Counterparty Risk

Counterparty Risk

Master Thesis, MSc BA Finance, University of Groningen

Management summary

Preface

Writing a thesis is like running marathon: the last phase is the hardest phase. During the last 12 km you know you will finish but it still takes a long time to get to the finish. The same goes for this thesis. The last phase took some time but I have finished this ‘marathon’. From here I would like to thank everybody within Company X Paris for their support during my internship. More specifically, I would like to thank Evert van Setten and Victor Helmond for their instructive feedback during my research. Also special thanks to Baptiste Langenstein for his support during my stay in Geneva. Moreover, I would like to thank Leo Hoogerheiden for his feedback on the English grammar. His comic feedback was a bright spot in the last phase. Finally, I owe special gratitude to Peter Smid for his helpful support from the department of finance of the faculty.

Rick Marsman

Contents Management summary ... 2 Preface ... 3 Contents ... 4 1 Introduction ... 6 2 Literature ... 10 2.1 Counterparty risk ... 10

2.2 Counterparty risk in commodity trading ... 12

2.2.1 Contract basics ... 12

2.2.2 Exposure basics... 13

2.2.3 Margin, MTM and credit in a contract life ... 14

3 Current Situation... 17

4 Data & Methodology ... 20

4.1 Methodology for analysis ... 20

4.2 Methodology for developing model and further calculations ... 24

5 Analysis ... 26

5.1 Analysis risk profile ... 27

5.1.1 Sugar market ... 27 5.1.2 Freight ... 27 5.1.3 Experience in Business ... 28 5.1.4 Quantities ... 28 5.1.5 Customer portfolio ... 29 5.1.6 Origination of buyers ... 30 5.1.7 To Sum up ... 31 5.2 Contract conditions ... 32

6 Model for setting trading limits ... 34

6.1 The new model ... 35

6.2.1 Deposit levels ... 40

6.2.2 Overlimits requests ... 40

6.3 Back testing the model ... 41

7 Conclusion, recommendation and suggestions for further research ... 42

Bibliography ... 44

Appendix ... 47

Appendix 1 Durations of shipment of main shipping routes in days ... 47

Appendix 2 Summary data white sugar spot returns ... 48

Appendix 3 Maximum one-day trade losses for certain security levels ... 48

Appendix 4 Statistical test ... 49

Appendix 5 Minimum deposit level for main shipping routes ... 51

1 Introduction

Company X is a successful family-owned food business with its roots in the Midwest of the US. In 1865 Company X started with a simple grain storage facility to become a prominent international player in the food industry. Company X is active in the trade, storage and transhipment, transport and industrial processing of primarily agricultural raw materials for the livestock feed and food industries with more than 100,000 employees in 43 countries (www.companyx.com , 29-10-07).

To adjust quickly and correctly to these changes, the BU started a new office in Paris in 2003: Container Unit (CU). This office is responsible for trading white sugar in containers with its origination in South and Central America, Asia and European Union, and various destinations around the globe, such as Africa, Middle East, Caribbean, and Balkans. The Geneva office still trades bulk sugar. CU has the intention to make profit on the service to move sugar from A to B and not on price movements of sugar. Thus they try to minimize commodity price risk exposure as much as possible. This means they try to hedge the full commodity price risk on their contracts, so that in an ideal situation every tonne of sugar is hedged with futures contracts. As the businesses of CU and the Geneva office are fundamentally different the counterparty risk is different as well. There are several specific customer characteristics for the container business that emphasise the change in counterparty risk. First of all, smaller quantities are sold with containers than with bulk. While e.g. Geneva sells on average more than 50,000 metric tonnes (MT) per counterparty annually, Paris sells on average only 5,500 MT. Secondly, CU’s customers have a higher risk profile. 70% of CU’s customers are distributors or small wholesalers. Since these types of companies resell sugar again, these companies could be more sensitive to changes in the sugar price than for example industrials. Thirdly, as mentioned CU is responsible for deliveries to destinations such as Africa, Middle East, Caribbean, and Balkans. Doing business in these regions is riskier than doing business in developed countries. Reasons for this are lack of information availability, different cultures, instable economies, etcetera. Additionally, although CU minimizes its commodity risk exposure, customers might not have any commodity risk mitigating policies. This makes it more plausible customers could default due to changes in sugar prices. Therefore trading with these customers makes the business riskier and strengthens the need for good counterparty risk management. Moreover, in such a starting business it is important to develop or change processes and procedures in ways they fit the business. Otherwise the probability of mistakes becomes too high.

volumes to one specific customer or selling on credit, could lead to higher profits but it increases the probability of counterparty default also. This leads to more volatile cash flows. Such volatile cash flows could lead to:

- a higher probability of forgoing investments (Myers and Majluf [1984]);

- higher costs of funding (Froot, Scharfstein and Stein [1993], Minton and Schrand [1999]);

- an increase of the expected costs of financial distress (Smith and Stulz [1985]); - an increase of the expected cost of taxes (Smith and Stulz [1985]).

Reducing the probability of those problems adds value. Moreover, although empirical studies are not unified whether the attitude of managers is related to the firm value or not (Tufano [1996], Géczy, Minton, and Schrand [1997], Fok et al. [1997]) risk management could add value as well by influencing the risk attitude of trading managers. Trading behaviour might not be in line with maximizing shareholders’ utility. Suppose a trading manager receives a basic income plus a profit related bonus. This means this manager is certain to have an income with only an upside potential. This manager might take more irresponsible risk to increase profit and thus increase his income. Furthermore, an asymmetry of information between risk manager and traders could increase the probability of this behaviour and make risk managers reactive. When traders do have more information about counterparties than risk managers, the risk management is highly dependent on the opinion of traders. With such information asymmetry a trader has the opportunity to influence the general judgment in such a way that he gets more flexible trading limits knowing that he takes more risk. Since more risk means more volatility and thus a higher probability of larger profits, the trader could utilize the upside potential of his income structure. This behavior could result in irresponsible risk for the company. Birnberg et al. [1983] classify several possibilities that traders could use to manipulate information in such situation:

- smoothing;

- biasing;

- focusing;

- filtering;

- ‘illegal’ acts.

customers. In the end of the process CU as well as the BU want to be proactive in managing the challenge to balance risk and profit. To get a well balanced risk the research objective is to develop a model to manage counterparty risk for CU in such a way that it:

- provides counterparty risk insight;

- avoids/ minimizes losses;

- limits risk to reasonable levels;

- maximizes trading opportunities.

Therefore the following main question needs to be answered:

Which model should CU use to measure counterparty risk?

To answer the main question the following sub questions need to be answered: What are the characteristics of CU’s counterparties/business?

What are the risks of business of CU?

Against which contract conditions that affect credit and mark-to-market risk are traders currently trading to mitigate risk?

What are the levels of risks that CU wants to take? What trading limits fit the business?

Note: All the above should fit in corporate risk policies

To get a good in-depth examination and detailed answers on these questions a case study design is used since the purpose of this study is to analyse specifically sugar trading in containers. CU is unique in the sugar trading business with shipping sugar bags in containers. Therefore, using data of other companies that do not only trade sugar in containers might impact the influence of shipping sugar in containers on counterparty risk. Another reason is that the data of other companies in the business were not available.

2 Literature

This section contains the background literature for this thesis. The first sub section describes counterparty risk. The second subsection explains the basics of counterparty risk in this type of business.

2.1 Counterparty risk

Counterparty risk comes from non-performance of a customer. The non-performance may arise from counterparty's refusal to oblige agreements due to price movements caused by systematic factors, or resulting from other political or legal constraints not anticipated by the principals [Oldfield & Santemero, 1997]. With only its downside risk, counterparty risk is an example of pure risk. From a statistical perspective counterparty risk is uncertain. The statistical perspective distinguishes three possibilities: certain, uncertain or impossible. The first one means that the probability of the event that the counterparty fails to perform its obligations is one. In case of uncertainty the probability is between one and zero. The latter one occurs when the probability is zero. It is clear that counterparty risk only appears in case of uncertainty. Sub-risks that influence this uncertainty and thus affect counterparty risk are:

- commodity risk: risk arising from changes in commodity prices;

- currency risk: risk arising from fluctuations in currency rates;

- country risk: risk arising when a counterparty may not be able to perform its contract due to political or economic developments;

- rating transition risk: changes in rating of counterparties.

value is relatively more exposed to changes in world market prices than that of customers with relatively more fixed assets. When such a company has taken a certain position on the market and the market moves to the opposite way as this company expects, it could lose its firm value and get into liquidity or solvency problems. This leaves no other option than to default on its contracts. As a consequence, companies with such limited assets increase the chances of default and thus increase counterparty risk. Therefore counterparty risk should be taken seriously in such cases. On the other hand, commodity trading with counterparties with many assets, e.g. confectionary or beverage producers, is more secure for several reasons. First of all, they buy commodities for own use as raw material. As a result, the market price influences the business not as much as it does for trading companies. This decreases the possibility that companies with much fixed assets will default. Furthermore, the relatively large portion of fixed assets makes the firm value less dependent on price changes in their resources. Therefore the probability of liquidity or solvency problems due to market movements is lower, which makes business more secure.

Moreover, it is important to know how market prices affect companies. Several studies prove that hedging (market) risk indeed adds value (e.g. Smith & Stulz [1985], Cassidy, Constand & Corbett [1990], Allayannis & Weston [2001], Carter, Rogers, & Simkins [2003], and Mackey & Moeller [2007])1. Thus, reducing market risk exposure reduces the probability of default of counterparties. Additionally, the origination of counterparties affects the possibility of using techniques like hedging. Several studies done for the Worldbank (e.g. Larson & Varangis [1996]; Larson, Varangis & Yabuki [1998]; Larson Anderson & Varangis [2004]) show that companies in developing countries do not have sufficient knowledge of good risk management to reduce their market risk exposure by hedging for example. Even if companies in developing countries do have the sufficient knowledge about risk management tools they still experience other difficulties in using these tools. Many companies in developing countries do business on a relatively small scale. This makes it difficult to use derivatives. Because they only need small quantities of derivatives, using these risk mitigating instruments is too costly for these companies. Also the accessibility of futures markets is more difficult for those companies. Compared to the Western world they do not have the facilities, such as the Internet, to get up-to-date information about changes in the market or to buy futures to hedge their risk exposure easily. Other characteristics that make it difficult for them to use risk-management tools are credit worthiness, lack of incentives for providers, or lack of correlation between local and international prices. With his lack of risk management tools

1

companies in developing countries are fully exposed to changes in market prices. This increases the probability of defaulting customers significantly.

Not only risk management characteristics of those countries do affect counterparty risk. Each country has characteristics that affect counterparty risk: country risk. There are four types of risk that affect country risk: political risk, commercial risk, macroeconomic risk and external risk (www.dnbcountryrisk.com, 19-12-07). The political risk includes factors that determine the internal and external safety situation, policy competency and other such factors that affect the business environment. Commercial risk encloses factors influencing the reliability of doing business in a country, such as the degree of corruption, sacredness of contracts, etcetera. Macroeconomic risk measures all macro economic factors that determine whether a country has the fundamentals for sustainable economic growth to increase investment opportunities. Examples are inflation rate, fiscal deficit, and money supply growth. Finally, the external risk measures factors, such as the current account balance, size of external debt capital flows, and foreign exchange reserves, that determine if a country can generate enough foreign exchange.

2.2 Counterparty risk in commodity trading

This section describes counterparty risk and mark-to-market (MTM) exposure. This subsection is based on internal documents of Company X. The first subsection gives the basics about contracts and counterparty risk and the second subsection describes the MTM and credit exposures during a contract life and how to mitigate it for one contract.

2.2.1 Contract basics

Regularly, traders sell sugar with standardized contracts. Contract terms used in the following sections that affect several risk exposures are:

- Contract date: the date when the contract is signed; - Quantity: the quantity sold to the customer;

- contract price: the price per MT;

- basic futures contract: the futures contract on which the basic price is based; - premium: the premium over the basic futures contract price paid for CU’s

service;

- pricing method:

o unpriced: only the premium is set on the contract date and the price of basic futures contract fluctuates till the moment the customer sets the price in an agreed pricing window;

- pricing window: the period in which the customer may set the contract price; - payment terms: the way of payment. They are described in the subsection 3.2; - shipment period: the period in which the sugar should be shipped from

origination to destination. It is possible to ship the contracted sugar with different ships with different arrival dates;

- origination: the location where the sugar is produced and shipped from; - destination: the destination of the contracted sugar.

2.2.2 Exposure basics

During the contract period, there are two possibilities in which counterparty risk could occur: 1) a failure to receive or deliver promised goods. This is defined as the mark-to-market

(MTM) risk;

2) a failure to receive money owed in full and on time. This is defined as credit risk. The essential difference between these two risks is that the first risk may occur in case of unpaid goods that are undelivered and the second type of risk in case of unpaid goods that are delivered. MTM-risk exposure occurs due to price movements that cause differences between contract and market prices. Depending on market movements, exposure arises from purchase and sales contracts. In case of a purchasing contract MTM-exposure arises when the market price is higher than the contracted price. Since the supplier is able to sell his sugar against a higher price to another customer he might not perform its contracted obligations. The probability of this event becomes more likely when the contracted price is lower than the real costs of producing the goods. The opposite is true for selling contracts. In case of selling MTM-exposure only arises when the market price is lower than the selling price. When prices drop with such an amount that the buyer cannot resell his goods without getting at least its costs back, the counterparty tries to skip the contract. In the end this could lead to counterparty failure. It follows that when the market price is above the contracted price this contract has no MTM-exposure. MTM-exposure for one contract can be calculated as follows: i c i c i c p MT MTM _, =∆ _, * _{, [2.1]}

Where MTM c,i is the MTM-exposure, ∆p_c,i is the change in market value per MT, and MTc,i

price of $350 per MT with a total amount of 1,000MT. Today the market value per MT is $345 of contract X. So the change in market value is $5 per MT. The MTM-exposure for contract X is $5 times 1,000MT. Thus $5,000. Furthermore, the total MTM-exposure for a customer is the sum of the MTM-exposures for each open contract:

cn cn c c c c MT p MT p MT p TMTM =∆ ₁* ₁+∆ ₂* ₂ +K+∆ * _[2.2]

Where TMTM is the total MTM-exposure with a customer, ∆_{pc1 till} ∆_{pcn are the changes in} market value per MT for each contract, and MTc1 till MTcn are the priced contracted tonnages.

For example, Customer Y has two open contracts with CU: contract X and contract Z. When MTM-exposure for contract X is $5000 and contract Z has a MTM-exposure of $2700 the total MTM-exposure for customer Y is $7700. Credit risk exposure occurs when goods are released and delivered but the counterparty did not pay the goods yet. This type of exposure can be caused by the same factors as MTM-risk exposure.

Credit risk exposure extends the risk exposure after the delivery of the goods. It brings along extra risks since the goods are already delivered. In such cases the goods are lost and the money is not received. In case of MTM-exposure the goods are still in possession of the seller. MTM-risk exposure and credit risk exposure depend on the payment terms used in contracts. This will be explained in the next subsection.

2.2.3 Margin, MTM and credit in a contract life

The contract price consists of a certain futures contract price plus a premium. Therefore the contract price is:

PR p p_c = _w +

[2.3]

Where pc is the contract price per MT, pw is the world price of the used basic future contract

per MT for pricing, and PR is the premium per MT for a contract. The premium is divided as follows: M Bogey Freight FOB PR= + + + [2.4]

are not known and therefore have to be estimated. The real prices for most of those costs (FOB and Freight) are only known on the moment the containers are booked at the shipping line. Depending on the forward position, which is the period between the contract date and the expected delivery date of the commodity (normally release date), there could be a long period between the contract date and the moment of booking the containers. In this period the amounts of those costs could change and affect the margin. When these costs have increased the margin will be lower since the premium has already been set. A longer duration of a contract increases chances of those costs. Furthermore, the margin is exposed to a possible default of the buyer. The period in which the margin is exposed to such possible default is called the margin-exposure. This is the period between the contract date and the date of delivery. Note that the margin-exposure starts from the contract date no matter if it is flat priced or not. In case a customer defaults on his contract before he sets the price, the seller loses the potential margin for this specific contract. The margin exposure occurs till the contract is fully paid (see figure 2.1). This exposure can be minimized by setting limitations for the total contracted tonnage and the forward positions of contracts.

Figure 2.1 Exposure on contract life

Source: Langenstein [2007a] (modified by author)

The margin-exposure for one contract can be calculated as follows:

i c i c i c M MT Mar_, = _, * _{, [2.5]}

Where Marc,i is margin-exposure, Mc,i is de margin, and MTc,i is the contracted tonnage. To

calculate the total margin exposure for one customer equation 2.6 can be used:

cn cn c c c c MT M MT M MT M TMar = ₁* ₁+ ₂* ₂ +K+ * _[2.6]

Invoice Release Payment

Where TMar is the total margin-exposure with customer Y, Mc1 till Mcn are the margin per MT

for each contract, and MTc1 till MTcn are the contracted tonnage for each contract.

MTM and credit exposure depend on several contract terms (see figure 2.1). First of all, the forward position affects the duration of MTM exposure window. Further, the way of pricing affects the MTM exposure. As mentioned there are two options for pricing: 1) pricing on contract date (flat pricing) or 2) pricing during a certain agreed period (pricing window). When contracts are flat priced MTM-exposure starts from the contract date, so the forward position is the same as the MTM-exposure. In case of pricing in a certain pricing window, the MTM-exposure window depends on wishes and characteristics of customers. When using a pricing window MTM-exposure starts from the pricing date. Finally, the payment terms could affect the MTM- and credit-exposure. Depending on the characteristics of the customers, traders can increase or decrease risk by using specific payment terms. The seller could use the following payment terms:

- cash against faxed documents. With this payment term the goods will only be released to counterparties when payments are received on the seller’s bank account. When the payments are received the documents are faxed to the customer. This payment term only covers credit exposure, using this payment term MTM-exposure will not be covered;

- cash against documents (CAD) via Bank. This payment term covers the same exposures as the former payment term. The only difference is that the seller sends the documents not directly with a fax but via its bank. In this case, when the seller’s bank receives the money from the buyer the seller’s bank sends the documents to the buyer’s bank and this bank hands them over to the buyer. This payment term gives the buyer some extra security that he really gets his goods; - cash against documents (CAD) in trust. This payment term is in fact some kind of

selling on credit. The goods are released before the payments are received. Normally the time between releasing sugar and receiving payments is 24 hours. This payment term does not covers any exposure at all;

- deposits/prepayments. To make sure counterparties will oblige its contract they are asked to give some deposit or prepayment that (partly) covers the contracted price. This payment term reduces MTM-exposure;

- CAD with a bank guarantee. This means that goods are sold CAD via bank with a

guarantee that the bank covers a certain amount in case the buyer defaults. Such a guarantee reduces MTM-exposure;

obligation. For example, when the seller is not able to deliver the goods at the agreed date the L/C is worthless and there is no guarantee anymore. This payment term covers MTM-exposure for 100% as long as the seller performs its contract. For the rest of this thesis cash against faxed documents and cash against documents via bank are put together as CAD, since these payments terms have the same MTM-exposure and no credit exposure. Table 2.1 shows the possible exposures for each payment term.

Table 2.1 Payment Terms and its Exposure

Payment term MTM-exposure Credit exposure Cash against faxed documents Yes No

CAD via Bank Yes No

CAD in Trust Yes Yes

CAD via bank with bank guarantee Partly/No No

Deposit Partly No

100% prepayment No No

L/C No No

Source: made by author

3 Current Situation

The former sections give the background for a model to set trading limits. With this background it is possible to check whether the current model fits the business of CU. Currently the CU is using the model with three status types:

- the occasional local buyer (OLB);

- mitigated risk business (MRB);

- rating by due diligence.

The MRB is meant as a status between the OLB and the due diligence and is recently implemented. It is created especially for CU. The main reason was that the outcome of the rating system that will be described below did not give trading limits that fit to the business of CU with most of their customers. For the MRB customers, new counterparty forms should be duly filled in. Further, traders have to ask for a copy of the certificate of incorporation. This is to be sure these companies could not affect Company X’s reputation. These customers are regular customers and will do business more times a year. Instead of 5,000MT per annum these customers may have a maximum open priced tonnage limit of 5,000MT at one time. The maximum sold quantity per annum is not limited (see table 3.1). This status maximizes the forward position to two months and the maximum unsecured pricing window is one week. 59 customers of the 134 customers did comply with this status in fiscal year 2006-2007 (FY06-07) (including customers that fit within the OLB trading limits).

Table 3.1 OLB and MRB explained

Occasional local buyer (OLB) Mitigated risk business (MRB) Max. 5,000MT per annum Max. 5,000MT contracted at any time

Max. one month forward position Max. two months forward position Max. one week unsecured pricing window Max. one week unsecured pricing window New c/p form with name and address of the

customer only

New c/p form duly filled in with copy of the certificate of incorporation

No analysis No analysis

Source: B. Langenstein [2007b]

The last status is based on a due diligence that distinguishes the customer type (trading house, distributor or industrial), net worth, and an analysis done on the characteristics of the customer based on:

- financials;

- business & market model;

- management quality;

- shareholders;

- structural risk factor;

- country risk;

- Company X’s historical trading experience with the customer.

trading limits does not fit the requested trading limits of traders with a specific customer. In that case traders can do a request for special limits (overlimit). Overlimits are currently based on the opinion of trading managers. They do not make calculations to measure extra contracted tonnages, MTM-exposure or a combination of both. The objective for the BU is that the model should fit for 90% of the customers. The current model fits for 78 customers (58%) for CU. Since literature suggests that risk management indeed adds value it is important to have a well fitting risk model to manage CU’s counterparty risk. Consequently, an analysis should be done on the trading limits to get a model that reaches the 90%-level. This new model should be in line with Company X counterparty risk policies. Company X has set up policies with a top-down approach (see figure 3.1).

Figure 3.1 Credit Policy Approach

Source: made by author

Three levels of policies are used. First, Company X uses a corporate credit policy set up by the financial risk committee (FRC) containing guidelines for accountabilities to ensure adequate consideration for risk, approval requirements and reporting requirements for lower levels. On the second level the corporate guidelines are specified to the platform business, in this case the Company X Agricultural Supply Chain (CASC). On the third level the policy is specified for the BU in detail and applied to its counterparties. The CASC platform demands that trades with each counterparty should have limits. These limits should bound contracted tonnage, open contract exposure on the current and stressed mark-to-market basis, and forward month duration of contracts. They make an exception for counterparties with a tonnage limit below 25k per annum. For this group it is up to the BU whether limits are set. These limits should be determined after doing a credit assessment on the counterparty.

Sugar Counterparty market & credit policy Credit Policy

CASC Platform Credit Requirements CASC Open contract

Requirements

- new contract policy - payment policy - financing policy

- letter of indemnity policy - procedures cargo release - trading ground rules Sub policies and procedures: Sub-policies and procedures:

FRC

Platform

Further, CASC demands that counterparties need to be segmented into credit quality tiers, which govern the size of limits granted, and stressed open contract limits should be linked to these tiers. On a BU level policies used for managing counterparty risk have the following primary goals:

- avoid/minimize losses due to collection on trades receivables or non-performance of contract obligations;

- avoid insolvent and unethical buyers/seller while maximizing profitable sales opportunities;

- ensure credit worthiness of customers;

- limit operations to reasonable and secured level;

- manage market and stress exposure, open positions, credit and trade receivables; - describe responsibilities and accountabilities.

4 Data & Methodology

Due to the uniqueness of trading in containers within Company X it is difficult to ‘copy-paste’ processes and procedures from other business units or sugar entities. Thus, to have a well fitting model to measure counterparty risk it is important to describe the risk profile of CU in more detail.

4.1 Methodology for analysis

position, the payment term and the historical volatility of the daily returns of closing white sugar spot prices traded on the LIFFE#5 in the period from the 1st of June 2004 till the 11st of January 2008. To measure the possible loss on contracts, assuming that sugar prices evolve according to a random walk and thus the prices of the sugar market cannot be predicted, the following equation as mentioned in Alexander [1996] can be used:

t VaR=

ασ

[4.1]

Where VaR is the maximum possible loss at certain confidence level, α _{is the one-tailed} confidence interval for a standardized normal distribution, σ _{is the standard deviation of} returns, and t is the horizon period in days. Transforming this equation in contract terms on the contract date, VaR is the maximum possible decrease of the contract value at a certain confidence level, α _{the one-tailed security interval for a standardized normal distribution,} σ the standard deviation of the daily returns of the white sugar spot closing prices (daily returns), and t is the forward position of a contract. Spot prices are used since most of the contracts are priced against the spot rate. The daily returns are calculated as follows:

1 1 − −

−

=

p

p

p

r

Where r is the daily return, pt is the closing spot price on day t, and pt-1is the closing spot

price on the working day before day t. According to this transformation equation 4.1 can be rewritten as:

F

M =

ασ

lucrative for the buyer to default on the contract. Using the payment term, deposit level, and forward position it is possible to measure the security level of the contracts:

F

D=

ασ

Where D is the deposit level as a % of the total contract value, α _{is the one-tailed security} interval for a standardised normal distribution, σ_{spot is the standard deviation of the daily} returns, and F is the forward position as the horizon period in days. For example, a customer bought sugar with a prepayment of 13% of the total contract value and the sugar will be delivered 25 working days after the contract date, i.e. the forward position is 25 days. With a historical σ_{spot is 0.0162} α _{is 1.64. In case of a standardised normal distribution this means} that it is 95% sure the market value of the contract will not be lower than the amount still to be paid by the customer. However, the daily returns are not normally distributed (see appendix two). Therefore, the ασ_{spot-part is inappropriate to use. To solve this problem and} assuming that sugar prices evolve according to a random walk and thus the prices of the sugar market cannot be predicted, all the daily returns are ranked from low to high to check which percentage of the historical daily return observations is higher than the deposit percentage. To compare the deposit percentage, it is calculated back to a daily return level using the following equation that is based on equation 4.4:

F D

s = [4.5]

Where D is the deposit level as a % of the total contract value, F is the forward position in working days of a contract, and s is the deposit calculated back to a one-day trade deposit. s replaces the ασ_{spot–part in equation 4.4. After that for each flat priced contract s is compared} with the historical daily returns to check which percentage of the observations is higher than

4.2 Methodology for developing model and further calculations

After the analysis, the results are used to answer the questions: ‘What are the levels of risks that CU wants to take?’ and ‘What trading limits fit the business?’. In a discussion with CU’s management appropriate trading limits are set to build a well fitting model. In this discussion the management weighs up the pros and cons for several decisions that need to be made to develop the model. Questions that need to be answered are:

- What forward positions are appropriate?

- What is the maximum quantity sold to customers?

- What is the security level for different customers? - Is a distinction between different customers useful?

- If yes, how to make this distinction?

After that, the model is developed that fits with the outcome of the discussion. However, the model needs supporting calculations to give exact contract conditions. First of all, traders need exact deposit levels that are needed to comply with the security level for the specific customer. These deposit levels can be calculated by rewriting equation 4.5:

F

s

D =

level s is 0.023 (see appendix three). Thus D is 18%, i.e. the trader should ask for a deposit that covers at least 18% of the total contract value.

Furthermore, as mentioned in section three the new model should fit for 90%. Thus, 10% of the customers need different trading limits than the model gives. Traders need to ask for an overlimit for these customers. Overlimits he could request for are longer forward positions, higher OPCT or a lower security level. The trading manager can be comfortable with a certain level of MTM exposure, in order to grant extra priced tonnage to a customer. When a trader wants to do business with a customer without MTM-exposure mitigating payment terms, i.e. the security level is 50%, he could ask for a MTM-exposure request. The trading manager could give an approval for a certain maximum exposure. This maximum MTM-exposure sets a boundary on the maximum open priced contracted tonnage. For calculating the maximum open priced contracted tonnage, the maximum MTM-exposure with a specific customer is derived, assuming no MTM-exposure mitigating payment terms are used. To calculate the OPCT-limit an equation is used that is founded on equation 4.1. In this case, where the white sugar spot returns are not normal distributed, VaR is the maximum MTM-exposure (MTMc,i) for a contract CU wants to have, N the forward position in working days

for a contract (Nc,i), and l, which is the maximum possible daily decrease of the daily returns

(calculated with a equation 4.2) at a certain security level, replaces the ασ_{-part. Furthermore,} to get hard figures instead %, the equation is multiplied with the market value of the contracted sugar. Thus, the equation is:

i c i c i c m s N MTM _, = _, * _, [4.7]

Where mc,i can be calculated as follows:

i c w i c p MT m _, = * _, [4.8]

Where pw is the world white sugar price, and MTc,i is the priced contracted tonnage for a

contract. So for example, a trader wants to sell to customer Y sugar three months forward (Nc,x

is 64) with no MTM-risk exposure mitigating payment terms. The trading manager agrees with these contract terms for this contract X under the condition that it is 95% sure that the maximum MTM-exposure (MTMy,x) for this customer is $500,000. The only way to bound the

maximum MTM-exposure is by limiting the contracted quantity (MTy,x), because the price

(pw) is given by the market. To calculate the maximum contracted quantity first the maximum

0.023 (see appendix three), and my,x will be $2,717,391. Now assuming that the sugar price is

$302, MTy,x is 8,998MT. Therefore, to be 95% confident that the maximum MTM-exposure is

$500,000 the maximum priced contracted tonnage is 8,998MT. However, the customers could have more open contracts with no MTM-exposure mitigating contract terms at one time. Therefore, the total open priced contracted quantity (OPCTc) should be limited. OPCTc is the

sum of all MTc,i with one customer. To compute OPCT c easier, the assumption is made that

all contracts have the same contract price and expire on the same date the total pw and Nc are

equal for all open contracts with a specific customer. So that equation 4.7 can be rewritten as:

c c w c p OPCT l N TMTM = * * [4.9]

Where TMTMc is the maximum total MTM-exposure of all contracts CU wants to have with a

certain customer, pw is the world sugar price, and OPCTc is the sum of all open priced

contracted tonnages with a customer, Nc is the forward position in working days for a

customer, and l is the maximum possible daily decrease of the daily returns, calculated with a equation 4.2, at a certain security level. Now using the average historical white sugar spot price the OPCT-limit is:

c c c

N

l

p

TMTM

OPCT

*

=

Where OPCTc is the open price contracted tonnage limit, TMTMc is the maximum

MTM-exposure CU wants to have for a certain customer, p is the historical average spot price, and l and Ncare defined as above. For example, the trading manager is confident with a specific

customer that buys sugar with no MTM-exposure mitigating payment terms, e.g. cash against faxed documents, and maximum two months forward. But he wants to be 95% confident that the maximum total MTM-exposure is $250,000. Assuming that the historical spot price is $302, the maximum open priced contracted tonnage is 5,519MT. In other words, this customer is allowed to price 5,519MT two months forward.

5 Analysis

5.1 Analysis risk profile

5.1.1 Sugar market

The current sugar market is as other (agricultural) commodity markets a volatile market. Last year’s volatility of the daily return white sugar spot closing prices was about 21%. The sugar market is mainly influenced by supply and demand imbalances. These imbalances find their origination in many things. First of all, the sugar market is one of the most protected commodity markets. Japan, EU, US, Indonesia, and Eastern European countries do all have their policies to protect their own sugar production. The World Trade Organization (WTO) is against such policies since they prevent developing countries to join the market. Therefore a couple of years ago, the WTO forced the EU to change their sugar protection policy to open the world market for developing countries. Since then the EU decreased its sugar export and European sugar producers closed factories. Nowadays countries as Thailand or Brazil are leading export countries for sugar. Further, due to the high oil price and the global heating the demand for bio energy increases. This increasing demand strongly influences agricultural commodity prices. Quantities and assets that were used for food production are now used for bio energies. Therefore prices of agricultural commodities as wheat, sugar, and rice increase. In case of white sugar, the price is also dependent on the demand for ethanol as bio fuel. Ethanol and white sugar are both refined from raw sugar and thus substitutes for refiners. A higher demand for ethanol increases white sugar prices. Furthermore, the weather patterns influence the world prices. In case countries have worse or better crops than expected sugar prices change. Moreover, changes in the dollar price affect the sugar price as well. A cheaper US Dollar will increase the sugar price. Finally, in the last couple of years financial institutions and other investors increased their business on the commodity markets. These companies could cause more speculations on the markets and increase the volatility of the market. All these events make the sugar market an insecure market.

5.1.2 Freight

container space nontransparant, resulting in more difficulties with predicting price movements. These difficulties make it hard to determine the right freight prices for contracts with a long forward position and therefore they increase chances of losses on contracts.

5.1.3 Experience in Business

CU is doing business in containers for almost 5 years now. Although people have business experience in sugar, trading sugar in containers creates new opportunities and threats, where the only way to get experienced is learning by doing. One threat is the lack of trading history for many customers. Many customers are new customers and do not have any business with other parts of Company X. Therefore it is hard to know the background of a new customer before making deals. For some companies it is possible to get some idea of a customer’s background through credit reports, statements, or experience of other suppliers/brokers. But for most companies the best way to know more about the customer is by doing business and making trips to customers for a better understanding of their business, capabilities, etcetera. Since traders make such trips most background information is available from the trading side and trip reports are available on an information system of the BU. This causes an information asymmetry between risk manager and traders, and consequently makes the risk management highly dependent on the opinion of traders. As mentioned in the introduction, this could result in irresponsible risks for the company. In case of such manipulations of information, mentioned in the introduction, risk managers are not able to get a good perception of the risk profile of customers. This makes it difficult to take correct decisions. Also it makes risk managers reactive in managing risk. Even if traders do not manipulate information by purpose, the difference in background between traders and risk managers could lead to different interpretations of the information available. Traders could be more focused on the opportunities while risk managers are naturally more focused on the threats of certain trades. Therefore, it is important to reduce this information asymmetry and give risk managers the opportunity to know the customer in the same way as traders do. Combining these two points of view makes it easier to take well-balanced decisions about whether certain trades are riskier or not, and makes risk management more proactive. In this way CU has well-balanced limits on trades.

5.1.4 Quantities

the smallest sold quantity to a specific customer and the most right the largest quantity, shows that almost 75% of CU’s customers bought less than 5,000MT during FY06-07. Comparing the Geneva and CU business models, it is obvious that CU sells much smaller quantities per customer than the Geneva office. For CU, the quantity per (sub) contract is on average 1,063MT.

Table 5.1 Summary Quantities to non-shared customers in FY06-07

Geneva per customer2 CU per customer CU per contract

Average 50,811MT 3,827MT 1,063MT Median 15,310MT 2,000MT 780MT Standard Deviation 91,858MT 5,603MT 1,364MT Max 790,239MT 33,969MT 13,500MT Sum 11,991,388MT 401,825MT 401,825MT Count 237 105 378

Figure 5.1 Sold quantities per customer for CU during FY06-07

0 5000 10000 15000 20000 25000 30000 35000 40000

Customer ranked on sold quantity

S o ld Q u a n ti ty 5.1.5 Customer portfolio

Different customer types have different characteristics that affect the risk of the business. CU sells sugar to customers that do not need much fixed assets for their business as well as to customers that need many fixed assets. Customers that do not need much fixed assets are trading houses and distributors. Both types trade sugar with the only difference that trading houses deal sugar mostly on a worldwide basis and distributors do business on a specific local market. 27% of CU’s customers are trading houses and they are responsible for 31% of the sugar sold over the last three book years. 46% of CU’s customers are distributors and they are responsible for 47% of the sugar sold over the last three book years. Further, companies with many assets can be distinguished in industrials, refineries and millers. Industrials process

2

sugar in their end products. Examples are PerfettiVanMelle, Coca Cola, PepsiCo or other local confectionary or beverage producers. 26% of the customer portfolio is industrial and they bought 20% of the sold sugar over the last three book years. Trades with refiners and millers are exceptional and not common business. This happens when these customers produced less white sugar than they have sold. To follow their obligations to their customers they buy white sugar to compensate their shortage. Table 5.2 gives the share for each customer type in the portfolio. Thus, most sugar is sold to customers with a relatively high exposure to changes in market prices (78%) and therefore it is really important to measure market risk, especially for these customers.

Table 5.2 Customer portfolio

% share in number of customers % of sales in terms of quantity Number Share FY04-05 FY05-06 FY06-07 Grand Total

Distributors 91 46% 63% 34% 45% 47% Trading Houses 54 27% 27% 41% 27% 31% Industrials 52 26% 4% 21% 28% 20% Millers 3 1% 3% 3% 0% 1% Refiners 1 0% 3% 1% 0% 1% 5.1.6 Origination of buyers

Apart from the characteristics of the type of customers, also their origination could influence the risk profile. Due to the characteristics of the sugar market where countries with a sugar deficit are mostly developing countries, it is important to zoom in on how this affects the business. As table 5.3 shows CU sold 79% of its sold quantity to companies from developing countries over the last three book years. For distinguishing developing and developed countries, the distribution of the International Monetary Fund (IMF) (www.imf.org, visited 18-01-08) is applied. As mentioned in subsection 2.1, companies in developingcountries have difficulties with risk management and thus they are highly exposed to market risks. Further, doing business in such countries brings along the problem that it is difficult to generate information about their financials, market position, reputation, background etc. It could even be possible that a customer has a background with a terror group like for example Hezbollah. This lack of information makes it hard to do a good due diligence on the customer risk profile.

Table 5.3 Share sales developing countries

Also the other country risk characteristics mentioned in subsection 2.1 affect the risk profile of customers. Table 5.4 shows the share of quantity sold to countries with the specific rating of the total sales. Each rating is based on the Dunn and Bradstreet country risk indicator. Table 5.5 gives a short interpretation of each rating (www.dnbcountryrisk.com, visited on 19-12-07). For 60% of the sold quantity there is enough uncertainty about returns and risk should be at least actively managed.

Table 5.4 CU Country Risk

Table 5.5 Dunn & Bradstreet (DB) Rating DB rating Company X DB rating Degree of Risk Interpretation

DB1 1 Lowest risk Lowest degree of uncertainty associated with expected returns, such as export payments, and foreign debt and equity servicing.

DB2 2 Low risk Low degree of uncertainty associated with expected returns. However, country-wide factors may result in higher volatility of returns at a future date.

DB3 3 Slight risk Enough uncertainty over expected returns to warrant close monitoring of country risk. Customers should actively manage their risk exposures.

DB4 4 Moderate risk

Significant uncertainty over expected returns. Risk-averse customers are advised to protect against potential losses.

DB5 5 High risk Considerable uncertainty associated with expected returns. Businesses are advised to limit their exposure and/or select high-return transactions only.

DB6 6

Very high risk

Expected returns subject to large degree of volatility. A very high expected return is required to compensate for the additional risk or the cost of hedging such risk.

DB7 6 Highest risk Returns are almost impossible to predict with any accuracy. Business infrastructure has, in effect, broken down.

Source: www.dnbcountryrisk.com, visited on 19-12-2007

5.1.7 To Sum up

CU is a relative inexperienced business in which relations with the repeating customers are in a starting phase. The sugar market is a volatile market that makes the business riskier, certainly due to the fact that CU mainly sells to customer types that are relatively highly affected by changes in sugar prices. Further, the market destination of the buyers makes the ‘Company X’ Dunn & Bradstreet Rating FY 04-05 FY05-06 FY06-07 Grand Total

business risky as well. In general, companies in the developing countries are less informed about hedging possibilities. When these companies do not hedge they are fully exposed to market risk. This makes it even more necessary to actively monitor and mitigate market risk exposure. Moreover, the lack of available information about customers makes it difficult or even impossible to do a due diligence on their credit worthiness. The only way to measure credit worthiness is to use generalizations about customer types. Furthermore, current freight prices are hard to measure and to predict. This makes it difficult to set the good premium on contracts with long forward positions and to prevent losses. According to this profile the assumption is made that for all non-shared customers it is difficult to do a good due diligence. Therefore traders and credit analysts do not know enough about variables, e.g. the business experience of the customer, their management capabilities, financials, credit worthiness, etc., that it could influence the contract conditions. They only know the customer type. Also country risk is not measured because traders say they do not make a difference in contract conditions between originations of customers. Finally, trading experience is not measured since it is difficult to have significant differences in trading experience in only three year business. Therefore only the customer type is used as independent variable for analyzing the contract conditions.

5.2 Contract conditions

Table 5.6 Payment Terms Payment term All

customers Distributors Trading Houses All Industrials Affiliated Industrials Non-affiliated industrials CAD* 23% 14% 18% 44% 83% 23% CAD in Trust 2% - - 7% 15% 2%

CAD via bank with bank guarantee** 2% - 7% - 11% Deposit** 54% 63% 73% 25% 2% 38% 100% prepayment** 6% 7% 2% 3% - 4% L/C ** 14% 16% 8% 14% - 21% Total 100% 100% 100% 100% 100% 100%

*: Credit-exposure mitigating payment term; **: MTM-exposure and credit exposure mitigating payment term

hypothesis at a 99% confidence level. When the data for affiliated industrials are kept out of the test, the hypothesis that the distribution of distributors, trading houses and non-affiliated industrials are equal cannot be rejected at a 95% confidence level. This means that the security levels of these three segmentations do not differ. Therefore these results show that traders indeed take into account the riskier profile of distributors and trading houses, i.e. these customer types have more risk mitigating contract terms.

Table 5.7 Summary data contract conditions Customer

type

Affiliated to big industrial

Average Median Standard deviation

Maximum Minimum Count

Margin- exposure (days) Industrials Yes 202 164 143 590 8 75 No 97 71 84 478 0 132 Distributors 71 49 67 406 0 459 Trading houses 61 45 57 508 3 238 MTM-exposure (days) Industrials Yes 170 127 134 510 8 53 No 97 67 90 478 12 98 Distributors 57 45 53 406 1 312 Trading houses 53 41 38 374 3 167 OCT (MT) Industrials Yes 3,742 1,500 5,789 16,660 0 13 No 2,413 1,495 3,110 16,881 0 33 Distributors 2,989 1,936 3,051 13,600 260 49 Trading houses 2,630 1,080 2,884 11,937 0 37 OPCT (MT) Industrials Yes 3,508 765 5,889 16,660 0 13 No 2,244 1,495 2,467 12,098 0 33 Distributors 2,615 1,725 2,623 13,600 0 49 Trading houses 2,331 1,080 2,638 11,937 0 37 Security level Industrials Yes 51% 50% 6.66% 98% 50% 53 No 84% 96% 20.8% 100% 50% 98 Distributors 89% 93% 16.3% 100% 50% 312 Trading houses 86% 93% 18.1% 98% 50% 167

6 Model for setting trading limits

subsection 6.1. Further, extra calculations are given to specify the limits in more detail. The last subsection back tests the new model.

6.1 The new model

There are two options: make a totally new model or tailor the current Geneva model for CU. Using a total different model makes it very difficult to implement it into the current processes and decreases the visibility for risk management on a BU level. When CU has its own unique model they need different operation or management systems, which makes it difficult to monitor risk on a BU level. Besides that, CU still has shared customers with the Geneva office. A unique model would make it very difficult to monitor risk exposure for these customers on a BU level. Therefore the current model should be tailored as some kind of spin-off. Such a tailored model has several conditions based on the needs of the higher policies and the analysis in section five. The proposed model should take into account:

- reputation risk;

- limits on stress MTM open positions;

- to use payment terms that avoid or minimize non-collection on trade receivables;

- credit worthiness of customers;

- to set tiers to several levels for companies;

- that it is hard to do a good due diligence on customers; - that most customers are highly exposed to market risks; - that CU sells in relatively small quantities;

- that it is possible to ask for deposits or other MTM-mitigating payment terms;

- the shared customers with Geneva;

- that traders make a distinction between trading houses and distributors at one side and affiliated industrials, and non-affiliated industrials on the other.

occasional basis. To do business with these customers it is crucial to keep the business simple and leave the paperwork of doing a due diligence or getting documents of incorporation. Though knowing so less about your customer, risks, e.g. reputation risk, will be much higher. Therefore the model needs a stage for these occasional buyers. Finally, the model should distinguish shared and non-shared customers: shared customers are measured with the Geneva risk model and non-shared customers have their own CU risk model. The CU risk model should have three stages:

- stage 1: the occasional buyer;

- stage 2: the regular customers;

- stage 3: the rated customers.

already bought the sugar whereas the customer knows that he gets the sugar without paying the (total) amount. During this period he can get a return on the amount of money whereas CU cannot invest this amount. Therefore CU finances its customers. They want to reduce these costs since these costs have too much negative impact on Company X’s working capital, certainly during times with high market prices. This decreases investment possibilities for Company X. The last motive for a general forward position is that the risks become too high with forward positions longer than three months. Certainly for the risk of changes in costs for FOB and freight. These costs cannot be measured or hedged and with forward positions longer than three months these costs can change too much, resulting in losses on contracts. Thirdly, when non-affiliated industrials, distributors or trading houses do not fit within stage two these customers move into stage three and get a rating based on the due diligence mentioned in section three. After the due diligence the rating is granted for one year. After that, a new due diligence has to be done. This is to cover rating transition risk. CU’s management is comfortable with a maximum OCT and OPCT of 10,000MT, and a forward position of maximally three months for customers in stage three. The minimum security level depends on the rating. The security levels for each rating are given in table 6.3. Notice that the customers with rating 6 have a higher security level than customers in stage two. This is to cover the OPCT of 10,000MT. The management of CU is convenient to have a maximum OPCT of 10,000MT when they are 99% sure the market price will not drop more than the deposit level.

Table 6.3 Rating and its minimum security level Rating Minimum security level

1 85% 2 85% 3 95% 4 95% 5 95% 6 99%

Table 6.4 Risks CU wants to take

Stage 1 Stage 2 Stage 3

Max. OPCT 5,000 max per annum 5,000 10,000 Max. OCT 5,000 max per annum 10,000 10,000

Min. security level 95% 95% 85%-99%

Max. forward position 2 months 3 months 3 months

Unsecured window 1 week 1 week 1 week

Doc. of incorporation No Yes Yes

Due diligence No No Yes

Figure 6.2 Risk process for CU No Use Geneva risk model Yes No Use GCB risk model stage 1 Yes Yes No Affiliated industrial? Shared customer with Geneva? No Yes No more trading this year Use limits stage 2 Do stage 2 limits fit? Yes No Not Accepted _{Do a due}

diligence Yes Ask for overlimit request Use limits stage 3 granted

for one year

Do these limits fit? Yes No Ask for overlimit request Accepted Overlimit granted for one year Not accepted Accepted Overlimit granted for one year Limits one year old? Yes Overlimit one year old? Yes No Rating one year old? No No New customer No Ask for document of incorporation Do these limits fit?

6.2 Further calculations

6.2.1 Deposit levels

The former subsection gives the minimum security levels needed for CU’s non-shared customers. To calculate the minimum deposit, traders should ask for a certain security level, equation 4.6 is used. Table 6.5 shows the results based on the spot prices used in section five from the 1st of June 2004 till the 11th of January 2008, and based on a 255 working days in one year. To clarify, to be 99% confident that the market value of a flat priced contract will be higher than the amount still to be paid by the customer during a contract period (forward position) of two months, the customer should prepay 24% of the total contract value.

Table 6.5 Minimum Deposit needed

Forward position (in months) 1 2 3 4 5 6 7 8 9 10 11 12 Volatility of Spot prices 7% 10% 13% 15% 17% 18% 20% 21% 22% 23% 25% 26% Security level 99% 17% 24% 29% 34% 38% 42% 45% 48% 51% 54% 56% 59% 95% 11% 15% 18% 21% 24% 26% 28% 30% 32% 33% 35% 36% 90% 8% 11% 13% 15% 17% 19% 20% 22% 23% 25% 26% 27% 85% 6% 9% 10% 12% 13% 15% 17% 17% 18% 19% 20% 21% 75% 4% 5% 6% 7% 8% 9% 10% 10% 11% 11% 12% 13% 67% 2% 3% 4% 4% 5% 5% 6% 6% 7% 7% 7% 8% 50% 0 0 0 0 0 0 0 0 0 0 0 0

In case of unpriced contracts it is difficult to determine the MTM-exposure window on the date the trade. Therefore it is also impossible to determine the exact deposit needed to cover changes in market prices. To give an indication about what the minimum deposit should be when contracts are priced on the B/L-date a calculation is made to show the minimum deposit for all main shipping routes for several security levels. To calculate these deposits equation 4.6 is used. The only difference is that the forward position is replaced with the average number of days needed to ship the sugar from A to B after the B/L-date. The number of days and results could be found in appendix one and five. The optimal deposit is to handle unpriced contracts as flat priced contracts. This has the advantage that when the customer defaults before contract is priced, the margin is (partly) covered. Therefore it reduces losses on defaults. Another option is to agree with a customer that the deposit is a certain percentage of the contract value the moment he wants to price its contract.

6.2.2 Overlimits requests

for 10% of the business: overlimits. These overlimits are granted for one year. After that they should be reviewed to cover rating transition risk. In case the trader wants a longer forward position at the same security level, the minimum required deposits can be calculated with equation 4.6 and are shown in table 6.5. Also table 6,5 gives the minimum deposit needed when the trader wants a lower security level. Furthermore, sometimes traders prefer to sell to a specific customer with no MTM-exposure mitigating payment terms, e.g. selling cash against faxed documents. In that case customers could have MTM-exposure. To set boundaries on the MTM-exposure the maximum OPCT should be limited. Table 6.6 shows the OPCT-limit for a 95% security level calculated with equation 4.10 and using the historical average market price of $302. The spot prices from the 1st of June 2004 till the 11th of January 2008 are used to estimate the historical average price. For other security levels see appendix six.

Table 6.6 Tonnage limit in MT for 95% security level

1 2 3 4 5 6 7 8 9 10 11 12 1,500,000 45,154 33,113 27,594 23,652 20,695 19,103 17,739 16,556 15,522 15,051 14,191 13,424 1,000,000 30,102 22,075 18,396 15,768 13,797 12,736 11,826 11,038 10,348 10,034 9,461 8,949 750,000 22,577 16,556 13,797 11,826 10,348 9,552 8,869 8,278 7,761 7,526 7,096 6,712 500,000 15,051 11,038 9,198 7,884 6,898 6,368 5,913 5,519 5,174 5,017 4,730 4,475 250,000 7,526 5,519 4,599 3,942 3,449 3,184 2,956 2,759 2,587 2,509 2,365 2,237

6.3 Back testing the model

It is important to check whether the trading limits really fit the business. To check, the proposed limits are checked with the historical data. For back testing this model the assumption is made that all customers are within stage two. Since the current MRB system was not implemented on the date of acquiring the data the risk system did not have the correct

ratings. As a result it is impossible to allocate the customers to the right stage. Note that this assumption extends the forward position and sold quantities limits for

have a maximum OPCT of more than 10,000MT. Finally the OPT fits for 95% of the business. Thus, according to the data the new model will limit traders in their business. This makes it important to know if these changes are not too radical. Based on the data 90% of the customers did have one or more contracts with a forward position within three months. Therefore it is not impossible for traders to change their behavior. Additionally 67% of the customers did have one or more contracts with a minimum security level of 95%. This is still a big gap. 23% of the customer should get stricter limits in terms of security levels. To fit for 90% table 6.7 shows the minimal levels for the contract conditions. Even the security levels and forward position of the highest rating in stage three are not flexible enough to fit for 90% of the customers. To conclude, in order to fit to the risks CU’s management wants to accomplish that traders will make a significant change in the contract conditions they currently use: shorter forward position and higher deposits.

Table 6.7 Back testing on historical data

Security level Forward position OPCT OPT

Suitability 50% 77% 87% 95%

Levels to fit for 90% 72% 4 months 5,400 MT 5,400 MT

7 Conclusion, recommendation and suggestions for further research