2015/16 Assessment of ASX Clearing and Settlement Facilities A1.1 ASX Clear Standard 6: Margin

A central counterparty should cover its credit exposures to its participants for all products through an effective margin system that is risk based and regularly reviewed.

ASX Clear applies initial and variation margin to both derivatives exposures and cash securities transactions, using margin systems that are tailored to the particular attributes of these product types (CCP Standard 6.1). Timely price data are available for most products subject to ASX Clear's margin systems, and ASX Clear applies appropriate models to estimate prices when timely and reliable data are not available (CCP Standard 6.2). ASX Clear's margin models for both cash securities and derivatives target a single-tailed confidence level of at least 99 per cent of the estimated distribution of future exposure, applying appropriate and conservative assumptions regarding holding periods, product risks, portfolio effects, product offsets and floors to limit the need for procyclical changes (CCP Standards 6.3, 6.5). In addition, ASX Clear applies variation margin to both securities and derivatives positions daily, and may call intraday margin on derivatives positions in the event of significant market movements (CCP Standard 6.4).

ASX Clear performs daily and periodic backtesting of its margin models to assess the adequacy of initial margin against the targeted level of cover, and performs an annual review of margin policy. ASX Clear uses sensitivity analysis to validate the assumptions underpinning margin models, including to test the reliability of implicit or explicit product offsets (CCP Standard 6.6). ASX Clear regularly reviews and validates its margin models. An external expert commenced a comprehensive review of ASX Clear's margin models in 2014/15; the first validation of the CMM model and the second validations of the DPS and the SPAN model were completed in June 2016 and July 2016 respectively (CCP Standard 6.7). The operating hours of ASX Clear's margin systems are consistent with those of related payment and settlement systems in Australia (CCP Standard 6.8).

6.1 A central counterparty should have a margin system that establishes margin levels commensurate with the risks and particular attributes of each product, portfolio and market it serves.

ASX Clear applies initial and variation margin to both derivatives products, and cash securities transactions. Initial (risk) margin provides protection to a CCP in the event that a participant defaults and an adverse price change occurs before the CCP can close out the defaulted participant's positions (potential future exposure). Variation margin is levied on cash market positions, long and short LEPOs, and all futures positions to reflect observed price movements (current exposure); it is collected from the participant with a mark-to-market loss and, depending on the product, either passed through in cash to the participant with a mark-to-market gain, or recognised as a credit (see CCP Standard 6.4). ASX Clear also levies so-called ‘premium’ margin on short ETO positions, updating this daily to reflect mark-to-market changes in the close-out price.

Cash equities

ASX Clear's CMM approach involves the collection of initial margin and mark-to-market margin in respect of unsettled cash securities transactions. The selected methodology for initial margin calculation is primarily based on HSVaR. The HSVaR methodology uses historical price moves to calculate hypothetical changes in the value of a portfolio of securities, and determines a margin requirement from these taking into account the desired degree of confidence (see CCP Standard 6.3). For less liquid stocks, or securities with an insufficient price history to apply HSVaR, ASX Clear applies flat rate margins. Currently 78 of the 500 stocks that make up the All Ordinaries Index are margined on a flat-rate basis. Margins calculated using HSVaR currently make up around 59 per cent of initial margins collected through the CMM system. Around 59 per cent of flat-rate margin collections relate to trades in warrants and stocks outside the All Ordinaries Index, which attract higher margin rates. Transactions in Australian Government securities depository interests are margined according to the flat rate applied to fixed-interest products. CMM margin rates are reviewed on a three-monthly cycle.

Cash securities transactions generated by exercise of ETOs or LEPOs are also margined using CMM between exercise and settlement. The settlement obligations of the ETO or LEPO buyer include the exercise price, final margin payments and the outstanding balance of the premium. Prior to exercise, ETOs and LEPOs are margined using SPAN (described below).

Derivatives

ASX Clear uses a variant of the internationally accepted SPAN methodology for the margining of derivatives positions (see CCP Standard 6.3). ETO margin rates are reviewed on a monthly cycle. Regular margin rate reviews are supplemented with ad hoc reviews during especially volatile market conditions.

ASX Clear predominantly clears standardised, exchange-traded products with risks that are well known to both ASX Clear and its participants. The only OTC products cleared by ASX Clear are equity options that share similar characteristics to exchange-traded options.

6.2 A central counterparty should have a reliable source of timely price data for its margin system. A central counterparty should also have procedures and sound valuation models for addressing circumstances in which pricing data are not readily available or reliable.

ASX Clear has access to timely price data for the majority of its exchange-traded products. For less liquid stocks (e.g. stocks outside the All Ordinaries Index and warrants) and new stocks for which there is insufficient historical price data, ASX Clear applies flat rate margin. This is based on available price information for individual stocks in the All Ordinaries Index, or for grouped categories of other products. The settlement value of ETOs is calculated throughout the day using the DPS. Where available, the DPS uses traded prices, but the system is able to extrapolate prices from previous pricing periods or untraded bids and offers where traded price data are not available. For less liquid stock options, DPS cross-checks calculated prices against trades in similar options, takes into account limits on implied volatilities, and smooths and imposes restrictions on the slope and convexity of deemed volatility curves. For OTC equity options, ASX Clear interpolates the value using the prices of similar ETOs.

The DPS is considered a key risk model and accordingly will be subject to annual validation under ASX's Model Validation Standard using an external expert. A validation of the DPS was completed in June 2016.

6.3 A central counterparty should adopt initial margin models and parameters that are risk based and generate margin requirements sufficient to cover its potential future exposure to participants in the interval between the last margin collection and the close out of positions following a participant default. Initial margin should meet an established single-tailed confidence level of at least 99 per cent with respect to the estimated distribution of future exposure. For a central counterparty that calculates margin at the portfolio level, this requirement applies to each portfolio's distribution of future exposure. For a central counterparty that calculates margin at more granular levels, such as at the sub-portfolio level or by product, the requirement should be met for corresponding distributions of future exposure. The model should: use a conservative estimate of the time horizons for the effective hedging or close out of the particular types of products cleared by the central counterparty (including in stressed market conditions); have an appropriate method for measuring credit exposure that accounts for relevant product risk factors and portfolio effects across products; and to the extent practicable and prudent, limit the need for destabilising, procyclical changes.

ASX Clear applies different margin models to securities and derivatives transactions.

Securities

For most securities transactions, ASX applies an HSVaR-based model, which is calibrated and adjusted to meet a single-tailed confidence interval of 99.7 per cent of the estimated distribution of future exposure. Estimates of the distribution of future exposure under this model are based on 2 years of 1-day price moves applied to current participant portfolios (see CCP Standard 6.5). While ASX targets 99.7 per cent coverage of the distribution of future exposures from CMM, as a first step ASX identifies the 99th percentile of the sample distribution. Since HSVaR requires reliable and uninterrupted price data, it is only applied to transactions in sufficiently liquid securities, namely those in the ASX 500 All Ordinaries. Even so, the small number of observations of price movements beyond the 99th percentile makes it difficult to construct reliable estimates of the desired 99.7 per cent margin coverage. ASX therefore applies a Portfolio Add-on Factor (currently 30 per cent) to the HSVaR estimate of potential future exposure at a 99 per cent confidence level to achieve the desired 99.7 per cent level of cover.

For securities that do not have the required price history to apply HSVaR, ASX applies, consistent with its overall CMM approach, flat rate margin intended to cover 1-day price moves with a 99.7 per cent confidence at a portfolio level. In order to achieve the desired confidence level at the portfolio level, confidence intervals and holding periods applied to individual stocks differ according to liquidity and available price information. Stocks in the ASX 200 target a 99.7 per cent confidence interval applied to a 1-day holding period; other stocks in the All Ordinaries target a 97 per cent confidence interval over a 2-day holding period; and all other products target a 95 per cent confidence interval over a 3-day holding period. The lower confidence intervals for the latter two groups reflect the difficulty of constructing reliable estimates of the extremities of the distributions of price movements for securities with limited price history and/or liquidity. However, longer holding periods for these securities are assumed. Backtesting seeks to verify that the flat rates for less liquid securities provide cover both to the target confidence interval and holding period at an individual security level, and to at least a 99.7 per cent confidence interval at the portfolio level (see CCP Standard 6.6).^[13]

Derivatives

For derivatives transactions, ASX Clear calculates initial margin requirements using the SPAN methodology. The SPAN methodology calculates initial margin requirements that reflect the total risk of each portfolio – for ASX Clear, each house or client account is considered a separate portfolio. The key parameters in the SPAN methodology are the ‘price scanning range’ (PSR) and ‘volatility scanning range’ (VSR). These scanning ranges are individually calibrated to the distribution of price and volatility movements for a set of related contracts under normal market conditions. The scanning ranges inform a set of 16 hypothetical risk scenarios used to measure the loss from a portfolio under alternative combinations of changes in price and volatility. For example, in one risk scenario, price increases by one-third of the PSR and volatility falls by the full VSR, while in another scenario price falls by the full PSR and volatility rises by the half of the VSR. The margin rate is then based on the highest estimated loss across the 16 scenarios.

ASX Clear bases the scanning ranges on key volatility statistics; namely, the higher of 2.75 standard deviations (a confidence interval of 99.7 per cent) of a 60-day or 252-day sample distribution, using the higher of one- or two-day price movements.^[14] The sample periods seek to balance incorporating recent market conditions with avoiding destabilising procyclical changes. The inclusion of two-day price movements reflects an assumption that a defaulter's positions may take up to two days to close out.

ASX Clear also applies a series of adjustments within SPAN to account for correlations and specific risks.

• Inter-commodity spread concession. ASX Clear also applies offsets designed to account for reliable and economically robust correlations across different contract types or across different stock option positions (see CCP Standard 6.5). These offsets reflect that, while the scanning risk for each related contract – a ‘combined commodity’ in SPAN terminology – is set based on the worst-case risk scenario for that combined commodity, it may be highly unlikely that the set of worst-case scenarios occurs simultaneously. This is particularly the case if a participant holds net long and net short positions in different related contracts that have a robust positive correlation, although net long/long and net short/short positions also have low level diversification offsets applied. The inter-commodity spread concession is calculated by applying (in a defined order) a spread ratio and concession rate to a participant's actual net positions in pairs of related contracts. The spread ratio determines the number of net positions in one related contract required to offset a position in another related contract. The concession rate is specified as a percentage of the scanning risk for both contracts in the pair. ASX calculates these parameters in the same manner as the price movement for the intra-commodity spread charge.
• Other adjustments. ASX Clear applies an adjustment to cover its exposure on the day of contract expiry, since expiring positions are otherwise not included in that day's initial margin calculations. ASX also maintains a minimum margin requirement on short positions to ensure the collection of margin on deep out-of-the-money options that would otherwise return no scanning range.

ASX targets the major inputs (including the PSR and VSR) to a 99.7 per cent confidence level. Other inputs are calibrated to the requirements of the standard to exceed a 99 per cent confidence interval.

To the extent that correlation assumptions (and hence offsets) across products in ASX Clear's margin models are conservative, diversified portfolios can provide additional protection for ASX. However, portfolios consisting of highly directional or concentrated positions do not benefit from the same degree of additional protection. In recognition of this, ASX Clear is developing risk-based margin add-ons that will apply to participants with highly concentrated exposures. ASX expects to implement these add-ons during 2017 (see Section 3.5.1.3).

Under ASX's internal Margin Standard, management discretion can be used if the application of the standard statistical analysis would result in inappropriate outcomes; for example, if the backward-looking statistical analysis does not take appropriate account of expected future price movements. Other reasons for using management discretion include insufficient historical data (e.g. where a product is new), seasonality in some products, and isolated spikes in price movements that result in a distortion of statistical recommendations. Where such a statistical override is made outside of a RQG approved margin review, it must be authorised by the Senior Manager of CRPM or the Manager of Exposure Risk Management with notification to the CRO and the General Manager, CRQ. The ASX Margin Standard also allows exceptions to the normal margin rate-setting process based on a broader risk assessment – where such exceptions are made outside of a RQG approved margin review, they require the approval of the Senior Manager of CRPM and the General Manager of CRQ.

6.4 A central counterparty should mark participant positions to market and collect variation margin at least daily to limit the build-up of current exposures. A central counterparty should have the authority and operational capacity to make intraday margin calls and payments, both scheduled and unscheduled, to participants.

Margin requirements are calculated overnight, with variation, mark-to-market, and premium margins based on closing prices each day. These are notified to participants the next morning. All margin obligations are settled via Austraclear and regular calls must be met by 10.30 am.

• For cash market transactions, mark-to-market margin is calculated in respect of securities in the All Ordinaries Index and added to initial margin if prices have moved against the participant. If prices have moved in favour of the participant then an offset may be applied to the participant's initial margin requirement, but this is capped by the level of initial margin. Mark-to-market margin is not called on flat rate-margined securities that are not within the All Ordinaries as up-to-date price data may not be available for all of these securities. However, adjustments are made to the overall margin called for these securities to more adequately cover ASX Clear's potential exposures in the case of adverse market movements.
• Variation margin is also levied on all LEPO and futures positions to reflect observed price movements. Variation margin is collected from the participant with a mark-to-market loss and passed through in cash to the participant with a mark-to-market gain.
• ASX Clear levies ‘premium’ margin on short ETO positions, updating this daily to reflect mark-to-market changes in the close-out price. Since premium margin is collected only on short positions, it is not passed through to participants with mark-to-market gains on long positions.

ASX Clear calculates intraday margin requirements on derivatives positions when market movements on derivatives positions exceed a defined threshold. That is, ASX calculates the net mark-to-market losses on all positions, and the initial margin on any new positions opened during the day. To determine if intraday margin is required, a nominal call amount is calculated for each portfolio of the participant (house and client) based on the combined initial and variation margin that would be due at the time of the intraday calculation. This is compared with the total margin posted by the participant. Where a participant's margin shortfall relative to the calculated requirement is greater than $100 000 and represents an erosion of initial margin of 40 per cent or more, ASX Clear calls intraday margin. This must be met by participants within two hours of notification.

Under ASX Clear's AIM methodology (discussed above in relation to CCP Standard 4), a participant is also required to post additional collateral should stress test outcomes reveal potential losses that exceed a predetermined STEL, or if participants have large portfolios relative to their capital (see CCP Standards 4.3 and 4.7).

If a margin payment is not made by the required time, ASX contacts the participant to determine the reasons for the delayed payment. Delayed payments are not common. When they do occur, they are typically the result of communication or technical issues involving the participant and/or its Payment Provider. Early communication by ASX aims to ensure that, in such cases, payment can still be made within a short period of the required time. In the event that the matter was more serious, ASX would investigate to decide whether a default event should be declared and, if so, how the default should be managed (see CCP Standard 12 and Section 4).

6.5 In calculating margin requirements, a central counterparty may allow offsets or reductions in required margin across products that it clears or between products that it and another central counterparty clear, if the risk of one product is significantly and reliably correlated with the risk of the other product. Where a central counterparty enters into a cross-margining arrangement with one or more other central counterparties, appropriate safeguards should be put in place and steps should be taken to harmonise overall risk management systems. Prior to entering into such an arrangement, a central counterparty should consult with the Reserve Bank.

ASX Clear applies margin at a portfolio (clearing participant) level for its cash market securities using its HSVaR methodology. This implicitly reduces the margin requirements for the participant where individual securities within the portfolio have displayed negatively correlated risks over the previous two years. The use of historical simulation over a two-year period establishes the significance and reliability of these correlations. ASX has identified that the direction of participant portfolios is a more significant contributor to the events captured in HSVaR margin settings than implicit price correlations, while flat rate margins do not assume any price correlations between securities. The significance and reliability of the correlations underlying the implicit offsets in HSVaR, which do not represent a significant proportion of ASX Clear's overall risk exposure, are subject to regular verification through backtesting and sensitivity analysis (see CCP Standard 6.6).

W In applying the SPAN methodology to derivatives transactions, ASX allows offsets in the form of ‘inter-commodity spread concessions’ (see CCP Standard 6.3). These offsets reduce margin requirements to account for reliable and economically robust correlations observed across related contracts. Inter-commodity spread concessions are applied where the correlation is significant and based on economic fundamentals. This concession is subject to a cap of 40 per cent for offsetting positions and 20 per cent for net long/long or net short/short positions. ASX uses sensitivity analysis to verify the reliability of assumed correlations between products used in calculating inter-commodity spread concessions (e.g. analysis of the effect of retaining stressed data from 2008 in the historical simulation period (see CCP Standard 6.6)). Changes to inter-commodity spread concessions must be approved by the RQG, which considers whether changes identified by SPAN appropriately reflect underlying economic relationships, including in periods of market stress.

ASX Clear does not have any cross-margining arrangements with any other CCP.

6.6 A central counterparty should analyse and monitor its model performance and overall margin coverage by conducting rigorous daily backtesting and at least monthly, and more frequent where appropriate, sensitivity analysis. A central counterparty should regularly conduct an assessment of the theoretical and empirical properties of its margin model for all products it clears. In conducting sensitivity analysis of the model's coverage, a central counterparty should take into account a wide range of parameters and assumptions that reflect possible market conditions, including the most volatile periods that have been experienced by the markets it serves and extreme changes in the correlations between prices.

Backtesting

Under ASX's Model Validation Standard, daily backtesting of both the SPAN and CMM margin models is used to test, on an ongoing basis, whether the margin models reliably cover price movements to at least a 99 per cent confidence interval. Daily backtesting is performed against both dynamic and static actual portfolios. Backtesting against actual dynamic portfolios involves the comparison of actual initial margin collected from a representative participant or client against actual variation margin calculated over the following one or two days, depending on which is the larger amount. One limitation of using variation margin on dynamic portfolios to model changes in the value of a portfolio over the holding period is that it is influenced not only by market movements but also by changes in the composition of the portfolio. To address the limitations of dynamic portfolio analysis, static portfolio backtests are also used to hold the portfolio composition constant over time. When static portfolios are used, ASX calculates hypothetical variation margin obligations for each day of the validation period based on historical price movements, and compares these to initial margin calculated on the static portfolio. The static portfolio used may be an actual portfolio held by a representative participant or client, or it may be a purely hypothetical portfolio; for example, one designed to examine the implications of directionality in positions or concentrations of exposures. Under both types of backtest, when variation margin is greater than initial margin an ‘exception’ is recorded. CRPM compares the number of exceptions to the expected number of exceptions, based on a 99.7 per cent confidence interval for static portfolios and a 99 per cent confidence interval for dynamic portfolios.

A report summarising the results of backtesting is automatically generated and circulated to relevant staff in the Risk division. Further analysis is undertaken when an exception is recorded, both to investigate model performance and to investigate the potential financial implications of the exception given the particular participant and portfolio affected. Where an exception is recorded against an individual client account, this investigation will proceed only if the dollar value of the exception breaches a materiality threshold. Further investigation also takes place if the actual number of exceptions exceeds the expected number. By investigating further, ASX determines whether any follow-up actions are required, such as the calling of additional margin or the managing down of positions.

Daily backtesting reports are aggregated into a monthly backtesting report which compares the number of observed exceptions to expected exceptions for the previous month, quarter and year. This report, which also includes the results of sensitivity analysis (see below) is reviewed by the RQG and used to identify the need for further investigation of margin model performance. RQG will take into account the frequency and magnitude of any breaches in determining whether to commission additional analysis from CRQ.

On a periodic basis, approximately every four months, ASX performs a more comprehensive backtesting analysis of each of its margin models. The periodic reviews allow ASX to examine the model in more detail and provide a basis for recommending changes to the model or further analysis. Hypothetical portfolios extend the analysis, allowing ASX to test the performance of margin models when applied to portfolios with certain characteristics (e.g. mix of contracts, concentrations, directionality) that may be particularly adversely affected by market conditions during the validation period.

Sensitivity analysis

ASX applies sensitivity analysis to its SPAN and HSVaR margin models on a monthly basis. Sensitivity analysis allows ASX to test the performance of a model beyond the boundaries of its existing assumptions, potentially also examining the implications of assumptions that would not reasonably be expected to hold. ASX has developed internal guidance setting out its approach to sensitivity analysis for margin models, which highlights three main assumptions that it varies when conducting sensitivity analysis: the confidence interval, holding period and look-back period. In addition, ASX investigates the impact of varying the historical simulation period for CMM, the impact of varying the key inputs and methodology used in determining inter-commodity spread concession rates, and the application of floors to model parameters in SPAN. If varying particular inputs reveals weaknesses in the model, as evidenced by a larger number of exceptions than expected, ASX considers whether to make adjustments to the model. Where sensitivity analysis identifies potential weaknesses in margin models, the RQG will consider recommended changes to address these.

6.7 A central counterparty should regularly review and validate its margin system.

ASX Clear's margin methodologies are also subject to a comprehensive annual validation and ongoing review under ASX's Model Validation Standard (see CCP Standard 4.5). The RQG is responsible for reviewing the regular reviews of models carried out by CRQ, while Internal Audit coordinates the independent validation process with CRQ input. ASX's Model Validation Standard requires that all models that are critical to ASX (as measured against a series of risk factors) undergo a full annual validation (see CCP Standard 2.6). Under the current framework the SPAN and DPS models must be validated annually using an external expert, while CMM must be validated once every two years using an external expert. During the 2014/15 Assessment period, ASX engaged external experts for a three-year period to conduct annual validations of ASX's key risk models, including both the SPAN and DPS models. The first validation of the CMM model and the second validations of the DPS and the SPAN model were completed in 2016. The Bank will continue to monitor the outcome of these validations.

At ASX, the margining process is governed by an internal Margin Policy and Margin Standard which is reviewed annually, with material changes approved by the Clearing Boards. The authorisation and documentation process for margin parameter changes and guidelines for the application of management discretion are also reviewed annually.

ASX publishes detailed margining information on its website, including descriptions of the margining methodology, schedules of margin rates, and daily SPAN margin parameter files. These files allow participants to perform margin calculations on hypothetical or actual portfolios. A number of third-party vendors use this information to provide margin estimation software to participants. ASX also maintains a web-based margin estimator that participants and their clients can use to calculate margin requirements on ETO positions.

6.8 In designing its margin system, a central counterparty should consider the operating hours of payment and settlement systems in the markets in which it operates.

ASX Clear's services are limited to CCP clearing of ASX-quoted cash securities and derivatives transactions executed on the ASX markets, as well as ASX- and non-ASX-quoted cash market securities transacted on AMO platforms under the TAS. ASX Clear's operating hours are consistent with the relevant payment and settlement systems (ASX Settlement, Austraclear and RITS).

Footnotes

Flat rates effectively assume independence of price movements between securities subject to flat rates. Unless a portfolio is highly concentrated in a small number of flat rate securities, it is likely that this assumption would lead to coverage at the portfolio level that exceeds the targeted confidence interval for individual securities. [13]

ASX assumes a normal distribution of prices in specifying its desired confidence level in standard deviations. [14]