Inference Based on Conditional Moment Inequalities

By Donald W. K. Andrews, Xiaoxia Shi. 2013

268 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/74fd69f68c927669446f267360c20031a31eb7fd

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Statistics and Probability

Citations in the corpus

When Is Parallel Trends Sensitive to Functional Form? (2020, methodology): “When Y has continuous support, one can apply a similar approach by first discretizing the outcome variable, or by using methods for testing a continuum of moment inequalities (e.g. Andrews and Shi (2013)).”
Inference on Directionally Differentiable Functions (2014, background): “We discuss the relationship between these conditions in detail in Section 3.2; see Remark 3.1.”
Asymptotically exact inference in conditional moment inequality models (2011, result): “1 Related Literature The results in this paper relate to recent work on testing conditional moment inequalities, including papers by Andrews and Shi (2009), Kim (2008), Khan and Tamer (2009), Chernozhukov, Lee, and Rosen (2009), Lee, Song, and Whang (2011), Ponomareva (2010), Menzel (2008) and Armst”
Specification tests for partially identified models defined by moment inequalities (2013, background): “5 is satisfied if the function φ is any of the the functions φ−φ described in Andrews and Soares (2010) or Andrews and Barwick (2012).”
Who should be Treated? Empirical Welfare Maximization Methods for Treatment Choice (2015, background): “(2010) and Donald and Hsu (2016), as well as the literature on moment inequalities with moment selection (Andrews and Shi (2013), among others).”
Weighted KS Statistics for Inference on Conditional Moment Inequalities (2011, methodology): “Whereas the KS type statistics considered by Andrews and Shi (2009) and Kim (2008) are defined as the supremum of a random process that converges to a tight random process, this does not hold with the increasing truncation points for the inverse variance weighting used here. Thus, while the statisti”
TESTING FOR A GENERAL CLASS OF FUNCTIONAL INEQUALITIES (2013, methodology): “2A working paper version (Andrews and Shi 2009) of Andrews and Shi (2013) covers testing moment inequalities indexed by τ ∈ T , but their framework does not appear to be easily extendable to deal with functions of multiple conditional quantiles such as differences in conditional quantiles.”
Inference on Causal and Structural Parameters using Many Moment Inequalities (2013, background): “j = 1,...,p ′ (3) where (Y,Z) is a pair of random vectors and g1,...,gp′ is a set of functions with p′ being large. Let Z be the support of Z and assume that Z is a compact set in Rl. Then, following Andrews and Shi (2013), one can construct an inﬁnite set I of instrumental functions I : Z → Rsuch t”
Testing Many Moment Inequalities (2013, background): “Then, following Andrews and Shi (2013), one can construct an infinite set I of instrumental functions I : Z → R such that I(z) ≥ 0 for all z ∈ Z and (3) holds if and only if E[gj(Y )I(Z)] ≤ 0 for all j = 1, .”
Adaptive test of conditional moment inequalities (2011, methodology): “The tests of Andrews and Shi (2010) and Lee et al. (2011) have nontrivial power against n−1/2-local one-directional alternatives whereas my method only allows for nontrivial testing against (n/ logn)−1/2-local alternatives of this type.”
Partial identification using random set theory (2010, methodology): “4, estimation and statistical inference can be carried out using the methods proposed by Andrews and Shi (2009), Chernozhukov, Lee, and Rosen (2009) and Ponomareva (2010). Estimation of sharp identification regions of types (b) and (c) with conditional or unconditional moment inequalities can be car”
Improved central limit theorem and bootstrap approximations in high dimensions (2019, unknown): “ to E[m(W)]. We will use this property extensively below without explicit mentioning. Now, ﬁx y∈ Rpand observe that Iy= X|Id| i=1 my Wσ i + V√σ(i) n −my Wσ i + Z√σ(i) n . Hence, letting f: [0,1] → Rbe a function deﬁned by f(t) = X|Id| i=1 E my Wσ i + tV√σ(i) n −my Wσ i + tZ√σ(i) n , for al”