Financial ratios as a powerful instrument to predict insolvency; a study using boosting algorithms in Colombian firms/Indicadores financieros como instrumento poderoso para predecir la insolvencia; un estudio usando el algoritmo boosting en empresas colombianas/Indicadores financeiros como poderoso instrumento para prever insolvencia. Um estudo usando o algoritmo boosting em empresas colombianas.
| Autor | Correa-Mejia, Diego Andres |
| Cargo | Research article texto en ingles |
-
Introduction
Insolvent companies and their creditors are affected when they enter the insolvency process. Effective insolvency prediction is relevant for creditors to make appropriate decisions and in order to reduce credit risk (Liang, Lu, Tsai, & Shih, 2016). In Colombia, insolvency is part of the bankruptcy system and is regulated by law 1116 (2006). This law allows companies which are having financial problems more time to pay outstanding sums, as well as the possibility to refinance their debts. Wilches (2008) remarks that the regulation was introduced with the aim of keeping the economy working and avoiding mass layoffs. Nevertheless, companies going into insolvency directly affect the financial performance of their customers, creditors, suppliers and investors. As a result, the financial situation of every company asking for credit or investment should be evaluated (Fonseca, 2007).
According to Amendola, Giordano, Parrella and Restaino (2017) financial ratios provide relevant information that can help to define whether companies are likely to incur bankruptcy or other financial problems. By evaluating liquidity, profitability and debt (Lopez & Sanz, 2015) firms can be classified as potential borrowers. Lartey, Antwi and Boadi (2013) indicate that through liquidity analysis, creditors and suppliers are able to determine whether a specific company has the capacity to pay its debts on time. Furthermore, since profitability should converge with liquidity as part of a firm's broader financial health (Nissim & Penman, 2003), profitability analysis provides the tools to evaluate firms' efficiency and capacity to sustain their financial results in the future. Moreover, debt levels show the level of support from owners. Yazdanfar and Ohman (2015) have shown that there is an inverse relationship between investors' participation in capital structure and credit risk. The less involved investors are in the capital structure of a company, the higher the level of credit risk.
This study aims to fill the gap in the literature identified by Amendola et al. (2017). On one hand, statisticians have focused their efforts on developing prediction models, but they usually develop these models with several financial ratios and do not define bankruptcy according to the relevant regulations. On the other hand, although financiers have defined the most accurate ratios to evaluate companies' financial performance while using legal definitions of bankruptcy, they usually employ traditional methodologies in their predictions. Since traditional methodologies assume the presence of symmetrical datasets (Calabrese & Osmetti, 2013), these researchers are required to bias the sample in order to reach accurate predictions.
Through a sample of 11,812 companies during the period 2012-2016, of which 99.5% were non-insolvent and 0.5% were insolvent firms, the objective of this study is to predict insolvency for Colombian firms one, two and three years beforehand through financial ratios, while keeping the original sample structure. The prediction was developed using a boosting algorithm proposed by Freund and Schapire (1997). According to Le et al. (2018), this algorithm allows researchers to make predictions in imbalanced data sets, as is the case for insolvent and non-insolvent companies. In addition, a study carried out by Kim, Kang and Bae (2015) showed that results using boosting an algorithm are generalizable at different imbalance rates.
This study contributes to the literature because unlike many studies; insolvency legislation, financial analysis and sample characteristics were considered when making the prediction. The experimental results also prove that boosting algorithm has an advantage over traditional methodologies for predicting insolvency in imbalanced data sets. The results show, in agreement with Du Jardin (2015), that predictions are less accurate when models are estimated with more years of anticipation. However, the results from using the algorithm show that it is an effective tool for evaluating insolvency risk (Kim et al., 2015) in real conditions for Colombian firms. This study offers important information for investors, suppliers, bankers, and governments. With the proposed model, organizations can reduce their credit risk and avoid running into losses.
The rest of the article is organized as follows: in section 2, a literature review is presented alongside a description of insolvency legislation in Colombia and a characterization of financial ratios. In section 3, the method and the sample of the research are described. In section 4, the results of the prediction are presented and analyzed, and in section 5 the conclusion is given.
-
Literature review
This section provides a literature review of insolvency prediction, taking into account Colombian regulations and previous studies that have predicted bankruptcy using financial ratios and boosting algorithms.
2.1. Insolvency legislation in Colombia
In Colombia, law 1116 (2006) regulated the bankruptcy system and created two stages. On the one hand, there are companies which close their operations definitively due to a decision of the owner(s) or the authorities (Mora, 2014). This stage is called judicial liquidation. On the other hand, there are companies which are in non-payment due to financial difficulties (Ochoa, Toro, Betancur, & Correa, 2009). This stage is called restructuring, but it is also known as insolvency. Companies can enter restructuring for two reasons: either they default, or they are unable to pay their obligations (Ley 1116, 2006). Figure 1 shows the bankruptcy system in Colombia.
The insolvency stage was created with the purpose of avoiding patrimonial liquidation of companies (Wilches, 2008). Insolvent firms continue to receive support from their creditors (Rodriguez, 2008) to avoid any interruption in their normal operations. Forecasting insolvency is important not only for banks (Hernandez & Wilson, 2013), but also for creditors in general, as it would be inappropriate for them to provide credit to a company that will be unlikely to repay it (Ben, 2017). Furthermore, when companies enter restructuring, their creditors are obligated to continue providing them with credit (Fonseca, 2007). The only obligation for companies in restructuring is that they cannot stop paying their new debts from the moment it begins (Wilches, 2009). In other words, debts incurred before companies enter insolvency can be renegotiated or they can pause the payment of these debts for up to eight months (Rodriguez, 2008).
Judicial liquidation indicates that a company will cease to operate (Nishihara & Shibata, 2016). In this case, the assets of the company are sold (Romero, Melgarejo & Vera, 2015) in order to pay their liabilities according to the law. The main difference between restructuring and judicial liquidation (Rodriguez, 2008) is that companies that begin restructuring have financial problems, and this situation can be evaluated through financial statements. On the other hand, companies can enter judicial liquidation for different reasons (Romero et al., 2015) which are not necessarily related to financial statements.
This study is focused on predicting insolvency as our principal information sources are financial statements. Furthermore, restructuring has a stronger financial effect on creditors than judicial liquidation because, according to Wilches (2008), when companies enter insolvency, their creditor cannot recover their accounts receivable immediately, this situation affects their cash flow and in some cases it can affect their operations as well (Bauer & Agarwal, 2014).
2.2. Insolvency prediction using financial ratios
Financial statements provide relevant information (Amendola et al., 2017) related to companies' investments, finances and dividend decisions (Cultrera & Bredart, 2016). This information is usually the principal source for evaluating the main financial objective, which is firm value maximization (Ng & Rezaee, 2015). Nevertheless, due to bankruptcy being the opposite situation of firm value maximization (Bauer & Agarwal, 2014), information provided by financial statements can help to warn of impending insolvency. According to Altman (1968), financial ratios emerge from the relations between variables in financial statements, and one of their most useful applications is to measure the credit risk (Mongrut, Fuenzalida, Alberti, & Akamine, 2011) of a specific firm.
Financial ratios have been used since early studies carried out by Beaver (1966) and Altman (1968) to predict bankruptcy. According to Hernandez and Wilson (2013) bankruptcy researches have been focused on developing the best statistical model to predict it using financial ratios. However, Amendola et al. (2017) affirm that some subjects are not explored enough in bankruptcy prediction and an appropriate selection of financial ratios (Wang, Ma, & Yang, 2014) is necessary to provide accurate predictions.
Several categories of financial ratios have been used in financial literature. Beaver (1966) classified them in six groups: cash-flow, net-income, debt to total-asset, liquid-asset to total asset, liquid-asset to current debt, and turnover ratios. Later, Liang et al. (2016) divided financial ratios into nine categories, based on Beaver (1966), but with the addition of three new categories in accordance with the work of Fedorova, Gilenko and Dovzhenko...
Para continuar leyendo
Solicita tu pruebaCOPYRIGHT GALE, Cengage Learning. All rights reserved.
