Hello Statlisters,

I apply a DID approach on trade diversion in regard of antidumping duties imposed by the EU. I've a treatment group of the 10 product codes and 10 products in a control group. Everything is based on CN8 CIF data. I use a LSDV estimator which is equivalent to a FE estimator. I'm puzzled as controlling for different clusters within standard errors leads varying significance.
The literature on DID estimators ("How Much Should We Trust Differences-In-Differences", Bertrant et al, 2004) suggests that se are biased for small number of clusters. Having read this, the assignment into only two clusters based on the not randomly assigned groups seems untenable. Though, this would represent the cluster structure caused by group selection and leads to outstanding significance of the coefficients of interest d_gi d_gt d_gr.

1. reg ln_value_2 $t i.product2 d_gi d_gt d_gr d_treatment_initiation d_treatment_time d_treatment_repeal , robust
2. reg ln_value_2 $t i.product2 d_gi d_gt d_gr d_treatment_initiation d_treatment_time d_treatment_repeal , vce(cluster product2) - 20 clusters
3. reg ln_value_2 $t i.product2 d_gi d_gt d_gr d_treatment_initiation d_treatment_time d_treatment_repeal , vce(cluster d_treatment_group) - 2 clusters


Code:
* Example generated by -dataex-. To install: ssc install dataex
clear
input float(ln_value_2 period2) long product2 float(d_gi d_gt d_gr d_treatment_initiation d_treatment_time d_treatment_repeal)
15.698238 540 1 0 0 0 0 0 0
15.514578 541 1 0 0 0 0 0 0
15.483043 542 1 0 0 0 0 0 0
15.618363 543 1 0 0 0 0 0 0
15.482442 544 1 0 0 0 0 0 0
15.476822 545 1 0 0 0 0 0 0
15.433095 546 1 0 0 0 0 0 0
 15.38793 547 1 0 0 0 0 0 0
15.684005 548 1 0 0 0 0 0 0
15.556653 549 1 0 0 0 0 0 0
15.652306 550 1 0 0 0 0 0 0
15.163828 551 1 0 0 0 0 0 0
15.630563 552 1 0 0 0 0 0 0
15.515283 553 1 0 0 0 0 0 0
15.451436 554 1 0 0 0 0 0 0
15.295702 555 1 0 0 0 0 0 0
15.516436 556 1 0 0 0 0 0 0
15.451324 557 1 0 0 0 0 0 0
15.420912 558 1 0 0 0 0 0 0
 15.68658 559 1 0 0 0 0 0 0
15.731094 560 1 0 0 0 0 0 0
15.666362 561 1 0 0 0 0 0 0
 15.72442 562 1 0 0 0 0 0 0
15.512174 563 1 0 0 0 0 0 0
15.919752 564 1 0 0 0 0 0 0
 15.73842 565 1 0 0 0 0 0 0
 15.87069 566 1 0 0 0 0 0 0
 15.81873 567 1 0 0 0 0 0 0
15.794265 568 1 0 0 0 0 0 0
15.734446 569 1 0 0 0 0 0 0
15.813786 570 1 0 0 0 0 0 0
 15.71266 571 1 0 0 0 0 0 0
15.716444 572 1 0 0 0 0 0 0
15.648077 573 1 1 0 0 1 0 0
 15.58461 574 1 1 0 0 1 0 0
  15.1336 575 1 1 0 0 1 0 0
 15.76287 576 1 1 0 0 1 0 0
15.593745 577 1 1 0 0 1 0 0
15.368608 578 1 1 0 0 1 0 0
15.520124 579 1 1 0 0 1 0 0
15.428104 580 1 1 0 0 1 0 0
15.502225 581 1 1 0 0 1 0 0
15.668934 582 1 1 0 0 1 0 0
15.562196 583 1 1 0 0 1 0 0
15.778552 584 1 1 0 0 1 0 0
15.653923 585 1 1 0 0 1 0 0
15.487323 586 1 1 0 0 1 0 0
 15.38042 587 1 1 0 0 1 0 0
15.570642 588 1 1 0 0 1 0 0
15.346223 589 1 1 1 0 1 1 0
 15.63006 590 1 1 1 0 1 1 0
  15.4768 591 1 1 1 0 1 1 0
15.575624 592 1 1 1 0 1 1 0
15.848676 593 1 1 1 0 1 1 0
  15.7995 594 1 1 1 0 1 1 0
15.320642 595 1 1 1 0 1 1 0
15.727983 596 1 1 1 0 1 1 0
15.749194 597 1 1 1 0 1 1 0
15.925272 598 1 1 1 0 1 1 0
 15.70578 599 1 1 1 0 1 1 0
15.994413 600 1 1 1 0 1 1 0
16.024767 601 1 1 1 0 1 1 0
16.053577 602 1 1 1 0 1 1 0
 16.07567 603 1 1 1 0 1 1 0
16.101507 604 1 1 1 0 1 1 0
16.251797 605 1 1 1 0 1 1 0
16.272213 606 1 1 1 0 1 1 0
16.115023 607 1 1 1 0 1 1 0
16.289412 608 1 1 1 0 1 1 0
16.223812 609 1 1 1 0 1 1 0
 16.31194 610 1 1 1 0 1 1 0
 16.04048 611 1 1 1 0 1 1 0
16.333113 612 1 1 1 0 1 1 0
 16.26534 613 1 1 1 0 1 1 0
16.257803 614 1 1 1 0 1 1 0
16.097738 615 1 1 1 0 1 1 0
 16.38007 616 1 1 1 0 1 1 0
16.266039 617 1 1 1 0 1 1 0
16.232845 618 1 1 1 0 1 1 0
16.342772 619 1 1 1 0 1 1 0
16.456781 620 1 1 1 0 1 1 0
16.445354 621 1 1 1 0 1 1 0
16.260475 622 1 1 1 0 1 1 0
 16.06071 623 1 1 1 0 1 1 0
 16.41099 624 1 1 1 0 1 1 0
16.282658 625 1 1 1 0 1 1 0
16.266817 626 1 1 1 0 1 1 0
16.140076 627 1 1 1 0 1 1 0
16.263208 628 1 1 1 0 1 1 0
16.226574 629 1 1 1 0 1 1 0
16.197224 630 1 1 1 0 1 1 0
16.224636 631 1 1 1 0 1 1 0
16.079723 632 1 1 1 0 1 1 0
 16.29428 633 1 1 1 0 1 1 0
16.165276 634 1 1 1 0 1 1 0
15.904102 635 1 1 1 0 1 1 0
16.253433 636 1 1 1 0 1 1 0
16.198286 637 1 1 1 0 1 1 0
16.157755 638 1 1 1 0 1 1 0
16.154299 639 1 1 1 0 1 1 0
end
format %tm period2
label values product2 product2
label def product2 1 "73181290", modify
Array