Skip to main content

Table 5 Nonparametric fuzzy regression discontinuity (RD) estimates—education heterogeneity

From: The effect of retirement on physical and mental health in China: a nonparametric fuzzy regression discontinuity study

 

Self-reported health

Life satisfaction

CESD scale

Low educated

High educated

Low educated

High educated

Low educated

High educated

MSE Optimal

 Bandwidth

2.6

3.8

3

2.7

3.1

3.3

 Conventional Point Estimate

0.0866

0.1072

0.0211

-0.1519

4.6093

-0.7161

 Conventional 95% CI

[-0.5828, 0.756]

[-0.3637, 0.5780]

[-0.3572, 0.3993]

[-0.5738, 0.2701]

[-4.6402, 13.8588]

[-7.7306, 6.2984]

 Robust Point Estimate

0.0527

0.1343

0.1302

-0.1776

5.7265

-1.9297

 Robust 95% CI

[-0.7494, 0.8548]

[-0.4333, 0.7019]

[-0.3128, 0.5733]

[-0.6840, 0.3289]

[-5.4867, 16.9396]

[-10.1049, 6.2455]

 Effective obs L|R

394|437

573|419

422|525

377|296

434|543

480|365

CER Optimal

 Bandwidth

1.7

2.6

2

1.8

2

2.3

 Conventional Point Estimate

-0.0584

0.1973

0.2163

-0.2383

6.5837

-1.5142

 Conventional 95% CI

[-1.0249, 0.908]

[-0.3613, 0.7559]

[-0.3146, 0.7473]

[-0.8462, 0.3696]

[-7.6886, 20.8559]

[-9.8794, 6.851]

 Robust Point Estimate

-0.0853

0.2070

0.2675

-0.2484

7.0552

-2.0292

 Robust 95% CI

[-1.1280, 0.9574]

[-0.3988, 0.8127]

[-0.3055, 0.8405]

[-0.9018, 0.4050]

[-8.3616, 22.4719]

[-10.9561, 6.8977]

 Effective obs L|R

281|306

388|307

297|331

257|225

316|347

333|255

Obs L|R

1407|1938

1594|864

1339|1905

1529|842

1355|1909

1537|842

  1. (i) All estimates are computed using a triangular Kernel and nearest neighbor heteroskedasticity-robust variance estimator as suggested by CCT framework. (ii) Effective number of observations depends on the size of the data-driven selected bandwidth
  2. All the p-values > 0.05