Events of the past few months have illustrated just how quickly global market conditions can change, and therefore how critical it is to have a regular profit and cash re-forecast. However, the prospect of a rolling forecast - going through the equivalent of an annual budget process every month - just seems too horrific for many finance managers to contemplate. FSN contributing editor Steve Bows talks to Challenger Financial Services Group, an organisation that dared to think the unthinkable, and who have been able to successfully adapt their business plans to the new economic world as a result.
The Challenger Rolling Forecast Framework
Challenger Funds Management Forecast – a Simplified Example
{manual input}
Rolling forecasts – making clearer decisions in turbulent times
Rolling Forecasts
What is a ‘rolling forecast’? Depending on who you talk to, you get a few different answers to this question. Some would classify any forecast process, being a budget revision undertaken part-way through a year, as a ‘rolling forecast’. The rolling element here is incremental addition of ‘actual’ monthly data to the budget as it becomes available, so that it ‘rolls’ to the end of the year. Others would insist that a true ‘rolling forecast’ requires a rolling timescale as well, so that it always looks a certain number of months into the future, rather than being fixed within the straitjacket of the financial year.
Richard Burton, Head of Performance Management at Challenger Financial Services Group, explains why this rolling timescale is important to the planning process: “We wanted to move people away from the artificial nature of the annual budget process, or even the quarterly reforecast. We were finding that business managers did not consider the full financial ramifications of the changing marketplace until they were forced to perform a reforecast for Group Finance, and this was leading to sudden spikes in profit forecasts and cash requirements every quarter or half-year. By implementing a system that took the pain out forecasting, we could incorporate these processes into the daily habits of our managers, and hence get an up-to-date Group forecast at the end of every business day.”
So it seems that there is a need for a rolling time horizon, and less rigidity in our forecasting processes, encouraging ‘event-based’ rather than ‘process-driven’ planning, but are there any other issues that the ideal rolling forecast should address?
“There was one other cultural issue that I felt we needed to address in our approach to rolling forecasts. Research shows that 80% of the focus of planning effort should be devoted to increasing revenue, and only 20% to reducing costs, whereas the reality is often the other way round. We wanted to re-educate our business managers to concentrate more on revenue, and not to ‘sweat the small stuff’ on minor expense items. Of course, with a driver-based, integrated planning model, modifying revenue projections is far quicker than detailed line item expense planning, so the system we have reinforces the management strategy.”
Walking the Talk
Strategy is one thing, but many organisations have found that implementation is quite another. How did Challenger manage to achieve a rolling forecast system that met their criteria for daily reporting where so many others have failed? Stuart Amiss, manager of Challenger’s Cognos Competency Centre, explains some of the critical decisions that led to project success:
“First and foremost, we needed to sort out the business dimensions before we started any of this. There was a lot of confusion around the use of products, business units and account codes throughout our divisions, so we spent a fair amount of time cleaning up the structures, removing redundant codes, and then ensuring we had a standard set of codes, names and (most importantly) hierarchies that all the divisions were happy with.”
“Once we had a set of coherent structures, we were able to build a central data repository for all of our forecasting and reporting data, the ‘hub’ into which all of our end-user applications connected.”
Richard picks up the story: “Then came the hard part – we had to take the time to understand all of the planning processes of our disparate divisions, and try to work out how we could incorporate them all into one connected group-wide framework. I don’t think you could say that we got it right from day one – I don’t think anyone does. But it was close enough to be workable, and we managed to introduce successive iterations that gradually made it smoother and more in tune with the overall rolling forecast philosophy.”
“The key thing was generating enough user buy-in, both at a senior management level, and down at the coalface, to be able to move on to each successive stage. I think this is where many systems implementations fall down – a failure of willpower to keep pushing forward and to finish the job. I was determined that we would set the standard for other organisations to follow, no matter what obstacles we faced along the way.”
So what sort of problems did they face, and how did they overcome them? Stuart lists a few: “One of our biggest problems was weaning users off spreadsheets. One of our divisions had a culture of unstructured creativity, with lots of accountants developing their own self-contained spreadsheet models. The models were very impressive, but because they didn’t use the centralised structures or data, they were giving inaccurate and inconsistent results. We battled long and hard to get them to use a centralised system, and it wasn’t until we proved that we could do all their spreadsheets could, and then some, that they eventually came on board.”
“Another major challenge was dealing with a constant flood of enhancement requests that started coming through as soon as we went live. Obviously this is a good thing, as it shows that people are using the system and want to do more with it, but it was a logistical headache for us at first. We tackled it by developing an in-house Cognos Competency Centre, staffed with both technical experts and business analysts, to regulate our response to these requests, and get new pieces of functionality out to the users as quickly as we could. We still get stretched at peak periods, but by and large we are free of any need for external consultants these days.”
Automating the Forecasting Process
Earlier on we mentioned the need to make the forecasting process as pain-free as possible. Are there any specific examples that Challenger can point to?
“We realised early on that our Funds Management division had the most ‘automatable’ process” says Stuart. “Fee revenue earned, and many of the variable costs associated with managing a fund, are driven by fixed percentages of the FUM (Funds Under Management). We already had a ‘clean’ view of actual FUM data from one of our source systems, so it was clear that we wouldn’t have too many problems in driving out a projection based on this data, with very little need for manual intervention.”
“The only real challenge we had was getting our calculations to work one way for actual months (calculating net flows and effective earnings percentages from FUM and recorded revenue) and another way for forecast months (calculating closing FUM from projected flows, and revenue from the earnings percentage). Fortunately the tool we used was very flexible, and we came up with quite an elegant way of doing it.”
| Ref | Fund P&L Item | Calculation (Actual) | Calculation (Forecast) |
|
A |
Opening FUM |
{b/f balance} |
{b/f balance} |
|
B |
Net In/Outflow |
B=E-(A+D) |
{manual input} |
|
C |
Fund Earnings % |
C=D/(A+E)/2 |
{fund parameter} |
|
D |
Fund Earnings |
{actual data uploaded} |
D=C*((A+E)/2) |
|
E |
Closing FUM |
{actual data uploaded} |
E=A+B+D |
|
|
|
|
|
|
F |
Management Fee % |
F=G/(A+E)/2 |
{fund parameter} |
|
G |
Management Fee Revenue |
{actual data uploaded} |
G=F*((A+E)/2) |
|
H |
Performance Fee % |
H=I/(A+E)/2 |
{fund parameter} |
|
I |
Performance Fee Revenue |
{actual data uploaded} |
I=H*((A+E)/2) |
This table shows just how automated the forecast is for each fund – there is really only one input needed every month, the projected net flow (B). All the other items are driven from the uploaded historical data (providing a moving baseline of actual FUM balance), combined with a fixed set of ‘fund parameters’ which can be tweaked occasionally as required.
Stuart continues: “This model has really come into its own over the last few months, with the extreme market volatility. We are constantly re-planning the net fund flows based on market projections. The system automatically aggregates the results for each fund and gives us a view of the P&L for the whole division very quickly. From this we can pick out non-performing products and mandates, and take decisions to maximise our profit and cashflow in these tough times.”
Where Next for Challenger?
Richard outlines the lessons learnt and next steps for Challenger: “Looking back, we certainly made a few mistakes in the early stages. We tried to put too much detail into the models, to make them as accurate as possible, and this had some performance issues. We’ve ended up with a compromise between high and low-level planning, trading in some accuracy for fast consolidation and performance. I would say that this shows the importance of a pragmatic approach – there are rarely ‘right’ answers when it comes to complex financial modelling.”
“Our biggest project now is continuing the process of cultural change – we don’t see a lot of systems development happening in the near future, especially given the market conditions, but fortunately we have pretty much everything we need already. However, there is still work to do on managing the expectations of our stakeholders and user group. People have been so used to preparing and reviewing budgets in the traditional annual cycle that it will take a long time to really move them away from it. Having said that, we have come a long way in a short space of time, and I’m pleased with our progress.”
