It’s the holiday season again and time to brace yourself for all the predictions about whether or not sales this year will exceed last year’s and by how much.

This story first appeared in the November 17, 2003 issue of WWD. Subscribe Today.

These forecasts always start up around September or October when no one — particularly the consumer being analyzed — really has the slightest clue what will happen in December. Ask yourself if you knew back in September how much you would spend in December. Chances are you didn’t, you still don’t and you won’t be able to tally it all up even after you’ve gone and spent it.

That won’t stop research organizations and pollsters from taking your top-of-mind responses during a telephone survey (possibly administered during your dinner or as your four-year-old strangles the baby) and using them as a key input —perhaps the only input — into their holiday sales forecasts.

“Prediction is very difficult, especially if it’s about the future” quipped Nobel Prize-winning physicist Nils Bohr. Forecasts by economists have always been easy to make fun of because they are frequently well argued, empirically based, stated with a high degree of precision and assurance, and yet still wrong.

Economists, however, are not fools. Like weather forecasters, they know that the closer you are to an event, the better chance you have of getting the prediction right. So they try to get around the uncertainty of their science by revising their forecasts as frequently as possible, and doing so as closely as possible to the actual event being forecasted. This has served to increase the accuracy of the forecasts while simultaneously reducing their usefulness.

Even so, “predictions” about growth and recession frequently produce howlers even when they are made during the event itself, such as the failure to predict the 2001 recession on the very eve of its start, long before Sept. 11.

We see predictions about consumer spending generally, and about sales growth specifically, before every major seasonal “event” including Easter, Mother’s Day and Father’s Day, back-to-school and Halloween. Long before we even begin the critical fourth fiscal quarter and holiday season we start getting a wave of forecasts about the growth rate of holiday sales. Are any of these forecasts worth anything?

One of the most commonly used research methods is having a telephone polling organization find out what consumers themselves plan to spend, what they plan to spend it on and where they plan to spend it. These surveys are ubiquitous. You’ll read something like, “The average household plans to spend $350 on back-to-school items this year, up from $330 last year.” Or “The favorite destination for shoppers this holiday season will be discount stores, with 75 percent of households planning to do most of their gift-buying there.”

These surveys are easy to devise, cheap to execute, have a quick turnaround time and are invariably wrong. However, both the methodology and results can be easily understood by the news media and the public. And better yet for the survey sponsor, no one can be bothered trying to figure out if the forecast was accurate after the event, since either the media are no longer interested (such as after a minor retailing occasion like Mother’s Day) or there is good data available from official sources, as is the case beginning on Dec. 26.

At first blush, going straight to consumers doesn’t seem to be a bad idea, and studies about the value of this consumer input indicate that there are very specific conditions under which consumer purchase intentions are useful for forecasting sales. First, the purchase event being predicted needs to be imminent at the time of the survey and of short duration. Secondly, the purchase itself should be for an expensive item that requires relatively more planning and consideration.

So purchase intention surveys may be useful for forecasting near-term sales of products like cars or houses but of questionable usefulness in predicting consumer spending over the course of an event or season. In the cases of back-to-school and end-of-year holidays, the seasons are protracted. Consumers being interviewed even immediately prior to these seasons can’t be expected to know what purchases they will have made by the season’s end. Even if a consumer consciously starts out with an overall dollar budget for these events, the sheer duration of the events and the impact of impulse buying are likely to have a seriously adverse effect on his or her initial prediction. Things change.

But are consumer surveys entirely worthless? That depends on what you are trying to measure. If you want to know how the entire holiday season is going to measure up, forget it. We have no business asking people how much they are going to spend on gifts over the course of a two-month period and expecting reliable answers.

If you have a new product to put on the market and want to forecast demand, that’s a different matter, particularly if the survey is designed well. However, when it comes to predicting the outcome of retail seasons, sometimes rules of thumb do as good a job as anything. One such rule is that wherever the back-to-school season goes, the holiday season follows. As it turns out, a careful examination of the data reveals that for some retail categories, it’s a pretty good rule indeed.

If you take some broad retail aggregates such as the Bank of Tokyo-Mitsubishi’s chain-store sales index, or the U.S. Census Bureau’s GAFO (general merchandise, apparel, furniture and other store sales) category, the relationship between third-quarter year-over-year sales growth and holiday season year-over-year sales growth is usually quite close. Third-quarter results aren’t all in yet for 2003, but if you take an arithmetic average of same-store sales growth for July, August and September, you get an increase of about 5 percent. That is by no means an implausible forecast for the holiday season.

But beware — if you take a highly discretionary category like home electronics or jewelry, the usefulness of back-to-school sales growth as a predictor of holiday sales is quite poor. That makes sense, because willingness to commit to highly discretionary purchases is very sensitive to small and rapid changes in psychology, as may occur copiously during the intervening months between back-to-school and Christmas.

And apparel? In the last six years, holiday sales growth has differed from back-to-school sales growth by an average of only 80 basis points, making it quite a useful guidepost. Six years isn’t much to hang your hat on, but it beats calling 1,000 consumers at dinner time.

Michael Baker, the former research director of the International Council of Shopping Centers, is a retail and economic consultant who helps retail clients in the U.S. and Australia better understand consumer trends and identify underserved market niches and opportunities. He can be contacted at and promises not to call during dinner.

load comments
blog comments powered by Disqus