Professor Barry Keating of Notre Dame, coauthor of a college textbook on business forecasting which has more buyers from businesses than from college students, was interviewed recently on the topic of using macroeconomic data to adjust forecasts down for coming recessions, or up when economies begin to grow again.
One of the points he made is that companies and forecasters in particular need to pay attention to the Index of Leading Economic Indicators (LEI), which is calculated by The Conference Board, a non-governmental organization. The index is based on ten key variables, including the number of manufacturers' new orders for consumer goods and materials and the amount of new orders for capital goods unrelated to defense.
If the overall Index falls three months in a row, it is a good indicator that the economy will go into recession within three to six months. However, the Index is not perfect – while it correctly forecast each of the 7 recessions that occurred between 1959 and 2001, it has also forecast 5 recessions that did not occur.
There are also many industry-specific indexes that are apt to be much more useful than LEI. For example, many industry trade associations have economists and/or statisticians that create their own index to track variables relevant to their specific industry. So, for example, a plywood manufacturers association might use an index that places considerable weight on new housing starts but does not factor in new orders for capital goods at all.
The next step is to apply these inputs into your forecasts. There are many robust demand management software solutions that use a large number of forecasting algorithms. The problem is that many times companies don't use the algorithms that factor in economic downturn predictions. Time series decomposition, for example, decomposes demand into trend, seasonality, and where an industry is in an economic cycle.
Using historical data, these software solutions can calculate trend and seasonality automatically, but since economic trend data is external to the system, forecasters must add these inputs separately. This is one reason why companies with large numbers of SKUs often don't use this algorithm. Similarly, regression can be a powerful tool for including macroeconomic data into a forecast, although Professor Keating warns that this is more complex and requires highly-skilled users.