SAN FRANCISCO – For years, one of the big puzzles in economics has been accounting for declining productivity growth in the United States and other advanced economies. Economists have proposed a wide variety of explanations, ranging from inaccurate measurement to “secular stagnation” to questioning whether recent technological innovations are productive.

But the solution to the puzzle seems to lie in understanding economic interactions, rather than identifying a single culprit. And on that score, we may be getting to the bottom of why productivity growth has slowed.

Examining the decade since the 2008 financial crisis – a period remarkable for the sharp deterioration in productivity growth across many advanced economies – we identify three outstanding features: historically low growth in capital intensity, digitization, and a weak demand recovery. Together these features help explain why annual productivity growth dropped 80%, on average, between 2010 and 2014, to 0.5%, from 2.4% a decade earlier.

Start with historically weak capital-intensity growth, an indication of the access labor has to machinery, tools, and equipment. Growth in this average toolkit for workers has slowed – and has even turned negative in the US.

In the 2000-2004 period, capital intensity in the US grew at a compound annual rate of 3.6%. In the 2010-2014 period, it declined at a compound annual rate of 0.4%, the weakest performance in the postwar period. A breakdown of the components of labor productivity shows that slowing capital-intensity growth contributed about half or more of the decline in productivity growth in many countries, including the US.

Growth in capital intensity has been weakened by a substantial slowdown in investment in equipment and structures. Making matters worse, public investment has also been in decline. For example, the US, Germany, France, and the United Kingdom experienced a long-term decline of 0.5-1 percentage point in public investment between the 1980s and early 2000s, and the figure has been roughly flat or decreasing since then, creating significant infrastructure gaps.

Intangible investment, in areas such as software and research and development, recovered far more quickly from a brief and smaller post-crisis dip in 2009. Continued growth in such investment reflects the wave of digitization – the second outstanding feature of this period of anemic productivity growth – that is now sweeping across industries.

By digitization, we mean digital technology – such as cloud computing, e-commerce, mobile Internet, artificial intelligence, machine learning, and the Internet of Things (IoT) – that is moving beyond process optimization and transforming business models, altering value chains, and blurring lines across industries. What differentiates this latest wave from the 1990s boom in information and communications technology (ICT) is the breadth and diversity of innovations: new products and features (for example, digital books and live location tracking), new ways to deliver them (for example, streaming video), and new business models (for example, Uber and TaskRabbit).

However, there are also similarities, particularly regarding the effect on productivity growth. The ICT revolution was visible everywhere, the economist Robert Solow famously noted, except in the productivity statistics. The Solow Paradox, as it was known (after the economist), was eventually resolved when a few sectors – technology, retail, and wholesale – ignited a productivity boom in the US. Today, we may be in round two of the Solow Paradox: while digital technologies can be seen everywhere, they have yet to fuel productivity growth.

MGI research has shown that sectors that are highly digitized in terms of assets, usage, and worker enablement – such as the tech sector, media, and financial services – have high productivity. But these sectors are relatively small in terms of share of GDP and employment, whereas large sectors such as health care and retail are much less digitized and also tend to have low productivity.

MGI research also suggests that while digitization promises significant productivity-boosting opportunities, the benefits have not yet materialized at scale. In a recent McKinsey survey, global firms reported that less than a third of their core operations, products, and services were automated or digitized.

This may reflect adoption barriers and lag effects, as well as transition costs. For example, in the same survey, companies with digital transformations under way said that 17% of their market share in core products or services was cannibalized by their own digital products or services. Moreover, less than 10% of the information generated and that flows through corporations is digitized and available for analysis. As these data become more readily available through blockchains, cloud computing, or IoT connections, new models and artificial intelligence will enable corporations to innovate and add value through previously unseen investment opportunities.

The last feature that stands out in this period of historically slow productivity growth is weak demand. We know from corporate decision-makers that demand is crucial for investment. For example, an MGI survey conducted last year found that 47% of companies increasing their investment budgets were doing so because of an increase in demand or demand expectations.

Across industries, the slow recovery in demand following the financial crisis was a key factor holding back investment. The crisis increased uncertainty about the future direction in consumer and investment demand. The decision to invest and boost productivity was correctly deferred. When demand started to recover, many industries had excess capacity and room to expand and hire without needing to invest in new equipment or structures. That led to historically low capital-intensity growth – the single biggest factor behind anemic productivity growth – in the 2010-2014 period.

But, as more companies adopt and learn through digital solutions, and as new forms of employment and investment opportunities strengthen the demand recovery, we expect productivity growth to recover. Myriad factors contribute to productivity gains, but it is the twenty-first century’s steam engine – digitization, data, and its analysis – that will power and transform economic activity, add value, and enable income-boosting and welfare-enhancing productivity gains.

CAMBRIDGE – In recent years, productivity growth in developed economies has been stagnating. The most prominent explanations of this trend involve technology. Technological progress is supposed to increase economies’ productivity and potential growth. So what’s going on?

Harvard’s Martin Feldstein has argued persuasively that productivity growth is actually higher than we realize, because government statistics “grossly understate the value of improvements in the quality of existing goods and services” and “don’t even try to measure the full contribution,” of new goods and services. Over time, he asserts, these measurement errors are probably becoming more important.

Northwestern University’s Robert Gordon is less optimistic. He has argued – also persuasively – that today’s innovations in areas like information and communications technology (ICT) cannot be expected to have as big an economic payoff as those of the past, such as electricity and the automobile.

But it’s possible that ICT and other new technologies are not just doing less to boost productivity than past innovations; they may actually have some negative side effects that undermine productivity and GDP growth. One need not be a modern-day Luddite to acknowledge the potential productivity pitfalls of technological innovation.

The first might seem obvious: technological disruption is, well, disruptive. It demands that people learn new skills, adapt to new systems, and change their behavior. While a new iteration of computer software or hardware may offer more capacity, efficiency, or performance, those advantages are at least partly offset by the time users have to spend learning to use it. And glitches often bedevil the transition.

The fast-changing nature of today’s digital technologies also raises security challenges. Spam, viruses, cyberattacks, and other kinds of security breaches can impose major costs on businesses and households.

Then there is the impact that connectivity has on our daily lives, including our ability to work and learn. Non-work emails, social media, Internet videos, and videogames can easily distract employees, offsetting at least some of the productivity-raising potential of that same connectivity. Such disadvantages may become even more pronounced when workers telecommute.

Similarly, the smart phone has shaped the minds of young people, who barely remember what it was like before addictive activities – from video games to social media – were constantly at their fingertips. According to one recent study, recreational computer activities partly explain a decline in labor supply among men ages 21 to 30. Moreover, research shows that laptops in the classroom slow student learning, even when used to take notes, rather than surf the web.

Moreover, smart phones undermine physical safety in some contexts. In the United States, the National Highway Traffic Safety Administration reports that 3,477 people were killed and 391,000 were injured in motor vehicle crashes involving distracted drivers in 2015, with texting being the biggest culprit, particularly among young people.

Digital currencies like Bitcoin have also so far failed to live up to the hype surrounding them. Far from being more efficient as a means of payment or store of value than conventional money, cryptocurrencies seem to encourage the diversion of resources away from productive uses. They also harm the environment, owing to the energy-intensive “mining” process, while the total anonymity they offer undermines law enforcement.

Beyond new technologies’ direct and indirect negative effects on productivity, there is a risk that they are undermining people’s quality of life. Few people have positive feelings about, say, the automatic phone calls that have come to plague many of our lives.

Then there is the ever-present “fake news” problem. The advent of digital “new media” was once heralded as a democratizing trend that would give ordinary people a measure of control over the “air waves,” at the expense of big companies or established institutions. But it has lately become apparent that “democratizing” information may not actually be good for democracy. For example, fake news has been found to spread faster on Twitter than true news. This has not only made citizens less informed in many cases; it has also enabled public figures – most notably, US President Donald Trump – to dismiss the truth as “fake.”

And these are just the downsides of information technology. Other technological innovations with major obvious drawbacks include opiate painkillers and increasingly advanced weaponry.

To be clear, I am not suggesting that the net effects of recent technological advances are negative. On the contrary, many have delivered huge benefits, and that will probably continue to be the case.

Technologies may have productivity-raising potential that is yet to be tapped. Historians like Paul David and technology experts like Erik Brynjolfsson, Daniel Rock, and Chad Syverson argue that it has always taken time for major breakthroughs (like the steam engine, electricity, or the automobile) to yield net economic gains, because businesses, buildings, and infrastructure need to be re-configured. Presumably the same will happen with recent technologies.

But this is not a reason to ignore the negative consequences of new innovations. As a group of Silicon Valley technologists has warned, “Technology is hijacking our minds and society.” We must take back control, ensuring that we do not just make our world “smarter,” but also make sure we are smart about how we use it.

LONDON – Everybody agrees that better education and improved skills, for as many people as possible, is crucial to increasing productivity and living standards and to tackling rising inequality. But what if everybody is wrong?

Most economists are certain that human capital is as important to productivity growth as physical capital. And to some degree, that’s obviously true. Modern economies would not be possible without widespread literacy and numeracy: many emerging economies are held back by inadequate skills.

But one striking feature of the modern economy is how few skilled people are needed to drive crucial areas of economic activity. Facebook has a market value of $374 billion but only 14,500 employees. Microsoft, with a market value of $400 billion, employs just 114,000. GlaxoSmithKline, valued at over $100 billion, has a headcount of just 96,000.

The workforces of these three companies are but a drop in the ocean of the global labor market. And yet they deliver consumer services enjoyed by billions of people, create software that supports economy-wide productivity improvements, or develop drugs that can deliver enormous health benefits to hundreds of millions of people.

This disconnect between employment and value added reflects the role of information and communications technology (ICT), which is distinctive in two crucial respects. First, in line with Moore’s Law, the pace of hardware productivity improvement is dramatically faster than it was at earlier stages of technological change. Second, once software is created, it can be copied limitless times at almost zero marginal cost. Taken together, these factors enable low-cost automation of ever more economic activities, driven by the high skills of only a tiny minority of the workforce.

Despite this phenomenon, more people than ever seek higher education levels, evidently motivated by the fact that higher skills bring higher pay. But many higher-paid jobs may play no role in driving productivity improvement. If more people become more highly skilled lawyers, legal cases may be fought more effectively and expensively on both sides, but with no net increase in human welfare.

The economic consequences of much financial trading are similarly zero-sum. But so, too, may be much of the activity devoted to developing new fashions or brands, with high skill and great energy devoted to competing for consumer attention and market share, but none of it necessarily resulting in an increase in human welfare.

More people receiving higher education does not therefore mean that their higher skills in all cases drive productivity growth. And rising university tuition and fees – growing in the US at a trend annual rate of about 6% in real terms – may not indicate that ever-higher skills are needed to perform specific jobs. Rather, future job applicants may simply be willing to spend a lot of money to signal to employers that they have high-value skills.

Universities, in turn, can become caught in a zero-sum competition of ever-increasing expenditure to attract paying students. And rapidly rising student debt – up from $400 billion to $1.3 trillion in the US alone since 2005 – may partly be financing more intense competition for high-paid jobs, not socially required investments in human capital.

Likewise, at the lower end of the income scale, it is not clear that better skills can significantly offset rising inequality. New jobs can always be created as we automate away many existing jobs, but the new jobs often pay less.

Projections by the US Bureau of Labor Statistics (BLS) for job creation over the next ten years illustrate the pattern. Of the top ten occupational categories that account for 29% of all forecast job creation, only two – registered nurses and operational managers – pay more, on average, than US median earnings, while most of the other eight pay far less.

Employment is growing fastest in face-to-face services such as personal care. These jobs are more difficult to automate than manufacturing or information services; but, according to the BLS, they require only limited formal skills or on-the-job training. And job categories that require specialist ICT skills do not even make the top ten. The BLS foresees 458,000 more personal-care aides and 348,000 home health aides, but only 135,000 more software and application developers.

But wouldn’t better skills enable people currently in rapidly growing but low-pay job categories to get higher paid jobs? In many cases, the answer may be no. However many people are able to code, only a very small number will ever be employed for their coding skills. And even if someone currently in a low-skill job is equipped to perform a high-skilled one at least adequately, that job may still go to an employee with yet higher skills, and the pay differential may still be great: in many jobs, relative skill ranking may matter more than absolute capability.

So “better education and more skills for all” may be less important to productivity growth and a less powerful tool to offset inequality than conventional wisdom supposes. But that would not undermine in the least the personal and social value of education.

As many people as possible should be highly literate, aware of and fascinated by the basics of science, and enthused by and able to understand good design or music. After all, in a world where automation can free us from the drudgery of endless work, a good education will better equip us to live satisfying lives, regardless of whether it increases individual pay or measured prosperity.

As for inequality, we may need to offset it through overt redistribution, with higher minimum wages or income support unrelated to people’s price in the job market, and through generous provision of high-quality public goods.

In a world where robots can increasingly do the work, education and skills are more important than ever – not because they can raise everyone’s price in the labor market, but because they can equip us for lives in which many jobs no longer deliver adequate income, satisfaction, or status.

NEW YORK – From the early nineteenth century to the early twentieth century, Western countries attributed their economic growth to the discoveries of “scientists and navigators.” A country needed only the “zeal” to develop “obvious” commercial applications, and build the facilities to meet demand for new products.

Until recently, the Chinese believed the same thing. But now, Chinese businesspeople and entrepreneurs are increasingly showing not only the entrepreneurial drive to adapt to new opportunities, but also the desire and capacity to innovate for themselves, rather than simply copying what’s already out there.

Indeed, more and more Chinese companies are realizing that they must innovate in order to get – and stay – ahead in the global economy. Several companies – notably Alibaba, Baidu, and Tencent – made breakthroughs, by offering digital-age infrastructure that facilitates innovative activity. And industrial firms have recently moved into robots and artificial intelligence.

For its part, China’s government is evidently supportive of Chinese businesses developing a capacity to produce indigenous innovations. It no doubt recognizes that such innovations are all the more valuable when innovation remains weak in the West, where growth in total factor productivity (TFP) has continued its long slowdown.

In recent years, China’s government has introduced initiatives aimed at increasing both entrepreneurship and innovation. It has shortened dramatically the process for forming a new company. It has built a vast number of schools, where Chinese children learn more about the world they will face. And it recently facilitated the entry of foreign experts to work on new projects in the business sector.

The authorities have also recognized the importance of allowing more competition in the economy. Individuals should be freed up to start new companies, and existing companies should be freed up to enter new industries. Competition solves a lot of problems – a point that is increasingly lost on the West.

At the World Economic Forum’s annual meeting in Davos, Switzerland, in January, Chinese officials discussed basic reforms that the government introduced two years ago to increase competition. Under the new policy, excess capacity now signals that supply should be allowed to contract and prompt redundant firms to exit the market. Of course, excess demand signals that supply should be allowed to increase, leading to the entry of new firms.

The key insight is that when existing enterprises are protected from new market entrants bearing new ideas, the result will be less innovation and less “adaptation” to a changing world, to use Friedrich Hayek’s term.

Another argument can be made. In any modern economy, virtually every industry operates in the face of a largely unknowable future. The more companies an industry has thinking about a problem, the more likely a solution is to be found. A company that has been kept out of an industry might know something that all the companies in the industry do not. Or some unique experience may have furnished an individual with “personal knowledge” that is impossible to transmit to others who have not had the same experience. Whatever the case, society benefits – through lower prices, more jobs, better products and services, and so forth – when outsiders with something to add are free to do so.

All of this was known to the great theorists of the 1920s and 1930s: Hayek, Frank Knight, and John Maynard Keynes. And now it is known to the Chinese, who understand that a country benefits when companies – each with its own thinking and knowledge – are free to compete.

The West seems to have forgotten this. Since the 1930s, most Western governments have seen it as their duty to protect established enterprises from competition, even when it comes from new firms offering new adaptations or innovations. These protections, which come in myriad forms, have almost certainly discouraged many entrepreneurs from coming forward with new and better ideas.

History is rife with evidence of the value of competition. In post-war Britain, into the 1970s, industries were controlled by exclusive clubs within the Confederation of British Industry, which barred new entrants. By the time Margaret Thatcher became prime minister in 1979, TFP had stagnated. But Thatcher put a stop to the Confederation’s anti-competitive practices, and Britain’s TFP was growing again by the mid-1980s.

We are now seeing something similar in China. By 2016, China’s TFP growth rate had been slowing for a number of years. But since the reforms that year, it has been increasing.

The West must address its great TFP slowdown, which has lasted since the late 1960s. Ending protection of incumbents from new entrants possessing ideas for new adaptions and innovations is a good place to start.

MILAN – Economists concern themselves not only with addressing difficult questions thoughtfully, but also with formulating the questions themselves. Sometimes, rethinking those questions can hold the key to finding the answers we need.

Consider the productivity debate. Economists trying to explain the apparent structural slowdown in productivity growth have been asking the following question: Where is the missing increase? Their response covers concerns about measurement, structural shifts in the labor market, a potential paucity of investment opportunities, productivity-diluting technological innovations, and technology-driven skills mismatches.

But it may also be useful to consider a more fundamental question: How much productivity growth do we really want, and at what cost?

There is no doubt that productivity growth is desirable. It is a primary driver of GDP growth (especially in countries where labor-force growth is slowing) and income gains. Strong GDP growth and rising incomes can then support the fulfillment of fundamental human needs and desires.

This link is particularly obvious in developing countries, where economic expansion and rising incomes are preconditions for poverty reduction and improvements in health and education. But the link between aggregate growth and individual welfare is no less visible in advanced countries – particularly those now struggling with slow growth, high unemployment, output gaps, debt overhangs, misaligned exchange rates, and structural rigidities.

But this does not mean that policymakers’ primary goal should be more productivity growth. Societies – including governments and individuals – care about a range of things, from health care and security to fairness and freedom. Inasmuch as productivity growth – and, in turn, GDP and income growth – advances these societal objectives, it is highly desirable.

There is, however, a tendency among economists and policymakers to overemphasize such market-related measures of performance, while overlooking the reason why that performance matters: human wellbeing. Efforts to implement a more comprehensive framework for assessing economic performance, one that reflects social needs and desires, have been largely unsuccessful.

In order to determine how much productivity growth we want, we need to take a broader view, one that enables us to decide how best to allocate society’s limited resources, especially its most valuable human resources. Such a perspective should recognize the possibility that market-related measures, particularly real (inflation-adjusted) income growth, may no longer be as important as they were in the past. And it must account for a society’s priorities, revealed in the ways in which its members use their resources.

Health-related discoveries and advances, for example, have brought massive societal benefits since World War II: increased longevity and reduced child mortality and morbidity, not just higher productivity and GDP. That is why the government of, say, the United States invests so much in medical research: the National Institutes of Health alone has an annual budget of $32 billion with which to fund infrastructure and research projects that employ a subset of the country’s greatest scientific talent. Similarly, the National Science Foundation and the scientific research arm of the US Department of Energy receive a combined total of about $12 billion per year, which they use to advance a wide variety of goals in engineering, energy efficiency, and green energy, and the natural and social sciences.

The economic return on public investment is even more difficult to calculate for security-related spending, where the total resources allocated to enhancing it and the effectiveness of those resources may be unknowable. But there is little doubt that security has a powerful claim on people’s wellbeing and thus on resource allocation.

In some cases, people’s desires may actually clash with the goal of improving productivity. Social media, for example, has often been derided as a feeble or even negative contributor to productivity. But productivity is not the point of social media. What people value about it is the connectivity, interaction, communication, and diversion that it enables.

In fact, for many individuals, particularly in wealthier countries, the top priority is not simply becoming richer, but rather living a richer life, and it is toward the latter goal that they will channel their time, income, and creativity. As societies become richer, the relative value placed on different dimensions of life may shift.

Societies’ allocation of resources will imprecisely but persistently follow these shifts. This is especially true when it comes to human resources, but public-sector resources also tend to respond to the same preferences and values over the longer term, regardless of the imperfections in our mechanisms of social choice.

This kind of evolution is not unique to high-income countries. China has reached – or perhaps passed – the stage during which a laser-like focus on productivity and GDP growth corresponds with ordinary citizens’ sense of wellbeing. As a result, China’s resources are increasingly being redeployed toward a more balanced portfolio that still includes growth, but adds environmental protection, social welfare, security, and innovation in a wide range of fields that overlap only partly with productivity and income growth.

All of this suggests that a substantial share of the decline in productivity growth may not be the result of some deep problem with resource allocation or some consequence of exogenous technological innovation cycles over which we have little control. Rather, it could reflect a natural shift in priorities to other dimensions of wellbeing.

This shift is not without its risks. Without productivity growth, the incomes of those at the lower end of the distribution will likely remain flat, exacerbating inequality and, as we have been seeing lately, jeopardizing social and political stability. Given this, governments should devote resources to reducing inequality, regardless of the shifting preferences of the average citizen.

Societies could, we have little doubt, elevate productivity and income growth substantially, if they managed to redeploy their resources entirely in that direction. But whether bucking revealed preferences embedded in private and public investment choices would make us individually and collectively “better off” is dubious, at best. More likely, it is simply not true.