More on John Lott and hyper-rationality
I recently made a short post about John Lott after watching a CSPAN presentation he made on his recent book, Freedonomics.
I had said he seemed to be assuming that humans are hyper-rational, and that's a scary mindset for someone who had been in a position to influence US Sentencing Commission policy.
Lott responded.
Of course what he says here is true. But that doesn't mean it has relevance to actually making government policy choices. And that's where the scary part comes in, he uses his simplistic assumptions and arguments to make policy.
His apple example is an example of the dangers of his hyper-rationality. "Everything else being equal" is his argument. That pretty much negates the relevance to policy determination of anything he's saying. The world is not linear, just because something might be true if everything else is held constant does not mean it's going to be true when everything else isn't actually equal, and we know it's not going to be equal.
Let's think about drug policy as an example. Everything else being equal, if we raise the price of drugs to the recreational users then they'll use less. Right?
No, that's not true. It's not true because many of the policies we use to try to raise the price of drugs tend to raise the search costs. When you're buying an illegal substance the search cost of acquisition is a big deal. You have to actually find the product and that might take all day. Maybe even a couple of if your usually supplier just got arrested.
If the search cost component of a purchase is high then the buyer will tend to want to buy in bulk. If he knows he can easily find some tomorrow then he'll just buy what he wants for a day today. But if he doesn't know he'll want to buy in bulk and stockpile.
Him having a stockpile means two things. 1. He'll probably sell some to his friends. 2. He's more likely to engage in binge usage.
It's not clear one way or another whether increasing the cost of drugs by making them harder to get will actually decrease drug usage. It might. But if you follow the line of thought above it's also clear it might not. It will tend to increase binge usage however, which almost certainly tends to make the effects of drug usage worse.
This is just an example of how simplistic arguments based on simplistic economic models of hyper-rationality lead to bad policy.
I'd never read anything by Lott before. I knew he'd written some anti-gun law stuff, but I didn't really know much about him other than that. After seeing his CSPAN presentation and seeing his comment on my blog post I went to his site and read some of his op-ed stuff linked to from there.
He has an op-ed about murder rates in Philadelphia that illustrate further how simplistic economic thinking can lead to bad policy.
Philadelphia was experienced high homicides and low homicide clearance rates. The city was arguing that they needed strong gun control to reduce the homicides. I think that's just silly, and I agree with Lott when he says it's silly. But Lott's proposed solution to the perceived problem is just as bad. He thinks they need to hire more cops.
The idea that increasing arrest rates for homocides will automatically reduce the homocide rate is something right out of the economic model of deterrence. The idea being that if you increase the probabilty of arrest you'll increase the expected cost of the crime and reduce the number of the crimes committed.
But there's actually a possibility it will have exactly the opposite effect. Why are these murders being committed? Are they over turf wars for illegal drug distribution?
If that's the case then more cops will tend to increase the immediate profits from owning drug turf. And criminals tend to have a high discount rate, putting more weight on the immediate benifet and not much wieght on the costs they might have to pay later.
If turf control becomes more valuable it may well increase the incentive for killing potential competitor drug dealers.
So, depending on what is driving the homocides, increasing clearnace rates might cause a decrease in homocides or it might not.
But it's very likely that stopping the drug war and having fewer cops will cause a decrease in Philidelphia homocides.
But of course Lott doesn't want to consider that. The model gets to complicated to fit into an op-ed.
I had said he seemed to be assuming that humans are hyper-rational, and that's a scary mindset for someone who had been in a position to influence US Sentencing Commission policy.
Lott responded.
I definitely am not assuming that people are hyper rational in the way that you describe. All I say is that if something becomes more costly, people do less of it. If you get a greater return from doing something, you do more of it. I do not say how much they change their behavior, just that they do. If the price of apples go up, everything else equal, you buy fewer apples. If the cost of committing crime increases (higher arrest rates or conviction rates or longer prison terms), fewer crimes will be committed. What is hyper rational about that? What is "scary" about that? I am not saying that all criminals will change their behavior, but that on net there will be a reduction in the amount of crime.
Of course what he says here is true. But that doesn't mean it has relevance to actually making government policy choices. And that's where the scary part comes in, he uses his simplistic assumptions and arguments to make policy.
His apple example is an example of the dangers of his hyper-rationality. "Everything else being equal" is his argument. That pretty much negates the relevance to policy determination of anything he's saying. The world is not linear, just because something might be true if everything else is held constant does not mean it's going to be true when everything else isn't actually equal, and we know it's not going to be equal.
Let's think about drug policy as an example. Everything else being equal, if we raise the price of drugs to the recreational users then they'll use less. Right?
No, that's not true. It's not true because many of the policies we use to try to raise the price of drugs tend to raise the search costs. When you're buying an illegal substance the search cost of acquisition is a big deal. You have to actually find the product and that might take all day. Maybe even a couple of if your usually supplier just got arrested.
If the search cost component of a purchase is high then the buyer will tend to want to buy in bulk. If he knows he can easily find some tomorrow then he'll just buy what he wants for a day today. But if he doesn't know he'll want to buy in bulk and stockpile.
Him having a stockpile means two things. 1. He'll probably sell some to his friends. 2. He's more likely to engage in binge usage.
It's not clear one way or another whether increasing the cost of drugs by making them harder to get will actually decrease drug usage. It might. But if you follow the line of thought above it's also clear it might not. It will tend to increase binge usage however, which almost certainly tends to make the effects of drug usage worse.
This is just an example of how simplistic arguments based on simplistic economic models of hyper-rationality lead to bad policy.
I'd never read anything by Lott before. I knew he'd written some anti-gun law stuff, but I didn't really know much about him other than that. After seeing his CSPAN presentation and seeing his comment on my blog post I went to his site and read some of his op-ed stuff linked to from there.
He has an op-ed about murder rates in Philadelphia that illustrate further how simplistic economic thinking can lead to bad policy.
Philadelphia was experienced high homicides and low homicide clearance rates. The city was arguing that they needed strong gun control to reduce the homicides. I think that's just silly, and I agree with Lott when he says it's silly. But Lott's proposed solution to the perceived problem is just as bad. He thinks they need to hire more cops.
On the other side of the spectrum, House Speaker John Perzel (R., Phila.) wants more police, with the state picking up half the cost of any new hires.
So why are Philadelphia's crime rates increasing so dramatically? To put it bluntly, the city isn't doing a very good job at law enforcement. While the arrest rate for violent crimes such as murder has fallen across the state, arrest rates have plummeted in Philadelphia. Criminals are simply not being caught. The drop has been stunning. While 81 percent of murderers were arrested in 2000, just 61 percent were arrested in 2005. And the rate has continued falling this year.
Over the next four years, Perzel's police program, if enacted, could help fund as many as 1,345 officers in Philadelphia - a 20 percent increase from today. Up to 10,000 police could be hired statewide. Hiring more police is one proven way to reverse much of the recent decline in arrest rates, though one must be careful to ensure that the money isn't diverted by localities into other activities. Perzel also seems serious about avoiding many problems that plagued President Clinton's police program, where buying items such as computers were counted as hiring police or the money was spent planting trees or doing other non-police work.
The idea that increasing arrest rates for homocides will automatically reduce the homocide rate is something right out of the economic model of deterrence. The idea being that if you increase the probabilty of arrest you'll increase the expected cost of the crime and reduce the number of the crimes committed.
But there's actually a possibility it will have exactly the opposite effect. Why are these murders being committed? Are they over turf wars for illegal drug distribution?
If that's the case then more cops will tend to increase the immediate profits from owning drug turf. And criminals tend to have a high discount rate, putting more weight on the immediate benifet and not much wieght on the costs they might have to pay later.
If turf control becomes more valuable it may well increase the incentive for killing potential competitor drug dealers.
So, depending on what is driving the homocides, increasing clearnace rates might cause a decrease in homocides or it might not.
But it's very likely that stopping the drug war and having fewer cops will cause a decrease in Philidelphia homocides.
But of course Lott doesn't want to consider that. The model gets to complicated to fit into an op-ed.
Labels: cops, nutcase economists, War on Drugs
posted by Gary Carson at 7:33 PM
2 Comments:
Thanks for your comment on my blog through to be honest I can't even find what heading you put it under. It just seems that it wasn't one of the recent posts that I put up.
1) I don't think that the everything else equal negates the usefulness of understanding that if something becomes more costly that you do less of it. You break something down into the different effects so you can find out better what effects what.
2) I don't see the problem with the search cost issue. If the price of drugs does go up (in the same way that you just discussed the price of apples), indeed less drugs would be purchased. If the price of drugs remains the same but the costs of finding drugs for sale goes up, people will buy less because fewer people will even get to the sale, but I can accept that those who do buy the drugs will make larger purchases when they do get it.
I don't really see the problem. Nor do I really understand the claim of "hyper" rationality.
All that said, I do appreciate you taking the time to look through things. Thank you.
One way the drug war increases costs is by increasing the search cost. Search cost is part of the cost of the item, there's more to the cost to the consumer than the price paid to the supplier.
(That's another problem with they standard economist oversimplification, but that's another story).
Increasing the search cost may mean fewer users, but it may mean more. Because the bulk buyers themselves become suppliers where they wouldn't be doing that if they just bought a days supply every day.
Even if it means fewer users the amounts bought in bulk may well be larger than the total purchses made if more users made small purchases.
It isn't clear at all how it all washes out, the exact functional forms of these supply/demand equations isn't clear.
Buyers and sellers are not completley disjoint, they don't act independently, and things aren't linear.
By hyper-rationality I mean you're assuming all these functional forms are simple and uncomplicated -- continious, well behaved functions, independent actors, etc.
In the deterrence model you assume that increasing future costs has no effect on immediate benefit and that criminals use a discount rate that makes those future costs relevant. They don't. And if the future costs are not independent of the current benefits your thinking tht no one will do more crime if the cost of crime goes up doesn't hold up.
I'll try to expand on this in a couple of days.
Post a Comment
<< Home