Contents of Volume 6, Number 2
Special Issue: Multi-Attribute Methods in Finance and Insurance (Guest Editor: Alejandro Balbás)
December 2010
Special Issue Papers
Portfolio Selection through the Goal Programming Model: An Overview
B. Aouni
Abstract: The main objective of the mutli-attribute portfolio selection methodology is to aggregate several conflicting and incommensurable objectives in order to choose a set of stocks (assets, securities) that will form the best financial portfolio. The attributes can be the return on investment, risk and liquidity. These objectives cannot be optimized simultaneously. Hence, the most satisfactory portfolio is the one of the best compromise. The compromises are guided by the preferences, the experience, the intuition and the judgement of the Financial Decision-Maker. In fact, multiple objectives and multi-criteria decision aid tools have been utilized for many years in finance. Goal Programming (GP) is one of these models that has been widely applied to solve portfolio selection problem. The aim of this paper is to highlight the different GP variants that have been applied for the portfolio selection problem.
Solving Stochastic Multi-Objective Programming in Multi-Attribute Portfolio Selection through the Goal Programming Model
B. Aouni, C. Colapinto,
Abstract: The aim of this paper is to present an approach for solving the Stochastic Multi-Objective Programming (SMOP) and formulate a Stochastic Goal Programming model that will be applied to multi-attribute portfolio selection problem. The concept of satisfaction function will be utilized to incorporate explicitly the financial decision-maker’s preferences for selecting the best financial portfolio based on several conflicting objectives. The proposed approach will be illustrated through numerical examples from the Tunisian stock exchange market.
Mutual Funds Portfolio Selection for Emerging Market: The Case for Egypt
M. Tamiz, R. Azmi
Abstract: This paper examines portfolio selection for mutual funds in Egyptian financial market based on Sharpe, Treynor and Goal Programming methodologies. The mutual funds are ranked and portfolios are constructed from the top 5 funds obtained from Sharpe and Treynor methodology respectively. Goal Programming is used to construct a portfolio of 5 mutual funds. The portfolios from the three approaches are compared against each other as well as against the benchmark. The performance comparison reveals that the portfolios constructed based on Sharpe and GP have the minimum deviations from each other as well as the benchmark. Although Sharpe and Treynor methodologies of ranking and selecting best mutual funds are well established historically, this research finds GP performing as good, if not better, than both Sharpe and Treynor in the Egyptian financial market.
Set-Valued Vector Risk Functions
A. Balbás, P.J. Guerra
Abstract: The concept of (real valued) risk measure is becoming more and more important and applied in Mathematical Finance. As a consequence recent literature has extended the analysis in order to introduce the notion of vector valued risk measure. Some interesting contributions have shown that the “set valued function approach'” may be a useful alternative to deal with vector risk measures. This paper draws on general Banach spaces and proves that set valued vector risk measures can also be considered in this wide framework. The article also presents a fundamental representation theorem and uses it in order to address several financial issues (pricing, hedging, portfolio choice) related to some classical optimization problems.
TSIR and Credit Risk Estimates with Goal Programming
B. Balbás, A. Heras
Abstract: The correct estimation of the Term Structure of Interest Rates (TSIR) is crucial in many financial and economic problems. Indeed, the TSIR has significant influence in Political Economy, Interest Rate Risk Hedging, Credit Spread Estimate, Valuation and Hedging, Derivative Pricing and Hedging, etc. Bond markets are usually incomplete, which provokes the existence of infinitely many solutions for the real TSIR. The final selection is implemented by a Parsimonious Principle, leading to an “as regular and smooth as possible” TSIR. However, the proposed methodologies usually minimize the committed error between market prices and those prices provided by the chosen TSIR. Consequently, the final result may be inaccurate, since it does not generate right prices. For instance, many authors have pointed out that the selected TSIR may lead to negative embedded option prices, or negative credit spreads of private corporations. This paper proposes a new methodology in TSIR and credit spread estimates. It uses Goal Programming Methods. Though it still respects the Parsimonious Principle it seems to present several advantages. Firstly, it generates exact market prices. Secondly, it generates smooth TSIRs and credit spreads even for companies with a small number of issued bonds and data. Thirdly it detects possible price errors in the available database. Finally, it is very easy to implement in practice.
Trading Volatility in Energy Markets
R. Balbás Aparicio
Abstract: Variance, volatility and correlation derivatives (henceforth volatility derivatives) are becoming more and more traded in practice because they are new instruments to diversify risks and, as pointed out by the empirical evidence, they provide adequate hedging when facing market turmoil. Besides, and also according to the empirical evidence, hedge funds and other non very risk adverse investors usually sell these products due to their high risk premium. There are two types of approaches that may apply in order to price and hedge volatility derivatives. One can draw on a general (complete) stochastic volatility model, for instance the Heston model, or use replicating portfolios composed of infinitely many options. The second way has the advantage that most of the options can be traded in the real market, so we already have a market price, even with the transaction costs generated by the classical bid/ask spread. However, it is impossible in practice to use infinitely-many options, which provokes the classical "hedging-errors". This paper deals with this kind of replicas and provides optimal approximations only composed of a finite number of options, providing alternative bid/ask prices for the volatility derivative as well as dynamic hedging strategies. We use general risk functions (coherent or expectation bounded measures of risk, deviation measures, etc.) in order to construct both the hedging portfolio and the bid/ask spread of the volatility derivative.
Modelling the Disability Severity Score in Motor Insurance Claims: An Application to the Spanish case
M. Santolino, J.-P. Boucher
Abstract: Bodily injury claims have the greatest impact on the claim costs of motor insurance companies. The disability severity of motor claims is assessed in numerous European countries by means of score systems. In this paper a zero inflated generalized Poisson regression model is implemented to estimate the disability severity score of victims involved in motor accidents on Spanish roads. We show that the injury severity estimates may be automatically converted into financial terms by insurers. As such, the methodology described may be used by motor insurers operating in the Spanish market to monitor the size of bodily injury claims. By using insurance data, applications with financial implications are presented in which the score estimate of disability severity is of value to assist the insurer decision maker, either for negotiation the claim compensation or for reserving reported claims.
Multivariate Stochastic Covariance Models and Applications to Pricing and Risk Management
P. Olivares, M. Escobar, A. Alvarez, L. Seco
Abstract: We study some elements of pricing and risk management under multivariate models considering stochastic covariance. In particular, we price Mountain Range Derivative, Spread and Quantos under some of these models. Models based on Principal Components and Factor Analysis are considered in order to reduce dimension.
Regular Papers
Credit Risk Modelling and Change in Accounting Standards: Evidence from the Adoption of IFRS in Greece
A. Apostolou, A.I. Dimitras
Abstract: The adoption of new accounting standards and their differences from the previously accepted ones, lead to changes that are reflected in the information presented in the financial statements. These changes assume great importance to financial analysts. The financial decision makers acquire this information mainly in the form of financial ratios and utilize it either directly or more commonly, as an input to decision aid models like credit scoring models. This study aims to investigate the importance of the differences in the disclosed financial information to develop credit scoring models and the credit rating of the firms that emerged by adopting the IFRS in accordance with the European Regulation n° 1606/2002 in Greece. This paper investigates the differences in the accounting ratio values, the models and the decisions that arise when financial statements are used conforming to the IFRS instead of to the Local GAAP. The results are comprehensively discussed, drawing important conclusions.
IFRS Voluntary Compliance: The Case of
A. Apostolou, K.A. Nanapoulos
Abstract: The European Regulation n° 1606/2002 introduced under conditions the IFRS, for the European listed companies, for each financial year starting on or after January 1st, 2005. In compliance with this Regulation, it was legislated in Greece (Law 3229/2004) the mandatory use of IFRS for the all the listed companies (consolidated or no) and the voluntary use of IFRS for the other companies. The primary objective of this paper is to present and examine the financial characteristics of Greek firms which voluntary adopt IFRS. Four types of firm financial characteristics (size, leverage, liquidity, efficiency) and six non-financial qualitative characteristics are evaluated using as sample all Greek firms which voluntary adopt IFRS, in year 2006. The study uses a cross-section model, in which a quantitative measure of firm qualitative characteristics was regressed on proxy financial characteristics in order to detect the existence of a statistically significant relationship. The findings of the study show that the financial characteristics of liquidity and size, as it can be measured by profitability and the number of employees had a significant relationship with the depended variable.