GDPcast

ENABLING DATA DRIVEN DECISION MAKING

MEASURING ECONOMICACTIVITY CONTINUOUSLY

ABOUT USCheckRisk is a trusted provider of risk services to over $70bn of risk assets globally. CheckRisk’s team of investment professionals has decades of combined experience within asset management, trading and qualitative/quantitative modelling disciplines. Our consulting services, risk models and commentaries provide a high level of value to any firm that is interested in having an independent view to aid decision making, formulating risk strategies, to put in place risk management systems or manage assets. Our work with the University of Bath, School of Management, and University of Bristol Systems Engineering Department means that investors wishing to understand risk have the backing of top level academic institutions and the pragmatic investment risk skills of CheckRisk’s team.

OUR APPROACHtrying to answer is: are you being paid to take risk?

Computers and new theories have transformed the way in which we can try and answer that question. We bring together the latest forward-looking methods and translate them into output that is easy to understand.

We recognise that there is no single ‘perfect’ approach and that there will always be a certain amount of irreducible uncertainty: this is why you need to use simulations and different techniques to build resilience. This offers us the best chance of achieving our objectives.

However, we should not leave it to

computers alone. We believe that the future lies in combining their power with practitioner experience. Working closely with our clients, we have been successful in customizing our risk knowledge to provide specific client solutions.

The cross-fertilisation of the sciences and big data offers us new and exciting insights into how the world works. CheckRisk is at the leading edge of this new frontier.

Our approach to risk is innovative, forward-looking and multi-dimensional. We think of risk in the widest possible sense: from the macro global economy, down to the micro securities level, using both quantitative and qualitative analysis.

Legacy systems such as Value at Risk (VaR) are kn to have failings and have not served the industry well. We aim to fill that gap.

Most investors spend their time chasing the thing they can’t control: returns. They spend little time thinking about the things they can control that make a difference: risk, cost and time.

The question CheckRisk is really

2

3

OUR SERVICES

INVESTMENT RISK GOVERNANCE

RISK PRODUCTS RISK MANAGEMENT CONSULTANCY

STRUCTURED PRODUCTS

We advise professional investors, boards and trustees on one of the most important and least understood risks: investment risk and its proper governance.

We do this by assisting our clients in reviewing their existing investment processes and looking for improvement in areas such as asset allocation, stress-testing and model validation.

We help decision makers get a clearer picture on the global economy and financial markets with our risk products.

We have off-the-shelf products and fully customisable systems to suit individual needs.

Our cutting edge products are primarily focused on forecasting financial conditions and reflect the interconnected world we live in.

We provide bespoke, cost -efficient risk solutions to assist institutions in managing financial and operational risk.

We partner with our clients to help them deliver specific projects for internal, commercial or marketing purposes.

Our experience and modelling skills are useful in different parts of an organisation and across industries.

CheckRisk creates bespoke structured products for institutional clients. These provide a diversified alternative to traditional asset classes, especially in an environment of low interest rates.

We adopt a multi-strategy approach and seek to create the best possible product. We also provide full ongoing after sales support.

GDPCASTSFOCUSING ON ECONOMIC GROWTHCheckRisk GDPcast indices help you to get a highly accurate picture of current economic growth. This is particularly useful when the official GDP figure is produced quarterly.

The high level indices we produce are rich with information and offer a clearer picture that can be missed with the huge number of data points that are released each month. Alerts and reporting can be set with CheckRisk so that our clients are kept up to date as events unfold.

The charts and dashboards are visually powerful and can be understood by anyone in an organisation.

The real time forecast model CheckRisk employs enables us to forecast economic growth, and produce probabilities of recession and strong growth at any given moment in time for any given country or region.

KEY FEATURESIdentify emerging risks and opportunities in different regions with statistical confidence

Receive high level information, or explore the detail. See the rates of change. Set the alerts and explanatory reporting from CheckRisk

Communicate risks in a visual language that is easily understood by quants and senior managers alike

Use the output to develop predictive stress scenarios that CheckRisk can integrate with your portfolio and risk management systems

CheckRisk GDPcast Indices filter signals from noise from numerous indicators, dealing with mixed frequency data and asymmetric start and end dates

Asset Managers

Hedge Funds

Wealth Managers

Pension Funds

WHO IS IT FOR?Regulators

Banks

Insurers

4

CheckRisk GDPcast: US Real Time Estimate Underlying Economic Activity

CheckRisk GDPcast: US Real Time Estimate Underlying Economic Activity MoM %

BACKGROUND

There are two main types of Real time models that have differing goals:

1. GDP Tracking Models - like the Atlanta Fed GDPNow models try to calculate the GDP figure prior to its release for a given quarter.

2. Dynamic Factor Models (DFMs) that aim to continuously measure underlying economic activity from numerous economic time series.

These models are based on the work of Evans (2005) and Giannone, Reichlin and Small (2008). The CheckRisk model and the New York Fed model fall into this latter category, although the NY Fed only present results for the current quarter.

The CheckRisk model does not try to forecast GDP specifically as a result, but rather a continuous measure of underlying economic activity. The beauty of this is, therefore, that you can get a snapshot of economic growth at any moment in time.

It is important to understand the differences between the models as there can be quite significant variations in the output.

The GDP tracking type models aggregate key elements in the economic data to produce an estimate of GDP itself. However, as the quarter proceeds to the GDP announcement the model places greater weight on out-of-date data skewing the result.

GDPcast, CheckRisk’s real time forecast model, is calculated daily and tries to rapidly capture any changes in the business cycle. It does this by filtering out the noise and finding the best combination of factors to determine what is driving economic growth. The result is an output that is more stable than GDP and far less prone to revision.

5

“We take a lot of variables and turn them into one accurate signal for a single country

”

The commonly used industry measure to assess economic activity is GDP (Gross Domestic Product). This is a form of total national income and acts as a proxy for ‘true’ economic growth. It is calculated on a quarterly basis by various governmental agencies in their native countries, and also by several non-governmental economic bodies, such as the IMF and OECD.

GDP is an imperfect measure as it is subject to volatility and revision. Nonetheless, it is the most widely used. Real time forecasting models aim to fill the gap that GDP leaves. They have become an increasingly popular method for investors and policymakers seeking to understand the latest fluctuations in economic activity.

The output of real time forecasting models can be particularly useful in understanding the movements in financial asset prices, which can be subject to sudden shifts.

TYPES OF REAL TIME MODELS

HOW DOES IT WORK?

Traditional models that forecast economic growth tend to rely on ‘hard’ pieces of data that are released monthly or quarterly like imports and exports, consumption, labour productivity, and unemployment figures.

Real time forecasting models are not designed to replace these models per se. However, academic literature and our own testing confirm that they tend to be more accurate than the traditional approach up to a forecast period of six to nine months.

The traditional DSGE (dynamic stochastic general equilibrium) models are often calibrated with manual inputs that reflect an anticipated policy response. This is a useful feature in creating a realistic forecast.

Real time forecasting models are not designed to do this, but they do incorporate the large amount of ‘soft’ survey data which is released every month that the standard models do not.

It is clear that risk assets respond to this information to establish pricing in the market environment. The academic literature also indicates that survey data can be used to create a superior forecast.

Real time forecasting methodology, therefore, reduces the amount of noise that decision-makers are subject to, and frees them up from devoting too much attention to a single data point.

It does this by performing a form of trend and cycle decomposition: establishing the common

longer-term factors that drive economic growth and the shorter-term cyclical elements that are having an effect now, and that will likely have an impact in the immediate future. The model uses a sophisticated form of filtering to combine the mixed frequency data together resulting in a single signal.

We use a form of vector autoregression model to establish confidence intervals and create forecasts about the future.

The panel of data below is an example for the US. This is illustrative, although each country’s ‘soft’ data does vary.

6

Monthly ‘Hard’ Indicators

Real Personal IncomeIndustrial ProductionNew Order of Capital GoodsReal Retail Sales & Food ServicesLight Weight Vehicle SalesReal Exports of GoodsReal Imports of GoodsBuilding PermitsHousing StartsNew Home SalesPayroll Empl. (Establishment Surv.)Civilian Empl. (Household Surv.)UnemployedInitial Claims for Unempl. Insurance

Monthly ‘Soft’ Indicators

Markit Manufacturing (PMI)ISM Manufacturing PMIISM Non-Manufacturing PMINFIB: Small Business OptimismU. of Michigan Consumer SentimentConf. Board: Consumer ConfidenceEmpire State Manufacturing SurveyRichmond Fed Manufacturing Surv.Chicago PMIPhiladelphia Fed Business OutlookSentix US Economic Expectations

Quarterly Time Series

Real GDPReal Consumption (exc. durables)Real Investment (inc. durable cons.)Total Hours Worked

MODEL DESCRIPTIONCheckRisk’s instantaneous forecast is achieved by the process of state-space based dynamic factor modelling which measures the co-movement of many time series and identifies the underlying component representative of GDP and forecasts forward in time taking into account a variable long-run growth rate and recent stochastic volatilty.

“We use a state-space dynamic factor model and maximum likelihood estimation for the real time forecast

”

7

1. Data CleansingThe model is produced by obtaining and transforming a panel of mixed frequency hard and soft indicators. These indicators are transformed to obtained QoQ or MoM % differences and where relevant are seasonally adjusted.

2. Model SpecificationThe model is formally specified to represent three components:

• time varying long-term growth• latent underlying factor• stochastic variability

The time varying long-term growth is represented as a time-varying intercept. The latent common factor is represented by a set of factor loadings that map onto lags of the underlying common factor. Similarly, stochastic volatility is represented by a set of factor loadings that map onto lags of the underlying stochastic volatility factor. For robustness, the dynamics of the model’s time-varying parameters are modelled as small variance driftless random walks thus providing confidence bands and uncertainty about future breaks or drifts in long-run growth.

3. Model CalibrationThe model is calibrated to the panel of mixed frequency indicators using maximum likelihood estimation. This calibration process identifies:

• the loadings on the underlying common factor • variance parameters for long-term growth• variance parameter for latent common factor• variance parameter for stochastic variability• variance parameters of observed mixed frequency indicators

4. Estimation of Economic ActivityOnce calibrated, the model uses forward filtering and backward smoothing to estimate the state of the underlying factors and their uncertainty given real-time data, and then forecasts future values and confidence intervals for underlying economic activity.

THE PRODUCT

CRGDP CRGDP+

If you purchase access to the CheckRisk GDPcast Indices (CRGDP) you will receive:

• A monthly report• A quarterly report• Share in any model improvements• Ability to set e-mail alerts

The monthly report is a brief synopsis of events, and informs the reader of any material changes in the underlying indices and regions. You will receive this by email in PDF format.

The quarterly report goes into much more depth, both in terms of data and narrative. Not only do we seek to explain the changes in risk across regions in a quantitative way, but we also apply our subjective opinion.

We are constantly researching ways of improving what we do. We will automatically update our models and share the improvements with you. We are also open to feedback as to how we can improve our service.

The regions covered in CRGDP are the US, Europe, China, Japan and the UK. We are developing further indices for the emerging markets.

Please contact us if you would like to set specific alerts and we will email you if there are any breaches between reporting cycles.

Cost: £10,000 per annum

Many of our clients involved in investment management and trading require a more sophisticated approach. For this reason, we have expanded the standard CRGDP and have created CRGDP+.

If you purchase the enchanced CheckRisk GDPcast Indices (CRGDP+) you will receive the standard service, but also with the ability to fully customize our models to your objectives.

The standard CRGDP is sufficient for a large proportion of our clients. However, if you have a specific requirement e.g. to forecast 5m economic growth - this requires ongoing calibration.

You may be interested in creating trading strategies based off our models.

We will share the model testing statistics with you on an ongoing basis, to aid your quantitative teams. We can also collaborate on the most suitable delivery method for you e.g. API, business intelligence etc.

We can also add further countries to the standard list in CRGDP.

The cost is therefore dependent on the complexity of your needs. Please contact us to discuss a quote.

Cost: POA

1 2

8

RESULTSThe results below show the performance of the CheckRisk GDPcast model versus the Atlanta Fed GDPNow model over the period May 2014 - June 2016. They show that the Atlanta Fed model tends to over-estimate the amount of economic growth.

The CheckRisk model has a much lower estimation error over both one-month and three-month time periods.

9

”The model also allows us to add levels of confidence to our forecasts, aswell as a base prediction. This enables us to produce a probability of either strong growth or recession ahead. This can be instructive for the behaviour of risk assets.

The results below show the performance of the CheckRisk GDPcast model versus the Atlanta Fed GDPNow model over the period May 2014 - June 2016. They show that the Atlanta Fed model tends to over-estimate the amount of economic growth.

The CheckRisk model has a much lower estimation error over both one-month and three-month time periods.

“CheckRisk GDPcasts show high accuracy

CheckRisk GDPcast Estimation Accuracy - Forecasting 3 Months CheckRisk GDPcast Estimation Accuracy - Forecasting 1 Month

CheckRisk GDPcast: US Real Time Estimate Underlying Economic Activity

9

The model also allows us to add levels of confidence to our forecasts, aswell as a base prediction. This enables us to produce a probability of either strong growth or recession ahead. This can be instructive for the behaviour of risk assets.

CheckRisk GDPcast: Negative Growth and Strong Expansion MoM %

CheckRisk GDPcast: US Real Time Estimate Underlying Economic Activity

FURTHER OUTPUTSCheckRisk can produce instantaneous forecasts for any country, although initially we have constructed these for the major regions. We are the only provider that can provide full access to this information in the public arena at present.

The probability of strong expansion in the Netherlands is negligible because the rate of growth has been near 0.5% MoM for a prolonged period of time.

We have witnessed a softening of growth across Europe and the advanced economies since the second quarter. Momentum has stalled. The emerging market economies, however, have accelerated to above trend levels. Overall, world growth is very close to its long-term trend.

“Growth has been slow in the Netherlands at 0.5% MoM

”

CheckRisk GDPcast: Netherlands Real Underlying Economic Activity MoM% CheckRisk GDPcast: Netherlands Real Time Estimate Underlying Economic Activity

CheckRisk GDPcast: Negative Growth and Strong Expansion MoM % Netherlands - probability of current quarter underlying activity

This document is provided to you on a confidential basis for your information and discussion only. It is not a solicitation or an offer to buy or sell any security or other financial instrument. Any information including facts, opinions or quotations, may be condensed or summarized and is expressed as of the date of writing. The information may change without notice and CheckRisk LLP (“CheckRisk”) is under no obligation to ensure that such updates are brought to your attention.

The price and value of investments mentioned and any income that might accrue could fall or rise or fluctuate. Past performance is not a guide to future performance. If an investment is denominated in a currency other than the base currency, changes in the rate of exchange may have an adverse effect on value, price or income. This document and any related recommendations or strategies may not be suitable for you; you should ensure that you fully understand the potential risks and rewards and independently determine that it is suitable for your given objectives, experience, financial resources and any other relevant circumstances. You should consult with such advisor(s) as you consider necessary to assist you in making these determinations. Nothing in this document constitutes investment, legal, accounting or tax advice, or a representation that any investment or strategy is suitable or appropriate to your individual circumstances, or otherwise constitutes a personal recommendation to you.

This document may relate to investments or services of an entity/person outside the UK, or to other matters which are not regulated by the FCA, or in respect of which the protections of the FCA for retail clients and/or the UK Financial Services Compensation Scheme may not be available. Further details as to where this may be the case are available on request in respect to this document. Certain information contained herein has been obtained from third party sources. While such information is believed to be reliable for the purposes used herein, no representations are made to the accuracy or completeness thereof and CheckRisk takes no responsibility for such information.

This document is intended only for the person to whom it is issued by CheckRisk. It may not be reproduced either in whole, or in part, without CheckRisk’s written permission. The distribution of this document and the offer and sale of the investment in certain jurisdictions may be forbidden or restricted by law or regulation.

Investments may have no public market or only a restricted secondary market. Where a secondary market exists, it is not possible to predict the price at which investments will trade in the market or whether such market will be liquid or illiquid. As such investments will not be listed or traded on any exchange, pricing information may be more difficult to obtain and the liquidity of the investments may be adversely affected. A holder may be able to realise value prior to an investment’s maturity date only at a price in an available secondary market.

CheckRisk LLP is authorised and regulated by the Financial Conduct Authority (Firm reference number: 490575). The registered address of CheckRisk LLP is 4 Miles’s Buildings, George Street, Bath, BA1 2QS (Company number: OC338765).

DISCLAIMER

info@check-risk.com01225 422 583

4 Miles’ BuildingsBathUK

BA1 2QS

CheckRisk LLP is authorised and regulated by the Financial Conduct Authority (Firm reference number: 490575)

SIMPLIFYING THE COMPLEXITY

OF INVESTMENT RISK