Paris Innovation Review – McKinsey & Company has published a report about possible automation of hundreds of millions of intellectual services jobs round the world by 2025. Should we be concerned by this trend?
El Mouhoub Mouhoud – The globalization phenomenon does not only affect unskilled personnel. It impacts on all operative tasks in every skilled segment. Witness the case of radiologists. They possess a high level of professional qualification but, in their services, only a few activities require physical proximity with the patients. Other tasks can be carried out remotely, for example, reading routine X-ray images. In some solicitors’ offices, certain very advanced contract research can be delocalised to India, almost in real time. The same applies to some computer science application. Lists are needed, not only for activities but also for individual tasks: analysis should focus on the task characteristics.
Whatever the level of qualification of the members of the labor force, questions of delocalization or process automation are very closely tied to task standardization. The same holds in the case of high level intellectual jobs, where a number of tasks can be codified and therefore the associate processes can be automated, as happens for industrial activities. Tasks in this category can be carried out remotely or can be automated, which in a sense amounts to the same thing. Wherever productivity gains can be made, there are possibilities to delocalize the activities.
However, we must not imagine that the very places where these tasks are performed are without relevance or are totally interchangeable. In real-life situations, many intellectual services need a favorable environment, especially when they relate to knowledge-intensive and immaterial investment services, which need access to resources and specific interactions. It is no easy process to decompose the tasks involved. Not everything can be codified, nor standardized. Creating and sharing knowledge differ from other activities and the automation/offshoring analysis grid does not allow you to fully apprehend this set-up. All services will not necessarily become industrial information, i.e., information that can be decomposed and mass-produced. Lots of new forms of knowledge are closely associated with the human capital that generates the knowledge: when you launch an innovative approach, you need time before the return on experience can be formalized to the point that the method can be imitated by others. In the meantime, the knowledge remains where it is, viz., in the team that implemented it. It is clear that more and more sectors will be automated. Increasing fields of knowledge which to date have been “tacit,” i.e., they owe their very existence to interactions between the persons who implemented them, in what could be termed the coffee-corner syndrome, will be codified and therefore automated, i.e., carried out by intelligent machines, software or robots.
What sort of jobs are under most threat?
I have investigated the impact of offshoring in terms of task destruction in some 4,000 French companies. A decision to delocalize has a very strong impact on all levels of unskilled or skilled work, but especially impacts engineers and technicians when concrete production and basic operational tasks are involved. In contradistinction, logistics and coordination tasks are not affected and management tasks are on the increase. In short, globalization can be seen to segment the skills spectrum and impact preferentially certain tasks. The effects can be felt now by knowledge workers and by an increasing proportion of the working middle class, who hitherto has been scared. The notable rise in political populism is partly due to this economic demolition of the middle classes.
There is a hiatus on the direct effects of offshoring. Economists continue to assert that at a macro-economic level, very few jobs are destroyed by offshoring. Globally speaking, this is true, but the destructive effects are highly concentrated on certain activities and in certain geographic regions; in Europe, for instance, it was the sectors of textile mills, toys, shoemaking, some iron & steel mills that were the most impacted and the effects were especially painful for territories with historic specialties and located outside the conurbations (we can cite here the coal-mining sites in North England or a French township such as Romans, with its concentration of shoe-makers). The conclusion here is that it is vital to identify one’s weak points in time.
To ascertain whether a territory is efficient or not in the context of globalization, our analysis must not be limited to generic activities but also to the breakdown of the latter into various tasks. This applies to both services and industrial activities. In the world of cloth and clothing, seen as an ensemble of activities that works with soft materials, the entire sewing assembly process cannot be automated and this makes the sector high delocalizable. In contrast, activities that manufacture, machine and assemble solid materials can all be automated and carried out by robots. The cost of labor therefore has a lesser impact on decisions to offshore. A European cast-iron founder who has delocalized this activity to China will relocalize it when the cost of transporting the iron billets become prohibitive. Reshoring applies to activities that handle large volumes of heavy goods, those that can be automated, and those for which offshoring negatively impacts a corporate or product image. Reshoring near the end-user markets – not necessarily the originating countries, think of nearshoring – is the consequence of recovering competitive edges.
Let us come back to the service sectors. How are we supposed to identify those sectors that can be delocalized?
Well, Paul Krugman offered a simple definition: a service can be delocalized when production can be concentrated in a given territory and demand can be satisfied remotely. In contrast, a service cannot reasonably be delocalized if production is distributed proportionately to the end-user household locations. We can therefore create a chart of services assessing their vulnerability faced with possible delocalization decisions.
The common typology for services – merchant or non-merchant, public or private, enterprise or household, exchangeable or non-exchangeable – does not prove especially useful when it comes to assessing economic performance or the competitiveness and the added-value of delocalizing service activities. To evaluate the degree of vulnerability of a given service activity, I would suggest three questions that include the relevant criteria, as I see them.
Firstly: does the activity need direct access to specific resources (manpower, physical assets, location, land)?
Secondly: does the activity need conurbation scale economies? To reframe this, when you localize an activity, will network effects and economy of scale make a difference, or can you locate the service almost anywhere? Activities such as telecoms, postal services and academic research need and use a specific environment if one wants them to be efficient.
Thirdly: does the activity need direct access to demand?
By cross-referencing these criteria, we obtain a typology for service activities which is far more operational when certain fundamental questions of economic policy are raised: what qualifies as a fragile territory or, in contradistinction, as a dynamic territory? How best can we accompany the dynamics of comparative advantages of given territories? How could we create so-called ‘granular growth,’ building on favorable market niches?
Territorial prospective approaches as implemented in Europe today are seeking to identify these dynamic aspects, which are not limited to metropolitan areas. For instance, the regions that produce wines in South Europe can develop a series of service activities to valorize vineyard production: in R&D, in marketing, in packaging and advertising; activities like these can be integrated to the value chain and indeed become key components.
Among the economic policy questions, one is of special interest to policy-makers and executives: how are they supposed to anticipate crises? Are national level programs, for example, relevant to such situations?
Majors industrial contingency plans are useful when the aim is to attain a world scale technological leading edge. For example, when national authorities launch a robotics sector initiative.
Notwithstanding, reasoning on a national or sectorial scale has its limits. Let’s return momentarily to the textile markets. When the World Trade Organisation was established in 1994, one of the decisions was to repeal the earlier “multi fibre” agreements which enabled certain countries to protect national industries against imported clothing textiles. At the time, the schedule for the cancellation procedures was set for year 2005, viz., with a full decade to prepare. But some countries waited for the crisis to occur before they took action. On the contrary, this should have called for drafting maps of the employment basins in order to anticipate crisis situations occurring in regions mono-specialized in textiles, to implement a diversification plan, and to concentrate resources on training and thus help the local workers switch to other activities. But in most cases, the policy makers tend to intervene after the fire outbreak to assist enterprises in difficulty via subsidies. Policy decisions like these are short-sighted and consequently inefficient. Reasoning (and intervention where appropriate) must be organized on the scale of employment basins. The challenge is to turn declining areas into dynamic areas.
What are the bases for a territorial-scaled economic policy?
Proponents of political economy must beware of two classic shortcomings in debates on territorial competitiveness and fragility: firstly, exaggerating the fragile features by restricting one’s vision to the manufacturing activities present in the territory. Secondly, in considering services as a form of activity that is protected naturally against offshoring. In real life, some services are very fragile. Others, by way of a contrast, are well rooted in the territories and display very strong levels of attractiveness, including for manufacturing activities.
Knowledge-intensive services represent 14% of so-called “industrial” jobs in a country like France. Indeed, they are drivers of industries, companies and jobs creation. These are not services that serve intermediate enterprise or household consumption: they serve immaterial corporate investment. These services are R&D, consulting, but also marketing or advertising. Any research engaged outside a company also comes under this heading, for example research conducted in self-standing public or private laboratories. I should also add higher education institutions, plus the fact that marketing is an important activity: analyzing consumer expectations is not R&D per se nor experimental science, no patents are involved or any measurable added value, but the contribution marketing makes to business success is important. It comes under the heading economists call “Invisible Innovation.”
If we follow your reasoning correctly, could service discrepancies explain different levels of competitiveness?
Yes, absolutely. If, for instance, we compare France and Germany, we discover an interesting situation, inasmuch as these two countries have made different policy decisions in terms of services offered to enterprises.
In Germany, deregulating the labor market did not have any serious impact on industry. There was a notable impact on the service sector, with high pressure on salary rise claims. In industry, stricto sensu, the industrial hourly pay rate is almost identical in France and in Germany, about 37 euros, to the nearest few cents. But in Germany the rate is much lower in the service sector. And industry uses a lot of service activities (interim workers, computer services and data processing, finance…): industrial labor costs only represent 20% of the price off an automobile. If we include this factor, we must add 15% to the degraded competitiveness of costs in French industries.
This logic is reinforced by organization choices. French companies have outsourced service activities much more than their German equivalents. Outsourcing has led to the creation of intermediate companies, specialized in marketing, R&D, consulting, and of course these services must be paid for. In choosing to outsource certain activities, companies are losing their grip on innovation driven by uses and the said services become more expensive because of the price margins practiced by the intermediate companies. There is a cost inherent to applying standardized methods to a given process. If the services are very specific, their productive value is higher when they are integrated to the production process than when they are not included, because in the latter case there is an additional cost to adapt to the case.
Outsourcing services to these intermediate companies is certainly not a panacea, hence the good results in Germany where very high quality services are integrated in the industrial sectors. In France, the downsizing of companies and policies to reduce production costs led, as of the 1980s, to the rise of outsourcing solutions. Another noteworthy point is that the demographic corporate structure also plays a role. It is commonplace to see that the larger German SMEs can retain services within the company far better than the smaller French SMEs.
Germany is not the only country that has developed efficient services and thereby build and draw on industrial advantages. For other countries, the main part of the quality of the non-cost advantages relies on the quality of “good” service arrangements. This is true notably for Ireland, South Korea and also, but this is largely unknown, for the USA. If we leave the national scale and continue to reason at a territorial level, this is also true as illustrated by innovation clusters, designed as a concentration of high quality services, around R&D activities. For certain territories, betting on immaterial investment services can prove to be a winning strategy.
For policy makers, what would be the appropriate policy in terms of development of service activities?
When you install a service in a given territory, there is a “location mix”: the service must have an access to resources, must benefit from a conurbation scale economy and from customer proximity. If we look at R&D, of course you can delocalize this activity but there will be a high line-loss. There is a strong logic in favor of installing large research centers close to major cities. Knowledge-intensive workers demand this sort of stimulating environment. But, even medium-sized cities have their chance. By way of contrast, the information services associated with the support functions (maintenance of the computer infrastructures, call centers, service activity operative functions) can be offshored more easily. The question of the environment is primordial. Former industrial semi-rural areas which were tempted by a reconversion policy to host call centers lost out when the service activities went elsewhere. But call centers in main city areas did well. They are more sophisticated, connected as they are to other activities using higher levels of skill (for which there are less substitutes). Decision makers must take care not to specialize a territory in activities that are both isolated and delocalizable.
Then productivity gap between service and industrial sectors is currently closing, with higher productivity gains being observed in service activities. And nothing in particular precludes these productivity gains leading to creation of new ancillary activities to accompany innovation. These may be industrial activities: for example in Europe, some of the jobs lost in the traditional textile industries have shifted to techno-textiles. The major wave of automation and delocalization will not lead to straightforward destruction of existing employment. New economic activities will create different jobs and professions, in maintenance and coordination, for instance.
The issue for a given territory therefore is not to resist offshoring, but to make optimum use of its advantages and not be afraid of seeing certain activities leave the area. There can be advantages in offshoring when industrial estates are specialized in those delocalizable activities that incorporate positive “external performance factors”, i.e., they export a lot and import intermediate part-finished goods. This I would qualify as “good” delocalization, as implemented in many cases by German enterprises, in their trade relations with Eastern Europe, while retaining in Germany their strong export-intensive activities and industrial skills. Knowledge and immaterial services tend to install their premises close to such industries. In contradistinction, “bad” delocalization can be seen as defensive: high imports, low exports and the focus falls on the difficult features such as unskilled manpower resources and repetitive operative tasks. The worst case is that of mono- sectorial specialized activities that run the risks of defensive delocalization.
The lesson here is that one must avoid mixing and confusing one’s objectives. The main risk can be exemplified by the French competitiveness clusters that wished to mix two contradictory aims: reducing corporate vulnerability, on one hand, and gaining technological competiveness on the other. In a short term and mid-term perspective, the main competitiveness clusters will destroy employment positions, which can be seen as normal inasmuch as their mission focuses on higher productivity gains and skilled labor. But they will create tomorrow’s activities, tomorrow’s wealth and hence new employment possibilities.