Every 28 days–13 times a year–each department or unit has to revise its forecast and send the updated template to the finance staff. Finance then aggressively reconciles forecasts against actual performance each period and asks questions when a significant gap occurs, she explains. When facing the prospect of transportation costs for 11 elephants, the unit responsible for acquiring the animals simply reflected the change of events in their template and rolled it up into the consolidated forecast spreadsheet that Brock's team maintains. The Excel-based templates are generated by Timeline Inc. software using a data warehouse. Department figures are populated from the data warehouse into each template, Brock explains. Using the template, each department manager adjusts his or her forecast. Then, the numbers are submitted directly to the data warehouse through Timeline's writeback process, she notes. Reports can then be run, pulling the data from the data warehouse. The current forecasting goes through the end of each year. The plan for 2005 includes converting forecasting into a 13-period rolling forecast, she says.

With practice and corrections has come success. "It's not a perfect process and never will be," Brock concedes, "but we've made it a priority and become pretty good at it. With good forecasts, we can time the maturity of our short-term investments and, more importantly, we can use our credit line efficiently and draw the right amount for the right period of time. It's critical to minimizing our borrowing costs."

For many treasury staffs, a working capital forecast that goes out beyond a week or so worth of cash needs is the metaphoric elephant at the cocktail party–the large presence that cannot be ignored but that somehow doesn't fit into the graceful elegance of an otherwise automated system. Theoretically, long-range forecasting should be a success story, given that the computing power is available to most treasuries through workstations and ERP systems and access to greater and greater amounts of relevant data is possible. Yet, treasuries generally are dissatisfied with their forecasting capabilities and are even reluctant to talk about them. "Companies are pushing to make their longer-term forecasts better. A lot of them have developed some forecasting tools, but there's still a lot of room for improvement," reports consultant Mike Gallanis, a Chicago-based principal at Treasury Strategies Inc. "It's a high priority, but more companies are deficient at this point than are proficient at forecasting."

One reason: there are no real plug-and-play solutions, Gallanis says, because the factors that affect each business's liquidity forecast–e.g., elephant transport–will be unique. "For some elements of cash flow, regressive analysis is very effective. For others, a time series works best. It takes testing and trial and error to find the methodologies that work best with each company's pattern of cash inflows and outflows," he says.

Gallanis and Treasury Strategies work with companies to build customized forecasting models, but he admits it takes effort and time and may prove too expensive for some treasuries. Typically, a company that builds a forecasting model maintains it as a separate application (not part of an ERP system or treasury workstation) and feeds data into it from other systems.

But it doesn't need to be that high-powered to produce meaningful results. Certainly, that has been Brock's experience working with cash flow elements that are more manageable than those for most companies.

When forecasting the Zoological Society's revenue of about $160 million a year, for instance, there are a limited number of revenue streams to take into account: $5 million from a decades-old tax on San Diego property owners; another $4 million from grant money; $24 million from fund-raising campaigns; and the remainder from memberships, admissions and sales of auxiliary items like food and Zoo merchandise, Brock explains.

While San Diego is a powerful tourist magnet and the Zoo enjoys a world class reputation, revenue is not always a straight, upward-sloping line. "Ticket sales depend on people getting here," Brock says. That takes disposable income, so recessions will usually dampen ticket sales. Even more disruptive are disasters like 9/11 and the destructive forest fires that ravaged the San Diego area last year. For instance, the Zoo was forced by security authorities to shut down and evacuate for less than a day shortly after 9/11, for the second time in its 88-year history, she recalls. That was clearly an expense not anticipated in the forecast. "But because of our process, we were able to make the appropriate adjustments necessary to our operations in a timely manner so that we were able to land on our feet," Brock observes.

To assure liquidity, the Zoo keeps a credit facility with Bank of America and draws on that line as needed to cover cash shortfalls. Having a viable forecast is "essential" to making efficient use of the bank credit, Brock says. As a result she has been able to significantly reduce borrowings over the last year.

There's plenty of pressure, coming from the CEO, CFO and the analysts and shareholders who question them, to forecast liquidity further out with greater accuracy, reports Lisa Rossi, head of U.S. liquidity management services for Deutsche Bank Global Treasury Services. And companies are trying to leverage the technology available from banks and from ERP and treasury workstation vendors to help them do this, but progress generally has been mixed, she explains. Companies that have formalized, consistent processes like electronic invoice presentment and payment generally fare best. And of course the job is easier for some companies than others. Companies that get most of their revenue under contracts, for example, can better forecast incoming cash, Rossi adds.

Longer-term working capital forecasts need a data pull that spans and penetrates the organization. "You need clear, timely input from the parts of your organization that interact with your customers and suppliers so you get a sense of what's happening out in the supply chain," Gallanis says.

ALMOST IN REACH

The simplest way is to parse out the forecasting duties and make each unit or department continually revise its forecast, then let the piecemeal forecasts roll up into a consolidated corporate forecast. But that strategy relies on coordinating lots of pieces, and it can be labor-intensive at the unit level. Treasury doesn't control the process unless senior management mandates participation, and treasury doesn't control the quality except through after-the-fact reconciliations and pressure on units to improve faulty forecasts.

The vision is tempting: forecasting software that mines all the relevant data in a company's ERP system, capturing contract data, purchase orders as soon as they are created and future receivables as soon as orders are entered. Then it taps external databases to pull in future prices of key commodities. If it ships by truck and its contracts allow carriers to pass on fuel price increases, it factors in oil price projections. If it relies on parts made from aluminum and its contract with its key supplier expires in the next six months, it factors in probable price increases for inventory after that point. And so it goes up and down the supply chain: Elements must be identified that affect cash intake and outflow; historic correlations must be found and then all of these must be built into the forecasting model. So far, reality falls short of the vision, and no one is confident enough to forecast when that is likely to change.