There is a lot of discussion about robots taking our jobs. Much of it is speculative, usually forecasts based on the opinions of AI experts, who tend to be too enthusiastic about the potential of digital technologies. Such estimates take little account of job redesign possibilities, or social and economic barriers against automation.

We can see these barriers in our daily lives. In a trip to Singapore three years ago, some colleagues and I went to a restaurant we had visited before, but there was a difference this time: we were greeted by a waiter who handed us not a paper menu and some suggestions, but gave us tablet to place our orders. We struggled. A year later we returned and saw, with relief, that the tablet was gone. The paper menus had returned! I asked why. The waiter told me that customers had difficulties navigating the pages, especially (but not only: we were living proof of that!) if they could not speak good English. Some inadvertently added food –perhaps also the odd drink?- that they did not want to their orders.

But even if we take the most pessimistic estimates around automation, one thing we know is that not all jobs will be replaced by machines. Workers, and governments, will be interested in what those “resilient jobs” may be. We asked ourselves: what if company contexts matter?

We used data from over 3,800 companies in Singapore –one of the most flexible labour markets in the world, where the costs of firing workers are very low- to explore what firm characteristics may shelter employees against job losses when digital technologies are introduced at work. Our analysis found that most changes to production processes tend to be labour saving. This was not a distinctive characteristic of technological change; other changes also led to reductions in workforce numbers. However, firms that introduced technological changes in the previous 12 months were more likely to reduce their workforce than those that introduced changes that were not related to technology..

But some firms introduced technological change without reducing their workforce. What we found about those firms is surprising. Firms with high proportions of high-skilled workers and flexible work organization –which provide much autonomy to workers- were not less prone to technology-related labour substitution. By contrast, two other characteristics acted as shelters: firms’ competitive strategies centred around high value-added products, rather than cost, and positive perceptions of management on the competence and commitment of the workforce.

Our results challenged the overwhelming interest in the academic and policy literature on the role of high skills and flexible forms of work organisation that provide workers with a high degree of autonomy as protective factors against automation. The results, instead, show that it is not only the individual worker and their skills that matter when we look at the how technology substitutes labour at the workplace – the firm level does too.

If workers want to reduce the insecurity associated with replacement by technology, they would be well advised to look at their workplace: how the firm competes and how management thinks. These are findings that governments seeking to achieve sustainable increases in job quality and stability should attend to, moving beyond an exclusive focus on increasing skills supply.

There is a bigger point to be made on management perceptions of their workforce. Studying social relations, and not only economic calculations or technological possibilities, is crucial to better understand the future of work.

Access our full paper here