FSN White Paper
Contents
Why is forecasting accuracy important?
External pressures on forecasting accuracy
Balancing financial and non-financial KPIs
The role of risk and probability in forecasting
The importance of enabling technology
The need for integrated planning, budgeting and forecasting
Cultural impediments and remuneration systems
Since the beginning of time, mankind has been obsessed by the ‘future’ and what it holds in store. Our desire to look beyond the horizon is driven by our need for certainty and a greater level of confidence that we are prepared for whatever life has to offer.
Biblical Pharaoh fretted about the meaning of his dreams, but armed with Joseph’s interpretation alerting him to “seven fat years” followed by “seven lean years” he was able to conserve grain in silos and protect ancient Egyptians from starvation. These days of course governments and businesses do not rely on dreams to set strategy yet there remains a nebulous quality to forecasting performance. “Aspirational targets” have entered into business vocabulary and many CFOs acknowledge that forecasting performance is more of an art than a science1. This begs the question can business learn from science?
The answer appears to be “Yes”. Recent research shows that businesses which take a more scientific approach to developing forecasts are more likely to get it right1. So what does science do that businesses eschew? Take some every day examples of scientific forecasts and the answer begins to become apparent. The weather forecast which manipulates more data and variables than the average business is explicit about risk and probability.
Phrases such as; “There is a 20 percent probability of precipitation in the afternoon” gives a fairly precise view of the likelihood of rainfall, or; “There is a 4 percent probability of a bird flu pandemic in the next decade with a 70 percent mortality rate” tells us in no-nonsense terms about the risks we face. Whether it is epidemiology or meteorology science couches its predictions in terms of risk and probability. Even a ‘hard’ science like physics, which is steeped in a tradition of direct observation and quantifiable data has come to appreciate the crucial role that probability theory has played in new discoveries, in areas such as particle physics.
By contrast business talks about risk assessment and probable outcomes but is rarely explicit about it. Earnings guidance may refer to potential downside or upside events but the probability of occurrence is rarely quantified.
But there is another reason to adopt best practice forecasting processes. In the face of increasing shareholder scrutiny and more onerous regulation around forecasting it is vital that Boards of Management can demonstrate a robust process supporting the generation of forecasts. Most market participants recognise that forecasting is inherently risky but they expect management to have taken reasonable steps in deriving their projections. A forecast that is supported by a dependable, auditable and documented process not only benefits the business but it also strengthens managements’ hands when outcomes are not in line with projections.
Why is forecasting accuracy important?
The most obvious answer is that it allows us to take action with a greater degree of confidence. Pharaoh filled his grain houses, weather forecasters tell us if we need an umbrella and epidemiologists inform governments whether to stockpile vaccines. But in a business context there are deeper reasons as well.
Forecasting appears to play a significant role in share valuation and reputation. Clearly it is not feasible to correlate share price and earnings forecasts with absolute precision since so many qualitative and quantitative factors contribute to share valuation, such as macroeconomic conditions and the performance of peer group companies. Nevertheless, recent research says that companies whose forecasts came within five percent of actual saw share price increases of 46 percent over the last three years compared with 34 percent for others1(an improvement of 35 percent).
With increased forecasting accuracy comes a reputation for a tightly run ship with good corporate governance, risk management and compliance (GRC). Although abstract in concept, reputation has real world consequences for the financial health of a company2. The benefits of a strong reputation include the ability to attract customers, employees and investment; to motivate employees and suppliers; and to differentiate the company from its competitors. A strong reputation also helps protect value, as it can lessen the impact of scrutiny, crises and competitive attack. Forecasting accuracy contributes to a strong business reputation.
On the other hand a surprise profit warning can leave a company undervalued and vulnerable to takeover, whilst a euphoric upgrade can leave management saddled with unrealistic expectations they cannot fulfil.
External pressures on forecasting accuracy
In an increasingly regulated and ‘stakeholder inclusive’ society the pressure is on for businesses to respond to a wide range of external interests. The view popularised by the economist Milton Friedman, that the sole concern of business is to deliver shareholder value appears considerably outdated.
“There is only one social responsibility of business – to use its resources and engage in activities designed to increase its profits so long as it stays within the rules of the game…if businesses do have a social responsibility other than maximum profits for their stakeholders, how are they to know what it is?”3
Regulators, governments and special interest groups have jettisoned this principle and imposed duties and obligations which extend well beyond the narrow interests of shareholders. The Accounts Modernisation Directive in Europe and Management Discussion and Analysis in the US have tilted financial reporting towards future guidance and broadened the agenda to reporting in novel areas such as the environment, employees and corporate social responsibility.
The broadening of disclosure requirements, allied to the increasing frequency of reporting driven by the SEC in the US and the EU Transparency Directive in Europe is pressuring the very core of financial processes, especially budgeting, planning and forecasting. A recent study4 suggests that rising compliance costs have halted the downward trend in finance departments’ running costs, at least for the top 1,000 global companies. Tighter regulation appears to have offset any recent gains from installing better systems or hiring smarter people, so that the average Global 1,000 companies’ spending on the finance function has risen 12 percent over the past three years.
When it comes to forecasting accuracy, commerce generally does not cover itself with glory. All recent surveys point in the same direction, namely that the majority of companies do not get anywhere near the mark. According to one study1, over the last three years only 1 percent of firms have hit forecast exactly and just 22 percent have come within five percent either way. On average, forecasts have been out by 13 percent (calculated as the mean absolute deviation from actual results). Executives in this survey estimate that such errors have directly knocked 6 percent off their share prices over the last three years, a significant part of which resulted from investor reaction.
A separate survey5 says that two out of every three companies are unable to accurately forecast earnings for the next quarter, missing the mark by anywhere between 6 percent and over 30 percent. It appears that companies do only slightly better when forecasting sales according to the same study. More than 50 percent of companies were unable to accurately forecast sales for the next quarter (accurate being defined as being within plus or minus 5 percent of actual results).
So why are companies so bad at forecasting their performance? Once again the evidence seems to point to several common threads, as follows;
- An over-reliance of spreadsheets which are error prone, difficult to maintain and provide little in the way of specialised functionality or process support.
- An inappropriate balance between leading and lagging indicators and between financial and non-financial measures.
- Failure to recognise formally the impact of risk and probabilities on potential outcomes.
- Lack of enabling technology with which to accelerate the process, enable strategic alignment, improve organisational reach and data quality.
- A disaggregated approach to forecasting which fails to align financial and operational plans so that resources can be optimised and contention between different demands resolved.
- Cultural impediments and remuneration systems which reward over achievement rather than forecast accuracy.
- The low uptake of rolling forecasting techniques which illuminate opportunities and risks beyond the normal year end horizon.
Balancing financial and non-financial KPIs
For most companies the focus of financial measurement is deeply rooted in traditional accounting techniques and it only in recent years that preparers and readers of financial statements have begun to question the value of accounts prepared using a technique developed for businesses more than five centuries ago.
At the heart of the debate is the realisation that strict financial measures merely provide information about past performance and do not necessarily provide a sound basis for extrapolating performance into the future. Yet increasingly, company valuations and shareholders’ willingness to invest are inextricably tied to future prospects.
More importantly, the market capitalisation of major companies around the globe is increasingly becoming uncoupled from their underlying net assets. The worth of companies recorded on the balance sheet often bears little relationship to their share price. The difference in these two valuations is largely attributable to intangibles, such as know how, intellectual property and brand – aspects of business performance that no accounting system is equipped to record and manage.
Unfortunately, traditional financial measures such as “profit”, “cash generated from operations”, and “revenues booked” so called, lagging indicators, provide little insight into future prospects and outcomes. In the search for more reliable harbingers of performance, business managers are turning to so called non-financial indicators or KPIs (Key Performance Indicators) that are often tightly correlated with future financial performance. For example, measures of customer satisfaction are often linked with a propensity to buy goods and services in the future. Similarly, measures around innovation, such as the percentage of sales derived from new products inform a company’s medium to longer term prospects for success. Likewise, employee commitment gives insights into future workforce attrition and, by implication, the ability to earn revenues in the future.
Despite this growing recognition that non-financial performance data is important, tracking it remains a problem for most. According to another survey6 while 87 percent of CEOs and senior executives describe their ability to track financial performance as excellent or good, just 29 percent of them describe their non-financial record as excellent or good.
The role of risk and probability in forecasting
Most businesses are familiar with the relationship between risk and reward but in assessing potential opportunities rarely acknowledge risks and probability in a formal way. That is not to say that there is not a role for intuition and experience, it is just that the record shows that of the businesses that regularly get within 5 percent of forecast, 51 percent of them perform scenario planning and sensitivity analysis compared to only 41 percent of the others1.
Even then, sensitivity analysis tends to be a limited exercise concentrating merely on three common scenarios, “best case”, “worse case” and something “in between”. Usually, worked up in a spreadsheet these scenarios tend to flex one or two variables (assumptions) at a time and provide little insight into probable outcomes which in the real world reflect the collective influence of many variables (good and bad) impacting at the same time. So is there a better way?
There is a growing appreciation of the value of mathematical techniques taken for granted in other forecasting environments but rarely applied in every day business strategy setting5. Engineers and scientists regularly make use of Monte Carlo simulation to refine their forecasts and set realistic expectations about the range of possible outcomes.
It works by modelling a number of business assumptions in parallel which each have an assigned range of input values and probabilities. For example, a house builder may decide to model “House Sales” based on assumptions about interest rates, inflation, and unemployment, setting for each of these variables the range of likely values and their probability. Monte Carlo simulation then uses random number generators in combination with various mathematical probability density functions to generate thousands of scenarios. By summing the scenarios, the simulation provides a forecast of the results expected from integrating the interactions of all of selected variables (interest rates, inflation, and unemployment in this case – but it could be more), presenting the probability of achieving each of the builder’s desired levels of “House Sales”. Furthermore, models of this type can identify the most influential (sensitive) factors to take into account. For example, bad news on unemployment may in the majority of cases outweigh any other considerations and management can focus on this as the key determinant in deciding whether or not to turn its land bank into new builds.
In minutes the house builder benefits from a comprehensive assessment of risk and a ‘real world’ simulation that could not be achieved by traditional financial forecasting techniques. The same technique can be applied to investment appraisal, cash flow forecasting, and portfolio optimisation.
Northrop Grumman finds that Monte Carlo simulation helps the business to manage uncertainty and improve forecast accuracy.
Northrop Grumman is a multi-$billion information systems integrator and managed services provider, rated by outside industry analysts as the number one government systems integrator and the premier provider of software and information technology services to the defense and intelligence sector.
To maintain its leadership position, the business has to continually invest in a variety of projects which ensure that it is endowed with leading edge capabilities, such as unique intellectual property and prototype systems that enable it to compete successfully for large government programs. But as Dr Robert Brammer, Vice President and Chief Technology Officer, Northrop Grumman Information Technology points out, traditional investment appraisal techniques are left wanting when it comes to managing multi-million dollar investment appraisal decisions in a complex, volatile and uncertain environment.
“In years past we used classic discounted cash flow techniques to compare investment opportunities, but the impact of probability on anticipated cash flows was only recognized in an informal and subjective way. For example, a bid manager might claim that a particular cash flow had a probability of 40 percent but there was nothing to back up the assertion other than his business judgment.”
Several years ago Brammer decided he needed a more objective and transparent approach to project appraisal which would lead to more robust and confident decision making. He introduced Monte Carlo simulation into the mix, a technique familiar to mathematicians and scientists but which had yet to make its mark in a business setting. However, the value of the technique quickly became apparent.
“We needed a better way of handling uncertainty in investment appraisal. The size and complexity of our programmes meant that probability distributions did not conform to normal probability distribution curves. We have to cope with multiple probability peaks and simultaneously with a very large numbers of project variables,” says Brammer.
What Brammer finds particularly valuable in Monte Carlo simulation is the availability of ‘Tornado’ diagrams which highlight the most significant variables, i.e. the ones that drive project sensitivity. “One cannot feasibly work with all of the variables but knowing which ones matter the most gives a better picture of the sensitivity of forecasts and therefore which variables to focus on.”
The technique has proved to be very accurate and over time Brammer has been able to demonstrate a correlation between early investments in R & D projects and the ability of the Group to win competitive tenders down the line. “If an investment costs a $1million but helps secure a contract of $500 million it is easy to demonstrate a return on investment,” he says.
Another benefit is that decision making stands up to scrutiny. “I have to be able to convince myself that an investment is justified as well as field questions from others. Classic discounted cash flow is subjective but using Monte Carlo simulation we can determine the probability distribution of winning a contract, the probability of the customer coming up with the funding and the probability of being able to agree terms and conditions.”
The success of Monte Carlo simulation is catching on and it is likely that the technique will be rolled out to other areas of financial forecasting - perhaps using the method directly in the bidding process.
The importance of enabling technology
Recent research points to the importance of enabling technology (i.e packaged applications not spreadsheets) in accurate forecasting1. Key determinants of forecast accuracy include forecast frequency, speed/ease of submission, multiple iterations and engagement with management at the sharp end.
Speed and frequency of forecasting are critical in a rapidly moving and volatile market. Budgets and infrequent re-forecasts are quickly overtaken by events and rendered meaningless. Forecasting is essentially a collaborative process and lends itself ideally to web based tools, supported by workflow. Earlier studies7 indicate that accelerating the process with specialised tools and allowing people to collaboratively review and approve forecasts leads to a greater number of iterations because each one is less time consuming and more productive. In turn, this higher level of challenge enhances the quality of the forecast. Again, of the businesses that regularly get within 5 percent of forecast, 58 percent of them update forecasts monthly or more often compared with 44 percent of the businesses that fail to get within 5 percent of forecast.
Similarly better performers are more likely, by a 14 percent margin to have forecasts conducted (more often) by operational managers who are closer to where the business takes place. The ease with which Web based tools can be deployed and their affordability seems to encourage best practice in forecasting. Although not explicitly referenced in any recent research it is clear that a driver-based approach to budgeting, which distils inputs into a simple range of parameters expressed in familiar business terms, shields end users from accounting complexity, assists data quality and accelerates the process.
The need for integrated business planning
Very few organisations truly achieve forecasting on an enterprise scale. Ideally, long term plans, budgets and forecasts should be vertically aligned so that strategic objectives and the related performance measures flow unremittingly from the very top of the organisation to the workforce at the sharp end of the business. This alignment has long been recognised as fundamental to communicating strategic intent and ensuring that managers make decisions with an appreciation of how they affect the execution of the strategy.
But there is another important dimension to budgeting, planning and forecasting which is often overlooked. This is the need to consider horizontal alignment across the different functional areas of the business so that plans in one area are consistent with another and performance measures are rationalised so that satisfying a performance objective in one place does not have unforeseen consequences in another. For example, a sales demand plan which projects increased volume and a different product mix needs to be tightly integrated with a production/capacity plan to confirm that demand can be satisfied. Both plans should be integrated with a financial plan that provides comfort, among other matters, that working capital is available to fund growth and assesses the implications of funding denied elsewhere.
Unfortunately the lack of appropriate enabling technology has encouraged businesses to budget and forecast in operational silos. Many rely on a patchwork of disconnected spreadsheets which often meansthat a financial forecast is derived separately from other critical areas. Ultimately nearly all forecasts have financial implications and thus integration is a pre-requisite of forecast accuracy. Similarly operational plans from different functional areas need to be integrated, for example, ensuring that a sales forecast is balanced with a production forecast, or that a recruitment plan for business development managers is consistent with a sales plan.
Integrating plans in this way not only requires a high level of collaboration but also impeccable data integrity so that the definition of metadata (structural information about, say, cost centres, dimensions, accounts and currencies) is consistent across the enterprise. Fortunately, advanced software products can centrally manage the metadata whilst supporting collaboration with workflow, commentary, and document attachments.
Monte Carlo simulation, mentioned earlier, also has a role to play in operational planning, for example, balancing manufacturing supply and demand, or optimising raw materials and capacity plans.
Integrated Business Planning needs robust enabling technology
The principle at the heart of integrated business planning is to join up the budgeting and forecasting efforts of different functional areas of an organisation in a concerted effort to achieve a co-ordinated and intellectually consistent plan. Although in principle this may sound straightforward, delivering a practical solution is full of process, people and technology challenges – a situation only too familiar to Bill Nienburg, who is responsible for Sales and Operations planning (S & OP)at Sara Lee, the global consumer products company.
S & OP is being recast as integrated business planning, spanning traditional demand planning, trade and promotion forecasting, financial planning and other functional areas where a common view of demand is critical. “One key to achieving an integrated plan is to ensure that the data structure is strictly governed by master data (information such as accounts, products and financial hierarchies), that the data structure is consistent applied across the various functional applications, and that the organizational design makes sense against that data structure,” says Nienburg. “And you need a strong toolset and process to manage this otherwise it can quickly get out of sync,” he adds. Nienburg recalls how master data was aligned when a new enterprise SAP application was implemented but that consistent review against business rules has been necessary to maintain the integrity of the data and therefore, the process.
Another critical component of integrated planning is the linkage with finance. “Gross Sales Volume in the constrained demand plan should flow and integrate with the equivalent Gross Sales Value in the financial plan. The right way to assemble the most accurate statement of demand volume, including base consumption, initiatives, new products, etc is through the S & OP process. It helps to drive the right visibility, transparency and alignment through your organisation”
But embedding the process takes time and Nienburg highlights the importance of recognising the critical nature of people issues in developing an enduring process. “Although people were signed up to an integrated planning approach and would attend joint planning meetings, complete adoption required ‘letting go’ of previous, less integrated planning and decision making processes. You must not underestimate the cultural and change management components,” he warns.
However, having faced these challenges Nienburg reports that Sara Lee is on the right path to greater functional alignment in the planning process, greater insight into the business and more pro-active business decisions. Although the S&OP has still has greater potential to drive business strategy, even tactical measures like forecast accuracy both at a national level (how much product to make) and local level (where to deploy it) are improving.
Reflecting on what has been achieved, Nienburg points out that bringing about change is the usual combination of people, process and technology, and that all are equally important. “Sometimes the importance of the toolset is understated in S&OP conversations,” he says. “People have to trust that the data is consistent and that the analyses are objective and accurate. You need good enabling technology to achieve this.”
Cultural impediments and remuneration systems
Naturally, modern systems alone are not a cast iron guarantee of forecasting accuracy. Achieving enduring improvements in forecasting requires a shift in mind-set as well. Few organisations measure and monitor forecast accuracy yet setting benchmarks is a vital tool in the context of continuous process improvement.
Furthermore corporate culture generally celebrates and remunerates over-achievement of budget rather than forecast accuracy. An honest appraisal of market conditions (positive and negative) affecting performance is absolutely critical if forecasting is to become reliable. The wide spread practice of “sandbagging” which focuses on setting soft targets which are easily exceeded simply promotes an inefficient allocation of resources. Recent research backs up this view and suggests that the best performing organisations (i.e. with actual results within 5 percent of forecast) incentivise their managers for forecast accuracy.
Finally, forecasting should not be regarded as the sole preserve of the finance function. Casting the net more widely allows managers, drawn from different functional areas to engage with the business and improve the quality of intelligence incorporated in forecasts.
The artificial horizon of the annual budget and attendant forecast revisions acts as a brake on forecasting accuracy. Rolling forecasts, which as the name suggests are based on a rolling 12 month timeframe are regarded as delivering superior accuracy for businesses facing constant change. 14 percent of all companies in a recent study5 characterised themselves as high risk/high volatility, a seven-fold increase in just three years.
But rolling forecasts should not be regarded as merely a mathematical extrapolation of the normal budgeting process beyond the year end. Reliability comes from a genuine appreciation of what drives value creation in the business, a willingness to focus on forecast accuracy and reward people accordingly. The effort seems to be well worthwhile. According to one survey1 over two thirds of organisations use rolling forecasts in some form and forty percent of them consider that they either already have or will increase forecasting confidence.
Forecasting in turbulent markets is a significant challenge with far reaching consequences for performance, resource allocation, reputation and shareholder value. Yet relatively straightforward measures can greatly improve forecasting success. Recognising the role of risk and probability is paramount in putting business forecasting and future guidance on the same footing as other fields, such as, meteorology and epidemiology. Equally important is the need to monitor forecast accuracy rather than “exceeding target” as a better measure of forecasting success.
However, many companies need to consider the fundamental basis of forecasting before embarking on process change, such as what factors really drive value creation and a more balanced approach to financial and non-financial measures, leading and lagging indicators.
But enterprise deployment is key, pushing the forecasting effort out to all functional areas and ensuring that it is supported by an integrated planning platform so that performance measures are applied consistently, financial and operational plans are combined and the consequence of a change in one area of the business is reflected automatically and accurately in all other forecasts.
In recent years, technology improvements have ensured that the building blocks for success are in place, but a holistic approach to organisational culture, process and technology is necessary if future guidance is to become consistently more accurate. Today’s enterprise performance management solutions provide robust planning, forecasting and modelling applications that can support web-based budgeting and planning, rolling forecasts, predictive modelling, and integrated business planning across functions. An integrated EPM system can eliminate an organization’s dependency on spreadsheets for its planning and forecasting process, provide more control and
confidence in the process, improved business predictability and the quality of guidance provided to internal and external stakeholders.
Bibliography
Note 1 Report: Forecasting with Confidence; Economist Intelligence Unit/KPMG 2007
Note 2 Report: The Rise and Rise of Non-Financial Reporting; MORI 2004
Note 3 Capitalism and Freedom; Milton Friedman 1962
Note 4 Hackett's 2007 Finance Book of Numbers™ research 2007; The Hackett Group
Note 5 Hackett’s 2008 Finance Book of Numbers™ research 2008; The Hackett Group
Note 6 Report: In the Dark 11 “What many Boards and executives still don’t know about the health of their companies, Deloitte 2007
Note 7Deloitte Survey of 2001 survey of budgeting and forecasting systems
Acknowledgements
Leading author; Gary Simon Group Publisher of FSN and Managing Editor of FSN Newswire
Ivo Bauerman, Senior Director Product Marketing, Oracle
Dr Robert Brammer, Vice President and Chief Technology Officer,
Northrop Grumman Information Technology
Larry Goldman, Senior Manager – Product Marketing, Oracle
Mike Malwitz, Director, Product Strategy, Oracle
Bill Nienburg, Sales and Operations planning (S & OP), Sara Lee
John O’Rourke, Senior Director, EPM Product Marketing, Oracle
About FSN
FSN Publishing Limited is an independent research, news and publishing organisation catering for the needs of the finance function. The report is written by Gary Simon, Group Publisher of FSN and Managing Editor of FSN Newswire. He is a graduate of London University, a Chartered Accountant and a Fellow of the British Computer Society with more than 23 years experience of implementing management and financial reporting systems. Formerly a partner in Deloitte for more than 16 years, he has led some of the most complex information management assignments for global enterprises in the private and public sector.
About Oracle
Oracle is a leader in Enterprise Performance Management (EPM), unifying Performance
Management applications and Business Intelligence (BI), supporting a broad range of strategic, financial and operational management processes. Oracle provides a complete and integrated system for managing and optimizing enterprise-wide performance. This allows organizations to achieve a state of management excellence, which provides competitive advantage and leverages their operational investments.
Whilst every attempt has been made to ensure that the information in this document is accurate and complete some typographical errors or technical inaccuracies may exist. This report is of a general nature and not intended to be specific to a particular set of circumstances. FSN Publishing Limited , the author and Oracle Corporation do not accept responsibility for any kind of loss resulting from the use of information contained in this document.
© FSN Publishing Limited. All rights reserved 2008.
