Subject: Re: Teasing Out the Michigan Study Data
Date: Wed, 26 Mar 2014 11:53:31 -0400
From: Michael Insler <insler a t usna d o t edu>
Thanks for the email. You are exactly right in that the hardest part in tackling this question is disentangling the effects from folks who are forced into retirement (due to a health issue) from the folks who voluntarily retire. A main task in my analysis is to address this challenge. I am interested in the influence of retirement on health regardless of their motives for retiring, so it is not appropriate to outright "remove" such individuals from my sample. Instead I use a technique that requires being able to reliably predict their intention to retire, whether or not they receive a retirement-causing illness down the road. To do so,
the Michigan data contains information on subjects' retirement plans from the first period in which I see them (when they are still working).
The statistical technique is known as "two stage least squares" estimation, in which the "1st" stage is one in which I predict individuals' probability of retiring based on factors that are NOT related to their health. In the "2nd" stage, I use these "clean" predictions to capture the true impact of retirement on health.
I'm not sure how well I'm explaining these technical details, but I am very happy to answer any more questions you might have. Thanks again for your interest!
Best,
Mike
--
Michael A. Insler
Assistant Professor of Economics
United States Naval Academy
Annapolis, MD
On Tue, Mar 25, 2014 at 5:13 PM, wrote:
Dr. Insler,
I heard the interview you did for NPR concerning retirement, health/longevity.
One of the common discussions on the early-retirement.com forum site concerns this idea that as soon as the paycheck stops, people die. Most of the participants, of course, don't buy it, or we wouldn't be retired or striving to retire. The idea many of us use to dismiss the idea is that many of the people in the study quit because they were sick. Were you able to remove the people who quit because they "knew" they only had a few good years left from your analysis?