Using accounting‐based information on young firms to predict bankruptcy

• Author: Christian Lohmann,Thorsten Ohliger
• Date: 01 December 2019
• DOI: http://doi.org/10.1002/for.2586
• Published date: 01 December 2019

RESEARCH ARTICLE

Using accounting‐based information on young firms to

predict bankruptcy

Christian Lohmann

1

| Thorsten Ohliger

2

1

Junior Professorship in Managerial

Accounting and Control, University of

Wuppertal, Gaußstraße 20, 42119

Wuppertal, Germany

2

parcIT GmbH, Bayenwerft 12–14, 50678

Cologne, Germany

Correspondence

Christian Lohmann, Junior Professorship

in Managerial Accounting and Control,

University of Wuppertal, Gaußstraße 20,

42119 Wuppertal, Germany

Email: lohmann@wiwi.uni‐wuppertal.de

Abstract

This study analyzes the nonlinear relationships between accounting‐based key

performance indicators and the probability that the firm in question will

become bankrupt or not. The analysis focuses particularly on young firms

and examines whether these nonlinear relationships are affected by a firm's

age. The analysis of nonlinear relationships between various predictors of

bankruptcy and their interaction effects is based on a structured additive

regression model and on a comprehensive data set on German firms. The

results of this analysis provide empirical evidence that a firm's age has a con-

siderable effect on how accounting‐based key performance indicators can be

used to predict the likelihood that a firm will go bankrupt. More specifically,

the results show that there are differences between older firms and young firms

with respect to the nonlinear effects of the equity ratio, the return on assets,

and the sales growth on their probability of bankruptcy.

KEYWORDS

accounting‐based key performance indicator, bankruptcy prediction, firm age, nonlinear interaction

effect, structured additive regression model

1|INTRODUCTION

Accounting‐based information can be very useful for

predicting whether a firm will become bankrupt within

a specific period. Particularly in cases where there is no

readily available market‐based information on a firm, to

assess that firm as a going concern and predict the prob-

ability of its going bankrupt it is necessary to use

accounting‐based information instead.

The explanatory power of accounting‐based key per-

formance indicators that are obtained from a firm's

annual financial statements has been studied extensively,

including in the early literature (Martin, 1977; Ohlson,

1980). In addition to studying such indicators, researchers

have sought to develop empirical and statistical methods

and models for predicting as accurately as possible the

probability of a firm, or a debtor in general, going bank-

rupt (for an overview see e.g., Altman & Saunders, 1997;

Balcaen & Ooghe, 2006; Bellovary, Giacomino, & Akers,

2007; Dimitras, Zanahkis, & Zopounnidis, 1996; Scott,

1981). The present study contributes to this body of

research by analyzing the nonlinear interactions between

accounting‐based key performance indicators and a firm's

age and their effects on the probability of bankruptcy. For

that purpose, this study will apply a structured additive

regression model.

Several studies in the literature assume that there are

non‐monotonous relationships between the key

accounting‐based indicators of a firm's performance and

predicting that firm's probability of bankruptcy (Atiya,

2001; Erlenmaier, 2011; Saunders & Allen, 2010). These

assumptions are partially supported by a body of empiri-

cal evidence. Specifically, several studies have found that

when a generalized linear model (GLM) or a generalized

additive model (GAM) is applied, it is indeed possible to

detect nonlinear relationships between accounting‐based

Received: 30 November 2018 Revised: 14 February 2019 Accepted: 8 March 2019

DOI: 10.1002/for.2586

Journal of Forecasting. 2019;38:803–819. © 2019 John Wiley & Sons, Ltd.wileyonlinelibrary.com/journal/for 803

key performance indicators and the predictors that mea-

sure a firm's probability of going bankrupt. Escott,

Glormann, and Kocagil (2001), Falkenstein, Boral, and

Carty (2000), and Sobehart, Keenan, and Stein (2000)

have all provided such evidence with respect to a number

of factors; namely, a firm's finance structure (see also

Estrella, Park, & Peristiani, 2000), profitability (see also

Estrella et al., 2000; van Gestel et al., 2005), liquidity

(see also Serrano‐Cinca, 1997), turnover, growth (see also

Hayden, 2011), and size (see also Altman, Sabato, & Wil-

son, 2010). Statistically speaking, however, this empirical

evidence is rather weak, because the respective works

tend to apply univariate methods and analyze nonlinear

effects only with respect to quantiles of classified data.

Nevertheless, Lohmann and Ohliger (2017), who used a

GAM to estimate nonlinear relationships as accurately

as possible, have confirmed and described in more detail

the existence of nonlinear relationships between

accounting‐based key performance indicators and the

predictors used to predict whether a firm will go bank-

rupt or not.

The empirical methods and models available to

researchers can estimate the direct effects of accounting‐

based key performance indicators on predicting the prob-

ability of a firm becoming bankrupt. A GLM strictly

assumes a linear relationship between such an indicator

and the predictor; in contrast, a GAM can estimate more

accurately such relationships. Both models, however, rely

on the assumption that there are no statistically signifi-

cant interactions between the independent variables that

they use. If in practice the independent variables in ques-

tion do interact in a statistically significant manner, to

estimate accurately whether a firm will go bankrupt or

not, it is necessary to obtain detailed information on

those interactions.

The objective of this study is to analyze whether the

interactions between accounting‐based key performance

indicators and a firm's age affect the prediction of

whether a firm will become bankrupt or not and, if so,

to what extent. Research has shown that a firm's age, par-

ticularly in the case of young firms, determines to a con-

siderable extent that firm's growth (e.g., Fort,

Haltiwanger, Jarmin, & Miranda, 2013) and, therefore,

its accounting‐based key performance indicators. Our

starting point is that accounting‐based key performance

indicators affect either linearly or nonlinearly the predic-

tor that measures a firm's latent probability of bank-

ruptcy. On that basis, we hypothesize that a firm's age

affects the relationships between these factors. In particu-

lar, we expect that in the case of young firms the relation-

ship between accounting‐based key performance

indicators and a firm's probability of bankruptcy is more

pronounced.

In this paper we use a structured additive regression

model to analyze how a firm's age interacts with its

accounting‐based key performance indicators and how

this interaction affects the prediction of bankruptcy. A

number of studies have applied GAMs to examine credit-

worthiness (Alp et al., 2011; Burkhard & de Giorgi, 2006)

and bankruptcy prediction (Berg, 2007; Cheng, Chu, &

Hwang, 2010; Dakovic, Czado, & Berg, 2010; Hwang,

Cheng, & Lee, 2007; Lohmann & Ohliger, 2017, 2018).

However, these studies focus on comparing several

empirical models from a strictly statistical perspective

and do not analyze or describe existing interaction effects

between a firm's age and its accounting‐based key perfor-

mance indicators. A firm's age, as we explain below, is

nevertheless a potentially important factor when it comes

to assessing its creditworthiness and predicting its proba-

bility of going bankrupt.

Given that young companies and older companies are

at different stages of their life cycle, they are likely to dif-

fer with respect to their accounting‐based key perfor-

mance indicators and their rates of change. We expect

that this is particularly true for indicators that relate to

equity and firm growth. If we assume that a young firm's

opportunities for growth are much greater than those of

an older firm, we can expect to see this difference

reflected in the two firms' financial structures. Both the

signaling theory (Myers, 1977) and the pecking‐order

theory (Myers, 1984; Myers & Majluf, 1984) argue that

a younger company's opportunities for growth are associ-

ated with lower equity and higher debt. In contrast,

agency theory (Jensen & Meckling, 1976) argues that in

younger firms these opportunities are associated with

higher equity and lower debt. In either case, what mat-

ters from our point of view is that these differences

may also affect a firm's probability of bankruptcy. If this

hypothesis is correct, estimating this probability requires

that we take into account the interaction effects between

a firm's age and accounting‐based key performance

indicators.

To test whether a firm's age indeed affects the nonlin-

ear relationship between that firm's accounting‐based key

performance indicators and probability of bankruptcy, we

will apply structured additive regression models with a

two‐dimensional spline function that captures the inter-

action effects in question. To the best of our knowledge,

this is the first study to do so. Our study furthermore eval-

uates the validity of the structured additive regression

models that we apply. As our analysis is based exclusively

on a set of data on German companies, our work also

contributes to research on predicting the probability of

bankruptcy specifically in the case of German companies

(Anders & Szczesny, 1998; Kaiser & Szczesny, 2003;

Schuhmacher, 2006).

LOHMANN AND OHLIGER

804