Falsifiability

Jane Galt and USS Clueless both have recent posts on "falsifiability", which tangentially touch on why humanities students seem so stupid and out of touch with reality so often. I assume that some of this comes out of the deep cognitive splits laid bare by the recent war with Iraq, and also with the observation that commentators who were wrong before haven't seemed to factor that into their most recent prognostications.

It is a fact that people do not automatically seek disconfirming evidence. To do so requires a deeper understanding of underlying premises than most of us have time for and is of dubious value if the (incorrect) rules of thumb we live with are good enough. A natural side effect is that this tendency makes it easy to live in a fantasy world, particularly in realms where our decisions have no real consequences.

I am well aware of how deeply my time at Chicago has influenced the way I think. Chicago-school micro is very powerful, and it is easy to see how it can displace other, flabbier, modes of thought. But it is helpful, in the spirit of falsifiability, to see what could undermine these beleifs. At its heart, if you had to sum it up, microeconomics believes that people respond to incentives, ie. they want what is best for themselves, their family, and their friends. All else flows from this -- you get demand curves that slope downwards (people buy more as stuff gets cheaper), supply curves that slope upwards (people produce more as prices rise) etc. etc. So I guess to disconfirm micro, I'd need to see demand curves that slope upwards, supply curves the slope downwards, and people willing to sell you an apple for $5 and instantly agree to buy it back for $10.

Kahneman, Taversky (Nobel prize last year) and Dick Thaler (Chicago) show that this can happen in lab settings.

And I don't know of any better way to attract attention at Chicago than provide evidence for upwards sloping demand curves and (broken) transitivity.

I'm taking a class with Thaler now, and it has me seriously thinking about where markets reward rational behavior and where the human mind can enjoy its biological idiosyncrasies without having to whip out a calculator to see if a hunch makes any sense.

Noting the payoff structure in academia, I think Thaler is dead wrong on the consequences and applicability of his findings (although I believe his results are real). He has a paper called "Save More Tomorrow" where he shows how some simple changes to a savings scheme increases savings rates dramatically. I would argue that it is so difficult to know what the right savings rate it is, that it is impossible to know whether Thaler's changes have made people better off or not. Nonetheless, you can't get around that fact that people's savings rate can change dramatically given how the system (not the incentives) are structured, so who's to say that the initial savings rate was any less artefactual?

I don't think this has the ramifications people think it does (including Thaler). I hope to discuss this with him before the term ends, but it does seem to me that you would expect these behavioral effects to be more pronounced when it is less costly to make wrong decisions. Far from limiting the ability of markets to make optimal decisions, this suggests the limits of democratic central planning.

For example, people naturally assess probability of an incident by how frequently they can recall such an incident happening. This works fine when the two are similar, but does poorly when the two are not. For example, most people think more people die of car accidents than colon cancer, but the opposite is true. Public policy decisions made on assuaging public fears will tend to overinvest in things which are rare but people panic about, and underinvest in things which are common but don't get media attention. So we spend $2,000,000 per life saved in nuclear power, and $140,000 per life saved on motorways. This is another way of saying that we, as a society, choose more people dead over less people dead, for equal cost.

But this does not give way to the libertarian dream of complete freedom with complete consequences either. The more we understand about how the brain's natural wiring leads us astray, the better we can figure out how to best make which different kinds of decisions. But given that these are still people out there who don't believe that it is inimical in human nature to want the best for ourselves, our family, and our friends (or understand what that neccessarily implies) it's going to be a long journey.

Good Lord

I never expected to see an article like this one on 2Blowhards. Friedrich really has been taking his economics study seriously.

Guns, Germs, and Steel

Jared Diamond has a new book out about why some societies made themselves extinct. He sums it up as basically 1) they made bad predictions about the future and/or 2) made bad decisions now. If you think I'm oversimplifying, trawl through the transcript yourself.

His bete noir is catastrophic environmental destruction, and he draws the obvious parallels between deforestation by the Easter Island society in the Pacific and, well, chopping down trees today.

His suffers from a serious case of status quo bias. While it is true that a society which impoverishes itself through poor environmental management has made an error, it is also true that a society which fails to get rich at its potential rate through poor environmental management has made an equal error, but in the opposite direction. Certainly if we tallied up the two columns, failing to get rich at your potential rate would be more common than impoverishing yourself. Unsurprisingly, since Diamond can't see the costs of the status quo, he urges us to embrace it.

Sheep vs Wolves

Yesterday I went to see Judge Posner debate Prof Jamin Raskin on democracy and the Constitution. I did not know this, but there is nothing in the US Constitution that guarantees citizen's right to vote. Posner argued that there are two poles of thought on what democracy is/should be: deliberative democracy where people educate themselves on the issues, debate with an open mind, and then vote for the common good, and essentially democracy as spectator sport, where "wolves" fight it out in the political arena (less bloody than them fighting it out in the field) and if the "sheep" get bored to tilling the field, they can go and watch for a couple of innings. Posner said that this view was clearly better descriptively, but was also better normatively as regular folks are bad at making decisions and so a system where everything was decided by plebiscite would be burdensome, fractious, and error prone. Raskin's view was more usual -- he preferred "top dogs" and "underdogs" and saw the democratic process as a dialectic where progressives slowly improved people's lives through democratic change.

My main reaction was that Posner is a really really funny guy, while Raskin is just angry. Being angry is not attractive. Since everyone (except maybe Raskin, and those who think deliberative democracy is a viable idea) knows debates never change anyone's mind, being funny is probably a good strategy.

The behavioral economics class I'm in with Richard Thaler has many convincing examples for why people are bad at things like probability and statistics,etc. which he argues means they do not behave like the classical rational economic actors. My take is that in a market, people have an incentive to study the issue and correct their biases, while in a political system, where you vote your identity, biases will go uncorrected and lead to wrong decisions. Watch me pull numbers out of thin air to buttress my case: the US spends $140,000 to save each life in highway accidents, but $2,000,000 to save each life from nuclear power. Public policy, therefore, has chosen to save fewer lives over more lives (or, if you prefer, kill more people over kill fewer people). Hardly an optimal social outcome.

Nobel prizewinning economist Ken Arrow showed how no democratic system can produce truly democratic outcomes, and given the inherent biases we all carry, it's very hard to design a system where decisions are made by those who enjoy all the subsequent benefits and pay all the subsequent consequences.

Distributed Filesharing

A judge in an LA court has ruled that distributed filesharing networks are not liable for copyright violation by end users. The key distinction here is "distributed"--Napster is still not OK because it had centralized servers while Grokster is OK because it doesn't.

This argument reveals how stupidly narrow legal reasoning can be because to an end user (and to the RIAA), there is no difference between Napster and Grokster. This case, literally, hinges on a technicality. I am sure the RIAA will appeal if it can, and who knows what the ultimate ruling will be, but filetrading, at least for songs, seems to be here to stay.

Moreover, even invasive and pervasive DRM will not help because that will just create a market for non-DRM songs, the market indie-kids keep insisting is just around the corner now that Napster has democratized distribution and no we can't understand why people insist on download Brittney Spears. Tie Brittney up with DRM and suddenly The Exploding Buzzlivers start looking better.

This is what I want in an electronic music system: 1) a pricing scheme that reflects the fact that I download lots of stuff initially, and then download very little, 2) programming to help me find new songs or cool stuff I hear on the radio, 3) music that I can keep on my computer and/or iPod for 4) as long as I want. We'll see what Apple announces in the upcoming months.

Metal Gods 2003

Rob Halford kicked off his 2003 tour in Chicago, and I was glad to be there. He ended his set with "Breaking the Law", which was a college favorite of mine, Stumblingtongue's, and our German roommate's.

Google and advertising

When I worked in the Human-Computer Interaction field, there was this one study that posted the top search results in big, flashing red letters. Users ignored it, thinking it was a banner ad. This study coined the term "banner blindness" and argued that users were literally blind to banner advertising.

Doc Searles has a nice Suitwatch column out today where he chats with Richard Cooper, who runs Google's Adwords program. Google's Adword strategy comes right out of a Chicago micro textbook:

People bid on placement based on keywords. They set the maximum cost per click (CPC) they're willing to pay. In effect, they set their true reservation price: the maximum they're willing to pay. They pay no more than slightly above the next lowest competitor, so there's no winner's curse where you outbid everybody by an extreme margin

Ah, a Vickrey auction (almost)! And not only do they get the economics right, but they also listen to customers and improve position based on clickthrough, while stripping out those irritating, bandwidth stealing banners. Very impressive.

Doc takes his case too far though, signing the death knell for regular, image based advertising. Many products are image products, and we would value them less if they did not come with fancy image marketing. Examples include soft drinks, beer, snacks, clothes, and cars. People like to consume image in some cases, but not in others. Doc opened his piece by moaning how tech companies wanted their ads to focus on features that would drive sales, and ignored coolness. Looks like he still doesn't know when people care about image and when they don't.

False positives

If you have a rare occurrence, then even very sensitive and accurate tests will give you bad results because they will create many false positives. Suppose you have a test which is 99% accurate for some disease which occurs only 0.01% (one hundredth of one percent) of the time. In a population of 10,000 people you should expect one instance of the disease, but the test will flag 100 "positive" candidates. So a 99% accuracy test gives you 100 positives, but only one of them is actually ill. That's a 1% success rate.

The Chicago Boyz criticize TIA because it generates so many false positives, but don't seem to realize this is because terrorists are very rare (thank god), not because the TIA is particularly incompetent (which it may be).

They compound their error by claiming that adding more information to the patchy databases will make things worse by generating more false positives. This depends on whether the new data improves the sensitivity or not, but as the simple algebra above shows, the sensitivity of the test is not the problem, it's the low incidence of occurrence in the population.

Suppose, instead of testing everyone for disease X, we tested a subpopulation that had a higher incidence of the disease, so 10%. Testing 100 people would yield 10 instances of the disease, and you would get 1 false positive. The "99% sensitive", which was performing at just 1%, is now operating at 91%. Much better.

I have no idea how many people in the US are terrorists, but it's a tiny number. Increasing the sensitivity of any tests is good, but if you want to reduce the false positive rate you must be more selective about how you screen. Sadly, the only way I can think to do this is to racially profile in some way, which is unpopular for all the usual reasons. But the consequence is a high false positive rate unless you abandon screening altogether. All seem to be bad options.

(BTW: I would not take stories about TIA targeting anti-Bush people very seriously. Given that around 50% of the country is Democrat, and therefore dislike Bush, you would expect about half the people flagged to be anti-Bush).

Agriculture

This article has some nice facts which further reveal Stiglitz's cheap theatrics as being just that.

Stiglitz states that US (and European) agricultural subsidies depress the value of agricultural goods, reducing the money third world farmers can make from farming. But third world non-farmers benefit from cheaper food, and they would be worse of if American tax payers stopped tying $20 notes to their lettuce. According to the article, the net effect would be negative in the very poorest countries, and positive in the less poor countries.

A cheap clown

Nobel Prize winner Joe Stiglitz spoke in Chicago today about what was wrong with the main two instruments of economic globalization, the IMF and World Bank. Perhaps because his audience was not made up of academics, his speech was shallow and filled with cheap shots, and he grinned impishly at every one. Clearly he loved the attention.

Here were his substantive points:
1) Protectionism in the US is bad. This is clearly true and an example of how special interest groups can hijack public policy to their own benefit. Many people see this as a limitation to the political process, but Stiglitz feels that the world would be better served by more government not less, all the while decrying how narrowly political import policy has become.

2) Intellectual Property regulation has gone too far. Again, no arguments here, but he used drug patents and AIDS as an example. Drug patents are clearly required to motivate medical research, and the reason pharmas don't offer them cheap in poor third world countries is because they need to price discriminate and maintain high prices in the US to pay for all the research. The issues around letting Botswana reimport from South Africa go to the very heart of this, but instead of treating the topic like an honest adult, he called it "murder" by the pharmaceutical industry.

3) Third world capital markets are destabilizing. I used to work in a hedge fund, and in my experience this is 100% true because the markets for exotic financial instruments, which includes everything out of third world countries, is very thin and therefore weird and volatile. The price you pay for excluding this volatility is higher cost of capital and less investment, but it may be worth it.

4) Monetary policy in the US does not take Latin America into account. True, but this is also the least worst alternative. I cannot see how the US refusing to lend to Latin America helps either of them out, and they both have alternatives. This type of interest rate risk is very real, but all the solutions I have heard to date are worse than the problem.

5) Fiscal policy is a mess. This is true, and if he had suggestions on how to improve it I'd love to hear them.

6) Privatizing social security costs the government money. Stiglitz says that since a government can't count private contributions to social security as an inflow, it makes government deficits look worse. In his next breath, the condemned Enron accounting, which specialized in counting inflows while hiding liabilities, exactly what government social security accounts do (to make them look better).

Cheap. Clownish.

The last question was on what the policy should be for Iraqi reconstruction. He said that Iraq needs to develop markets and institutions that are close fits with local needs, and therefore require lots of local input. He also said that debt forgiveness would be critical to any rebuilding and the US would reverse its usual position on that but be blocked by France and Russia, the two biggest creditors.

All Kling, all the time

It's been too long since I visited Arnold's pages. He has a nice piece on how the more diverse an economy becomes, the more resistant it is to shocks. This is exactly right -- people who complain about how uncertain the economy is today have no comprehension of the magnitude of the business cycle in years gone by. The US economy is smoother by far than it has ever been in its history, although it may not feel like to right now.

It's also worth knowing that, for all the talk of globalization, the US economy is essentially closed. Only about 12% of US GDP comes from trade, and 90% of that is between Canada and Mexico. This is not true for Canada and Mexico, however, where a substantial portion of their GDP comes from trade with the US. In terms of trade, the rest of the world does not register on the US radar screen, but I don't know what it looks like from the other side.

The Brain of a Rational Actor II

A while ago I wrote about how economics (seems to) struggle with people who have conflicting desires -- how do you explain an anxious dieter who padlocks the refrigerator, but then unlocks it to binge? In the end I said that "that people's desires come from biology as much as logic" and not to get caught up on the Spock-like connotations of the word "rational".

I chatted with Chicago behavioral economist Thaler yesterday during a class on how strong people's beliefs can be, and how folks can come up with all sorts of explanations for anomalous behavior. I then stumbled upon an essay by chemist/philosopher Michael Polanyi who writes eloquently about the same thing (via 2blowhards). (School buddy Stumbling Tongue also chimes in).

How does this tie together? A key question in Behavioral Economics is "how much does all this anomaly stuff actually happen where it matters?" Thaler does not know. Neoclassical micro would say it happens most where it is cheapest, that is, when the costs of getting things wrong are low, and if they are high, are paid by someone else. The examples that spring to my mind are public policy made after a traumatic event, surgeons recommending unpleasant, risky procedures, and academics, particularly in the social sciences and humanities.

Behavioral research also shows how new evidence tends to polarize belief based positions (Polanyi makes this point also). I used to wonder why people were so bad at micro, and now I know. Then I wondered why some people hated it, and that's pretty clear now too.

Red Hat

I ran into Redhat CEO Matt Szulik in the Atlanta airport yesterday. We chatted about how dramatically open source software has gone from "something kooky" to "yeah, of course", particularly in the financial sector. I know a bunch of guys in IBM's Linux groups whose mission in life is to "kill Sun in banks". Their weapon: Linux.

Matt also mentioned that adoption in Europe was very good, but middle America was still holding out. He was pleased with Microsoft's draconian License 6.0 as it has encouraged folks to reevaluate their relationship with that company, opening the door for Lintel systems.

Back from Key West

Song of the trip: "Shine On You Crazy Diamond" by Pink Floyd.

Misplaced overconfidence

Landsburg from Slate claims that economics shows how everything goes worse than expected. A little economics can be a dangerous thing, and this muddled thinking from Landsburg is a case in point.

There is a phenomenon in economics called the "winner's curse" which works just as Landsburg says. Auction winners are, by definition, those who were most optimistic about the value of the prize. Given that the true value is likely to be the mean of all the bids, it means that winners were those who overpaid the most. (This is not lost on auction goers. In industries where this stuff really matters, like oil exploration, they conduct auctions under different rules to deal with this effect).

Landsburg then applies this to war, saying that wars of choice are where the chooser feels most optimistic about winning, and therefore is most likely to be overoptimistic, and therefore finds things harder going than first expected. While it is true that war involves overoptimism (both sides think they can win, but only one is right) I don't see why this is limited to the one who picked the conflict. After all, unless you are dealing with nutcases, the other side usually has a list of demands which you can capitulate on and therefore avoid war (Clauswitz makes a persuasive argument that war is merely an extension of diplomacy). Since both sides have freedom of action in a war of choice, overconfidence can come from either side.

The be frank, the main overconfidence I've seen over the last three weeks have been from news commentators thinking they know what they are talking about.

People on NPR get it wrong again

There was some discussion on NPR this morning about the role luck plays in people's success in the US. While we've all heard about how you need to work hard to get ahead, suppose some folks just got lucky? Both academics on the show were arguing that if luck played more of a role in personal success, then it was OK to redistribute money more since it was all just a lottery anyway.

One of the benefits of mathematics is that it keeps you from falling prey to cognitive illusions, like the one above. If you say that output is some combination of effort AND luck, then the outcome of effort is more uncertain, so you need to give people even stronger incentives to work. This means that redistributive taxes, which reduce the share of output you get to keep, become more harmful the more luck driven the output becomes. This is the exact opposite of the point the NPR folks were making.

The error in their thinking was that they equated getting money through luck with getting money through zero effort. Gold prospecting, which is about as luck driven as you can get, requires you to go out and look for gold. The NPR folks presumably thought wealth occurred while you sat at home watching TV, because they ignored the effect of incentives on work and how those incentives would have to be strengthened if the output from effort grew more uncertain.

Microsoft ships first app for Linux

Windows Media Player is now available to third party device makers running Linux. This increases the value of WMP, but decreases the value of embedded Windows (WinCE).

The newest addition to the family

My sweetie named him Percy.

The Brain of the Rational Actor

One of the classic "puzzles" in economics is why people break diets. By starting a diet, they have clearly expressed a preference to eat less and therefore lose weight, but then they snack in the middle of the night. In economics, people are modeled has having some combination of preferences and a budget, and then they maximize the "utility" they get from these preferences by moving around their resources. As breaking a diet is clearly contrary to stated preferences, some economists belive that the "utility maximizer" model is not useful.

Winterspeak reader NP writes in:

Anyway, you mentioned you were ruminating over a problem about how people sometimes seem to have conflicting preferences -- you gave the example of a person who says explicitly that they want to lose weight, yet consistently exhibit behavior contrary to that goal.

The problem is that said person has a dissonance between his conscious and unconscious desires. The classic method of demonstrating this is to take a subject, hypnotize him and give him the post-hypnotic suggestion to kick his leg when he hears a certain bell ringing. Then, the subject is taken out of hypnosis, and in a little while somebody rings that bell. He kicks his leg.

Then, they ask the subject: "Why did you kick your leg?" and without taking a breath, he responds "Oh, well I had an itch" or "My leg was falling asleep" or "I needed to stretch." Never an "I don't know." Their conscious mind plays poodle to the unconscious and makes something up.

Now, there's the basic problem of the labeling -- whether the actors should be limited to two things called the "unconscious" and "conscious" -- but there seems to be a great deal of evidence that the human mind has different "parts" which can want completely different things at the same time, and that it also appears to have faculties which attempt to compensate for the conflicts, sometimes yielding puzzling results.

I don't know if this actually helps you with your problem. Now we're into philosophy -- where is the "real" economic actor in there?

DP is right -- the real economic actor is both the "concious" and "unconcious" desires. People don't think in math, which is all that models are, so they might disagree with the model if it does not seem to fit with their "word" interpretation of the world, even though it really remains consistent and useful. Gary Becker's work extending the standard econ models to the family, emotions, addiction etc. shows how effective this can be.

People's "word" model of the rational economic utility maximizer posits a ghost in the machine, an essential self that sits in control and makes decisions (the "concious" actor interpretation). And since people are a jumble of imperatives, some cognitive and some biological, this detached ghost seems to be a bad way of thinking about things. But there is nothing in the math that requires a ghost in the machine. People's preferences can be a jumble of conflicting values that are pretty stable over time, but still emotional as well as clear headed, and mathematically it still all operates comfortably within the standard economic models. People declare diets because they want to lose weight. They cheat because fatty food tastes good. Nothing mathematically irrational there, just an honest observation that people's desires come from biology as much as logic. The utility function is broad, it can welcome all kinds of things.

J'African

My good buddy from school, Joel, has been living in Cape Town and having all sorts of wonderful adventures for about 4 years now. He finally has a blog where he talks about the ins and outs of South African politics (he's a speech writer for the opposition party there) and other usual suspects. I encourage you to check it out.

Thaler unplugged

Had my first class with Dick Thaler today, Chicago's Behavioral Economist of choice. Two of my buddies are TAs, and I'm looking forward to the material. Thaler trotted out some optical illusions, as well as puzzles demonstrating how people are bad at complex trigonometry, statistics, and probability. The optical illusions he used also happened to be in Pinker's Blank Slate so I asked him if any type of biological/evolutionary foundation underpinned his work. He said no, and that he found evolutionary explanations to be useless because they could be used to explain anything, and offered the example of how signaling arguments explaining peacocks' tails, or thrill seeking in young men, could be cooked up for anything and go both ways.

He is, of course, exactly right, but only in the narrow sense that asymmetric information/signaling stories can explain anything and are usually unfalsifiable wherever they appear, be it evolutionary genetics or economics. (My PhD buddies tell me that this is why the field is virtually dead in academia, even though it won a Nobel Prize a couple of years back.) In other areas this is not true. The fact that eyes get tricked by optical illusions show that they are not simply light meters, but that they process visual signals in specialized ways as well. These specializations are genetic, so it is not crazy thinking they've evolved under adaptive pressure. I have a feeling that the answers Thaler et al are trying to dig up will come out of experimental neurocognition, not empirical behavioral economics, but we'll see. I'm looking forward to the class.

Coase Unplugged

I had the rare privilege of meeting Nobel Prize winning Chicago economist Ronald Coase today. Coase came up with the theory of the firm showing how transaction costs determine industry structure, and also was behind the "Coase Theorem" which demonstrates how, under certain conditions, the economic outcome is independent of how the property is allocated.

The guy is really old, but he was still sharp and witty, and remains focused on looking at real data to figure out what actually happens instead of solely relying on blackboard economics and then looking for corroborative examples. I asked him if the questionnaire based data sets behavioral empirical economists come up with were any good, and he said they were worthless because they don't test what people do when it matters, and that's when folks start to think seriously about costs. He also mentioned that he was unimpressed by a recent Behavioral Law & Economics book for precisely this reason.

(On a semi-related note, a friend and I were discussing how to make sense of people who don't do what they want to, for example, they declare they want to lose weight by going on a diet but then gorge on ice cream. Between the Coase lecture and The Blank Slate, I think it has something to do with how people think about the preference function, but I need to mull on this some more. More later.)

As a brief example of how insightful he is, he casually mentioned that "covering costs and maximizing profits is the same thing." Now everyone believes that a business has to cover its costs, but people view the profit motive as being evil. If you take opportunity cost into account (which you should), i.e. the next best thing the business could be doing if it weren't doing what it was, then the two are the same. So a business that burns $10 a month and could be making $40 but instead chooses to make $20 actually has an economic profit of -$30 ($20 - $10 - $40) even though it has an accounting profit of $10 ($20 - $10). If it chooses to make $40, it has an accounting profit of $30 and an economic profit of $10. Who loses money and runs away if a business doesn't maximize economic profit? Investors.

Getting it exactly wrong

I've been reading The Blank Slate by MIT neuropsychologist Steven Pinker where he discusses what modern advances in neural anatomy, evolution, and genetics have to say about human nature. Long time readers of this site know that I've been trying to figure out why people find economics so counterintuitive, and I have a much better sense of that now that I know something about how evolutionary pressures have shaped how the mind processes information. The thing that I like most about the book is that it only needs a few, basic insights that explain a lot of behavior and are pretty widely applicable. (I also did some biochemistry in a former life so I appreciate the chemical functions underlying life processes).

This piece in the Guardian discusses how language frees people from their genetic drivers allowing us to "forge our own futures." This is exactly wrong, since 1) people's ability to learn language is genetic and 2) language is distinct from thought. Chomsky, a big believer in man's ability to improve human nature, showed that the brain is born with the ability to learn and process language, although the particular language a child ends up learning is determined by the environment. Moreover, language is distinct from thought based on how people sense lies, get outraged at hypocrisy, and deal with ambiguous sentences like "Young Children Appeal to Catholic Priest". So our ability to speak languages is as genetic as the color of our eyes, and while language gives us more ways to mediate conflict, the world has zero sum games which, ultimately, you need to fight to win. More on this later.

Back from Vegas

Song of the trip: "In Da Club" by 50 Cent.