Product Analysis FAQ. October

Transcription

1 Product Analysis FAQ October

2 Contents Why are you presenting risk metrics to customers?... 2 How are products tested?... 2 What are the underlying assumptions of the test?... 2 What is your sampling size?... 2 How is the risk rating determined?... 2 What scale does the risk rating use?... 2 How do you calculate the return potential of a product?... 3 Do you consider liquidity risk?... 3 Do you consider default risk?... 3 Do you consider historical simulations?... 3 Is the risk rating appropriate for cross asset comparison?... 3 When is the research conducted?... 3 Why do you not use a pricing model to calculate probabilities?... 4 Do you stress test products?... 4 Do you update your methodology and assumptions?... 4 How does your testing fulfil regulatory requirements?... 4

3 Why are you presenting risk metrics to customers? For the benefit of transparency and to reduce the risk of information asymmetry. As the product manufacturer, we want to make sure that the presence of any material risks that we ascertain during the product development process are communicated to potential investors. This means that they can be confident in the fact that they have access to all the relevant information when they are considering products. How are products tested? All our retail products are subject to a rigorous product testing process that will involve a high level qualitative fundamental analysis of shares and indices, quantitative analysis of the risk/return profile of the product shapes, an analysis of issuer risk and a technical analysis of potential risk mitigation in stress testing environments. What are the underlying assumptions of the test? The forecasts on which the risk rating, and much of our product development statistics are based on, rely on the statistical procedure known as bootstrapping or random sampling with replacement. There are many other methods of forecasting used in the financial and actuarial world but this is a widely accepted statistical inference methodology utilising relatively simple mathematics. The idea is that we try to infer the potential returns of a market variable (in our case, this is likely to be share prices or levels of an index) by modelling a sample of the data. To model the potential returns of a product for example, we would take a sample of the historical daily returns of the relevant underlying shares/indices and then use that data to create many hypothetical future paths for the underlying shares/indices. We then see how the product shape performs based on each of these paths. The more simulations we do, the more consistent our results. Imagine a game where a hat contains 9 red balls and 1 blue ball but the players do not know this and they want to find the probability of picking a red ball. They are only allowed to pick and return one ball at a time. By performing this task many times we would expect that the frequency at which a red ball is chosen will converge on to the true 90% probability. The more repetitions, the more confident the players will become in stating their assessment of the probability. In our tests, we produce 10,000 independent forecast simulations. This number is sufficient enough to see immaterial variance in the statistics. An increasing number of simulations will not increase the accuracy of the forecast as the mean and variance of the returns distribution remains the same. What is your sampling size? For our forecasts, we use the historical 5 year daily returns of the relevant underlying shares/indices. Arguably, this sample, being some of the most recent, could be the most relevant and realistic for inferring future returns. Larger samples incorporate more information but smaller samples give the most relevant innovations. In addition, we have used 5 years to be consistent with European standards and to increase synergy with other investment classes. By using the same sample size as the official SRRI for funds, we maintain the prevailing volatility of the past 5 years. Indeed, when we use our model, for example, on the historical price data of funds to our modelling process, we get a risk rating that closely approximates the official quoted SRRI in the KIID (Key Investor Information Document) used by funds. How is the risk rating determined? The risk rating is based on the volatility of the simulated product returns. The higher the volatility, the higher the risk rating. Volatility is a measure of the dispersion of a returns series. A common assumption is that the higher the volatility, the riskier an investment. This is because volatility refers to the amount of uncertainty in the size of the potential changes in the value of an investment. A high volatility means that there is a potentially large range in the extent of change in value of an investment. We would expect a high volatility product to have the ability to either rise or fall in value in relatively large amounts. A low volatility means that the range of changes will be relatively smaller. We would then expect changes in value to be relatively more consistent. What scale does the risk rating use? For the benefit of familiarity and consistency when compared to other investments, we assign our ratings based on the Synthetic Risk and Return Indicator (SRRI) established by the European Securities and Markets Authority (ESMA). The SRRI scale is a mandatory element of the Key Investor Information Document (KIID) and aims to demonstrate the potential risk and reward profile of a product in one simply presented indicator. Because fund

4 providers/managers are obligated to provide a KIID, both experienced advisers and retail investors in funds alike should be familiar with the SRRI. How do you calculate the return potential of a product? Expected returns are calculated on a standardised per annum basis to increase comparability and transparency. Returns are quoted on a compound annual growth rate basis. For example, consider a product that pays 10% for each year the product has been in force and allows for early maturity. If the product matures early after two years this would lead to a 20% holding period return. This is equivalent to 9.545% per annum. This is because an investment with a growth rate of 9.545% per annum would grow to a 20% holding period return in two years. A 1000 investment for example, would earn in the first year and be worth at the end of that year. At the end of the second year would be earned from the and so the total holding period value would have risen to Because structured products allow for different maturity scenarios, we need a method to account for both the weighted probability of certain payoff conditions occurring as well as the weighted probability of different maturity schedules occurring. Our expected return is therefore a function of the average payoff simulated and the average time to maturity simulated. Do you consider liquidity risk? The simulations used in our analysis and product development do not account for liquidity risk. Liquidity is often considered a product feature rather than a risk. For example, many of our products allow for weekly early encashment, subject to the counterparty being able to make a market for the transaction. Do you consider default risk? The simulations used in our analysis and product development do not account for default risk. Including default risk into our risk/return metrics would skew the presentation of market risk, which is arguably the most important risk when considering a product shape. Default risk, however, should be a fundamental consideration and is therefore presented too, albeit separately from market risk measures. Information on credit ratings and default probabilities can be found in the retail facing and adviser facing materials. Do you consider historical simulations? Yes. We consider the historical simulations in our product development and also provide the results to advisers in our adviser facing materials. Is the risk rating appropriate for cross asset comparison? The sample size used in our forecasts are based on historical returns in the past 5 years. The sample size used in the official SRRI risk rating currently being used in the KIID are based on the historical returns in the past 5 years. In our sanity testing of the model, we subjected the historical price data of some popular funds to our forecasting model. We obtained risk ratings that closely approximated the SRRI published in their official fund literature. It should be remembered however, that the methodology we use for obtaining the volatility and the methodology used in the official ESMA methodology for obtaining the volatility are fundamentally different. This is because the official methodology cannot be applied to structured products, as there is no historical price data from which to calculate the historical volatility of returns. The risk rating we show therefore, may not necessarily be appropriate for a direct comparison to the risk ratings you might find in a KIID. The choice to treat them as an equivalent volatility measure, however, is up to your discretion. When is the research conducted? Research is always conducted on products prior to them being launched. Only when a product is deemed to have passed our strict criteria do we trade a product and market it for retail distribution. The information we provide to investors and advisers is the same information that we use to judge the return potential and value for money in a product.

5 Why do you not use a pricing model to calculate probabilities? The Financial Conduct Authority (FCA) Thematic Review of Structured Products in March 2015 highlighted the fact that some of the selection and calibration of modelling approaches did not reflect the statistical properties of prices observed in the real world and that some firms used unrealistic and/or optimistic growth rates. Using derivatives pricing models, we can calculate the fair value of a security; interestingly for us, we would be able to calculate the price of a structured product based on prevailing market conditions. The price is a function of said market inputs and the resultant probability and present value of the potential payoffs. The model however, is based upon an environment which assumes risk-neutrality. Such an environment only exists when the market is arbitragefree and when investors demand no risk premium. All assets in this environment have the same expected return; the risk free rate. The constructed probability distributions from these models therefore, are artificial and do not show the real world probability of events occurring. Accordingly, we do not believe that these models are reflective of the real world. Do you stress test products? Yes. We simulate the FTSE 100 Total Return Index as a proxy for general market equity performance alongside the product simulations. After our normal testing is completed, we filter the simulations in which the market proxy was simulated to have produced a negative return. This subset of simulations is a sample in which we can analyse the performance of the product when faced with adverse market conditions. Do you update your methodology and assumptions? Yes. The industry is in a state of transition at the moment and there is no standard for the calculation of risk in structured products. We regularly review our methodology and will change our processes if we find it beneficial for our potential investors. In particular, any future standards provided by regulatory bodies or trade bodies may prompt changes. How does your testing fulfil regulatory requirements? The Financial Conduct Authority (FCA) Thematic Review of Structured Products in March 2015 highlighted the fact that products need to be stress tested and modelled for their return potential. The key is that the product development process should analyse a product s value for money, potential risks and resilience. For a number of years now, we have utilised a product development procedure that produces the above statistics. The process involves sufficiently demanding criteria and is constantly being monitored for its robustness.