Do the problem set without limit of words and page.

**Note:

• 2b: The theory of change doesn't have to separately examine the two mechanisms through which the program was meant to operate.
• 2d-e: I've clarified that we are thinking about children's health as an outcome. Along those lines I've provided a link to a systematic review in part d(iii) that focuses on health impacts rather than education impacts. If you've already done this assuming the outcome was education impacts, please make a note of it on your problem set.

This problem set is the first of two parts; the second part will cover poverty traps that we will discuss after spring break. Each counts as half of a "regular" problem set.

It has been argued that conditional cash transfer programs, while effective, are expensive. Based on the J-PAL policy publication here: https://www.povertyactionlab.org/sites/default/files/publication call-getting-children-into-school.pdf, what is the basis for this claim?

Inspiteofrelativelypoorcost-effectivenessasameanstoincreaseschoolparticipation,ithas been argued that CCTs can be justified based on a broader set of impacts. One important im- pact is child health. In fact, the original Progresa evaluation included data collection on the program's impacts on children. The results are presented in Gertler (2000) "The Impact of Progresa on Health," available here: https://ebrary.ifpri.org/utils/getfile/collection/p15738coll2/id/125436/fil The following questions relate to the article.

(a) The paper lays out two mechanisms through which the program can impact childhealth. What are they?

(b) Sketch out a theory of change for the intervention's impacts on child health. Identify the needs and intervention, as well as outputs, intermediate outcomes, and final out- comes. Incorporate both mechanisms from part (a) into your theory of change. (A visual similar to what we discussed in class would be helpful, but is not necessary.)

(c) Before we think about the results of the paper, a quick note on methods. The paper uses a "differences in differences" methodology to estimate impacts. The Millennium Vil- lages Project evaluation had the same strategy. This compares the change in outcomes before/after Progresa was implemented between Progresa and non-Progresa villages. Because village assignment to Progresa was random, this is basically equivalent a sim- ple comparison of average outcomes between the two types of villages. In other words, you can treat the difference-in-differences estimates as if they were simple differences in averages.

The paper presents some (but not much) evidence that the theory of change "worked" as hypothesized. What is it? What additional data would be useful in tracing out the theory of change? [Note that circa 2000, RCTs focused more on final outcomes and less on processes.]

(d) Now let's think about CCTs more broadly. Suppose that the government of Haiti is considering adopting a CCT program similar to Progresa. Let's follow the steps of the generalizability framework to understand whether this is a worthwhile endeavor.

i. Theory of Change: We've already gone through the theory of change in part (b), so no need to rewrite it here.

ii. Local conditions: What information would be useful in understanding whether the underlying issues and mechanisms of change are important in the new context?

iii. General Lessons for existing evidence: What does the existing evidence say about the effectiveness of CCT interventions? Are there situations where it seems to work better than others? Or are the results consistent across contexts and variants of the programs? [The Garcia and Saavedra (2022) article may be useful here, available at https://www.nber.org/papers/w29758.]

iv. Local Implementation: What factors are important in whether the organization can successfully implement the intervention in the new context?

(e) Based on your answers to part (d), would you recommend a CCT program for Haiti? Is there additional information you would need to make your decision (e.g. specific descriptive data on the local context)? Explain.