Intangible assets evaluation: The machine learning perspective

Highlights

• Different machine learning techniques for intangible assets evaluation are compared.

• Particularly, single classifiers, classifier ensembles and hybrid classifiers are compared.

• Hybrid classifiers by k-means+DT ensembles provide the best performance.

Abstract

The lack of regulations and disclosures regarding intangible capital has made it rather difficult for investors and creditors to evaluate a firm׳s intangible value before making the associated investment and loan decisions. This study represents an initial attempt to compare/contrast different types of machine learning techniques and identify the optimal prediction model for intangible assets. In addition, this paper shows that machine learning can be used effectively for the problem of intangible assets evaluation. To be specific, five classification algorithms are considered: decision trees (DT), artificial neural networks (ANN), naïve Bayes, support vector machines (SVM) and k-Nearest Neighbors (k-NN). Consequently, thirty prediction models are constructed for comparison, including five single classifiers, boosting and bagging based classifier ensembles, and the combination of k-means clustering, single classifiers and classifier ensembles. The experimental results show that prediction models combining k-means with boosting/bagging based classifier ensembles perform much better than the other methods in terms of prediction accuracy, ROC Curve, as well as Type I and II errors. In particular, while the best single classifier, k-NN provides 78.24% prediction accuracy, k-means+bagging based DT ensembles provide the best performance to predict intangible assets with a prediction accuracy of 91.60%, 96.40% of ROC Curve and 18.65% of Type I and 6.34% of II errors, respectively.

Introduction

In the knowledge-based economy era, companies have to pay attention to the capability and efficiency for the creation, expansion, and application of knowledge [31]. The primary method available for creating firm value has changed from traditionally physical production factors to intangible knowledge. Given this fact, a large part of a firm׳s value may be reflected in its intangible assets. However, financial reporting cannot adequately reflect intangible asset value, because of fewer regulations and less disclosure in the area of intangible capital and creates an information gap between insiders and outsiders [59]. In order to provide other useful information different from financial statements for investors or creditors to measure the firm׳s value in investment opportunities or loans, and also to assist them make more accurate decisions more effectively, it is important to build a more effective and reliable intangible assets value evaluation or prediction model.

Related studies in many business domains have shown that machine learning techniques, such as neural networks and support vector machines, are superior to traditional statistical methods. They can be used to discover interesting patterns or relationships from a given dataset and predict or classify new unknown instances [13], [28], [7]. Therefore, the aim of this study is to examine the performances of single classification, classifier ensembles, and hybrid classifiers techniques in terms of intangible assets prediction. Specifically, five well-known classification techniques, including multilayer perceptron (MLP) [26], decision trees (DT) [47], naïve Bayes [4], support vector machines (SVM) [58], and k-Nearest Neighbor (k-NN) [14] methods are employed to develop the prediction models.

Since economic development has become more dependent on intangible assets, it is important to investigate the role of intangible assets in terms of a firm׳s market-based value. For this reason, the contribution of this study is two-fold. Firstly, this research represents an initial attempt to employ several different data mining techniques including single classifier, hybrid classifiers, and classifier ensembles to identify the optimal prediction model for evaluating intangible assets. From the technical point of view, the performance of these techniques has not been fully assessed in the domain of intangible assets evaluation. Otherwise, many prior researches have examined the impact of different factors in developed countries [19], [36], but they did not consider the extent to which the degree of various factors affects intangible assets in emerging countries. As [39] indicate that the capital markets are less developed and ownership concentration is higher in the emerging market, these researches for firms headquartered in emerging countries could offer different results for the relationship between various factors and intangible assets. Emerging markets are attention of the whole world since their current high growth rates and potentials for the future. Taiwan is one of those emerging countries with its fast growing economy. Thus, this study extends the prior work by examining the impact of different factors on intangible assets in the emerging market context.

Secondly, in practice, since fewer regulations and less disclosure of intangible capital is the norm, financial reporting cannot actually be utilized to reflect the value of intangible assets or recognize intangibles. The issue of recognizing intangible assets is the subject of active debate in the current literature from the conventional statistical point-of-view [3], [29]. This study applies data mining technology and demonstrates that it can be used effectively to evaluate intangible assets as well as provide additional information which is not ever disclosed in financial statements. Due to the lack of regulations about recognizing intangible assets, outsiders cannot use financial statements alone to accurately assess the market-based value of a firm prior to making decisions for new investment in general, or evaluate initial public offerings (IPO) firms in particular. This study will help investors and creditors to better evaluate the potential of new investment or lending opportunities, and also assist them to make the relevant decisions more precisely.

The remainder of this paper is organized as follows. Section 2 reviews related studies about intangible assets. Section 3 describes the experimental methodology, and experimental results are presented in Section 4. Finally, in Section 5 some conclusions are offered and their implications.