Local Speed Factories: when the need for speed is vital

In the 1990s, Adidas moved its entire production to low-wage countries in Asia, but in 2015 it re-opened a factory in Germany and recently also opened one in the US. These highly automated Speed Factories in Ansbach and Georgia are only responsible for a few per cent of Adidas’ total yearly production, but they guarantee extra short delivery times – one week instead of two months, to be precise. However, most of the brand's shoes are still manufactured in Asia. And Adidas is not alone. Nike also combines small-scale robotised local production with large-scale offshore factories. The textile sector is also exploring the possibilities. But why are companies moving a small share of their manufacturing activities back home? Professor Robert Boute and some of his colleagues developed a unique algorithm to answer this question.

Asia is becoming more expensive

“The phenomenon of Speed Factories is closely linked to the trend of manufacturing coming home, closer to the local market”, Robert explains. “In the past ten years, labour costs in some regions of China, for example, have more than tripled, and rental prices have doubled. Production may still be cheaper in Asia for the time being, but transport and stocking costs are on the rise. Long travel distances mean longer delivery times, so companies require larger stocks - after all, e-commerce demands prompt delivery to consumers. However, they risk not selling all their stocks and their working capital is tied up until they do. Moreover, long delivery times mean companies cannot respond to rapid market changes, which is a problem because demand is becoming less and less predictable. However, there is good news too: technological advances in automation and robotisation have accelerated, and they enable us to cut down the higher labour costs and long delivery times and lower working capital costs.”


Robert and his colleagues have developed an algorithm that investigates when such Speed Factories are useful, and to what extent. It takes account of the impact of the demand pattern, capacity and factor costs, stock levels, replenishment smoothing rules, delivery times and labour flexibility.

The algorithm applies to companies that rely on a supplier with long delivery times in low-wage countries and are considering setting up an additional automated Speed Factory closer to home, to respond quickly to market fluctuations. Among other things, the team modelled the impact of three typical demand patterns, each with a different type of unpredictability: (1) autocorrelated demand1, (2) stationary demand, which is unpredictable but fluctuates around an average value, and (3) non-stationary demand that can increase or decrease out of the blue. “That last pattern in particular is becoming increasingly common,” Robert explains, “and social media have a lot to do with it. An influencer posts a pair of sneakers on Instagram, and orders peak from one day to the next: that the company cannot keep up, but it all changes with the next post.”

The less predictable, the better

What did the analysis show? An automated Speed Factory combined with large-scale production abroad is not always equally useful. The demand pattern plays a key part in this: the more unpredictable the demand – as is the case with non-stationary demand – the greater the added value. The optimal Speed Factory is small and can even be a useful option if local production costs are higher than the price charged by the foreign manufacturer. The short delivery times allow the Speed Factories to save on stocks, and as a result also on working capital. Those savings compensate for manufacturing costs that may be higher. The greater the difference in delivery times and the more flexible the local labour force, the more attractive a Speed Factory is. The foreign manufacturer may also benefit from the Speed Factory: it will receive fewer orders, but demand will fluctuate less, resulting in lower costs for unused capacity and overtime.


How can you use the algorithm yourself? “We are currently developing a Shiny app2, an interactive online app that will allow companies to analyse where and when a Speed Factory would be viable in a fairly simple manner. You basically need to determine the break-even point by following the evolutions in foreign purchasing prices and local manufacturing costs. If the price abroad exceeds the break-even price or if the local manufacturing costs decrease below the break-even costs, it makes sense to bring part of the production home. The app will help you determine when and where to take this step.”


Speed Factories are also relevant from a social point of view, as Robert explains. “The technological advances that we call ‘Industry 4.0’ for simplicity’s sake allow us to bring part of production back home, even if only on a very small scale. It makes no sense whatsoever to keep complaining about manufacturing activities moving abroad, because what is lost will not come back. However, we can attract a new type of production. That is a positive message that does not often receive enough attention. When people claim that highly automated factories do not create jobs, my response is: “They do create jobs, but not the traditional ones. They require operators and maintenance technicians with specialised skills that we need to stimulate. Moreover, you should not lose track of the multiplier effect: an extra job in the high-tech sector offers over four jobs in other sectors in the same region - in logistics, catering and other related services.”

1 In the case of an autocorrelated demand, the subsequent values of a variable are correlated according to a repeated pattern. A high demand during a specific period, for example, leads to a high demand during the subsequent period.
2 An interactive online app developed in R, a user-friendly programming language for data analysis. Shiny apps present the results of sophisticated analyses in a visually attractive manner.

Source: The paper ‘Dual Sourcing and Smoothing under Non-Stationary Demand Time Series: Re-shoring with Speed Factories’, including the detailed mathematical appendixes. You can request a copy from the authors.
About the authors: Robert Boute is a Full Professor in Operations Management at Vlerick Business School and at the Faculty of Economics and Business of KU Leuven. Stephen M. Disney is a Professor in Operations Management at Cardiff Business School (UK). Jan A. Van Mieghem is a Harold L. Stuart Distinguished Professor in Managerial Economics and a Professor in Operations Management at the Kellogg School of Management at Northwestern University (USA).

& Rankings

Equis Association of MBAs AACSB Financial Times