Introduction The main goal of a business is usually improvement of its value and profit. To achieve this, the business deals with its customers to sell them goods or services. However, the business process may vary in different companies. In some cases, the payment or partial payment is made in advance before the goods are delivered (typically in case of B2C deals). In B2B deals the goods are usually delivered and then an invoice is issued with a specified due date. In that case, a problem may arise if the delivering company fails to receive the payment because the other party of the deal got into financial troubles and went bankrupt. For the delivering company the situation is very unpleasant because it has neither the goods nor the money. Moreover, if the company diversified the business risks inappropriately and delivered large quantities of goods or services without receiving any payment it may negatively influence its financial stability or lead to secondary insolvency. This situation was quite common in the Czech Republic in 1990s. Bankruptcy and solvency models and their information capacity One of the methods to avoid non-payment of invoices and thus potential problems is to select customers who are expected to be sufficiently stable and not likely to go bankrupt in foreseeable future. There are several methods to predict future standing of companies, e. g. financial analysis, pyramid of ratio analysis, bankruptcy models, etc. Each of the mentioned methods has its pros and contras. The main shortcoming of financial analysis and pyramid of ratio analysis is that the resulting data are difficult to interpret. On the other hand, they provide an overall view of the company and, consequently, if the analysis is performed by a professional their reliability is very high. Despite the fact that bankruptcy and solvency models fail to provide a comprehensive view of a company because their result is usually only one number, their results are easy to interpret even for a person who is not a professional in the field. Selection of businesses suitable for B2B can be based on both the above-mentioned methods, but the former ones are very time-consuming and highly dependent on whether the business has an expert to perform them. For this reason, I will outline the options for selection and other parameters for business management based on bankruptcy and solvency models. Solvency index Solvency index (IB) is one of the most widely used solvency models in German speaking countries . The index was developed based on multivariate discriminant analysis and it consists of six ratio indicators. The resulting value of the index categorizes businesses into solvent ones and those endangered by bankruptcy. If the index value is positive then the business is solvent. If the result is negative then the business is very likely at risk of bankruptcy. Some references  provide a more detailed categorization of index results. Values smaller than - 1 indicate extremely weak solvency and values greater than 2 indicate a highly stable business. The categories are shown in the tabl. 1 below. Table 1 Categories of results based on the solvency index (IB) [1, 2] Result Evaluation Business IB є ( -∞ ; -2) Extremely bad economic situation Bankrupt business IB є < -2 ; -1) Very bad economic situation Bankrupt business IB є < -1 ; 0) Bad economic situation Bankrupt business IB є < 0; 1) Problematic economic situation Solvent business IB є < 1; 2) Good economic situation Solvent business IB є < 2; 3) Very good economic situation Solvent business IB є < 3; ∞) Extremely good economic situation Solvent business Kralicek´s quick test Kralicek´s quick test differs from the other models by its simple calculation. There is no complicated formula, whose result would indicate categories, but relative indicators are categorized separately, their results are averaged and the result is again categorized. Weights of the individual categories are provided in the tabl. 2 below. Table 2 Weights of the individual indicators [1, 2] Evaluation Indicator Interval Grade Evaluation of revenues situation Equity quota. Equity/Assets (30 %; ∞) 1 (20 %; 30 %> 2 (10 %; 20 %> 3 (0; 10 %> 4 (-∞ ; 0) 5 Time of debt repayment from cash flow. Payables/Operating cash flow (-∞ ; 3 years) 1 <3 years; 5 years) 2 <5 years; 12 years) 3 <12 years; 30 years) 4 <30 years; ∞) 5 Evaluation of financial stability Cash flow in revenues. Operating cash flow/revenues (10 %; ∞) 1 (8 %; 10 %> 2 (5 %; 8 %> 3 (0 %; 5 %> 4 (-∞; 0 %> 5 Return on assets (ROA). (Net profit + Interest paid x (1 - tax) / Revenues (15 %; ∞) 1 (12 %; 15 %> 2 (8 %; 12 %> 3 (0 %; 8 %> 4 (-∞; 0 %> 5 The resulting values can be classified into 3 categories: - <1; 2) - solvent business; - <2; 3) - grey zone; - (3; 5> - bankrupt business. Apart from the basic variant, also a modified variant has been developed. The latter does not compare the results of the indicators with previously specified values but with percentiles for the individual industries. Consequently, the calculation is significantly more demanding because, in addition to the calculation, it is also necessary to determine the percentiles for the industry. The limits of the individual categories are also different. Altman´s models Specialized literature usually describes four versions of this type of analysis  designed for: - joint-stock companies with publicly traded shares (the so-called Z - Score); - companies not publicly traded on financial markets (the so-called Z’ - Score or ZETA); - non-manufacturing companies (the so-called Z’’ - Score); - Czech companies. Big companies traded on the stock exchange usually have a high quality controlling management system and for this reason, i will focus on smaller companies [1-3]: Based on the resulting values, the businesses can be classified into the following categories: - Z’є < 2.9; ∞ ) - solvent business, - Z’є ( 1.23; 2.9 ) - grey zone, - Z’є ( - ∞; 1.23 > - bankrupt business. Altman reports that the resulting analysis will predict bankruptcy of a business in 91 % of cases and only in 3 % of cases, a business is classified as prospering, but actually goes bankrupt in the following period . Index IN The index has been developed for the local market in the Czech Republic and therefore it should have a high information capacity in the local environment. The index has gradually developed from the variant IN95 (95 means the year of its development), to variants IN99, IN01 and IN05. The variant IN95 was a bankruptcy model. The variant IN99 was a solvency model, which determined whether the monitored business creates value. The variants IN01 and IN05 have merged both the approaches and they reflect both the aspects . Information capacity of the last mentioned index IN05 calculated in the form : The result is further categorized based on the limits 0.9 and 1.6 (Fig. 1). If a business is classified under the lower limit of 0.9 the probability of its bankruptcy is 97 %. If the index is in the grey zone (between 0.9 and 1.6) the probability of bankruptcy is 50 %. If the index exceeds 1.6 then the probability of avoiding bankruptcy is 95 %. This information capacity was verified using 1526 industrial businesses . Fig. 1. Classification of businesses based on the IN05 index [4; Own interpretation] Shortcomings of the bankruptcy models Despite the above-mentioned information capacity of the individual models and simple interpretability of the results, there is a number of problematic issues. However, the objective of this article is not to address those individual problematic issues and therefore we will provide only a basic list to point to the fact that at present it is impossible to rely absolutely on the models and that they have to be incorporated onto a whole palette of other parameters. In real life, they may often include non-financial and informal indicators, such as a simple conversation with a sales representative of the other party, who may be complaining about late payment of his salary etc. The fundamental problems of bankruptcy models include : 1. Bipolar dependent variables. The models use simplification that cannot be achieved in real life. Even for a simple definition of business failure there may be different viewpoints, which subsequently lead to differences in the calculation and interpretability of the result. 2. Set of input data. Most models use non-random data, which means that during application in real life the result will be different from that, on which the model was developed. 3. Stationary character and instability. Models are usually developed retrospectively. When a model is used to make a forecast then the dependence between dependent and independent variables is different, which affects the results from the model. 4. Selection of independent variables. Most models select indicators, which are popular and frequently used. Those models, however, are often subject to "window dressing" (deceitful conduct in order to improve a given indicator in the accounting). 5. Annual financial reports. The reports are used for computation despite the fact that different countries use different methods for their drawing and many data may be missing, which are necessary for correct calculation. 6. Time dimension. Сlassical models ignore the fact that companies change in the course of individual years. Consequently, the calculation is always independent on time, which is in conflict with reality. Credit limit and its determination Despite the above-mentioned shortcomings, bankruptcy and solvency models can be used for practical business management thanks to the easy interpretability of the results. One of the ways of potential use of those models is selection of suitable companies to deal with and determination of the rules to decide, which companies will be allowed to pay invoices and which companies will be required to pay high advance payments or to provide other securing. A credit limit can be used for this purpose and it can be employed in several ways. Segmentation based on categories The business can identify individual segments of customers based on its own needs and subsequently specify overall credit limits. In this case, it is very important to make sure that the segments and credit limits meet the needs of the company. The limits and segments will be determined differently by the company, whose goods are in high demand, and the company is not able to meet the demand. In that case, the company will not probably deal with any customers endangered by bankruptcy. A different situation will be in the company struggling to find its place on the market. Such a company will probably need to take a greater risk to sell its goods. This differentiated approach may not affect the company as a whole but only its individual products. An example of limit determination is provided in the tabl. 3 below. Table 3 Possible categories of the segments Category Solvent Uncertain Risky Product A 200 thous 50 thous 30 % advance payment and max. 20 thous Product B 300 thous 50 thous 100 % advance payment Product C 300 thous 100 thous 50 thous Tabl. 3 shows that the demand is probably high for Product B and therefore the company can afford not to deal with risky customers at all or it may require 100 % advance payment for it. For Product C the situation is reverse. In case of segmentation the goods are delivered and an invoice is issued. If a customer orders more goods but failed to pay previous deliveries exceeding the credit limit then the orders are suspended until the debt is paid at least partly. The total amount of orders in progress should never exceed the credit limit. Matrix for frequent deliveries In case of very frequent and repeated deliveries, the monitoring of orders and related amounts in real time may be complicated and therefore the customers in the segments may be viewed in a matrix with credit limits and terms of payment. Customers are allowed to take bigger deliveries but the term of payment is shorter or the deliveries are smaller and the term of payment is longer. If you multiply potential deliveries per day with the payment time and the maximum order then you will get the maximum permitted credit limit expressed as the area of the rectangle (Fig. 2). Fig. 2. Matrix view of the credit limit - businesses within the same segment Fig. 2 shows a situation, in which businesses in the same segment have differently set-up parameters but their credit limit is identical (area of the rectangle). Fig. 3 shows a comparison of the individual segments, while the segment category A includes the most stable businesses with a minimum risk of bankruptcy and, on the other hand, the segment category C poses a high risk. C Fig. 3. Comparison of the individual segments If a company considers trading with customers of C type, with high risks of bankruptcy in the following year, and if the company is not willing to take the risk but at the same time does not want to lose the opportunity to sell its goods, then it can use additional securing. The securing should cover a potential loss, which should never exceed the amount of the credit limit (area of the square). Selection of a model to determine the segments Regardless of whether the business determines the individual segments for credit limits using a classical method or whether it uses a matrix to deal with frequent deliveries, it is necessary to have a specified procedure to classify individual businesses into the segments. As suggested above, it is possible to use various approaches from the classical financial analysis, to pyramid of ratio analyses to bankruptcy models. For reasons of simple interpretability and relatively good information capacity of the results, my recommendation to all businesses is to use bankruptcy and solvency models unless they have an expert available (financial analyst). If a business decides to specify segments and classify its customers into them based on bankruptcy or solvency models then the question remains about which of the models should be used, because there are many of them. Some of them have been used globally, e. g. the above-mentioned solvency index or Altman´s analysis. Other models are more local, such as the IN index developed by the Neumaiers. Since the local bankruptcy models are able to take into account various regional specific features, which the globally used models do not need to respect, I recommend the local models as long as reliable studies are available about their information capacity. IN05 is such an example of the local model for the Czech Republic, it was developed based on the analysis of 1 526 businesses, which is by an order of magnitude more than the number used for development of the Altman´s analysis. Moreover, the information capacity of the model is, as mentioned above, above 90 %. Therefore, the use of this index seems to be the most suitable variant in case of businesses based in the Czech Republic. The classification into categories should respect the limit values 0.9 and 1.6 that have been empirically tested. However, when establishing the segments a person should be appointed within the business, who shall be authorized to reassign companies from one segment to another under certain circumstances. This competence shall be naturally associated with responsibility for potential problems if a business is classified into a higher category and subsequently goes bankrupt. The competence may be used in various situations. 1. Shifting a customer into a lower category - provided other than accounting sources, e. g. internal sources, indicate financial problems of the customer, crisis of liquidity etc. The same applies in the case that other partners in business provide information that the customer has many overdue payables to them. 2. Shifting a customer into a higher category - this may occur in a situation if the business is aware of financial problems of the customer and considers its takeover and in that case the business may intentionally purchase overdue payables from the other suppliers. By assignment into a lower segment we can reduce the risk of non-payment, but we increase the risk that the goods or services will not be sold. Therefore, each manager will have to decide, which of the potential risks is more serious. It is also necessary to realize that the tax must be paid even if the money is not collected and that overdue receivables, recovered by litigation, may be gradually written off and thus reduce the profit, which will subsequently result in return of the paid tax. In this case, a secondary loss is incurred due to the time value of money and also other negative impacts on liquidity etc. Conclusions Systematic controlling is essential nearly for any business practicing B2B, in order to reduce the risk of secondary insolvency as a result of bankruptcy of a business partner. For this purpose, it is necessary to classify customers into categories based on probability of their bankruptcy or insolvency. Businesses may use a number of tools for this, while bankruptcy and solvency models can be recommended thanks to their quality results and simple application; the models should be ideally developed for local markets, subject to the condition of their high information capacity. When deciding about controlling the rules it is always essential to take into account the objectives of the business and view the results as inputs for a resolution and not as the resolution as such. In other words, if a calculation shows negative values it does not mean that the business should automatically stop dealing with the customer. On the other hand, even if the cooperation continues it is necessary to be aware of the risks and to manage them.