The Stata Journal Editors’ Prize 2019: Matias D. Cattaneo

1 Prize announcement

The editors of the Stata Journal are delighted to announce the award of the Editors’ Prize for 2019 to Matias D. Cattaneo.

The aim of the prize is to reward contributions to the Stata community in respect of one or more outstanding articles published in the Journal in the previous three calendar years. For the original announcement of the prize and its precise terms of reference, see Newton and Cox (2012), which is accessible at the following website: http://www.stata-journal.com/sjpdf.html?articlenum=gn0052.

Matias Damian Cattaneo was born in 1978 in Buenos Aires, Argentina. He received a licentiate in economics from the Universidad de Buenos Aires in 2000, a master’s degree in economics from the Universidad Torcuato Di Tella in 2003, a master’s degree in statistics from the University of California at Berkeley in 2005, and a PhD in economics also from Berkeley in 2008. He worked at the University of Michigan between 2008 and 2019, rising to professor in both economics and statistics. He now works at Princeton University as a professor in the Department of Operations Research and Financial Engineering and as an associated faculty in the Department of Economics and is also affiliated with its Center for Statistics and Machine Learning. He enjoys playing soccer competitively, hiking, running, and climbing. He also likes traveling with his family and eating many asados (barbecues) in Buenos Aires.

His research ranges over statistics, econometrics, data science, and quantitative methods, with particular interests in program evaluation and causal inference. Much of his work is interdisciplinary and motivated by empirical problems. Most notably, Matias has developed novel semiparametric and nonparametric inference procedures showing superior robustness to tuning parameter and other implementation choices.

At the time of writing, Matias was serving as associate editor for eight journals in statistics and economics. His many prizes and awards include a Statistical Software Award from the Society for Political Methodology. As of August 2019, his citations on Google Scholar are well past 4,000, his most cited article being Calonico, Cattaneo, and Titiunik (2014b) with nearly 1,200 citations.

The award recognizes specifically three outstanding articles by Cattaneo and coauthors (2016, 2017, 2018):

Inference in regression discontinuity designs under local randomization (Stata Journal 16: 331–367)

rdrobust: Software for regression discontinuity designs (Stata Journal 17: 372–404)

Manipulation testing based on density discontinuity (Stata Journal 18: 234–261)

The references detail all of Cattaneo’s publications in the Stata Journal. In addition, Calonico, Cattaneo, and Titiunik (2014a), his second most cited article, is approaching 400 citations on Google Scholar.

We turn now to detailed comments on the articles behind this award. Each combines the introduction of major new Stata commands with a thorough but accessible account of the associated theory and substantive examples based on real data.

In the 2016 article, using local randomization, the authors introduce the rdlocrand package, which contains four commands to conduct finite-sample inference in regression discontinuity designs under a local randomization assumption, following the framework and methods proposed earlier by Cattaneo and his coauthors. Assuming a known assignment mechanism for units close to the cutoff, these commands implement a variety of procedures based on randomization inference techniques. The rdrandinf command uses randomization methods to conduct point estimation, hypothesis testing, and con- fidence interval estimation under different assumptions. The rdwinselect command uses finite-sample methods to select a window near the cutoff where the assumption of randomized treatment assignment is most plausible. The rdsensitivity command uses randomization techniques to conduct a sequence of hypothesis tests for different windows around the cutoff, which can be used to assess the sensitivity of the methods and to construct confidence intervals by inversion. Finally, the rdrbounds command implements Rosenbaum’s sensitivity bounds for the context of such designs under local randomization.

The 2017 article describes a major upgrade to the rdrobust package, introduced in Calonico, Cattaneo, and Titiunik (2014a). The package provides a wide array of estimation, inference, and falsification methods for the analysis and interpretation of regression discontinuity designs, using local polynomial and related nonparametric methods. The main new features of this upgraded version are covariate-adjusted bandwidth selection, point estimation, and robust bias-corrected inference; cluster–robust bandwidth selection, point estimation, and robust bias-corrected inference; weighted global-polynomial fits and pointwise confidence bands in regression discontinuity plots; and several new bandwidth-selection methods, including different bandwidths for control and treatment groups, coverage error-rate optimal bandwidths, and optimal bandwidths for fuzzy designs. In addition, the upgraded package has superior performance because of several numerical and implementation improvements.

In the 2018 article, two further commands, rddensity and rdbwdensity, are introduced that implement automatic manipulation tests based on density discontinuity and using local polynomial density estimators. These new tests exhibit better size properties (and more power under additional assumptions) than other conventional approaches currently available in the literature. The first command, rddensity, implements manipulation tests based on a novel local polynomial density estimation technique that avoids prebinning of the data (improving size properties) and allows for restrictions on other features of the model (improving power properties). The second command, rdbwdensity, implements several bandwidth selectors specifically tailored for the manipulation tests in rddensity.

Each article gives a detailed and thorough explanation of the commands in question. They share as a major source of illustrations a dataset on 1,390 U.S. Senate elections between 1914 and 2010. The overarching aim is to analyze the effect of the incumbent status of a political party on the probability of winning future elections. The running variable in this dataset is the Democratic margin of victory in a statewide Senate election, defined as the difference in vote share between the Democratic party and its strongest opponent. A positive value of the running variable indicates that the Democratic party won the election, thus implying a cutoff of zero.

The idea of regression discontinuity design has long roots, starting with its introduction by Thistlethwaite and Campbell (1960). In the last two decades, it has become one of the central methods for causal inference and program evaluation in econometrics and other statistical sciences. The bold aim, tempered by caution and calculation, is to capture the “experiments”, whether witting or unwitting, embedded within observational data but structured by changes of circumstance. Researchers within this exciting and expanding territory need toolkits, and Matias Cattaneo and his colleagues have made major contributions of well-thought-out tools within Stata (and, we should mention, also within R).

In sum, we salute Matias Cattaneo for outstanding contributions to the Stata community through his recent publications in the Stata Journal.

As editors, we are indebted to the awardee for biographical material and to necessarily anonymous nominators for most helpful appreciations.

H. Joseph Newton and Nicholas J. Cox Editors, Stata Journal