Estimation uncertainty in repeated finite populations
Abstract
Draft
Standard errors need to be adjusted down when the sample is a large fraction of the population of interest (a finite population setup). I consider the empirically relevant case where a finite population coexists with a measurement problem,
in that the features of interest are not necessarily observable even if the entire population is sampled. I show that conventional standard errors remain generally conservative in this context and propose Finite Population Corrections (FPCs)
that guarantee non-conservative inference when repeated measurements are available. FPCs rely on weak dependence across measurements and are very simple to implement. I apply these methods to two empirical settings where uncertainty has
been previously understood in different ways: predicting lethal encounters with police using data on all U.S. police departments, and studying firm misallocation with a census of large Indonesian firms. Finite-population inference leads to
confidence intervals that are up to 50% shorter in the former and illustrates the need to account for measurement uncertainty in the latter.
Micro responses to macro shocks
w/
Martín Almuzara (
submitted)
Abstract
Draft
Supplement
Slides (NBER SI)
Code
We study panel data regression models when the shocks of interest are aggregate and possibly small relative to idiosyncratic noise. This speaks to a large empirical literature that targets impulse responses via panel local projections.
We show how to interpret the estimated coefficients when units have heterogeneous responses and how to obtain valid standard errors and confidence intervals. A simple recipe leads to robust inference: including lags as controls and then clustering at the time level.
This strategy is valid under general error dynamics and uniformly over the degree of signal-to-noise of macro shocks.
Estimating flexible income processes from subjective expectations data: evidence from India and Colombia
w/
Manuel Arellano,
Orazio Attanasio and
Sam Crossman (
submitted)
Abstract
Draft
NBER version
Slides
Code
We develop a methodology for modeling household income processes when subjective probabilistic assessments of future income are available. This allows us to flexibly estimate conditional cdfs directly using elicited individual subjective probabilities,
and to obtain empirical measurements of subjective risk and persistence. We then use two longitudinal surveys collected in rural India and rural Colombia to explore the nature of income dynamics in those contexts. Our results suggest linear income processes
are rejected in favor of more flexible versions in both cases; subjective income distributions feature heteroskedasticity, conditional skewness and nonlinear persistence.
Heterogeneity in impulse response functions
w/
Martín Almuzara
Abstract
Slides
The extent of heterogeneity in impulse responses is a key empirical target for many substantive questions. For instance, it is helpful to understand the transmission of monetary policy shocks to household- and firm-level outcomes.
We propose a method to estimate the cross-sectional distribution of impulse responses in a panel data context. Our method combines the simplicity of estimating separate standard local projections for each
horizon with additional common structure that is naturally present across units and across horizons. From a practical point of view, this structure facilitates documenting and summarizing heterogeneity;
from a methodological one, it affords dimensionality reduction that we exploit in estimation. Preliminary simulation evidence supports these findings.
How fixed are fixed-effects? Evidence from the Spanish labor market
Abstract
A popular approach to account for labor market heterogeneity views aggregate fluctuations as driven by a particular group of workers with low attachment.
Behind this is an assumption that a worker's type is approximately constant over the relevant horizon. I propose an econometric framework to evaluate type stability that exploits the nature
of longitudinal data and is based on comparing the posterior probability that a worker belongs to a particular unobserved type over short, consecutive panel segments.
I find evidence of substantial type dynamics in an application to labor market transitions in Spain over 2016–2018, although preliminary evidence suggests that these are attributable to a relatively small fraction of units.
