Health Econ Problem Set 2

This pertains to the demand for health care.
1. Draw a vertical demand curve on a graph.
2. Using the formula for the elasticity of demand, show why this is a perfectly inelastic demand?
3. Can subsidy used to promote access to health care work if the demand is perfectly inelastic? Why or why not?
4. Next, consider the demand equation $q=-\frac{1}{4}\times p + 200$, where $q$ is the number of doctor visits, p is price. Plot this demand curve.
5. Calculate the point elasticity of demand when $p=\$5$ and $p=\$100$, respectively.
6. Comment on the elasticity estimates from part e.

$~$

The demand for doctor visits can be written as:

$doc\;visits=g(Insurance,\;Education,\;Income,\;Demographic,\;\epsilon)$,

where $\epsilon$ contains unobserved factors that a researcher cannot perceive. One example is risk preference – as a researcher you cannot observe the risk preference of an individual. Now, assuming a linear functional form, you can express the demand for doctor visits as:

$doc\;visits=\alpha + \beta Insurance + \gamma Education + \delta income + \kappa Demographic + \epsilon$.

After estimating the regression, say you get $\beta>0$. Does this mean that having an insurance increases doctor visits compared to those without insurance? Explain.
Based on part a. can you say that insurance has a causal effect on increasing doc;visits? Why or why not?
Based on the evidence that insured individuals face lower price compared to uninsured and insured individuals are more likely to go to the doctor ($\beta>0$), can you state that the demand for doctor visits is downward sloping? Why or why not?

$~$

This problem pertains to the randomized control trial. Say, you want to find out whether the demand for doctor visits is downward sloping among young adults. To do so, you do a lottery and randomly assign health insurance for 25 people, who are technically termed as the ``treated group." The other 25 people who did not receive insurance are control group.
1. Consider the specification: $visits=\alpha + \beta Treat + \epsilon$.
Where $visits$ represent the total number of doctor visits, $Treat$ is the group that was assigned insurance through the lottery draw, and $\epsilon$ is the error term. Say, after estimating this specification, you find that $\beta>0$. What does this suggest?
1. Does the estimate of $\beta>0$ (in this case of randomization) suggest that there is a causal relationship between getting insured and increased doctor visits? Why or why not?
2. Now, in a different analysis based on the survey data, you find that those insured tend to have higher doctor visits. From this finding can you say that insurance leads to more doctor visits? Explain in resonance to your answer from part b.
3. Briefly describe the Oregon Health Insurance experiment.

$~$

Card et al. (2008) uses a natural experiment to identify the responsiveness to prices on health care and services.
1. Explain how eligibility criteria set for Medicare can be used as a quasi natural experiment.
2. By using a figure that corresponds to the eligibility criteria (age), briefly describe the findings of their study.
3. What are some potential drawbacks of their study?