productivity – thinking liberal

Productivity, growth and wellbeing – the awkward triangle

Two recent developments have tickled this cynical old veteran of office work. There was a successful trial of a four day working week. And there is general excitement at the latest thing in Artificial Intelligence – ChatGPT (generative AI, apparently). Both seem to point to improved productivity. But if that’s true it doesn’t follow that economic growth will result.

To my cynical mind office work can be divided into two broad categories: problem-solving and bullshit. The latter seems to take up most of people’s time: talking about solving problems rather than actually solving them. In any office-based environment remarkably few people in organisation actually seem to be productive problem-solvers. The others supervise, communicate, convene meetings, make calls, write presentations, set deadlines, monitor project plans, strategise and so on. Doubtless a lot of the activity I am describing as bullshit contains an element of necessary work, but it often doesn’t feel that way.

So it’s no surprise that some businesses have found that they can reduce office hours without impacting adversely on output when implementing a four-day week. The saving seems to have been in the region of four hours in a five-day week – four eight-hour days making up for five seven-hour ones, for example, though that’s a saving of three hours. ChatGPT, meanwhile, automates the production of bullshit. It manufactures a lot of plausible but unreliable verbiage that you would be unwise to stake much on. Since producing such verbiage is what so many people spend such a lot of effort doing, it’s not hard to see why people are getting so excited. Both ideas offer ways of spending less time doing pointless things. So productivity should improve.

But, of course, it is much harder to see how either invention increases the production of useful things. The idea of a four-day week isn’t to give people the time for side-hustles. The idea is that people get more time for unpaid domestic things (“leisure” is probably an mis-description of this). The study reported high levels of improved wellbeing among employees – which was seen as the main benefit. As for ChatGPT, it’s not meant to solve tough problems or make hard professional calls – the things you most want service providers to do for you – or provide the warmth of human company, though doubtless some people hope that it will help robots to do that job, it sounds a poor substitute.

Doubtless I exaggerate. But there is a more substantial point here. A lot of improvements made to workplace efficiency – improved productivity in economic speak – won’t have much impact on the sort of economic growth you can measure in money and tax – the holy grail for economists and politicians. But that doesn’t mean that people won’t be better off. Wellbeing and per capita economic income or consumption are quite different things. Some people have been saying this for quite a while – Professor Richard Layard for one, and he still is. I met him when I was part of a Liberal Democrat policy working group looking at the issue more than a decade ago. Lord Layard’s big idea is to use self-reported wellbeing as a measure of progress. I am more sceptical – I don’t think the measure is robust enough to do heavy lifting, though it is interesting nevertheless. Still I wish politicians would take up the mantra of improving wellbeing a lot more. The Lib Dem policy paper I co-authored was adopted as official policy and then forgotten. But people are voting with their feet. If growth is slowing because people are opting out of the money economy and improving their health and wellbeing, then that’s to be celebrated. Economists rarely consider this possibility, though. And Conservatives who advocate cutting taxes don’t suggest this so that people can afford to work for fewer hours – though this could be the result. Indeed they think it will increase GDP rather than reduce it.

In my youth I remember a story of some western development experts and going to an African rope factory. They gave them a machine that improved output per poker ten-fold. A year later they returned and were surprised to find the factory empty. “Why aren’t people working,” they asked. “Well, we finish the production in an hour, and then everybody can go home,” was the replay. Doubtless the original story was play on African stereotypes, but even at the time, we weren’t clear the the joke was supposed to be on.

The goal of advancing wellbeing while economic growth remains lacklustre is a perfectly feasible one. Improvements to workplace organisation and continued automation have their part to play. But public services and infrastructure can be better directed towards this goal too. And political reform to reduce the feelings of powerlessness will also help. This remains a long way off – but eventually public pressure will force it. If the four-day working week starts to take hold, it will be a major step forward.

Economists are failing to understand the 21st century economy

When is a fact not a fact? When it is an economic statistic. Economists use statistics like GDP, productivity or inflation, as if they were facts. And because economists set the terms of public policy debate, the rest of the world follows them. But they are abstract artefacts, designed to help us see the facts, but which conceal as much as they reveal. There may have been a time when this didn’t matter much. That is certainly not the case now. We are missing something very important.

Economists like to accuse others of the fallacy of composition. That is the assumption that the finances of a nation work on the same principles as that of the households that make it up. For example, the budget constraints in a household work in an entirely different way to those for state finances – which doesn’t stop politicians lecturing us about the absence “money trees”. But economists are the world’s worst at a very similar fallacy: assuming that national level statistics represent the truth for individual households and businesses.

This is very striking with discussions about productivity, such as on this morning’s Radio 4 Today. This is newsworthy here in Britain. Growth in productivity grew at about 2% a year in Britain until 2008, during the financial crisis. It then stopped growing – creating what is often called the “productivity puzzle”. This is happening, to a lesser extent, in other developed countries too. It is the news now because the Office for Budget Responsibility has admitted that its assumption that productivity growth would revert to historical norms has failed to materialise for yet another year. That matters because it affects forecast tax revenues, around which the Chancellor of the Exchequer must base his annual budget, due next month. It is also the explanation offered for the fact that rates of pay seem to have stagnated for many Britons.

What is productivity? It is output per worker per period worked (typically a day or an hour). If you work in a ball-bearing factory it is relatively easy to understand what this actually means. You count the number of ball-bearings produced in a month, say; you count the number of hours worked in that month; you divide one by the other. Because of this simplicity it is easy to imagine the national economy made up entirely of ball-bearing factories or close equivalents, so that if productivity rises, it means that more ball bearings are being produced for the same number of hours. And most of the discussion about productivity uses something like this mental picture. Economist speculate about businesses using cheap labour as a substitute for upgrading capital equipment, for example. This is like the butler in Kazuo Ishiguro’s The Remains of the Day who immerses himself in his familiar daily duties so as not to confront the realities of the changing world around him.

To understand this, consider two groups of problems. The first type of problem is how do you actually measure output? The more you consider this, the harder it gets. Start with some obvious problems: what is the output of a squadron of jet fighters? Then how do you compare organic carrots with carrots grown on a farm pumped with environment-degrading chemicals? Or an iPhone 6 with an iPhone 8? Or is a loss-making factory with few people and overpriced robots really more efficient that a profitable one with lots of labourers and primitive machinery? Economic statisticians wrestle with such questions, but largely hidden from view. Economics undergraduates are barely troubled with such issues, and are quickly ushered on to the certainties of supply and demand curves and medium term fiscal policy. The result is a series of estimates and fixes, but not enough discussion about what they all add up to. Occasionally a conservative politician will pop up to suggest that lower-paid workers should be much happier because improvements in technology aren’t properly reflected in inflation. But that’s about it.

The second group of problems is conceptually somewhat easier: it is variations in the composition of the economy. Overall productivity may be improving not because individual businesses, or even industries, are becoming more efficient, but because industries with a higher measured productivity are taking a larger share. In fact it turns out that this was largely the case in Britain before 2008, with the expansion of banking and professional services. Just how productive these industries were in reality was thrown sharply into question in the following year, when banking threw up eye-watering levels of losses, from which government finances have never recovered.

These difficulties have been known about for a long time, and professional economists are more aware of them than the public whom they lecture. But they are shrugged off, with the assumption that it all comes out in the wash. Policymakers should be doing something about productivity, they say. This needs some serious challenge. And the consequences of that challenge are profound.

My first challenge is known as the Baumol effect, or more usually known, revealingly, as “Baumol’s cost disease”, as if it was something to be eradicated rather than an ordinary physical constraint. Suppose a legal firm adopts a highly efficient artificial intelligence system and makes most of its workers redundant. And then those workers take up jobs such as being personal trainers or producing craft pottery. All the positive productivity effects of the firm’s investment in AI is neutralised by the displaced workers moving into low-productivity jobs. Perhaps we are at state where most productivity improvements are being statistically neutralised in this way? After all, the more efficient an industry becomes, the fewer jobs in that industry there are overall.

Now let’s get into territory that mediocre economists really want to avoid: human alienation. One way of looking at productivity is that it advances by two alternative means: cutting wastage and cutting human content. By and large nobody will argue with cutting wastage. Vacuum cleaners have liberated home-keepers; time spent in queues is good for nobody; and so on. But cutting human content, and human contact, is distinctly two-sided. This is the world of standardised products, specialisation and automatic interfaces that may reduce costs but leave workers and users alike cut off from their fellow human beings, and being forced to conform to somebody else’s idea of what they should be. That is alienation, an idea first made famous by none other than Karl Marx in the Communist Manifesto – a document of startling insight. Alienation can be to the good overall, but it often isn’t. And there are bad signs all around us, from the obesity epidemic, the poor mental health of teenagers (especially girls), to the breakdown of community activities in our cities. Now, to back to my example of the legal firm, what if one of those displaced admin assistants finds life much more fulfilling as a personal trainer, even if the pay is miserable? A conventional economist would fret that surely it is better for her clients to download a demo video from You-Tube so that the trainer can work in a soulless call-centre? Well it would for those aggregated statistics, but not necessarily for the state of the human condition. In fact many economists suffer from acute double-think. On the one hand they praise markets and the wisdom of freely made choices of individuals over bureaucratic planners. And yet when those freely made choices go to more leisure, low productivity work and locally-sourced vegetables, they moan like mad.

But there is one sense in which those economists are right. Those aggregate statistics have one useful purpose: in planning taxes. Taxes are based on the money economy, which is the foundation of those statistical measures. If the stagnation of productivity is a fact of life, and actually represents a struggle against human alienation, then tax revenues are going to stagnate unless the rates increase. And since demand for tax funded services is liable to keep rising, we are going to have to think very hard how we order people’s relationship to the state. We are surely heading for an era of higher taxes. But how to design these taxes?

Instead we just get useless calls for action to raise productivity. Time to move on.

Productivity statistics expose deep weaknesses in theoretical economics

I hadn’t intended to post for another couple of weeks, but this article in the Financial Times is too good to miss. It tackles one of the central issues in modern economic debate: why productivity growth is so slow. Productivity lies at the heart of the conventional view of public policy – and yet it is very poorly understood. This article sheds light on what is happening in the UK – and it should give politicians and economists pause.

Productivity is in principle a very simple idea. It is the amount produced by a unit of labour in a unit of time – the number of widgets per person per hour, for example. This immediately conjures up a clear mental picture of a factory producing cars, say. Count the number of cars produced, and the number of hours of labour required and it is easy-peasy, surely? Alas in a modern economy it is a much more difficult idea. What if your car factory is producing both Ford Fiestas and Mondeos, and switches to the smaller car? Has productivity gone up if more are produced? And how do you distinguish product enhancement from inflation? And then there are problems treating capital outputs and inputs, research and development, and so on. In the end the productivity measured across an economy is a bit of a balancing figure, as we accountants would call it – or a bit of a dustbin – what’s left when you’ve taken everything else out. It is just a number relationship without a coherent meaning in its own right. It is not like the concepts that physical scientists are used to dealing with – such as the temperature and pressure of a gas. Macroeconomics is heterogeneous, to say nothing of being subject to capricious social forces that tend to corrupt all attempts at measurement.

Now, what is the productivity puzzle? It is that productivity growth, as measured by macroeconomic statisticians, has slowed markedly since 2008, when the financial crash caused a dislocation in measured income. This applies to all developed economies, but to the British economy most of all – UK productivity growth, according to the article, fell from 1.6% per annum before 2008 to just 0.3% after. This has profound implications, since in the long term productivity growth is what drives income per head, alongside the average hours people work (influenced strongly by workforce participation – such as how many women are in paid employment). And this drives tax revenues, from which public services are funded. Since we assume that quality of life is mainly driven by income, and that public services can constantly be enhanced by extra spending (apart from occasional periods of “austerity”), this has profound implications. Prior to 2008 most economists assumed that productivity growth of 1-2% pa was a law of nature and main driver of “trend growth”, which could be baked into economic models. The corollary was that weak growth since 2008, and the failure of GDP to catch up with the pre 2008 trend-line, was a failure in macroeconomic policy.

But given the dustbin nature of the productivity statistics, it is very hard to drill down into them to find out just where the problem is – though that there is a problem of some sort is clear. This is licence for all manner of people to project their speculations into a fact-free zone. Mostly these are based on the intuitively obvious idea that the changes to the productivity figures represent trends in the efficiency of workers. Recently Bank of England bigwig Andrew Haldane moaned that the problem was that efficiency was stuck in a rut, especially in a swathe of mediocre firms. He based this on sectoral analysis which showed that the productivity had stagnated across all sectors – with economic growth mainly attributed to rises in employment, not efficiency.

The FT article, authored by Chris Giles and Gemma Tetlow, challenge that. A close examination of the numbers shows that the crash in productivity growth arises from changes in a small number of economic sectors, accounting for just 11% of income. These are banking, telecoms, electricity and gas, management consultancy, and legal and accounting services. Actually Mr Haldane’s and Mr Giles/Ms Tetlow’s analysis can be reconciled. Mr Haldane was taking a general view across the economy since 2008, where productivity growth is now very limited. The FT writers are looking at the transition from before and after 2008. The curious point is why productivity growth was so high in that small number of industries before 2008 – and the realisation that this is what was driving so much of the figures for productivity growth before that date.

And that leaves this blogger asking whether that pre-crash productivity growth – and by implication the pre-crash trend rate of overall economic growth – was in any sense real, other than statistically. In banking we know that in 2008 massive state resources were required to keep the industry alive, and that since then the industry has been much better controlled. This suggests that “productivity” would more correctly be described as “recklessness”. And in each of the other industries you can point to factors that demonstrate that growth was not simply incremental improvements in efficiency. For example in electricity and gas productivity was based on high inputs of fossil fuels and nuclear energy – and the switch away from these destructive sources of power has caused a decline in measured productivity. And how on earth do you assess the output of management consultancy, and accountancy and legal services? The transition may simply be from high margins in boom economy conditions to higher scrutiny when times were harder – or to put it another way, what was supposedly economic growth prior to 2008 was in fact concealed inflation.

All this supports the narrative that I have been promoting for quite a few years about the transition from growth to austerity. This is that the supposed growth of the economy of the early to mid noughties in the UK was down to excess demand, of which reckless fiscal policy was a part – though you might alternatively argue that it was reckless borrowing by the private sector that the government turned a blind eye to. It also suggests that the lacklustre economic performance of the UK economy since 2008 reflects a lot more than just weak demand management: it is chickens coming home to roost.

This takes me to two very important conclusions. The first is that we have to be very careful about the recommendations of macroeconomists – and the eco-system of commentators and policy types that use macroeconomics as their starting point. The bandying about of aggregate statistics is all very well – but the aggregates hide as well as reveal – and we need to base economic prescriptions on the complexities of the real economy. That is hard, but necessary.

The second point is that overall productivity is indeed stuck in a rut, and has been since well before 2008. It must reflect structural issues in real economy – and not simply laziness amongst mediocre firms or poor macroeconomic management. There is no shortage of potential culprits: demographics; the nature of modern technology; the temporary nature of gains from trade with Asian economies. The world may still be becoming a better place – but because of things that are not captured in GDP, and hence productivity statistics. The problem for public policy is that tax revenues are largely driven by GDP (which is why it is an important statistic) – so we can’t expect an ever increasing flow of tax revenue to fund public services. In the long run we must either reduce the demand for public services (healthier people, fewer crimes, less skewed income distribution, etc.), raise taxes, or compromise what level of services and benefits we think that a civilised state should provide.

And that is a completely new way of thinking about public policy. The political right have grasped this (for the wrong reasons, perhaps) – but the left has not.

Lesson from the banking industry: sometimes people need to be treated as people.

This article from the Economist struck me like a bullet on reading it today. Not so much for the subject matter itself (US banking practices) but what the whole episode says about the modern world. We have never had more data readily available on people – but we seem less able than ever to take decisions on their individual merits. More data, less information. This problem is usually shrugged off y economists and reformers with a laugh; it shouldn’t be.

The story starts in the US property boom, when banks were falling over themselves to offer mortgages, based on the vague idea that since these loans where secured on property, and property values always go up, you couldn’t have too much. The banks stand accused of approving loans robotically, without any consideration of individual merits – and as a result often lending to people who could not afford to keep up with the repayments. This accusation was commonplace, but, as the article points out, little effort seems to have been made to substantiate it against hard evidence.

Then came the crash, and many people who had taken out loans could not or would not keep up with the repayments – and stood at risk of having their homes repossessed. And the banks once again stood accused of carrying out repossession without due care and attention, again on mainly anecdotal evidence. This became a hot political issue, and the individual US states set about suing the banks, with the Federal government becoming involved too. And now an umbrella settlement is proposed, to which the five main US banks and 49 out 50 state Attorney Generals have agreed to. The banks are making a blanket payment to make the problem go away.

What remains characteristic of the whole story, from the original alleged malpractices right up to the settlement, is a failure to reconcile it to what actually happened to real people in real homes. No attempt is made to distinguish between whether some banks are more culpable than others; and no attempt to distinguish between arrears that arise from people in genuine hardship, and those who are trying to beat the system. All that is just too difficult.

And this type of thing is happening all around us. Decisions are made about us using computer algorithms based on data that may or may not be accurate – or based on our membership of some or other broad group of people (men, women, over 50, etc.) and the law of averages. Companies calculate that it is cheaper that way. To consider people as real people, and base decisions on the individual merits of the case, well that requires the intervention of skilled staff, and they cost a lot of money.

And so the flip side to ever advancing productivity (one of the things that makes skilled people cost so much) is that we are subjected to an increasing volume of de-personalised services and arbitrary decisions; and around the fringe a spectrum of fraud arises, as people learn to take advantage of system weaknesses. I have been the subject of mild identity theft several times; this looks quite safe for the people who perpetrate it, since nobody bothers to find them – it’s just a cost of doing business.

But what’s the moral of the story? We gain a lot from the increased wealth that arises because of all this added productivity. And what’s more part of becoming a more equal society is that well off people like me can’t expect to have armies of people running around fawning on their every need. So should I just stop whinging, and get on with all the things I can now do that would have been unthinkable in a previous age?

Up to a point. I think there are two important consequences that many people overlook. One big picture, and the other of more urgency. The big picture point is that are are physical limits to economic growth, and it is no wonder that the pace of growth slows in developed societies. Higher productivity means we consume more services with diluted human content. But huge part of the pleasure we derive from some services is exactly because we get one-on-one attention from somebody (hairdressing perhaps, a personal trainer, dinner at a posh restaurant, and so on); as productivity advances, the proportion that these non-negotiable services comprise in the total economy rises – and so growth slows. Economists refer to this as “Baumol’s Disease” after the economist who originally pointed it out. But it is not a disease; it is the product of success – it’s the process of arriving at the promised land, so to speak – the place that is so good that progress is impossible. An increasing proportion of services cannot be improved without detracting from their value, and people will resist buying them at any price; and that’s saying nothing of the distortion to incentives that arises from making decisions based on averages. We can’t rely on economic growth to wash away society’s problems – we need to confront them more directly.

The more urgent point applies to the reform of public services. Too many people assume that to make these more effective we must follow a similar process of sucking the human content out of them as we see in so many commercial services. In some cases I’m sure that’s true; some Indian organisations are doing amazing things to improve the productivity and effectiveness of certain medical procedures by using economies of scale. But in most cases the effectiveness of public services depends on joining up the dots; seeing people as people rather than collections of unrelated needs that can be picked off one by one. An individual who is committing serial antisocial behaviour offences, may have mental health problems, addiction issues, a dysfunctional family life, educational under-achievement, and inadequate housing. Just from listing them you can see how all these problems are interrelated and feed off each other. We stand a much better chance of making progress if we design solutions based on looking at this individual and his exact personal circumstances and negotiating with him as a human being. Productivity in public services is not about rate of throughput, its about solving problems and reducing demand. This needs a completely different mindset than that needed from the commercial world. Alas too much (though certainly not all) public service reform misses this key point.