Friday, June 6, 2014

A question about health insurance

James asked:

I saw your Youtube video on Demand for Insurance and it led me to here. With that said, there are many articles written about health coverage choice e.g.National Institute of Health:
Beginning on page 5 and,

Urban institute:
Beginning on page 12.
RAND Compare has an overly complex model, but for discussion purposes, will use NIH's example.
NIH is taking into consideration (i) Risk aversion utility maximization (ii) expected healthcare needs (iii) variance in OOP cost from a set of health plans with consideration of variance in premiums for the health plans.

Is there a model or someone who can help me create a supplemental excel model that reflects the NIH's Model? In this model, they are using limited data such as Age, Income, and gender to impute a cost (lognorm function?)

There is so much written about this subject, however, I am having a very hard time modeling what is being discussed. If you could point me in a direction to someone who can help or may be interested in taking this on as a project, I would appreciate your feedback.

My response:

The video of mine that you mention just considers risk aversion and then only in the two-state case: loss or no-loss. So it is done very simply in that model.  I have another video on adverse selection, that looks at the problem graphically in a simple model and one on moral hazard that does it algebraically.  (It is really a model of the principal and agent, but that is essentially the same as an insurance model with moral hazard.) 

The MIT paper that you mention puts all these factors together in one model by considering two periods, where in the second period there is moral hazard, given the insurance policy choice in the first period.  Then, understanding the second period solution, they fold it back and look at the selection problem in the first period.  And the do this noting that demographic characteristics of the insurance purchaser will impact both the moral hazard and the selection problems.  This is substantially more complex then any of my videos.  It is also more realistic.   

Whether you can build an Excel model that imitates the MIT paper, I'm not sure.  But why would you want to do that?  To understand the MIT paper because it is too hard to work through the equations that are presented there?  Please not that papers published in the AER are meant for an audience of professional economists and assume the reader has crossed a prior threshold of understanding - both on the issues and the math modeling.  I would not teach this paper to an undergraduate class.  


  1. Right, this paper has a very specific objective and one that doesn't concern me that you had addressed, i.e. "folding back" to re-assess. The reason this interest me is that it provides a means to assign one choice out of many based on utility and adverse selection for individuals characterized only by age/income/and gender given the trade-off of premium vs lower cost share. Thank you so much for your time and videos. I will definitely continue to review your videos.

  2. Thank you so much for your time. I would like to have something MUCH more simple that allows someone to choose a plan out of many health plans based on an individuals distinguishing characteristics (age/sex/income) by imputing risk aversion and utility. I'm trying to learn how to create a model that can use expected usage, (e.g. expected health cost, or minutes on a cell phone or kilowatts for energy) can determine whether someone pays for a high utilizer plan or low utilizer plan (but more than 2 plans.) I like the MIT article because it imputes a lot of unique characteristics based on limited data (age/sex/income) and wallah! chooses a plan.

  3. I'm afraid that neither the adverse selection issues nor the moral hazard issues are conceptually easy to understand, but if you have a model that ignores those elements you miss some of the essential economics that is going on.