Publication: dqd: A command for treatment effect estimation under alternative assumptions
Loading...
Identifiers
Publication date
2014-04-01
Defense date
Authors
Advisors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Export
Abstract
Conventional difference-in-differences (DID) methods that are used to estimate the effect of a
treatment rely on important identifying assumptions. Identification of the treatment effect in a
DID framework requires some assumption relating trends for controls and treated in absence of
treatment, the most common being the assumption of Parallel Paths. When several pre-treatment
periods are available, Mora and Reggio (2012) show that treatment effect identification does not
uniquely depend on the Parallel Path assumption, but also on the trend modeling strategy. They
further define a family of alternative Parallel assumptions and propose a more flexible model
which can be a helpful starting tool to study robustness to alternative Parallel assumptions and
trend dynamics. In this paper we introduce a Stata command that implements the fully flexible
model presented in Mora and Reggio (2012), producing tests for the equivalence of alternative
parallel assumptions and for the dynamic effects of the treatment. The standard DID in model
with or without polynomial trends can also be obtained.
Description
Keywords
Difference-in-differences, treatment effect, identification, fully flexible model, dqd