Vacancy Chains as Strategy: Inter-Administration Mobility of Political Elites in Reform China
Shilin Jia
University of Chicago
Bigger picture
- Black box of Chinese Communist Party (CCP)’s elite politics
- Raw Data: Huge volumes of high-level Chinese political elites’ CVs
- RQ: What can we say about the careers of CCP elites? And what can we say about the CCP’s staffing strategy?
- Analytic strategy: Computational & organizational analysis
Two perspectives for studying mobility of elites
- Individual perspective
- views mobility from the perspective of the individual who moves through positions
- important factors: human capital (merit), social capital (connections), structural opportunities
- Organizational perspective
- views mobility from the perspective of the organization responsible for filling positions, e.g. institutionalism
- Our perspective: “Structure follows strategy” (Chandler 1990: 14)
Our central claim
- Individuals’ careers cannot be studied as independent when they are endogenously generated by the organization.
- We bring in organizational strategy to explain career patterns and success.
How careers appear in CVs
How careers are generated
Let’s switch perspectives
- People cannot freely move inside a system, but vacancies can (White 1970)
Data
- Source: CVs of political elites of the Chinese Communist Party(CCP) (http://cped.nccu.edu.tw/)
- Who are they?
- Chinese political elites who have reached the level of vice-minister or vice-governor after 1966 (N ~ 4000)
- ~ 11.7k job entries
- Period: 1978-2011
- Method: computer-assisted coding
Computer-assisted coding
Goal: Recognizing the organizational unit (provincial or functional) where job entries took place
At least 8 rounds of manual examination of the results via sampling (N = 100 each)
Error rates estimated from the most recent sample check:
- 3%, 17 missed and 7 wrong in 948 jobs of 100 people. 5 missed (false negatives) and 2 over-coded (false positives) in 51 transfers
Computer-assisted coding
Why inter-organizational transfers?
- Internal hires: specialization
- External hires: integration
Isolated vs. chained transfers
Tracking vacancy chains
- Predecessors and successors must occupy the exact same posts
- Maximum vacancy window: 6 months
- Fuzzy match followed by manual checks by me and 3 research assistants
- All chains and isolated transfers are complete.
White (1970)’s original assumption
- Vacancy chains as Markov chains
Key distinction of our study
- Isolated transfers
- more likely initiated by locals on ad hoc bases
- Chained transfers
- more likely orchestrated by the state and part of a strategy
China’s transition to market economy
Organizational perspective
- Departure point: Soviet planning system as a giant factory organized by central planning agencies
- Decentralization as a solution (Chandler 1990; Fligstein 1985) and M-form transformation (Xu 2011; Qian and Weingast 1996, 1997)
- Piecemeal transition, a ”dual-track” system (Naughton 1995; Shirk 1993)
- Centrifugal force: decentralization (Landry, 2008); regional protection and bureaucratic fragmentation (Lieberthal 1992; Li and Bachman 1989)
- Centripetal force: personnel management (Landry, 2008; Naughton and Yang 2004; Xu 2011)
CCP’s organizational strategy
- nomenklatura (职务名称表)
- reserved cadres (后备干部) and “third echelon” (第三梯队)
- long-planned sponsorship through orchestrated transfers
Vacancy chains as strategy
Simultaneous replacement of vacancies
Longer chain lengths over time
Markovian analysis
- Null hypothesis: Long chains were successions of independent isolated transfers
- Where a vacancy visits next only depends on where it currently is
- We can calculate expected chain lengths based on
- observed initial vacancies plus
- estimated transitional probabilities (parametric) or
- simulations via random walks (non-parametric)
Orchestration: Longer lengths than expected by chance
Expected chain lengths under Markovian models
More lateral transfers in vacancy chains
chain length
|
demotion (%)
|
lateral (%)
|
promotion (%)
|
1
|
3.2
|
56.5
|
40.3
|
2
|
4.6
|
62.2
|
33.2
|
3
|
6.8
|
63.4
|
29.8
|
4
|
10.7
|
60.5
|
28.8
|
5
|
5.1
|
67.7
|
27.2
|
- Administrative ranks
- -1: No.1 person 一把手 (e.g. provincial party secretary, minister)
- -1.5: No.2 person (e.g. governor, standing deputy minister)
- -2: All positions at the vice-governor/deputy-minister level
- -3: All sub-province/ministry-level positions
Where do vacancies start and where do they terminate?
- In early periods, vacancies start in ministries and terminate in provinces
- In later periods, vacancies start in provinces and are less likely to terminate.
Where do vacancies visit?
- Centripetal force
- The central state tries to penetrate its subunits through external transfers via vacancy chains
- Centrifugal force
- Powerful local strongholds (e.g. Shanghai and Guangdong) try to internally absorb their vacancies as much as possible
Estimating organizational ranks through organizational hieararchy
- Not all organizations are equal even when administrative ranks are the same.
- Hierarchy of organizations can be observed through transfer patterns (Padgett 1990).
- The hierarchy should be the most salient at the very top level (admin rank = 1) where competition for limited number of posts is extremely fierce.
Estimation
- Assumption: head(unit \(i\)) \(\rightarrow\) head(unit \(j\)) implies \(i \leq j\) where the arrow means a top-level lateral transfer.
- Ranking provinces and ministries through network triangularization (Jia and Xu 2018; Schiavinotto and Stützle 2005)
Validation
- Alternative method: Bradley-Terry-Luce (BTL) model
- produces very similar rankings (corr > 0.9 )
- Org ranks correlate with provincial per capita GDP at 0.6~0.7
Vacancy chains tend to start in high-ranking places, cascade downwards, and terminate in high-ranking places
They tend to terminate in high-ranking provinces
But not high-ranking ministries
Terminal probability by per capita GDP rank
Per capita GDP quintile
|
1
|
2
|
3
|
4
|
5
|
1978-1991
|
0.44
|
0.3
|
0.5
|
0.62
|
0.59
|
1992-2001
|
0.42
|
0.44
|
0.31
|
0.63
|
0.58
|
2002-2011
|
0.37
|
0.35
|
0.28
|
0.47
|
0.48
|
All
|
0.4
|
0.37
|
0.32
|
0.54
|
0.53
|
Let’s switch to the individual perspective
What happened to the elites involved in these transfers?
- Suppression hypothesis: VCs transferred powerful elites out of their local bases
- Sponsership hypothesis: By moving people (presumably pre-screened officials on reserved-cadre lists), VCs put them into faster career tracks
Short-term tradeoff
- Mean des-ori org rank differential by admin rank differential by decade
decade
|
1978-1991
|
1992-2001
|
2002-2011
|
demotion
|
0.43
|
0.42
|
0.3
|
lateral
|
0.19
|
0.06
|
0.09
|
promotion
|
-0.09
|
-0.13
|
-0.2
|
- Higher trade-off for chained transfers (regression not shown)
Long-term benefit
- The involved elites can also expect that their coming stops are not their final destinations but stepstones for further career advancement
Hypothesis: everything else being equal, an official involved in a longer chain should have a brighter future than his counterpart.
Regression
- Unit of analysis: transfer-person
- DV: Whether the person involved in a transfer is promoted in the next Party Congress.
- Party rank: The ultimate rank of power in the CCP
- None
- Alternate Central Committee (ACC)
- Central Committee (CC)
- Politburo (PB)
- Politburo Standing Committee (PBSC)
- IV: VC Length
- Control:
- Transfer-level characteristics
- Person-level characteristics and special connections
Weak correlation with existing human and social capital variables
VC effects with matching
Temporal effects
Summary
- As China transitioned into decentralized market economy, the Party employed orchestrated vacancy chains to transfer its trusted elite members through a wide range of subunits
- Strong provinces were, to some degree, more able to shield themselves
- The strategy worked because elites involved in vacancy chains, in general, benefited from the transfers through either getting promotions or moving to better places
- The VCs put those elites into strategic positions, not only for the present, but also for the future.
Can we theorize above the Chinese case?