · #
I see you addressed some of my points already in your response to Blair…thanks. Teach me to read before I post!
— MQ · Oct 2, 07:05 PM · #