How bad a downturn should you assume when developing your recession contingency plan? I was talking with two bankers about their efforts, which were excellent, but they began with that question. As a first pass, they assumed a recession half as bad as 2008-09. That felt right to me, but I had to go luck at actual numbers—that’s the kind of guy I am.

The average recession since 1950 has led to a drop in inflation-adjusted GDP of 2.0 percent, whereas our last recession pushed GDP down 4.0 percent. So that rule of thumb—half the last recession—is right in line with historical averages. (Note that inflation-adjusted GDP cannot be translated directly into revenue changes, but the “half the last recession” concept can be.)

GDP decline in recessions (Dr. Bill Conerly, Based on data from the National Bureau of Economic Research and Bureau of Economic Analysis).

There aren’t many businesses that can look at their own experiences over 60 years, as I looked at the economy, but most companies can look back ten years. Half of the last recession experience would be a good benchmark. Keep in mind, of course, that most recessions are not average. Heck, hardly anything is average. In addition, each recession has its own texture. The last recession was especially harsh on housing, but 2001’s recession was very kind to housing. Your own industry may do well or do poorly next time.

What these bankers had done was take the decline in loan quality that they experienced in the last recession and projected half of that decline forward from their most recent quarter. Non-financial businesses would do best projecting a sales drop in the coming quarters. Depending on the company, the sales drop may be accompanied by a decline in cost of goods sold or other drop in expenses. This modeling doesn’t have to be perfect, just reasonable.

My bankers looked at the adequacy of their loan loss allowance and their capital. Non-financial companies should run a sources and uses of funds projection along with a balance sheet. Compare the results against loan covenants and the desire of equity owners for dividends.

We cannot take too much comfort on results that are just barely adequate. The next recession may be worse than average, or it might coincide with another problem, such as heightened government regulations or new competition. When I work with clients I remind them that we are just ballparking the numbers. The exercise is valuable, however. It indicates the magnitude of corrective action that might be needed. Maybe a short-term hiring freeze, with normal attrition, will keep expenses down below revenue. But maybe severe cuts will be necessary. It’s hard to tell without running the numbers.

In my business contingency planning video, I tell a real story of how one company used its recession contingency plan to deal with the last recession, then used an upside contingency plan to reap all possible benefits from the recovery. Planning ahead, the story illustrates, enables a faster response to recession, as well as a smarter response.

Before diving into the analysis, set goals for the process, such as determining how much corrective action would be needed to maintain positive earnings in a recession. Then outline your procedure. You’ll probably have to iterate through a few rounds, such as testing a small expense cut, seeing the results, then revising your first estimate. At the end, write out the steps you took so that it will be easy to update the analysis next year.