Sunday, September 04, 2005

Why won't empirical study solve policy issues?

Kash defines liberalism:
So if I had to encapsulate in a few words why I describe myself as a liberal, I would simply say this: I believe in bad luck. I think that a huge number of the forces that affect most people's lives are outside of their control - the parents that they were born to, the quality of their local educational opportunities, the management of the company that they happen to work for, the fortunes of the city or town in which they happen to live, or the industry in which they happen to find work - and that individuals who suffer from a bad family, poor education, being laid off, or a hurricaine, should not be left to live with the consequences of their plain bad luck without help from society at large.
That's certainly a fair characterization of the liberal ideology. But isn't the basic issue here - "a huge number of the forces that affect most people's lives are outside of their control" - empirically resolvable? Perhaps not as a whole, but in pieces. I think there are two pieces to be puzzle that have to be resolved separately.

Stripping the issue down to its basics, liberals think that most of a person's well-being is determined by luck, conservatives think that most of a person's well-being is determined by choice. The welfare debate reflects this. Liberals argue that people on welfare are there only due to circumstance. Conservatives argue that people on welfare are there because they have chosen not to try hard. But this is just an empirical argument. You could design a study to see who was correct. I think that if the liberals are right, welfare should exist out of compassion for those who were simply unlucky. If the conservatives are right, welfare shouldn't exist because it reduces the incentive to expend effort. The normative arguments are essentially correct, it's just the positive, empirical arguments that are called into question.

Likely, there are people on welfare for both reasons. This would necessitate some sort of system that could sort out the types of people. That could be very hard. The point is, we don't want to distort any of the natural incentives that govern decision-making while helping those who, purely by luck, became immiserated. But almost all issues have some mix of luck and choice and so the normative issues are also very difficult. Even the relief after Hurricane Katrina has fallen into this sort of debate. Some say that people should have realized they were living in a potential deathtrap. But nearly all decisions, like the decision about where to live, are made under uncertainty. Conversely, nearly all events have some element of choice, even Katrina. Consider the problem of people building in hurricane-prone areas, collecting on insurance and rebuilding each year after the house is blown away. That is so clearly wrong while sending help to New Orleans is so clearly right.

How do we know where the line should be drawn? Whenever you are bailing someone out there is always an element of moral hazard. Yet it is impossible to be morally ignorant and not help those in need. So we have a tough normative issue in addition to a separate empirical issue. To the degree that liberals and conservatives agree on where to draw the normative line (and I think that degree is high), empirical study could seemingly solve most debates.

But in practice, it doesn't. Why? We can think of ideology as a sort of Bayesian prior. If I'm liberal it's going to take a mountain of evidence to convince me that there are "welfare queens," but if I'm conservative it'll just take a few anecdotes. To sum up: you could explain different policy choices between parties by assuming that one or other other is irrational and the other rational, or you could assume both are rational but have different objective functions (drawing the line differently), but all you really need is that the parties have different priors or initial beliefs.


Blogger Alcuin Bramerton said...

On the subject of "ideology as a sort of Bayesian prior", a Norfolk koan comes to mind:

A man called Bentley
Wants to enrol
In a post-graduate course
In joined-up thinking
At Oxford University.

He goes to a public lavatory
In Berlin
To speak to the academic registrar.
The academic registrar is not there.
The academic registrar
Is in a public lavatory
In Oxford.

"Why do you want to enrol
In a post-graduate course
In joined-up thinking
At Oxford University?"
He is asked.

"Because I am impressed
With the blue colour
Of yellow daffodils
As perceived by Polar Bears
In the Sahara Dessert,"
he replies.

"When you say Sahara Dessert,
Do you mean Sahara Desert?"
"Yes, I mean Kalahari main course,"
he replies,
"With noodles
And pachyderm depth psychology."

"Polar Bears
Are not pachyderms,"
Says the academic registrar.

"Some Polar Bears
May exhibit an internal predisposition
To pachydermatous behaviours
When aroused by joined-up thinking."

"What is your academic background,
Mr Bentley?"
"Zoology, Botany
And Crustacean Flower Arranging."

"Do you have an IQ?"
"I have several IQs"
"Which is your favourite IQ?"
"My favourite IQ is 156.
That is the IQ
Of a Giant Panda
On the crest
Of a whole-body orgasm."

"Your manner of speech is curious,
Mr Bentley."

"May I have an application form?"
"What kind of application form
Would you like?"
"I would like an application form
For post-graduate entry
To the select course
In joined-up thinking
At Oxford University."

"Would you like the application form
To have coloured pictures on it?"
"No. I would like the application form
To have black-and-white pictures on it
Which I can colour-in with wax crayons."

"At Oxford University,
You are allowed a choice
Of a special free gift
To go with your application form,"
Says the academic registrar.

"Thank you.
I would like a bee orchid
To go with my application form."
"Of course."
"Or a vegetable object
That is bee-orchid-like
In its visual characteristics,
Such as a tennis shoe
Or a traction engine."

"You have a fine mind, Mr Bentley."

5:19 AM  
Blogger Isaac said...

That has to be one of the weirder things I've ever read...

I guess I had never thought of labelling Kuhn a Bayesian, but he is (we hold onto our priors until it is completely impossible to, because a better explanation has been introduced that explains the anomalies). Ever more reason to take 111...

12:17 PM  
Blogger henry said...

Well, that's certainly a first.

2:19 PM  
Blogger David Schraub said...

Consider this the low-tech equivilant of a trackback.

6:12 PM  
Blogger Jim Satterfield said...

It's not whether "welfare queens" exist. It's about whether or not they are a majority or at least a large enough number of the people receiving benefits to justify not having welfare. One of the main differences affecting the different attitude between liberals and conservatives is also a belief in the ability of the system to provide everyone who can work a job that pays enough for them to live on. I've never seen any evidence that it does.

9:45 PM  

Post a Comment

<< Home