Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Dynamic network signatures of Labor Flows

Dynamic network signatures of Labor Flows

■イベント
NetSci 2019
http://vermontcomplexsystems.org/events/netsci/

■登壇概要
タイトル:
Dynamic network signatures of Labor Flows

登壇者:
DSOC R&Dグループ 西田貴紀

▼Sansan Builders Box
https://buildersbox.corp-sansan.com/

Sansan

May 31, 2019
Tweet

More Decks by Sansan

Other Decks in Technology

Transcript

  1. Dynamic Network Signatures of Labor Flows:
    Evidence from a Large Business Social Network
    Takanori Nishida
    (Sansan Inc.,
    Tokyo, Japan)
    Lav R. Varshney
    (University of Illinois
    at Urbana-Champaign)
    Yoshiki Ishikawa
    (Habitech, Inc.,
    Tokyo, Japan)

    View full-size slide

  2. Strength of Weak Ties Hypothesis
    Weak Ties Strong Ties
    Results are mixed!

    View full-size slide

  3. Bird’s-Eye View
    Static
    Dynamic
    Online Offline (Face to Face)
    Previous
    Studies
    Facebook
    LinkedIn
    Our
    Research

    View full-size slide

  4. Data
    ▶ The longitudinal dataset includes anonymous user’s profile and daily
    trajectories among over 1.5 million individuals
    over 3 years (from 2015 to 2017)
    Data that is anonymized within the permission scope of the "Eight Service Terms of Use" is analysed only statistically. The result of this research will be used for improving the service.

    View full-size slide

  5. Data on job change
    Current Job
    Previous Jobs

    View full-size slide

  6. Empirical Definition of Weak ties and Strong Ties
    Job Change
    Strong Ties
    T-1 T
    T-2
    = Number of
    New Bridging Ties
    =Δ Clustering coefficient
    Weak Ties
    Bridging Ties
    Ties in same community

    View full-size slide

  7. Empirical Model
    ▶ We estimate the following Fixed effects model
    ̈
    "#$
    = &' ̈
    (#$)* /$),
    + ̈
    .#$
    /ℎ121 ̈
    "#$
    = "#$
    − 4
    "# (within transformation),5
    "#
    = 1/7 ∑#9*
    : "#$
    Fixed effects model
    ▶ Using Variables
    ▶ Dependent Variable: Dummy variable for job change (T)
    ▶ Explanatory Variables: Weak ties and Strong Ties (T-1)
    ▶ Control Variables: Dummy variable for job change (T-1); Number of newly formed inner
    community ties (T-1); degree (T-2); clustering coefficients (T-1); and time fixed effects

    View full-size slide

  8. Results
    Weak Ties
    Job Change
    Strong Ties
    T-1 T
    T-2
    -0.006***
    (0.001)
    0.003***
    (0.001)
    R-squared:0.067
    Num. obs.: 4,357,566
    - Standard errors in parentheses
    - ***p < 0.001, **p < 0.01, *p < 0.05

    View full-size slide

  9. Discussion (1)
    An increase in strong ties is associated with an
    increase in the probability of job change.
    ▶One explanation could be that individuals in tightly-knit
    communities tend to form new ties with those in the
    same community who deliver time-critical information on
    desirable jobs.

    View full-size slide

  10. Discussion (2)
    Contrary to the strength of weak tie hypothesis, we
    found a negative impact of new weak ties on job change
    ▶Allocating limited resources to connect with weak tie may reduce
    spending time with strongly connected people, which are more
    likely to deliver time critical information (e.g., with job opportunity)
    ▶Due to social capacity limits (Aral, 2011), new weak and strong
    ties are complementary.

    View full-size slide

  11. Future work

    Our motivation:
    ▶We live in the world where knowledge drives innovations, which
    lead to economic growth.
    ▶We are interested in job change as one driver for knowledge
    transfer across social network.
    Dynamic change
    in network structure
    Job change
    ???

    View full-size slide