Bryan Caplan  

Do Econlog Comments Pass the Rational Expectations Test?

PRINT
Jason Furman's Health Care Pla... Robert Fogel Interview...

Yes! The correct answer to the ice cream demand question is that sales went up by 200%. Here's the histogram of reader's responses:
icecream.jpg
When people gave a range, I took the midpoint. When they gave a lower or upper bound, I took the boundary. Thus, "more than x" and "less than x" both got coded as x - I couldn't think of any better solution.

Here are the summary stats:
icecream2.jpg
If you do a simple regression of the responses on a constant, the constant equals 173.3, with a standard error of 70.1. So we can't reject the rational expectations hypothesis that the constant equals 200.

As an added bonus, three readers were dead on the money. Well done.


Comments and Sharing






COMMENTS (8 to date)
Gabriel M. writes:

I think it makes sense to eliminate those two outliers on the right tail.

With those extreme values removes, the starts will look even better.

economister writes:

Note that observations are not independent: the last commenters could see all previous comments (the first published comment was 200%).

Even though it is not a pure information cascade, you cannot ignore some cascading....

Swimmy writes:

It's true that there's some cascading, but I think that's an implication of rational expectations anyway: they don't exist in a vacuum.

Brandon Berg writes:

Eliminating the outliers would push the mean down to around 80%.

SheetWise writes:

Your initial question was -

It was 85 degrees in Fairfax last Monday. When I bought an ice cream cone at Ben and Jerry's, I asked the owner how much extra business he was doing that day compared to the typical day in January.

What makes you think the weather had anything to do with it? Is it possible more people were out enjoying the good weather? Was this ice cream shop located in a place people would go to enjoy the good weather? Did ice cream shops in the malls have increased sales?

I would still posit that the sales were correlated to traffic. They could have been generated by a well publicized Elvis sighting in the area as easily as the weather.

Carl Marks writes:

just because you asked the question I believed the answer would be different from what would typically be expected.

Remember, you can't observe something without changing it.

Jacob Kearns writes:

I declare myself the winner! Do I get a prize?

SheetWise writes:

Jacob -

Of course you don't win the prize -- hands down, I win.

I'll have a double chocolate mint dip, thank you.

Comments for this entry have been closed
Return to top