[1]
A Social-Spatial Network Choice Model with Application to Pop-Up Store Operations
Submitted to Management Science, 2026
[Abstract]
Social interactions are increasingly shaping consumers'
movements and visits to physical venues. Visits are also
spatially linked, as consumers frequent nearby locations and
demand spills over across adjacent sites. So, visit decisions
may depend jointly on (i) social relationships among
individuals and (ii) spatial relationships among locations.
Modeling and estimating dual-network dependence is
challenging because utilities spill over through both
networks, inducing high-dimensional dependence across a choice
space that grows with the product of the number of individuals
and locations. We develop an integrated framework for modeling
and estimating individual-location choices under social and
spatial dependence and use it for operational decisions. The
framework begins with a social-spatial network choice model in
which both dependence channels jointly propagate through
individual-location utilities. For scalability, we exploit the
dual-network structure and construct a copula surrogate that
preserves model-implied first- and second-order information,
yielding a pairwise likelihood estimator and closed-form visit
probabilities for downstream optimization. We apply the
framework to pop-up store operations, where a firm jointly
selects a temporary location and seed users for promotion. We
develop an exact reformulation and a scalable approximation
method for this joint decision problem. Synthetic experiments
show our estimator accurately estimates parameters across
multiple data generating processes. Using real-world
location-based social network data, we document positive,
statistically significant social and spatial dependence in
users' recorded check-in behavior. In this application,
coordinating both decisions substantially improves expected
demand over location-only optimization, and the optimal seed
set balances broad social reach against promotional
effectiveness at the selected site.
[SSRN]