sim_check

Checking input for the sim_paneldata function

Testing for parallel trends is crucial in the Difference-in-Differences framework. To this end, this package performs equivalence testing in the context of Difference-in-Differences estimation. It allows users to test if pre-treatment trends in the treated group are “equivalent” to those in the control group. Here, “equivalence” means that rejection of the null hypothesis implies that a function of the pre-treatment placebo effects (maximum absolute, average or root mean squared value) does not exceed a pre-specified threshold below which trend differences are considered negligible. The package is based on the theory developed in Dette & Schumann (2024) <doi:10.1080/07350015.2024.2308121>.

Ties Bos

EquiTrends

Equivalence Testing for Pre-Trends in Difference-in-Differences
Designs

Martin Schumann

sim_check function

<dl><dt>N</dt>
<dd>The number of cross-sectional units in the panel-data</dd>
<dt>tt</dt>
<dd>The number of time periods in the panel-data</dd>
<dt>beta</dt>
<dd>The vector of coefficients for the placebo variables. Must be of size tt.</dd>
<dt>p</dt>
<dd>The number of additional regressors</dd>
<dt>gamma</dt>
<dd>The vector of coefficients for the additional regressors</dd>
<dt>eta</dt>
<dd>The vector of fixed effects. Must be of size N.</dd>
<dt>lambda</dt>
<dd>The vector of time effects. Must be of size tt.</dd>
<dt>het</dt>
<dd>The heteroskedasticity parameter. Must be 0 or 1: <code>het = 1</code> indicates that the error terms are generated under heteroskedasticity, <code>het = 0</code> indicates the error terms are generated under homoscedasticity.</dd>
<dt>phi</dt>
<dd>The AR(1) parameter for the error terms. Must be in the interval [0,1).</dd>
<dt>sd</dt>
<dd>The standard deviation of the error terms. Must be a positive number.</dd>
<dt>burnins</dt>
<dd>The number of burn-ins for the AR(1) process. Must be a positive integer.</dd></dl>

Arguments

Checking input for the sim_paneldata function — sim_check

<dl>

<dt>N</dt>
<dd>The number of cross-sectional units in the panel-data</dd>


<dt>tt</dt>
<dd>The number of time periods in the panel-data</dd>


<dt>beta</dt>
<dd>The vector of coefficients for the placebo variables. Must be of size tt.</dd>


<dt>p</dt>
<dd>The number of additional regressors</dd>


<dt>gamma</dt>
<dd>The vector of coefficients for the additional regressors</dd>


<dt>eta</dt>
<dd>The vector of fixed effects. Must be of size N.</dd>


<dt>lambda</dt>
<dd>The vector of time effects. Must be of size tt.</dd>


<dt>het</dt>
<dd>The heteroskedasticity parameter. Must be 0 or 1: <code>het = 1</code> indicates that the error terms are generated under heteroskedasticity, <code>het = 0</code> indicates the error terms are generated under homoscedasticity.</dd>


<dt>phi</dt>
<dd>The AR(1) parameter for the error terms. Must be in the interval [0,1).</dd>


<dt>sd</dt>
<dd>The standard deviation of the error terms. Must be a positive number.</dd>


<dt>burnins</dt>
<dd>The number of burn-ins for the AR(1) process. Must be a positive integer.</dd>

</dl>

sim_check: Checking input for the sim_paneldata function

Description

Usage

Value

Arguments