Instrumental variable estimation of truncated local average treatment effects

Instrumental variable (IV) analysis is used to address unmeasured confounding when comparing two nonrandomized treatment groups. The local average treatment effect (LATE) is a causal estimand that can be identified by an IV. The LATE approach is appealing because its identification relies on weaker assumptions than those in other IV approaches requiring a homogeneous treatment effect assumption. If the instrument is confounded by some covariates, then one can use a weighting estimator, for which the outcome and treatment are weighted by instrumental propensity scores. The weighting estimator for the LATE has a large variance when the IV is weak and the target population, i.e., the compliers, is relatively small. We propose a truncated LATE that can be estimated more reliably than the regular LATE in the presence of a weak IV. In our approach, subjects who contribute substantially to the weak IV are identified by their probabilities of being compliers, and they are removed based on a pre-specified threshold. We discuss interpretation of the proposed estimand and related inference method. Simulation and real data experiments demonstrate that the proposed truncated LATE can be estimated more precisely than the standard LATE.

Evidence of Neighborhood Effects from Moving to Opportunity: Lates of Neighborhood Quality

This paper estimates neighborhood effects on adult labor market outcomes using the Moving to Opportunity (MTO) housing mobility experiment. We propose and implement a new strategy for identifying transition-specific effects that exploits identification of the unobserved component of a neighborhood choice model. Estimated local average treatment effects (LATEs) are large, result from moves between the first and second deciles of the national distribution of neighborhood quality, and pertain to a subpopulation of nine percent of program participants.

The deterrent effects of Brazilian child labor law

Brazilian law prohibits all forms of work for children under the age of 14 years old. Therefore, work performed by children under 14 years of age is subject to sanctions that do not apply to work performed by those over 14 years of age. We use this quasi-experiment generated by Brazilian law to test the deterrent effects of such sanctions. For this purpose, we use the 2013 Pesquisa Nacional de Amostra por Domicílios (PNAD) data to estimate the local average treatment effects (LATE) using the regression discontinuity approach. The results indicate that on average, this law results in 88% fewer weekly working hours when individuals living in rural and urban areas are considered. The effects in rural areas are thus inconclusive. The paper concludes that the law has a deterrent effect and reduces child labor in Brazil, but the effects are ambiguous in rural areas, where law enforcement is weaker and more children work.

Collective property rights reduce deforestation in the Brazilian Amazon

In this paper, we draw on common-pool resource theory to argue that indigenous territories, when granted full property rights, will be effective at curbing deforestation. Using satellite data, we test the effect of property rights on deforestation between 1982 and 2016. In order to identify causal effects, we combine a regression discontinuity design with the orthogonal timing of homologation. We find that observations inside territories with full property rights show a significant decrease in deforestation, while the effect does not exist in territories without full property rights. While these are local average treatment effects, our results suggest that not only do indigenous territories serve a human-rights role, but they are a cost-effective way for governments to preserve their forested areas. First, obtaining full property rights is crucial to recognize indigenous peoples’ original right to land and protect their territories from illegal deforestation. Second, when implemented, indigenous property rights reduce deforestation inside indigenous territories in the Amazon rainforest, and could provide an important positive externality for Brazil and the rest of the world in terms of climate change mitigation.

Dual Language Education and Student Achievement

Two-way dual language (DL) classrooms enroll students of two different language backgrounds and teach curriculum in both languages. I estimate the effect of attending a dual language school on student achievement using school choice lotteries from Charlotte Mecklenburg School District, finding local average treatment effects of 0.04 and 0.05 standard deviations per year in math and reading, respectively. Attending a DL school increased test scores for English learners (ELs) and for non-ELs by a similar magnitude. The positive effects of winning the lottery to attend a DL school are substantially larger than the average effect of assignment to other magnet schools, and several peer and school characteristics are ruled out as explanations for the increased test scores. There is no statistically significant evidence that attending a dual language school changed the probability of having EL status throughout elementary school.

Too LATE for Natural Experiments: A Critique of Local Average Treatment Effects Using the Example of Angrist and Evans (1998)

There has been a fundamental flaw in the conceptual design of many natural experiments used in the economics literature, particularly among studies aiming to estimate a local average treatment effect (LATE). When we use an instrumental variable (IV) to estimate a LATE, the IV only has an indirect effect on the treatment of interest. Such IVs do not work as intended and will produce severely biased and/or uninterpretable results. This comment demonstrates that the LATE does not work as previously thought and explains why using the natural experiment proposed by Angrist and Evans (1998) as the example.

Inference on finite-population treatment effects under limited overlap

Summary This paper studies inference on finite-population average and local average treatment effects under limited overlap, meaning that some strata have a small proportion of treated or untreated units. We model limited overlap in an asymptotic framework, sending the propensity score to zero (or one) with the sample size. We derive the asymptotic distribution of analogue estimators of the treatment effects under two common randomization schemes: conditionally independent and stratified block randomization. Under either scheme, the limit distribution is the same and conventional standard error formulas remain asymptotically valid, but the rate of convergence is slower the faster the propensity score degenerates. The practical import of these results is two-fold. When overlap is limited, standard methods can perform poorly in smaller samples, as asymptotic approximations are inadequate owing to the slower rate of convergence. However, in larger samples, standard methods can work quite well even when the propensity score is small.