Business Cycles and Future Transport Demand: How Use of a Leading Indicator Improves Forecasts of Rail Passenger Transport

Business Cycles and Future Transport Demand: How Use of a Leading Indicator Improves Forecasts of Rail Passenger Transport


M de Bruyn, P Spijkerman, Netherlands Railways, NL


Netherlands Railways developed a new short term forecast model: Q6 Plus, which is used for deployment of rolling stock. It is based on a combination of trend extrapolation and leading indicators: variables that are running ahead of business cycles.


Whoever can predict the future with certainty has the world at his/her feet. After all, reliable forecasts are highly important to policy makers, investors and companies. Also Netherlands Railways (NS) employs various models to forecast future passenger transport demand, such as the so-called Q6 Plus model that plays an important role in short-term forecasts (up to two years in advance). This new Q6 Plus model combines trend analysis and extrapolation with so-called leading indicators: variables and sectors that are running ahead of business cycles, such as employment agencies and freight transportation. This research is exploratory by nature, since, as far as we know, the leading indicator has not been applied before to company-specific forecasting in the transport sector - even though it has already been a tested method of forecasting in the literature on business cycles.

Forecasts derived from the Q6 Plus model are used to determine marketing policy and the deployment of rolling stock in particular. A more accurate prediction of the rolling stock requirement implies lower costs (because running excess capacity is avoided), and increased customer satisfaction (with not enough being deployed occurring less often). As satisfied customers are known to travel more frequently, this has a positive effect on revenues.

The new forecasting model uses three different regression models, each with their own speciality. The first model is predominantly suitable for short-term forecasts of passenger kilometres (approximately six months in advance). The second model provides highly accurate predictions for the somewhat longer term (up to two years in advance). The third model, finally, is specialized in predicting turning points in passenger kilometre data for the entire prediction period. The Q6 Plus model successfully combines the strong points of these three regression models: the mean prediction error of the Q6 Plus model is significantly lower than the prediction error of the existing trend extrapolation model. Even in comparison with other prediction models in the forecast literature, the Q6 Plus model performs exceptionally well.

With initiatives such as the development of Q6 Plus NS presents itself as a company that is able to actively incorporate recent developments in the academic world. With the aid of advanced technology, existing models are constantly being improved. In addition, current know-how and skills are applied in new ways. The resulting innovations are actively disseminated within NS to allow the company to adapt to new developments in a timely manner.


Association for European Transport