ECC Margining Version 1.7.8

Transcription

1 ECC Margining Version Contact European Commodity Clearing AG Risk Controlling & Compliance Phone:

2 Table of Contents 1. Introduction Overview of Margin Types Regulatory Requirements 6 2. Calculation of Margin Parameters Calculation of Single Margin Parameters Margin Parameter Procyclicality and Period of Stress Buffer Expiry Month Factor Calculation of Spread-Margin Parameters Conservative Corrections Calculation of Spot Margin Daily Exposures Algorithm to Compute the Initial Spot Margin Special Holiday Adjustment Additional Exposure Dates Consideration of delivery risk in the exposure generation Example for exposure aggregation Calibration of Parameters Current Exposure Spot Market Financial Resources Default Waterfall 22 ECC Margining Page 2 Version 1.7.8

3 5. Margin Reports SPAN Calculation Combined Commodity Scan Risk Volatility Scan Range Short-Option Minimum (SOM) Spreads Perfect Spreads Regular Spreads SPAN Initial Margin Delivery Margin Premium Margin and Additional Collateral % Margin CAP Sample Calculations with PC SPAN Prerequisites Loading SPAN Parameters Building-Portfolio Calculating SPAN Initial Margin Requirements Checking Margin Requirements 35 ECC Margining Page 3

4 1. Introduction This document provides the documentation of ECC's margining system both for derivatives and spot markets. The current values of used calculation parameters, if not set in this document, can be found in the risk parameter file on Overview of Margin Types Margining at ECC distinguishes between derivatives and spot markets: For derivatives markets, ECC employs a statistical approach that calculates potential changes in the value of a trading member s portfolio over a time horizon that is needed to liquidate the portfolio using SPAN. 1 For spot markets, the counterparty risk comprises the payment obligation from concluded transactions that have not been settled. ECC uses an in house model based on the time-series of a members trading behavior to calculate the margin requirement. ECC clears spot markets for commodities on which mainly non-financial market participants are active. Settlement cycles between the non-financial market participants and their Clearing Member are usually longer than between ECC and the Clearing Members. On some spot markets (power, natural gas) trading and clearing takes place 24/7 including times where settlement of payments is not possible due to TARGET II closure. ECC measures credit exposure on spot markets 2 near to real time on a 24/7 basis using the Current Exposure Spot Market. This Current Exposure Spot Market has to be covered with collateral at all times. In order to avoid frequent margin calls due to collateral shortfalls and to cover exposures that might arise from trading activities during TARGET II closure times ECC has established an additional margin component, the Spot Initial Margin. This Spot Initial Margin is an additional buffer. It is designed to cover exposure from potential spot transactions in the future. 1 'SPAN ' is a registered trademark of Chicago Mercantile Exchange Inc. Chicago Mercantile Exchange Inc. assumes no liability in connection with the use of SPAN by any person or entity 2 Non-storable commodities (e.g. gas and power transactions) have current exposure only; storable commodities (e.g. EUA certificates) also have potential future exposure ECC Margining Page 4

5 Potential Future Exposure Current Exposure The parameters of both methods are calibrated to cover ECC's exposure with a confidence level of 99%. The following table gives an overview of the different margin types: Exposure Type Margin Type Description Variation Margin Mark-to-market value (change) of all open positions in futures using the latest market prices received from the markets Premium Margin ECC's options are Premium Style (i.e. no daily Variation Margin is calculated). Therefore Premium Margin has to be deposited for net short positions. For net long positions, credits from Premium Margin are used to offset other margin requirements Current Exposure Spot Market 3 (CESM) The net value (payment amount) of all concluded transactions on the spot markets during the day that have not been settled SPAN Initial Margin SPAN Initial Margin covers the risk in open positions in futures and options Supplementary Initial Margin (MCAP) Covers the difference between the allowed 80% margin reduction and the current margin reduction if the latter is higher Delivery Margin Delivery Margin covers the risk in positions in physically settled futures during the delivery period Initial Margin Spot Market (IMSM) The IMSM is called for expected spot in the future and serves as a buffer to reduce intraday margin calls. Margin requirements on spot markets and margin parameters for derivatives are adjusted on each business day thus allowing ECC to quickly adopt its risk management to new market conditions. Stress testing according to EMIR Article 42, where the default of one or more clearing members under extreme but plausible market scenarios is simulated, is performed daily. Its results are used to determine the default fund ECC maintains to cover counterparty risk in extreme market conditions. ECC performs daily back testing for single and spread margin parameters as well as portfolio margin and spot margin. Daily historical stress testing is used to assess the adequacy of margins. ECC performs an annual validation of methods, models, and model assumptions. 3 Non-storable commodities (e.g. gas and power transactions) have current exposure only; storable commodities (e.g. EUA certificates) also have potential future exposure ECC Margining Page 5

6 1.2 Regulatory Requirements ESMA Article 24 requires that for the calculation of initial margins the CCP shall at least respect the following confidence intervals: (a) For OTC derivatives, 99.5 %; (b) For financial instruments other than OTC derivatives, 99 %. ESMA Article 25 requires that a CCP shall ensure that initial margins cover at least with the confidence interval defined the exposures resulting from historical volatility calculated based on data covering at least the latest 12 months. A CCP shall ensure that the data used for calculating historical volatility capture a full range of market conditions, including periods of stress. ESMA Article 26 requires that the liquidation period shall be at least two business days for financial instruments other than OTC derivatives. According to ESMA Article 28 a CCP shall ensure that its policy for selecting and revising the confidence interval, the liquidation period and the look back period deliver forward looking, stable and prudent margin requirements that limit procyclicality to the extent that the soundness and financial security of the CCP is not negatively affected. This shall include avoiding when possible disruptive or big step changes in margin requirements and establishing transparent and predictable procedures for adjusting margin requirements in response to changing market conditions. In doing so, the CCP shall employ at least one of the following options: (a) Applying a margin buffer at least equal to 25 % of the calculated margins which it allows to be temporarily exhausted in periods when calculated margin requirements are rising significantly (b) Assigning at least 25 % weight to stressed observations in the look back period calculated in accordance with ESMA Article 26 (c) Ensuring that its margin requirements are not lower than those that would be calculated using volatility estimated over a 10 year historical look back period. ECC Margining Page 6

7 2. Calculation of Margin Parameters 2.1 Calculation of Single Margin Parameters Margin Parameter Single margin parameters are the result of a filtered historical simulation over the past 255 days with non-zero returns. The single margin parameter (without buffer) is given by a multiple of a contract's returns' standard deviation 4 M X (t) = R X (t) σ X (t) p X (t) 2. The standard deviation of futures daily returns is computed as an exponentially weighted moving average (EWMA) 5 of the last 255 daily relative non-zero returns from the observed daily settlement prices. For options all relative changes of the implied volatility are used. ECC uses the concept of constant maturities, i.e. the returns and margin parameters are calculated for time buckets with a fixed time to expiry. The use of an exponentially weighted average allows for quicker reaction of the margin parameters to changes in market volatility than the equally weighted estimator for the empirical standard deviation. R X is calculated as a quantile of the volatility normalized returns: R X = ( q(x τ /σ X (τ 1)) α + q(x τ /σ X (τ 1)) 1 α )/2, where α = 0.99 and q(x τ ) α is the empirical α-quantile of X using data from the past 255 days with non-zero returns. There is both a lower and an upper cut-off, R max and R min, for the calculated values of R X. For time series with less than 100 values, R X is set R X = R max. The factor 2 is used for scaling to a liquidation period of 2 days. For contracts with less than 30 values the volatility is linearly interpolated between the calculated volatility and the maximal volatility found in the contract class where the time series have more than 30 values Procyclicality and Period of Stress Buffer ESMA allows for three different methods to prevent procyclical margining. 1. Applying a 25% weight for stress volatility, 2. Using all available data, 3. Applying a 25% buffer which can be temporarily exhausted. ECC has decided to adopt the buffer-method in the following way: ECC first calculates the minimal and maximal volatility σ min and σ max of a contract's return time series using all available data up to the day for which margin is calculated. 4 Always valid for day t. 5 Under the assumption that the mean return is zero, the implemented formula is τ τ σ X (τ) 2 = (X k λ τ k ) / λ τ k. k=max{τ 255,1} k=max{τ 255,1} ECC uses λ = Zeros and missing values are not taken into account. ECC Margining Page 7

8 ECC then calculates a stressed volatility by adding a weighted stress part into the calculation σ S X = 255 w σ 255 X + w σ 255 max, where the weight is currently set 6 w = 5. Below a threshold of σ crit = σ min + a (σ max σ min ), the margin parameter is increased by 25%. The value of a is currently set to 0.2. Above the threshold σ crit, the buffer of 25% is linearly reduced to zero: 0.25 σ X σ crit b X = { 0.25 (1 σ X σ crit. ) else σ max σ crit ECC now calculates the final single margin parameter by taking the maximum of the requirements of the stressed volatility and the buffer method: w σ max σ X M X = R X σ X 2 (1 + max {b X ; }) p 255 σ X X with p X being the contract's current settlement price Expiry Month Factor To cover the increased price risk and / or delivery risk in positions in physically settled futures during the delivery period, a delivery margin is called. Price risk results from the fact that there is no variation margin payment during the delivery. Delivery risk results only in areas where ECC s nomination has no priority and therefore ECC could be imbalanced in the default of a trading participant. To cover the delivery risk, the single margin parameter is adjusted by the expiry month factor (EMF). The EMF is set as follows: - For delivery areas where ECC s nomination has priority or with single sided nomination the EMF is set for natural gas and power futures separately. The current values can be found in the risk parameter file on the website. - For delivery areas where ECC s nomination has no priority the EMF is derived depending on the maximum number of calendar days between last successful settlement and suspension by a TSO following default to deliver (this takes into account local holidays and weekends). The EMF is subject to annual validation. 2.2 Calculation of Spread-Margin Parameters Spread-Margin parameters are calculated in the form of credit C P for selected bivariate portfolios with spread positions, i.e. one asset being held long and the other being held short. These portfolios are not assigned a gross margin, i.e., the sum of the margin requirements for the individual contracts, but a net portfolio margin. ECC's approach to calculate the net margin for such portfolios is similar to the 6 This equals a stress period of one week which has been observed in the Fukushima Event ECC Margining Page 8

9 approach for single margins. The margin is seen as the 99%-quantile of a volatility-normalized historical simulation 7 of the absolute portfolio returns in over the past 255 days with non-zero returns of both contracts. The portfolio is constructed for day 8 t. Its absolute returns (with respect to the current price level and denoted in ) are a time series indexed by τ: P τ (t) = a(t)x τ p X (t) b(t)y τ p Y (t), where X τ, Y τ are the returns on day τ, a(t), b(t) are the position sizes and p X (t), p Y (t) are the prices on day t. The times τ are defined such that both time series (X τ, Y τ ) are non-zero for each τ, 9 as shown in the following table: Target-2 days p X ( ) X p Y ( ) Y τ The time series comprises the past 255 τ-values before t. In the following a(t) and b(t) are set to equal the margin parameters of Y and X, respectively. Standard deviations and correlations are computed using exponentially weighted averages, i.e., EWMA. The gross margin for the portfolio is the sum of the individual assets' margin requirements M P(t) = a(t)m X (t) + b(t)m Y (t), where the margins M X and M Y are calculated as shown in chapter 2.1, i.e. as the product of the EWMA standard deviation and a dynamically adjusted risk multiplier. The risk multiplier also has to be computed for the portfolio itself, yielding the net margin M P (t) = R P (t) σ P (t). Here, the portfolio's standard deviation can be expressed by the assets' correlation coefficient ρ obtained by statistical averaging 10 over τ σ P (t) = (a(t) σ X (t) p X (t)) 2 + (b (t)σ Y (t) p Y (t)) 2 2ρ(t)(a(t) b(t))(σ X (t)p X (t) σ Y (t)p Y (t)). When calculating the net margin, a conservative correction is applied to the correlation coefficient, such that σ P (t) σ P corr (t) and thus M P (t) = R P (t) σ P corr (t). Details are explained in the next section. Like for the single margin parameters, the risk multiplier is given as 7 Also known as "filtered historical simulation". 8 Under the assumption that both contracts are quoted on t. If this is not the case, the most recent credit is taken. 9 Given the individual time series X t and Y t, only values are taken into account where both time series are simultaneously non-zero. This can lead to time series being different from those used for the calculation of the single margin parameters of X and Y. 10 The statistical averaging is again exponentially weighted. ECC Margining Page 9

10 R P (t) = ( q(p τ (t)/σ P (τ 1)) α + q(p τ (t)/σ P (τ 1)) 1 α )/2, where again α = 0.99, and q(x τ ) α denotes the empirical α-quantile of X τ, i.e., with respect to the time-index τ. However, the conservative correction of the correlation coefficient is not applied when the risk multiplier is calculated. Finally, the credit is given by the ratio of gross to net margin both without margin buffer C P = 1 M P. M P SPAN calculates the gross margin M P based on the single margin parameters M X, M Y, which already include the margin buffer b X, b Y, and subtracts C P M P leaving the net margin requirement also including the required margin buffer: M P = M P (1 C P ). 2.3 Conservative Corrections To account for statistical uncertainty in the estimation of volatility for short time series in particular, ECC has implemented a bootstrapping method to apply conservative corrections to the correlation coefficients that are used to calculate the net portfolio standard deviation and hence the margin credit. Drawing from a sample of 100,000 time series with defined correlation, the 10% quantile of the observed EWMA-correlations are determined. These values are stored with respect to length of the time series and underlying correlation coefficient. In the determination of the net margin, the standard deviation is obtained using the corrected correlation coefficient, thereby increasing σ P, yielding a larger net margin and accordingly a smaller margin credit. Given the length N of the time series and the sample correlation coefficient ρ to be corrected, two time series, x t and y t, of length N with independent Gaussian random numbers (zero mean and unit standard deviation) are constructed. From these a third time series is constructed as z t = ρ x t + 1 ρ 2 y t, such that the time series x t and z t are correlated with Pearson-coefficient ρ. In the next step the sample correlation r of x t and z t is computed using exponentially weighted moving averages (EWMA). Repeating this for 100,000 samples of x t and z t produces a distribution of sample correlations. ECC measures the 10% quantile, q 0.1, of this distribution. Sample correlations between the pre-recorded values are linearly interpolated, i.e., given a sample correlation ρ and both the next lower and the next higher value for which data is stored in the look-up table ρ < ρ < ρ >, and r <, r > being the corrected values q 0.1 for ρ < and ρ >, respectively, the corrected value for ρ is r = r < + r> r < ρ > ρ < (ρ ρ< ). ECC Margining Page 10

11 3. Calculation of Spot Margin Spot markets for commodities such as power or natural gas are different from securities markets for a number of reasons: - On power and natural gas spot markets transactions are concluded 24 hours a day, 7 days a week. Due to this 24/7 trading for power and gas credit exposures on the spot markets can arise 24/7. The credit exposures from those transactions are measured on near-to-real time basis with the Current Exposure Spot Market (CESM) as described in chapter Most of ECC s trading participants on the spot markets are non-financial counterparties. Payment cycles between the Clearing Member and his Trading Participants might be slower than between financial counterparties. - A delivery versus payment (DVP) standard cannot be employed to limit counterparty exposures as delivery of power and natural gas takes place shortly after trading and those commodities are not storable: a) Delivery (nomination) of power spot transactions occurs before gate closure of the respective Transmission System Operator (TSO), e.g., 2:30 pm CET on the day prior to delivery for day-ahead contracts (e.g. German Power), minutes before gate closure for intraday transactions. b) Nomination for natural gas spot transactions occurs regularly 2 hours before gate closure. Payment is effected on the following business day (t+1) for all transactions concluded before 4:00 pm CET and t+2 for transactions after 4:00 pm CET. For these reasons ECC has introduced the concept of Spot Initial Margin (IMSM) which is based on an in-house developed margin model. The IMSM is a margin which is called for expected transactions i.e. before any transactions are concluded and before any credit exposure has arisen. It serves as a buffer to reduce the number of margin calls during the day and to ensure that credit exposure (measured with the CESM) on days where margin calls are not possible (e.g. weekends) i.e. the potential exposure from expected transaction will be covered with a high degree of confidence. 11 On the cross border deliveries between France and Germany the lead time is 15 min only; this will become the future lead time for all power spot transactions. ECC Margining Page 11

12 The calculation of IMSM is based on a statistic of the daily total exposures. The daily exposure 12 at day t comprises all spot transactions that have been concluded between the accounting cut-off of the previous ECC business day and the latest point in time where a trading participant in default would be suspended from trading. The following graphic gives an overview of the total exposure: t - 1 t t :00 8:00 16:00 8:00 12:00 Accounting cut-off Payment Accounting cut-off Payment Latest possibility for suspension Total Exposure of day t For power and natural gas, ECC has to cover the full payment risk and computes the spot margin by a time-series statistic of its exposure towards each Trading Member. ECC s margin model for the spot market of non-storable commodities is based on statistical analysis of the historical trading exposure of trading participants, to estimate the 99% quantile of the exposure from spot market transactions between two regular margin calls during a look-back period of 1 year. Exposures from different commodities are netted as ECC only assumes the financial risk occurring due to the trading behavior of the participant, and the settlement cycles are identical for all commodities. The maximum exposure from the last 20 days is used if it is greater than the exposure from the lookback period to account for sudden changes in trading behavior. In general an absolute minimum of 30,000 Euro is applied. The Spot Initial Margin is called with admission of a new Trading Participant and is only returned when the trading admission is terminated and all pending payments with the trading participant have been settled. For emission certificates 13 and other storable commodities no Spot Initial Margin is calculated as there is no 24/7 trading. The certificates can be used in the default of a counterparty (similar to securities) to remediate the loss, only the price-change risk has therefore to be covered. This is done by calculating an intraday margin (Current Exposure Spot Market CESM) on a near-to-real time basis during the trading hours of the respective storable commodities. 12 In EUR, used EOD FX rates can be found at 13 Currently including all EUA, PEUA and CER contracts ECC Margining Page 12

13 3.1 Daily Exposures In the following a detailed mathematical description of the variables used in the calculation of the spot margin is given. The latest parametrization can be found in the ECC risk parameter file. i. ECC is using several calendars a. Calendar days D cal. This is the full calendar. All other sets are subsets of D cal. b. ECC days D ECC = D Business_Dates D Special_Dates 14. ii. When counting days in a calendar, e.g. t + 1 in D ECC, it is implied that both t and t + 1 are elements of D ECC, and t + 1 is the day after t in that calendar 15. Exposure: iii. E i (t) = { P i (s)q i (s) S = [(t 1) ; (t + 1) ]; t counted in D ECC }, s S where t is counting ECC days and the index set S contains timestamps 16, and P i (s) is the price paid by NCM i for that trade at datetime s and Q i (s) the quantity bought (Q > 0) or sold (Q < 0). For calendary days E i (t cal ) = E i (max t tcal { t D ECC }), i.e., the most recent ECC day counts for the calculation of exposures, e.g., Friday's exposure is the same as Saturday's and Sunday's. The payment amount is defined similarly, however shifted in time and containing less trades, PA i (t + 1) = { P i (s)q i (s) S = [(t 1) ; t ]; t counted in D ECC }, s S where the symbols are the same as in the definition of exposure. iv. Spot margin: M i (t + 1) is the spot margin called at ECC day t + 1. The margin M i (t + 1) is constructed to be the actual margin available at day t + 1. v. Outlier: An outlier is counted, when the available margin M i (t) < E i (t) and t is an ECC day 17, t D ECC. The probability p out is the empirical probability (frequency) over all days in a certain test window (at ECC a business year with 250 days) and all NCMs i. 14 More information about this dates can be found in section A more precise definition is achieved by imposing an order on the calendar such that t j > t i for all j > i. Then our notation of t + 1 corresponds to t i+1 if t corresponds to t i. The simplified notation is used for convenience. 16 Here, for convenience the timestamp is assumed to be a unique identifier of a member's trade, although in ECC's settlement instructions there is a unique primary key for each trade, because, e.g., all trades belonging to an auction are settled simultaneously. The subscript indicates the time at that particular day. 17 By construction an outlier at a calendar weekend is preceded by an outlier at a calendar working day. Furthermore, on calendar weekends, ECC has no means of reacting to the outlier, e.g., by intra-day margin calls, such that these outliers cannot be counted against ECC's margin methodology. ECC Margining Page 13

14 vi. The margin efficiency E = 1 t max{e i(t); M min } N NCM M i (t) i is the average efficiency of the margin by the exposure over a specified time window (at ECC a business year with 250 days). Exposures less than the minimum Margin M min are replaced by the minimum Margin M min, because by construction these exposures do not enter the margin algorithm and thus the margin could not have been more accurate in such a case. At present E is computed over the set t D cal. t 3.2 Algorithm to Compute the Initial Spot Margin Each ECC day t D ECC, an NCM's margin requirement for t + 1 D ECC is given by M i (t + 1) = max {μ(e i (s)) S + α σ(e i (s)) S ; β max s S {E i(s )}; M min ; M min _first Ι 18 {t+1<t 0 +29} } + M min _additional The exposure is only used in the statistical component (mean plus standard deviation) if it is larger than the minimum margin M min, otherwise it is replaced by a missing value and hence ignored. The index set S is given by S = [t ; t] counted in D ECC. The standard deviation is calculated using an exponentially weighting algorithm with λ=0.99 to provide a faster adaption to recent changes in trading behavior: σ(e i (s)) S = ((E i (s) μ(e i ) S ) 2 λ t s+1 ) s S / λ t s+1. s S To reduce the statistical error for short time series a safety add-on 19 is calculated and applied to the standard deviation depending on the number of exposures greater than M min. where I stands for the the indicator function 19 The security add-on (or better to say security factor for the standard deviation) is a conservative adjustment to the calculated standard deviation to cope with possible estimation errors when working with small data samples. It is obtained by calculating n (exponentially weighted and unbiased) standard deviations of a ksample of standard normal distributed random variables via bootstrapping and taking the 90% percentile of the values. This procedure is repeated for k = 1 to k = 255 with n = to get a security factor for every possible sample size of exposure values. ECC Margining Page 14

15 Number of exposures Safety Add On I 1 n/a A list of all add-ons can be found in the ECC Risk Parameter file The look-back index set for the maximum-part is similarly given as S = [t d; t] counted in D ECC where d is currently set to 20. In contrast to the statistical component all exposures in S are used. The component M min _first is only applied if the admission date t 0 of NCM i is closer than 30 days to the calculation date t. Since the exposure on calculation day t is not known completely, ECC calculates E i (t) from all trades after (t 1) 4pm and concluded and registered until the calculation point of time to improve the forecast of the model. This incomplete exposure is often referred to as T0-Exposure since it includes the most updated trades on calculation t = Special Holiday Adjustment This methodology covers exposures for periods with a maximum duration of three days considering the regularly occurring risk on weekends. However, on Easter and Christmas holidays the exposure can increase up to five days. To cover this risk ECC Risk controlling amends the current methodology by increasing IMSM before special holidays to cover the additional exposure on the extra non TARGET2 days. The following methodology for a conservative estimation of possible exposures on dates 20 prior the holiday period is implemented: E t0 =λ μ(e s) S + max s S {E (s )} ECC Margining Page 15

16 where λ is 2 for a five day period or 1 for a four day period respectively. μ(e s) S describes the mean of all exposures for single days 20 of the last calendar year which are greater than the minimal margin Additional Exposure Dates The inclusion of the additional exposure dates 21 is a direct consequence of the special holiday adjustment. As the holiday adjustment is used to cover exposures for periods larger than three days it is not necessary any more to include exposures into the IMSM calculation containing trading data for more than three days as well. In order to split these bigger exposures the additional exposures date were added to the calendar being used to determine the exposure ranges. Example: Easter 2016 Calendar date Current Calendar Exposure New Calendar Exposure Wednesday, 2016/03/23 Yes (2016/03/ /03/24) Yes (2016/03/ /03/24) Thursday, 2016/03/24 Yes (2016/03/ /03/29) Yes (2016/03/ /03/25) Friday, 2016/03/25 n/a Yes (2016/03/ /03/28) Saturday, 2016/03/26 n/a n/a Sunday, 2016/03/27 n/a n/a Monday, 2016/03/28 n/a Yes (2016/03/ /03/29) Tuesday, 2016/03/29 Yes (2016/03/ /03/30) Yes (2016/03/ /03/30) Consideration of delivery risk in the exposure generation For markets where ECC faces a delivery risk, i.e. where ECCs physical nomination can be cancelled by external parties the delivery risk is included in the exposure generation by taking a relative portion 22 of the financial exposure. Example: For the UK market sellers are facing a curtailment risk by the TSO. In order to cover the potential payments of imbalances 30% of the financial exposure is taken into account. Exposure OLD = GBP 20 Tuesday, Wednesday, Thursday 21 which can be found in the ECC Risk Parameter File 22 which can be found in the ECC Risk Parameter File ECC Margining Page 16

17 Exposure NEW = GBP * -0.3 = 300 GBP Example for exposure aggregation Consider the following example of one weeks' trading: Day and time indicate when a trade has been concluded. The trade value is the traded quantity times the price. All trades up until the booking cut are included in next day s payment amount. ECC's exposure towards the trading participant contains all trades between the last booking cut and the next morning (12:00). This quantity is the basis of ECC's Spot Margin. ECC Margining Page 17

18 ECC 4pm Booking Cut Latest possibility for suspension 11:45 am Day Time Trade Value (k ) Exposure PaymentAmount Mo 22: :00 Tue Wed 12: : : :00 13: : : Thu Fri Sat Sun 12: : : : : : : : : : : : : : : Mo 13: : : Tue 12 : In order to better illustrate the IMSM algorithm and how the exposures are processed by the daily margin calculation ECC provides an IMSM calculator as an excel version. The current version can be downloaded on the ECC website ECC Margining Page 18

19 3.3 Calibration of Parameters The algorithm used to calculate ECC's spot margin has four free parameters: α, β, M min _first and the length of the look-back period d. These parameters have been determined by a calibration on the latest one-year time-window at the initial calibration date (September 2014). The calibration is an optimization algorithm that maximizes the margin efficiency E while keeping the outlier probability p out 1%. The current values of the parameters α, β and d can be found in the risk parameter file on the website. 3.4 Current Exposure Spot Market On spot power and natural gas markets trading takes place 24/7. During an ECC business day ECC's current, i.e. intraday, risk will be reflected in the Current Exposure Spot Market (CESM). The CESM refers to the already accrued trade risk exposure, the remaining spot risk exposure from e.g. certificate-based products, and delivery risk which is not included in the IMSM buffer. The interpretation of the CESM at day t depends on the point in time when it is monitored: Intraday CESM until CET: The CESM describes the intraday monitor value for the net payment amount on t + 1. In addition, the CESM includes unresolved delivery risk for UK Power as well. End of Day CESM: The CESM describes the already accrued payment amount for trades conducted after 16:00 CET with payment on t + 2 for products not included in the IMSM buffer. In addition, unresolved delivery risk of is included as well. If financial settlement is deferred (e.g. due to non- Target2 holidays in the corresponding market) the EOD CESM additionally includes these deferred payments. This margin is uploaded intraday into the EUREX Clearing system and is updated every 10 minutes. The margin will be released, as soon as the corresponding payments have been instructed in the payment system at the end of the business day (currently 18:00 CET).It is reported in EUREX's reports so Clearing Members can detect accumulating exposure on the spot market intraday. The time schedule is illustrated in the following graph 23 : 23 Note: As the figure is already complex and is intended to show only timing, only the case is shown where exposure equals payments. ECC Margining Page 19

20 For all unsettled gas and power transactions and net sells of EUA and similar products 24 the current exposure is the net payment amount of all yet unpaid transactions. For net purchases of EUA and similar storable products, the exposure is the price change risk, encoded in a margin parameter calculated along the lines of section based on the daily returns from EEX's settlement prices for emission certificates. For market areas where ECC faces a curtailment risk (e.g. by trading on a non virtual balancing point or by a local limit implementation of the TSO) or currency risk the margin parameter will be adjusted to cover all possible financial and physical balancing actions ECC will execute. A list of adjusted margin parameters for affected market areas can be found in the ECC risk parameter file. The amount of CESM being reported is the sum of all current exposures of a member: 24 Storable products like EUA, CER, Guarantees of Origin ECC Margining Page 20

21 Current Exposure Spot Market Max (0; Storable Commodities + Non-Storable Commodities) Storable Commodities n { PA i MP i if PA i > 0 PA i if PA i < 0 i=1 i Storable Commodity, e.g. EUA, CER PA i Outstanding payments resulting from trading Storable Commodity i MP i Margin Parameter for Storable Commodity i Storable Commodites with additional risk drivers n PA j MP j j=1 j Non-Storable Commodity, e.g. POWER, NATGAS PA j Absolute value of outstanding payments resulting from trading Non-Storable Commodity j Non-Storable Commodities n PA j j=1 j Non-Storable Commodity, e.g. POWER, NATGAS PA j Outstanding payments resulting from trading Non-Storable Commodity j The CESM is floored at 0 so no credit will be granted to other margin classes. The current values of the used margin parameters can be found in the risk parameter file on the website. ECC Margining Page 21

22 4. Financial Resources According to EMIR Article 42 a CCP shall maintain a pre-funded default fund to cover losses that exceed the losses to be covered by margin arising from the default of a clearing member. The CCP shall establish a minimum amount below which the size of the default fund is not to fall under any circumstances. A CCP shall establish the minimum size of contributions to the default fund and the criteria to calculate the contributions of the single clearing members. The contributions shall be proportional to the exposures of each clearing member. According to EMIR Article 43 a CCP shall maintain sufficient pre-funded available financial resources to cover potential losses that exceed the losses to be covered by margin requirements and the default fund. The margin requirement, the default fund and the other financial resources combined shall at all times enable the CCP to withstand the default of at least the two clearing members to which it has the largest exposures under extreme but plausible market conditions. The financial resources of ECC are therefore comprised of: financial resources = margin requirement + default fund + ECC dedicated own resources + other financial resources 4.1 Default Waterfall Calling margins is ECC's first mechanism to protect itself against the risks from a clearing member s default. The margin requirement is the capital equivalent of the available collateral (including haircut, capped at the margin requirement as per the current margin method) at the time of the stress calculation. As a first (conservative) approximation it is assumed that, in a stress case, excess collateral would largely be reclaimed prior to default as for banks insolvency is largely triggered by illiquidity, such that this conservative assessment of the margin requirement is plausible. The default fund covers losses exceeding the losses to be covered by margins. The clearing members contribute to the default fund in proportion to their exposure, which is equivalent to the margin provided. The default fund has to be replenished up to its contribution at the time of a default. The default fund has sufficient resources to cover losses arising out of a default of the two clearing member to whom ECC has the largest exposure to. The total volume of the default fund is determined by ECC based on daily stress tests. The stress tests simulate the default of one or more clearing members under the assumption of extreme but plausible market conditions as defined in the ECC Stress Test Framework. Moreover, ECC holds dedicated own resources to cover potential losses exceeding the ones covered by the margin requirements and the default fund contributions of the defaulting clearing member. Dedicated own resources are freely available and amount to 25% of the capital requirements defined under 16 EMIR. If ECC s dedicated own resources should fall below the limit required by EMIR, ECC will reinstate sufficient capital within one month at the latest. The combined resources of margins, default fund and dedicated own resources are sufficient to cover losses arising out of a default of the two clearing members to whom ECC has the largest exposure to under market definitions defined in the ECC Stress Test Framework. ECC Margining Page 22

23 ECC may hold additional financial resources (other financial resources) to cover potential losses exceeding the resources described above. Other financial resources are defined as the part of the ECC's own capital not needed to comply with the minimum requirements under supervisory legislation (solvency ratios, liquidity ratios). The sequence of usage of the financial resource in the default waterfall is stipulated as follows: 1. ) Indiviual Margins, excess collateral or other amounts due to the defaulting member 2. ) Default fund contribution of defaulting member 3.) ECC dedicated own resources 4.) Default fund contribution of non-defaulting members 5.) ECC other financial resources Only if the loss from the default of a member exceeds its individual margins, individual default fund contribution and ECC dedicated own resources, the default fund contribution of other non-defaulting members will be used. Individual margins of non-defaulting members are never used to cover any losses of defaulting members. ECC does not use any powers of assessment or tear up of existing contracts. ECC Margining Page 23

24 5. Margin Reports Margin Requirements are reported end-of-day in Eurex report CC050. Report Name Content Availability CC050 Margin Overview End-of-Day overview of all calculated ECC Margin Requirements. Chapter 7 describes the detail reports that can be created using PC SPAN. Within margin reports the following abbreviations are used: Shortcut Margin SPAN IMSM UKPF HUPF NLPF AMPO AMCO AMEM NLGF NBPF CEGF DMEM CESM SPAN Initial Margin Spot Margin (CESM Buffer) Daily before the start of new ECC business day via Eurex Delivery Margin for delivery instructions of expired UK Power Futures Product Families: F5BM, F5PM, F6BM Delivery Margin for delivery instructions of expired Hungarian Power Futures Product Families: F8BM, F8PM, F8B1-F8BM Delivery Margin for delivery instructions of expired NL Power Futures Product Families: F4B1-F4B5 Additional Margin for financial Power contracts expiring before the final settlementprice is determined. The margin is equal to the SPAN margin of the contract. The margin will be released on the next payment day after the final settlement price was calculated. Additional Margin for a coal contract expiring on a EEX holiday where the final settlementprice is not determined. The margin is equal to the SPAN margin of the contract. The margin will be released on the next payment day after the final settlementprice was calculated. Additional Margin for emisson contracts, which is called for the buy- side only in order to cover the time period between contract expiry and delivery. The margin is equal to the SPAN margin of the contract. The margin will be released with the payment of the delivery day. Delivery Margin for delivery instructions of expired TTF Natural Gas Futures Product Families: G4W1-G4W5 Delivery Margin for delivery instructions of expired NBP Natural Gas Futures Product Families: G9BM Delivery Margin for delivery instructions of expired CEGH Natural Gas Futures Product Families: G7BM Delivery Margin Emissions Certificates and storable commodities Current Exposure Spot Market ECC Margining Page 24

25 6. SPAN Calculation ECC uses the standard SPAN 25 methodology to account for portfolio effects on derivatives markets. The methodology allows ECC to optimally align margin requirements with risk, thereby realizing efficient margining. ECC updates SPAN risk parameters daily which are available on ECC s homepage for download. ECC recognizes the diversification effect in large portfolios by granting margin credits of up to 99% for opposing positions in highly correlated products. 6.1 Combined Commodity Products (futures and options) with the same underlying, load profile, delivery period and maturity form a combined commodity, e.g. all power futures of the same delivery area and delivery period with the same maturity including all option series of the same maturity form a combined commodity Scan Risk SPAN uses a configurable range of price and volatility movements to calculate the worst-case loss of a combined commodity. SPAN comes with 16 pre-defined scenarios of combinations of price and volatility movements over an assumed liquidation period; at ECC these are used without further customization. The scenarios are so-called scan points, each of which is characterized by a price change (multiple of price scan range), volatility change (multiple of volatility scan range) and the weight attached to the scan point. In the case of futures, the worst-case loss is determined by the price scan range only. To comply with standard methods, ECC bases this price scan range on the single margin parameter defined before. Price Scan Range = M X ContractVolume For simplicity, the term price scan range is also referred to as scan range. For options, ECC sets the price scan range to the price scan range of the underlying, uses a look-ahead period according to the assumed liquidation period (i.e. the time to maturity is decreased by the liquidation period) and an adequate volatility scan range. In SPAN Options are priced using the Black76-model 27. The value of the volatility scan range, the appropriateness of the 16 scenarios and the inclusion of interest-raterisk are determined by Risk Controlling and are subject to at least annual validation. 25 SPAN being short for The Standard Portfolio Analysis of Risk system is a methodology that calculates margin requirements by analyzing the "what-ifs" of different market scenarios. Developed and implemented in 1988 by Chicago Mercantile Exchange (CME), SPAN was the first system ever to calculate margin requirements exclusively on the basis of overall portfolio risk at both clearing and customer level. In the years since its inception, SPAN has become the industry standard for portfolio risk assessment. 26 Due to technical constraints at the Clearing Members' vendors a combined commodity across two exchanges is not possible; the same effect is achieved by setting the intercommodity spreads to Black, Fischer. The pricing of commodity contracts, Journal of Financial Economics (1976) ECC Margining Page 25

26 These scenarios are applied to all products of each portfolio. The scenario with the greatest loss is called active scenario and is considered for the calculation of the SPAN Initial Margin. The scan risk is calculated by multiplying the active scenario loss by the net position. The scan risk of a combined commodity is calculated by summing over the respective scan risks of the constituent product families. 6.3 Volatility Scan Range The volatility scan range is the expected change of the implied volatility of an option over the assumed liquidation period. It is expressed as percentage change from the current implied volatility of the option. The volatility scan range is determined from analyzing the daily changes in implied volatility across all existing liquid 28 options with the same underlying (across all maturities and strikes). The maximum day-to-day change (rounded up to the next 10%) is then used as volatility scan range for all options. 6.4 Short-Option Minimum (SOM) For short options that are deep out of the money, the theoretical risk calculated can be near to zero. In market situations where the underlying price changes significantly these options may move intothe-money and may generate large losses for holders of short positions in these options. Therefore a minimum margin requirement for net short positions in options is implemented, called short option minimum. If the short option minimum of a combined commodity exceeds the scan risk after spreading, it constitutes the SPAN Initial Margin of the combined commodity. The short-option minimum is subject to annual validation. 6.5 Spreads A spread contains offsetting positions in correlated instruments. Due to ECC s product portfolio offering arbitrage-free prices for futures in months/quarters/years, ECC uses not only correlation to form regular spreads but also arbitrage-free prices to form so-called perfect spreads. Both kinds of spreads allow ECC to reduce margin requirements without compromising risk coverage. Contracts in delivery are not included in spreading Perfect Spreads Opposing positions with the same underlying and completely overlapping delivery periods, which differ only in delivery profile or delivery period, are called perfect spreads. Such positions are nearly risk free i.e. the daily variation margin of all positions in a perfect spread is zero because the settlement prices are arbitrage free. To account for differences from rounding effects the margin credit is set to 99%. Thus, the intercommodity credit for each position in a perfect spread is 28 Liquid Options are all options that have been traded at least once ECC Margining Page 26

27 Intercommodity Credit = Scan Risk 0.99 Perfect spreads can contain different combined commodities. Perfect spreads are obtained by decomposing all products in subproducts, such as years in seasons/quarters/months, seasons in quarters and months, quarters in months, base in peak and offpeak. The decomposition is implemented using a Chomsky-type-2 algorithm Regular Spreads Regular spreads exploit the correlation between time series to reduce the margin requirement. The granted margin reduction is calculated in SPAN using the credit as introduced in section2.2. Credits less than 0.01% are deleted, and the maximum applied margin credit is 99%. If a portfolio consists of two opposing positions X and Y, the intercommodity credit for each of both positions is given by Intercommodity Credit = min (Scan Risk X, Scan Risk Y ) Margin Credit Extraction of regular spreads from portfolio data using the margin credits given in the span-file is done by the SPAN -software. 6.6 SPAN Initial Margin To summarize, the SPAN initial margin is calculated per combined commodity in the following steps: 1. Calculation of the overall scan risk for each combined commodity. 2. The scan risk is then reduced by the intercommodity credits to reflect the reduced risk in portfolios with opposing positions. This process is called spreading. Perfect spreads are applied first. Then ordinary spreads are applied to the portfolio in descending order of intercommodity credit. If necessary, the short option minimum is applied afterwards. 3. The resulting amount per portfolio is called SPAN initial margin. 6.7 Delivery Margin Delivery Margin (DM) is called for positions in physically-settled power and natural gas futures during the delivery period on the day after the expiry of the contract and for net short positions in storable commodities two business days before expiry of the contract. The Delivery Margin for power and natural gas futures is given by 29 : DM Power / Natural Gas = Scan Range Front_Month Expiry Month Factor Net Position The Expiry Month Factor is set as introduced in section 2.1. It should be noted that the reduction in Contract Volume over the delivery month is not considered in the calculation of the Delivery Margin. For power and natural gas futures for which positions are kept in the EUREX system during delivery, 29 The Scan Range of the last trading day before the expiry date is used ECC Margining Page 27

28 the Delivery Margin is reported under the future s respective Margin Class (e.g. HUPF for Hungarian Power Futures). The Delivery Margin for storable commodities is given by: DM Storable Commodities = Last Spot Price *(1+HC Storable Commodities)* Volume * Net Short Position Each holder of an open short position in storable commodities such as emission rights or guarantees of origin is obliged to pre-deliver the respective commodities to ECC's storable commodities account before the settlement day of the position. In case of a shortage of holdings ECC will demand securities in the form of a Delivery Margin. ECC calculates the Delivery Margin two business days before expiry of the contract and adjusts the Delivery Margin on a daily basis until expiry of the contract. ECC will mark-to-market the exposure from the net short balances intraday and add a haircut for potential fluctuations in market value. The current value of this haircut can be found in the risk parameter file on the website. The Delivery Margin for storable commodities is reported under the Margin Class DMEM. ECC Margining Page 28

29 Example (using the parameters as of ) Power and Natural Gas Future Delivery Margin: Contract F8BM Net Position 10 Scan Range Expiry Month Factor 2.5 Delivery Margin 10 * * 2.5 = 100, Emission Certificates Future Delivery Margin: Last Spot Price 12 /t Net Short Position 10 Contract Size 1000 t Delivery Margin 12 *1.3 * 10 * 1000 = 156, Premium Margin and Additional Collateral At ECC, options are not subject to variation margin. Instead a premium margin is defined as the product of net position, contract size, and current option settlement price. For short options, the premium margin is called daily: for long options, the premium is credited to the member's account but not paid out. EMIR Article 46 (2) allows the acceptance of the underlying of a derivative or the financial instrument that gives rise to a risk to be used as collateral (additional collateral). Currently, ECC accepts EUA as collateral, which is applied similar to premium margin i.e. an additional collateral amount reduces the margin requirement. Different models are offered which have to be selected by the Clearing Member: # Formula per selected Model incl. Risk Management Limits 1 := Min ( Available Collateral Value ; 0) 2 := Min ( Available Collateral Value ; x % * Req BASE ; y ; Req IMSM ) 3 := Min ( Available Collateral Value ; x % * Req BASE ; y ; Req IMSM + Req SPAN ) 4 := Min ( Available Collateral Value ; x % * Req BASE ; y ; Req IMSM + Req SPAN + Req PREM ) 5 := Min ( Available Collateral Value ; x % * Req BASE ; y ; Req IMSM + Req SPANcapped ) 6 := Min ( Available Collateral Value ; x % * Req BASE ; y ; Req SPANcapped ) Where: Req BASE := Req IMSM + Req SPAN ECC Margining Page 29

30 Req SPANcapped := Min ( Min ( Pieces Available ; Net Short Future Position ) * ( Value * (1 Haircut ) ) ; Req SPAN ) Concentration limits on ECC level according to the current concentration risk policy apply to limit the maximum share of EUA that can be used as collateral % Margin CAP ECC calculates margin on a gross basis for each Clearing Member (i.e. without taking into account netting effects between different clients). Regular Spreads are calculated only for opposing positions (i.e. long vs. short) and the margin reduction is based on the historical behavior of the regular spreads. If the margin reduction after all those measures still exceeds the 80% cap set by RTS 153 Article 27 (4), a supplementary margin is calculated. According to RTS 153 Article 27 (4), where portfolio margining covers multiple instruments, the amount of margin reductions shall be no greater than 80 % of the difference between the sum of the margins for each product calculated on an individual basis and the margin calculated based on a combined estimation of the exposure for the combined portfolio. ECC has a two-step approach: In the first step, margins are calculated on a gross basis and on a net basis for each Clearing Member. The difference between the gross and net margin is the highest possible margin reduction for each Clearing Member. In the second step the supplementary margin is calculated using the current initial margin, the gross margin and the net margin for each Clearing Member (CM). If the margin reduction granted by the current initial margin exceeds 80% of the highest possible margin reduction, the supplementary margin ensures coverage of this exceedance.step 1: The gross margin is the SPAN Initial Margin without spreading except for Perfect Spreads. Perfect Spreads are applied to risk-free positions and therefore a 100% margin reduction is allowed. The gross margin of a Clearing Member is the sum of the gross margins of its accounts (proprietary and Non Clearing Members accounts): GrossMargin CM = GrossMargin Accounts The net margin is calculated similarly to the regular SPAN Initial Margin but Margin Credits are granted for opposing and concurrent positions and there is a netting of all positions on Clearing Member level (non-segregated collateral). The net margin represents the portfolio VaR of a Clearing Member. Step 2: The difference between the gross and net margin is expected to be the highest possible margin reduction of a Clearing Member. ECC now calculates the lower margin limit for each Clearing Member: Lower Margin Limit CM = GrossMargin CM (GrossMargin CM NetMargin CM ) 0.8 = 0.2 GrossMargin CM NetMargin CM If the current initial margin of a Clearing Member is less than the lower margin limit, a supplementary margin is calculated at the end of each day to ensure that the 80% limit is observed: SupplementaryMargin CM = max{lower Margin Limit CM Current Initial Margin CM ; 0} The supplementary margin is introduced as additional minimum requirement for the default fund. ECC Margining Page 30

31 7. Sample Calculations with PC SPAN 7.1 Prerequisites To calculate SPAN Initial Margin, the following requirements have to be fulfilled: 1) The PC SPAN tool can be downloaded from CME free of charge. This can be accomplished in the following way: a. Go to and create a new account b. Login and navigate to Portfolio&Risk/CoreMarginCalculator/Download Center/Software c. Choose the Version of PC SPAN suiting your needs (in general the latest version) 2) Download and install the newest PC SPAN Orgmaster, which is available on the CME public Span FTP site, (ftp://ftp.cmegroup.com/pub/span/util/ ; no login required) 3) The current SPAN parameter file can be obtained from: a. ECC s FTP site (ftps.ecc.de, login required) b. ECC s website ( no login required) c. CME s FTP site (ftp://ftp.cmegroup.com/span/data/ecc/; no login required) It should be noted that only the SPAN parameter file on ECC s FTP site contains the full set of settlement prices. 4) A position file in the SPAN format or the user can enter positions manually The example files are available on ECC s website ( as of 4th July 2012 (which has been generated on 3rd July 2012) ECC Margining Page 31

32 7.2 Loading SPAN Parameters The first step is to load parameter files into PC SPAN via File > Load File(s). Detailed information concerning products and exchanges can be viewed by expanding the relevant exchange path. 7.3 Building-Portfolio Portfolios can be loaded via File Open Portfolio. Select the specific SPAN position file or enter the positions manually. Click on File New Portfolio and enter details of your account. The options ( Qualified Institutional Buyer and Normal for GSCIER ) are not used for ECC margin calculation and should be left unchanged. Please double check that Euro is used as your portfolio currency or otherwise select Euro as your portfolio currency. By adjusting the entries for native and conversion currency to EUR on the Tools Preferences Calculation Parameters tab, no change for portfolio currency will be necessary. ECC Margining Page 32

33 By selecting the Positions tab, you can choose the specific products and type in the actual position size. In the example the ECC Test Positions ( ) contain five positions of F1BQ with settlement date Oct 2012: By checking the boxes Contracts with Positions and All Positions for Selected Exchange Complex you get a complete overview of all current positions. ECC Margining Page 33

34 7.4 Calculating SPAN Initial Margin Requirements Choose Calculate Portfolio(s) Requirement from File Menu either for all portfolios or just for one specific portfolio by selecting a portfolio and right clicking and then selecting calculate requirements. The different components of the SPAN Initial Margin and the Net Option Value (for portfolios with option positions) are shown in a popup window: They can also be accessed by selecting the specified portfolio under the Performance Bond Requirement tab. ECC Margining Page 34

35 7.5 Checking Margin Requirements A detailed margin calculation can be viewed, printed and exported by selecting the tab Reports. The Scan Risk can be obtained within Scan Tiers section which shows the result from the 16 scenarios: The report Inter Tiers shows Inter Commodity Credit per Combined Commodity (in the Intercomm Credit column): ECC Margining Page 35

36 Original Delta and Remaining Delta denote position sizes before and after spreading. In summary, the SPAN Initial Margin Requirement (without Available Net Option) consists of all Scan Risks less the sum of all Intercommodity Credits, except the lines denoted with Overall in the column Tier#. ECC Margining Page 36