Thursday, October 17, 2013

abstract principles

If an abstract principle is any good it should coordinate available concrete information in an unambiguous, coherent way. This works best in maths and some of the sciences (though some branches of physics don’t do as well as you might think). In social science it’s basically impossible to articulate abstract principles that will satisfy everyone, but that doesn’t mean we should try or that we think of better or worse abstract models. In the case of economics I often see people moving quickly from noting shortcomings of economic models as if that somehow supported their own opinions. The weaknesses of the models show that economics is hard, not that you know better.

In the case of economics, there are a bunch of economic principles which economists basically agree on and many people can accept when stated as a general principle, but get totally overridden when comes to particular cases. Here’s a list of 15 economic principles from a speech given by a journalist I really like. Some of the principles are obvious enough to seem like common sense rather than “economics” but people, including politicians, deny them all the time! My past few posts may be boring but I don’t really expect that there is a lot to get worked up about one way or another. This post is different though. I’m trying to argue that we’re wrong to let the emotional effect of special cases override abstract ideas we should believe are true. 

Maybe the most obvious one is that demand curves slope downward. In general, raising the price of something lowers the demand for it. There are some obvious exceptions to this rule like luxury products where low prices can be seen as a bad signal about quality which can affect demand. There are also less obvious exceptions to the principle like the idea of a “Giffen good”. If the price of a staple food like rice increases people might buy more of it because they have less money for nicer food and rice is still the cheapest thing around. How far do these exceptions go in invalidating the idea that demand curves slope downwards? Not that far I say, maybe we need to stay on our toes when applying the principle but still, I think that not assuming it’s generally true would be crazy. 

Now how about the demand curve for labour? Does that slope downwards? If putting it that way is a little boring, how about, “should we get rid of the minimum wage?” If it seems really obvious that we shouldn’t get rid of it, why is that? Does raising the price of labour not reduce demand for it? Or do we think of rich employers and poor employees and see that it’s just obvious that employers should pay more?
Now maybe the potential downsides of a minimum wage are worth the costs, but if you think it’s obvious that a larger minimum wage won’t result in less job seekers finding low skilled jobs, your belief is in tension with a pretty fundamental economic principle. I have seen lots of arguing in favour of the minimum wage, but I see much less recognition of this basic tension.

On Monday Eugene Fama shared the Nobel Prize partly (primarily) for his work on the efficient markets hypothesis. “Efficient” makes it sound like markets are “good” but the EMH is compatible with bad outcomes. The idea is that markets are good at taking the available information into account. Basically, when someone tells you about a hot new stock that you have to buy you can basically just ignore them. The trouble is that in conversation when someone is trying to convince you about their stock, they will definitely have more information about it than you. They will be able to give you a detailed argument in favour it that you will not be in a position to pick apart, but you are still justified in ignoring their advice. The world is a complicated place and there are lots of people out there smarter than you trying to be clever about what stock to pick. The odds that your salesman is the special one, who really knows, are really low.

I’d guess that most educated non-economists (and plenty of economists) think that the EMH is just plain wrong and many think that it’s so obviously wrong that they are genuinely surprised  to discover that it’s actually a thing. There are some people who do seem to be special and there are apparent exceptions where the EMH doesn’t seem to hold. But the fact remains that it is incredibly difficult to beat the market. Lots of very clever people devote their lives to doing this and very few succeed. Just like some exceptions to demand curves sloping downwards doesn’t mean we should chuck away the basic idea, same with EMH, even if the dude selling you his stock is really convincing. Once you give up on the quest for exceptionless certainty these principles are very powerful and can help understand the world and I think we should resist the emotional appeal of cases that seem to clash with them. 

P.S. Link to the 15 principles is there now. 

Monday, October 14, 2013

concrete and abstract thinking

In my last post I argued that we often know something in a general, abstract sense, but we often let let the influence of specific events override these more abstract beliefs. The general truth that we’re unlikely to die in a terrorist attack is not refuted by the occasional attack killing some people, but people act as if is. This is bad.  


This is not to say that it’s always better to think abstractly. Stalin knew that “one death is a tragedy, a million death is a statistic”. Thinking abstractly about “ a million deaths” means we don’t appreciate the horror implied. We’d actually do better to imagine specific people being affected.


One way to think of what science (and philosophy?) is trying to do is to discover abstract models which are generally applicable. People might become extremely skilled at something and this skill involves implicit understanding of deep principles, but a scientist wants to make these principles explicit so they can be applied in different situations, or even create different situations.


There are some domains where we have been able to do this very well, like maths and quantum physics (I was tempted to write “hard” sciences, but that would be circular since I take it that what make a science hard is the lack of ambiguity in how the model relates to lower order building blocks of the domain). Developing an unambiguous conceptual framework for observations is difficult because the world is very complicated. Economists have tried really hard to make economics more scientific but basically it really isn't a science in anything like the sense of physics or chemistry. Most economists know this well enough, but some don’t. The mathematical nature of academic economics can be pretty misleading, especially with macroeconomics.


Another reason why constructing abstract models that coherently integrate available data is that it is just so easy to think up general principles and then to screen info that you consider to confirm your big idea. Conspiracy theorists are the ultimate masters of this. Conspiracy theorists are usually armed with a ton of “facts” while normal people do not actually know much about the supposed events and are not usually in a position to refute the crazy person point by point. Conspiracy theorists imply that if you can’t refute their points you should believe their conclusions if you want to call yourself a rational person. But usually we are perfectly justified in supposing that a man really has walked on the moon and that 9/11 was not actually plotted by George Bush.


A while back I wrote this blog post comparing evolution and religion as conceptual frameworks explaining how life got to be the way it is. People who believe in evolution are usually not phased when a new evolutionary puzzle pops up because they’re confident that the explanation will fit within the framework of natural selection. We’re right to think this, but it’s not that paranoid for sceptics to think that the detailed story is being constructed specifically to fit with evolution when a different theory might fit more naturally (God, of course).

Finally, our minds are just not designed for our concrete actions to be consistent with our abstract beliefs. Some animals have some sense of of the future and past and slightly distant places and humans have developed this capacity much more than any other animal. The very fact that we are able to think abstractly comes from our need to think about the future and far away things. Initially this would have had adaptive value because it helped us to plan for the future and coordinate our actions, which is good for hunting and avoiding being killed by other tribes. But the ability is mostly used for presenting good images of ourselves to improve our status and get desirable mates. It’s easy to think of examples of people who profess idealistic values but whose day to day behaviour is not consistent with these values. So long as we don’t get caught, it is usually to our benefit that our concrete behaviour is not consistent with our abstract beliefs.

Monday, October 07, 2013

trees and forests

I started writing this intending to make a particular point but then the post got long and I can’t be bothered to edit it back down to size, so maybe I’ll write a follow up.


Sometimes we get so worked up about specific facts that we lose sight of more general truths. We should want the big picture and the low level detail to be part of the same, coherent story but this is really difficult, especially when we want to argue for a particular conclusion.


Take bad calls in sport. Our reactions are wildly inconsistent if they go for or against us. When they go against us we’ll go into detail about why this particular case is particularly inexcusable. When a bad call goes our way (and there is no possibility of arguing that the call was actually fine) we are more likely to get philosophical, “Well, that’s the nature of the game. These things balance out in the end.” Unless the ref is actually fixing the match, the second reaction is actually sensible and the first reaction is just plain wrong. We all know that refs make mistakes all the time and we simply don’t care when the mistakes are not important, but these mistakes are no less blameworthy than the important ones. It’d would be a sign of a genuine problem if the mistake rates were different in big situations (even if they were lower!), but I’ve never seen this argued for.


The conceptual framework you use should take the available information into consideration. If you know that refs makes mistakes then your general beliefs about refs should reflect this. Another example is free speech. Most people I know think the principle of free speech is a good thing. This is not undermined by the cases where people use the right to say or write ridiculous, immoral things. But I quite often see people saying things like, “the right to free speech doesn’t include the right to be offensive.” Not only does it include that right, that is almost the very definition of free speech. Since it would be the government that enforced any prohibition on offensive speech, a definition of “Free speech = speech the government decides is not offensive” is hardly reassuring. Now perhaps free speech is actually a very bad thing because of all the offence flying about, but then you need to argue against the principle of free speech, not try to stop individual assholes spouting hateful crap.


How about terrorism, or shark attacks? Individual instances of both have such a powerful grip on our imaginations that actual policy responses to them are *ridiculous*. I’ve read articles suggesting that in the US many people would choose to drive instead of fly if they could so the extra number of deaths from car accidents may be higher than the number of deaths from the 9/11 attacks. If the goal is not to die, we are mistaken to drive when we would otherwise fly.  


We are not always wrong to let individual facts influence our general beliefs about something or someone. It doesn’t matter if a person doesn’t murder people as a general rule; one murder is enough to justify a generally negative opinion of him. And of course we are not justified in dismissing particularly salient events if it turns out that they are norm. This is the whole point of the saying, “the plural of data is not anecdote” (*It definitely isn't! What is the plural of data?) . There is a reason why we collect proper data instead of just telling compelling stories.


There are plenty of cases where the visceral, detailed events seem to directly contradict what the data tells us. In these cases we need to actively resist the emotional force of particular events and let more impersonal, abstract truths guide our beliefs. Failure to do this has bad consequences, like massive overreaction to terror attacks.

But if you agree with all this, in what other areas do you think we tend to make the same mistake?