Hacker News Reader

Comment by yummyfajitas | Hacker News Reader

We've seen this article before: http://news.ycombinator.com/item?id=156177

It was crap then and it's crap now. Karelis doesn't understand econ, but criticizes it because econ disagrees with his thought experiments and political leanings. His theory is nothing but traditional econ with a modified utility function. It has rather perverse implications (i.e., we should tax the poor but not the rich), and as the article notes it does not agree with experiment.

The best reason for completely ignoring Karelis: "ultimately, he believes, the strength of his arguments is less in how they fit with the economic work that's been done to date on poverty - much of which he is suspicious of anyway - but in how familiar they feel to all of us, rich or poor."

View on HN

Karelis doesn't understand econ, but criticizes it because econ disagrees with his thought experiments and political leanings.

Sounds like a model economist to me - because of his political leanings, he doesn't like what the oversimplified toy models of the more mainstream theories tell him, so he invents another toy model with different oversimplifications, and claims that it's more fundamental. Surprise, surprise, it happens to reinforce his pre-existing beliefs!

Isn't this the way the game is always played in the soft sciences, where we can't actually figure anything out for reals?

I say he doesn't understand econ, simply because he says "traditional economics doesn't apply" and then describes a theory which is traditional econ with a simple mathematical modification.

Also, the theory of diminishing marginal utility has been tested empirically many times. While quite a lot of econ is the theory spinning you describe, DMU is pretty solid and accepted by economists of all political stripes (it's the primary argument for progressive taxation, among other things).

"In many cases, Karelis says, diminishing marginal utility certainly does apply:"

Either way, I'd like to know why exactly he and the article are crap. I'm not completely sold on it myself, but you've really only managed to appeal to authority so far.

The main reason his theory is crap is it just doesn't agree with experiment. The article mentions several, all of which tend to encompass the income regime Karelis is presumably talking about. One example is the negative income tax experiment of the 70's, another is welfare to work.

If Karelis has evidence of his theory, he should present it. But as it stands, he is pushing thought experiments and analogies, and telling us to ignore the empirical evidence when it disagrees with his theory.

If traditional economics applied to the poor, a whole lot of things would be working a whole lot differently now.

So it seems he's right about that. Whether he actually understands economics or not. And it's likely that he does understand traditional economics to some degree, it's not hard, he prob took intro, micro and macro like everyone else. Those models quite clearly do not apply to the poor or else we'd see a different poor.

Those models quite clearly do not apply to the poor or else we'd see a different poor.

Could you explain this statement in more detail?

If traditional economics, specifically "Econ 101" (which is actually what he disparages in the article) applied, the poor wouldn't be so dumb with their money, wouldn't be the biggest buyers of lottery tickets, drop out of school disproportionately just for being poor, etc.

According to "Econ 101", people are supposed to pursue their rational self-interest and maximize utility regardless of their conditions. Which is a bit of a problem with all of economics, and a huge problem when it comes to applying it to the poor.

Does that make sense?

I read an interesting paper a few months back about the lottery ticket thing. This guy's theory was that it actually was rational, because many poor people don't have access to bank accounts, they just buy a scratch ticket every day and that's effectively a probabilistic bank account. Dunno that I buy it. But a neat thought.

Traditional economics characterizes the poor as having a very high discount rate. Thus, they assign a low value to events which might happen in the future - that is, a high paying job in 5 years is less valuable than using meth now.

In much the same way, I (a middle class person) would rather have $100 now than $102 in 50 years.

As for lotto tickets, traditional econ says the poor derive utility from playing lotto, not just from winning. Am I foolish and not obeying the laws of economics for buying video games, which have an expected payoff of $0?

So, the original author's point was that the discount rate thing applies differently than you're applying it.

They're not choosing less satisfaction now rather than more satisfaction later. They see things as hopeless (too many bee stings), throw their hands in the air and say screw it. A 0% chance of a high-paying job in the distant future discounts to 0 utility at present for any discount rate.

Karelis isn't talking about discount rates at all.

He is claiming the poor have a flat utility function with a kink at zero. I.e., the ability to cure 1 bee sting (out of 5) now is worth nothing to them, only the ability to be bee sting free holds utility.

I think to understand the point I'm trying to make, we could re-examine your previous statement: "Am I foolish and not obeying the laws of economics for buying video games, which have an expected payoff of $0?"

If the laws of economics predict that you shouldn't buy that video game, then you're not foolish -- they are.

People's behavior changes radically when you give them experimental situations that involve gaining or losing money -- they're way more risk averse with losing 100$ then they are with changing their odds of winning 100$. This isn't baked into the assumptions that economics makes, and neither are a million other wrinkles of human behavior.

"What economists say" isn't the be-all and end-all, it's just another prism to be used.

If the laws of economics predict that you shouldn't buy that video game, then you're not foolish -- they are.

The laws of economics make no such prediction, and they make no corresponding prediction about lottery tickets. That's the point.

There are places where the laws of economics don't work, I just don't believe that your previous post (or Karelis article) identifies any of them.

Thus, they assign a low value to events which might happen in the future - that is, a high paying job in 5 years is less valuable than using meth now.

That's two different classes of events and you are misrepresenting what was claimed by attempting to further devalue poor peoples choices by saying they would rather use meth than work. It's also insulting towards poor people to imply they are all meth users. A proper analogy would be that they disproportionaly value a job they can have right now over a higher paying job they can have in 5 years.

I thought his main policy suggestion was just solving poor people's problems without strings attached (has the same effect and is cheaper to run with no bureaucracy).

His main theory is that poor people's utility function is flat.

The policy implications are more nuanced than he says. We should do one of two things: 1.) give the poor so much money that they get into the region with an upward sloping utility function, assuming such a region exists. 2.) Tax the poor, since taking some of their money will not make them worse off (what's one more bee sting?).

Mysteriously, he never mentions 2).

Because 2 doesn't solve the problem the problem of poverty. I don't find that particularly mysterious or dishonest.

1.) The Bee Sting model doesn't imply that the poor have a completely flat utility function - just a huge drop-off at some point and then a gradually decreasing plateau from there on. Indeed, the poor do seek to improve their utility functions - with booze and drugs and short-term gratification - in ways that would never allow them to overcome that seemingly insurmountable drop-off. The initial drop-off is the chief characteristic of the Bee Sting model.

2.) Utility functions are mathematical tools, and our interpretations of and use for them don't exist in a moral vacuum. Why should tax policy seek to minimize it's effect on our utility functions at all? Why not take $40000 from Warren Buffet AND the primary school teacher? Local movement on one's utility function is important, but so is absolute position, with all of the attendant moral implications.

1.) If the utility function of the poor wasn't flat, they would make some effort to cure a few bee stings.

2) Tax policy should minimize impact on our utility because the goal is (in principle) to harm people as little as possible. This is why we consider it fair to take $1 million from Buffet but nothing from a poor person - the loss of $1 million will harm Buffet less than the loss of $50 would hurt a poor person.

If Karelis is right, taking $50 from a poor person won't hurt them at all - what's one more bee sting?

1.) No, if the utility function of the poor wasn't flat, they would make some effort to address, not cure, a few bee stings. And in fact they do address those bee stings - they booze and drug themselves into numbness.

There's no point in curing anything less than all the bee stings, because of the huge drop-off in their utility function. Would it make any sense at all for you to attend 1/2 of medical school if you wouldn't eventually enjoy the pay-off of becoming a doctor? Would doing half the monumental amount of work to start a successful business make you rich? And yet you probably wouldn't think that you have a flat utility function, would you?

2.) This is only relevant if you make the flat-utility function assumption, which I don't think is true.

If U(4 bee stings) > U(5 bee stings), a poor person would attempt to eliminate a bee sting since they would increase utility by doing so. The fact that there is a spike at x=0 bee stings is irrelevant, it does not make it pointless to go from 5 stings to 4, it just means that going from 1 to 0 makes you happier than going from 5 to 4.

If Karelis asserts there is no utility gain in going from 5 to 4 bee stings, then Gain = 0 = U(4 bee stings) - U(5 bee stings), or U(4 bee stings) = U(5 bee stings). That's a flat utility function.

Would it make any sense at all for you to attend 1/2 of medical school if you wouldn't eventually enjoy the pay-off of becoming a doctor? Would doing half the monumental amount of work to start a successful business make you rich? And yet you probably wouldn't think that you have a flat utility function, would you?

You've just described exactly what a flat utility function means. U(0 years of med school) = U(half of med school). And similarly, there would be no harm in kicking people out of med school who would not finish anyway.

Your argument misses the fact that there is a huge difference between not having money and being impoverished.

Sure, adding one more impossible task on top of everything else a person who views the world as a set of impossible challenges does not significantly change that person's miserable life. (Your odds of collecting are admittedly low though.)

However there are plenty of other people with similar incomes who manage to make ends meet. For those people, adding the additional burden can easily make all the difference between holding everything together and having their lives fall apart. That's a huge change in utility. (I can name two of these off of the top of my head.)

Furthermore there are other people who are temporarily impoverished. They weren't so. Now they are so. They may be better soon. Punitive measures like the ones you suggest can greatly lessen their odds of coming out of poverty. (I can easily name multiple friends of mine who were temporarily homeless with no resources, who managed to reconstruct their lives.)

There are two excellent reasons why the bee sting theory of poverty can be correct, which would invalidate your conclusion that it is not bad to heavily tax the poor.

If Karelis is right, taking $50 from a poor person won't hurt them at all - what's one more bee sting?

Karelis's viewpoint is that they wouldn't have spent the $50 to alleviate the bee sting. They would have spent it on something that did give them utility. That nice thing (that they really couldn't afford!) what is taken away with the extra taxation.

That said, I think the point you make is reflected in a lot of real public policy. Poor areas get the worse public services, the most pollution, etc. They have so many bee stings that they can't be bothered to complain about one more.

1) It doesn't have to be completely flat, just fairly close to it. A "bee sting" is perhaps too mild a way of describing the problems of a poor person. Is it easier to understand what Karelis is talking about if you imagine a poor person who, instead of four bee stings, has three out-of-wedlock children, ignorance of contraception, a third-grade education, and a drug habit? Solving any one, or any two, or three, of these problems, still wouldn't be enough to lift them out of poverty, which seems like a pretty obvious point.

2) I suppose this might be seen as biting the bullet and admitting to an indefensible point, but I do not find it hard to imagine that if I were poor and desperate, and some asshole came and took my last $50, and I had absolutely nothing, nothing else with no hope of being able to improve things, I wouldn't care. I was already fucked anyway.

Amen. Since when do we accept arguments just because they sound nice? Hasn't a whole lot of work gone into avoiding this black hole of rationality?

Hey, maybe supposedly being able see Russia from her house really DOES give Ms. Palin foreign policy experience.

If a theory sounds promising, then go gather some data. If it still looks good after careful scrutiny, then go write your book.

> Hey, maybe supposedly being able see Russia from her house really DOES give Ms. Palin foreign policy experience.

You do know that that's from SNL, not Palin. (Since you're suggesting, correctly, that one should go with data, it's curious that you used fiction as supporting evidence.)

Border state governors do negotiate with neighboring countries, just as other governors negotiate with neighboring states. Alaska mostly deals with Canada, but there's some stuff with Russia for the Bering Strait. (Palin's house isn't near the strait so she can't see Russia from her house.)

You do know that the SNL thing was a riff on something she actually said, right?

Actual quote, http://abcnews.go.com/Politics/Vote2008/Story?id=5782924&...:

GIBSON: What insight into Russian actions, particularly in the last couple of weeks, does the proximity of the state give you?

PALIN: They're our next door neighbors and you can actually see Russia from land here in Alaska, from an island in Alaska.

GIBSON: What insight does that give you into what they're doing in Georgia?

PALIN: Well, I'm giving you that perspective of how small our world is and how important it is that we work with our allies to keep good relation with all of these countries, especially Russia. We will not repeat a Cold War. We must have good relationship with our allies, pressuring, also, helping us to remind Russia that it's in their benefit, also, a mutually beneficial relationship for us all to be getting along.

This claim is true.

http://www.slate.com/id/2200155

Yup, and what she said was true. The riff, not so much.

Governors do deal with neighboring countries and states. What part of that do you find scorn-worthy?

Senators may be asked by the president to deal with other countries but have no formal role in foreign policy beyond voting on legislation and overseeing the state department.

I concede the point. Though upon looking it up, she did make a similarly bogus statement.

I watched her first big address and thought that I remembered this statement from it, but it looks like in fact Tina Fey was really that convincing.

Her first big televised address -- the one that had the related statement -- actually had the middle of the statement edited out. The statement, as aired, was pretty bogus, but it makes a lot more sense when you read the whole transcript.

Reply on news.ycombinator.com