PDA

View Full Version : What do you think about maximizing utility over time?



philecon
11-23-2010, 04:33 AM
Hi, I had a talk with one of my professors and he encouraged me to read up on dynamic optimization while I still have the time (he's assuming that I will get admitted to even one of the schools that I'm applying to). He mentioned something about it being very useful for macroeconomics, or in situations wherein you have to maximize utility over time. Then he quipped that, "then again, maximizing utility over time is pointless."

He didn't explain any further, but I surmise that it may be because preferences change over time, so the idea of maximizing one's utility function over time is pointless? I'm not too familiar with it, nor have I been acquainted with dynamic optimization, so what are your thoughts?

rthunder27
11-23-2010, 05:22 AM
I think a combination of discounting the future and the increasing uncertainty make maximizing expected utility too far into the future "pointless" . But it really depends on the application.

Andronicus
11-23-2010, 05:24 AM
Was he a microeconomist? Many microeconomists (including me) think all macro is pointless, mainly because it's boring. There are problems with standard assumptions like additive separability and exponential discounting, but there are problems with standard assumptions in static models too.

But, understanding dynamic optimization is necessary for getting through the first year. Just about everything in my first-year macro sequence used these concepts. It can show up in some micro topics also, like in labor economics and in theoretical IO.

dreck
11-23-2010, 05:41 AM
Hi, I had a talk with one of my professors and he encouraged me to read up on dynamic optimization while I still have the time (he's assuming that I will get admitted to even one of the schools that I'm applying to). He mentioned something about it being very useful for macroeconomics, or in situations wherein you have to maximize utility over time. Then he quipped that, "then again, maximizing utility over time is pointless."

He didn't explain any further, but I surmise that it may be because preferences change over time, so the idea of maximizing one's utility function over time is pointless? I'm not too familiar with it, nor have I been acquainted with dynamic optimization, so what are your thoughts?

Well he's definitely right about the fact that you'll use it in your macro core course.

I think that assuming utility functions have certain properties that are time-invariant (like a constant coefficient of absolute or relative risk aversion, or a constant discount rate) is a good assumption for modelling. But I agree that the assumptions necessary are unrealistic (see hyperbolic discounting and a million other reasons). What's arguably worse is ignoring heterogeneity of preferences between households, but that's my personal beef. But Macro is so big and complex you've got to make simplifying assumptions to get models that make any sense at all (and can be solved within this century), so we're kind of stuck....

philecon
11-24-2010, 04:49 PM
@ Andronicus: Yes, he is an applied micro prof, hence the bias against macro. He's actually pushing for replacing the third macro core course with an applied econometrics course in the graduate core at the school where I did my undergrad.

@ dreck: I understand what you mean. It just sucks, though, that the current macro models have little predictive power.

Anyway, I'll probably take a stab at reading up on dynamic optimization. What books should I look into?

Cuak2000
11-24-2010, 07:43 PM
I haven't done any serious analysis on the subject, but infinite horizon optimisation seems to me too far of a stretch and I'm inclined to doubt the conclusions that are obtained that way (I'm sure, though, that there are many ways to motivate it that would make sense)
On the other hand, dynamic optimisation over a finite period seems more reasonable to me

dreck
11-24-2010, 08:38 PM
I think it might be worth noting here that modern macro does sometimes deal with these problems...sometimes the complexity problem can be addressed using agent based modelling. But I know next to nothing about this area, so I'll stay away from commenting on it :)

Edit: also, Leonard and van Long is a pretty good book, but it's hard to find a book with everything you need in it.

Diplomer
11-24-2010, 10:18 PM
He didn't explain any further, but I surmise that it may be because preferences change over time, so the idea of maximizing one's utility function over time is pointless? I'm not too familiar with it, nor have I been acquainted with dynamic optimization, so what are your thoughts?

Well, even if preferences were stable, it appears that people do not have the cognitive skills to make a complicated calculation over a large horizon, not to mention over an infinite horizon. Honestly, people just don't reason like that, at least it would be new to me, and often they lack the computational power to do simple tasks, so how could they plan all their future decisions so as to maximize their discounted lifetime utility? That's why many people think that macro misses the point.

dinosaurus
11-24-2010, 10:46 PM
Two very common (advanced) textbooks that cover "dynamic optimization for macro theory" are "Recursive Methods in Economic Dynamics" by Lucas, Stokey, Prescott, and "Recursive Macroeconomic Theory" by Ljungqvist and Sargent. I have used them a little and while I think they are generally good, there are certainly not light reading. Maybe you can also check some Maths for Economics book, like "Further Mathematics for Economic Analysis" Snydsaeter, Hammond, Seierstad and Strom.