The Hazards of Forecasting Auto Sales
Forecasting new-vehicle sales in the U.S. automotive market is notoriously difficult, as all of us automotive economists know all too well. Monthly levels fluctuate widely and even the underlying trend rate of growth seems to accelerate and slow down over the years with unpredictable regularity. Add in some potentially severe shocks, ranging from skyrocketing gasoline prices to housing slumps, and the challenge is daunting under the best of circumstances.
Consider, for example, that the most recent Blue Chip Consensus Forecast that includes predictions for sales in 2007 by economists and business analysts. Forecasts vary from as low as 15.7 million unit sales per year (or, the lowest annual performance in almost a decade) to as high an estimate as 17.3 million unit sales (or, just short of a record level, if it is achieved). That’s a difference of nearly 1.6 million vehicles. The problem is not so much the skills of the forecasters, but the lack of data on what is really driving car and truck sales.
In today's automotive market, two factors are essential to predicting how many new vehicles will be sold in a year: incentives, which includes rebates and special loan rates including the popular 0% financing, and fleet sales, which are purchases by rental car agencies, such as Hertz and those vehicles sold to government and corporate fleets.
During the last several years, changes in incentive programs have explained much of the month-to-month volatility in vehicle sales. Yet, not only is there little data on incentives, but the decision to change incentives by the Detroit-based automakers, including General Motors, Ford and the Chrysler Group, may have little to do with economic factors that are normally used to forecast sales, such as income, interest rates, or even changing gasoline prices. Incentive strategy is more likely determined by automakers’ sales and market share concerns. Manufacturers must worry about keeping their workers active in the plant (many of these workers get paid whether they build anything or not). Fleet sales play a similar role, increasing and declining, depending partly on the need to offset changes on the consumer side of the automotive market. And, like incentives, there are little data that allow forecasters to separate retail and fleet sales activity and review their behavior separately.
The result of missing data is that automotive forecasters today are often flying blind. Our firm, J.D. Power and Associates, is in the process of building new databases that will soon allow forecasting modelers or builders to incorporate incentives and fleet sales in their crystal ball gazing. However, for now, we must resort to guessing about manufacturers’ intentions. When you hear on the business news a forecast about how many vehicles will be sold this year, you may now have a little sympathy for the poor economists who have to make their livings reading tea leaves and holding séances just to figure out how many cars and trucks consumers will buy each month.
