Informs 2015

209 / 552

INFORMS Philadelphia – 2015
207
3 - Influence Maximization Revisited
Paramveer Dhillon, MIT Sloan School of Management,
77 Massachusetts Avenue, Cambridge, MA, 02139,
United States of America, , Sinan Aral
Most research on influence maximization has focused on a single task: to devise
algorithms with better approximation guarantees for the NP-Hard discrete
optimization problem. The influence models over which the optimization
operates, however, remain simplistic and disconnected from empirical evidence
on influence in real networks. We propose extensions to existing models of
influence propagation that incorporate the most recent empirical evidence and
study the implications of these extensions.
4 - Is Exercise Contagious? Evidence from a Global
Natural Experiment
Christos Nicolaides, Postdoctoral Fellow, MIT Sloan School of
Management, 100 Main St, E62-489, Cambridge, MA, 02142,
United States of America, , Sinan Aral
Health-related behaviors, such as fitness habits, cluster amongst connected peers
in social networks. Clustering of behaviors is in part attributable to causal social
influence but can also arise through alternate means like homophily of
preferences. Using fined grain individual running data from Nike+ we devise a
natural experiment to quantify social contagion, identify influential members and
groups and determine under which conditions influence is the dominant factor in
behavior clustering.
MC06
06-Room 306, Marriott
INFORMS Section on Finance Student
Paper Competition
Sponsor: Financial Services
Sponsored Session
Chair: Jim Bander, Toyota Financial, Chandler, AZ,
United States of America,
1 - Revisiting Eisenberg - Noe: A Dual Perspective
Deung-geon Ahn, KAIST, #2111, E2-2, 291 Daehak-ro, Yuseong-
gu, Daejeon, Korea, Republic of,
Kyoung-kuk Kim
In this paper, we consider the Eisenberg-Noe framework for systemic risk with
random shocks. Using duality, we characterize the amount of shock amplification
due to the network structure and find the region for the shock vector that makes
a specific bank default. These results enable us to improve some of the existing
results of the network effect on systemic risk. More importantly, we propose
efficient simulation schemes for the systemic risk measurement based on the
characterization.
2 - A Partitioning Algorithm for MDPs and its Application to Limit
Order Books with Stochastic Market Depth
Ningyuan Chen, Columbia University, S. W. Mudd 321, 500 W
120th Street, New York, NY, 10027, United States of America,
Linear-quadratic control plays a central role in control theory, but its analytical
solution, the so-called linear-quadratic regulator, fails in the presence of
constraints. We consider a class of Markov decision processes (MDPs), with linear
inequality constraints, non-convex quadratic cost, and linear state dynamics,
governed by a Markov chain. By the proposed partitioning algorithm, we find the
explicit solution to this class of MDPs: The value function and the optimal policy
have analytical quadratic and linear forms, respectively, subject to a linear
partition of the state space. The algorithm is applied to two applications. In the
main application, we present a model for limit order books with stochastic market
depth to study the optimal order execution problem. As a feature of our model,
stochastic market depth is consistent with empirical studies and necessary to
accommodate various order activities, such as limit order submission at and
outside the best quotes and order cancellation, which may account for a large
proportion of limit order activities. As a result, the optimal order execution policy
is also stochastic and adapted to the random changes of market depth.
3 - An Optimization View of Financial Systemic Risk Modeling:
The Network Effect and the Market Liquidity Effect
Xin Liu, Doctoral Student, The Chinese University of Hong Kong,
609, William Mong Engineering Building, Hong Kong,
Hong Kong - PRC,
Abstract not available at this time.
4 - Accounting for Estimation Risk when Pricing under
Adverse Selection
Richard Neuberg, Columbia University, 1255 Amsterdam Avenue,
Dept of Statistics, 10th Floor, New York, NY, 10027,
United States of America,
Financial product prices often depend on unknown parameters. Their estimation
introduces the risk that a better informed counterparty may strategically pick
mispriced products. We discuss how overall estimation risk can be minimized by
selecting a probability model of appropriate complexity. Such a model has small
bias, which allows measuring product-specific estimation risk. We illustrate how
to determine a premium for estimation risk, using a simple example from pricing
regime credit scoring.
5 - Combined Estimation-Optimization (CEO) Approach for High
Dimensional Portfolio Selection
Chi Seng Pun, PhD Candidate, The Chinese University of Hong
Kong, Department of Statistics, Lady Shaw Building, Shatin, N.T.,
Hong Kong - PRC,
We propose a combined estimation-optimization (CEO) approach that directly
estimates the optimal trading strategy (optimal control), instead of separating the
estimation and optimization procedures. This paper investigates a constrained
$\ell_1$-minimization for estimating the optimal control and applies it to the
mean-variance portfolio (MVP) problems under static and dynamic settings when
the number of assets (p) is larger than the number of observation times (n). We
prove that the classical sample-based MVP strategy makes the probability that the
optimal portfolio will outperform the bank account tend to 50% for p>>n and a
large n. The CEO approach, however, converges to the true optimal solution. In
addition, the CEO scheme automatically filters out unfavorable stocks based on
historical data, and works for dynamic portfolio problems and non-Gaussian
distributions. Simulations validate the theory and the behavior of the proposed
approach. Empirical studies show that the CEO-based portfolios outperform the
equally-weighted portfolio, the MVP with shrinkage estimators and other
competitive approaches.
MC07
07-Room 307, Marriott
Modeling and Quantification of Risk
Cluster: Risk Management
Invited Session
Chair: Patrick Cheridito, Princeton University, ORFE, Princeton, NJ,
United States of America,
1 - Assessing Financial Model Risk
Pauline Barrieu, Professor, London School of Economics,
Statistics Department, Houghton Street, London, WC2A2AE,
United Kingdom, , Giacomo Scandolo
Model risk has a huge impact on any risk measurement procedure and its
quantification is therefore a crucial step. In this paper, we introduce three
quantitative measures of model risk when choosing a particular reference model
within a given class: the absolute measure of model risk, the relative measure of
model risk and the local measure of model risk. Each of the measures has a
specific purpose and so allows for flexibility.
2 - Multivariate Shortfall Risk and Monetary Risk Allocation
Samuel Drapeau, Shanghai Jiao Tong University, 211 West
Huaihai Road, Shanghai, 200030, China,
, Stephane Crepey, Yannick Armenti,
Antonis Papapantoleon
We present a measure designed to address the global and intrinsic risk of
interconnected system (banks, CCP...). The goal is twofold: first, provide the total
amount of liquidity to be reserved to overcome financial stress situations. Second,
address its allocation to each member in function of the systemic risk they put on
the system. Finally, we present how these high dimensional computations can be
solved in an efficient manner using Fourier methods.
3 - Variable Annuities with Guaranteed Withdrawal Benefits
Patrick Cheridito, Princeton University, ORFE, Princeton, NJ,
United States of America,
Variable annuities with withdrawal benefits have become popular over the last
couple of years. Their cost to the issuer not only depends on market conditions
but also on policyholder behavior. In this talk we discuss a contract whose
withdrawal guarantees are based on the running maximum of the account value.
The optimal withdrawal strategy is derived, and the cost of the contract to the
issuer is determined.
MC07