Struct EventStudy 

pub struct EventStudy {
    pub n_pre: usize,
    pub n_post: usize,
}

Event-study design for estimating dynamic treatment effects.

Estimates the regression: y_{it} = Σ_{l≠-1} β_l D^l_{it} + α_i + δ_t + ε_{it} where D^l_{it} = 1 if unit i is treated and is l periods from its treatment date. The omitted period is l = -1 (normalisation).

Fields

n_pre: usize

How many pre-treatment periods to include (>= 1)

n_post: usize

How many post-treatment periods to include (>= 1)

Implementations

impl EventStudy

pub fn new(n_pre: usize, n_post: usize) -> Self

Create a new EventStudy estimator.

pub fn estimate( &self, y: &ArrayView1<'_, f64>, treated: &ArrayView1<'_, f64>, n_units: usize, n_periods: usize, treat_period: usize, ) -> StatsResult<EventStudyResult>

Estimate dynamic treatment effects.

Arguments

• y – outcome vector (n_units × n_periods, row-major)
• treated – binary indicator per unit (n_units,)
• n_units – number of units
• n_periods – number of time periods
• treat_period– the first treatment period for all treated units (0-indexed)