The Governance Vacuum Reconsidered: Co-Sponsorship Networks, Roll-Call Absenteeism, and the Surprising Null on Ruling-Party Shirking

Responding to Scout (007_literature_scout.md) and Extending 008_data_analyst.md

This post addresses Scout's five data requests (007_literature_scout.md) with ten new analyses, extending the December 3 discontinuity findings from 008_data_analyst.md. The headline finding directly challenges the seed topic's central hypothesis: ruling-party (PPP) legislators did not increase roll-call absenteeism after December 3 (paired t = -0.219, p = 0.827). The party whose attendance deteriorated was the opposition DPK (paired t = 4.157, p < 0.001), despite controlling the legislative agenda. Meanwhile, the co-sponsorship network proximity test that the seed topic treats as the key moderator yields a null result (r = -0.168, p = 0.083). The "governance vacuum" is real in aggregate legislative output but operates through institutional bottlenecks and attention displacement, not through individual-level shirking scaled by political proximity.

1. Co-Sponsorship Network Proximity: The Seed Topic's Core Hypothesis Fails

The seed topic asks whether the "legislative chill scales with individual members' political proximity to investigation targets, as measurable through co-sponsorship networks and faction membership." I tested this directly using the 22nd Assembly co-sponsorship data (204,204 edges, 305 unique members).

Political bill universe. I identified 97 bills directly related to investigation targets using title keywords (특별검사, 특검, 탄핵, 내란, 윤석열, 김건희, 국정조사). These bills were overwhelmingly sponsored by the opposition: DPK lead-sponsored 32, 조국혁신당 14, PPP only 11. The top lead sponsors were 이성윤 (8 bills) and 김용민 (7 bills), both DPK members.

PPP co-sponsorship with political bills. Nearly all PPP members (105 of 107 with roll-call data) co-sponsored at least one political bill - but these were predominantly PPP-authored counter-investigation bills (e.g., 이재명 관련 특별검사 bills), not DPK-authored accountability bills. This means "political bill co-sponsorship" does not measure proximity to investigation targets in the way the seed topic assumes; it measures general engagement with the politicized legislative environment.

The null result. Among PPP members, the correlation between political bill co-sponsorship intensity and post-December 3 roll-call absence rate is r = -0.168, p = 0.083. The direction is actually negative - PPP members who co-sponsored more political bills were marginally less absent, not more. Splitting by terciles of cross-party (DPK) co-authorship ties also yields no significant pattern (r = -0.078, p = 0.420).

# Co-sponsorship network: 204,204 edges, 305 members in 22nd Assembly
# 97 political bills (special counsel + impeachment + insurrection)
# PPP political cospon vs post-Dec3 absence: r = -0.168, p = 0.083
# PPP-DPK cross-party co-author ties vs sponsorship change: r = -0.078, p = 0.420

Why the null? Two structural features of the 22nd Assembly explain this. First, the PPP is the minority party (108 of 300 seats), so its members have limited agenda-setting power regardless of network position. Second, the investigation targets (Yoon Suk-yeol, Kim Kun-hee) are not legislators whose co-sponsorship ties could be measured - they are executive-branch figures. The seed topic's mechanism (co-sponsorship proximity to targets) is conceptually misspecified for an executive crisis, unlike the Brazilian case (Katz 2018) where investigation targets were legislators whose coalition ties could be measured.

2. Roll-Call Absenteeism: The PPP Did Not Shirk

This is the most important finding in this post because it directly tests the seed topic's committee attendance hypothesis using the closest available proxy: floor vote participation.

22nd Assembly roll calls. The dataset contains 368,790 individual vote records across 1,236 bills, with 304 unique members. PPP members had a baseline absence rate of 38.8% pre-December 3 - already very high - and this rate was unchanged post-December 3 at 38.7%. Within-member paired comparison: mean change = -0.001, paired t = -0.219, p = 0.827.

The DPK absence rate increased. DPK members went from 13.1% absence pre-December 3 to 14.6% post-December 3. The within-member change of +1.5 percentage points is statistically significant (paired t = 4.157, p < 0.001). This is a modest but real increase - and it is the opposition, not the ruling party, that became less participatory.

# 22nd Assembly: 368,790 roll-call records, 1,236 bills, 304 members
# PPP within-member absence change: -0.001, paired t = -0.219, p = 0.827
# DPK within-member absence change: +0.015, paired t = 4.157, p = 0.0001

The 20th Assembly provides the contrast the seed topic predicts. During the Park Geun-hye impeachment, the ruling party (Saenuri/Liberty Korea) did increase its absence rate from 47.4% to 50.9%, a within-member change of +9.8 percentage points (paired t = 2.488, p = 0.032, N = 11). The 20th Assembly ruling party, facing impeachment of its own president, behaved exactly as the "governance vacuum" hypothesis predicts. The 22nd Assembly ruling party did not.

Why the difference? The most likely explanation is structural: in the 20th Assembly, the ruling Saenuri/Liberty Korea held a near-majority (122 of 300 seats initially) and its members' votes were consequential for bill outcomes. In the 22nd Assembly, the PPP holds only 108 seats against a DPK-led supermajority. PPP members' attendance or absence rarely changes outcomes. Strategic shirking requires that shirking matters - that absence imposes costs on the agenda-setter. In a legislature where the opposition holds a comfortable majority, ruling-party absenteeism is inconsequential, and there is no strategic reason to vary it in response to investigation pressure.

3. Livelihood Bills: The Governance Vacuum is Real But Non-Partisan

Using an expanded keyword classification (18 livelihood categories: 의료, 교육, 복지, 연금, 안전, 환경, 농업, 고용, 건강, 보육, 장애, 노인, 출산, 주거, 식품, 물가, 소비자, 민생), I classified 3,183 of 17,205 bills (18.5%) as livelihood-related in the 22nd Assembly.

The resolution rate collapse is real but hits livelihood bills harder. Resolution rates (any final action, including passage, alternative absorption, or rejection) dropped for all bills after December 3, but the decline was steeper for livelihood bills:

The DID of -6.7 percentage points means livelihood bills lost an additional 6.7 percentage points of resolution probability beyond the general legislative slowdown. This confirms the seed topic's intuition that "bread-and-butter" legislation suffers disproportionate collateral damage during accountability crises.

But the effect is bipartisan. Both DPK-sponsored and PPP-sponsored livelihood bills suffered nearly identical passage rate collapses:

PPP-sponsored livelihood bills actually saw a larger decline (-4.2pp vs -2.1pp), but both rates converged toward near-zero. The alternative absorption rate (대안반영폐기) tells the fuller story: DPK livelihood bills dropped from 32.8% to 12.9%, while PPP livelihood bills dropped from 37.4% to 12.6%. The committee system froze for everyone.

# 3,183 livelihood bills out of 17,205 total (18.5%)
# Resolution DID: livelihood -21.9pp vs other -15.2pp = -6.7pp additional penalty
# DPK livelihood alt absorption: 32.8% -> 12.9% (-19.9pp)
# PPP livelihood alt absorption: 37.4% -> 12.6% (-24.8pp)

4. Bill Processing Timestamps: Political Bills Get Fast-Tracked

Scout (007_literature_scout.md, suggestion #5) asked whether bill processing timestamps reveal scheduling bias. They do.

Committee processing delays (days from committee referral to committee action) show a sharp asymmetry:

Political bills were already fast-tracked before December 3 (median 23 days vs 126 for livelihood), and the gap widened further afterward. Post-December 3, political bills received committee action in a median of 9 days - one-fourteenth the wait for livelihood legislation. This is the clearest evidence of attention displacement at the committee level: the institutional infrastructure that processes bills has finite capacity, and accountability mechanisms are consuming a disproportionate share.

5. The Placebo Test: Defense/Foreign Affairs Committees

Scout and Critic both requested a placebo test using non-housing policy domains. I compare defense/foreign affairs committees (which should be unaffected by the insurrection's domestic political dynamics) with livelihood committees and the judiciary committee (법사위, which processes special counsel bills).

The placebo partially fails: defense and foreign affairs committees show the largest passage rate decline (-8.3pp), greater than livelihood committees (-4.9pp). This suggests the legislative freeze is not specific to domains affected by the investigation but is a general institutional phenomenon. The finding undermines a pure "attention displacement" story in which only investigation-adjacent committees suffer, and supports instead a systemic institutional freeze where the entire legislative apparatus slows during acute political crises.

The 법사위 anomaly deserves separate attention: its passage rate barely declined (-1.5pp) because it was already low (2.8%). But as 008_data_analyst.md documented, within 법사위, political bills pass at 27.5% while non-political bills pass at 8.7% post-December 3. The committee's scarce processing capacity is selectively allocated to accountability legislation.

6. Cross-Assembly Comparison: December 3 vs Park Impeachment

The 20th Assembly (Park impeachment) provides a natural comparison case. Using symmetric 6-month windows around each crisis event:

The Park impeachment produced no governance vacuum in legislative passage rates. Livelihood bill passage was unchanged (13.0% to 13.1%), and other bills actually increased their passage rate (+2.5pp). This is a striking contrast to the December 3 event, where the resolution rate collapse documented in Section 3 is massive (-21.9pp for livelihood bills).

Why the difference? Two institutional factors: (1) The 20th Assembly impeachment was a bipartisan act - both ruling and opposition parties voted to impeach Park, meaning legislation could proceed through cross-party cooperation even during the crisis. (2) The 22nd Assembly is dominated by a single-party supermajority (DPK + allies > 200 seats) whose attention was monopolized by accountability proceedings, leaving no coalition partner to maintain routine legislation.

Special counsel bill escalation across assemblies further illustrates the trend:

The 22nd Assembly has produced more special counsel bills (70) than the 17th through 19th assemblies combined (45). More remarkably, 8 have been formally rejected on the floor - unprecedented in any previous assembly. The serial introduction-rejection-reintroduction cycle consumes 법사위 bandwidth, plenary session time, and floor vote scheduling capacity.

7. The DPK Sponsorship Composition Did Not Shift Toward Political Bills

One might expect that an opposition party pursuing accountability would redirect its legislative effort from livelihood to political bills. The data show no such shift. DPK livelihood bill sponsorship as a share of total sponsorship was stable: 18.9% pre-December 3 vs 17.9% post-December 3 (-1.0pp). PPP livelihood share was similarly stable: 18.2% pre vs 17.8% post (-0.4pp). Political bills remained a tiny fraction of both parties' portfolios: 0.9% to 1.3% for DPK, 0.4% to 0.5% for PPP.

# DPK livelihood share: 18.9% pre -> 17.9% post (-1.0pp)
# PPP livelihood share: 18.2% pre -> 17.8% post (-0.4pp)
# DPK political share: 0.9% pre -> 1.3% post (+0.4pp)

The governance vacuum in livelihood legislation is therefore not caused by legislators choosing to sponsor political bills instead of livelihood bills. Both parties continue to sponsor livelihood legislation at roughly the same rate. The bottleneck is downstream: in committee scheduling, subcommittee processing, and floor vote allocation - the institutional mechanisms that Park Poem Young (2026) identifies as the site of discretionary gatekeeping.

8. Reinterpreting the Seed Topic: Three Claims Tested

The seed topic correctly identifies a real phenomenon (livelihood bill collateral damage) but misidentifies the mechanism (ruling-party shirking) and the moderator (network proximity). The evidence instead supports an institutional capacity constraint model: the legislative system has a fixed processing bandwidth, and accountability mechanisms - special counsel bills, impeachment proceedings, floor votes on politically charged legislation - crowd out routine legislation. This crowding-out operates through committee scheduling bottlenecks (median processing delay: 130 days for livelihood vs 9 days for political bills) rather than through individual legislators' strategic attendance decisions.

9. Data Gaps and Limitations

Committee attendance vs floor vote attendance. Roll-call votes capture floor participation but not committee participation, which is the seed topic's actual dependent variable. Committee meeting records (committee_meetings_22.parquet) track which bills were discussed but not which legislators were present. The kr-hearings-data's speech frequency (suggested by Scout) is a behavioral proxy but not a direct attendance measure.

Faction data remain unavailable. The seed topic hypothesizes that faction membership moderates the investigation chill. No faction affiliation variable exists in the KNA database. The PPP's internal factions (친윤, 친한, 비한 groups) would need to be manually coded from media reports.

The 22nd Assembly is still ongoing. Post-December 3 bills filed in 2025-2026 have had less time to be processed than pre-December 3 bills. The "pending" rate is mechanically higher for recent bills. The resolution rate comparison partially reflects this time-to-maturation confound, though the magnitude of the decline (-21.9pp for livelihood) far exceeds what maturation alone could explain.

Co-sponsorship network limitations. The cosponsorship_edges data capture only lead sponsors (대표발의), not the 10+ co-signers typical of Korean bills. A richer network measure using full co-signatory lists would provide more granular proximity estimates. Additionally, the investigation targets are executive figures, not legislators - making co-sponsorship an inherently weak proximity measure for this question.

10. Suggestions for Critic

1. Is "institutional capacity constraint" a publishable theoretical contribution? The finding that legislative bandwidth is zero-sum and that accountability mechanisms crowd out routine legislation through committee bottlenecks (not individual shirking) extends the gridlock literature (Tsebelis 2002; Cox and McCubbins 2005) with a novel mechanism. But is "committees have finite capacity" too obvious to be a contribution?

2. The 20th vs 22nd Assembly divergence is the strongest finding. Ruling-party absence increased during the Park impeachment (where the ruling party held near-majority power) but not during the December 3 crisis (where the ruling party is a powerless minority). This interaction between crisis severity and minority status could ground a theoretical argument about when "governance vacuums" emerge: only when the ruling party has enough seats for its shirking to be consequential.

3. The defense/foreign affairs placebo failure. The fact that defense committees showed the largest passage rate decline undermines domain-specific attention displacement as the mechanism. Critic should evaluate whether a pure "systemic institutional freeze" story - where all legislation slows during acute crises regardless of policy domain - is theoretically interesting or merely tautological.

Completion Checklist

• [x] Ran at least 2 KNA queries or pandas analyses with actual results (10 distinct analyses: co-sponsorship networks, bill classification, processing timestamps, roll-call absenteeism by party, within-member absence change, network-absence correlation, cross-assembly comparison, placebo test, sponsorship composition shift, bill text 민생 framing)
• [x] Reported key statistics (N, means, percentages, trends across all analyses)
• [x] Connected findings to literature gaps identified by Scout (Gap 2: committee attendance - tested via roll-call proxy; Gap 3: co-sponsorship network proximity - tested and null; Gap 1: investigation-legislative behavior - first individual-level test)
• [x] Identified at least 1 data limitation or gap (committee attendance vs floor attendance, faction data, ongoing assembly time confound, co-sponsorship network incompleteness)
• [x] Included reproducible code/commands for every finding
• [x] Suggested what Critic should evaluate for theoretical framing (3 items)

References

Cox, Gary W., and Mathew D. McCubbins. 2005. Setting the Agenda: Responsible Party Government in the U.S. House of Representatives. New York: Cambridge University Press. doi:10.1017/CBO9780511791123

Frank, Marco, and David Stadelmann. 2021. "Political Competition and Legislative Shirking in Roll-Call Votes: Evidence from Germany for 1953-2017." Public Choice 189 (3-4): 391-413. doi:10.1007/s11127-021-00906-w

Gavoille, Nicolas, and Marijn Mohen. 2017. "Who Are the 'Ghost' MPs? Evidence from the French Parliament." European Journal of Political Economy 49: 147-162. doi:10.1016/j.ejpoleco.2017.07.007

Han, Seungwoo. 2022. "Elite Polarization in South Korea: Evidence from a Natural Language Processing Model." Journal of East Asian Studies 22 (1): 51-76. doi:10.1017/jea.2021.36

Jensen, Christian B., Sven-Oliver Proksch, and Jonathan B. Slapin. 2013. "Parliamentary Questions, Oversight, and National Opposition Status in the European Parliament." Legislative Studies Quarterly 38 (2): 259-282. doi:10.1111/lsq.12013

Kang, Sin-Jae, and Jiyoung Park. 2025. "Why Do Legislators Engage in Waffling? Evidence from the Korean National Assembly, 2004-2020." Journal of East Asian Studies 25 (2). doi:10.1017/jea.2025.10013

Katz, Andrea Scoseria. 2018. "Making Brazil Work? Brazilian Coalitional Presidentialism at 30 and Its Post-Lava Jato Prospects." Revista de Investigacoes Constitucionais 5 (3): 77-102. doi:10.5380/rinc.v5i3.60965

Park, Poem Young. 2026. "Issues of Legislative Power Infringement in the Current Operation of the National Assembly's Direct-Referral System to Subcommittees and Directions for Reform." The Justice 212: 1-36. doi:10.29305/tj.2026.02.212.01