2016 INFORMS Annual Meeting Program

SC42

INFORMS Nashville – 2016

2 - Dea Computation For The Big Data – A Proactive Approach Wen-Chih Chen, National Chiao Tung University, Hsinchu, Taiwan, wenchih@faculty.nctu.edu.tw, Yueh-Shan Chen This talk presents a computation strategy to determine the DEA efficiencies of a massive data set. The strategy proactively searches for the references of a data point under evaluation by solving small-size linear programs (LPs). The size of each individual LP solved is controlled within a guarantee upper bound. The approach does not rely on the data density, and can improve the computational performance significantly. 3 - Segmented Concave Least Squares: An Automatic Classification Method With An Application To The Analysis Of The Room Rates Of Hotels In Finland Abolfazl Keshvari, Aalto University School of Business, Helsinki, Finland, abolfazl.keshvari@aalto.fi In this paper, segmented concave least squares estimator is introduced. It estimates a piecewise linear concave function, wherein the number of linear segments (k) is pre-specified. Two extreme cases of this problem are ordinary least squares (k=1) and concave least squares (k=n, the number of observations). The estimator is used to analyze the room rates of hotels in Finland and to classify them into three groups based on their pricing strategies. SC41 207C-MCC Finance Section Student Paper Competition Sponsored: Financial Services Sponsored Session Chair: Rafael Mendoza, McCombs School of Business, 1, Austin, TX, 1, United States, rafael.mendoza-arriaga@mccombs.utexas.edu Co-Chair: Tim Siu-Tang Leung, Columbia University, New York, NY, United States, tl2497@columbia.edu 1 - Robust Versus Sparse Portfolio Selection: Insights And Alternatives Yufei Yang, Singapore University of Technology and Design, In this talk, we will provide an in-depth discussion on the robustness and sparsity trade-off in finding the mean-variance portfolio. We extend the classical mean- variance framework by incorporating an ellipsoidal uncertainty set and fixed transaction costs. We demonstrate that the optimal portfolio can be approximated by a linear combination of three benchmark portfolios, and discuss how the number of traded assets changes with respect to uncertainty level and transaction cost. 2 - Portfolio Liquidity Estimation And Optimal Execution Kai Yuan, Columbia Business School, Broadway, New York, NY, 10027, United States, kyuan17@gsb.columbia.edu We develop a tractable model to estimate portfolio liquidity costs through a multi- dimensional generalization of the optimal execution model of Almgren and Chriss. Our model allows for the trading of standardized liquid bundles of assets (e.g., ETFs or indices). We show that in a “large universe” asymptotic limit, where the correlations across a large number of assets arise from relatively few underlying common factors, the liquidity cost of a portfolio is essentially driven by its idiosyncratic risk. Moreover, the additional benefit of trading standardized bundles is roughly equivalent to increasing the liquidity of individual assets. 3 - Spectral Portfolio Theory Shomesh E Chaudhuri, Massachusetts Institute of Technology, Cambridge, MA, 0, United States, shomesh@mit.edu Andrew W Lo Economic shocks can have diverse effects on financial market dynamics at different time horizons, yet traditional portfolio management tools do not distinguish between short- and long-term components in alpha, beta, and covariance estimators. In this paper, we apply spectral analysis to quantify stock- return dynamics across multiple time horizons. Using the Fourier transform, we decompose asset-return variances, correlations, alphas, and betas into distinct frequency components. These decompositions allow us to identify the relative importance of specific time horizons in determining each of these quantities, as well as to construct mean-variance-frequency optimal portfolios. Singapore, Singapore, eeyufei@gmail.com Selin Damla Ahipasaoglu, Jingnan Chen

4 - Long Term Risk: A Martingale Approach Likuan Qin, Northwestern University, Evanston, IL, 60208, United States, likuan.qin@gmail.com This paper extends the long-term factorization of the stochastic discount factor introduced and studied by Alvarez and Jermann (2005) in discrete time ergodic environments and by Hansen and Scheinkman (2009) and Hansen (2012) in Markovian environments to general semimartingale environments. The transitory component discounts at the stochastic rate of return on the long bond and is factorized into discounting at the long-term yield and a positive semimartingale that extends the principal eigenfunction of Hansen and Scheinkman (2009) to the semimartingale setting. The permanent component is a martingale that accomplishes a change of probabilities to the long forward measure, the limit of T- forward measures. The change of probabilities from the data generating to the long forward measure absorbs the long-term risk-return trade-off and interprets the latter as the long-term risk-neutral measure.

SC42 207D-MCC Stochastic Systems in Finance Sponsored: Financial Services Sponsored Session

Chair: Alexandra Chronopoulou, Assistant Professor, University of Illinois, Urbana-Champaign, 117 Transportation bldg. MC-238, 104 S. Mathews Ave., Urbana, IL, 61801, United States, achronop@illinois.edu 1 - Statistical Inference For Long Memory Stochastic Volatility Models Alexandra Chronopoulou, University of Illinois, Urbana-Champaign, achronop@illinois.edu Long memory stochastic volatility (LMSV) models have been used to explain the persistence of volatility in the market, while rough stochastic volatility (RSV) models have been shown to reproduce statistical properties of low frequency financial data. In these two classes of models, the volatility process is often described by a fractional Ornstein-Uhlenbeck process with Hurst index H, where H>1/2 for LMSV models and H<1/2 for RSV models. The goal of this talk is to propose a general methodology for the estimation of the parameters of the above models, the filtering of the volatility process, and the calibration of the Hurst index, H, which will then be applied to the option pricing on the S&P 500 index. 2 - Optimal Randomized Unbiased Monte Carlo Simulation Of Discounted Costs Zhenyu Cui, Stevens Institute of Technology, zcui6@stevens.edu In this talk, we consider the problem of estimating the expected infinite-horizon cost of running a stochastic system with stochastic fluctuations using Monte Carlo simulation. We propose a randomized unbiased estimator based on truncating the simulation horizon at an independent random time. The problem of determining the optimal randomization distribution of the truncation random variable is formulated as minimizing the “work-variance product” proposed by Glynn and Whitt (1992). We solve this optimization problem explicitly and prove that it is always optimal to use a “shifted” distribution. Numerical experiments illustrate our findings. (This is joint work with Lingjiong Zhu). 3 - Monte Carlo Estimation Of Sensitivities From Analytic Characteristic Functions Runqi Hu, University of Illinois, Urbana-Champaign, runqihu2@illinois.edu, Liming Feng Sensitivity analysis is transformed into simulating a probability expectation through the likelihood ratio method (LRM). In this paper, we apply Hilbert transform inversion in evaluating a cdf on a uniform grid from its characteristic function and provide explicit bound for estimation bias. In one dimension cases, the bound allows one to determine the size and fineness of the grid and numerical parameters for the inversion. For multidimensional cases, the parameters can be determined by a procedure that will be proved to converge, and work well practically. In the numerical experiments part, the method is applied in estimating both European and Asian option deltas under CGMY model. 4 - Sensitivity Of The Eisenberg-Noe Network Model To The Relative Liabilities Mackenzie Wildman, University of California, Santa Barbara, CA, United States, mackenzie.wildman@gmail.com Zachary Feinstein, Weijie Pang, Birgit Rudloff, Eric Schaanning, Stephan Sturm The Eisenberg-Noe algorithm gives a clearing payment vector for a system of interconnected financial institutions in which some banks are unable to fulfill their obligations to other banks in full. The network model takes as input a relative liability matrix which gives the liabilities owed from each bank to its counterparties. In practice, these liabilities are generally unknown and must be estimated. We perform sensitivity analysis on this relative liability matrix and obtain a worst-case scenario in terms of the payoff to a “society” node. We illustrate our results on an EBA dataset of European banks.

83