Bankruptcy case study

Would be possible to guide a business in such a way to control the factors that contribute to going out of the market?

Bankruptcy is a wide topic and the root causes of this phenomenon vary greatly. In this post, I analyse qualitative data of bankrupt and non-bankrupt companies.

Experts graded companies considering their perception of the company on specific factors. These factors are used here to assess the likeliness of a company going bankrupt. More details about the data are shown later.

My initial questions are:

1) Which factors analysed by experts are the biggest contributors to forecast a case of bankruptcy?

2) Is there any factor that does not contribute at all in foreseeing such cases?

3) How can I use any of the gathered insights to inform businesses to make better decisions?

The data

The data used was presented on Myoung and Ingoo 2003. There are 250 records of bankrupt and non-bankrupt companies along with its corresponding qualitative analysis. The analysis was done by experts working for a large Korean bank and cover a period from 2001 to 2002. The data set is made available by the University of California.

A sample of the file can be seen in table 1.

As stated before, each row represents a specific company, IR, MR, FF, CR, CO, OP are factors subjected to the evaluation of experts and can be evaluated as Positive, Negative or Average. The last row informs if the company is or not bankrupt. Bellow it is clarified what each of these factors represent.

Industry risk (IR): Is measured by the stability, growth , the degree of competition, and the overall conditions of the industry.

Management risk (MR): Is concerned with the efficiency and stability of management and organization structure. It is measured by the ability of management, the stability of top management, the stability of organization structure, management performance, and the feasibilities of the business plan.

Financial flexibility (FF): Mean the firm’s financing ability from direct and indirect financial market and other sources such as affiliates and the third parties.

Credibility (CR): Is concerned with the reputation of a company associated with credit history, the reliability of the information provided by the company, and the relationship with financial institutions.

Competitiveness (CO): Means the degree of competitive advantage determined by the market position and the capacity of core technology.

Operating risk (OP): Is the volatility and stability of procurement, the efficiency production, the stability of sales, and the efficiency of collection policy of accounts receivable.

Frequency

Understanding the distribution of these variables is key to get the first ideas about how the variables behaves. On picture 1 it is plotted the absolute and relative frequency of all variables involved.

What is noticeable is that not all factors are good indicators of whether or not the business is going out of the market. All negative evaluations for CO are assigned to bankrupt companies while negative evaluations for IR account only for half of the bad cases.

Competitiveness (CO) and financial flexibility (FF) seem to be the best variables to be used indicators.

Although the visual analysis can inform something about the data, I use Chi-Square analysis to confirm my hypothesis.

As confirmed by this analysis, CO and FF are the best variables but CR is pretty good as well.

Another point to investigate is how many positive evaluations would mean that the company if free from bankruptcy, or how many negative evaluations I need to be certain that the company is going to default. Picture 3 presents this information.

What the graph tells us is that companies need 3 positive evaluations to be sure it is running well. Again, I confirm my visual impressions with a statistical evidence.

The three kinds of evaluation are useful for separating bankrupt and non-bankrupt companies but negative evaluations are more accurate at predicting this kind of scenarios as shown in picture 4.

The rules behind the numbers

Besides understanding which variables or combination of variables are more important, it is interesting to understand which rules can be applied to predict it the company is going to or not to be bankrupt.

For this purpose, I applied a “decision tree” algorithm.

In the first situation (picture 5), I use the responses given by the experts for each one of the factors as inputs to the model, in the second situation (pictire 6)I use the number of Positive, Negative and Average responses as inputs. On both situations, the output is whether a company can be classified as bankrupt or not.

As one could already have assumed, CO is the big factor to be used to predict bankruptcy among these variables. Note also that the negative evaluations are strong predictors as well. It seems that if a company have a competitive advantage over its competitors the other factors seem way less important.

I have decided to exclude CO as input for the model and run the model again, just to see check how the other variables influence the results. The results can be seen on picture 7.

The results are quite interesting because they basically tell that if you are not financially flexible (FFN>=0.5) you should have a good credit score (CRP>0.5) to avoid bankruptcy. Intuitively we can think that a company that doesn’t have access to financial markets needs to be able to finance itself with banks for example, of course this is a big assumption and needs further investigation.

I tried to run the model using only IR as an input (Our weaker variable), but as expected it didn’t generate a decision tree because it was not able to be used to divide bankrupt and non-bankrupt companies.

What can be learnt from this?

Based on this sample we can start to think that having a competitive advantage is key and we could forget the other factors. This is probably misleading since the advantage may emerge from a well managed and stable company (Measured on the other factors). On the other hand, the data shows that is hardly possible that one will succeed without a competitive advantage.

Competitive advantage is a vague term, the paper uses three criteria to assess it: market position, the level of core capacities and differentiated strategy. These terms are more suitable for a self-assessment if a company is competitive or not. Comparing sales, market share, sales, could give enough information to decide if a company is competitive.

Thanks for reading!

Nelson