The puzzle of…

Ed Humpherson, Director General for Regulation, writes about stagnant productivity and stalling life expectancy.

An excellent blog on life expectancy by Veena Raleigh of the King’s Fund, published by the Health Service Journal, makes for sobering reading.

The blog, called “UK’s stalling life expectancy: where do we go from here?”, sets out the evidence on life expectancy in the UK. It outlines how the latest data shows that life expectancy at birth is stalling – after decades of steady improvement. It also shows how the UK may be falling behind other countries.

The blog goes on to outline a range of factors that could be contributing to these phenomena.

All of them are important in their own right as factors to consider when looking at life expectancy.

But I was struck by another thought, an echo from a completely different domain.

There is another area where experts are concerned with a break in the long-term UK trend, and the UK falling behind other countries.

That is productivity.

Productivity is an economic concept that describes how much economic output is produced for a given set of inputs. The productivity puzzle deals with the concern that the UK’s labour productivity has, since 2008, behaved in ways that economists haven’t expected.

Here’s the Bank of England’s Andy Haldane on the productivity puzzle in 2017:

“Productivity growth has consistently underperformed relative to expectations, since at least the global financial crisis.  This tale of productivity disappointment, in forecasting and in performance, has been extensively debated and analysed over recent years.  Some have called it the ‘productivity puzzle’ [ ]

He followed this in 2018 by adding “..it is cold comfort that the UK shares this problem with much of the Western World…because the UK’s productivity slowdown appears to have been larger than almost any other country….Fact one is that UK productivity has flatlined for a decade. This means that UK productivity is running almost 20% below its level had it continued its pre-crisis trend… Fact two is that there is a second gap…between levels of productivity in the UK and in our main competitors, the US, Germany and France” [  ]

Or, more succinctly, the FT’s Neil Collins: “British productivity used to grow at a stolid 2 per cent a year but in the last decade it’s hardly grown at all.”

To be clear. I’m not arguing for a second that there is any causal link between the productivity slowdown and the apparent life expectancy slowdown. Economist talk of “secular stagnation” in productivity, and demographers talk tentatively of “stalling” life expectancy improvements. But I’ve seen no evidence of links between the two phenomena.

I’m more interested in it as an example of problems highlighted by statistics: the economic statistics on the one hand, and mortality statistics on the other. In both cases, official statistics have revealed a set of issues that have puzzled commentators. So in a way I’m interested in whether there are any common patterns in the structure of the discourse around productivity that may help in the discourse around life expectancy.

And my tentative view is that there are things to learn from the these parallel debates:

  • The common structural feature of the discourse is that a problem is highlighted by very aggregated statistics. In both cases, people start from the presumption that trends established over a long time series should be expected to continue – and if they don’t, and it isn’t entirely clear why, then there is a “puzzle”.
  • A range of explanations is put forward to explain the aggregate phenomenon (in the 2017 Andy Haldane speech on productivity, he lists 5 candidate explanations of that puzzle, including credit scarring and slowing innovation). In the life expectancy debate, issues as varied as austerity and virulent flu have been put forward.
  • Some hypotheses can become quite established as explanations, before the evidence is really available to support or challenge them. A good example is the issue of “zombie companies” in the case of productivity – poor performers that should be swept aside by more productive competitors, but which are kept afloat by low interest rates. This was for a time quite a common narrative about productivity. The debate seems to have moved on now. Commentators on the life expectancy puzzle should be wary about plumping for a single explanation too soon, and perhaps even more wary of suggestions that there is a single explanation at all.
  • There’s no substitute for detailed analytical engagement with the data. Diving down into the datasets can identify the drivers of puzzling results. It’s great that Public Health England and other public health agencies are doing exactly this in the case of life expectancy. Their report is expected soon. Curiosity about the data is key.
  • Even the microdata will not answer everything. Imagine a (made-up) scenario in which the micro data tell us –for example– that a factor in the productivity puzzle is the performance of small, family owned business. But this wouldn’t answer the question as to why – why (in this example) do some kinds of firms behave differently to others? That might be hard to ascertain from the micro data alone. There may always a degree of kicking the analytical can down the road, and “More research needed” may well be the verdict of a lot of reports on these puzzles.

I think all this points to a set of lessons from the productivity case for the life expectancy “puzzle”. Keep a wide range of explanations in mind; don’t get too fixed on one explanation to exclusion of others; remember that the best way of understanding the issue lies in the data themselves; but recognise that even a close analysis of the data at a micro level might not yield all the answers. And be ready to challenge presumptions that long-term trends automatically continue into the future. There is however one difference: in productivity, there has been some question as to whether the contemporary economy has become much harder to measure, and this may be effecting estimates of productivity. It’s possible there are measurement issues in life expectancy but they have been less prominent in discussions to date.

But there is also a cause for optimism. After a period of bafflement, and pet theories, and false starts, there is an emerging consensus supported by data in the UK that the primary underlying driver of the productivity puzzle is the weakness in the diffusion of innovation from leading to laggard firms. Focus on improving that diffusion, economists argue, and the productivity problems may start to be addressed. Measuring diffusion may therefore become an important area for official statistics.

I only hope that the life expectancy questions prove easier to resolve that the productivity puzzle. It is a very important issue.

After all, what Paul Krugman said of productivity is even more true of life expectancy. In the long run, it’s not everything. But it’s nearly everything.