Wehrspohn Risk Management

How do you assess credit risk from receivables with a Monte Carlo simulation?

Case study

By Dr. Uwe Wehrspohn.

Industrial, trading and service companies also carry credit risk as soon as they deliver on invoice, and they often hold thousands of open receivables. Using a receivables portfolio in Excel, we show how to assess receivable defaults with a Monte Carlo simulation, how a correlation of defaults through the economy changes the result and what receivables management with debt collection achieves after a default.

Credit risk is a cross-cutting risk for all companies. Below we show how to model a portfolio of 1,000 receivables in Excel and assess it with a Monte Carlo simulation. We calculate the portfolio three times: first with independent defaults, then with a correlation of defaults through the economy and finally with receivables management that recovers part of the receivable after a default.

Why does credit risk affect every company?

The word credit risk first makes us think of banks. That is natural, because the business model of banks is to lend large sums of money to complete strangers. It is only natural that something sometimes goes wrong and such an amount is not repaid. That is credit risk.

Interestingly, industrial, trading and service companies are in a similar situation. They, too, may provide goods or services to their customers before these have paid. They may ask small customers and private customers for prepayment, but this is usually not possible with larger customers. These expect the delivery to come first and the invoice to be paid afterwards. This creates the possibility that the company delivers and in the end is not paid. So it also has a credit risk.

In addition, a company does not have just a single credit risk. As a rule, thousands, possibly even hundreds of thousands or millions of such receivables are open. So we have to model not only the individual credit risk but also a large number of credit risks, which we then combine.

What information do you need for each receivable?

Figure 1 shows the receivables portfolio in Excel. From row 10 onwards, each row holds one receivable; in the example there are 1,000 of them. Three pieces of information are available for each receivable. Column A contains a plain ID that identifies the receivable and does not actually affect the risk at all. Column B contains the receivable amount, that is, the amount at stake. This is a very simple figure, because the amount is printed on the invoice and is therefore known from the outset. The first receivable in row 10, for example, amounts to 1,512 euros.

Column C contains the probability of default, which the next section is about. Column D “Ausfall?” (default?) is still empty, and column E “Verlust” (loss) shows zeros. In column D we will shortly simulate whether a receivable defaults, and in column E the resulting loss. The yellow field in cell E6 will later hold the portfolio loss, that is, the sum of the losses of all receivables.

Figure 1: The receivables portfolio in Excel. Each receivable has an ID in column A, a receivable amount in column B and a probability of default (PD) in column C. The columns “Ausfall?” (default?) and “Verlust” (loss) are filled in the course of the article (German interface).

How high is the probability of default of a receivable?

The uncertainty of the receivable lies in column C with the header PD. PD stands for probability of default. It is a much more difficult figure than the receivable amount, because we need an assessment of the probability with which the receivable will not be paid.

We cannot simply make up this probability. That is why there are firms that specialise in estimating probabilities of default for a large number of companies and customers. They offer these estimates as a service to companies that want to manage their receivables. Examples are Creditreform, Schufa or Experian. For large companies there are also rating agencies that give the companies an external rating. Such a rating is also associated with a probability of default.

A one-year PD for receivables with short terms

Probabilities of default are usually standardised to one year. The PD of 1.1% for the first receivable therefore means that the customer defaults with this probability within the next twelve months. The term of a receivable, however, is usually much shorter. Typical terms are 30 days after receipt of the invoice, perhaps 45 or 60 days, but not a year. That would be very unusual. The probability of default of the actual receivable is therefore much smaller. It also depends on the payment term actually agreed and on whether that term is exceeded somewhat.

In practice, however, this is less of a problem. Most companies do not necessarily want to know what losses the receivables portfolio existing today faces over its term. Rather, they want to know what receivables risk they have over a whole year. Over the year, individual receivables are paid and disappear from the portfolio. At the same time, however, new invoices are sent out, so that new receivables are added to the portfolio.

It is like a lake into which water flows on one side and out of which water flows on the other. The lake is constantly in motion, but the water level always stays roughly the same. So we can say: here is a receivable of 1,512 euros that defaults with 1.1%. If it is paid, another, similar receivable newly enters the portfolio. With today’s state of the portfolio, a snapshot so to speak, we can therefore actually make a representative statement about the whole year.

Banks use exactly the same approach, because the same thing happens there. A loan is repaid month by month, and interest is paid. One deal or another expires, and the portfolio shrinks. At the same time, however, the bank keeps granting new loans, so that the portfolio fills up again and remains just as stable over time as the receivables portfolio.

How do you simulate the default of a single receivable?

In short: Risk Kit is an add-in for Microsoft Excel that lets you build and analyse Monte Carlo simulations directly in the spreadsheet. What we show here applies in general, though: assessing credit and receivables risks with a Monte Carlo simulation works on the same principle in any simulation tool.

The default of a receivable is a 0-1 event: either the receivable is paid or it defaults. This is practically a coin toss, though not with a 50% probability for heads and tails but with the probability of default, so 1.1% for the first receivable. This representation is associated with the name of Jakob Bernoulli. In Risk Kit we use the Bernoulli distribution for it.

More on this: Our article “When do I use a Bernoulli distribution in a Monte Carlo simulation?” explains how this distribution describes events with exactly two outcomes, such as whether a receivable defaults or not.

The Bernoulli distribution is found on the Risk Kit toolbar under the button “Beliebt” (popular). Clicking on it opens the dialog in Figure 2. Below the selection “Funktion” (function), which is set to “Zufallszahl” (random number), it has two input fields. The important one is the probability, the parameter P, and that is exactly our PD. There we refer to cell C10 with the probability of default of the first receivable. To the right of the field, the dialog shows its value, 0.0112813767, that is, about 1.1%. We do not need the other field, “Name”, at the moment.

Figure 2: The dialog of the Bernoulli distribution for cell D10. The probability P is the PD of the first receivable from cell C10 (German interface).

After inserting, cell D10 contains a worksheet function that refers to the probability of default:

=Bernoulli_variate(C10)

Copying the function to all receivables

With a large portfolio, we could proceed in exactly the same way for every single receivable. In this example we would then have to open the dialog a thousand times, select the function, set its parameters and confirm. That is of course very laborious. We therefore simply copy the cell function down with Excel. You can work with all Risk Kit functions in exactly the same way as with the standard Excel functions: you can copy them or type them directly into the cell. Double-clicking the small square at the bottom right of the selected cell copies the formula down to the end of the filled column.

Column D now contains many zeros for receivables where the risk did not occur. Here and there, however, it also contains ones for risks that actually materialised, where the receivable defaulted.

How do you calculate the loss of the receivables portfolio?

The column “Ausfall?” (default?) gives us a default indicator. It is not yet our default risk, however. For that we need to know what loss we suffer when a default occurs. Here the loss is the receivable that was at stake. Its size therefore depends on which receivable actually defaults.

Figure 3 shows column E with the loss, in which we evaluate the indicator. If the default indicator has the value 1, the loss is the receivable amount. If it has the value 0, the loss is also 0. In the figure, for example, the receivable with ID 10 in row 19 has defaulted: column D shows a 1 and column E the loss of 3,369 euros, the full receivable amount from column B. The formula of cell E10 is shown at the top in the formula bar:

=+WENN(D10=1;B10;0)+OutputName("Forderung "&A10)+OutputUnit("EUR")

Figure 3: Column E holds the loss per receivable, and the formula of cell E10 is shown at the top. On the right, the dialog of the “Eingabe” (input) button is open with the output name and the unit EUR. In row 19, the receivable with ID 10 has defaulted (German interface).

WENN is the German name of Excel’s IF function. We added the functions OutputName and OutputUnit at the end of the formula with the button “Eingabe” (input) on the Risk Kit toolbar. It marks the loss cell as an input cell. In the accompanying dialog, shown on the right of Figure 3, we enter an output name, here “Forderung” (receivable) together with the ID from cell A10, and EUR as the unit.

The portfolio loss as the target figure

Once we have calculated the loss for every single receivable, we can form the sum. It is our target figure, the portfolio loss of all receivables together. It is in cell E6 below the heading “Portfolioverlust” (portfolio loss) in cell E5:

=+SUMME(E10:E1009)+OutputName(E5)+OutputUnit("EUR")

SUMME is the German name of Excel’s SUM function. Here, too, we find the functions OutputName and OutputUnit, this time added with the button “Ausgabe” (output). This button marks the cell as an output cell, and on that occasion we give it the output name “Portfolioverlust” and the unit.

Starting the simulation

The model is now complete. All input data are in place, we have a model for the default of the individual receivable, we have calculated the loss from it, and we have modelled the sum of all losses as the portfolio loss. Now we can evaluate the model. We set up the simulation with the button “Konfiguration” (configuration). The most important figure is the number of simulation runs, here 10,000, and as the simulation mode we choose “Aktive Arbeitsmappe” (active workbook). Clicking “Simulation” starts the simulation, which then runs in the background. In each of the 10,000 runs, Risk Kit draws anew for every receivable whether it defaults and calculates the portfolio loss from that.

How large is the default risk of the whole portfolio?

After the simulation, the result statistics shown in Figure 4 appear to the right of the table. The middle column contains amounts in euros, the right column “Relativ zum Exposure” (relative to exposure) the same amounts as a share of the total exposure. Below them is the histogram of the loss distribution.

Figure 4: The results with independent defaults. At the top are the statistics of the portfolio loss, in absolute terms and relative to the exposure, and below them the histogram of the loss distribution (German interface).

Let us first look at the total exposure. Exposure is, so to speak, the receivable amount that is on the line. The sum of all receivable amounts is 6,403,211 euros, about 6.4 million euros.

The probability that a loss occurs in the portfolio as a whole is shown in the row “Ws von Verlusten” (probability of losses) and is 100%. The individual receivable may have a small probability of default, such as 1.1%, 0.3% or 6.2% like the first three receivables in Figure 1. But that something goes wrong with 1,000 such receivables is certain or almost certain. In this simulation it happened in every run.

The expected loss, that is, the mean over all simulation runs, is 167,409 euros or 2.6% of the total exposure. Below it are the smallest simulated loss (minimum), the value at risk (VaR) at various confidence levels and the largest simulated loss (maximum). The VaR is a quantile of the loss distribution. The VaR at 99%, for example, is the loss that is not exceeded in 99% of the simulation runs, here 459,178 euros.

The histogram shows the shape of the distribution. The horizontal axis shows the portfolio loss in euros. The vertical axis shows the density, that is, how often losses of this size occurred. Its numbers are so small because the density refers to a single euro. Very small losses have a very small probability, down to zero. Then the probability rises, and most of the time the loss lies between 0 and 320,000 euros. Above that, the losses continue thinly up to the maximum of 743,811 euros, the largest result of this simulation.

How does the economy affect receivable defaults?

An important property of this simulation is that we have represented all defaults as independent of each other. One may well think that this is not a realistic assumption. Companies naturally depend on the macroeconomic environment they operate in, that is, on the economy. If the economy in which a company is based gets into a crisis, does not grow or even shrinks, and costs rise, then the money to pay receivables is lacking. The probability that companies do not pay their receivables then rises across the board. This creates a dependence between the defaults.

Conversely, things may also go very well for everyone. When the economy is booming and there is plenty of money around, a receivable default is no longer really a big issue. We add this dependence in the next step.

How do you correlate the defaults of receivables?

So that we can see later what effect the dependence has, we first save the results so far. To do this, we select “Ergebnisse speichern” (save results) under “Simulationsergebnisse” (simulation results). Here we save only the model outputs and name them “Unabhängig” (independent). When we click “Exportieren” (export), Risk Kit adds another worksheet in Excel that contains the simulated values from all simulation runs.

Now let us assume we want to introduce a correlation of 20% between the credit risks. A positive correlation means that the receivables tend to default together. In bad times many receivables then default, in good times few.

Correlations of this order of magnitude are also common in banking. Banking regulation prescribes them for banks when they calculate their capital requirements, in Europe in the Capital Requirements Regulation (CRR).

Risk Kit has a dedicated function for the correlation. Figure 5 shows where to find it. It is under the button “Weitere” (more) in the submenu “Mehrdimensionale Verteilungen” (multidimensional distributions). The distribution is called multidimensional because we now no longer look only at the individual receivable but at all receivables together. So we have 1,000 dimensions. The function we need is called NormalCopulaWithDistributions. On the left of the screenshot, the correlation of 20% is already in cell D6 under the heading “Korrelation” (correlation).

Figure 5: The function NormalCopulaWithDistributions is found under “Weitere” (more) and “Mehrdimensionale Verteilungen” (multidimensional distributions). On the left, cell D6 holds the correlation of 20% (German interface).

The copula separates dependence and distribution

The name of the function is very complicated. It refers to the fact that in a multidimensional distribution you can always separate the dependencies between the individual dimensions from the distribution of the individual dimensions. The distributions of the individual dimensions are also called marginal distributions; here they are the Bernoulli distributions of the defaults. The part that describes only the dependence is called the copula. If we want to use correlations, the normal copula is the method of choice, because it is parameterised precisely by correlations. This is the “Normale Copula mit Randverteilungen” (normal copula with marginal distributions), as the tooltip in Figure 5 says.

Setting the parameters of the function

The function first needs the correlation matrix, here the 20%. Second, it needs the distributions, namely the name of the distribution as a new parameter. We therefore insert a column, give it the heading “Verteilung Risikofaktor” (distribution of risk factor) and enter the name “Bernoulli” for every receivable, because that is the name of the distribution for all individual risks. The new column D moves the other columns to the right. The correlation is now in cell E6, the column “Ausfall?” (default?) is now column E and the column “Verlust” (loss) column F.

Figure 6 shows the completed dialog for cell E10. The individual fields mean the following:

  • The field CorrelationMatrix contains E6, the correlation of 20%. In the background, Risk Kit assumes that every single correlation coefficient has the same value of 20%. Strictly speaking, this field takes a correlation matrix of 1,000 by 1,000 numbers, that is, one million numbers. Because they all have the same value, this one number is enough. If we wanted different correlations between the individual receivables, we would have to enter the large matrix.
  • The field Distributions contains D10:D1009, the vector of distributions with “Bernoulli” in every position.
  • The field Parameters contains C10:C1009, the vector of parameters with the probabilities of default.
  • In the field ResultInRow we specify whether we want the results in a row. We do not, which is why it says falsch (false).
  • In the field IsRowWise we specify whether the parameters that belong to a risk factor are in one row. That is the case here, because the PD and the name of the distribution are in the same row. That is why it says wahr (true).

Figure 6: The completed dialog of the function NormalCopulaWithDistributions for cell E10. On the left is the new column D “Verteilung Risikofaktor” with the name “Bernoulli” for every receivable (German interface).

This completes the parameters of the function, and we can insert it. Cell E10 then contains:

=NormalCopulaWithDistributions(E6;D10:D1009;C10:C1009;FALSCH;WAHR)

FALSCH and WAHR are the German FALSE and TRUE. Excel now returns the whole vector as a single result. We delete the old Bernoulli formulas in the cells below. The one function call then fills the whole column automatically, and here too some positions show a one for a default. Then we run the simulation again and see how the result develops.

What does correlation change in the result?

Figure 7 shows the new statistics. The total exposure has naturally not changed. It is not random at all but simply the sum of the receivables.

Figure 7: The results with a correlation of 20%. The histogram is cut off on the right at just over 1.6 million euros because the camera image covers that area in the video (German interface).

The probability that losses occur is no longer 100% but only 98.4%. The correlation between receivable defaults has increased the probability that nothing happens. With a probability of 1.6%, the economy in the country is so good that not a single receivable in the portfolio defaults. Accordingly, the minimum is now 0 euros. The distribution thus reaches further to the left than before.

The high losses at the end of the table show that the same has happened on the right side of the distribution. Before, the largest loss was 743,811 euros. Now the VaR at 99.9% is 1,509,993 euros or 23.6% of the total exposure, and the maximum is 1,805,293 euros, which is about 28% of the total exposure.

This is also reflected in the histogram, which now looks completely different. The hill that used to lie around the expected value has been pulled apart towards the edges. On the left there are now many more very small losses, and the tallest bar is on the far left at zero. On the right there are very high losses, much higher than before.

Independence and correlation side by side

To compare before and after directly, we also save this result via “Simulationsergebnisse” (simulation results) and “Ergebnisse speichern” (save results), this time under the name “Korrelation” (correlation), and export it. We get a second worksheet with results. We can import both results again by clicking “Ergebnisse laden” (load results) and “aus Excel-Tabelle laden” (load from Excel sheet). Then they appear directly side by side.

The expected portfolio loss under independence and under dependence is relatively close, at 167,409 euros and 163,187 euros. The difference between the two means is a pure Monte Carlo effect, that is, a random fluctuation of the simulation. The correlation does not change the expected value, only the shape of the distribution around it.

This becomes immediately clear from the range, that is, the distance between the smallest and the largest simulated loss. It has risen from about 720,000 euros (743,811 minus 23,790 euros) to about 1.8 million euros and is thus two and a half times as large. The value zero now occurs too, and in return there are very large results at the high quantiles.

Figure 8 shows both distributions on top of each other as a histogram, orange for independence and blue for correlation. Where the two overlap, the area appears dark blue. Under independence we had a hill around the expected value. With correlation, the distribution is polarised: it has a lot of weight on the left at very small losses and reaches out on the right to very large losses.

Figure 8: The loss distribution under independence (orange) and under correlation (blue) compared. The chart is cut off on the right at just under 1.5 million euros because the camera image covers that area in the video (German interface).

This is exactly the effect of dependencies: they polarise. That is paradoxical, because with dependencies the probability of small losses also rises. Dependencies therefore do not automatically make losses only larger; first of all they also make them considerably smaller. The phases in which little or nothing happens become longer, although the crises are worse when they do occur.

What can you do about receivable defaults?

In these calculations we have not done any risk management yet. So far we have only modelled the gross risk, so to speak, that is, the risk before any kind of measures. This is quite necessary, because for the German auditing standard IDW PS 340, too, you model the gross risk. But of course we do not want to be helpless in this situation; we want to reduce the receivables risk.

Prepayment shifts the risk

We have already mentioned one solution: we could ask for prepayment. This is an active measure and completely safe. If someone has already paid, the risk shifts from the supplier to the buyer. The buyer may then have the problem that the goods do not arrive after paying for them. If you can insist on prepayment, it works.

Precisely because of this situation, however, large customers are usually not willing to order only against prepayment. They assume that, because of the lasting business relationship, the company is willing to tolerate the receivables risk and deliver on invoice. These receivables therefore remain at risk for us.

Debt collection recovers part of the receivable

How can we deal with this? A classic option is debt collection. You do not simply give up when the customer says that they cannot pay the receivable. Instead, you commission a collection company, such as a law firm or another specialised company, to collect the receivable after all. We then assume that the collection company manages to recover at least parts of the receivable, perhaps not the whole receivable, but at least something. This recovered part is called the recovery. This is exactly the approach we now pursue.

How do you model the recovery after a default?

For this we insert two more columns, the recovery rate in column E and the recovery vola in column F. The recovery rate is the share of the defaulted receivable that is still collected. The recovery vola describes the uncertainty of this recovery rate, that is, its volatility. As a result, the columns “Ausfall?” (default?) and “Verlust” (loss) move to columns G and H.

Where the values come from

Like the PD, the recovery rate is not an easy number to determine. We would ask the collection company about its experience: in what order of magnitude are you typically able to still collect amounts when you take care of it? The company will answer something like this: within the first twelve months after a receivable defaults, we typically still manage to collect this or that share of the receivable amount. Typical values lie between 30 and 40%. Here we assume 30%.

The associated volatility is typically in the order of 15 to 20%. We assume 15%. These, too, are essentially measured values from experience. If you have been managing receivables yourself for some time, you can also evaluate your own recovery experience and determine these numbers from it. So these are not expert estimates plucked out of thin air. We assume that the values are the same for all receivables and copy the parameters down.

The beta distribution for the recovery rate

Which distribution can we use for this? A distribution that is very well suited is the beta distribution. It takes values between 0 and 1 and is therefore ideal for describing a share, that is, a percentage of something. Besides the standard parameterisation, Risk Kit offers a second parameterisation that is based precisely on the expected value and the standard deviation with which we describe the recovery rate here. This is the function Beta2_variate.

Figure 9 shows the dialog. Under “Parameter”, “Erwartung + Standardabw.” (expectation + standard deviation) is selected. In the field Expectation we refer to the recovery rate of 30% in cell E10, in the field Std to the recovery vola of 15% in cell F10. To the right, the dialog shows a preview of what the distribution looks like: the density as a chart and a table of statistics. The recovery rate will essentially lie between 0 and 80%, with the expected value at 30%. In the table you can see this from the expectation of 0.30 and the 99.9% quantile of 0.80.

Figure 9: The beta distribution for the recovery rate is parameterised by the expected value of 30% from cell E10 and the standard deviation of 15% from cell F10. On the left are the new columns E and F (German interface).

The loss after recovery

With this we can weight the receivable. The loss in the event of default is no longer the full receivable but only the part that was not recovered. We therefore multiply the receivable by the share that is lost. This share is exactly 1 minus the recovery rate. In technical terms, it is called the loss given default (LGD). The formula for the loss in cell H10 now reads:

=+WENN(G10=1;B10*(1-Beta2_variate(E10;F10));0)+OutputName("Forderung "&A10)+OutputUnit("EUR")

Beta2_variate draws its own random recovery rate for each default. We copy the formula down to all receivables. With this we have generalised the model to take debt collection into account, and we can run a new simulation on this basis.

What does receivables management achieve after a default?

Figure 10 shows the statistics of the simulation with recovery rates. The probability of losses has stayed practically the same, 98.6% compared with 98.4% before. The small difference is again a Monte Carlo effect. The recovery rate does not affect the mere occurrence of a default at all, and the receivables default just as before. Receivables management only comes in afterwards and determines what we lose once the default has occurred.

Figure 10: The results with correlation and receivables management, that is, with a recovery rate of 30% and a recovery vola of 15% (German interface).

The distribution has a similarly polarised shape as before, with very small and also very large losses. The minimum is again 0 euros. But the VaR at 99.9% has fallen considerably, from 1,509,993 euros or 23.6% of the exposure to 1,080,467 euros or 16.9%. This is because we have actively managed our receivables.

The maximum, on the other hand, has not changed all that much: it is 1,782,867 euros compared with 1,805,293 euros before. This is because two very bad events can come together here. Large losses arise when we have the bad luck that very many receivables in the portfolio default and, in addition, we can recover little on precisely these defaults.

The distribution therefore has a similarly long tail as before, that is, a similarly long right edge with rare, large losses. But this tail becomes much thinner, which means that the extreme events occur less often. This is an essential effect of receivables management.

The second effect is that the expected loss has changed considerably. Before, it was in the order of 160,000 to 170,000 euros; now it is only 116,181 euros. That is about 30% less. In principle, we can therefore expect to be considerably better off with this receivables management.

With and without receivables management compared

To take a closer look, we also export these simulation results, this time under the name “Forderungsmanagement” (receivables management). A third worksheet is created. After importing again, we can compare the unmanaged portfolio under correlation with the managed one.

Figure 11 shows the two distributions as a histogram, orange without and blue with receivables management. Where the two overlap, the area again appears dark blue. On the left side the blue bars are higher, so small losses occur more often with receivables management. On the right side, the tail has become considerably thinner. There we mainly see orange, that is, the portfolio loss without measures. The blue bars are still there, but they have become rarer.

Figure 11: The loss distribution under correlation without receivables management (orange) and with receivables management (blue). The chart is cut off on the right at just under 1.5 million euros because the camera image covers that area in the video (German interface).

The effect shows very nicely in the distribution function in Figure 12. For each loss amount on the horizontal axis, it shows the probability that the portfolio loss does not exceed this amount. The blue curve with receivables management lies clearly to the left of the orange curve of the unmanaged portfolio at every height, that is, for every confidence level. At a height of 0.80, for example, the blue curve is at just under 190,000 euros and the orange one at just over 260,000 euros. These are the VaR values at 80% from Figure 10 and Figure 7, namely 187,825 euros and 264,618 euros.

Figure 12: The distribution function of the portfolio loss without receivables management (orange) and with receivables management (blue). At every height, the blue curve lies to the left of the orange one. This chart, too, is cut off on the right at just under 1.5 million euros (German interface).

What do the three assessments show together?

We have now assessed the portfolio three times. The first model made the simplifying assumption that all receivables default independently of each other. We corrected this and included macroeconomic effects through a correlation. In doing so, we followed an approach that banks also use. This gave us the gross risk without measures, that is, the risk of the portfolio if we do nothing about it.

After that, we actively pursued debt collection and receivables management: when a receivable defaults, we hand it over to a specialised company that tries to recover as much as possible for us. We have seen exactly what effect this has. From this we can also judge whether our receivables management works, whether it pays off financially and whether we need to think about further measures.

This was an introduction to the topic of receivables management. As a next step, the distribution of the portfolio loss, into which we have condensed the individual losses here, could be transferred into enterprise risk management. It can then also be properly taken into account in company-wide risk aggregation.

More on this: Our case study “How does risk aggregation work under IDW PS 340 and StaRUG?” shows how to aggregate risks across the company and compare them with the risk-bearing capacity.