2
$\begingroup$

I just don't understand how the theory, particularly pertaining to parameter identification is relevant when conducting applied work. Why must we present a sound identification strategy when carrying out an applied project.I hope this is clear. Thank you.

Edit: I guess I wasn't very clear in my question. I understand how identification strategies work in applied papers. I was more interested in how, say, somebody introduces a new estimation method in an academic paper. Then they propose the conditions on which the estimated parameters are identifies using this method. There is usually some talk of the dimensions of some parameters, and technical things like that. I just wanted to know why these kinds of concerns are relevant when one wants to apply said estimator empirically.

$\endgroup$
1
$\begingroup$

I am not exactly sure what you're asking. So...I hope we both understand identification strategy to mean the same thing. If so:

The identification strategy serves as the bedrock upon which you build your argument.

For example, consider some policy implemented by a school district that focuses on improving academic outcomes for low performing students. Say this school district sets a cutoff of a 1.5 GPA. Any student at or below this GPA is placed on academic probation. This assignment to AP triggers a set of responses designed to increase a student's academic success. However, these responses are time consuming. Now, suppose at the same time that a neighboring school district with a similar GPA distribution does not implement this policy.Suppose students can move freely between the two districts.

There are several ways to analyze the efficacy of such a policy. I will consider two.

Assume researcher A decides to investigate the efficacy of the policy by simply conducting some simple trend analysis. Let's suppose he decides that it is reasonable to determine the efficacy of the policy by evaluating the amount of students with a "low" GPA both before and after the policy implementation. This seems like it could serve as an intuitively sound approach.

Let's assume now that when the school implements this academic probation policy that half the students placed on academic probation leave and begin attending the neighboring school to avoid the time demands associated with the academic probation.

The trend analysis by researcher A likely now overestimate the efficacy of the academic probation program. This is because there will be a large decrease in the number of students with a low GPA. However, this will be largely due to those students leaving the school and not because the policy helped raise their GPAs.

Suppose researcher B uses a difference-in-difference approach using the neighboring school district that does not implement the academic probation policy as a counterfactual (where the similar trends assumption holds). His identifying assumption and research strategy will allow him to identify and control for the change in composition brought about by the policy implementation.

$\endgroup$
  • $\begingroup$ I guess I wasn't very clear in my question. I understand how identification strategies work in applied papers. I was more interested in how, say, somebody introduces a new estimation method in an academic paper. Then they propose the conditions on which the estimated parameters are identifies using this method. There is usually some talk of the dimensions of some parameters, and technical things like that. I just wanted to know why these kinds of concerns are relevant when one wants to apply said estimator empirically. $\endgroup$ – Carlos Restituyo Vassallo Apr 14 '16 at 7:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.