Duarte Trigueiros

Neural Networks and Empirical Research in Accounting

Abstract This article seeks to provide an overview of the potential role of neural network (connectionist) methodology in empirical accounting research. It highlights how the accounting task domain differs substantially from those for which neural network techniques were originally developed. A non-technical overview of neural network methodology is given, along with guidelines to help accounting researchers interested in applying these new tools to recognise the potential dangers and strengths underlying their use. An illustrative example is provided. The paper suggests research areas in accounting where neural network approaches could make a potential contribution. Explicit recommendations for prospective authors are made.

Proportionate Growth and the Theoretical Foundations of Financial Ratios

The article proposes a theoretical framework for understanding financial ratios, showing that the multiplicative character of the financial variables from which financial ratios are constructed is a necessary condition of valid ratio usage, not just an assumption supported by evidence. Also, by assuming that firm size is a measurable statistical effect, the article offers an informed reappraisal of the limitations of financial ratios, particularly the well–known limitation of proportionality. The article is divided into two parts, one where ratio components are viewed as deterministic vari– ables and the other where they are random. Such an approach allows the characteristics of ratios to be more easily understood before generalizing the relationship between ratio components to encompass randomness. In the second part, when variability introduced by firm size is treated as a random effect, it is shown that if the accounting variables Y and X used to calculate a financial ratio Y/X are exponential Brownian motion, and if continuous growth rates are equal and proportionate to firm size, this may lead to ratios which are asymmetric but which do not necessarily drift.

Incorporating complementary ratios in the analysis of financial statements

Abstract Ratios are routinely used for extracting information from accounting reports. However, ratios present only part of the information available and can be easily complemented provided that information-technology facilities are accessible. This study infers the functional form of the information discarded by ratios. Then it develops extensions of ratios incorporating the discarded information, showing examples of their use and discussing the benefits obtained. Extensions of ratios seem promising as a facility attached to computerized databases of accounting reports. The concepts developed here are a step toward a more technology-supported analysis of financial statements.

Attention, emotions and cause-related marketing effectiveness

Purpose – The purpose of this study is to explain how cognitive and emotional responses may influence decisions to purchase cause-related products. Design/methodology/approach – An experimental design clarifies how autonomic reactions determine altruistic choices in a simulated shopping environment. Eye-tracking and electrodermal response measurements were set to predict choices of hedonic vs utilitarian cause-related vs unrelated products. Findings – Emotional arousal, pleasure and attention to the cause-related bundle are associated with altruistic behaviour in hedonic choices. When facing utilitarian choices, customers focus on brand logo and donation amount while experiencing pleasure, but emotional arousal does not increase marketing effectiveness in this case. Research limitations/implications – The experiment may be replicated in the real-world shopping environment, but spurious influences will be difficult to control. Distracting cues such as background music and scents used to increase positive e...

A unified approach to the extraction of rules from artificial neural networks and support vector machines

Support Vector Machines (SVM) are believed to be as powerful as Artificial Neural Networks (ANN) in modeling complex problems while avoiding some of the drawbacks of the latter such as local minimae or reliance on architecture. However, a question that remains to be answered is whether SVM users may expect improvements in the interpretability of their models, namely by using rule extraction methods already available to ANN users. This study successfully applies the Orthogonal Search-based Rule Extraction algorithm (OSRE) to Support Vector Machines. The study evidences the portability of rules extracted using OSRE, showing that, in the case of SVM, extracted rules are as accurate and consistent as those from equivalent ANN models. Importantly, the study also shows that the OSRE method benefits from SVM specific characteristics, being able to extract less rules from SVM than from equivalent ANN models.

An empirical extension of Rock's IPO underpricing model to three distinct groups of investors

This article examines earned returns and allocation details of more than 200 new offerings (Initial Public Offering, IPO) from companies that went public in Hong Kong during the period 1988 to 1995. Three distinct groups of investors are identified, each exhibiting a particular type of returns pattern. Each pattern seems to correspond to a specific level of information. This finding is of particular interest as it shows the level of return that an investor can expect from IPO investments, also being an extension of previous studies where, following Rock (1986), two, not three, groups of investors are identified. This article also finds that expected returns from IPOs remain positive and highly significant after adjusting for the allocation bias. With the exception of the smallest application sizes, results are invariant to adjustments such as transaction costs and the risk-free rate of return.

Renal adenoma: identification of two histologic types.

In this paper the morphologic characteristics of renal adenoma are studied in 4 groups of kidneys from 500 unselected autopsies performed for legal reasons, 5,000 autopsies from the files of a general hospital, 16 kidneys with adenoma and without carcinoma, and 200 kidneys with carcinoma. All the tumors were classified by pattern, cytology and sclerosis of the renal nontumoral stroma. The relationship between tumor size, tumor histological type and sclerosis was evaluated by statistical analysis. The differences in frequency were significant between the parameters size and histological type (p < 0.003), namely for solitary adenomas (p < 0.001). The differences in histological type observed in multiple adenomas were not significant (p > 0.05). The differences were significant between the extent of sclerosis and histological type (p < 0.002) and tumor size (p < 0.009). This study identifies two distinct groups of adenomas according to size, histological type and extent of sclerosis, and suggests that small adenomas of the mixed tubulopapillary type and with basophilic cells have little or no malignant potential. Conversely, adenomas with a solid or papillary pattern and clear cells are potentially malignant.

Rule induction for financial modelling and model interpretation

The paper discusses the possibility of applying a specific rule induction algorithm, ID3 to various financial data analysis tasks. These tasks include not only model building but also interpreting the outputs of financial models. The algorithm is shown to have major drawbacks as a modelling tool, some of which carry over to the post processing task. The fact that financial variables are often measured on a ration scale also causes problems. The paper examines solutions to the key problems and provides the basis on which analysts can judge the suitability of the algorithm for their own applications.<<ETX>>

Reversal in the relative performance of state- and legal person-owned companies during the Chinese split share structure reform

The split share structure reform was started in 2005 with the object of re-designating state-related, nontradable shares into tradable shares. The article compares the two major forms of state ownership in China (direct or indirect ownership) showing that, close to the reform period, companies directly held by the state experience a significant increase in market performance relative to indirectly held companies. Results suggest that investors’ perception about the worth of these two forms of state ownership may have suffered a reversal, thus bringing to light value-related consequences ensuing from protective schemes usual in China and elsewhere. The article also addresses a recurrent pitfall relating to the use in empirical models of fractions of the same total and shows that U-shaped patterns found in the relationship between ownership and performance are transient rather than stable.

Discovering the optimal set of ratios to use in accounting-based models

Ratios are the prime tool of financial analysis. In predictive modelling tasks, however, the use of ratios raises difficulties, the most obvious being that, in a multivariate setting, there is no guarantee that the collection of ratios eventually selected as predictors will be optimal in any sense. Using, as starting-point, a formal characterisation of cross-sectional accounting numbers, the paper shows how the multilayer perceptron can be trained to create internal representations which are an optimal set of ratios for a given modelling task. Experiments suggest that, when such ratios are utilised as predictors in well-known modelling tasks, performance improves on that reported by the extant literature.