1. One of the first causal questions we examined in this class was whether or no
ID: 3312466 • Letter: 1
Question
1. One of the first causal questions we examined in this class was whether or not health insurance has a causal effect on health. Because of selection bias, simply comparing the average health of people with health insurance to the health of people without health insurance will not accurately estimate the true causal effect. A potential solutions is to use a regression discontinuity design (RD) to study a government program that gives a subsidy for health insurance to individuals bclow a certain level of incomc. Suppose the government program works as follows: 1. Individuals with an income over $20,000 do not qualify for a government subsidy 2. Individuals with an income of $20,000 or less do qualify for a S5,000 subsidy that can only be used to purchase health insurance, but individuals who qualify do not have to take the subsidy if they do not want health insurance la. Identify the outcome, running variable, and cutoff that could be used in a RD in this context Is this a sharp or fuzzy RD? lb. Why is it problematic if individuals can manipulate the running variable? It makes the RD estimate non-causal, but I want you to explain why. lc. Do you think individuals are able to finely manipulate the running variable in this context? ExplainExplanation / Answer
1.a The outcome variable will be health insurance because we are assessing the effect of government subsidy on health insurance and consequently on the health of an individual. The running variable is the income of the individuals and the cutoff is the presenence or absence of subsidy with cutoff rule as" income above $20,000" and "income below $20,000". This could be a fuzzy regression design as there might not be a sharp change in the number of people buying health insurance policies just above or just below the income level of $20,000.
1.b If individuals can manipulate the running variables then it violates the continuity assumptions and other standard identification arguments used in the analysis of regression discontinuity design break down if this condition is violated and the treatment effects determined in the analysis remain only partially identified even if some units in the study are able to manipulate their running variable. .
1.c YES. Since the individuals can change their income levels by taking up more jobs or leaving present jobs, and thus can very fairly alter the running variable which is the level of income that they have. So, the individuals can alter the running variables in this context.
Related Questions
Navigate
Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.