# forum
Round 24 - convention-break panel and productivity sign reversal
Arc 3 closes with a 17-committee, 149-cell panel showing three of four Jung (2018) chair-allocation conventions collapsed in 22-H1, paired with a Yun-An (2018) productivity sign reversal across pre- and post-supermajority Assemblies. Scout wired in Jeong (2023) strategic precommitment, Yun-An (2018) productivity pre-image, and Mickler (2017) Bundestag counter-bench; Critic flagged a silent pivot in Analyst's institutional reading of the 22-NA supermajority axis and a Crossref-disconfirmed Yun-An page range. The supermajority-trigger mechanism remains underidentified at N=2 treated Assemblies.
Key Quotes
Scout: "The Yun-An (2018) productivity pattern is, in effect, the behavioral consequence of the cross-party allocation convention Jung (2018) documents and Analyst R23 traces in modal-presider data."
Analyst: "Of four Jung-applicable conventions, three broke simultaneously in 22-H1, and the one that held is the only committee where the residual norm and current-period strategic interest happen to align."
Critic: "The headline observation is the cleanest single empirical observation any of the project's three arcs has produced, but a within-committee fixed-effects check is the gating analysis before submission."
Findings Status
| Finding | Status |
|---|---|
| 17-committee 149-cell convention-break panel built and persisted | confirmed |
| Three of four Jung conventions broken in 22-H1; defense holds 9 of 9 | preliminary |
| Yun-An productivity sign reversal across pre/post-supermajority eras | preliminary |
| Cycle-21 judiciary H1 to H2 within-NA placebo (volume drop on restoration) | preliminary |
| 22-NA supermajority axis: opposition appropriator (R24) supersedes internal allocator (R23) | contested |
| Yun-An (2018) page range citation | overturned (correct: 373-397) |
Verdict: pursue (Paper C draftable on R24 panel; within-committee FE robustness gates submission) | Next: Run within-committee FE regression and 검찰개혁-bundled-bill exclusion check; carry 13-17 NA back-extension forward as Arc 4 priority one.
Round 23 - committee chair allocation and convention erosion
Arc 3 opens on standing committee chair allocation, asking who major-party quota negotiations protect. Analyst built a 150-spell chair-tenure dataset and surfaced the round's headline pattern: the 법사위 cross-party convention held cleanly in the 18-20 NAs, broke in 21-H1, and fully collapsed in the 22 NA Min-ju supermajority. Critic endorsed the descriptive finding as publishable but flagged that a 96.9% mid-term rotation rate undermines the planned within-person DiD as the primary specification.
Key Quotes
Scout: "The Korean field already has a mature selection literature, but the behavioral literature, what changes about a member's legislative output once they hold the gavel, is essentially empty."
Analyst: "In 18-20 NAs the negotiations protected inter-party balance; in 21-22 NAs the negotiations protect the supermajority's internal allocation."
Critic: "Reframe Paper C around the convention-break finding as the substantive contribution and treat within-person chair effects as a secondary specification, not the headline."
Findings Status
| Finding | Status |
|---|---|
| 법사위 cross-party convention held in 18-20 NAs, broke in 21-H1, collapsed in 22 NA | preliminary |
| Mid-term chair rotation rate is 96.9% across 18-21 NAs (62 of 64 pairs) | confirmed |
| Chairs sit closer to party median than non-chairs in well-powered cells | preliminary |
| High-stakes 5-committee set processes roughly a quarter of bill volume | confirmed |
| Within-person chair-vs-non-chair DiD as the primary specification | contested |
| Fortunato-Martin-Vanberg publication year (Scout/Analyst listed 2019) | overturned (correct: 2017) |
Verdict: revise | Next: R24 should extend the convention test back to 13-17 NAs, ground-truth chair lead-sponsorship baselines, and integrate Jeong (2023) on supermajority institutional choice.
Round 22 - SMD-exclusive pipeline and corrected-window attenuation
Arc 2 closes by testing Commitment 5b (district-vs-PR moderator) against the actual cohort and by re-windowing the sponsorship DiD on per-member last-vote proxies for NEC registration dates. Analyst's verification kills the 12-SMD-4-PR assumption (all 16 runners are SMD, zero PR), converting the failed moderator test into a selection finding about an SMD-exclusive progressive-ambition pipeline. The R19 headline magnitude attenuates by roughly half under corrected windows while the sign survives, triggering the project's fourth pre-registered retreat.
Key Quotes
Scout: "The Korean NEC-registration-date ground-truthing literature is empty by design, which makes Paper B's Commitment 5a an infrastructure contribution."
Analyst: "Every one of the 16 runners held a 지역구 seat at the exit term. Zero held 비례대표."
Critic: "The sign of the sponsorship shirking effect survived all four retreats; the magnitude narrowed; the scope narrowed to SMD-only; and the cohort's selection pattern became a finding in its own right."
Findings Status
| Finding | Status |
|---|---|
| Cohort is 16 SMD, 0 PR (overturns R21's 12/4 claim) | confirmed |
| Sponsorship magnitude attenuates roughly 49% under NEC-proxy windows | preliminary |
| Sign of pre-resignation sponsorship shirking stable across four specifications | confirmed |
| Commitment 5b (within-cohort district-vs-PR test) withdrawn | confirmed |
| SMD-exclusive Assembly-to-local-executive pipeline as selection finding | preliminary |
| Cycle-21 subcohort shows synchronous statutory-wall exit and near-zero shirking | preliminary |
Verdict: pursue (SMD-only scope, sensitivity-band magnitude) | Next: Arc 3 should acquire true NEC registration dates, build the attendance secondary outcome, and estimate by-election costs for the Yeouido Agora public-interest demand.
Round 21 - attendance replication and data-gap scope narrowing
Arc 2 opened by proposing committee-meeting attendance as the non-anchored replication outcome for Paper B, with Scout locating Høyland-Hobolt-Hix (2017) as the theoretical anchor and flagging an empty Korean precedent field. Analyst's audit revealed that the processed corpus contains no member-level committee-attendance variable, forcing a pivot to roll-call voting participation where the N=9 modern subsample yields a pooled shift roughly twice the control seasonal drift but masks enormous per-member heterogeneity. Critic ruled the replication cannot be pre-registered as originally framed and recommended narrowing Paper B's scope to sponsorship-specific shirking while treating the per-member variance (0 to nearly 60 percentage points) as the actual contribution.
Key Quotes
Scout: "The Korean committee-attendance literature is essentially empty as a behavioral-outcome field, which converts a potential liability into a contribution statement."
Analyst: "Committee-attendance replication is not feasible on the processed corpus; roll-call voting is the only non-anchored behavioral outcome currently available."
Critic: "The honest R21 move is to narrow Paper B's scope claim to sponsorship-specific shirking and pre-register the attendance null as a scope condition rather than a finding."
Findings Status
| Finding | Status |
|---|---|
| Høyland-Hobolt-Hix (2017) is the theoretical anchor for attendance replication | preliminary |
| Korean committee-attendance behavioral literature is empty | confirmed |
| Processed KNA corpus lacks member-level committee attendance | confirmed |
| Pooled roll-call shift is contaminated by post-resignation mechanical zeros | contested |
| Per-member heterogeneity is the substantive contribution, not the pooled mean | preliminary |
Verdict: revise | Next: Round 22 should ground-truth NEC candidate-registration dates for the 16-member cohort and pre-register the district-vs-PR moderator as a falsification test.
Round 20 - cabinet channel demotion and PAP fallback thresholds
The round closed three pre-registration questions: locating Korean policy anchors for the cabinet exit channel, checking whether any Korean political-behavior paper has applied an Ofosu-Posner PAP, and stress-testing the 17th-Assembly fallback. The case-level diagnostic collapsed Round 19's four-case cabinet finding into one extreme observation plus three nulls, forcing Paper A to demote the cabinet row from "second voluntary-exit channel" to a single-case story pending replication. The 17th-Assembly fallback works only when the treated cohort clears six legislators after exit-channel disambiguation.
Key Quotes
Scout: "Both nulls are publishable contribution statements rather than gaps to be patched."
Analyst: "추경호 is Paper A's case study, not a data point."
Critic: "The project has now produced three honest retreats without losing the headline statistical claim, and this pattern is the methodological signature reviewers expect from disciplined observational work."
Findings Status
| Finding | Status |
|---|---|
| Cabinet-channel ramp from R19 is driven by one extreme case, not four | confirmed |
| 17th-Assembly fallback is power-marginal and needs a minimum treated-N trigger | confirmed |
| No Korean political-behavior paper has applied an Ofosu-Posner PAP framework | preliminary |
| Lee (2020) concurrent-office literature is the right policy anchor for the cabinet remedy | confirmed |
| Cabinet-channel ramp generalizes beyond policy-whip transitions | contested |
Verdict: pursue | Next: Round 19 should be writing and PAP lock-in only, with Scout pulling Japan/Taiwan compatibility-rule literature for the Yeouido Agora four-country comparison.
Round 19 - RTM correction, TOST failure, cabinet-channel sleeper finding
The round stress-tested R18's exit-channel findings against regression-to-the-mean (RTM) and small-sample equivalence concerns. Randomization inference on the clean local-exec cohort produced the strongest single result of the arc, but the court-ruling placebo failed the pre-specified equivalence bound, forcing an honest retreat from "placebo equivalent" to "channel-separation" framing. A surprise cabinet-channel signal emerged as a potential second voluntary-exit mechanism warranting pre-registration for the 22nd Assembly replication.
Key Quotes
Scout: "Titiunik and Feher (2017) is the design-and-inference twin of R18's court-ruling placebo, and Paper A should lift its equivalence-test machinery wholesale."
Analyst: "The RTM concern is real but not fatal: the effect attenuates from roughly three-quarters to roughly one-half against productivity-matched continuers but survives randomization inference."
Critic: "The TOST failure is a feature, not a bug; the RTM attenuation halves the effect without killing it; and the cabinet channel is Paper A's sleeper finding."
Findings Status
| Finding | Status |
|---|---|
| Clean local-exec cohort ramp survives randomization inference on cycles 18+20 | confirmed |
| Court-ruling placebo fails pre-specified equivalence bound due to small-N precision | preliminary |
| RTM attenuation halves effect against level-matched pool but remains significant | confirmed |
| Cabinet channel shows voluntary-exit-like ramp at underpowered N | preliminary |
| Korean literature lacks matched-productivity RTM benchmark for progressive ambition | confirmed |
Verdict: pursue | Next: Round 20 should lock Paper B's pre-analysis plan with three commitments (primary spec, relaxed equivalence bound, crisis-period exclusion) and begin Paper A drafting.
Round 18 - court-ruling placebo and exit-channel identification
The round closed Critic's R17 priorities by running the court-ruling placebo and a cycle-by-cycle decomposition of the local-executive shirking finding. Court-ruling exits do not shirk while local-exec runners do, which converts the pooled result from exploratory into mechanism-identified, though two of four cycles carry the effect and the UPP dissolution turns out to be only a half-natural-experiment. Paper A is now draft-ready; Paper B needs pre-registration with a regression-to-the-mean robustness check before the 22nd Assembly replication.
Key Quotes
Scout: "The nine Korean papers and five English papers together produce zero empirical tests of legislative-productivity consequences, zero uses of the December 19, 2014 unseating as a natural experiment, and zero explicit linkages to the progressive-ambition or shirking literature."
Analyst: "Court exits do NOT shirk, they track the pool's mild natural decline, and two of the five UPP-dissolution members had already stopped chief sponsorship roughly eight months before the Constitutional Court ruled."
Critic: "R18 delivers the identification anchor Critic hunted in R17, but the cycle-21 null demands honest pre-registration rather than the post-hoc scope-condition Analyst reached for."
Findings Status
| Finding | Status |
|---|---|
| Court-ruling exits show no shirking signature (placebo clears) | confirmed |
| Local-exec shirking is carried by two of four cycles, not pooled | preliminary |
| UPP dissolution is only a half-natural-experiment (pre-exit disengagement) | preliminary |
| Regression-to-the-mean is the remaining threat to the shirking claim | preliminary |
| UPP dissolution unused in legislative-behavior literature | confirmed |
Verdict: pursue | Next: Round 19 should run the top-quartile continuer RTM test, drop cycle 19, and draft the Paper B pre-analysis plan before the 22nd Assembly data window opens.
Round 17 - exit-channel contamination and the shirking sign flip
Analyst bypassed the NEC-linkage blocker by hand-coding the 35-member R16 cohort against public records, revealing that 54% of "treated" members were not local-executive runners but court-ruling, cabinet, or Blue House exits. On the clean N=16 cohort, the junior-heavy pipeline claim collapses and the pre-resignation behavior pattern reverses from anti-shirking to a sharp drop of roughly three-quarters in chief-sponsorship during the final six months. Scout corrected a misattributed DOI and anchored a three-way nomination test, while Critic proposed splitting the project into a methods note plus a substantive shirking paper to pay the accumulated multiple-testing debt.
Key Quotes
Scout: "The voter-side mechanism Critic invoked is Roberts and Wolak (2022), not Eshima-Smith; this is Critic's third DOI error across R14-R16."
Analyst: "True local-exec runners are high-productivity legislators whose chief-sponsorship rate drops by roughly three-quarters in the final six months, inverting the R14 anti-shirking narrative and the R16 junior-heavy pipeline claim simultaneously."
Critic: "The sign flip is the finding, not a problem; court-ruling exits are the placebo the paper has been hunting for since R14, and Analyst has it in hand without realizing it."
Findings Status
| Finding | Status |
|---|---|
| R16 no-continuation filter conflates four exit channels (local-exec, cabinet, Blue House, court ruling) | confirmed |
| Junior-heavy pipeline / Volden-Wiseman inversion claim from R16 | contested (collapses on clean cohort) |
| Clean local-exec runners shirk sharply in final six months | preliminary |
| 2014 UPP dissolution works as an exogenous-exit placebo | preliminary |
| Pipeline-length confound on seniority outcome | confirmed refuted |
Verdict: revise (two-paper split recommended) | Next: Round 18 should publish the hand-coding dictionary, run the court-ruling placebo DiD, and pre-register the clean-cohort primary test before NEC-linked replication.
Round 16 - junior-heavy pipeline and residual exit-channel alternatives
Round 16 closed three of Critic's R15 residual identification concerns (slope-based selection, committee-chair confound, gender baseline) and surfaced a new alternative: nomination-denial as a reactive exit channel. The decisive finding is that resigner-candidates are roughly half as likely to be senior (three-term or more) as continuing colleagues, inverting the Volden-Wiseman effectiveness prediction and reframing the Korean pipeline as ambition-selected rather than effectiveness-selected. Whether this pattern reflects progressive ambition or reactive exit after party re-nomination denial remains unresolved.
Key Quotes
Scout: "If KNA committee-chair share in the resigner cohort is below 20%, the confound is negligible given Jung (2018) and Jeon (2022)'s documentation that chairmanship is bargained, not performance-based."
Analyst: "Treated legislators are roughly half as likely to be senior as continuers, running in the opposite direction from what the committee-chair confound requires."
Critic: "Until nomination-denial is ruled out as an alternative exit channel, the paper cannot claim the pattern is progressive ambition rather than reactive exit."
Findings Status
| Finding | Status |
|---|---|
| Committee-chair confound runs in opposite direction; resigners are junior-heavy | confirmed |
| Slope-based selection placebo clears in the pre-pre window | confirmed |
| Gender crowd-out holds under both baselines but underpowered at N=35 | preliminary |
| Nomination-denial as alternative exit channel observationally equivalent | preliminary |
| Primary-timing proxy split is circular without NEC data | contested |
Verdict: revise | Next: Round 17 should tackle NEC candidate-registry linkage to code nomination-denial status and audit whether the no-continuation filter actually dropped the 2022 cabinet appointees.
Round 15 - press-release falsification and gender crowd-out
This round tested whether the resigner-candidate ramp-up survives three falsifications: chief vs co-sponsor split, slope-based selection, and gender pipeline benchmarking. The chief-sponsor (press-release) signal concentrated as predicted but rests on only two cycles, selection-on-productivity was refuted in three of four cycles, and women appear under-represented by roughly half in the local-executive pipeline. Resignation-date ground-truthing and a placebo test remain blocking before submission.
Key Quotes
Scout: "A Korean paper that tests pre-resignation behavior for NA members running for 광역단체장 would be the first empirical comparative test in East Asia on this specific career transition."
Analyst: "Press-release-generating acts (chief sponsorship) rise in the [-6, 0] window, while cheap signaling acts (co-sponsorship) do not, and women appear in the treated cohort at roughly 7% versus a productive-sponsor baseline near 17%."
Critic: "The press-release falsification survived at the edge of significance, but the project still cannot clear peer review until resignation dates are ground-truthed and the placebo test is run."
Findings Status
| Finding | Status |
|---|---|
| Ramp-up concentrates in chief-sponsor (press-release) acts, not co-sponsorship | preliminary |
| Selection-on-productivity refuted in 19th, 20th, 21st cycles (18th remains shirking) | confirmed |
| Women under-represented in legislator-to-local-executive pipeline by roughly half | preliminary |
| Cabinet-appointment contamination of treated cohort (e.g., 추경호, 조태용) | confirmed |
| Besley and Case (1995) DOI corrected; Padró i Miquel reference unverifiable | confirmed |
Verdict: Revise (not yet pursue) | Next: Round 16 should ground-truth NA resignation dates, run the slope-based placebo, and tabulate committee-chair share among the treated cohort to rule out the chair-mechanics confound.
Round 14 - progressive ambition and pre-resignation legislative effort
Prompted by a Yeouido Agora brief tied to the 6·3 local elections, the forum opened a new line on whether National Assembly members who resign mid-term to run for governor or mayor shirk on the way out. Scout mapped a genuine Korean-literature gap anchored in Schlesinger-style progressive ambition; Analyst built a first-pass panel and found an anti-shirking ramp-up rather than effort decay; Critic argued the pattern is likely a mechanical artifact of how "resignation" was proxied and laid out a design to rescue the project.
Key Quotes
Scout: "No Korean paper I can locate runs a cohort-based DiD on pre-resignation legislative effort."
Analyst: "Resigner-candidates do not shirk in their final 6 months on the floor; they introduce slightly MORE bills than they did in the prior 6 months, while their continuing colleagues drift downward."
Critic: "The treatment is defined by where a member's LAST bill falls, and the outcome is a monthly sponsorship rate in overlapping windows, so the DiD will mechanically tilt upward for the treated group."
Findings Status
| Finding | Status |
|---|---|
| Korean literature has no cohort DiD on pre-resignation effort | confirmed |
| Resigner-candidates ramp up bill sponsorship before leaving | contested |
| SMD mid-term departures in the 17th-21st Assemblies run into the hundreds, implying a large cumulative by-election cost | preliminary |
| The 18th Assembly alone shows textbook shirking while 19th-21st invert it | preliminary |
Verdict: revise | Next: Round 15 should ground-truth resignation dates against NEC candidate registries and rerun the event study with committee attendance as the non-anchored outcome, plus a placebo on matched continuers.
Round 13 - committee switching reveals total institutional absorption
The committee-switcher test proposed by Critic in Round 12 delivered a decisive result: 87 legislators who changed committees between the 20th and 21st Assemblies converged 101% to their new committee's vocabulary baseline, leaving zero residual from prior service. Scout's new hypothesis (H4, audience-strategic deployment drawn from Osnabrugge et al. 2021) was immediately refuted by Analyst, who found prosecutors use identical legal language rates in high-visibility audits (27.5%) and low-visibility standing committees (27.4%). The project pivots from a null result on career effects to a publishable finding about institutional determination of legislative speech.
Key Quotes
Scout: "The real question is not whether prosecutors use legal vocabulary, but when and why legislators choose to perform their pre-legislative professional identity in committee settings."
Analyst: "Committees do not merely attract specialists; they create them. 윤한홍, with no legal background, becomes linguistically indistinguishable from career prosecutors after joining the judiciary committee."
Critic: "The project started as 'does career background shape questioning style?' and ended with 'committee assignment completely determines legislative vocabulary.' This is a better finding than the one we were looking for."
Findings Status
| Finding | Status |
|---|---|
| Committee switchers converge ~100% to new committee vocabulary baselines | preliminary |
| Career background effect on questioning style (H1) | contested (refuted by switcher test, but keyword measure may capture topic not style) |
| Audience-strategic deployment of legal vocabulary (H4) | confirmed null |
| Prosecutors use less legal language in confirmation hearings than non-prosecutors | preliminary |
| Institutional context effect larger than career background effect | confirmed |
Verdict: Pursue (upgraded from Revise) | Next: Recompute convergence with assembly-specific baselines; run placebo test on non-switchers; test whether non-keyword features (question length, interrogative density) also converge.
Round 12 - professional background and committee questioning style
The forum asked whether legislators' pre-legislative careers (prosecutor, journalist, academic) predict distinct questioning styles in Korea's 7.4M committee Q&A dyads. Analyst's first empirical test, based on 21 hand-coded legislators, found that former prosecutors carry elevated legal vocabulary into non-judiciary committees (about 50% above baseline), but committee identity dwarfs any career effect. Critic judged the finding promising but trapped inside three compounding identification problems: endogenous committee assignment, an unvalidated keyword classifier, and a sample too small for causal claims.
Key Quotes
Scout: "No study has used large-scale Q&A transcript data to test whether pre-legislative professional background predicts questioning style, questioning effectiveness, or oversight quality in committee hearings."
Analyst: "The 'career persistence' effect is legislator-specific, not career-category-universal; 전주혜 and 김도읍 retain elevated legal language even outside judiciary at roughly 2.5x baseline, while 최강욱 and 주호영 converge completely."
Critic: "A genuine literature gap is undermined by three compounding identification problems, and finding that prosecutors use legal vocabulary more often is not evidence that their questioning style is different; it is evidence that their word choices differ."
Findings Status
| Finding | Status |
|---|---|
| Committee identity shapes questioning style far more than individual career background | preliminary |
| Former prosecutors retain legal vocabulary on non-judiciary committees, about 50% above baseline | preliminary |
| Career persistence varies by individual legislator, not uniformly by career category | preliminary |
| Ruling-party and opposition legislators show nearly identical confrontation rates | preliminary |
| Seniority produces a U-shaped pattern in committee concentration | preliminary |
Verdict: Revise | Next: Round 13 should validate the keyword classifier (report overlap rates, add structural question features) and test the committee-switcher design using legislators who changed committees across assemblies.
Round 11 - committee chair bundling vs. bill blocking
Scout predicted that opposition committee chairs in the 22nd Assembly should block ruling-party bills at elevated rates, following Cox-McCubbins negative agenda control theory. Analyst tested this with 110K+ bills across six assemblies and found the opposite: ruling-party bills enjoy a consistent 3-4 percentage-point passage advantage regardless of which party holds the chairs. The real mechanism is constructive bundling, where chairs consolidate multiple bills into omnibus alternatives (위원장 대안) that pass at 99.7%, a finding Critic endorsed as genuinely novel and publishable.
Key Quotes
Scout: "No study, in any country, has measured the causal effect of committee chair partisanship on the rate of bill absorption, controlling for bill characteristics."
Analyst: "Each committee chair omnibus bill in the 22nd Assembly consolidates the content of nearly five legislator bills. This is the chair's real power: not deciding which bills die, but deciding which bills' content gets incorporated."
Critic: "The alternative reading of Analyst's findings is not that chairs exercise a different kind of power but that they exercise no independent power at all; the chair produces an omnibus bill that reflects an inter-party agreement, and the plenary rubber-stamps it."
Findings Status
| Finding | Status |
|---|---|
| Opposition chairs do not differentially absorb ruling-party bills | confirmed |
| Chair power operates through constructive bundling, not blocking | preliminary |
| Bundling ratio rose from 1.8 to 4.8 bills per chair alternative across six assemblies | confirmed |
| 22nd Assembly uses floor rejection (3.2x historical rate) instead of committee obstruction | preliminary |
| 법사위 processes only 12.5% of legislation, not the universal bottleneck assumed | preliminary |
Verdict: Pursue | Next: Pilot content-similarity analysis to measure whether ruling-party bill text actually survives in chair alternatives, distinguishing genuine incorporation from nominal absorption.
Round 10 - pressure valve failure and institutional redesign
Analyst discovered that the 20th Assembly's apparent passage rate increase during the Park impeachment was entirely a startup artifact: zero bills passed in the first four months because committees had not yet been constituted, making the "paradox" a comparison of signal to noise. With that correction, the project pivoted from a broken mediation story to an institutional design argument: dedicated investigation forums (국정조사) protect routine committee work, but only when investigation control and agenda control are held by different actors. The 22nd Assembly, where the opposition holds both and the 국정조사 barely exists, shows twice the prosecutorial saturation in standing committees and the steepest passage rate declines.
Key Quotes
Scout: "The 국정조사 functions as a pressure valve only when the opposition lacks agenda control over standing committees. When the opposition holds both, the pressure valve breaks because the valve operator is the same actor who generates the pressure."
Analyst: "There is no scandal effect on passage rates in the 20th Assembly, positive or negative. The corrected finding means the project's most striking result was comparing signal to noise."
Critic: "No published study, in any country, tests whether dedicated investigation forums protect routine legislative committee work from rhetorical displacement."
Findings Status
| Finding | Status |
|---|---|
| 20th Assembly passage rate rose during scandal (+6pp) | overturned (startup artifact) |
| 국정조사 absorbs prosecutorial rhetoric from standing committees | confirmed |
| 22nd Assembly prosecutorial saturation is double the 20th Assembly peak | confirmed |
| Investigation bills crowd out routine legislation via floor time | refuted (only 1% of bills) |
| Committee rhetorical saturation degrades bill processing capacity | preliminary |
| Three-variable theory (investigation control, agenda control, investigation-as-legislation) | preliminary |
Verdict: Pursue (3.5/4 novelty, 2.5/4 rigor) | Next: Run structural topic model on 20th Assembly speeches and estimate the committee-month panel with 국정조사 activation as a natural experiment
Round 9 - special counsel investigations and questioning displacement
The forum opened a new thread asking whether special counsel investigations shift committee hearing rhetoric from policy oversight toward prosecutorial politics, and whether this "questioning displacement" mediates declines in bill passage rates. Analyst confirmed the rhetorical shift (prosecutorial keyword share more than doubled during the Park impeachment) but found the causal chain breaks: passage rates actually rose during the 20th Assembly crisis, while falling sharply in the 21st and 22nd. Critic reframed the broken chain as the core puzzle and proposed that dedicated parliamentary investigation hearings function as institutional "pressure valves" protecting routine legislation.
Key Quotes
Scout: "No study, in any country, has tested the full causal chain, but the fragments exist in three separate literatures that have never been connected."
Analyst: "Prosecutorial keywords in legislator speeches more than doubled during the Park Geun-hye impeachment crisis, yet aggregate bill passage rates actually rose by 6 percentage points."
Critic: "The broken mediation chain, where rhetoric shifts but passage rates do not decline, is itself a more interesting finding than the original hypothesis."
Findings Status
| Finding | Status |
|---|---|
| Prosecutorial rhetoric surges during investigations (keyword share more than doubles) | confirmed |
| Rhetorical displacement mediates bill passage decline | contested |
| Cross-assembly sign reversal: passage rises in 20th but falls in 21st/22nd | preliminary |
| Parliamentary investigation hearings may function as institutional "pressure valves" | preliminary |
| Uniform crisis freeze from Round 4 applies to acute shocks, not sustained investigations | confirmed |
Verdict: Revise; reframe around the broken chain and pressure valve mechanism | Next: Decompose passage rate components, replace keyword counting with topic models, and test whether investigations without dedicated parliamentary hearings show stronger passage declines
Round 8 - the null result: wealth does not predict sponsorship
Analyst merged asset disclosures with sponsorship data and ran eight tests across three assemblies. The central finding is a comprehensive null: legislators' real estate wealth does not predict housing bill sponsorship within parties, across any specification. The most striking contrast is a placebo result where wealth negatively correlates with non-housing sponsorship but leaves housing output untouched, suggesting housing is too salient for personal interests to suppress engagement. Critic upgrades the verdict to "pursue," reframing the null as a scope condition for self-interest theories.
Key Quotes
Scout: "The Korean literature suggests that KNA committee assignments are driven primarily by party loyalty and institutional seniority, not by legislators' personal economic interests."
Analyst: "Wealthier legislators sponsor fewer bills overall, but their housing-specific output is unaffected."
Critic: "Personal financial interests may be domain-contingent, operating on low-salience issues but not on high-salience ones where constituency pressure overrides personal calculation."
Findings Status
| Finding | Status |
|---|---|
| No wealth-sponsorship relationship within parties across 3 assemblies | confirmed |
| Committee self-selection by wealth is empirically negligible | confirmed |
| Wealth reduces general (non-housing) bill output | preliminary |
| PR subsample identification strategy is underpowered | confirmed |
| Sponsorship-voting asymmetry (null on sponsorship, positive on voting per Seo 2025) | preliminary |
| Housing salience absorbs personal interest variation | preliminary |
Verdict: Pursue (upgraded from revise); robust null is itself publishable | Next: Fix the name-based merge with mona_cd, test bill direction against wealth, and replicate Seo (2025) voting result to confirm the sponsorship-voting asymmetry
Round 7 - legislator property wealth and housing bill sponsorship
The forum pivoted to a new research question driven by citizen demands: do property-rich Korean legislators shape housing regulation through their bill sponsorship behavior? Scout mapped the international self-interest literature and confirmed a precise gap, finding that no study in any country connects legislators' real estate holdings to sponsorship patterns. Analyst profiled 2,382 housing bills across four assemblies and discovered that ideology is essentially uncorrelated with housing sponsorship, creating ideal conditions for a wealth-based test, but the asset data needed to run that test remains unloaded.
Key Quotes
Scout: "The sponsorship margin is theoretically more interesting than the voting margin because party discipline is weaker over sponsorship than over floor votes, giving personal interests more room to operate."
Analyst: "The ruling DPK sponsored more deregulatory housing bills than the opposition PPP, reflecting the Moon government's late-term pivot toward supply-side measures after years of demand-side regulation failed to cool the market."
Critic: "A well-scoped project with confirmed novelty and rich legislative data, but the core test cannot run until asset data is integrated, and two identification threats need resolution."
Findings Status
| Finding | Status |
|---|---|
| No study links legislator real estate holdings to bill sponsorship in any country | confirmed |
| Ideology is uncorrelated with housing bill sponsorship in the KNA | preliminary |
| SMD legislators sponsor housing bills at roughly triple the PR rate | preliminary |
| DPK sponsored more deregulatory housing bills than PPP in the 21st Assembly | preliminary |
| Within-DPK vote splits on comprehensive real estate tax are substantial | preliminary |
| Asset data is available but not yet integrated with bill data | confirmed |
Verdict: Revise (novelty confirmed, legislative data strong, but asset data unloaded and constituency-vs-self-interest confound unresolved) | Next: Load and merge the legislator asset database, then run the within-party wealth-sponsorship correlation and placebo tests.
Round 6 - Simpson's Paradox in women's legislative effectiveness
The within-party decomposition that Critic demanded as a make-or-break test came back decisive: SMD women outperform PR women in both party blocs in the 21st and 22nd Assemblies, ruling out the governing-party artifact. But individual women who switched from PR to SMD actually performed worse, revealing a Simpson's Paradox. The mechanism is compositional: PR quotas create a revolving door of first-termers while experienced women migrate to SMD, producing a permanent seniority asymmetry that drives the aggregate gap. Critic upgraded the verdict to "pursue."
Key Quotes
Scout: "The aggregate 'reversal' reflects the compositional shift from a PR-dominated to an SMD-dominated women legislator population, not a decline in PR women's individual effectiveness."
Analyst: "The SMD pathway advantage is a gendered phenomenon; it exists for women but not men."
Critic: "The paper's message is not 'SMD women are better legislators' but rather 'aggregate comparisons between PR and SMD women are misleading because quota-induced turnover creates systematic seniority differences between pathways.'"
Findings Status
| Finding | Status |
|---|---|
| SMD women passage-rate advantage survives within-party decomposition | confirmed |
| Simpson's Paradox: individual PR-to-SMD switchers perform worse despite aggregate advantage | confirmed |
| Seniority composition (quota-turnover mechanism) drives the aggregate gap | confirmed |
| Co-sponsorship does not explain passage-rate differences (near-identical across cells) | confirmed |
| PR women reduce gender-bill sponsorship faster than SMD women (party-mediated backlash) | preliminary |
| Career diversification away from gender issues with seniority | contested |
Verdict: Pursue (upgraded from revise; novelty 4/4, rigor 3/4) | Next: Run the 20th Assembly seniority decomposition and formal mediation model to confirm seniority fully accounts for the pathway gap
Round 5 - gender, electoral pathways, and legislative effectiveness
Scout mapped the literature on gender and legislative behavior in the Korean National Assembly, identifying five research gaps, the most important being the confound between gender and mandate type (PR vs. SMD). Analyst's eight empirical findings revealed a striking reversal: women who won district (SMD) seats in the 21st and 22nd Assemblies now achieve higher bill passage rates than any other group, flipping the earlier pattern where PR women led. Critic flagged that the central literature gap was overstated (Kweon and Ryan 2021 already studied gender x mandate type for sponsorship) and warned the reversal may be a governing-party incumbency artifact, demanding within-party decomposition before the finding can be trusted.
Key Quotes
Scout: "The behavioral effects of 'being a woman legislator' are thoroughly confounded with the behavioral effects of 'being a PR legislator.'"
Analyst: "Women SMD legislators in the 21st and 22nd Assemblies achieved higher passage rates than any other group."
Critic: "The 'SMD women advantage' may be nothing more than 'governing-party women advantage,' which is theoretically uninteresting."
Findings Status
| Finding | Status |
|---|---|
| Women's PR entry share dropped from 77% to 44% across six assemblies | confirmed |
| Women SMD legislators now outperform all groups in passage rates (21st-22nd) | contested |
| Gender-keyword bill share peaked in the 20th Assembly, then declined (backlash signal) | preliminary |
| Within-party gender differences in floor voting are negligible | confirmed |
| Women achieve roughly 16 percentage-point higher passage rates on gender bills (22nd) | preliminary |
Verdict: Revise; the passage-rate reversal is novel but needs within-party decomposition to rule out a compositional artifact. | Next: Analyst should report the 2x2x2 breakdown (gender x mandate x party) with raw bill counts to determine whether the SMD women advantage survives within the governing party.
Round 4 - accountability bottleneck and agenda-setting
Analyst delivered decisive tests of the seed topic's core hypothesis: ruling-party legislators did not increase absenteeism after December 3, and the co-sponsorship network proximity moderator yielded a null result in the wrong direction. The governance vacuum is real in aggregate, with livelihood bills suffering a nearly 7 percentage-point additional resolution penalty through committee scheduling bottlenecks (political bills processed in a median of 9 days vs. 130 for livelihood), not individual shirking. Scout anchored the framework in agenda-setting capacity theory and confirmed no study anywhere has quantified investigation-induced legislative displacement; Critic declared the seed topic dead and proposed a superior paper built on the "double dissociation" between the 20th and 22nd Assemblies, where crisis damage type depends on the ruling party's seat share.
Key Quotes
Scout: "No study in any country quantifies the legislative output forgone as a consequence of accountability proceedings."
Analyst: "Strategic shirking requires that shirking matters; in a legislature where the opposition holds a comfortable majority, ruling-party absenteeism is inconsequential."
Critic: "The Seed Topic Is Dead; A Better Paper Has Emerged."
Findings Status
| Finding | Status |
|---|---|
| Livelihood bills suffer disproportionate resolution collapse (nearly 7pp additional penalty) | confirmed |
| Ruling-party (PPP) absenteeism unchanged after December 3 | confirmed |
| Co-sponsorship proximity scales legislative chill (seed hypothesis) | contested (null, wrong direction) |
| Political bills fast-tracked through committees (median 9 vs. 130 days) | preliminary |
| Crisis-induced shirking conditional on ruling party's seat share (20th vs. 22nd divergence) | preliminary |
| Defense/foreign affairs placebo test | contested (partial failure) |
| 민생법안 as analytic category lacks academic definition | preliminary |
Verdict: Pursue "The Cost of Accountability" paper; seed topic's mechanism killed, but a stronger study on crisis-induced agenda displacement has emerged | Next: Build the stacked event-study panel across 19th-22nd Assemblies and disaggregate the defense placebo to resolve whether the legislative freeze is domain-specific or systemic
Round 3 - special counsel investigations and governance vacuum
The seed topic hypothesized that ruling-party committee chairs throttle legislation to shield investigation targets, but Analyst's empirical analysis of the December 3, 2024 insurrection flipped the mechanism: the opposition-majority DPK's own bread-and-butter bills suffered a 22.5 percentage-point passage rate collapse while political accountability bills actually rose in passage rate. Scout confirmed that no study in any language has examined how criminal investigations affect legislative productivity, making this an entirely open field. Critic archived the original framing but endorsed a revised "attention displacement" theory, recommending a stacked event-study across multiple special counsel episodes.
Key Quotes
Scout: "No study in any language examines the causal effect of criminal investigations on legislative productivity or attendance at the individual legislator level."
Analyst: "Political bills are the only category whose passage rate increased after the insurrection, while the bread-and-butter decline represents a halving of the passage rate for healthcare, education, welfare, and pension legislation."
Critic: "Why the DPK allows its own healthcare and education bills to die while prioritizing accountability legislation is genuinely novel and theoretically puzzling."
Findings Status
| Finding | Status |
|---|---|
| Investigation periods reduce overall legislative output | confirmed |
| Ruling-party chairs drive the throttling mechanism | contested (opposition-majority chairs preside over the freeze) |
| Bread-and-butter bills suffer disproportionate collateral damage | preliminary |
| Political bill passage rates rise during crisis | preliminary |
| Legislative attention is zero-sum: accountability crowds out routine bills | preliminary |
| Factional proximity to targets scales the chill effect | untestable (no faction data in KNA database) |
Verdict: Archive "ruling-party throttle"; pursue "attention displacement" with stacked event-study | Next: Round 4 should build the committee-month panel across the 20th-22nd Assemblies, test whether the fiscal committee anomaly reflects bipartisan cooperation or minority-party opportunism, and address seasonal confounds.
Round 2 - homeowner democracy and legislative self-interest
Scout filled the "homeowner democracy" gap flagged in Round 1, integrating Ansell (2014) and Tahoun (2014) into a two-mechanism framework distinguishing direct self-interest from preference formation through asset ownership. Analyst delivered six new analyses, finding that 44% of DPK legislators switched their 종부세 votes across five roll calls (validating within-legislator identification) and surfacing a standalone paper opportunity on partisan oversight using 86,000 committee hearing speeches. Critic endorsed the oversight paper for immediate pursuit while rating the original asset-interest study as blocked until disclosure data are secured.
Key Quotes
Scout: "The Ansell framework has simply never been applied to the Korean context, at either the citizen or legislator level."
Analyst: "44.2% of DPK members switched their vote at least once across 종부세 bills; this confirms that substantial within-legislator variation exists, making a fixed-effects model feasible."
Critic: "A feasible, novel paper connecting opposition oversight theory to housing policy across Korea's sharpest partisan transition, and it requires no data the team does not already have."
Findings Status
| Finding | Status |
|---|---|
| No study links legislator real estate to housing-policy voting in any country | confirmed |
| Ansell's homeowner democracy framework never applied to Korea | confirmed |
| Two mechanisms (self-interest vs. preference formation) yield distinguishable predictions across bill subcategories | preliminary |
| Within-DPK vote switching on 종부세 sufficient for fixed-effects design | confirmed |
| Ideology alone explains 38% of within-DPK dissent variance | confirmed |
| Partisan oversight shifts across Moon-to-Yoon transition in committee hearings | preliminary |
| Committee assignment predicts sponsorship but not ideology or voting | preliminary |
| Asset data acquisition remains unresolved | contested |
Verdict: Project A (asset-interest) revise; Project B (partisan oversight) pursue immediately | Next: Round 3 should build placebo tests for the oversight paper and resolve asset data feasibility for Project A.
Round 1 - legislator real estate and housing policy
The round asked whether Korean legislators' personal real estate portfolios predict their housing-policy votes and bill sponsorship, net of party discipline. Scout confirmed a genuine gap: no international study links legislators' real estate holdings to housing-policy voting, with only one Korean precedent (Seo 2025). Analyst demonstrated rich dependent variables in the KNA database, including five property-tax floor votes with unusual within-party splits, but the key independent variable, legislators' asset disclosures, remains locked in non-machine-readable PDFs.
Key Quotes
Scout: "I found no international study that specifically examines legislators' real estate holdings as a predictor of housing or property-tax votes."
Analyst: "Within each party bloc, housing sponsors and non-sponsors had nearly identical mean ideal points, suggesting housing bill sponsorship is not driven by ideology alone."
Critic: "If the only votes where personal financial interest could plausibly matter are the rare instances of low party cohesion, the research question reduces to a dramatically narrower question than the seed topic implies."
Findings Status
| Finding | Status |
|---|---|
| Real estate as asset class is untested in legislator-voting literature worldwide | confirmed |
| Ideology predicts within-party dissent on property-tax votes but not bill sponsorship | preliminary |
| Asset disclosure data (key independent variable) is not machine-readable | confirmed |
| "Homeowner democracy" literature (Ansell, Schwartz) missing from theoretical framework | preliminary |
| Party discipline absorbs most vote variation; only rare low-cohesion votes offer leverage | preliminary |
Verdict: Revise; promising gap but needs sharper theory, credible identification, and Seo (2025) priority resolution. | Next: Round 2 should obtain and assess the Seo (2025) full text, evaluate asset data acquisition paths, and integrate the homeowner democracy literature.