Sunday, July 31, 2011

Beer: The Solution, and Cause, of All Our Problems

In the long run, the USA is in the same situation as Greece but in the short run all the haggling over the debt ceiling seemed inefficient, the stakes a bit too high. Here's how it went down in my state. The Democratic governor and the Republican controlled legislature could not come up with a budget deal and the state government actually shut down for two weeks. At first, all I noticed was that construction sites were inactive, and you couldn't buy lottery tickets (?). Over the first two weeks the sides seemed only to stiffen their resolve. Then there was this (Star Tribune 7/13):
The Department of Public Safety told the brewer it must stop distribution in Minnesota and devise a plan to pull its product from the shelves, including Coors, Coors Lite, Miller Lite, Miller High Life and 35 other name-brand beers. That would decimate choices for consumers.
On June 30, one day before the government shutdown, the company received a letter from the state that its brand licenses had expired. State employees who would typically renew those licenses have been deemed noncritical during the shutdown and laid off.

That got people talking! The next day, this
The governor of Minnesota and the state’s Republican lawmakers announced on Thursday that they had, at last, reached a deal on the state’s budget, bringing what is expected to be a swift reopening of government services.

I can report that beer access is now safely back to its pre-crisis state. The key to ending an impasse is to have something really annoying but not catastrophic happen while the impasse occurs. Default was the current topic in Washington, but this is pretty severe. As was done over 2000 years ago in Lysistrata, a crisis should focus on something both more and less pressing at some levels, in this case have a clause that says 'no more beer until a compromise is reached.' That would provide the same incentives but not affect our credit rating.

Wednesday, July 27, 2011

Rating Agencies Causing the US Debt Crisis?

The Wall Street Journal's Holman Jenkins Jr. see a problem in Rating Agencies (Moody's, S&P, Fitch) actually anticipating default:
But now we have a new problem. The rating agencies, especially Standard & Poor's, have decided to join the politicians in turning an artificial crisis into a real one. S&P says it plans a U.S. debt downgrade, regardless of any debt-ceiling outcome, unless it sees a "credible" plan to reduce future deficits by $4 trillion over the next 10 years.

This has become the real worry for Wall Street, but why? America's spending debate does not remotely make it any more of a default threat than it was a week or month or year ago. America's IOUs are still completely acceptable to the markets.

Rating Agencies generally rationalize market opinion the way politicians rationalize whatever talking points are popular with their base. They aren't leaders, rather focal points who artfully persuade many who don't follow or trust markets, that the market is generally correct. When I worked at Moody's, we took a beating from KMV because this firm showed that their purely quantitative 'distance to default' outperformed Agency Ratings. A distance to default metric is basically a simple function of market cap (higher is better), volatility (lower is better), and debt (lower is better), which is basically a market cap metric (volatility and market cap are positively correlated). All I can say is that the ratings agencies said they ratings were not comparable, and then Moody's bought KMV and such embarrassing information is not mentioned much (yet never repudiated).

So, ratings agencies are symptoms, and when things trade at AAA they get those kinds of ratings. As the markets are now questioning the AAA rating of the USA, it's only reasonable the Ratings Agencies too would ask these questions. Jenkins remarks that
There's a reason the monumental unfunded liabilities of Social Security and Medicare are not counted as part of the federal debt—they don't have to be paid

Which is what many said about the obligations of Latin American before the 1980 debt crisis, or what Russian investors said before the 1998 default. They don't have to pay these obligations, but given the incentives to pay senior citizen and public employees, or foreign debtholders, I would bet on them defaulting. Debt payment is based on priorities, and often governments choose to pay others before their debt holders.

For Jenkins to imply that rating agencies should, as a rule, wait until the market trades debt at a certain level to downgrade or even discuss a downgrade only makes the ratings purely redundant. Without the option to draw a line proactively, the charade is over and Ratings Agencies lose whatever meaning they have.

Tuesday, July 26, 2011

Trust Your Intuition

I commented on Amartya Sen's book the Idea of Justice, which while I found it generally unpersuasive still had a few gems. For example, it is interesting that throughout history leaders unsure of the grounds of their actions have found reticence appealing. As Lord Mansfield, an 18th century English judge, advised a newly appointed colonial governor:
consider what you think justice requires and decide accordingly. But never given your reasons; for your judgement will probably be right, but your reasons will certainly be wrong.

I find this a very wise course of action for leaders, because the last thing they want to do is get mired in a debate where their opinion is analyzed just like any other. Best to let the subordinates make their case, hope for the best, and move on. Groups need people in key positions to make final decisions, even if flawed, and while they should be held accountable, that should be done at a meta level, not at every step in a decision making process. Thus with hindsight I now appreciate the wisdom of some very wealthy leaders I have worked with who would say little, if only on this tactical point.

Another interesting point brought out by this line of reasoning is that judgement has a higher chance of being correct than the reasoning. I find this true too, as Hayek noted many traditions, ethics and mores come down to us proven by their fruitfulness to believers. So, the institutions of freedom were not established because lawmakers foresaw the benefits they would bring, that would come later, when Smith noted the nonintuitive connection between selfish and collective interest, and when Hayek outlined the importance of decentralized incentives. Hayek wrote a lot about how much of what we know that is useful is impossible to articulate. It's provisional knowledge to be sure, but proven via its pragmatic value as opposed to rhetoric. Indeed, rhetoricians are often quite good at debating any side of an issue, often choosing what is most likely to be favored, highlighting this skill is not so useful in finding the truth as it is in persuading others one should be a leader (at which point, one should then become reticent, see above).

The Gettier problem is an epistemological problem introduced in a three page paper by Edmund Gettier that now has a Wikipedia page longer than three pages (showing what philosophers love to do!). It's best given by an example:

Farmer Franco is concerned about his prize cow, Daisy. In fact, he is so concerned that when his dairyman tells him that Daisy is in the field, happily grazing, he says he needs to know for certain.

Farmer Franco goes out to the field and standing by the gate sees in the distance, behind some trees, a white and black shape that he recognizes as his favorite cow. He goes back to the dairy and tells his friend that he knows Daisy is in the field.

The dairyman goes to the field and finds Daisy having a nap in a hollow, behind a bush, well out of sight of the gate. He also spots a large piece of black and white paper that has got caught in a tree.

Daisy is in the field, as Farmer Franco thought, but Franco's reasoning was faulty.

In this case Farmer Franco was correct that the cow was safe and had evidence, it was a Justified True Belief in his mind. He knew the truth but did not have 'knowledge', rather, luck.

The problem is that the believer (Farmer Franco) was right for the wrong reason, but this turned out irrelevant. It's likely that in many such cases, his reasoning was only a confabulation for a deeper intuition about how Daisy behaves, and so, it was not pure luck. Epistemologically the problem is that often beliefs are not based on 'if and only if'--aka necessary and sufficient--conditions, so your beliefs are often right for wrong reasons, and so you don't really understand not only those beliefs you have that are wrong, but those that are right as well. As Keynes said, right policies are invariably chosen for the wrong reasons, so one can't too worked up about the fact that 'someone on the internet is wrong.' Given any big debate has two sides (eg, raise or lower taxes to increase welfare), you have only a 50% chance of being correct. As there are only a few out of many plausible reasons for having the right belief, odds are you are wrong about your reasoning more than your are wrong about what to do.

Reason is to man a great gift over and above the raw instinct and emotions we share with the great apes. It should help us find better solutions faster, which is best demonstrated by our technology and increased life spans. But we shouldn't trust reason and rhetoric too much, as traditions and instincts have wisdom too.

Monday, July 25, 2011

Due Process in Practice

From a letter to the editor defending unions:
Unions are blamed for protecting the jobs of inferior teachers, causing economic collapse of government and electing corrupt candidates with their "secret" donations. Wrong, wrong and wrong. Unions ensure the rights of teachers being considered for firing to due process guaranteed by the Tenure Act, an act Republicans are eager to dismantle so that experienced, well-educated teachers can be replaced by cheaper, younger ones...

Here's how mob hit man Frank "the Irishman" Sheeran described it in I Heard You Paint Houses:
The Teamsters gave me good job security at Food Fair. They could only fire you if they caught you stealing. Let me put it another way, they could only fire you if they caught you stealing and they could prove it.

Failure at the micro level is essential for success at the macro level, whether it's the development of the brain, a company, industry, or economy. Regulations ensuring due process are why the documentary Waiting for Superman noted that in most major cities the teacher firing rate is 1 in 1000.

The growth in regulations surrounding development, education, health care, and finance, is hard to quantify, but I think they are really affecting things now. I know a CFO of a public company and he said they aren't developing in California because of all their regulations and obligations. Unfortunately, unlike taxes, regulations are much harder to scale back because they are so insidiously constructed, which is why it's very hard to turn around retarded countries like Greece and Spain that guarantee their workers so many things.

Saturday, July 23, 2011

Risk is Like Sex Because...

... its payoff depends more on context than quantity. The same investment, or sex partner, has a different value to different people. Investments that pay off have a synergy with an individual investor's particular human capital, which is why finding alpha is just as much a process of self-discovery as it is of raw ambition. If you simply allocate your money into objectively risky investments you will lose money because in general investments are like banner-ad business opportunities that play on people's naive hopes to get rich.

Obviously sex has different value depending on what, when, where, and especially who. Consider Jack and John, who have different utility payoffs of sex with the same people, sometimes even having negative value in certain cases--the sign in our utility function flips. While almost all humans enjoy sex and it is necessary for our species, the amount of sex in equilibrium is the result of millions of individual, carefully chosen transactions. Doubling it would not be an improvement even though everyone talks about how much they like it, though clearly there are institutional and ethical norms that could improve the average person's sex life. In general, sex is essential and omnipresent, but its payoff is not linear in some universal quantity metric.

Risk is also contextual. For John, who has a plausible strategy and some discipline, day trading may be a great idea, but for Jack it just churns money. CNBC-type chatter that discusses investment strategies, and companies offering trading programs, primarily churn money too, highlighting that the vast majority of information related to investing encourages bad investing (if it worked they wouldn't give it away for anything less than hundreds of thousands of dollars). For some investments like investing in broad ETFs, the returns are the same, yet these are exceptions, not the rule. Risk taking, like sex, is essential and omnipresent, and its payoff is not linear in any universal quantity metric either.

The standard theory is that risk--properly defined as the omnipresent, nondiversifiable, priced risk---is something that generates a linear payoff for everyone. If you take 2 units, you will receive 2 units of the risk premium regardless of who you are. Thus, if you think the equity risk premium is 5%, then investing in stocks with twice the equity beta should generate a 10% annualized premium, and stocks with a 0.5 beta should generate half the beta. Properly defined, it's all about the quantity of the dollar betas at work. After 45 years of intense search, no one has found a beta that has this property.

Consider that for equities, future returns are negatively correlated with inflows, so that on average investors are poor at market timing. Further, stock substitutes have the patina of a risky asset that has demonstrated a positive return, though in general they are sucker plays. This makes the simple arithmetic average of the S&P500 a best-case scenario for your average investor, who also face taxes and significant transaction costs when they trade a lot as they often do.

Just look at the total returns to the ProShares ETFs, highly popular trading vehicles that allow one to short and lever all sorts of sectors. For 90 funds with data since Jan 2010, the average return through July 2011 has been -4% while the boring SPY was up 21%. When you look at all the ways to invest, people are adversely selected by hucksters as they always have and always will. Risk takers need a realistic contingent strategy (including exit and a way to lever that strategy if it works). That's quite different than saying that once you've identified risk, anyone and everyone will get the same linear return based on how much money they throw into it.

Wednesday, July 20, 2011

Amartya Sen's Justice

I was reading The Idea of Justice by Nobel Laureate Amartya Sen, now in paperback, and he starts out with the following example.

Take three kids and a flute. Anne says the flute should be given to her because she is the only one who knows how to play it. Bob says the flute should be handed to him as he is so poor he has no toys to play with. Carla says the flute is hers because she made it.

Sen argues that who gets the flute depends on your philosophy of justice. Bob, the poorest, will have the support of the economic egalitarian. The libertarian would opt for Carla. The utilitarian will argue for Anne because she will get the maximum pleasure, as she can actually play the instrument. Sen states there are no institutional arrangements that can help us resolve this dispute in a universally accepted just manner.

This supposedly shows that there is no single theory of justice, rather one should look at enhancing the redistribution of life-saving goods and removing 'injustice'.

I thought his initial thought example rather curious. Instead of asking how to allocate the flute between the three children, why not ask first under which rules would the flute have come into existence? If Carla knew she would not get the flute, she would not have made it. Therefore, just add a time dimension to the puzzle, and there's no puzzle at all: only a libertarian form of justice is consistent with the flute existing.

I was rather surprised that Sen didn't address this, as if goods and technologies exist independent of our allocation rules, which we then choose between. It's a bit like John Stuart Mill, who reasoned that the laws of production are exogenous, but the laws of distribution are created and enacted by human choice. That's excusable for Mill, who wrote before the marginal revolution (around 1871) and the manifest failure of socialism, but for an accomplished social welfare theorist like Sen to make such an error highlights that experts are working within a paradigm, and a flawed assumption (ie, wealth is given) makes all the resulting argument lame.

Tuesday, July 19, 2011

Research on Relative Preferences

From an Newsweek article on winning:
An economist at the University of Bonn has shown that test subjects who receive a given reward for a task enjoy it significantly more if other subjects fail or do worse—a finding that upends traditional economic theories that absolute reward is a person’s central motivation.

Alas, they didn't give the full citation.

From the journal Science, an fMRI study on the effects of absolute and relative rewards:
Despite the importance of distinguishing the roles of absolute and relative income levels for subjective well-being, and thus for human decision-making, the underlying neurobiological basis of social comparison is not well understood...Similarly, context dependency of ventral striatal responses in humans has been demonstrated with fMRI...Our study introduces two critical new aspects of relative reward processing in the ventral striatum. First, it shows that social context is an important factor for reward processing. Second, in contrast to previous studies, the differences in reward activation cannot be explained by a mismatch between expected and received reward.

Another on social signaling:
Students in the award treatment were offered a congratulatory card from the organization honoring the best performance. The award was purely symbolic in order to ensure that any behavioral effect is driven by non-material benefits. Our results show that students in the award treatment outperform students in the control treatment by about 12 percent on average.

The unexpected implication of this research, which is growing, is that there is no risk premium. Ignore at your peril.

Monday, July 18, 2011

Why Elizabeth Warren Got Passed Over

Nietzsche said 'no one lies as much as the indignant do', which is why one should never try to reason with an angry person, as they have left their senses in righteous anger. Emotion is one of those curious things that seemingly short circuits our reason, yet, without it we all would be ineffectual and boring (Antonio Domasio writes that people with damage to the emotional part of their brains get stuck in indecision). As Aristotle cautioned, moderation in all things, including emotion.

Anyway, here's Elizabeth Warren, who was recently passed over for head of the CFPA, demonstrating why she was not up for the top position:

[The CFPB] still has enemies in Washington, D.C. And they have a plan...But this agency needs to have its full powers right now, and that means we need Rich in place as Director. Today, I'm celebrating -- but I'm not taking my eye off those who want to cripple this agency.

Now, every politician has opponents, but to publicly describe them as conniving, crippling enemies highlights a rather naive rhetorical style. While many in politics have ulterior or selfish motives, it's just more effective to give them the benefit of the doubt in public discussions because without superficial pleasantries like 'my good friend on the other side of the aisle,' it is too easy to get to the 'bad Nash equilibrium' where each side correctly assumes bad faith from the other side. Manners are not hypocritical, rather a way for us to become better people because pretending to be nice is often the first step to actually being nice.

I'm still confused by her statement that 'on Thursday, we will have cops on the beat -- making our first contacts with the 111 largest financial institutions in the country so we can monitor their compliance with the law.'

If you read the Dodd-Frank bill, you will see a lot of vague language that overlap other agencies (eg, the FRB, FDIC), so, anything new isn't obvious to anyone. I tend to think they will just like having everyone give them a bunch of reports to make them feel important, and then look for something obvious, but they will find obvious improvements rare because the status quo in our highly regulated financial sector has come about because of trade-offs between powerful interests. For example, will they argue for more mortgage lending to people with low credit scores to achieve diversity targets, or less to achieve safety?

When I was in litigation I was often indignant because I thought I was being unfairly persecuted by a vindictive person. I thought the charges against me were purely a pretext to keep me from starting my own business, which at that time was to start a fund based on going long low-volatility equities worldwide, and as I had consistently argued and applied this objective my entire professional life, preposterously considered off-limits as part of my confidentiality agreement with my old employer. Yet, each time I spoke or wrote to the court or my adversary while indignant I made statements that made my case that much harder for me, leading to the following costly insight: never communicate in anger with an adversary. This is because you aren't being fully rational, and so an opponent without good faith will cherry pick your flawed statements for a tactical advantage, which is why lawyers always tell their clients to shut up irrespective of the big picture.

Now, emotional pleas may generate sympathy and support from the masses, but often you aren't in a situation where people who elicit such sympathy matter. I sense Warren is trying to steamroll her opponents via PR, but the problem is that anything she does will pertain to details that don't matter to the public, such as how to trade off simplicity and some circumscription of liability. Simply casting banks as bad guys only works if your end game is nationalization, which even I doubt the Obama administration wants.

Sunday, July 17, 2011

Predicting Tsunamis

Prediction expert Phillip Tetlock and his colleague Andy Gardiner have a piece over at Cato Unbound, discussing the failure of experts to foresee this year's events:

Even now, only halfway through the year, The World in 2011 bears little resemblance to the world in 2011. Of the political turmoil in the Middle East—the revolutionary movements in Tunisia, Egypt, Libya, Yemen, Bahrain, and Syria—we find no hint in The Economist‘s forecast. Nor do we find a word about the earthquake/tsunami and consequent disasters in Japan or the spillover effects on the viability of nuclear power around the world. Or the killing of Osama bin Laden and the spillover effects for al Qaeda and Pakistani and Afghan politics. So each of the top three global events of the first half of 2011 were as unforeseen by The Economist as the next great asteroid strike.

Killing Osama, revolution in Tunisia, and the tsunami in Japan are all pretty specific outcomes that were unforeseen. But I would say that rather than this year being really crazy, it is rather similar to last year: US mired in Iraq and Afghanistan, deficits as far as the eye can see, no shovel-ready government spending, inner city America pathetic, Harvard outdoing Mississippi State academically, etc. Predicting broad trends is eminently doable but we often just take those for granted. Indeed, later they point out that experts tend to do worse than simple extrapolation models, which just highlights the futility of predicting outliers. So what should one do: predict specifics, or forecast broad trends that necessarily miss specifics?

When I worked as an economist I remember that the statistically optimal forecast was generally an exponential curve from where we are to the long-run historical average. But that's pretty boring, so we would add some little wiggles at the end based on some theory (A causes B which causes C in 18 months), because that kind of reasoning really resonated with the audience; they wanted to learn some new story to apply to their understanding of the world, something more novel than the future will be a lot like the past. Similarly, Tetlock appears to want tsunami forecasts, surely fun to have.

Later, Tetlock notes that statistical models outperform alternatives, but 'have an unfortunate tendency to work well until they don’t.' This, supposedly, is a deep thought. What Tetlock seems to lament is the absence of perfect foresight, because dominant approaches he finds poor. This isn't a very logical conclusion, because he is making the best the enemy of the good here.

Tetlock is fond of grouping experts into hedgehogs and foxes: 'the fox knows many things but the hedgehog knows one big thing.' The experts with modest but real predictive insight were the foxes, but sometimes not. In an interview Tetlock was asked whether he was a hedgehog or a fox, and replied, 'both!', as if everyone else would not also choose this description, highlighting the futility of his schema. Tetlock greatly admires Nassim Taleb, another who, ironically, criticizes pundits who make lots of nonfalsifiable predictions and selectively report their record. I think this highlights Nietzsche's warning that 'He who fights with monsters should look to it that he himself does not become a monster.' In other words, people who make a life out of criticizing overconfidence, dogma, hate, hypocrisy, are often filled with overconfidence, dogma, hate, and hypocrisy.

It's good to dislike error and vices, but to thirst too much for these points just highlights an intellectual diabetes.

Thursday, July 14, 2011

Trader Myths

One of the most misleading myths about traders was repeated in this Atlantic article:
Now think of what a trader does. A trader's job is to be smarter than the market. He converts a mess of analysis and intuition into simple bets. He makes moves. If his predictions are better than everyone else's, he wins money; if not, he loses it.
The article is a hagiographic view of traders, like writing that politicians are the best-of-the-best, constantly working on how they can best advance social welfare by seeing win-win exchanges mere mortals cannot. If you are a trader, you might want to send it to your mom or girlfriend.

Traders supposedly described in the article are not like day traders, more like bookies who make money off the vig, and have little insight about 'true value' but usually a lot of franchise value in their brand or network of contacts, which is why they aren't as smart as they would be if they were simply price-taking market wizards. Indeed, most of the first Market Wizards book naively conflated market making and speculating, but when the legend becomes folk wisdom, print the legend, because people obviously like comforting myths.

When traders buy and sell the same stuff right away, it's perfect, they make money without risk. When they have to inventory it, they need it hedged and they have to hope it was priced well, which shows up statistically in the inventory depreciation. Hedging is not insanely complicated for the risk factors you can hedge, and for those you can't you have understood limits (the Serenity Prayer is especially important).

Now for 'equity derivatives people', you have to take a collection of non-standard derivatives, price them, and add some profit to anticipate adverse selection and cover your fixed costs plus any transaction costs (eg, the cost of implementing necessary hedges). Consider the following equity derivative transaction described in the article:
"We want a monthly stream of 12% returns as long as price X stays above 95, a downside no greater than Y, and a coupon of 120% of the principal in the event that A, B, and C all happen on or before the close of trading on April 11th."

If you want a 12% return in this market, that's about 11% more than Libor. An 11% yield necessarily has about an 18% default rate in a risk neutral world assuming a 50% loss-in-event-of-default, higher if you believe in a risk premium. The structure he describes is like asking
I want an easygoing hetero Jewish blonde, 36-26-36 (or better), 700+ Quantitative GRE, hilarious, symmetric face, caramel complexion, small nose who loves sex, second-hand cigar smoke, and homebrewing.

Equity derivative traders are constrained by simple present values, the expected value discounted by libor, applying probabilities from market prices for options and underlyings. This will be the lower bound to any price quoted to the buyer; they can't create 12% yields without massive downside any more than a pool of subprime loans can create a AAA security without massive subordinate tranches.

Wednesday, July 13, 2011

Holder Wants Banks to Stop Discriminating

It seems they discriminate against borrowers with bad credit, which Eric Holder sees as massive civil rights violations. This kind of thinking was instrumental in Fannie Mae's influential MyCommunity Mortgage program and accompanying software that gave the Federal imprimatur to ninja loans. How could any regulator criticize loans its own agency was encouraging at Fannie Mae? How could any investor doubt loan criteria vetted by the historically low-risk Fannie Mae? They didn't. It didn't cause the crisis, but it was a leading cause.

Last month, the Supreme Court ruled that the Wal-Mart class action suit had to be refocused, but the opposing side said mere statistical disparity and anecdotes were sufficient and they only lost 5-4. If statistical disparities and anecdotes are proof of malfeasance it's hard to see any large organization not in violation.

A Department of Justice spokesperson named 'Xochitl' (demi-god spawn of Chthulu?) explained they are trying to make sure banks behave in methods that are sure to create a vast array of convoluted regulations, because it would be suicide for banks to not discriminate on the basis of creditworthiness, and impossible for this to not have disparate impact on historically disadvantaged minorities. From the IBD:

As part of settlement deals, prosecutors have required banks to sign "nondisclosure agreements" barring them from talking about the methods used to allege discrimination. Bank lawyers contend the prosecutors are trying to hide the shaky legal grounds on which the cases are built. "It's horrible what they're doing at the civil rights division," said Reginald Brown, a partner at Wilmer Hale in Washington, who has represented banks in connection to recent race-bias investigations. "They don't have any proof, just theories."

He added, "They want you to sign something saying you agree, under the condition of any settlement with them, that you won't disclose what their theories were. That's because their theories are loopy and wouldn't stand the light of day."

One such theory — "disparate impact" — holds that merely a difference in loan application outcomes is enough to prove racial discrimination — even if no intent exists on the part of loan officers to contrast based on the color of applicants, and even legitimate business factors — such as credit scores and down payments — help explain disparities in loan outcomes between white and black applicants.

And just in case you think I'm being paranoid about the race card being thrown around, consider this choice quote:
In announcing a recent $2 million settlement with Dallas-based PrimeLending, Civil Rights Division chief Tom Perez said, "We will require lenders to invest in the community that they've harmed."

Another Reno protege, Perez has compared bankers to Klansmen. Only difference is, he said, bankers discriminate "with a smile" and "fine print." He said this kind of racism, though more subtle, is "every bit as destructive as the cross burned in a neighborhood."

Tuesday, July 12, 2011

A Keynesian Concession?

Keynesian Paul Krugman, in his bi-weekly column arguing for more fiscal stimulus, gives us a new twist on his argument why we need to double down:
Everybody knows that President Obama tried to stimulate the economy with a huge increase in government spending, and that it didn’t work. But what everyone knows is wrong.

Think about it: Where are the big public works projects? Where are the armies of government workers?
So, not only has the stimulus spent $800B without any more stuff, it doesn't even hire more people! Alan Blinder makes a similar point--we have 500k fewer government workers at all levels--in spite of spending rising 16% over the past 3 years at all levels (from inflated 2008 levels, no less). Keynesians see that data and thinks it implies we need more spending, whereas others have looked and found that is what happens when you waste money on food stamps, Medicaid, and welfare. Private companies don't benefit from wasting money, and neither do governments.

Here in Minneapolis, we spend more on fixing roads, but they get worse:
In the third year of Rybak's five-year $19.25 million acceleration of street repairs, engineers rated the average condition of city streets at 70 on a scale of 100, down from an 82 average in 1995. The rating has slipped by a point in each of the first two years of Rybak's street-spending surge.

How is this possible? Milton Friedman called it Gammon's law: 'In a bureaucratic system, increase in expenditure will be matched by fall in production.'

Consider that the Omaha Public Schools used more than $130,000 in federal stimulus dollars to buy books on diversity.

The book says that teachers should acknowledge historical systemic oppression in schools, including racism, sexism, homophobia and "ableism," defined by the authors as discrimination or prejudice against people with disabilities...

The Omaha school board approved buying 8,000 copies of the book--one for every employee, including members of the custodial staff--in April.

With investments like that I'm not shocked we aren't seeing big increases in public works projects, employees, and especially productivity. The multiplier supposedly works independent of actually producing anything tangible, which is why I can't take this theory seriously. Now even Krugman admits there has been nothing created, which just means we should spend even more money, highlighting that the Keynesians will argue for more continually (like the series of ineffectual Japanese fiscal stimulus packages in the 1990's).

Monday, July 11, 2011

Why Everyone (else) is a Hypocrite

I loved most of Robert Kurzban's book by this title, as it's really similar to other books I liked in this genre, Geoffrey Miller's Spent or Jonathan Haidt's The Happiness Hypothesis. It's funny and profound. Kurzban gives the nice analogy that our conscious selves are basically like the press secretary for a large organization. In such organizations, press secretaries often don't know why their organization is really doing X, indeed, they might not want to know so they can maintain plausible deniability. Further, in large organizations, there often isn't a simple reason why they are doing X, as any leader knows, even the CEO can't control very much--he's not a dictator and if he tried to be one he'd be thrown out. Yet the press secretary must spin their story as well as possible, and such is the case when we try to explain our beliefs with inconsistent explanations, or explanations that conveniently avoid certain facts and exagerate others.

There are many really interesting points made in this book. For example, he points out scenarios where more information makes us unambiguously worse off. Consider if you were watching a house burn, and saw what may or may not be a little puppy in there. You are afraid of getting hurt in the fire, so don't want to save the puppy, but you appreciate a reputation for bravery, and so don't want to be seen as avoiding your duty to save the puppy. If a little boy comes up to you and says 'Hey mister, are you going to save that puppy?' you now have to make a choice between risking your life or your reputation, because the boy now made it clear to everyone that you know the puppy is there. Damn kid. This is an example of things 'I wish I didn't know now what I didn't know then,' and why people avert their eyes from evidence contrary to their thesis. If you really dug into your opponent's arguments, you might not like what you find, so better to just attack the dopes who are against you for obviously wrong reasons, and that way you don't have to deal with greater cognitive dissonance (eg, John Horgan defending Stephen Jay Gould without trying to understand the criticism or even look at work subsequent to 1850 on the matter in question).

Then there's the idea that we are optimally overoptimistic, because its optimal to present yourself as successful as plausible, because that makes others like you more (they anticipate your ability to generate more favors for them in the future), making you more successful. This also makes us see our successes as more due to our efforts, and our failures more due to bad luck or evil-doers, because we will appear more sincere in our overoptimism if we truly believe it. This was nicely put into academese by Albert Bandura: 'optimistic self-appraisals of capability, that are not unduly disparate from what is possible, can be advantageous, whereas veridical judgement can be self-limiting.'

Another point is that, while there are some things correlated with success like higher self esteem, merely raising your self esteem without anything else is like moving your gas gauge to Full when it's near empty, hardly an efficient solution. Feelings that are symptoms are not to be trifled with, but addressed more fundamentally.

As with many of these books, they spin out of control at the end. He also seems to think our principles are contradictory because of our mind's modularity. For example, he takes the idea that our minds are modular to explain why we both want to eat the cake, and diet, one hitting our short term pleasure, the other our long run interest to be healthy. I think this isn't modularity so much as our discount rates, and people with greater discipline forgo the quick route of indulging in leisure to get a long-term benefit. A lot of this can be explained as simply better understanding the true 'present value' of one's actions, as William James's pragmatism would argue. Further, every principle is only good in moderation, as honesty clashes with politeness at some point, and so we have to always balance them. The paradox of tolerance, than an unlimited tolerance must lead to the disappearance of tolerance, requires us at times to be intolerant. It isn't due to our modular minds, just the way in which any virtue must be applied moderately and in context, as opposed to absolutely.

As he wrote a book about the essential hypocritical nature of a mind that has to explain things it can't fully understand, that mainly wants to thrive as opposed to be correct, it should come as no surprise that he ties himself into knots as a liberal who loves liberty. He argues our morality is basically a pretext for self-interested reproductive success, conditional on our current situation. So, alpha males like promiscuity, lions kill the cubs in their new prides, etc. But he notes that applying moral rules to others infringes upon their liberty and is thus bad, using the idea that 'Rawlsian justice' then seems like the optimal solution. Alas, Rawlsian justice is radical egalitarian redistribution of wealth and even esteem, infringing on a lot of freedom of contract by other people merely because they happened upon having more stuff (the Rawlsian condition that forced redistribution should not hurt the least advantaged person basically makes it moot, because the least advantaged person in any society is a miserable wretch not helped nor hurt by anything). A Rawlsian utopian would equally reward, in all ways, the person who sits looking at butterflies all day and he who plays great basketball, the CEO and the janitor. This doesn't generate Sweden (prosperous and equal) rather, illiberal hell-holes like Cambodia and Russia at their worst. That Kurzban can celebrate liberty and Rawlsian justice is a nice example of the hypocrisy within us all (especially because it was not, it appears, intentional).

Friday, July 08, 2011

Behavioralists Think Everyone Else is Biased

Sendhil Mullainathan of Harvard won a MacAurthur genius grant in 2009, which makes him a celebrity among the New Yorker crowd who like to think they are on the cutting edge. He is also the Assistant Director for Research of the Consumer Financial Protection Bureau. He is a 'behavioral economist', a field supposedly generating what David Brooks calls golden age of behavioral research. I read Kahneman, Tversky and Slovic's Judgement under Uncertainty in the 80's (published 1982), which mainly discussed a series of papers published in the 1970s, and found it fascinating, but now it's now 30 year old stuff and pretty boring. There's a couple hundred academically based confirmed biases which are all kinda true, but not very profound. For instance, people tend to eat more in groups than when they eat alone. That's interesting, and has some implications, but they are pretty parochial. I suspect Freud had a similar popularity cycle, publishing around 1899, peaking in the 1950's, and now scientifically irrelevant although always useful for a rhetorical flourish ('penis envy', 'Oedipus complex').

In his Aspen Institute speech, Mullainathan recounts some familiar anecdotes, such as that default choices are very influential, that being distracted makes you less attentive. These are useful to be sure but hardly something to generalize upon. There aren't a lot of insights generated from the default choice bias, other than, for simple things like organ donation and 401ks, make those the default choice. These have been the staple examples since the 1990s, and they remain the primary regulatory examples.

Take something like irradiating eggs, something almost all scientific people agree is good (kills bacteria, no residual radiation). Why not make them the default eggs? They never would argue for that, because it's contrary to the Luddite-leftists who hate technology but love government. In practice behavioral insights are applied mainly to increase the scope and scale of government, which presumably is staffed by unbiased administrators. Consider than Al Gore cozied up to ethanol when he was trying to be President but now agrees it was counterproductive, and like the sugar cane quotas created a permanent policy that cost merely tens of billions in waste. Such are the results of presumably high-minded objectives, all twisted by the absence of competition. Thus, Mullinaith can only go to the 401k and organ donor example, but can't extrapolate to food irradiation or something else as simple (eg, don't encourage low down payment mortgages). Simple rational thinking doesn't generate better policy. He can prove me wrong with actually generating regulations that have positive benefit/cost ratios, but I'm not optimistic.

Sendhil also thinks that a major problem of the poor is that they inadequately sign up for welfare because the forms are so complicated. If you simplify forms you will have more people on welfare and more fraud, which isn't necessarily good for families in the long run. So, Sendhil likes to play to standard liberal pieties about how government can help and inequality is the result of evil and prejudice (he's big on the stereotype bias, another publication bias result gone viral). Yet he forgets the most profound behavioral bias in this literature that is delightfully recursive:
We think we are better than average at not being biased in thinking that we're better than average.

This is true especially for people who have studied this subject at college. Thus, the behavioralists neatly infer from the literature that everyone else is biased. Now, surely these are smart, educated people, and they are less biased on simple scientific questions that your average housewife doesn't know, but there are lots of big issues relating to politics and business strategy that don't have a monotonic relation between learning and the ability to discern the good. Just look at politics or education, areas subject to a lot of study over the past 50 years, which are hardly more efficient than before all this knowledge arrived.

Thursday, July 07, 2011

Disclosure is Nontrivial

I'm a fan of disclosure because I think it's one of those equilibrium like wearing hockey masks, where individuals may not want to wear one if no one else is because of marginal costs in eyesight, signalling toughness, but if everyone wears one they are all likely to have more teeth and functioning neurons.

Yet, it's a nontrivial problem identifying the data that might be useful. Good data tends to be very business-specific, different for credit card portfolios, direct vs. indirect lending, home equity, commercial real estate: each have different key risk factors. As previously, Moody's did not think it had to know what the loan-to-value ratio was on the mortgage collateral, and bank examiners also never mentioned this as a bad thing prior to 2007, one shouldn't be optimistic that the current bugaboos will be relevant.

Too often disclosures focus on things that don't matter, and they become fetshized. For example, many have pointed out that leverage for the largest investment banks was crazy-high in the last crisis, 25-1 in some cases. True, but they were high for decades for most of these investment banks, and cross-sectionally, the relationship between leverage and future failure rate or ratings is weak. It's a really weak signal because having 10-1 leverage in some assets is much safer than having 30-1 assets in another, and banks know this and game accordingly.

Another example of regulator focus is VaR, which is applied to a subset of traded assets within a bank. I'm a fan of VaR, having set up a VaR for KeyCorp's trading operations in the 1990's, and found it immensely useful as a monitor on our positions. Yet, for risk capital, or the risk of the bank in total, it was irrelevant. The assets relevant to this calculation are usually quite small as a percent of total assets (<5%), and often part of market-making operations that generally have pretty flat exposure to big factors like interest rates or the S&P500. Thus, in the past crisis the big 7 US money center banks showed an average daily VaR (99%) of about $120MM in early 2008, and then through March 2009 lost about $7T (trillion) dollars in market cap. Being of 1000x smaller magnitude highlights its irrelevance.

So, if you have a good idea, send a 3-5 page PDF version of your proposal or paper, with a separate title page and abstract, by October 15, 2011 to: Hamid Mehran ( and JAR ( specifying "FRBNY/JAR Conference" in the subject line ($50 submission fee). There's a conference the following year in NYC sponsored by the Fed and the Journal of Accounting Research entitled “Disclosure, Transparency, and Financial Reporting in the Financial Services Industry."

My idea would be to take the bank consultants at Oliver Wyman or First Manhattan, and certify them as our standard bearers. No penalties, just full reviews, with their own subjective criteria, which is as well explained and more rational than anything coming of the FRB/OCC/OTS/SEC/HUD/FDIC/NCUA/FTC. Sure that would give them a lot of power and a huge wealth transfer, but consider the billions spent on bank regulation that is so transparently pointless, we could do better for one tenth the cost.

Wednesday, July 06, 2011

When Priorities Dominate Effort

This video on why Russians are better wrestlers than Americans I found quite profound. In general Russians are known as being tough--think Siberia, the battle of Stalingrad--but in wrestling it's the Americans who are tougher, they work harder and are better conditioned. Yet Russians generally dominate American wrestlers.

Ken Primola argues that the Russians are better because they are more cerebral, they are more into technique. As I'm a coach for little kids, I see a lot of the high school and college kids, and know that in America they are taught that the primary attribute of a great wrestler is hard work. Your average great high school wrestler after a big loss will say 'I have to work harder', not realizing it's a cliche. Clearly, to be really good you have to work hard, and to be great, you need some natural athletic ability (see Jon Jones or Anderson Silva in MMA). But after a certain point, everyone has put in 10,000 hours, and so simply working harder has very little marginal benefit.

I agree that any excellence involves a good amount of disciplined effort, yet that only gets one so far. To really improve once at that level you need a better strategy, and that takes better higher-order priorities. Avoid the temptation to work so hard that there is no time left for serious thinking, in wrestling or anything else.

Tuesday, July 05, 2011

CFPA Still Fighting

While the Consumer Finance Protection Act seems like a no-brainer--who isn't for consumer protection?--the mission was recently bullet pointed as follows:

_ Created the consumer protection agency to oversee mortgages, credit cards and other financial products.

_Established a body of regulators to scan the economy for threats to the financial system.

_Required banks to hold back money for protection against losses.

_Curbed the trading of derivatives, speculative investments partly blamed for the 2008 financial crisis.

_Gave the Federal Reserve powers to oversee huge companies whose failures could jeopardize the entire financial system.

There's nothing particularly wrong with these objectives, but they are already covered by existing agencies. I'm all for replacing old objectives with new ones, but no one is proposing shutting down the old agencies, just adding this new one. Thus, the picture below is misleading, because it is merely adding another set of lines the the spaghetti graph at the bottom. I have found merely one 'dotted line' reporting function to be a disaster within large organizations, and with our financial regulation just about every law has multiple agencies enforcing it, leading to everyone and no one being in charge.

Spending $200MM (what Republicans want) or $329MM (what Obama wants) on this elephant highlights that US interests rates are way too low, because if we are still spending money on boondoggles as if The Multiplier will make it all pay off, we deserve to have our credit card canceled.

Monday, July 04, 2011

Mortgage Lottery Tickets

There appears to be a program that is targeting borrowers who haven't paid their mortgage in 3 months, have had a 15% loss in income, and live in their mortgaged property. As part of a $1B national program, 1500 lucky Minnesotans will receive a $50k no-interest loan after filling out a bunch of paperwork, and they don't have to pay it back at all if they then live in that home for 5 years.

This seems to have all the wrong incentives. Why pay on your mortgage if you have had a decline in income given this carrot? In lending, ability and willingness to pay are key variables, and it seems government is doing everything it can to make people believe those who pay are suckers. As traditional lending then dries up, leaving only FHA lending, the housing market remains depressed. Further, most people who don't pay will just make it harder for them to get loans in the future and won't receive the windfall, as it is decided via a lottery.

Recessions are bad things, but the key is that people need to find new lines of work where their services are needed, that will pay them what they think is their best wage. These programs simply slow this process down, as people hold out for the latest government give-away that encourages having lower income. Further, most people won't receive aid, so luring them into this complacency just makes them less motivated and ultimately poorer.

Sunday, July 03, 2011

Rotational Indeterminacy and Factor Analysis

A recent Bloggingheads has highlighted the 'Factor indeterminacy' problem remains a conspicuous if rarely highlighted item among IQ critics due to Stephen Jay Gould's Mismeasure of Man, which remains required reading for many college freshman. Gould set out to demolish all research relating to racial differences in IQ, arguing that neither concept exist, and noting that slave owners and Nazis also believed in these things (Hitler was a vegetarian too, but no one sees the obvious connection there). His main argument against IQ is the following (taken from his critique of The Bell Curve):
Spearman used factor analysis to find a single dimension—which he called g—that best identifies the common factor behind positive correlations among the tests. But Thurstone later showed that g could be made to disappear by simply rotating the dimensions to different positions...In any case, you can't grasp the issue at all without a clear exposition of factor analysis—and The Bell Curve cops out on this central concept.
Basically Gould argued that 'if you can't grasp the issue' you shouldn't have an opinion, as if there's some big math error at the bottom of IQ and if you can't do factor analysis you can't understand IQ. Meanwhile, IQs are remarkable stable over people's lives, belying the assertion they are arbitrary.

Arbitrage Pricing Theory (APT) proposed by Stephen Ross in 1976 generate a flurry of research in the eighties and 1990s, because it extended the basic single factor CAPM to include multiple factors. Such generalizations are popular among academics because they invite a lot of rigorous extensions, perfect for creating publications. These factors could include 'the market', but also something for 'oil', currency risk, and just about any bugaboo you can imagine. The basic idea is to estimate the equation


where r is an nx1 vector of returns over n stocks, and then use factor analysis to extract the factors (f) and the factor loadings (aka 'betas', b). In the CAPM, the factor is the excess market return, Rm-Rf, but this could be instead the vector of {Rm-Rf, Oil price Change}, etc.

Interestingly, there has never been a big worry about factor indeterminacy among finance professors who studied it so carefully. That is, say you multiply b by L, and f by L-1. This creates


Now, this simplifies to


So, you can do this with any nonzero L you want and the data are observationally equivalent. That is, you can say everyone had a low factor loading and the factor was really volatile, or the factor loadings were really large and the factor was rather quiet; these are observationally equivalent. Generally, you solve the problem by arbitrarily stipulating that E(ff')=I (the identity matrix), and then everyone knows what you are talking about.

There are ways to test for multiple factors, basically, by estimating k factors and their loadings in sample A, and show that the loadings from sample A explains more of the sample B variance! You can use likelihood ratio tests on k and k+1 factors models, but these I find much less convincing than the out-of sample performance. If you can identify an additional 'music' or 'verbal' IQ, or a 'non-market risk factor', together with their loadings, and then find this also explains subsequent performance on different samples, the world will beat a path to your door. Indeed, in finance people have left the latent factor approach, and chosen instead factor proxies such as 'value' or 'size'. Interestingly, the 'you can't measure the true market factor' issue-- a logical possibility just as possible as that g is unmeasurable--hasn't been a practical problem because either one discusses factors that others can also measure and thus have relatively unambiguous meaning, or you are irrelevant. There is no factor indeterminacy problem because people don't get bogged down on problems that have reasonable solutions (the Roll critique has been no cause for nihilism).

I found this contretemps interesting because it's an example of Euler's famous '(a + bn)/n = x, hence God exists' argument, a statement that seems technical and profound, but is pure squid ink, indicative of bad faith, especially in the context of Gould's assertion that the IQ research bullies people into believing results via abstruse technical arguments (ironic because that is exactly what he did here).

This came up because Jon Horgan was discussing his support for Stephen Jay Gould in light of criticism of his Mismeasure of Man, where one chapter focused on the work of a 19th-century physician, Samuel George Morton, who amassed a collection of almost 1000 skulls from around the world. Morton estimated the brain size of different racial groups by pouring seed and lead shot into the skulls. He concluded that whites have larger brains on average than blacks.

Gould argued that Morton applied an unconscious bias into his work by selectively including and excluding various skulls in his measurement. Recent researchers revisited the data, and found the reverse: Gould had a biased view of the data, not Morton. The issue centers on which of the 1000 skulls one includes in the sample. Read the paper, it's easy enough for a high school kid to follow. It is hard to see how Gould's selective sampling as opposed to the updated approach is more accurate.

Yet Horgan states that:
their analysis, far from being "straightforward," was highly technical and based on many judgment calls, as were those of Gould and Morton. The divergent results depend in part on whether to include or exclude certain skulls that could unduly skew estimates of brain sizes.
Now, I agree its not easy to put their critique of Gould into a single picture that leaves  room for no other interpretation, but if Horgan is going to state that the judgement calls in their analysis allows him to remain unconvinced then virtually no one can be convinced of anything, because the judgement calls were about as simple as could be (eg, include all skulls labeled 'American Indian' as American Indians and take the average).

Indeed, why Horgan defends Gould is mainly because he hates biological determinism (as stated, who is for that?), but also as he is now pushing the argumentative theory, that brains have evolved to persuade others rather than find truth. While I too think we all are more hard-wired to try and win arguments as opposed to find truths, I would add that the truth is still a good goal. It's simply more productive to have better theories underlying your observations than false theories, otherwise you end up arguing in circles trying to convince people that, say, 'the Hegelian dialectic' is the best way to explain history, which is possible but difficult because it's a crappy theory. Horgan has taken argumentative theory to rationalize his unwillingness to engage in analysis of the data simply because so many researchers are merely rationalizing their biases. He seems to have spent a life studying science and concluded that it's all a game, and he's going to play it as such.

The refutation of Gould is important because it highlights that those who are most strenuously against bias are often the most biased, just as those who say 'it's not about the money' really mean it's all about the money. For Gould this was just the tip of the tendentious iceberg, as he ignored all the contrary and more sophisticated data since Morton on brain size, race, and IQ, that really makes Morton irrelevant (see here, here, here, here, here, here, and here for some more updated analyses with references), and such errors of omission are most disingenuous because they reflect a strategy of avoiding the opposite's best arguments and instead attacking straw men. He ignored the fact that the higher the g loading a subtest the higher is its heritability, the higher its relation to elementary cognitive tasks like reaction time, and a lot of other such data.

It's possible to maintain a belief as long as the evidence is not common knowledge--something outsiders know you know they know that you know they know etc--because as long as it's not common knowledge, there's a chance you could be reasonably wrong. I understand that my prejudices are the most important determinant of how I see the world, but instead of simply accepting that, I try to tackle my adversary's best arguments as opposed to avoid them via the "it's complicated" dodge.

What we see is very theory-laden, we see what we believe. It's not only good to have priors, to have prejudices and beliefs, it's impossible not to. To have strong opinions, weakly held, is a good strategy in such an environment. Moderation in all things. I expect people to be biased, to tell lies, to occasionally betray friends with minor slights, but that doesn't mean I should do it unabashedly. Principles are what keep us from being totally uninteresting shills, in spite of the fact they are also unrealistic ideals.