The researchers found that the most productive firms have higher proportions of skilled workers. But then they looked more closely at the causes ...
Everyone knows about Gordon Brown's infatuation with Prudence. But lately he seems just as keen on her more cerebral sister, Productivity.
Brown has thrown all manner of tax breaks, credits and initiatives at the so-called productivity gap between the UK and some of its competitors, to little apparent avail. For a long time, middle managers took the blame, but they have just been exonerated by no less a figure than Michael Porter.
But one of the elements of the high-productivity drive that continues to animate ministers is the need to move to a high-skill economy, the presumption being that high skills will bring high productivity along on their coat-tails. And although Charles Clarke, the education secretary, and Brown are not the closest of political allies, on this they are joined at the hip - witness Clarke's recent rant against the 'medieval' concept of universities as places pursuing learning for its own sake, and his renewed call for educational establishments to produce skills 'required by the economy'.
There is a question about whether productivity growth matters all that much in already wealthy nations (leisure growth might be more important) but, leaving that knotty one aside, we need more compelling evidence that increasing the supply of skilled workers will increase productivity.
New research undertaken by Jonathan Haskel, Denise Hawkes and Sonia Pereira at the government-funded Centre for Research into Business Activity (CeRiBA) seems, at first glance, to support the Brown-Clarke orthodoxy. Analysing the first data-set to match individual employee skills with firm-level performance, Haskel and co find that the most productive firms have significantly higher proportions of skilled workers. So the top 10% of firms, measured by productivity, hire workers with, on average, two years more education compared to the bottom tenth.
But then the researchers dug a little deeper - which, as journalists could have warned them, all too often ruins the story. They investigated the ingredients of higher productivity, and found that the difference in skills levels accounted for less than 10% of the productivity gap between firms at the top and those at the bottom. Investment, capital availability and technology all seem to carry more weight. This latter finding may explain why the sponsoring government departments have not been bombarding Fleet Street with press releases hailing this groundbreaking research.
The CeRiBA findings might seem puzzling. How can it be that higher skill levels are correlated with higher productivity, yet higher skill levels do not account for much of that productivity? The answer may lie in one of the most dangerous misuses of research - mistaking correlation for causation. Proving that X is correlated with Y is not the same as saying that X is the cause of Y. Often, the causality lies in the opposite direction.
In this instance, it seems perfectly plausible that highly productive firms, having become more successful, are able to attract and pay workers with higher skills. Only by looking over time at the performance of firms, and the skills of their staff, could this be proved or disproved.
There are countless other examples of correlations that are immediately assumed to denote a causal process. Long hours may be correlated with divorce, but who's to say the long hours aren't caused by the toxicity of the relationship? People who watch lots of TV are more depressed - yes, but what else are depressed people supposed to do?
And the area of workplace performance is replete with correlations masquerading as causalities. We know that firms with more women on their boards are more profitable, but we have no idea if the fact of having more women caused the profits, or the fact of being profitable allowed or forced these firms to shape up on the gender front. We know that companies with progressive HR polices post better financial performance, but we have no idea which came first. The only way to answer these questions is to track organizations over time: but longitudinal research is long-winded, laborious and expensive. It's much more tempting to go quick and dirty.
The media must shoulder its share of the blame. Most reporters are only dimly aware of the difference between a cross-sectional study and a longitudinal one, and life would be much less interesting if stories had to carry the caveat 'or vice versa' at the end of almost every intro paragraph.
Especially in areas that influence public policy, there's a desperate need for deeper consideration of real research evidence. The work by Haskel and colleagues represents the first robust, empirical look at the question of whether skills drive up productivity. The fact that their tentative answer runs counter to prevailing wisdom should provoke an energetic public policy debate rather than a silence. John Maynard Keynes famously said: 'When the facts change, sir, I change my mind. What do you do?' It seems we plunge blindly on.