Cochrane on Alpha-Beta

Quantivity glowing quotes John Cochrane:

I tried telling a hedge fund manager, “You don’t have alpha. I can replicate your returns with a value-growth, momentum, currency and term carry, and short-vol strategy.” He said, “‘Exotic beta’ is my alpha. I understand those systematic factors and know how to trade them. You don’t.” He has a point. How many investors have even thought through their exposures to carry-trade or short-volatility “systematic risks,” let alone have the ability to program computers to execute such strategies as “passive,” mechanical investments? To an investor who has not heard of it and holds the market index, a new factor is alpha. And that alpha has nothing to do with informational inefficiency.

Most active management and performance evaluation just is not well described by the alpha-beta, information-systematic, selection-style split anymore. There is no “alpha.” There is just beta you understand and beta you don’t understand, and beta you are positioned to buy vs. beta you are already exposed to and should sell.

I don't find this very profound. If the carry trade or shorting the VXX is a beta trade and makes a risk premium, investors should be indifferent to it. The risk premium is an even trade of premium for risk (or rebate for insurance). He's assuming that most people would love to earn the returns from the carry trade or shorting the VXX, and I suspect he's right, but that's because it's not a risk premium, rather, it's simply an opportunity.

Ever since the small cap effect was discovered, people have been attracted to it as an asset class because it offered higher returns. Dimensional Fund Advisors started with a small cap fund. Now, if the 'risk premium' story were true, it would be just as interesting for investors to take the other side, to buy insurance against whatever risk these things were providing a return premium for (these other factors tend to have medians of zero, not 1 as in the CAPM beta). Such opportunities are never sold that way, and investors don't want to pay for these things by shorting them.

Aristotle's Philosophical Conceit

Aristotle wrote many years ago an anecdote that many erudite people take as gospel, that the thoughtful sages could be rich if they really wanted to. Here he recounts the story of Thales of Melitus (ie, the primordial philosopher 624-546 BC):

he knew by his skill in the stars while it was yet winter that there would be a great harvest of olives in the coming year; so, having a little money, he gave deposits for the use of all the olive-presses in Chios and Miletus, which he hired at a low price because no one bid against him. When the harvest-time came, and many were wanted all at once and of a sudden, he let them out at any rate which he pleased, and made a quantity of money.

Now, this is total bull because the stars don't predict weather. Reading chicken entrails and the like was popular back then and they had lots of other wacky beliefs, but the idea that they traded on them and prospered is clearly a self-serving lie. Today, we have active mutual fund managers who claim to outperform indices even though this has never been the case, so I guess every age has its own myths.

Lewin's Love of Physics

Walter Lewin is an MIT physics professor, and wrote a pretty fun book entitled For the Love of Physics. It's a nice romp through fun facts of physics, such as why rainbows only appear when the sun is behind you. One of the more interesting parts concerns his investigation of Galileo's Square-Cube law.

Galileo's Dialogue Concerning Two New Sciences contained what he considered to be one of his most profound insights: the square-cube law. If two cubes are made of the same material then they will have the same density. Yet since the two cubes have different area to volume ratios they will likewise have different stress at the base of each cube. If too much stress is placed on an object then it will fail, or in this case a large cube has a much greater possibility of collapsing. This is why sandcastles can only be a few feet high.

Galileo applied this to animals, what we now call allometry, and noted that a this implies the diameter of bones should be proportional to their length so that length³=k•diameter², or diameter=k•length^1.5. The simple support required of bones implies limbs get thicker and thicker as animals get bigger, which is why rhinos and elephants are pretty thick. Bones comprise about 8% of the weight of a mouse, 14% of a goose or dog, and 18% of a man.

Also because of the Square-Cube Law, larger animals have less relative muscle strength than smaller animals. Both the muscle strength and bone strength are functions of the cross sectional area, while the weight of the animal is a function of volume. It is because of relative muscle strength that an ant can lift fifty times its weight while a human can lift an amount equal to its own weight, and an Asian elephant can only lift 25% of its own weight. The greater muscle to weight ratio of smaller animals is what allows them to jump higher than several times their own height, while at the other extreme an elephant can not even jump.

Back to Lewin, he actually looked at various animal bones from MIT's museum. Comparing a raccoon with a horse femur, he found the horse femur should be 6 times thicker than a raccoon's. It turned out 5 times thicker, which is close. Then he compared a horse to a mouse, and the thickness was only 70 times thicker, not the 250 times thicker as predicted by their lengths. The elephant femur was only 120 times thicker than a mouse's femur, not the 1000 times as predicted.

It appears that the structure of bones is different, bone chemistry changes along with its size. Thus, the fact that dinosaurs were the size of 12 bull elephants with leg bones of similar diameter, need not be explained by an expanding earth theory (the idea that the earth was smaller back then, so gravity wasn't as strong). So, Galileo's square-cube 'law' is really an approximation for the moderate spectrum of animal sizes, something that explains a lot but doesn't generalize like gravitation, which Newton used to explain the fall of an apple and the orbit of the moon.

Then again, perhaps even gravitation does not scale. Dark matter was introduced to explain the fact that galaxies rotate in a way very unlike our solar system. In our solar system the gravity vs. centripetal force generates the pattern where the period of a planet (T) and the mean distance from the sun (R) are related by a constant ratio for T²/R³, something observed by Kepler then proved by Newton 70 years later. Thus, Mercury has an orbit of 88 days, Neptune 165 years. Galaxies don't do that, with the outer regions moving almost as fast as the center stars. Dwarf galaxies are even worse, looking more like an evenly distributed swarm of bees, indifferent to clumping or spinning. The idea that the space is suffused with unobservable dark matter fixes this problem, but it's a fudge, because it 'appears' only as a solution to this problem. Perhaps gravitation doesn't scale at that dimension.

Chance, Effort, and Ability

Here's my local paper's Sunday major opinion piece on wealth inequality, discussing a researcher's model of wealth distribution:

He began his research with a simple question: Can chance alone account for wealth concentration?...He assumed that all entrepreneurs began with equal wealth. Returns varied, solely by chance...I'll spare you the calculus, but according to Fargione's model, by the "inexorable effect of chance," and chance alone, "a small proportion of entrepreneurs come to possess essentially all of the wealth...According to Fargione, greater variation in rates of return hastened the concentration of wealth.

That's not a model, that's an assumption. He assumed individual wealth varies randomly, and found the net inequality will be due to randomness. I understand that assumptions drive models, but the step between assumption and result has to be a little subtle or non-obvious. Here, he assumed everyone varied by randomness, and after doing this in Excel (really), it implied variation by chance is really random.

While it's important to remember that assumptions aren't models, they are perhaps more important, because as Darwin said,

False facts are highly injurious to the progress of science, for they often long endure; but false views, if supported by some evidence, do little harm, as every one takes a salutary pleasure in proving their falseness.

People who get excited about wealth being explained solely by effort or chance are making an important assumption about navigating one's life. Success is the result of randomness, effort, and ability. If you omit one of these, you will be miserable. A lot of growing up is about finding what you like that you are good at, and usually you like things you are relatively good at. Then practice that skill until you become excellent at it. The rest you can't really worry about even though that too is important, especially in explaining things like why certain people are really rich, which is often being in the right place at the right time. This should make us content because it's all we can control.

Anxiety should be not be ignored, but seen as what incents us to do our best--to observe the Serenity Prayer--because we worry all the time if we are doing our best given an uncertain future (in the end, it's the doers that prosper). For the existentialists Kierkegaard and Heidegger, this anxiety is the essence of consciousness (or sein), because we exist in time and are always thinking about an uncertain future in a way animals do not. They had very different solutions to this problem, and while I tend to find Heidegger's solution more fruitful, it clearly has more potential downside (Kierkegaard chose faith in God, Heidegger became an enthusiastic Nazi).

It's sad that some people see the disparities in income, and think this is all effort or all luck. In any case, this is a more damaging belief for their own self-actualization than any silly tax policy they envisage.

Real Investors Lag Indices by 6%

I recently discovered there is considerable amount of data documenting how much investors underperform indices, and it's being ignored. In the 2008 Journal of Pension Benefits, N. Scott Pritchard documented that that individual investors have done much worse than the indices that everyone assumes reflect investor returns. He looked at data from 401(k) plans, and found that from 1988 through 2007, while the S&P 500 returned 11.81 percent annually and Treasury bills returned 4.53 percent, the average investor achieved a return of only 4.48 percent.

Pritchard relied on the annual Dalbar study, which consolidates data from the Investment Company Institute, and is availble for investment advisers as a way to show them what the 'conventional wisdom' is on asset allocation and investor performance. More recent data found that over the twenty years ending in 12/31/2010, the average annual equity return for investors was only 3.27%, while the S&P500 was 9.14%.

So, this 6% investor underperformance you would think would be very interesting news, because the risk premium is one of the most important facts in all of economics, being the subject of thousands of research pieces, but instead it has had about zero impact. No one finds it interesting because it is not useful to academics or investment companies.

To put this into perspective, one of the most important research findings in twentieth century finance was when two professors at the University of Chicago, James H. Lorie and Lawrence Fisher, created what has become the preeminent database on stocks in the United States, what is now known as the Center for Research on Security Prices (CRSP) database.

The front page of the New York Times financial section heralded the pair’s findings, and their Journal of Business article reported the average of the rates of return on common stocks listed on the NYSE was 9 percent for the NYSE from January 30, 1926, to through 1960. Included in the NYT article was a flattering picture of them in a room with a big computer, emphasizing that this was very scientific. Interestingly, if you read their paper, you will see how academics bury their lede by noting that this final result is not conspicuous, highlighting that academics like to emphasize their technique, not the results.

Now, this was 3% higher than the average annual return on the Dow Jones index, but with dividends added, about the same, so why the big deal? It was important because while this old data probably still contains large amounts of survivorship bias, it meticulously corrected for all sorts of other issues and therefore added the patina of academic rigor to a fundamental financial constant: the equity risk premium.

This result stood for decades, and led to many to think the equity risk premium was on the order of 6-8% up to the internet bubble of 2001. It validated the new 'risk premium' paradigm that was being created in the early 1960s, where risk, properly measured, generates an observable return premium over time, because without higher expected returns no one would invest in risky assets. When combined with a stock's beta, it generated the expected return for each asset, making a seemingly qualitative problem one amenable to linear programming.

Much has happened since then, most importantly, the poor returns since 2000, which have reduced most estimates of the equity risk premium to around 3.5%. This excludes transaction costs, and adverse timing, which are captured in the Dalbar data. And of course this excludes taxes.

Happy Halloween!

Everyone likes to mock people who dress up their pets, but my daughter has been saying 'doggy Batman' over and over for the past few days, in pure delight over the sublime ridiculousness of combining a dog, a bat, and a man. Worth every penny. Boys have grown out of Star Wars, and are now 'scary/gross.'

Investors Underperform Indices

mistake (see above)

Shorting Leveraged ETF Pairs

ProShares offers a wildly successful array of Exchange Traded Funds (ETFs) that allows people to easily gain leverage and short targeted subsets of stocks. They trade a lot, and so clearly are satisfying consumer preferences, but probably the more delusional part of investor beliefs. Their explicit goal of targeting a daily benchmark return correlation leaves the secondary consideration less important: maximizing long term return. Thus, they tend to underperform their benchmark over longer durations, and this adds up. I suspect they end up burning money trading so much in a way that traders can anticipate and game.

If the Proshares are underperforming, there's a simple arbitrage here. Take the 'Ultras', which offer 2x leverage, and the 'UltraShorts', which offer the opposite, -2x exposure. Going short both stocks generates a very low-risk portfolio pair, because on a daily basis when one goes up 1% the other will almost surely go down about 1% by design; the positions will offset each other. But the drift for both is negative, as they burn money. Since ProShares has been very busy adding ETFs, this simulated strategy started with 23 pairs in 2008, and is now up to 45. Below is a graph of the total return to going short all the Ultra and UltraShort pairs offered by Proshares since 2008. I rebalanced every week. The annual return was 14%, and the annualized standard deviation was 12%.


Now, I am ignoring the short rebate, which for these may have been highly negative for some of these, but on average these have pretty meager short rates. As the S&P500 has a prospective Sharpe of 0.3 (excess return of 5% and standard deviation of 15%), this is a very dominant strategy. Notice that while the annualized vol is 12%, this really overestimates the risk here because most of this volatility is 'good': sometimes returns are much higher than average. It's a rather Madoff looking strategy

Another way to shade this is to notice that it works best during periods of high volatility, and among those pairs with the highest volatility. Notice that the October 2008 to March 2009 was a great time for this strategy, and so was August 2011, when markets were reeling.

Below are the returns to the various pairs I used, annualized. You can also use these pairs to simulate them yourself. Over time, I suppose these ETFs should start trading at a discount to their net asset value, but until then, it's a pretty simple strategy that seems to work.

Total Return to Short Pairs

pair1	pair2	AnnRet
AGQ	ZSL	31.8%
BIB	BIS	3.7%
DDM	DXD	9.0%
DIG	DUG	18.8%
EET	EEV	7.7%
EFO	EFU	8.0%
EZJ	EWV	6.6%
LTL	TLL	16.2%
MVV	MZZ	8.5%
QLD	QID	10.2%
ROM	REW	7.9%
RXL	RXD	5.2%
SAA	SDD	10.4%
SSO	SDS	9.6%
TQQQ	SQQQ	3.0%
UBR	BZQ	1.1%
UBT	TBT	4.8%
UCC	SCC	7.3%
UCD	CMD	6.5%
UCO	SCO	5.7%
UDOW	SDOW	5.3%
UGE	SZK	5.6%
UGL	GLL	8.1%
UKF	SFK	7.6%
UKK	SKK	12.8%
UKW	SDK	7.9%
UMDD	SMDD	5.9%
UMX	SMK	5.8%
UPRO	SPXU	4.5%
UPV	EPV	17.0%
UPW	SDP	13.8%
URE	SRS	51.8%
URTY	SRTY	13.9%
USD	SSG	10.2%
UST	PST	3.8%
UVG	SJF	10.7%
UVT	SJH	14.7%
UVU	SJL	20.1%
UWC	TWQ	2.5%
UWM	TWM	12.6%
UXI	SIJ	5.9%
UXJ	JPX	4.7%
UYG	SKF	30.8%
UYM	SMN	11.0%
XPP	FXP	7.1%

DailyKos on Good Intentions

I found this funny:

While Communists are certainly responsible for more deaths and misery than the Nazis could ever dream of, at least their intentions were good, so I'll give them a pass.

The entire article was filled with such observations, making me wonder whether the website was hacked. So, if you cause more death and misery than the Nazis, but have good intentions...it's ok? [I'm now told it was an attempt at sarcasm. If so, it's stupid sarcasm.]

Calomiris on Underwriting Problems

Charles Calomiris taught me Money and Banking in graduate school, and I thought then he was very wise. Here he is in today's WSJ:

In a painstaking forensic analysis of the sources of increased mortgage risk during the 2000s, "The Failure of Models that Predict Failure," Uday Rajan of the University of Michigan, Amit Seru of the University of Chicago and Vikrant Vig of London Business School show that more than half of the mortgage losses that occurred in excess of the rosy forecasts of expected loss at the time of mortgage origination reflected the predictable consequences of low-doc and no-doc lending. In other words, if the mortgage-underwriting standards at Fannie and Freddie circa 2003 had remained in place, nothing like the magnitude of the subprime crisis would have occurred.

Good Ideas Become Clearer over Time

Kant is known for his theory that there is a single moral obligation, which he called the "Categorical Imperative", and is derived from the concept of duty. Moral acts are those done in good will, which are done for the sake of duty. Duty is the necessity of acting out of reverence for universal law, something that you would want everyone in a situation to do.

Now, I find this reasoning rather flawed,* but my opinion on that isn't my point. In the Groundwork for the Metaphysics, of Morals Kant states that what he is saying is not the same as the Golden Rule; that the Golden Rule is derived from the categorical imperative with many important limitations. Many agree with Kant, such as radical egalitarians like Jurgen Habermas (see here), or John Rawls (see here). On the other hand, many argue that the Categorical Imperative is the same as The Golden Rule. Biologist and economist Peter Corning and game theorist Ken Binmore suggests that Kant's objection notwithstanding, the Golden rule is basically Kant's Categorical Imperative.

This seems like one of those ideas that is infinitely malleable, merely useful to give false authority to one's current pet idea. Habermas and Rawls did not want to engage in a debate on the practical issues of radical egalitarianism (its usefulness to tyrants, its impossibility, its assault on liberty), they preferred to simply defer to some famous philosopher's statement that is not evaluated merely by its consequences.

A good idea becomes clearer and more useful over time, while bad ideas become more subtle. Black-Scholes and Feynman diagrams are useful tools, taught to every beginner in finance and physics, because they explain things very parsimoniously, and because they are so clear can be extended or modified, which is the goal of every active mind. The invisible hand, the idea that inflation is ultimately a monetary phenomenon, that free markets decentralize knowledge and incentives, all simple, powerful, ideas. An idea's objective and quick decipherment enables us to avoid the systematic errors which invariably arise from prolonged entanglement. The longer we look at something vague and well-known, the more we qualify it to make it more sympatico with our prejudices.

In finance the risk premium started as volatility, became beta (covariance with the stock market divided by the variance of the market), and is now a covariance with some undefined set of proxies for our happiness (too be uncovered by powerful econometric techniques really soon). The 'risk premium' is a bad idea. Taleb's 'Black Swan' applies to anything unexpected, and as every specific outcome is in some sense unexpected, it applies to everything (except finance, says Taleb, which is ironic because presumably his 'buying cheap options' strategy supposedly reflects the profundity of his approach). He now says 'Ideas come and go, stories stay,' which makes about as much sense as anything else he says.

* I agree with Nietzsche that duty is not obvious, and often some self-serving platitude for some powerful interest, so it isn't helpful to state that something that is a duty is best. Further, it makes no sense to ignore context when applying a universal law, as when you should lie to keep Nazis from finding Jews in your basement, or when you kill a dangerous burglar, and so universal law is not obvious. Lastly, as Ayn Rand noted, ignoring the consequences of actions, and just focusing on the duty, is irrational, because we act to make things in a real world that we might as well believe really is real. Finally, I should note it is rather circular in practice, because in the end the 'universal law' is defended on utilitarian grounds anyway (eg, it will make society happier).

Haidt on the Moral Foundations of Occupy Wall Street

Happiness author Jonathan Haidt on the Occupy Wall Street/Tea Party difference:

We really hate cheaters, slackers, and exploiters. By far the most common message I saw at OWS was that the rich (“the 1 percent”) got rich by taking without giving. They cheated and exploited their way to the top. As if that wasn’t bad enough, we the taxpayers then had to bail them out after they crashed the economy, and so now they really owe us for saving their necks. It’s high time that they started giving back, paying what they owe.

As a point of comparison, a similar look at signs found at the Tea Party rallies suggests that protesters there are also chiefly concerned with fairness. The key to understanding Tea Partiers' morality, though, is that they want to restore the law of karma. They want laziness and cheating to be punished, and they see liberalism and liberal government as an assault on that project. The liberal fairness of OWS diverges from conservative and libertarian fairness in that liberals often think that equality of outcomes is evidence of fairness.

Those who think the market is generally fair and rewards virtue, who think that unequal ability is primarily from effort, discipline, and finding one's niche, believe in markets; those who think the market is generally a rigged game that rewards vice, that people are basically equal and become unequal mainly through forces beyond their control, believe in greater government control. Equality of outcomes is justice in one case, injustice in the other. Given these different assumptions are responsible for the most pressing political disagreements we have, and these are rather factual statements, the nice thing is that someday there may be more agreement on politics.

Bank Lending Dilemma

In the Financial Times, Larry Summers argues lenders should pay more for past bad loans, and also make more now:

First, and perhaps most fundamentally, credit standards for those seeking to buy homes are too high and rigorous.
...
Surely there is a strong case for experimentation with principal reduction strategies at the local level.
...
Fifth, there were substantial abuses by financial institutions and almost everyone in the mortgage industry during the bubble. Just compensation to the victims is a legitimate objective of public policy. But allowing negotiation over the past to dominate present policy creates overhangs of uncertainty that impose huge costs on the financial system and inhibits lending.
...
Bank regulators could facilitate inevitable restructuring of underwater mortgages by requiring banks to treat second mortgages and home equity loans in realistic ways.

Summers seems to recognize that punishing banks hurts new lending, but he also thinks some form of bank punishment would be just. Until they get over this and let the banks alone, lending will remain weak because banks are wary of the lookback option being giving to borrowers who 'bought' houses without the means or willingness to pay it back. In the US, loans are already 'non-recourse', meaning borrowers can walk away and let the lender eat most of the loss, but people want 'just compensation', which is a code word for an expropriation from banks to NINJA borrowers.

Until regulators, legislators, and the experts that advise them stop hounding banks for their old home loans, new home loans won't be forthcoming. It's all good and well to say we should just nationalize home lending, but if public housing is any guide that's a disastrous endgame.

Obama's latest housing effort seems like the last iterations (Hope Now, Hope for Homeowners, the Home Affordable Mortgage Program, the Home Affordable Refinancing Program, the Hardest Hit Funds), but until things like the Department of Justice's lawsuit against banks gets cleared up, banks won't consider homelending anything but toxic.

Short Powerful CEOs

A recently published article in the Journal of Finance by Morse, Nanda, and Seru argues that if you generate a metric of CEO overreach, their stocks underperform. They define CEO malfeasoance using the example of Home Depot CEO Robert Nardelli, who in 2005 changed his incentive pay to be based on average diluted earnings per share, from the tota return to shareholders over the prior 3-years compared to their peers. This was very convenient to Nardelli, because he did much better on the new comparison over the old, and it abrogated the prior performance contract; a lookback option, as it were. Nardelli presided over a 6-year period (2001-07) where the S&P500 was up 12%, Home Depot lost 17%, and Nardelli pocketed a $240MM for his stewardship.

The researchers construct three different metrics of CEO power. One is whether he is also President or Chairman of the Board. Secondly,they uses insider ownership, the amount of stock owned by the directors. Lastly they capture the percent of the board appointed by the CEO. They use regression analysis to find that firms with high CEO power face a 4.8% decrease in firm value going forward.

It would have been nice if they put this into a long-short portfolio and showed the portfolio return characteristics. As the data covered the infamous tech bubble (1992-2003), a lot could be explained by the rather singular 2001-2 tech bubble, which while interesting, is much less interesting than if this result was more persistent across time periods. In any case, another reason to read proxy statement footnotes, and hopefully people will invest on this information which would be the best way to regulate it. Home Depot got what they deserved, those responsible, shareholders, suffered most.

.

Aggregate Supply and Demand Nonsense

One characteristic of Keynesian thinking is to think the problem currently is with inadequate Aggregate Demand, as if this is some simple analytical construct that is just as meaningful as the demand for apples. This is nonsense. Aggregate Demand and Aggregate Supply are incoherent constructs that require heroic assumptions. You might as well talk about the current law of motion on the Hegelian Dialectic, which for decades was discussed as if it were real. Things exist before people know they exist (eg, nations, atoms), and things also don't exist even when many are certain they do (eg, anthropomorphic God, aether, phlogiston). About what one can not speak, one should remain silent, and you can't talk about something that is as logically vacuous as aggregate supply and demand.



In partial equilibrium, a good exists and its price represents its output relative to innumerable other consumer wants. This is why the demand curve slopes downward, because the more it costs, the more you have to forgo of other stuff. Demand curves are driven by consumer utility (which decreases as one consumes more of a specific good), and income (which shifts it about). For supply curves, the correlate to utility curves are cost curves. One generally produces where marginal cost equals price, and so marginal cost is increasing (if it were decreasing, you could increase supply and lower costs for a 'given' price). Thus, you only raise your output if the price rises.

You can assume demand and supply move separately, as when seasonal harvests of perishable goods arise in supply, or demand for cranberries increases before Thanksgiving. The logic of supply and demand curves in markets is firmly based on utility and cost functions, and is a very useful way to think about things.

Now consider Aggregate Demand. Here the 'price' on the vertical axis is not a relative price, but rather an absolute price level, which by itself is meaningless. If there was only one good in any economy people cared about, what would its 'price' even mean? So, right off the bat, something's fishy. Supposedly, in normal times the AD curve slopes down, we think, because other things equal a higher price level increases the demand for money, which drives up interest rates, which reduces investment and spending. But how does one increase the price level and leave 'other things equal?' One can imagine doing this in partial equilibrium, but it's a strange thing to contemplate over all goods. Further, at interest rate levels like today, it's not as if lowering interest rates is having any effect on investment (the liquidity trap, which is occurs always in real time for Keynesians, who always say this is why government spending is necessary now).

Then there's the 'Pigou's wealth effect', which affects wealth by changing people's real balances, because presumably their cash levels are constant but magically prices move, affecting real wealth, CashValue/Price. However, offsetting that is the 'Fisher real balance effect', where one's debts change in real value. These debt effects are generally thought of as more important, and why most macroeconomists believe a little inflation would be good right now, and perhaps always: it reduces the legacy debt in real terms.

Lastly there are Mundell-Fleming effects, which operate though capital inflows caused by changes in real rates from changing the price level. Supposedly lower interest rates lead to capital account deficit, which a trade surplus, which implies higher GDP. The data on this are mixed, but generally international trade is the tail, not the dog, for large countries like the USA.

Now consider aggregate supply, which classically is horizontal, and then in the Keynesian world vertical. It supposedly becomes positively sloped because 'prices' refers to outputs, not inputs. Why prices supposedly effect finished goods rather than wages or intermediate goods is strictly ad hoc.

This is why general equilibrium models, like those of Edward Prescott and Finn Kydland, whose Nobel Prize winning research emphasized shocks to utility or production functions, because one has to put in some ad hoc structure to get these aggregate demand and supply curves to work in the Keynesian paradigm. Now, I don't think Prescott/Kydland models work either, but most policy debates aren't predicated on these models (I don't know anyone to really believe our problems are currently a technology shock, or preferences for more leisure).

The result of this flawed paradigm is to continually assert that simply spending more on X increases aggregate demand via their spending, because one becomes inured to extrapolating partial equilibrium analysis into general equilibrium results. Consider this AFL-CIO press release, which conflates more spending on union jobs with greater prosperity. If only everyone worked for a government protected industry with market power, presumably, we could all work 9-4 with negotiated, predictable wage increases.

So, like discussions in Marxist economics, which often involves very learned, earnest, and prolix researchers, it's best just not to go there, because it's gibberish. Don't say Aggregate Demand, say, subsidies to investment of some kind, or more government spending, because that's more meaningful, and then ask, should we be subsidizing these investments, or having more government spending? The indirect effects are so speculative you might as well ignore them, and just ask if the direct effects are worthwhile.

Do Academics Overfit?

Yes. Academics are just as susceptible to this bias as anyone else. On one hand they have extra discipline from having to put their ideas out there, while on the other hand they often don't pay the price for creating overfit models in the way a poorly performing asset manager would. The big difference between academic overfitting and that from your average quant is that when academics do it they are much better at rationalizing such models.

I've worked with finance professors on consulting projects, and cherry-picking data recent data and pointing to something 'out of sample' when it is used iteratively is quite common. An important postulate to remember is that there are no true out-of-sample backtests, just tests of subsample stability. Invariably researchers know about the entire dataset in question, so out-of-sample results are really models that when fit on a subsample and applied to its complement generate the best fit. That is, quants try models sequentially until they find one that works well 'out of sample,' which means the data is not really out of sample.

That's not to say out-of-sample tests are meaningless, just that it takes a lot of self-discipline because a lot of this is done outside the box, and the easiest person to fool is often oneself because it's very tempting to believe things when they imply many self-serving benefits. This is why integrity is a virtue, because it's hard, uncommon, and helpful. It's tempting to over-promote your own pet idea as tendentious advocacy can seem necessary in the real world where 'everybody does it.' But, the biggest problem knowledge-workers make is not making a logical error or not being able to solve a complicated problem, but working on something that is a dead-end, because that implies you've just wasted a large part of your career: an expert on input-output models, Keynesian macro models, dynamic programming isn't valuable for making decisions. Fooling yourself into believing in a false model simply wastes your time.

Consider John Cochrane and Monika Piazzesi's Bond Risk Premia paper that purports a model that forecasts one year bond excess returns with a 44% R2. Both are competent academics who I generally respect, as I think they are smart, careful and do research with good faith. Their model suggests that if you look at the current forward rates from the US Treasury yield curve, the first five forwards predict year-ahead bond returns very well (to be precise, these are 'excess' returns, so they subtract the 1-year bond yields).

What is this model? Basically, if you run an ordinary least squares of the 10yr bond return over the next year (minus the 1yr yield), on the forwards. They looked at the 1964-2004 period, which has 467 monthly datapoints, but because these are year-ahead returns, we really only have 39 totally independent datapoints, which is not a very large sample (most year-ahead returns being highly correlated because they share much of the same data). So the basic pattern they found was

year ahead 10yrBondReturn-1yr Bond Yield=a+b1*f1+b2*f2+b3*f3+b4*f4+b5*f5

*here f1-f5 are the 1 through 5 year forwards.

You can download his data here. Now, the first problem I found is that his bond data is a bit fishy. He used bond data from CRSP, and 2 and 4 year USTreasury datapoints are pretty uncommon. His 4 and 5 year forwards yield changes have a suspiciously low correlation. In anycase, I took the H15 data using their 1, 3 and 5 year monthly bond yields, and generated pretty much the same result: tent-shaped set of coefficients on the forwards (approximately equal and negative for 1 and 5 year forwards, positive and larger for the 3 year forward), and my R2 for the 1964-03 period was a large 31%. My results look like this for the same sample period Cochran and Piazzesi use:



The coefficients suggest that there are higher returns the more concave the forwards are, and lower returns the more convex. This doesn't really make any sense, in that there's no intuition as to why this 'tent-structure' of coefficients is related to risk, or utility, it just comes out of a best fit of the data.

If we look at the subsequent 7 years, that same set of coefficients that worked so well in-sample for 64-03, don't work at all for 2004-2010 (last datapoint was for the return from 9/2010 through 9/2011). See below.


So, it seems a classic overfitting of the data. Sure, the pattern could have just stopped, but given the model had no intuition, no causal mechanism, just some unlikely set of coefficients, it almost surely was an overfit. Such results, prior to Freakonomics and Behavioral Finance, were considered rubbish for a while, as the development of CRSP into a data source led to a lot of stupid correlation papers in the 1980s and 70s, but the success of other atheoretical findings (momentum) has unleashed non-intutive correlations into top tier journals.

As an academic, this will always be a plus on Cochrane and Piazzesi's vitas because it made a top tier journal (AEA 2005), but as a practioner this would have gotten them fired. Thus for academics, overfit theories that generate publications have little downside compared to a practioner.

Zero-Sum Game

A while back I got a copy of Zero-Sum Game by Erika Olson. She worked for the CBOT when there was a merger battle with CME and the ICE exchanges, so it's basically a lot of inside baseball on the corporate politics of acquisitions. It sat on my bookshelf hidden for a while, because it didn't really leap up at me. But eventually I found time to read it, and it reminded me of the old saw that everyone has a story to tell.

What's most telling is that any large institutional realignment involves a lot of little issues, and various stakeholders all have different interests and power, and making them all happy involves a lot of politics that is difficult to do publicly. The idea that merely passing a law will change things is pretty naive because any explicit, complicated process invites its circumvention. You need to align incentives, and in the private market this is by giving various people carefully delineated property rights (shares, seats), often via having them buy-in explicitly in some way. When you involve 'stakeholders' who don't have any investment in the process, they just have some vague sense of a better structure, these people just create more anachronisms and complexities that lower transparency and help insiders.

One of the more telling anecdotes was about how when Amaranth Advisers was still alive, it was breaking various position limits in energy on the NYMEX, and the CFTC which regulates the NYMEX took 7 months to actually enforce these rules. Yet Amaranth merely moved its positions to another exchange, the ICE, whose natural gas swaps were not subject to position limits at that time. One should anticipate that any regulation that mandates certain positions must be in a certain contract on a particular set of exchanges, will simply move to highly correlated positions in different venues.

Partying with Actuaries

As they say, what happens at the Society of Actuaries annual meeting, stays there. It was a pretty good conference, one with literally a dozen streams, so it's pretty easy to find some talk that's interesting at any time (it's actually continuing until Thursday). My talk (see pdf of presentation here), was pretty well received. That is, I had a lot of people telling me it's interesting, thoughtful questions, etc. This is in contrast to academics, who either tell me I'm stupid or crazy, though usually just ignoring me. I have not met an active finance professor who thinks my 'no risk premium due to relative status' is interesting, let alone true.

As mentioned, some professors have even been quite defensive, which is understandable. It doesn't bother me because it's fun to have the facts on your side. They'll come around, as they did on the low return to high volatility stocks and distressed stocks.

Bernie Sanders Clueless on the Fed

Patron saint of Liberal radio and websites, Socialist senator Bernie Sanders, questioned Bernanke this week, and he asked one reasonable question, namely, why not break up the top 6 banks if they are all 'too-big-to-fail'. Bernanke said he thought incentives in Dodd-Frank would better address the incentive problems. I'm not so sure. I would prefer the simplicity of having a maximum asset size than the open-ended regulatory gibberish in Dodd-Frank.

Then Sanders asked why the Fed does not provide low-interest loans to small businesses, to help kick-start the economy (see video here). He even threw out a number, $15 Trillion. Bernanke noted that the Fed has no structure to underwrite or service general loans to businesses. Sanders thought that this would be no different than offering back-stop liquidity to banks, oblivious to the insanely large amount of infrastructure needed for that to work.

That's just insanely ignorant, and highlights how the socialist mindset fails to appreciate the reasons why markets, and the firms that comprise them, are more efficient than government. It reminds me of Lenin's assumption that nationalizing industry would be trivial because business was strictly accounting, or more recently, the faith in shovel-ready projects. There are many stupid Republicans, but this really sets a new bar.

We are the 99%!

Egalitarians may be a majority, but they aren't close to 99%. This reminds me of the False Consensus Effect, which states that individuals view their own preferences, behaviours and judgements as being typical, normal and common within a broader context; it also suggests we find alternative characteristics as being more deviant and atypical than they actually are.

The blogger Psycasm did a survey, and asked them about their phobias, and compared them to what they supposed the percent of people shared these specific phobias, and got these results:


So, people who are afraid of spiders, dogs, and heights, vastly overestimate the prevalence of their specific phobia. Sort of like how economists, who have no alpha and are inclined to statistical optimization, assume all investors would invest as they would: presuming no alpha, optimizing mean-variance preferences. It's not a problem limited to proles.

I feel pretty calibrated, in that I'm pretty aware of my many beliefs that are a distinct minority. It doesn't bother me too much because I believe in meritocracy, which is inherently elitist. This not only is a minority view (at least in public), but it by definition considers 'common' to mean 'crappy' in most cases. As Aristotle noted, just because people are equal in some respects does not imply they are equal all respects; men are equally free, but not absolutely equal

We are a democracy, which means majorities elect the lawmakers, so it is important to have a majority opinion, especially if you want to pass laws that make people do what they would otherwise not (eg, pay more to strangers in Washington, not marry their partner, not own a gun). But there's no reason to brag about it, because it's merely a sign you can be a bully, not that you are somehow more enlightened:

The fact that an opinion is widely held is no evidence whatever that it is not utterly absurd; indeed in view of the silliness of the majority of mankind, a widespread belief is more likely to be foolish than sensible.
~Bertrand Russel

To disagree with three-fourths of the British public is one of the first requisites of sanity.
~Oscar Wilde