Are large industrial sites the enemy of employment?

From the global Trendeo database, we selected 6009 investments where the amount invested and jobs created were known (for a total of US$2073 billion). We then determined the intervals that would allow us to separate the investments into deciles (each class of project represents 10% of the total amount invested).

We then calculated the number of jobs created per M$ invested. As the graph below shows, the result is fairly indisputable: the M$ invested in the largest projects creates 1.3 jobs, compared with 24 jobs for the smallest projects.

Without recalculating the investment classes, we looked for the corresponding ratios for France: from 4.4 jobs for small projects to 0.5 jobs for the largest. The number of jobs is not the same, as the French economic fabric is more developed than the global average (our global data include projects in Bangladesh, where several hundred jobs are easily created for a relatively low investment, notably in the textile sector).

Does this mean we should avoid large-scale, job-poor investments?

◼ The first thing to consider is the knock-on effects of a major investment. They are certainly more important than diffuse investments, which have no direct impact on the local fabric, whereas a major project will often attract local subcontractors. The three semiconductor investment projects in Dresden(Infineon Technologies, TSMC and GlobalFoundries) announced in the last three years, totalling $25 billion, are the equivalent of the Sarkozy government's December 2008 stimulus package (not even counting induced jobs).

◼ And don't forget that investment normally translates into production (although, as Noah Smith pointed out with regard to recently announced major U.S. projects, it's hard for them to get off the ground, sometimes because of administrative constraints). Production is proportional to investment, and the production/investment ratio is likely to vary little with project size. The ranking of productivity per job is therefore exactly the opposite of that of jobs per million invested. One million in a large project will be 10 to 20 times more productive. More productivity also means, more or less directly, better salaries and a means of limiting recruitment difficulties.

◼ Small and medium-sized investments are therefore necessary to maintain a diversified economic fabric, but at a time when we're wondering about our lost productivity, taking an interest in large and very large projects is also important.