Beyond the crystal ball
March 2009
The accuracy of forecasting can be enhanced by range forecasting. Caroline Stockmann demonstrates how...
It seems to be an accepted truth that the corporate world is very different from the charity/not-for-profit sector, but is it really that different when it comes to managing the finances?
Coming from a corporate, blue chip background, but with a passion for the arts (as a classically trained musician), I have come to the sector with a fascination for which practices can successfully be applied here, and which will simply fail. One key area where learnings seem to transfer effectively is that of forecasting, and in particular range forecasting.
Range forecasting has added significant value to Southbank Centre, and is shortly to be adopted by Sue Ryder Care. It can make a huge difference to the accuracy of forecasting, and stimulates discussion which can lead on to even more value-adding activity. It can also be introduced in a very simple, low-tech way, which needs no further investment.
Budgets v forecasts
Organisations can be at many different stages in their sophistication as far as financial planning is concerned. One conclusion I have drawn in all the organisations for which I have worked is that budgets and planning in general can have limited value. Sometimes one spends so long trying to explain differences versus a budget that was out of date the moment it was written that the organisation forgets to run its business and focus on necessary actions. Whereas departmental overheads budgets do have their purpose, the overall financial reporting may be better served with comparators versus prior year, complemented by forecasts looking at the most likely outcomes going forward, versus target.
Further, budgets can cause all kinds of behavioural issues which need to be managed. As former General Electric boss Jack Welch observed: ‘The budget never should have existed... if you make it you generally get a pat on the back... if you miss it you get a stick in the eye or worse... Making a budget is an exercise in minimalisation. You’re trying to get the lowest out of people... everyone negotiates to get the lowest number.’
Put another way, have you ever noticed any of the following:
- ‘lowballing’ of targets?
- building ‘sleeves’?
- negotiating bigger budgets?
- treating budgets as an ‘entitlement to spend’?
- resisting initiatives as they increase the target?
- starting initiatives to increase targets?
- moving money between years?
- cutting investment to meet targets?
- hiding good news for fear of increasing targets?... etc...
There are alternatives to the ‘classic budget’, which have been the study of the Beyond Budgeting Round Table [1], and which also formed part of my work at Unilever in developing and implementing what we called ‘Dynamic Performance Management’ (DPM). There our aim was to move from a set of fixed annual plans, typically arrived at by a process of negotiation, to a set of processes which helped the business to continuously adapt to a changing environment and where the key measure of performance was competitive success.
Southbank Centre is now exploring the principles of DPM and rolling forecasts, which could be the perfect planning solution for an organisation with multiple and different lead times for its artistic programme. A rolling forecast deals with the fact that activities do not naturally stop at our financial year-end and encourages management to think about opportunities and risks in a pro-active way.
However today I will concentrate on forecasting accuracy, with or without budgets, and how range forecasting
can transform the organisation’s approach and indeed have a significant de-risking impact.
can transform the organisation’s approach and indeed have a significant de-risking impact.
Forecasting and range forecasting
There are of course a number of issues around forecasting, which can impact its accuracy (leaving aside acts of God/terrorism etc!), which include:
- politics (personal or general, which can encourage forecasts which significantly deviate from reality);
- cultural tendencies (in terms of under or over-forecasting); and
- lack of/inappropriateness of supporting systems.
Range forecasting and clear forecasting guidelines are of help here, and I will give some specific examples.
Range forecasts describe possible outcomes around a single point forecast, with a specified likelihood of occurrence. For example, Unilever would normally specifically identify ranges with 90 per cent confidence levels, which means that in the judgment of management the outcome will fall within the range nine times out of 10. At Southbank Centre we do this more intuitively, trying to estimate ‘absolute worst’ and ‘absolute best’ case in each area of the income and expenditure statement – which, in most cases, we believe equates to a similar (90 per cent) confidence level.
Forecasts must also be unbiased – i.e. the probability of falling short should equal the probability of overshooting (being 5 per cent at this confidence level). The central (or ‘most likely’, as we call it here) forecast is where the likelihood of the actual outcome being above it equals that of being below, but this does not mean therefore that this forecast is in the middle of the range, i.e. the range is more often than not ‘skewed’(see figure 1 below).
Figure 1: Forecast of total profit to 2015
How to create a range forecast
A very practical approach is to work out the forecast through taking each level of the income and expenditure statement and, where appropriate, going down to a more detailed level (which can then be consolidated upwards). The forecast process should start with identifying the high and low points, and then the ‘most likely’ forecast can be estimated. I have never started with the central point then filled out the range, as one might be tempted to do. This ensures that there is an open discussion on all aspects and frees up the mind so as not to get ‘fixed’ on a particular outcome. Standard advice also is to beware of ‘experts’, who often give too narrow a range, due to (a natural) overconfidence; in my experience this has been proven again and again to be good advice.
Behind the discussion there may also be modelling and other detail to support the range proposed (say, marketing department modelling the best and worst ticket sales based on historical and consumer data, or development looking at recent trends of corporate sponsorship). These can be developed with simple tables and Excel analyses, or via the various software packages which support these areas of activity.
One can go into even greater depth and use statistical modelling, more appropriate for the large corporate, where there is a lot of data to crunch and the return on investment is justified. One type of modelling would be ‘Monte Carlo’ simulation; a spreadsheet simulation which randomly generates values for uncertain variables over and over to simulate a model. The uncertain variables in the model are then represented by a distribution. The simulation calculates multiple scenarios of a model by repeatedly sampling values from the probability distributions for the uncertain variables, and using those values for the cell, which then recalculates the spreadsheet.
The biggest impact of this kind of probabilistic analysis on finance is that it removes our obsession on getting to the ‘right number’, and forces us to really focus on the edges – i.e. how do we minimise the downside and maximise the upside (by shifting the curve, or changing the shape of the distribution)? This is where most discussions ought to be focused – not only arriving at a forecast, but getting us to think about the various options that we can draw from when the need arises, in order to react appropriately to our forecast.
There are a number of Excel add-ins/tools which are available to create Monte Carlo simulations, and these can easily be found via the Internet.
Another tool which many will be familiar with is the use of a ‘decision tree’. This enables you to choose from a number of different actions, where choices are also allocated a probability. It provides a highly effective structure within which you can lay out options and investigate the possible outcomes of selecting those options, and helps form a balanced picture of the risks and rewards associated with each possible course of action. From this a sensible range of outcomes can be predicted. Other tools include ‘decision hierarchies’ and ‘SWOT analyses’. However these cannot replace basic, informed conversations along the lines of ‘How bad could it be? What if no-one bought any tickets? What are the chances of...?’ and, conversely, ‘How good could it get? What would be our maximum capacity? What would make it sell that well?’
The basic principles of probability and range forecasting are easy to apply to any organisation, whether through use of underlying complex models, or via a more intuitive, simplistic approach. But the real key is for the organisation to find itself in a situation where it is assessing a range of possibilities, rather than discussing a single-point forecast.
I have used this approach to good effect at Southbank Centre. On a monthly basis we reforecast all our activities – artistic programme, development fundraising and events, commercial income, operational costs etc, as described above. By holding open discussions with managers in each area we can get a real feel for the risks and opportunities facing us.
Once all the data is consolidated, you can then narrow the range if appropriate. Our approach in the first instance is to reduce it by 50 per cent on each side, the theory being that it is highly unlikely that all bad news or all good news will happen to us at once. We report our single-point forecast (‘most likely’), with an indication of the range of outcomes possible, to our board and the Arts Council. They then have much greater insight also into our risks and opportunities, and jointly we can manage our decisions with reference to the range of outcomes predicted. So, we use forecasts to change activities and direction where necessary – to steer the organisation.
I have also applied ranges to our 09/10 budget ‘forecast’, but in this case no narrowing of range has been applied. Due to the high level of uncertainty in the current economic environment we have deemed it wise to look at gross risks and opportunities, in particular
to enable us to really consider what we will do in a worst case scenario to mitigate risk.
to enable us to really consider what we will do in a worst case scenario to mitigate risk.
We can then use a ‘tornado’ chart to provide an easily understood visual representation of the risk profile of the budget (see figure 2 below).
Figure 2: Risk profile of the budget
This can easily be created from Chart Wizard (go to ‘bar charts’ and, using negative and positive data, complete a table, then rotate the resulting chart), and is a tool which communicates the message effectively to all groups of stakeholders.
Guidelines
A simple point, but a key one, is that forecasting guidelines are a significant input to the whole process. When one is dealing with different people and different income/cost areas, it is key that there is consistency of approach. Guidelines include defining the kinds of sensitivities or confidence levels the organisation is working to, providing clarification of forecasting in general (how it differs from targets and budgets – which are not the same thing, but which many people confuse!), why forecasts are a catalyst for action, advice on how to express the range (e.g. ‘most likely’ case should not always equal ‘best’ case), and the benefits which can be derived from implementing a range forecast process.
When looking at probability, it is important to be aware of the following key biases:
- Anchoring and adjustment: initial estimates or data can anchor subsequent thoughts or judgements.
- Availability: probability estimates can be influenced by the ease with which instances can be brought to mind.
- Representativeness: this occurs when available information is not considered carefully or the logic of an estimate is not thought through!
- Implicit conditioning: experts may make unstated assumptions which may bias any estimate given.
What’s next?
Range forecasting has proven successful within Southbank Centre, and is soon to be adopted at Sue Ryder Care. The mechanics to arrive at the range forecast may be different to those I have used in large corporates in the past, but the principles work in exactly the same way, and derive the same benefits. Range forecasting not only ensures a more accurate ‘central’ forecast, but allows finance greater insights into the workings of the organisation, as well as creating a greater sense of teamwork cross-functionally. The beauty of it is that it is very simple to introduce, and even if the basics are all that can be achieved in the first instance, this will already take your organisation a long way towards providing better forecasts.
The next step in terms of financial planning could be a more dynamic approach through the year, to allow the organisation more flexibility – a key strategy in the uncertain environment in which we find ourselves. It comes back to the issues around fixed budgets. An analogy can be found in human behaviour: when going from A to B, having planned the route, if there is a hole in the road it would be normal behaviour to go around it – rather than try to walk through it, as originally planned. So why in so many business planning situations do we insist on pursuing our original route, rather than adapting to unexpected or changed circumstances?
Rolling forecasts can be the answer, whereby progress against targets can still be monitored, actions to exploit opportunities/mitigate risk can be taken, and future commitments can be made on a sound basis. Watch this space...
Author: Caroline Stockmann
Caroline Stockmann FCA DChA is finance and commercial director of Southbank Centre, London’s 21-acre arts centre including Royal Festival Hall, and a Trustee of Sue Ryder Care, a specialist palliative and neurological care provider.Previously Caroline has had a varied corporate background, working for Unilever, Novartis, Cadbury Schweppes and KPMG.
There are no comments on this article. Be the first to comment.