A Statistical Approach to Receivables Risk

by Helen Sanders, Editor

The evolution in the role of treasury is a perennial feature in industry conference programmes and seminars. For a number of years now, we have seen a gradual expansion of treasurers’ responsibility as they influence the elements that contribute to the financial supply chain, such as payables, trade finance and supply chain finance. In some cases, such as payables, the actual payment processing may take place in a separate shared service centre, but treasury takes ownership of issues such as bank relationships, bank connectivity and payment timing as part of a wider working capital responsibility. However, the area that is more difficult for treasurers to influence is collections, which is typically closer to the business and often less centralised. Achieving oversight and influence over collections at a strategic level, if not necessarily taking full operational responsibility, potentially has a far greater impact on the treasurer’s ability to manage working capital than payables. This article looks at some of the most recent developments in credit and collections management, specifically risk-based collections, and how these techniques contribute to an effective liquidity and working capital strategy.

Receivables metrics

The trade receivables portfolio is, for many companies, the first or second largest asset on their balance sheets and in the same way as any other asset, it makes sense to take care of it. Collections delinquency has a major impact on companies’ ability to forecast cash and manage working capital effectively. Uncertainty about when a customer will pay an invoice results in treasurers having to create large working capital ‘cushions’ which sit on the balance sheet providing little value to the company. Companies have developed sophisticated models to determine the probability of customer payment based on customer segmentation and historical data which they use to calculate the size of these cash cushions. In addition, many organisations have developed efficient collection procedures with automated payment reminders and dedicated collections professionals calling customers to remind them to pay. Despite the automation and sophistication of these techniques, being proactive in chasing payment is only one part of a credible collections management strategy. Assigning probability of non-payment for cash flow forecasting is also important, but neither technique contributes to reducing the incidence of late payment.

Invoice ageing is typically the metric used most frequently to determine receivables risk, with specific actions such as chasing invoices or referring amounts to a collection agency. However, using ageing alone as a way of measuring and addressing collections is effectively closing the stable door after the horse has bolted. Far better is to allocate credit limits, and decide on appropriate collection processes based on intelligence about customer behaviour. Indeed, it seems ludicrous that pre- and post-sale processes are often so poorly aligned. For example, having a reliable understanding of the customer risk profile can inform the way that the sales team work with a customer, and how they prioritise their focus on different customer groups. At the other end of the process, automated, timely reconciliation and posting of collections is vital to free up credit limits for lower risk customers to enable the company to do more business with them.

As companies start to centralise their collections and standardise their use of technology, they are increasingly able to leverage the benefits of proactive credit scoring techniques.

Anticipating late payment

Treasurers and finance managers should therefore be looking at how to anticipate and prevent late payment. This does not mean taking a draconian attitude to payment terms, but simply to set payment terms according to the risk associated with each customer and their behaviour. To do this requires a proactive and regular approach to credit scoring the receivables portfolio to adjust the way that customer credit terms are determined, and how invoices are followed up. However, according to a recent study undertaken by SunGard, while 34% of the study participants reported using their receivables as part of their overall capital structure, 46% never credit score their portfolio, and only 19% are performing monthly risk analysis. This has major implications on how a company is evaluating one of the largest assets on the balance sheet, and the decisions that are made accordingly.

But what approaches could a company take to credit scoring? Essentially, there are three general types. Bureau scoring, that uses bureau data and allocates statistical scores, is typically used for evaluating credit risk to new customers. Judgmental or rules-based scoring models combine internal and bureau data in a scorecard in order to make decisions on new accounts and credit lines. The scorecard variables, weights and score ranges are subjectively determined, with a ranking of customers as opposed to quantifiable risk. Finally, statistical portfolio or collection scoring models take into account internal A/R and performance data but can also include bureau or other external data, with definable scorecard variables, risk categories and weightings.[[[PAGE]]]

Challenges to receivables modelling

There are undoubtedly challenges that prevent companies from employing modelling and specifically statistical modelling, related to organisation, technology and data:

Organisation - While many companies have successfully centralised their payments and cash management functions, credit and collections are typically more commercially sensitive and closer to the business. Consequently, with collection departments and processes often decentralised, it is difficult to apply modelling on a consistent basis. Increasingly, however, we are seeing companies centralise these areas more, in order to improve collection metrics across the business, establish greater visibility at a group level, enhance working capital and improve credit risk management. For example, a company may deal with the same company in different regions and across different business lines. It’s impossible to identify total risk to a customer, and set workable credit limits, without central visibility over risk and control over credit limits utilisation.

Technology - A second, related challenge is technology. With different tools used across different regions and business lines, companies find it difficult to model credit risk on a consistent basis. Centralisation can obviously play an important part in addressing this, by standardising the credit and collections technology used by a central credit and collections function. In companies where credit and collections are managed at a regional or even in-country level, however, the use of a common technology platform is an important way of creating central visibility and common processes that facilitate effective risk modelling, even if the physical functions are not centralised.

Data - A third challenge relates to the data required for effective modelling, such as credit reference data. While this information may be readily available in Western Europe and North America, this is not necessarily the case in other regions, particularly countries of growing importance such as China and other countries in Asia. Incongruities, fragmentation and patchiness of data can make it very difficult to establish a consistent approach to risk modelling.

However, as companies start to centralise their collections and standardise their use of technology, they are increasingly able to leverage the benefits of proactive credit scoring techniques, using tools such as Predictive Metrics. In the experience of one senior credit professional,

“Each month we provide a file to Predictive Metrics, which is very similar to the trade files that we were already providing to a credit bureau. The file is then scored based on our set criteria. The scored file is then uploaded into our ERP system and from there it is updated in our receivables management system, AvantGard GetPaid.   

“The scores are mapped to risk categories that we also have defined. These are used to determine when accounts are placed on hold for past due balances and over credit limits. The higher the risk category, the tighter the hold criteria.  

“We then use the risk categories to drive collections priority. The higher risk accounts are actioned earlier in the collections cycle and period between actions is shorter. The lower risk accounts won’t go into a collectors queue until much later with longer periods between actions. We also use auto mass faxing for our lower risk accounts.”

The result of this is lower DSO, better customer intelligence and significantly reduced risk. Resources can be prioritised appropriately, and by predicting customer delinquency to influence customer behaviour, future delinquency can be reduced.

Statistical modelling differs from other types of credit scoring as one credit professional explains,

“Statistic-based credit scoring models are designed to predict the inherent risk of your customer, including the probability that the customer will become seriously delinquent, go to write-off or file for bankruptcy at some point in the future (usually within six months from the scoring date).

“Statistical models ‘quantify risk’ by indicating the odds or probability of the delinquency occurring is, allowing you to determine the value of your credit risk in dollar terms. Companies can benefit from leveraging pre-built models that are used across billions of dollars of AR portfolios in multiple industries and geographies. These models include a pre-validation process and compare the output with actual historical results in order to reach an acceptable level of confidence. It was this validation process that convinced us that the statistical models were better than relying solely on credit bureau data.

“We use this risk level derived from this model in conjunction with invoice ageing as the primary driver for order management and determining a collection strategy that is appropriate to the customer. This brings greater efficiency and effectiveness to the collection process, substantially reduces the cost of collections and optimises resources by focusing their efforts on priority accounts.”

[[[PAGE]]]

The impact of this approach has been considerable, with a 20% reduction in the number of orders referred to credit management, more proactive collection management for higher risk accounts and a major reduction in delinquency over 90 days.

Corportates are looking at how they can generate the highest return on their receivables portfolio.

Receivables as financing collateral

A related issue to proactive credit scoring and collection practices is how receivables can be used as financing collateral. With bank credit lines shorter and often more stretched, companies often prefer to use these borrowing facilities to fund strategic investment rather than financing working capital. In this environment, alternative financing becomes a very attractive means of sourcing funds without affecting the company’s capital ratios, credit ratings or bank lines. While there are a variety of different alternative financing mechanisms, these all rely on an asset as collateral. The more certain the receivables, the higher the proportion of the portfolio that can be financed. Furthermore, the more rigorous a company’s credit and collection processes, the more confidence the financier has and therefore the bigger the pool of assets that can be financed, and at a lower cost, particularly when combined with the use of credit insurance. As we saw from Copap (TMI issue 190, November 2010) credit insurance can be a vital component of an effective risk management strategy and also be valuable in determining the proportion of invoices that a bank is prepared to finance. Alternative financing techniques have historically been used by smaller companies, but as larger companies seek to diversify their funding sources, receivables financing is becoming increasingly important.

Conclusion

To summarise, therefore, trade receivables, classified as short-term assets on the balance sheet, are amongst the largest assets held by a company, so it makes sense to protect them, and then to use these assets to advantage, whatever the size of company. Corporates are therefore looking at how they can generate the highest return on their receivables portfolio. The difficulty is that companies often have customers with very different profiles. For example, a potential financier will want intelligence on customers’ ability to pay, credit risk, industry, location, open sales orders, remaining credit limit, and ability to use alternative methods of payment. Perhaps the single biggest factor in determining how to best use a specific receivable is the probability of payment.

Assigning a probability of late payment or of loss that take into account all these other factors provides an organisation with the ability firstly, to apply appropriate collection strategies and payment terms and secondly, to segment customer groups and identify which can be used to generate the most liquidity at the least cost. To do this requires considerable data and intelligence, based on the right organisation, technology and data, but the cost and strategic benefit of optimising these areas can be substantial.