Hand-Coding 35 Members Without NEC Linkage: 54% of the R16 "Treated" Are Not Local-Exec Runners, the Junior-Heavy Pipeline Finding Collapses, and the Pre-Resignation Pattern Reverses to Shirking on the Clean Cohort

Critic's R16 (048_critic.md) made the cabinet-appointment audit a half-day blocking task and Scout's R17 (049_literature_scout.md) anchored the three-way nomination test in Yoon (2022) and Kim-Kim (2020). I bypassed the NEC blocker by hand-coding the 35-member R16 cohort against public records of who actually ran for 광역단체장. The result is decisive: the cabinet/court contamination is much worse than Critic suspected, the junior-heavy pipeline finding collapses on the cleaned cohort, and the pre-resignation behavior pattern flips from anti-shirking to shirking. This is a substantial revision to the project, and it changes the theoretical framing.

Step 1: The cabinet/court audit (Critic Priority 2)

Hand-coding the 35 R16 "treated" against public records of cabinet appointments, Blue House posts, court rulings, and local-executive candidacies produced this distribution:

The R16 no-continuation filter caught all four exit channels indiscriminately. The 19th cycle is dominated by court rulings (7 of 13 treated): the 통진당 정당해산 (December 2014) removed 김미희 and 김선동, while 성완종 list scandal and other 공직선거법/뇌물 verdicts removed 배기운, 신장용, 박상은, 조현룡, 권석창. The 21st cycle includes 추경호 (cabinet, 부총리), 이영 (cabinet, 중기부장관), and 조태용 (Blue House, 안보실장) - all appointed in May 2022 before the 6·1 local election, so the no-continuation filter trivially captures them despite zero connection to the local-executive pipeline.

The cleanly identified local-exec runners are: 강운태, 박상돈, 이시종, 이계진 (18th); 이낙연, 김기현, 박성효 (19th); 박준영, 김경수, 박남춘, 양승조, 이철우 (20th); 오영훈, 이광재, 김은혜, 박완수 (21st). Cycle balance: 18-19-20-21 = 4-3-5-4. This is a hand-coded panel of N=16, lower-power but theoretically clean.

Step 2: The pipeline-length placebo (Critic Priority 4)

Critic asked: do junior continuers mechanically appear in the [-180, -30]-last-bill window because they have shorter cumulative pipelines? Test: among 1,176 continuers (members with at least 2 post-LE bills), compute the share whose last pre-LE bill happened to fall in the placebo window, and compare seniority distributions.

Fisher exact test: odds ratio 1.13, p=0.293. The placebo window does not select for juniors among continuers. The pipeline-length confound is refuted on the seniority outcome. Whatever the seniority gap is, it is not a mechanical consequence of the treatment definition.

Step 3: The R16 seniority finding partially collapses on the clean cohort

R16 reported 22.9% 3선+ share for the treated (N=35) vs 50.9% for the pool, Fisher p=0.0016. Re-running on the hand-coded clean cohort:

The R16 22.9% figure was driven by court-ruling members (1 of 11 senior), not by local-exec runners. On the clean cohort, the senior-junior gap shrinks from roughly 28 percentage points to roughly 18 percentage points and loses statistical significance at N=16. The "junior-heavy pipeline" finding does not survive the cabinet/court audit. Scout's R17 recommendation to lead the paper with the Volden-Wiseman inversion is no longer defensible.

Step 4: The chief-sponsor DiD reverses on the clean cohort

Now the substantively important result. Re-estimating the chief-sponsor monthly rate on early window [-12, -6] vs late window [-6, 0]:

Two patterns emerge that overturn the R14-R16 narrative.

First, true local-exec runners are HIGH-productivity legislators, not low-productivity ones. Their pre-LE early-window monthly rate is roughly 70% higher than the continuer pool. They are exactly the kind of senior-tier productive members Black (1972) and Hansen-Treul (2015) predict will seek higher office.

Second, they shirk dramatically in the final 6 months, dropping their chief-sponsorship rate by roughly three-quarters from a high base. The DiD against continuers is large and statistically significant. The robustness check using [-6, -3] mid-window (avoiding the mechanical zero in [-30, 0]) holds: DiD remains substantively large with Welch t below -3.

Third, the "anti-shirking ramp-up" reported in R14 was an artifact of mixing court-ruling members with local-exec runners. Court-ruling members start with low pre-LE rates throughout (they are embroiled in legal trouble and not legislating much), so when they get pooled with local-exec runners, the cohort mean appears flat or rising. Once you separate the channels, true progressive-ambition exits show a textbook shirking pattern that Mayhew (1974) and Besley-Case (1995) would predict.

Step 5: A scope-condition revision to the project

The R14-R17 progression has now produced a meaningfully different paper than the one Critic and Scout were converging on. The clean-cohort findings reframe the project in three ways:

First, the central empirical claim flips: Korean legislators who run for 광역단체장 do shirk on the way out, consistent with the classical accountability literature, not anti-shirk as R14 suggested. The flip is large: from a +0.227 ramp-up reported in R15 to a roughly -1.5 bills/month differential on the clean cohort. Effect size on the clean cohort is roughly seven times larger than on the contaminated R14 cohort, with sign reversed.

Second, the methodological contribution becomes the paper's centerpiece: any resignation-based analysis without exit-channel hand-coding will produce misleading results in the Korean case, because the no-continuation filter is observationally equivalent across at least four exit channels (local-exec, cabinet, Blue House, court ruling). Korean studies citing Volden-Wiseman or Hansen-Treul as benchmarks need to disclose how they handle court-ruling exits. The 19th-cycle 통진당 해산 episode is a particularly large source of confounding for any 2014-window analysis.

Third, the seniority finding is no longer publishable as a junior-heavy pipeline claim but becomes a useful null: at N=16 the data cannot reject the hypothesis that local-exec runners are seniority-balanced relative to continuers. The R16 22.9% figure should be reported as confounded, not retracted.

Step 6: Status of remaining R16-R17 priorities

Data limitations still binding

1. Hand-coding does not scale beyond N=35: I coded 35 members in roughly 2 hours using public records. Extending to all SMD mid-term resigners (hundreds across 17th-22nd Assemblies, per the Yeouido Agora demand) requires NEC linkage.
2. Court-ruling timing is irregular: 의원직 상실 dates need to be coded from 대법원 판결 records, not from last-bill timing. The current cohort-construction filter is too coarse.
3. The clean N=16 is underpowered for subgroup tests: cycle-level breakdowns (4-3-5-4) cannot support the cycle-cycle stability check Critic asked for in R14.
4. Cabinet appointees as a separate behavioral channel: their early-window rate (1.30) sits between the pool (1.40) and court-rulings (lower), suggesting they drift down before appointment. A separate paper on cabinet-appointment exits would be cleaner than mixing them in.

What Critic should evaluate for theoretical framing

Three theoretical pivots are now required.

First, the paper's headline finding flips: Korean legislators on the local-executive pipeline shirk in the final six months, consistent with Mayhew-Besley-Case, not the anti-shirk pattern Hansen-Treul (2015) found in European parliaments. This is a textbook accountability story but with a Korean institutional wrinkle: shirking shows up only after exit-channel disambiguation.

Second, the Volden-Wiseman inversion claim is dead, but a different scope condition emerges: the theoretical prediction that "high-effectiveness legislators are most recruitable for higher office" is not testable in Korean data without exit-channel coding, because the apparent counter-evidence (junior-heavy cohort) is largely an artifact of court-ruling exits.

Third, the methodological contribution is the paper's strongest novel claim. No paper in either Korean or English literature, as far as I have seen, hand-codes exit channels before estimating progressive-ambition models in the KNA context. The four-channel disambiguation (local-exec, cabinet, Blue House, court ruling) is a transferable contribution that future scholars can use without re-doing the hand-coding.

Critic should evaluate: (a) is the clean-cohort shirking finding (DiD roughly -1.5 bills/month, p=0.004) sufficiently large and identified to carry a paper; (b) does the methodological contribution warrant a separate methods note rather than being folded into a substantive paper; (c) what is the right placebo for the new shirking pattern, given that the contaminated cohort plausibly shows a different pattern for theoretically meaningful reasons rather than as a placebo benchmark.

Code

All cohort construction, hand-coding, placebo, and re-estimation at /tmp/r14_cohort.py, /tmp/treated_named.csv, /tmp/treated_clean.csv. Hand-coded exit_pathway dictionary in this round's analysis script. Public-record sources for the hand coding: 중앙선거관리위원회 후보자 명부 (광역단체장 2010, 2014, 2018, 2022), 대한민국 정부 인사발령 공식 기록 (cabinet appointments), 대법원 판결문 검색 (court rulings).

References

Besley, Timothy, and Anne Case. 1995. "Does Electoral Accountability Affect Economic Policy Choices? Evidence from Gubernatorial Term Limits." Quarterly Journal of Economics 110 (3): 769-798. doi:10.2307/2946699

Hansen, Michael E., and Sarah A. Treul. 2015. "Aiming Higher: The Consequences of Progressive Ambition among MPs in European Parliaments." European Political Science Review 7 (3): 373-395. doi:10.1017/s1755773915000260

Kim, Jaehoon, and Dohyung Kim. 2020. "공천제도와 입법행위 (Candidate Selection Systems and Legislative Incentive)." SSRN Electronic Journal. doi:10.2139/ssrn.3564175

Kim, Tae-hyoung. 2025. "Why did the two major Korean parties converge in their candidate selection methods? Focusing on the 2014-2022 local elections." Korean Journal of Political Science 33 (3). doi:10.34221/kjps.2025.33.3.8

Mayhew, David R. 1974. Congress: The Electoral Connection. New Haven: Yale University Press.

Volden, Craig, and Alan E. Wiseman. 2024. "Legislative Effectiveness in the American States." American Political Science Review. doi:10.1017/s0003055424000042

Yoon, Wang Hee. 2022. "A Study on Candidate Selection Methods in the 8th Korean Nationwide Local Election: Clash of Multi-layered Values and Cartel-type Nominations." Journal of Contemporary Politics 15 (2): 5-36. doi:10.52594/jcp.2022.08.15.2.5