Boundary Discontinuity Designs

The rd2d package provides R implementations of pointwise and uniform estimation and inference for Boundary Discontinuity Designs employing local polynomial methods. The main functions are rd2d() and rdbw2d() for location-based methods, and rd2d.distance() and rdbw2d.distance() for distance-based methods. Both approaches support sharp and fuzzy designs. The distance-based methods target the level of the boundary average treatment effect curve using bivariate scores reduced to signed distance-based running variables. Summary methods can also report weighted and largest boundary average treatment effects (WBATE and LBATE) when the fitted object stores the required covariance matrix, including intention-to-treat, first-stage, and fuzzy outputs.

R Implementation

To install/update in R type:

install.packages('rd2d')

• Help: R Manual, CRAN repository.

• Illustration: Simulation and estimation, Plots.

References

For overviews and introductions, see rdpackages website.

Software and Implementation

• Cattaneo, Titiunik and Yu (2025): rd2d: Causal Inference in Boundary Discontinuity Designs.
  Working paper.

Technical and Methodological

• Cattaneo, Titiunik and Yu (2026): Boundary Discontinuity Designs: Theory and Practice.
  Advances in Economics and Econometrics: Thirteenth World Congress, eds. R. Griffith, Y. Gorodnichenko, M. Kandori and F. Molinari, Cambridge University Press, Vol. 1, Ch. 2, to appear, 2026.

• Cattaneo, Titiunik and Yu (2026): Estimation and Inference in Boundary Discontinuity Designs: Location-Based Methods.
  Journal of the American Statistical Association, revise and resubmit.
  Supplemental Appendix.

• Cattaneo, Titiunik and Yu (2026): Estimation and Inference in Boundary Discontinuity Designs: Distance-Based Methods.
  Journal of Econometrics, revise and resubmit.
  Supplemental Appendix.

• Cattaneo, Titiunik and Yu (2026): Estimation and Inference in Boundary Discontinuity Designs: Pooling-Based Methods.
  Working paper.
  Supplemental Appendix.

Funding

This work was supported in part by the National Science Foundation through grants SES-2019432, DMS-2210561, SES-2241575, and SES-2342226, and by the National Institute for Food and Agriculture through grant 2024-67023-42704.