Description
Measures how many people cluster just at a threshold (free shipping minimum, plan limit, deadline) relative to a smooth counterfactual, revealing behavioural responses.
When a price schedule has a kink or notch, such as a tax bracket or a free shipping minimum, people bunch just at or above it. Bunching estimators fit a smooth polynomial to the distribution away from the threshold, estimate the counterfactual density in the excluded region, and compute excess mass. This quantifies responsiveness (an elasticity), and gaps above notches reveal dominated choices people avoid. Product pricing and limits are full of thresholds: free tier caps, minimum order values, character limits, streak deadlines. Bunching reveals how strongly users react and whether thresholds distort behaviour in desired ways.
Attributes
- Discipline
- Behavioral Science & Economics
- Loss type
- Unmigrated
- Stage
- Post-launch
- Data
- Quantitative
- Effort
- Days
Steps
- Identify a threshold in a product rule (e.g. free shipping at $50) and obtain the distribution of the running variable (basket value).
- Plot a fine histogram and look for excess mass at the threshold and missing mass nearby.
- Fit a flexible polynomial to bins outside an excluded window around the threshold.
- Estimate excess mass and its standard error via bootstrap of residuals.
- Translate into a behavioural elasticity or count of users changing behaviour, and check sensitivity to window and polynomial choices.
- Use findings to set or redesign thresholds and test changes experimentally.
Sources
- Kleven, Bunching (2016), Annual Review of Economics
- Saez, Do taxpayers bunch at kink points? (2010), American Economic Journal: Economic Policy
- Kleven & Waseem, Using notches to uncover optimization frictions (2013), Quarterly Journal of Economics
Also known as
- Bunching analysis
- Kink and notch analysis
- Threshold behaviour analysis
Categories
- Analytics & Behaviour
- Behavioral Science & Economics