Table 1

Effects of municipal council size in local elections

Electoral outcomes
(1)(2)(3)
SeatsTotal candidatesCandidates per seat
β bias-correcteds.e.0.726***11.494*0.582
(0.145)(5.046)(0.467)
Bandwidth1,7361,4691,109
Observations2,0751,8231,464
Racial representation
(4)(5)(6)(7)(8)
Non-white candidatesNon-white electedAt least 1 non-whiteShare of non-white candidatesShare of non-white elected
β bias-correcteds.e.7.2681.136**0.0355.715*5,503
 (3.849)(0.357)(0.022)(2.735)(3.051)
Bandwidth1,3361,4052,2221,7171,781
Observations1,7281,9262,6092,1032,276
Gender representation
(9)(10)11)(12)(13)
Women candidatesWomen electedAt least 1 womanShare of women candidatesShare of women elected
β bias-correcteds.e.3.718*0.0290.014−0.377−0.473
(1.745)(0.097)(0.036)(0.241)(0.754)
Bandwidth1,3702,9872,7841,4325,639
Observations1,7252,9182,7951,7884,552
Age representation
(14)(15)(16)(17)(18)(19)(20)(21)
Candidates under 30Share of candidates under 30Elected under 30Share of elected under 30Candidates over 60Share of candidates over 60Elected over 60Share of elected over 60
β bias-correcteds.e.1.636**0.6850.000−0.0420.911−0.278−0.046−0.974
(0.531)(0.360)(0.069)(0.670)(0.665)(0.406)(0.066)(0.796)
Bandwidth1,4553,7343,0243,5021,6352,5834,1422,701
Observations1,8313,4243,0703,3962,0152,7054,1823,144
Educational attainment representation
(22)(23)(24)(25)(26)(27)(28)(29)
High school graduates candidatesShare of high school graduates candidatesHigh school graduates electedShare of high school graduates electedCollege graduates candidatesShare of college graduates candidatesCollege graduates electedShare of college graduates elected
β bias-correcteds.e.7.358*0.6610.858***2.7862.1471.566*0.497*3.392
(3.416)(1.040)(0.236)(1.697)(1.225)(0.781)(0.194)(1.868)
Bandwidth1,4442,0061,3681,8851,3781,5861,6371,366
Observations1,8012,2631,7222,1841,7491,9612,0011,753

Note(s): This table reports the reduced-form effects of council size on the pool of candidates and elected councilors in local elections. The estimates are derived from regression discontinuity (RD) designs, utilizing a local polynomial estimator with a uniform kernel. The optimal bandwidth for each regression, minimizing mean squared error (MSE), was chosen following the method outlined in Calonico et al. (2014, 2018). Significance levels are denoted by: ***p < 0.01, **p < 0.05, *p < 0.1

Source(s): Table by authors

or Create an Account

Close Modal
Close Modal