Análisis de redes

The Quarterly Review of Economics and Finance 50 (2010) 191–201

Contents lists available at ScienceDirect

The Quarterly Review of Economics and Finance
journal homepage: www.elsevier.com/locate/qref

Analysis of the probability of deletion of S&P 500 companies: Survival analysis and neural networks approach
John M. Geppert a , Stoyu I. Ivanov b,∗ , Gordon V. Karels a
a b

Universityof Nebraska, Lincoln, United States San José State University, Accounting and Finance Department, College of Business, One Washington Square, San José, CA 95192-0066, United States

a r t i c l e

i n f o

a b s t r a c t
We examine the probability of deletion of a firm from the S&P 500 index due to a decision of the index committee because the firm did not satisfy the index committeecriteria. We study the probability of deletion with survival analysis and neural networks methods. We document that deletion might be predictable, which is contrary to the findings of most studies that the market cannot predict the timing of a company deletion from the S&P 500 index. It might also be beneficial to know ahead of time which company might be deleted from an index, to supplement the arbitrageopportunities that exist already in the announcement-effective date event window. Published by Elsevier B.V. on behalf of The Board of Trustees of the University of Illinois.

Article history: Received 12 June 2009 Received in revised form 21 November 2009 Accepted 11 December 2009 Available online 22 December 2009 JEL classification: G12 G14 Keywords: S&P 500 deletion Survival analysis Neuralnetworks

1. Introduction The “S&P game” is the profit generating trade in a stock that is about to be added to the S&P 500 index. The trading sequence is to buy the stock before the index funds rebalance their portfolios on or around the index change announcement date and to sell the stock when they rebalance on the index change effective date. Beneish and Whaley (1996) study the effects of the“S&P game” and document that it became prominent after Standard and Poor’s began pre-announcing (five days prior to the effective date of change) S&P 500 index changes in October 1989. The reason for the arbitrage opportunity is the inactivity of index fund managers. Blume and Edelen (2004) and Chen, Noronha, and Singal’s (2006b) document inactivity of index funds until the effective index change datedue to daily tracking error constraints imposed on index fund managers by their investors. Another variant of the “S&P game” would be to short a stock that was removed from the index and then to repurchase on the effective date. Chen, Noronha, & Singal (2006a) and Lynch and Mendenhall (1997) find significant negative cumulative abnormal

returns (CARs) of 8–9% in the event window, betweenannouncement and effective date of S&P 500 index deletions, and 1–4 days prior to the announcement date. Considering the documented evidence of significant profits to be made in the “S&P game” and the extensive literature attempting to analyze and explain the “S&P game” it is surprising that there are very few publicly available models attempting to predict the S&P 500 changes.1 After all, predictabilitymodels of the index changes will be beneficial to arbitrageurs. The reason for the lack of predictability models is likely due to the S&P statement that changes are made as needed and thus the wide-spread belief in the finance literature that changes in the index composition are unpredictable. In this study we attempt to fill the void in the finance literature of predicting index changes by addressingthe following issue: What is the probability of a firm in the S&P 500 index being removed from the index due to a discretionary decision by the index committee and can excess returns be earned by trading on such information.

∗ Corresponding author. Tel.: +1 408 924 3465. E-mail address: ivanov s@cob.sjsu.edu (S.I. Ivanov).

1 The “S&P game” is a risk arbitrage trading strategy typically...