Financial Outcome Winsorization
RCTClinical Trial
Financial remittance datasets frequently contain extreme high-value outliers that distort sample means and inflate standard errors in standard regression analyses. Applying top-tail truncation or winsorization preserves sample size while preventing extreme observations from driving treatment effect estimates.
Picture this
Imagine scoring a gymnastics competition where one judge gives a score of 100 while everyone else scores between 5 and 10. To prevent that single extreme score from ruining the average, you cap any score higher than 10 down to exactly 10 before calculating the final result.
What the evidence says
Winsorization bounded weekly remittance amounts (control mean 2,338.50 PhP), stabilizing regression estimates across full sample (p = 0.81) and low-baseline subsample (+174.94 PhP, p < 0.05) specifications.
- Who was studied
- 137,927 weekly remittance observations across 4,451 analyzed migrant study participants.
- How
- Capping non-zero weekly remittance amounts at the 99th percentile of the full distribution prior to panel regression estimation.
What to do
Cap skewed financial outcome distributions at the 99th percentile to prevent extreme top-tail outliers from distorting causal inference.
From the source
"Amount sent: total in Philippine pesos (PhP), summed across all transactions, winsorized at 99th percentile of the full distribution of non-zero weekly amount sent in the data."
A Field Experiment among Filipino Migrant Workers in the UAE