Literature on Recent Advances in

Applied Micro Methods∗

Christine Cai†

November 28, 2020

CLICK HERE FOR THE MOST RECENT VERSION

Contents
1 OLS 2

2 RCT 5

3 Diff-in-Diff & Event Studies 6

4 Standard IV 12

5 Shift-Share IV 16

6 RDD 17

7 Synthetic Control 22

8 Matching 24

9 Bunching 25

10 Decomposition 26

11 General 27
∗
The references listed here are mainly from the applied econometrics courses taught by Michal Kolesár
(ECO 539B, Fall 2019, Princeton University) and Arindrajit Dube (ECON 797B, Fall 2020, UMass Amherst).
I also added papers that have circulated on #EconTwitter and those that have been suggested to me after I
shared the first version of this draft. In particular, I thank Kirill Borusyak, Otavio Canozzi Conceição, Dylan
(@dc_pov), Kevin Grier, Sean Higgins, Pedro Picchetti, Esteban Quiñones, Jorge Rodríguez, and Ben Solow
for their suggestions. Additional suggestions are welcome (“more is better” after all ,).
†
Princeton University, Department of Economics. Email: [email protected]

1
1 OLS
• Acharya, Blackwell, and Sen (2016), “Explaining Causal Findings Without Bias:
Detecting and Assessing Direct Effects,” APSR
“Researchers seeking to establish causal relationships frequently control for variables
on the purported causal pathway, checking whether the original treatment effect then
disappears. Unfortunately, this common approach may lead to biased estimates. In
this article, we show that the bias can be avoided by focusing on a quantity of inter-
est called the controlled direct effect. Under certain conditions, the controlled direct
effect enables researchers to rule out competing explanations–an important objective
for political scientists. To estimate the controlled direct effect without bias, we describe
an easy-to-implement estimation strategy from the biostatistics literature. We extend
this approach by deriving a consistent variance estimator and demonstrating how to
conduct a sensitivity analysis. Two examples–one on ethnic fractionalization’s effect
on civil war and one on the impact of historical plough use on contemporary female
political participation–illustrate the framework and methodology.”

• Ibragimov and Müller (2016), “Inference with Few Heterogeneous Clusters,” RE-
Stat
“Suppose estimating a model on each of a small number of potentially heterogeneous
clusters yields approximately independent, unbiased, and Gaussian parameter estima-
tors. We make two contributions in this setup. First, we show how to compare a scalar
parameter of interest between treatment and control units using a two-sample t-statistic,
extending previous results for the one-sample t-statistic. Second, we develop a test for
the appropriate level of clustering; it tests the null hypothesis that clustered standard
errors from a much finer partition are correct. We illustrate the approach by revisiting
empirical studies involving clustered, time series, and spatially correlated data.”

• Imbens and Kolesár (2016), “Robust Standard Errors in Small Samples: Some
Practical Advice,” REStat
“We study the properties of heteroskedasticity-robust confidence intervals for regression
parameters. We show that confidence intervals based on a degrees-of-freedom correc-
tion suggested by Bell and McCaffrey (2002) are a natural extension of a principled ap-
proach to the Behrens-Fisher problem. We suggest a further improvement for the case
with clustering. We show that these standard errors can lead to substantial improve-
ments in coverage rates even for samples with fifty or more clusters.We recommend
that researchers routinely calculate the Bell-McCaffrey degrees-of-freedom adjustment

2
 to assess potential problems with conventional robust standard errors.”

• Abadie, Athey, Imbens, and Wooldridge (2017), “When Should You Adjust Stan-
dard Errors for Clustering?,” NBER WP
“In empirical work in economics it is common to report standard errors that account
for clustering of units. Typically, the motivation given for the clustering adjustments is
that unobserved components in outcomes for units within clusters are correlated. How-
ever, because correlation may occur across more than one dimension, this motivation
makes it difficult to justify why researchers use clustering in some dimensions, such as
geographic, but not others, such as age cohorts or gender. It also makes it difficult to
explain why one should not cluster with data from a randomized experiment. In this
paper, we argue that clustering is in essence a design problem, either a sampling design
or an experimental design issue. It is a sampling design issue if sampling follows a two
stage process where in the first stage, a subset of clusters were sampled randomly from
a population of clusters, while in the second stage, units were sampled randomly from
the sampled clusters. In this case the clustering adjustment is justified by the fact that
there are clusters in the population that we do not see in the sample. Clustering is an
experimental design issue if the assignment is correlated within the clusters. We take
the view that this second perspective best fits the typical setting in economics where
clustering adjustments are used. This perspective allows us to shed new light on three
questions: (i) when should one adjust the standard errors for clustering, (ii) when is
the conventional adjustment for clustering appropriate, and (iii) when does the conven-
tional adjustment of the standard errors matter.”

• Canay, Santos, and Shaikh (2018), “The Wild Bootstrap with a “Small” Number of
“Large” Clusters,” REStat
“This paper studies the wild bootstrap-based test proposed in Cameron et al. (2008).
Existing analyses of its properties require that number of clusters is “large.” In an
asymptotic framework in which the number of clusters is “small,” we provide condi-
tions under which an unstudentized version of the test is valid. These conditions include
homogeneity-like restrictions on the distribution of covariates. We further establish that
a studentized version of the test may only over-reject the null hypothesis by a “small”
amount that decreases exponentially with the number of clusters. We obtain qualita-
tively similar result for “score” bootstrap-based tests, which permit testing in nonlinear
models.”

• Cattaneo, Jansson, and Newey (2018), “Inference in Linear Regression Models

with Many Covariates and Heteroscedasticity,” JASA

3
 “The linear regression model is widely used in empirical work in economics, statistics,
and many other disciplines. Researchers often include many covariates in their linear
model specification in an attempt to control for confounders. We give inference meth-
ods that allow for many covariates and heteroscedasticity. Our results are obtained
using high-dimensional approximations, where the number of included covariates is
allowed to grow as fast as the sample size. We find that all of the usual versions of
Eicker–White heteroscedasticity consistent standard error estimators for linear models
are inconsistent under this asymptotics. We then propose a new heteroscedasticity con-
sistent standard error formula that is fully automatic and robust to both (conditional)
heteroscedasticity of unknown form and the inclusion of possibly many covariates. We
apply our findings to three settings: parametric linear models with many covariates, lin-
ear panel models with many fixed effects, and semiparametric semi-linear models with
many technical regressors. Simulation evidence consistent with our theoretical results is
provided, and the proposed methods are also illustrated with an empirical application.
Supplementary materials for this article are available online.”

• Abadie, Athey, Imbens, and Wooldridge (2020), “Sampling-Based Versus Design-

Based Uncertainty in Regression Analysis,” ECMA
“Consider a researcher estimating the parameters of a regression function based on data
for all 50 states in the United States or on data for all visits to a website. What is the
interpretation of the estimated parameters and the standard errors? In practice, re-
searchers typically assume that the sample is randomly drawn from a large population
of interest and report standard errors that are designed to capture sampling variation.
This is common even in applications where it is difficult to articulate what that popu-
lation of interest is, and how it differs from the sample. In this article, we explore an
alternative approach to inference, which is partly design-based. In a design-based set-
ting, the values of some of the regressors can be manipulated, perhaps through a policy
intervention. Design-based uncertainty emanates from lack of knowledge about the val-
ues that the regression outcome would have taken under alternative interventions. We
derive standard errors that account for design-based uncertainty instead of, or in addi-
tion to, sampling-based uncertainty. We show that our standard errors in general are
smaller than the usual infinite-population sampling-based standard errors and provide
conditions under which they coincide.”

• Colella, Lalive, Sakalli, and Thoenig (2020), “Inference with Arbitrary Clustering,”
WP
“Analyses of spatial or network data are now very common. Nevertheless, statistical

4
 inference is challenging since unobserved heterogeneity can be correlated across neigh-
boring observational units. We develop an estimator for the variance-covariance matrix
(VCV) of OLS and 2SLS that allows for arbitrary dependence of the errors across obser-
vations in space or network structure and across time periods. As a proof of concept,
we conduct Monte Carlo simulations in a geospatial setting based on U.S. metropoli-
tan areas. Tests based on our estimator of the VCV asymptotically correctly reject the
null hypothesis, whereas conventional inference methods, e.g., those without clusters
or with clusters based on administrative units, reject the null hypothesis too often. We
also provide simulations in a network setting based on the IDEAS structure of coauthor-
ship and real-life data on scientific performance. The Monte Carlo results again show
that our estimator yields inference at the correct significance level even in moderately
sized samples and that it dominates other commonly used approaches to inference in
networks. We provide guidance to the applied researcher with respect to (i) whether or
not to include potentially correlated regressors and (ii) the choice of cluster bandwidth.
Finally, we provide a companion statistical package (acreg) enabling users to adjust the
OLS and 2SLS coefficient’s standard errors to account for arbitrary dependence.”

• Słoczyński (2020), “Interpreting OLS Estimands When Treatment Effects Are Het-
erogeneous: Smaller Groups Get Larger Weights,” REStat
“Applied work often studies the effect of a binary variable (“treatment”) using linear
models with additive effects. I study the interpretation of the OLS estimands in such
models when treatment effects are heterogeneous. I show that the treatment coefficient
is a convex combination of two parameters, which under certain conditions can be in-
terpreted as the average treatment effects on the treated and untreated. The weights on
these parameters are inversely related to the proportion of observations in each group.
Reliance on these implicit weights can have serious consequences for applied work, as I
illustrate with two well-known applications. I develop simple diagnostic tools that em-
pirical researchers can use to avoid potential biases. Software for implementing these
methods is available in R and Stata. In an important special case, my diagnostics only
require the knowledge of the proportion of treated units.”

2 RCT
• Muralidharan, Romero, and Wüthrich (2019), “Factorial Designs, Model Selection,
and (Incorrect) Inference in Randomized Experiments,” NBER WP
“Cross-cutting or factorial designs are widely used in field experiments. Standard t-tests

5
 using the fully-saturated “long” model provide valid inference on the main treatment ef-
fects and all interactions. However, t-tests using a “short”model (without interactions)
yield greater power for inference on the main treatment effects if the interactions are
zero. We show that the assumption of zero interactions is problematic and leads to a
significant increase in incorrect inference regarding the main treatment effects relative
to a “business as usual” counterfactual. Further, we show that pre-testing the inter-
actions and ignoring them if they are not significant also leads to incorrect inference
(due to the implied model selection). We examine econometric approaches to improve
power relative to the long model while controlling size for all values of the interaction.
Modest “local” power improvements are possible, but come at the cost of lower power
for most values of the interaction. For the design of new experiments, an alternative is
to leave the interaction cells empty. This design-based approach yields global power im-
provements while controlling size and we recommend it for policy experiments where
a “business as usual” counterfactual is especially important.”

• Young (2019), “Channeling Fisher: Randomization Tests and the Statistical In-
significance of Seemingly Significant Experimental Results,” QJE
“I follow R. A. Fisher’s, The Design of Experiments (1935), using randomization statis-
tical inference to test the null hypothesis of no treatment effects in a comprehensive
sample of 53 experimental papers drawn from the journals of the American Economic
Association. In the average paper, randomization tests of the significance of individual
treatment effects find 13% to 22% fewer significant results than are found using au-
thors’ methods. In joint tests of multiple treatment effects appearing together in tables,
randomization tests yield 33% to 49% fewer statistically significant results than conven-
tional tests. Bootstrap and jackknife methods support and confirm the randomization
results.”

3 Diff-in-Diff & Event Studies1

• Brewer, Crossley, and Joyce (2017), “Inference with Difference-in-Differences Re-
visited,” JEM
“A growing literature on inference in difference-in-differences (DiD) designs has been
pessimistic about obtaining hypothesis tests of the correct size, particularly with few
groups. We provide Monte Carlo evidence for four points: (i) it is possible to obtain
tests of the correct size even with few groups, and in many settings very straightfor-

1
See also Andrew Baker’s blog posts on diff-in-diff methods and event-study analysis.

6
 ward methods will achieve this; (ii) the main problem in DiD designs with grouped
errors is instead low power to detect real effects; (iii) feasible GLS estimation combined
with robust inference can increase power considerably whilst maintaining correct test
size – again, even with few groups, and (iv) using OLS with robust inference can lead
to a perverse relationship between power and panel length.”

• Athey and Imbens (2018), “Design-Based Analysis in Difference-In-Differences Set-

tings with Staggered Adoption,” NBER WP
“In this paper we study estimation of and inference for average treatment effects in a
setting with panel data. We focus on the setting where units, e.g., individuals, firms, or
states, adopt the policy or treatment of interest at a particular point in time, and then
remain exposed to this treatment at all times afterwards. We take a design perspective
where we investigate the properties of estimators and procedures given assumptions on
the assignment process. We show that under random assignment of the adoption date
the standard Difference-In-Differences estimator is is an unbiased estimator of a partic-
ular weighted average causal effect. We characterize the properties of this estimand,
and show that the standard variance estimator is conservative.”

• Borusyak and Jaravel (2018), “Revisiting Event Study Designs, with an Application
to the Estimation of the Marginal Propensity to Consume,” WP
“A broad empirical literature uses “event study” research designs for treatment effect
estimation, a setting in which all units in the panel receive treatment but at random
times. We make four novel points about identification and estimation of causal effects
in this setting and show their practical relevance. First, we show that in the presence
of unit and time fixed effects, it is impossible to identify the linear component of the
path of pre-trends and dynamic treatment effects. Second, we propose graphical and
statistical tests for pre-trends. Third, we consider commonly-used “static” regressions,
with a treatment dummy instead of a full set of leads and lags around the treatment
event, and we show that OLS does not recover a reasonable weighted average of the
treatment effects: long-run effects are weighted negatively, and we introduce different
estimators robust to this issue. Fourth, we show that equivalent problems of under-
identification and negative weighting arise in difference-in-differences settings when the
control group is allowed to be on a different time trend or in the presence of unit-specific
time trends. We show the practical relevance of these issues in a series of examples
from the existing literature. We focus on the estimation of the marginal propensity
to consume out of tax rebates: according to our preferred specification, the marginal
propensity to consume is much lower than (about half of) the main estimates in the

7
 literature. The main message for practitioners is that because of identification issues
and negative weighting in event study designs, results from common specifications are
likely to seem non-robust. These problems can be alleviated in a principled way by
using parametric and semi-parametric estimators and tests.”

• de Chaisemartin and d’Haultfoeuille (2018), “Fuzzy Differences-in-Differences,”

REStud
“Difference-in-differences (DID) is a method to evaluate the effect of a treatment. In its
basic version, a “control group” is untreated at two dates, whereas a “treatment group”
becomes fully treated at the second date. However, in many applications of the DID
method, the treatment rate only increases more in the treatment group. In such fuzzy
designs, a popular estimator of the treatment effect is the DID of the outcome divided
by the DID of the treatment. We show that this ratio identifies a local average treatment
effect only if the effect of the treatment is stable over time, and if the effect of the
treatment is the same in the treatment and in the control group. We then propose two
alternative estimands that do not rely on any assumption on treatment effects, and that
can be used when the treatment rate does not change over time in the control group.
We prove that the corresponding estimators are asymptotically normal. Finally, we use
our results to reassess the returns to schooling in Indonesia.”

• Arkhangelsky and Imbens (2019), “Double-Robust Identification for Causal Panel

Data Models,” WP
“We study identification and estimation of causal effects of a binary treatment in set-
tings with panel data. We highlight that there are two paths to identification in the
presence of unobserved confounders. First, the conventional path based on making
assumptions on the relation between the potential outcomes and the unobserved con-
founders. Second, a design-based path where assumptions are made about the relation
between the treatment assignment and the confounders. We introduce different sets
of assumptions that follow the two paths, and develop double robust approaches to
identification where we exploit both approaches, similar in spirit to the double robust
approaches to estimation in the program evaluation literature.”

• Cengiz, Dube, Lindner, and Zipperer (2019), “The Effect of Minimum Wages on
Low-Wage Jobs,”2 QJE
“We estimate the effect of minimum wages on low-wage jobs using 138 prominent state-
2
Even though its title makes it sound like it is irrelevant, this paper has been added because it describes
another method to deal with heterogeneous treatment effects in event-study designs, by using stacked
diff-in-diff by event – see Online Appendix G of that paper for more detail.

8
 level minimum wage changes between 1979 and 2016 in the United States using a
difference-in-differences approach. We first estimate the effect of the minimum wage
increase on employment changes by wage bins throughout the hourly wage distribution.
We then focus on the bottom part of the wage distribution and compare the number of
excess jobs paying at or slightly above the new minimum wage to the missing jobs
paying below it to infer the employment effect. We find that the overall number of low-
wage jobs remained essentially unchanged over the five years following the increase. At
the same time, the direct effect of the minimum wage on average earnings was amplified
by modest wage spillovers at the bottom of the wage distribution. Our estimates by
detailed demographic groups show that the lack of job loss is not explained by labor-
labor substitution at the bottom of the wage distribution. We also find no evidence of
disemployment when we consider higher levels of minimum wages. However, we do
find some evidence of reduced employment in tradeable sectors. We also show how
decomposing the overall employment effect by wage bins allows a transparent way of
assessing the plausibility of estimates.”

• Ferman and Pinto (2019), “Inference in Differences-in-Differences with Few Treated

Groups and Heteroskedasticity,” REStat
“We derive an inference method that works in differences-in-differences settings with
few treated and many control groups in the presence of heteroskedasticity. As a leading
example, we provide theoretical justification and empirical evidence that heteroskedas-
ticity generated by variation in group sizes can invalidate existing inference methods,
even in data sets with a large number of observations per group. In contrast, our in-
ference method remains valid in this case. Our test can also be combined with feasible
generalized least squares, providing a safeguard against misspecification of the serial
correlation.”

• Freyaldenhoven, Hansen, and Shapiro (2019), “Pre-event Trends in the Panel Event-
Study Design,” AER
“We consider a linear panel event-study design in which unobserved confounds may be
related both to the outcome and to the policy variable of interest. We provide sufficient
conditions to identify the causal effect of the policy by exploiting covariates related to
the policy only through the confounds. Our model implies a set of moment equations
that are linear in parameters. The effect of the policy can be estimated by 2SLS, and
causal inference is valid even when endogeneity leads to pre-event trends (“pre-trends”)
in the outcome. Alternative approaches perform poorly in our simulations.”

• Abraham and Sun (2020), “Estimating Dynamic Treatment Effects in Event Studies

9
 with Heterogeneous Treatment Effects,” JE
“To estimate the dynamic effects of an absorbing treatment, researchers often use two-
way fixed effects regressions that include leads and lags of the treatment. We show that
in settings with variation in treatment timing across units, the coefficient on a given lead
or lag can be contaminated by effects from other periods, and apparent pretrends can
arise solely from treatment effects heterogeneity. We propose an alternative estimator
that is free of contamination, and illustrate the relative shortcomings of two-way fixed
effects regressions with leads and lags through an empirical application.”

• Callaway and Sant’Anna (2020), “Difference-in-Differences with Multiple Time Pe-

riods,” WP
“In this article, we consider identification, estimation, and inference procedures for
treatment effect parameters using Difference-in-Differences (DID) with (i) multiple time
periods, (ii) variation in treatment timing, and (iii) when the “parallel trends assump-
tion” holds potentially only after conditioning on observed covariates. We show that
a family of causal effect parameters are identified in staggered DID setups, even if
differences in observed characteristics create non-parallel outcome dynamics between
groups. Our identification results allow one to use outcome regression, inverse prob-
ability weighting, or doubly-robust estimands. We also propose different aggregation
schemes that can be used to highlight treatment effect heterogeneity across different
dimensions as well as to summarize the overall effect of participating in the treatment.
We establish the asymptotic properties of the proposed estimators and prove the validity
of a computationally convenient bootstrap procedure to conduct asymptotically valid si-
multaneous (instead of pointwise) inference. Finally, we illustrate the relevance of our
proposed tools by analyzing the effect of the minimum wage on teen employment from
2001–2007. Open-source software is available for implementing the proposed meth-
ods.”

• de Chaisemartin and d’Haultfoeuille (2020), “Two-Way Fixed Effects Estimators

with Heterogeneous Treatment Effects,” AER
“Linear regressions with period and group fixed effects are widely used to estimate
treatment effects. We show that they estimate weighted sums of the average treatment
effects (ATE) in each group and period, with weights that may be negative. Due to the
negative weights, the linear regression coefficient may for instance be negative while all
the ATEs are positive. We propose another estimator that solves this issue. In the two
applications we revisit, it is significantly different from the linear regression estimator.”

• Goodman-Bacon (2020), “Difference-in-Differences with Variation in Treatment

10
 Timing,” WP
“The canonical difference-in-differences (DD) estimator contains two time periods, “pre”
and “post”, and two groups, “treatment” and “control”. Most DD applications, how-
ever, exploit variation across groups of units that receive treatment at different times.
This paper shows that the general estimator equals a weighted average of all possible
two-group/two-period DD estimators in the data. This defines the DD estimand and
identifying assumption, a generalization of common trends. I discuss how to interpret
DD estimates and propose a new balance test. I show how to decompose the differ-
ence between two specifications, and provide a new analysis of models that include
time-varying controls.”

• Marcus and Sant’Anna (2020), “The Role of Parallel Trends in Event Study Set-
tings: An Application to Environmental Economics,” JAERE
“Difference-in-Differences (DID) research designs usually rely on variation of treatment
timing such that, after making an appropriate parallel trends assumption, one can iden-
tify, estimate, and make inference about causal effects. In practice, however, different
DID procedures rely on different parallel trends assumptions (PTA), and recover dif-
ferent causal parameters. In this paper, we focus on staggered DID (also referred as
event-studies) and discuss the role played by the PTA in terms of identification and es-
timation of causal parameters. We document a “robustness” vs. “efficiency” trade-off
in terms of the strength of the underlying PTA, and argue that practitioners should be
explicit about these trade-offs whenever using DID procedures. We propose new DID
estimators that reflect these trade-offs and derived their large sample properties. We il-
lustrate the practical relevance of these results by assessing whether the transition from
federal to state management of the Clean Water Act affects compliance rates.”

• Rambachan and Roth (2020), “An Honest Approach to Parallel Trends,” WP

“This paper proposes robust inference methods for difference-in-differences and event-
study designs that do not require that the parallel trends assumption holds exactly. In-
stead, the researcher must only impose restrictions on the possible differences in trends
between the treated and control groups. Several common intuitions expressed in ap-
plied work can be captured by such restrictions, including the notion that pre-treatment
differences in trends are informative about counterfactual post-treatment differences in
trends. Our methodology then guarantees uniformly valid (“honest”) inference when
the imposed restrictions are satisfied. We first show that fixed length confidence inter-
vals have near-optimal expected length for a practically-relevant class of restrictions.
We next introduce a novel inference procedure that accommodates a wider range of re-

11
 strictions, which is based on the observation that inference in our setting is equivalent to
testing a system of moment inequalities with a large number of linear nuisance param-
eters. The resulting confidence sets are consistent, and have optimal local asymptotic
power for many parameter configurations. We recommend researchers conduct sensi-
tivity analyses to show what conclusions can be drawn under various restrictions on the
possible differences in trends”

• Sant’Anna and Zhao (2020), “Doubly Robust Difference-in-Differences Estimators,”

JE
“This article proposes doubly robust estimators for the average treatment effect on the
treated (ATT) in difference-in-differences (DID) research designs. In contrast to alterna-
tive DID estimators, the proposed estimators are consistent if either (but not necessarily
both) a propensity score or outcome regression working models are correctly specified.
We also derive the semiparametric efficiency bound for the ATT in DID designs when
either panel or repeated cross-section data are available, and show that our proposed
estimators attain the semiparametric efficiency bound when the working models are
correctly specified. Furthermore, we quantify the potential efficiency gains of having
access to panel data instead of repeated cross-section data. Finally, by paying particular
attention to the estimation method used to estimate the nuisance parameters, we show
that one can sometimes construct doubly robust DID estimators for the ATT that are
also doubly robust for inference. Simulation studies and an empirical application illus-
trate the desirable finite-sample performance of the proposed estimators. Open-source
software for implementing the proposed policy evaluation tools is available.”

4 Standard IV
• Andrews and Armstrong (2017), “Unbiased Instrumental Variables Estimation un-
der Known First-Stage Sign,” QE
“We derive mean-unbiased estimators for the structural parameter in instrumental vari-
ables models with a single endogenous regressor where the sign of one or more first-
stage coefficients is known. In the case with a single instrument, there is a unique non-
randomized unbiased estimator based on the reduced-form and first-stage regression
estimates. For cases with multiple instruments we propose a class of unbiased estima-
tors and show that an estimator within this class is efficient when the instruments are
strong. We show numerically that unbiasedness does not come at a cost of increased
dispersion in models with a single instrument: in this case the unbiased estimator is less

12
 dispersed than the two-stage least squares estimator. Our finite-sample results apply to
normal models with known variance for the reduced-form errors, and imply analogous
results under weak-instrument asymptotics with an unknown error distribution.”

• Mogstad and Torgovitsky (2018), “Identification and Extrapolation of Causal Ef-

fects with Instrumental Variables,” ARE
“Instrumental variables (IV) are widely used in economics to address selection on un-
observables. Standard IV methods produce estimates of causal effects that are specific
to individuals whose behavior can be manipulated by the instrument at hand. In many
cases, these individuals are not the same as those who would be induced to treatment
by an intervention or policy of interest to the researcher. The average causal effect for
the two groups can differ significantly if the effect of the treatment varies systematically
with unobserved factors that are correlated with treatment choice. We review the im-
plications of this type of unobserved heterogeneity for the interpretation of standard IV
methods and for their relevance to policy evaluation. We argue that making inferences
about policy-relevant parameters typically requires extrapolating from the individuals
affected by the instrument to the individuals who would be induced to treatment by the
policy under consideration. We discuss a variety of alternatives to standard IV methods
that can be used to rigorously perform this extrapolation. We show that many of these
approaches can be nested as special cases of a general framework that embraces the
possibility of partial identification.”

• Andrews, Stock, and Sun (2019), “Weak Instruments in Instrumental Variables Re-
gression: Theory and Practice,” ARE
“When instruments are weakly correlated with endogenous regressors, conventional
methods for instrumental variables (IV) estimation and inference become unreliable.
A large literature in econometrics has developed procedures for detecting weak instru-
ments and constructing robust confidence sets, but many of the results in this literature
are limited to settings with independent and homoskedastic data, while data encoun-
tered in practice frequently violate these assumptions. We review the literature on weak
instruments in linear IV regression with an emphasis on results for nonhomoskedastic
(heteroskedastic, serially correlated, or clustered) data. To assess the practical impor-
tance of weak instruments, we also report tabulations and simulations based on a sur-
vey of papers published in the American Economic Review from 2014 to 2018 that use
IV. These results suggest that weak instruments remain an important issue for empirical
practice, and that there are simple steps that researchers can take to better handle weak
instruments in applications.”

13
• Evdokimov and Kolesár (2019), “Inference in Instrumental Variable Regression
Analysis with Heterogeneous Treatment Effects,” WP
“We study inference in an instrumental variables model with heterogeneous treatment
effects and possibly many instruments and/or covariates. In this case two-step estima-
tors such as the two-stage least squares (TSLS) or versions of the jackknife instrumental
variables (JIV) estimator estimate a particular weighted average of the local average
treatment effects. The weights in these estimands depend on the first-stage coefficients,
and either the sample or population variability of the covariates and instruments, de-
pending on whether they are treated as fixed (conditioned upon) or random. We give
new asymptotic variance formulas for the TSLS and JIV estimators, and pro- pose consis-
tent estimators of these variances. The heterogeneity of the treatment effects generally
increases the asymptotic variance. Moreover, when the treatment effects are heteroge-
neous, the conditional asymptotic variance is smaller than the unconditional one. Our
results are also useful when the treatment effects are constant, because they provide
the asymptotic distribution and valid standard errors for the estimators that are robust
to the presence of many covariates.”

• Lee, McCrary, Moreira, and Porter (2020), “Valid t-ratio Inference for IV,” WP
“In the single IV model, current practice relies on the first-stage F exceeding some
threshold (e.g., 10) as a criterion for trusting t-ratio inferences, even though this yields
an anti-conservative test. We show that a true 5 percent test instead requires an F
greater than 104.7. Maintaining 10 as a threshold requires replacing the critical value
1.96 with 3.43. We re-examine 57 AER papers and find that corrected inference causes
half of the initially presumed statistically significant results to be insignificant. We in-
troduce a more powerful test, the tF procedure, which provides F-dependent adjusted
t-ratio critical values.”

• Mogstad, Torgovitsky, and Walters (2020a), “Policy Evaluation with Multiple In-
strumental Variables,” WP
“Marginal treatment effect methods are widely used for causal inference and policy
evaluation with instrumental variables. However, they fundamentally rely on the well-
known monotonicity (threshold-crossing) condition on treatment choice behavior. Re-
cent research has shown that this condition cannot hold with multiple instruments un-
less treatment choice is effectively homogeneous. Based on these findings, we develop a
new marginal treatment effect framework under a weaker, partial monotonicity condi-
tion. The partial monotonicity condition is implied by standard choice theory and allows
for rich heterogeneity even in the presence of multiple instruments. The new frame-

14
 work can be viewed as having multiple different choice models for the same observed
treatment variable, all of which must be consistent with the data and with each other.
Using this framework, we develop a methodology for partial identification of clearly
stated, policy-relevant target parameters while allowing for a wide variety of nonpara-
metric shape restrictions and parametric functional form assumptions. We show how
the methodology can be used to combine multiple instruments together to yield more
informative empirical conclusions than one would obtain by using each instrument sep-
arately. The methodology provides a blueprint for extracting and aggregating informa-
tion about treatment effects from multiple controlled or natural experiments while still
allowing for rich heterogeneity in both treatment effects and choice behavior.”

• Mogstad, Torgovitsky, and Walters (2020b), “The Causal Interpretation of Two-

Stage Least Squares with Multiple Instrumental Variables,” NBER WP
“Empirical researchers often combine multiple instrumental variables (IVs) for a single
treatment using two-stage least squares (2SLS). When treatment effects are heteroge-
neous, a common justification for including multiple IVs is that the 2SLS estimand can
be given a causal interpretation as a positively-weighted average of local average treat-
ment effects (LATEs). This justification requires the well-known monotonicity condition.
However, we show that with more than one instrument, this condition can only be sat-
isfied if choice behavior is effectively homogenous. Based on this finding, we consider
the use of multiple IVs under a weaker, partial monotonicity condition. We character-
ize empirically verifiable sufficient and necessary conditions for the 2SLS estimand to
be a positively-weighted average of LATEs under partial monotonicity. We apply these
results to an empirical analysis of the returns to college with multiple instruments. We
show that the standard monotonicity condition is at odds with the data. Neverthe-
less, our empirical checks show that the 2SLS estimate retains a causal interpretation
as a positively-weighted average of the effects of college attendance among complier
groups.”

• Young (2020), “Consistency Without Inference: Instrumental Variables in Practical

Application,” WP
“I use Monte Carlo simulations, the jackknife and multiple forms of the bootstrap to
study a comprehensive sample of 1359 instrumental variables regressions in 31 papers
published in the journals of the American Economic Association. Monte Carlo simu-
lations based upon published regressions show that non-iid error processes in highly
leveraged regressions, both prominent features of published work, adversely affect the
size and power of IV estimates, while increasing the bias of IV relative to OLS. Weak in-

15
 strument pre-tests based upon F-statistics are found to be largely uninformative of both
size and bias. In published papers, statistically significant IV results generally depend
upon only one or two observations or clusters, IV has little power as, despite producing
substantively different estimates, it rarely rejects the OLS point estimate or the null that
OLS is unbiased, while the statistical significance of excluded instruments is substan-
tially exaggerated.”

5 Shift-Share IV3
• Adão, Kolesár, and Morales (2019), “Shift-Share Designs: Theory and Inference,”
QJE
“We study inference in shift-share regression designs, such as when a regional outcome
is regressed on a weighted average of sectoral shocks, using regional sector shares as
weights. We conduct a placebo exercise in which we estimate the effect of a shift-share
regressor constructed with randomly generated sectoral shocks on actual labor market
outcomes across U.S. commuting zones. Tests based on commonly used standard errors
with 5% nominal significance level reject the null of no effect in up to 55% of the placebo
samples. We use a stylized economic model to show that this overrejection problem
arises because regression residuals are correlated across regions with similar sectoral
shares, independent of their geographic location. We derive novel inference methods
that are valid under arbitrary cross-regional correlation in the regression residuals. We
show using popular applications of shift-share designs that our methods may lead to
substantially wider confidence intervals in practice.”

• Borusyak, Hull, and Jaravel (2020), “Quasi-Experimental Shift-Share Research De-

signs,” WP
“Many studies use shift-share (or “Bartik”) instruments, which average a set of shocks
with exposure share weights. We provide a new econometric framework for shift-share
instrumental variable (SSIV) regressions in which identification follows from the quasi-
random assignment of shocks, while exposure shares are allowed to be endogenous.
The framework is motivated by an equivalence result: the orthogonality between a
shift-share instrument and an unobserved residual can be represented as the orthogo-
nality between the underlying shocks and a shock-level unobservable. SSIV regression
coefficients can similarly be obtained from an equivalent shock-level regression, mo-

3
The references in this section are taken from a guest lecture that Peter Hull gave in Arindrajit Dube’s ECON
797B Fall 2020 course at UMass Amherst – see lecture slides here.

16
 tivating shock-level conditions for their consistency. We discuss and illustrate several
practical insights delivered by this framework in the setting of Autor et al. (2013).”

• Borusyak and Hull (2020), “Non-Random Exposure to Exogenous Shocks: Theory

and Applications,” NBER WP
“We develop new tools for causal inference in settings where exogenous shocks affect
the treatment status of multiple observations jointly, to different extents. In these set-
tings researchers may construct treatments or instruments that combine the shocks with
predetermined measures of shock exposure. Examples include measures of spillovers in
social and transportation networks, simulated eligibility instruments, and shift-share
instruments. We show that leveraging the exogeneity of shocks for identification gen-
erally requires a simple but non-standard recentering, derived from the specification
of counterfactual shocks that might as well have been realized. We further show how
specification of counterfactual shocks can be used for finite-sample inference and spec-
ification tests, and we characterize the recentered instruments that are asymptotically
efficient. We use this framework to estimate the employment effects of Chinese market
access growth due to high-speed rail construction and the insurance coverage effects of
expanded Medicaid eligibility.”

• Goldsmith-Pinkham, Sorkin, and Swift (2020), “Bartik Instruments: What, When,

Why, and How,” AER
“The Bartik instrument is formed by interacting local industry shares and national in-
dustry growth rates. We show that the typical use of a Bartik instrument assumes a
pooled exposure research design, where the shares measure differential exposure to
common shocks, and identification is based on exogeneity of the shares. Next, we show
how the Bartik instrument weights each of the exposure designs. Finally, we discuss
how to assess the plausibility of the research design. We illustrate our results through
two applications: estimating the elasticity of labor supply, and estimating the elasticity
of substitution between immigrants and natives.”

6 RDD
• Calonico, Cattaneo, and Titiunik (2014), “Robust Nonparametric Confidence In-
tervals for Regression-Discontinuity Designs,” ECMA
“In the regression-discontinuity (RD) design, units are assigned to treatment based on
whether their value of an observed covariate exceeds a known cutoff. In this design,

17
 local polynomial estimators are now routinely employed to construct confidence inter-
vals for treatment effects. The performance of these confidence intervals in applica-
tions, however, may be seriously hampered by their sensitivity to the specific bandwidth
employed. Available bandwidth selectors typically yield a “large” bandwidth, leading
to data-driven confidence intervals that may be biased, with empirical coverage well
below their nominal target. We propose new theory-based, more robust confidence in-
terval estimators for average treatment effects at the cutoff in sharp RD, sharp kink RD,
fuzzy RD, and fuzzy kink RD designs. Our proposed confidence intervals are constructed
using a bias-corrected RD estimator together with a novel standard error estimator. For
practical implementation, we discuss mean squared error optimal bandwidths, which
are by construction not valid for conventional confidence intervals but are valid with
our robust approach, and consistent standard error estimators based on our new vari-
ance formulas. In a special case of practical interest, our procedure amounts to running
a quadratic instead of a linear local regression. More generally, our results give a for-
mal justification to simple inference procedures based on increasing the order of the
local polynomial estimator employed. We find in a simulation study that our confidence
intervals exhibit close-to-correct empirical coverage and good empirical interval length
on average, remarkably improving upon the alternatives available in the literature. All
results are readily available in R and STATA using our companion software packages
described in Calonico, Cattaneo, and Titiunik (2014d, 2014b).”

• Grembi, Nannicini, and Troiano (2016), “Do Fiscal Rules Matter?,”4 AEJ Applied
“Fiscal rules are laws aimed at reducing the incentive to accumulate debt, and many
countries adopt them to discipline local governments. Yet, their effectiveness is disputed
because of commitment and enforcement problems. We study their impact applying
a quasi-experimental design in Italy. In 1999, the central government imposed fiscal
rules on municipal governments, and in 2001 relaxed them below 5,000 inhabitants.
We exploit the before/after and discontinuous policy variation, and show that relaxing
fiscal rules increases deficits and lowers taxes. The effect is larger if the mayor can be
reelected, the number of parties is higher, and voters are older.”

• Armstrong and Kolesár (2018), “Optimal Inference in a Class of Regression Mod-

els,” ECMA
“We consider the problem of constructing confidence intervals (CIs) for a linear func-
tional of a regression function, such as its value at a point, the regression discontinuity
4
Even though its title makes it sound like it is irrelevant, this paper has been added because it thoroughly cov-
ers the identifying assumptions of the “difference-in-discontinuities estimator,” which intuitively combines
a diff-in-diff strategy with an RD design.

18
 parameter, or a regression coefficient in a linear or partly linear regression. Our main
assumption is that the regression function is known to lie in a convex function class,
which covers most smoothness and/or shape assumptions used in econometrics. We
derive finite-sample optimal CIs and sharp efficiency bounds under normal errors with
known variance. We show that these results translate to uniform (over the function
class) asymptotic results when the error distribution is not known. When the function
class is centrosymmetric, these efficiency bounds imply that minimax CIs are close to
efficient at smooth regression functions. This implies, in particular, that it is impossible
to form CIs that are substantively tighter using data-dependent tuning parameters, and
maintain coverage over the whole function class. We specialize our results to inference
on the regression discontinuity parameter, and illustrate them in simulations and an
empirical application.”

• Canay and Kamat (2018), “Approximate Permutation Tests and Induced Order
Statistics in the Regression Discontinuity Design,” REStud
“In the regression discontinuity design (RDD), it is common practice to assess the cred-
ibility of the design by testing whether the means of baseline covariates do not change
at the cut-off (or threshold) of the running variable. This practice is partly motivated
by the stronger implication derived by Lee (2008), who showed that under certain con-
ditions the distribution of baseline covariates in the RDD must be continuous at the
cut-off. We propose a permutation test based on the so-called induced ordered statis-
tics for the null hypothesis of continuity of the distribution of baseline covariates at
the cut-off; and introduce a novel asymptotic framework to analyse its properties. The
asymptotic framework is intended to approximate a small sample phenomenon: even
though the total number n of observations may be large, the number of effective obser-
vations local to the cut-off is often small. Thus, while traditional asymptotics in RDD
require a growing number of observations local to the cut-off as n → ∞, our framework
keeps the number q of observations local to the cut-off fixed as n → ∞. The new test is
easy to implement, asymptotically valid under weak conditions, exhibits finite sample
validity under stronger conditions than those needed for its asymptotic validity, and has
favourable power properties relative to tests based on means. In a simulation study, we
find that the new test controls size remarkably well across designs. We then use our test
to evaluate the plausibility of the design in Lee (2008), a well-known application of the
RDD to study incumbency advantage.”

• Ganong and Jäger (2018), “A Permutation Test for the Regression Kink Design,”
JASA

19
 “The regression kink (RK) design is an increasingly popular empirical method for esti-
mating causal effects of policies, such as the effect of unemployment benefits on unem-
ployment duration. Using simulation studies based on data from existing RK designs,
we empirically document that the statistical significance of RK estimators based on con-
ventional standard errors can be spurious. In the simulations, false positives arise as
a consequence of nonlinearities in the underlying relationship between the outcome
and the assignment variable, confirming concerns about the misspecification bias of
discontinuity estimators pointed out by Calonico, Cattaneo, and Titiunik. As a comple-
ment to standard RK inference, we propose that researchers construct a distribution of
placebo estimates in regions with and without a policy kink and use this distribution to
gauge statistical significance. Under the assumption that the location of the kink point
is random, this permutation test has exact size in finite samples for testing a sharp null
hypothesis of no effect of the policy on the outcome. We implement simulation studies
based on existing RK applications that estimate the effect of unemployment benefits on
unemployment duration and show that our permutation test as well as inference proce-
dures proposed by Calonico, Cattaneo, and Titiunik improve upon the size of standard
approaches, while having sufficient power to detect an effect of unemployment bene-
fits on unemployment duration. Supplementary materials for this article are available
online.”

• Kolesár and Rothe (2018), “Inference in Regression Discontinuity Designs with a

Discrete Running Variable,” AER
“We consider inference in regression discontinuity designs when the running variable
only takes a moderate number of distinct values. In particular, we study the common
practice of using confidence intervals (CIs) based on standard errors that are clustered
by the running variable as a means to make inference robust to model misspecification
(Lee and Card 2008). We derive theoretical results and present simulation and empirical
evidence showing that these CIs do not guard against model misspecification, and that
they have poor coverage properties. We therefore recommend against using these CIs in
practice. We instead propose two alternative CIs with guaranteed coverage properties
under easily interpretable restrictions on the conditional expectation function.”

• Gelman and Imbens (2019), “Why High-Order Polynomials Should Not Be Used in
Regression Discontinuity Designs,” JBES
“It is common in regression discontinuity analysis to control for third, fourth, or higher-
degree polynomials of the forcing variable. There appears to be a perception that such
methods are theoretically justified, even though they can lead to evidently nonsensical

20
 results. We argue that controlling for global high-order polynomials in regression dis-
continuity analysis is a flawed approach with three major problems: it leads to noisy
estimates, sensitivity to the degree of the polynomial, and poor coverage of confidence
intervals. We recommend researchers instead use estimators based on local linear or
quadratic polynomials or other smooth functions.”

• Armstrong and Kolesár (2020), “Simple and Honest Confidence Intervals in Non-
parametric Regression,” QE
“We consider the problem of constructing honest confidence intervals (CIs) for a scalar
parameter of interest, such as the regression discontinuity parameter, in nonparametric
regression based on kernel or local polynomial estimators. To ensure that our CIs are
honest, we use critical values that take into account the possible bias of the estimator
upon which the CIs are based. We show that this approach leads to CIs that are more
efficient than conventional CIs that achieve coverage by undersmoothing or subtracting
an estimate of the bias. We give sharp efficiency bounds of using different kernels, and
derive the optimal bandwidth for constructing honest CIs. We show that using the band-
width that minimizes the maximum mean-squared error results in CIs that are nearly
efficient and that in this case, the critical value depends only on the rate of convergence.
For the common case in which the rate of convergence is n−2/5 , the appropriate critical
value for 95% CIs is 2.18, rather than the usual 1.96 critical value. We illustrate our
results in a Monte Carlo analysis and an empirical application.”

• Bugni and Canay (2020), “Testing Continuity of a Density via g-order statistics in
the Regression Discontinuity Design,” JE
“In the regression discontinuity design (RDD), it is common practice to assess the cred-
ibility of the design by testing the continuity of the density of the running variable at
the cut-off, e.g., McCrary (2008). In this paper we propose an approximate sign test
for continuity of a density at a point based on the so-called g-order statistics, and study
its properties under two complementary asymptotic frameworks. In the first asymptotic
framework, the number q of observations local to the cut-off is fixed as the sample size
n diverges to infinity, while in the second framework q diverges to infinity slowly as n
diverges to infinity. Under both of these frameworks, we show that the test we propose
is asymptotically valid in the sense that it has limiting rejection probability under the
null hypothesis not exceeding the nominal level. More importantly, the test is easy to
implement, asymptotically valid under weaker conditions than those used by compet-
ing methods, and exhibits finite sample validity under stronger conditions than those
needed for its asymptotic validity. In a simulation study, we find that the approximate

21
 sign test provides good control of the rejection probability under the null hypothesis
while remaining competitive under the alternative hypothesis. We finally apply our test
to the design in Lee (2008), a well-known application of the RDD to study incumbency
advantage.”

• Cattaneo, Jansson, and Ma (2020), “Simple Local Polynomial Density Estimators,”

JASA
“This article introduces an intuitive and easy-to-implement nonparametric density esti-
mator based on local polynomial techniques. The estimator is fully boundary adaptive
and automatic, but does not require prebinning or any other transformation of the
data. We study the main asymptotic properties of the estimator, and use these results to
provide principled estimation, inference, and bandwidth selection methods. As a sub-
stantive application of our results, we develop a novel discontinuity in density testing
procedure, an important problem in regression discontinuity designs and other program
evaluation settings. An illustrative empirical application is given. Two companion Stata
and R software packages are provided.”

7 Synthetic Control
• Abadie (2020), “Using Synthetic Controls: Feasibility, Data Requirements, and
Methodological Aspects,” JEL

• Arkhangelsky, Athey, Hirshberg, Imbens, and Wager (2020), “Synthetic Difference

in Differences,” WP
“We present a new estimator for causal effects with panel data that builds on insights
behind the widely used difference in differences and synthetic control methods. Rel-
ative to these methods, we find, both theoretically and empirically, that the proposed
“synthetic difference in differences” estimator has desirable robustness properties, and
that it performs well in settings where the conventional estimators are commonly used
in practice. We study the asymptotic behavior of the estimator when the systematic part
of the outcome model includes latent unit factors interacted with latent time factors,
and we present conditions for consistency and asymptotic normality.”

• Athey, Bayati, Doudchenko, Imbens, and Khosravi (2020), “Matrix Completion

Methods for Causal Panel Data Models,” WP
“In this paper we study methods for estimating causal effects in settings with panel
data, where some units are exposed to a treatment during some periods and the goal

22
 is estimating counterfactual (untreated) outcomes for the treated unit/period combi-
nations. We develop a class of matrix completion estimators that uses the observed
elements of the matrix of control outcomes corresponding to untreated unit/periods to
impute the “missing” elements of the control outcome matrix, corresponding to treated
units/periods. The approach estimates a matrix that well-approximates the original (in-
complete) matrix, but has lower complexity according to the nuclear norm for matrices.
We generalize results from the matrix completion literature by allowing the patterns
of missing data to have a time series dependency structure. We present novel insights
concerning the connections between the matrix completion literature, the literature on
interactive fixed effects models and the literatures on program evaluation under uncon-
foundedness and synthetic control methods. We show that all these estimators can be
viewed as focusing on the same objective function. They differ in the way they deal with
lack of identification, in some cases solely through regularization (our proposed nu-
clear norm matrix completion estimator) and in other cases primarily through imposing
hard restrictions (the unconfoundedness and synthetic control approaches). proposed
method outperforms unconfoundedness-based or synthetic control estimators.”

• Ben-Michael, Feller, and Rothstein (2020), “The Augmented Synthetic Control Method,”
WP
“The synthetic control method (SCM) is a popular approach for estimating the impact
of a treatment on a single unit in panel data settings. The “synthetic control” is a
weighted average of control units that balances the treated unit’s pre-treatment out-
comes as closely as possible. A critical feature of the original proposal is to use SCM
only when the fit on pre-treatment outcomes is excellent. We propose Augmented SCM
as an extension of SCM to settings where such pre-treatment fit is infeasible. Analo-
gous to bias correction for inexact matching, Augmented SCM uses an outcome model
to estimate the bias due to imperfect pre-treatment fit and then de-biases the original
SCM estimate. Our main proposal, which uses ridge regression as the outcome model,
directly controls pre-treatment fit while minimizing extrapolation from the convex hull.
This estimator can also be expressed as a solution to a modified synthetic controls prob-
lem that allows negative weights on some donor units. We bound the estimation error of
this approach under different data generating processes, including a linear factor model,
and show how regularization helps to avoid over-fitting to noise. We demonstrate gains
from Augmented SCM with extensive simulation studies and apply this framework to
estimate the impact of the 2012 Kansas tax cuts on economic growth. We implement
the proposed method in the new augsynth R package.”

23
8 Matching
• Otsu and Rai (2017), “Bootstrap Inference of Matching Estimators for Average
Treatment Effects,” JASA
“It is known that the naive bootstrap is not asymptotically valid for a matching esti-
mator of the average treatment effect with a fixed number of matches. In this article,
we propose asymptotically valid inference methods for matching estimators based on
the weighted bootstrap. The key is to construct bootstrap counterparts by resampling
based on certain linear forms of the estimators. Our weighted bootstrap is applicable
for the matching estimators of both the average treatment effect and its counterpart for
the treated population. Also, by incorporating a bias correction method in Abadie and
Imbens (2011), our method can be asymptotically valid even for matching based on a
vector of covariates. A simulation study indicates that the weighted bootstrap method
is favorably comparable with the asymptotic normal approximation. As an empirical
illustration, we apply the proposed method to the National Supported Work data. Sup-
plementary materials for this article are available online.”

• Adusumilli (2018), “Bootstrap Inference for Propensity Score Matching,” WP

“Propensity score matching, where the propensity scores are estimated in a first step,
is widely used for estimating treatment effects. In this context, the naive bootstrap
is invalid (Abadie and Imbens, 2008). This paper proposes a novel bootstrap proce-
dure for the propensity score matching estimator, and demonstrates its consistency. The
proposed bootstrap is built around the notion of ‘potential errors’, introduced in this
paper. Precisely, each observation is associated with two potential error terms, cor-
responding to each of the potential states - treated or control - only one of which is
realized. Thus, the variability of the estimator stems not only from the randomness of
the potential errors themselves, but also from the probabilistic nature of treatment as-
signment, which randomly realizes one of the potential error terms. The proposed boot-
strap takes both sources of randomness into account by resampling the potential errors
as a pair as well as re-assigning new values for the treatments. Simulations and real
data examples demonstrate the superior performance of the proposed method relative
to using the asymptotic distribution for inference, especially when the degree of over-
lap in propensity scores is poor. General versions of the procedure can also be applied
to other causal effect estimators such as inverse probability weighting and propensity
score sub-classification, potentially leading to higher order refinements for inference in
such contexts.”

24
9 Bunching5
• Kleven (2016), “Bunching,” ARE
“Recent years have seen a surge of applied work using bunching approaches, a devel-
opment that is closely linked to the increased availability of administrative data. These
approaches exploit the incentives for bunching created by discontinuities in the slope
of choice sets (kinks) or in the level of choice sets (notches) to study the behavior of
individuals and firms. Although the bunching approach was originally developed in the
context of taxation, it is beginning to find applications in many other areas, such as
social security, social insurance, welfare programs, education, regulation, private sector
prices, and reference-dependent preferences. This review provides a guide to bunch-
ing estimation, discusses its strengths and weaknesses, surveys a range of applications
across fields, and considers reasons for the ubiquity of kinks and notches.”

• Blomquist, Newey, Kumar, and Liang (2019), “On Bunching and Identification of
the Taxable Income Elasticity,” NBER WP
“The taxable income elasticity is a key parameter for predicting the effect of tax re-
form or designing an income tax. Bunching at kinks and notches in a single budget
set have been used to estimate the taxable income elasticity. We show that when the
heterogeneity distribution is unrestricted the amount of bunching at a kink or a notch is
not informative about the size of the taxable income elasticity, and neither is the entire
distribution of taxable income for a convex budget set. Kinks do provide information
about the size of the elasticity when a priori restrictions are placed on the heterogene-
ity distribution. They can identify the elasticity when the heterogeneity distribution is
specified across the kink and provide bounds under restrictions on the heterogeneity
distribution. We also show that variation in budget sets can identify the taxable income
elasticity when the distribution of preferences is unrestricted and stable across budget
sets. For nonparametric utility with general heterogeneity we show that kinks only pro-
vide elasticity information about individuals at the kink and we give bounds analogous
to those for isoelastic utility. Identification becomes more difficult with optimization
errors We show in examples how results are affected by optimization errors.”

• Marx (2020), “Dynamic Bunching Estimation with Panel Data,” WP

“Bunching estimation of distortions in a distribution around a policy threshold provides
a means of studying behavioral parameters. Standard cross-sectional bunching estima-

5
See this 2018 Bunching Estimator Workshop webpage for more references on bunching, including recent
applications (thanks to Ben Solow for sharing this link).

25
 tors rely on identification assumptions about heterogeneity that I show can be violated
by serial dependence of the choice variable or attrition related to the threshold. I pro-
pose a dynamic bunching estimation design that exploits panel data to obtain identifi-
cation from relative within-agent changes in income and to estimate new parameters.
Simulations using household income data demonstrate the benefits of the panel design.
An application to charitable organizations demonstrates opportunities for estimating
elasticity correlates, causal effects, and extensive-margin responses.”

10 Decomposition
• Fortin, Lemieux, and Firpo (2011), “Decomposition Methods in Economics,” HLE
“This chapter provides a comprehensive overview of decomposition methods that have
been developed since the seminal work of Oaxaca and Blinder in the early 1970s. These
methods are used to decompose the difference in a distributional statistic between two
groups, or its change over time, into various explanatory factors. While the original
work of Oaxaca and Blinder considered the case of the mean, our main focus is on other
distributional statistics besides the mean, such as quantiles, the Gini coefficient or the
variance. We discuss the assumptions required for identifying the different elements of
the decomposition, as well as various estimation methods proposed in the literature. We
also illustrate how these methods work in practice by discussing existing applications
and working through a set of empirical examples throughout the paper.”

• Chernozhukov, Fernández-Val, and Melly (2013), “Inference on Counterfactual Dis-

tributions,” ECMA
“Counterfactual distributions are important ingredients for policy analysis and decom-
position analysis in empirical economics. In this article, we develop modeling and in-
ference tools for counterfactual distributions based on regression methods. The coun-
terfactual scenarios that we consider consist of ceteris paribus changes in either the
distribution of covariates related to the outcome of interest or the conditional distri-
bution of the outcome given covariates. For either of these scenarios, we derive joint
functional central limit theorems and bootstrap validity results for regression-based es-
timators of the status quo and counterfactual outcome distributions. These results allow
us to construct simultaneous confidence sets for function-valued effects of the counter-
factual changes, including the effects on the entire distribution and quantile functions
of the outcome as well as on related functionals. These confidence sets can be used to
test functional hypotheses such as no-effect, positive effect, or stochastic dominance.

26
 Our theory applies to general counterfactual changes and covers the main regression
methods including classical, quantile, duration, and distribution regressions. We illus-
trate the results with an empirical application to wage decompositions using data for
the United States.”

11 General
• Abadie and Cattaneo (2018), “Econometric Methods for Program Evaluation,” ARE
“Program evaluation methods are widely applied in economics to assess the effects of
policy interventions and other treatments of interest. In this article, we describe the
main methodological frameworks of the econometrics of program evaluation. In the
process, we delineate some of the directions along which this literature is expanding,
discuss recent developments, and highlight specific areas where new research may be
particularly fruitful.”

27
References
Abadie, A. (2020). Using Synthetic Controls: Feasibility, Data Requirements, and Method-
ological Aspects. Journal of Economic Literature, forthcoming.

Abadie, A., S. Athey, G. W. Imbens, and J. Wooldridge (2017). When Should You Adjust
Standard Errors for Clustering? NBER Working Paper.

Abadie, A., S. Athey, G. W. Imbens, and J. M. Wooldridge (2020). Sampling-Based Versus

Design-Based Uncertainty in Regression Analysis. Econometrica 88(1), 265–296.

Abadie, A. and M. D. Cattaneo (2018). Econometric Methods for Program Evaluation.

Annual Review of Economics 10, 465–503.

Abraham, S. and L. Sun (2020). Estimating Dynamic Treatment Effects in Event Studies
with Heterogeneous Treatment Effects. Journal of Econometrics, forthcoming.

Acharya, A., M. Blackwell, and M. Sen (2016). Explaining Causal Findings Without Bias:
Detecting and Assessing Direct Effects. American Political Science Review 110(3), 512–
529.

Adão, R., M. Kolesár, and E. Morales (2019). Shift-Share Designs: Theory and Inference.
Quarterly Journal of Economics 134(4), 1949–2010.

Adusumilli, K. (2018). Bootstrap Inference for Propensity Score Matching. Working Paper.

Andrews, I. and T. B. Armstrong (2017). Unbiased Instrumental Variables Estimation

under Known First-Stage Sign. Quantitative Economics 8(2), 479–503.

Andrews, I., J. H. Stock, and L. Sun (2019). Weak Instruments in Instrumental Variables
Regression: Theory and Practice. Annual Review of Economics 11, 727–753.

Arkhangelsky, D., S. Athey, D. A. Hirshberg, G. W. Imbens, and S. Wager (2020). Synthetic

Difference in Differences. Working Paper.

Arkhangelsky, D. and G. W. Imbens (2019). Double-Robust Identification for Causal Panel

Data Models. Working Paper.

Armstrong, T. B. and M. Kolesár (2018). Optimal Inference in a Class of Regression Models.

Econometrica 86(2), 655–683.

Armstrong, T. B. and M. Kolesár (2020). Simple and Honest Confidence Intervals in Non-
parametric Regression. Quantitative Economics 11(1), 1–39.

28
Athey, S., M. Bayati, N. Doudchenko, G. Imbens, and K. Khosravi (2020). Matrix Comple-
tion Methods for Causal Panel Data Models. Working Paper.

Athey, S. and G. W. Imbens (2018). Design-Based Analysis in Difference-In-Differences

Settings with Staggered Adoption. NBER Working Paper.

Ben-Michael, E., A. Feller, and J. Rothstein (2020). The Augmented Synthetic Control
Method. Working Paper.

Blomquist, S., W. Newey, A. Kumar, and C.-Y. Liang (2019). On Bunching and Identification
of the Taxable Income Elasticity. NBER Working Paper.

Borusyak, K. and P. Hull (2020). Non-random exposure to exogenous shocks: Theory and
applications. NBER Working Paper.

Borusyak, K., P. Hull, and X. Jaravel (2020). Quasi-Experimental Shift-Share Research

Designs. Working Paper.

Borusyak, K. and X. Jaravel (2018). Revisiting Event Study Designs, with an Application
to the Estimation of the Marginal Propensity to Consume. Working Paper.

Brewer, M., T. F. Crossley, and R. Joyce (2017). Inference with Difference-in-Differences

Revisited. Journal of Econometric Methods 7(1).

Bugni, F. A. and I. A. Canay (2020). Testing Continuity of a Density via g-order statistics
in the Regression Discontinuity Design. Journal of Econometrics.

Callaway, B. and P. H. Sant’Anna (2020). Difference-in-Differences with Multiple Time

Periods. Working Paper.

Calonico, S., M. D. Cattaneo, and R. Titiunik (2014). Robust Nonparametric Confidence

Intervals for Regression-Discontinuity Designs. Econometrica 82(6), 2295–2326.

Canay, I. A. and V. Kamat (2018). Approximate Permutation Tests and Induced Order
Statistics in the Regression Discontinuity Design. Review of Economic Studies 85(3),
1577–1608.

Canay, I. A., A. Santos, and A. M. Shaikh (2018). The Wild Bootstrap with a “Small”
Number of “Large” Clusters. Review of Economics and Statistics, 1–45.

Cattaneo, M. D., M. Jansson, and X. Ma (2020). Simple local polynomial density estima-
tors. Journal of the American Statistical Association 115(531), 1449–1455.

29
Cattaneo, M. D., M. Jansson, and W. K. Newey (2018). Inference in Linear Regression
Models with Many Covariates and Heteroscedasticity. Journal of the American Statistical
Association 113(523), 1350–1361.

Cengiz, D., A. Dube, A. Lindner, and B. Zipperer (2019). The Effect of Minimum Wages on
Low-Wage Jobs. Quarterly Journal of Economics 134(3), 1405–1454.

Chernozhukov, V., I. Fernández-Val, and B. Melly (2013). Inference on Counterfactual

Distributions. Econometrica 81(6), 2205–2268.

Colella, F., R. Lalive, S. O. Sakalli, and M. Thoenig (2020). Inference with Arbitrary Clus-
tering. Working Paper.

de Chaisemartin, C. and X. d’Haultfoeuille (2020). Two-Way Fixed Effects Estimators with

Heterogeneous Treatment Effects. American Economic Review 110(9), 2964–96.

de Chaisemartin, C. and X. d’Haultfoeuille (2018). Fuzzy Differences-in-Differences. Re-

view of Economic Studies 85(2), 999–1028.

Evdokimov, K. S. and M. Kolesár (2019). Inference in Instrumental Variable Regression

Analysis with Heterogeneous Treatment Effects. Working Paper.

Ferman, B. and C. Pinto (2019). Inference in Differences-in-Differences with Few Treated

Groups and Heteroskedasticity. Review of Economics and Statistics 101(3), 452–467.

Fortin, N., T. Lemieux, and S. Firpo (2011). Decomposition Methods in Economics. In

Handbook of Labor Economics, Volume 4, pp. 1–102. Elsevier.

Freyaldenhoven, S., C. Hansen, and J. M. Shapiro (2019). Pre-event Trends in the Panel
Event-Study Design. American Economic Review 109(9), 3307–38.

Ganong, P. and S. Jäger (2018). A Permutation Test for the Regression Kink Design. Journal
of the American Statistical Association 113(522), 494–504.

Gelman, A. and G. Imbens (2019). Why High-Order Polynomials Should Not Be Used
in Regression Discontinuity Designs. Journal of Business & Economic Statistics 37(3),
447–456.

Goldsmith-Pinkham, P., I. Sorkin, and H. Swift (2020). Bartik Instruments: What, When,
Why, and How. American Economic Review 110(8), 2586–2624.

Goodman-Bacon, A. (2020). Difference-in-Differences with Variation in Treatment Timing.

Working Paper.

30
Grembi, V., T. Nannicini, and U. Troiano (2016). Do Fiscal Rules Matter? American
Economic Journal: Applied Economics, 1–30.

Ibragimov, R. and U. K. Müller (2016). Inference with Few Heterogeneous Clusters. Review
of Economics and Statistics 98(1), 83–96.

Imbens, G. W. and M. Kolesár (2016). Robust Standard Errors in Small Samples: Some
Practical Advice. Review of Economics and Statistics 98(4), 701–712.

Kleven, H. J. (2016). Bunching. Annual Review of Economics 8, 435–464.

Kolesár, M. and C. Rothe (2018). Inference in Regression Discontinuity Designs with a

Discrete Running Variable. American Economic Review 108(8), 2277–2304.

Lee, D. L., J. McCrary, M. J. Moreira, and J. Porter (2020). Valid t-ratio Inference for IV.
Working Paper.

Marcus, M. and P. H. Sant’Anna (2020). The Role of Parallel Trends in Event Study Set-
tings: An Application to Environmental Economics. Journal of the Association of Envi-
ronmental and Resource Economists, forthcoming.

Marx, B. M. (2020). Dynamic Bunching Estimation with Panel Data. Working Paper.

Mogstad, M. and A. Torgovitsky (2018). Identification and Extrapolation of Causal Effects

with Instrumental Variables. Annual Review of Economics 10, 577–613.

Mogstad, M., A. Torgovitsky, and C. Walters (2020a). Policy Evaluation with Multiple
Instrumental Variables. Working Paper.

Mogstad, M., A. Torgovitsky, and C. R. Walters (2020b). The Causal Interpretation of

Two-Stage Least Squares with Multiple Instrumental Variables. Working Paper.

Muralidharan, K., M. Romero, and K. Wüthrich (2019). Factorial Designs, Model Selection,
and (Incorrect) Inference in Randomized Experiments. NBER WP.

Otsu, T. and Y. Rai (2017). Bootstrap Inference of Matching Estimators for Average Treat-
ment Effects. Journal of the American Statistical Association 112(520), 1720–1732.

Rambachan, A. and J. Roth (2020). An Honest Approach to Parallel Trends. Working Paper.

Sant’Anna, P. H. and J. Zhao (2020). Doubly Robust Difference-in-Differences Estimators.

Journal of Econometrics 219(1), 101–122.

31
Słoczyński, T. (2020). Interpreting OLS Estimands When Treatment Effects Are Hetero-
geneous: Smaller Groups Get Larger Weights. Review of Economics and Statistics 0(ja),
1–27.

Young, A. (2019). Channeling Fisher: Randomization Tests and the Statistical In-
significance of Seemingly Significant Experimental Results. Quarterly Journal of Eco-
nomics 134(2), 557–598.

Young, A. (2020). Consistency Without Inference: Instrumental Variables in Practical

Application. Working Paper.

32