5 The End of Growth?

GROWTH ENDED ON October 16, 1973, or thereabouts, and is never to return, according to a wonderfully opinionated book by Robert Gordon.1

On that day, the member countries of OPEC announced an embargo on oil. By the time the embargo was lifted in March 1974, the price of oil had quadrupled. The world economy at this time had become increasingly reliant on oil and was generally facing raw material shortages that were pushing up prices. What followed in the rich countries of the West was a lackluster decade of “stagflation” (economic stagnation accompanied by inflation). Slow growth was supposed to go away but has been with us ever since.

This happened in a world where most citizens of these rich countries had grown up expecting endless and ever-expanding prosperity, where political leaders had grown accustomed to measuring their success in terms of a single yardstick: the rate of growth of the country’s gross domestic product, or GDP. And to a large extent this is still the world we live in, and in some sense we are still talking about that pivotal moment in the 1970s. What went wrong? Was there a policy mistake? Can we coax growth to return and stay? What magic button do we need to press? Is China immune to this slowdown?

Economists have been busy answering these questions. Countless books and papers have been written about them. Many Nobel Prizes have been awarded. After all that, what is it that can be said with confidence about how to make rich economies grow faster? Or does the fact so much has been written signal that we really have no idea? And should we even be concerned?

For the thirty-odd years that separated the end of the Second World War from the OPEC crisis, economic growth in Western Europe, the United States, and Canada was faster than it had ever been in history.

Between 1870 and 1929, GDP per person in the United States grew at a then unheard of rate of 1.76 percent per year. In the four years after 1929, GDP per person went down by a catastrophic 20 percent—it is not called the Great Depression for nothing—but it recovered fast enough. The average yearly growth rate from 1929 until 1950 was actually slightly higher than in the previous period. But between 1950 and 1973, the yearly growth rate went up to 2.5 percent.2 There is more difference than there might appear to be between 1.76 percent and 2.5 percent. It would take forty years for GDP per head to double with a growth rate of 1.76 percent, but only twenty-eight years at 2.5 percent.

Europe had a more checkered history before 1945, partly because of its wars, but after 1945 things really exploded. When Esther was born, late in 1972, France had about four times the GDP per capita than when her mother, Violaine, was born in 1942.3 This was typical of the Western European experience. GDP per capita in Europe increased by 3.8 percent every year between 1950 and 1973.4 It’s not for nothing that the French call the thirty years after the war les Trente Glorieuses (“the Glorious Thirty”).

Economic growth was driven by a rapid expansion in the productivity of labor, or the output produced per hour worked. In the United States worker productivity grew at 2.82 percent per year, which meant it would double every twenty-five years.5 This rise in labor productivity was large enough to more than offset a decline in hours worked per head that was going on at the same time. During the second half of the century, the workweek went down by twenty hours in the US and in Europe. And the postwar baby boom lowered the share of working-age adults in the population since the baby boomers were then, well… babies.

What made workers more productive? In part, they were becoming more educated. The average person born in the 1880s studied only up to seventh grade, whereas the average person born in the 1980s had on average two years of college education.6 And they had more and better machines to work with. This was the age in which electricity and the internal combustion engine came to assume their central role.

Making somewhat heroic assumptions, it is possible to guesstimate the contribution of these two factors. Robert Gordon reckons that rising education explains about 14 percent of the increase in labor productivity over the period, and the capital investment that gave workers more and better machines to work with explains a further 19 percent of the increase.

The rest of the observed productivity improvement cannot be explained by changes in things economists can measure. To make ourselves feel better, economists have given it its own name: total factor productivity, or TFP. (The famous growth economist Robert Solow defined TFP to be “a measure of our ignorance.”) Growth in total factor productivity is what is left after we have accounted for everything we can measure. It captures the fact that workers with the same education level working with the same machines and inputs (what economists refer to as capital) produce more output today for each hour they work than they did last year. This makes sense. We constantly look for ways to use our existing resources more effectively. This reflects in part technological progress: computer chips become cheaper and faster, so one secretary can now do in a few hours the work a small team used to do; new alloys are invented; new varieties of wheat that grow faster and require less water are introduced. But total factor productivity also increases when we discover new ways to reduce waste or shrink the time either raw materials or workers are forced to stay idle. Innovations in production methods like chain production or lean manufacturing do that, as does, say, the creation of a good rental market for tractors.

What made the few decades before 1970 extraordinary compared to much of history is that total factor productivity increased particularly rapidly. In the United States, TFP growth was four times faster between 1920 and 1970 than between 1890 and 1920.7 In fact, it was this rather than growth in education or capital per worker that gave the later period its special mojo. TFP growth in Europe was even faster than in the United States, especially after the war, partly because Europe adopted innovations already developed in the US.8

Rapid growth was not only to be seen in national income statistics. By any measured outcome, quality of life was radically different by 1970 compared to what it was in 1920. The average person in the West ate better, had more heat in the winter and better cooling in the summer, consumed a larger variety of goods, and lived a longer and healthier life.9 With a shorter workweek and earlier retirement, life was no longer quite so dominated by the drudgery of daily labor. Child labor, omnipresent in the nineteenth century, had more or less disappeared in the West. There, at least, children could now enjoy their childhoods.

But in 1973 (or thereabouts) it all stopped. On average, over the next twenty-five years, TFP has grown at only a third of the rate achieved in 1920–1970.10 What started with an economic crisis with a clear start date, and even a set of foreign powers to blame, became the new normal. The persistence of the slowdown was not immediately apparent. Born and bred during the golden age of economic growth, scholars and policy makers initially believed it was a temporary blip, soon to fix itself. By the time it became clear that slow growth was not just an aberration, the latest hope was that a new industrial revolution, spurred by computing power, was right around the corner. Computing power was increasing at a faster and faster speed, and computers were being introduced everywhere, much as electricity and the combustion engine once were. This would surely translate into a new era of productivity growth that would pull the economy with it. And indeed it finally happened. Starting in 1995, we saw a few years of high TFP growth (though still significantly less than in the go-go years). It faded quickly, however. Since 2004, TFP growth and GDP growth both in the United States and in Europe seem to be back to the bad days of 1973–1994.11 In the United States, GDP growth did pick up in mid-2018, but TFP growth remains slow. Over the year, TFP grew only at an average of 0.94 percent,12 compared to the 1.89 percent achieved during the 1920–1970 period.

This new slowdown has provoked a lively debate among economists. It seems difficult to reconcile it with everything we hear around us. Silicon Valley keeps telling us we live in a world of constant innovation and disruption: personal computers, smartphones, machine learning. Innovation seems to be everywhere. But how could there be all this innovation without any sign of economic growth?

The debate has revolved around two questions. First, will sustained fast productivity growth eventually return? Second, is the measurement of GDP, at best a bit of an exercise in guesswork, somehow missing all the joy and happiness the new economy is bringing us?

Two economic historians at Chicago’s Northwestern University are at the center of this discussion.

Robert Gordon takes the view that the era of high growth is unlikely to come back. We have only met Gordon once. He gives the appearance of being quite reserved; his book, however, is anything but. On the other side is Joel Mokyr, whom we know much better, an enormously vivacious man, with twinkling eyes and a kind word for everybody; he writes with infectious energy consistent with his generally positive outlook on the future.

Gordon has gone out on a limb and predicted economic growth will average a meager 0.8 percent per year over the next twenty-five years.13 “Everywhere I look,” he said during a debate with Mokyr, “I see things standing still. I see offices running desktop computers and software much as they did ten or fifteen years ago. I see retail stores where we are checking out with bar code scanners the same way we did before; shelves are still stocked by humans, not by robots; we still have people slicing meat and cheese behind the counter.’’ Today’s inventions, in his view, are simply not as radical as electricity and the internal combustion engine were. Gordon’s book is particularly daring. He gleefully takes on the set of future innovations futurologists predict and one by one explains why, in his opinion, none of them would be as transformational as the elevator or air conditioning, and why none would take us back to an era of fast growth. Robots cannot fold laundry. Three dimensional (3D) printing won’t affect large-scale manufacturing. Artificial intelligence and machine learning are “nothing new.”14 They have been around at least since 2004 and have done nothing for growth. And so on.

It is clear of course that nothing Gordon says precludes the possibility that something entirely unexpected, perhaps some hitherto unimagined combination of familiar ingredients, will prove to be transformative. It is just his hunch that it won’t.

Mokyr, on the other hand, sees a bright future for economic growth, spurred by nations competing to be the leader in science and technology, and the resulting rapid spread of innovation worldwide. He sees the potential for progress in laser technology, medical science, genetic engineering, and 3D printing. To Gordon’s claim that nothing much changed in fundamental ways in how we produced in the last few decades, he counters: “The tools we have today make anything that we had even in 1950 look like clumsy toys by comparison.”15 But mostly, Mokyr thinks that the way the world economy has changed and globalized produces the right environment for innovations to bloom and change the world, in ways we cannot even begin to envision. He predicts one factor that will accelerate growth: we will be able to slow down the aging of the brain. Which of course would give us more time to have better ideas. Mokyr, engaging and creative as ever at seventy-two, is a good example for his thesis.

The fact that two brilliant minds come to such radically different conclusions about growth highlights what a vexing topic it has been. Of all the things economists have tried (and mostly failed) to predict, growth is one area where we have been particularly pathetic. To name just one example, in 1938, just as the US economy was going back into high-growth mode after the Great Depression, Alvin Hansen (who was not a nobody; he was the co-inventor of the IS-LM model most students of economics will remember from their first macroeconomics class, and a professor at Harvard) coined the term secular stagnation to describe the state of the economy at the time. His view was that the American economy would never grow again because all the ingredients of growth had already played out. Technological progress and population growth in particular were over, he thought.16

Most of us today who grew up in the West grew up with fast growth or with parents used to fast growth. Robert Gordon reminds us of our longer history. It is the 150 years between 1820 and 1970 that were exceptional, not the period of lower growth that followed. Sustained growth was virtually unknown until the 1820s in the West. Over the period 1500 to 1820, annual GDP per capita in the West went from $780 to $1,240 (in constant dollars), a paltry annual growth rate of 0.14 percent. Between 1820 and 1900, growth was 1.24 percent, nine times more than in the previous three hundred years, but still much less than the 2 percent it would hit after 1900.17 If Gordon is right and we end up with a 0.8 percent growth rate, we would simply be returning to the average growth rate over the very long run (1700–2012).18 This is not the new normal; it is just normal.

Of course, the fact that sustained growth over a long time, the kind we saw over most of the twentieth century, was unprecedented, does not mean it could not happen again. The world is richer and better educated than ever before, the incentives for innovation are at an all-time high, and the list of countries that could lead a new innovation boom is expanding. It could well be the case, as some technology enthusiasts believe, that growth explodes again in the next few years, fueled by a fourth industrial revolution, perhaps powered by intelligent machines capable of teaching themselves to write better legal briefs and make better jokes than humans. But it could also be, as Gordon believes, that electricity and the combustion engine brought about a onetime shift in how much we can produce and consume. It took us some time to reach this new plateau and there was fast growth along the way, but we have no particular reason to expect this episode will repeat itself. Nor, we might add, do we have definitive proof it won’t. Mostly, what is clear is that we don’t know and have no way to find out other than by waiting.

Abhijit’s parents did not really believe in toys. He spent long afternoons playing war games with flowers. The buds of the ixora, with their long stems and pointy heads, were the enemy, purportedly throwing stones at his foot soldiers, the long and fleshy leaves of the portulaca. The tuberoses were the health workers, operating on the casualties of war with toothpicks and bandaging them with soft petals of jasmine.

Abhijit remembers these as some of the most pleasurable hours of his day. That should surely count as well-being. But none of his enjoyment was captured by the conventional definition of GDP. Economists have always known this, but it deserves emphasis. When a rickshaw puller in Abhijit’s native Kolkata takes the afternoon off to spend time with his lady love, GDP goes down, but how could welfare not be higher? When a tree gets cut down in Nairobi, GDP counts the labor used and the wood produced, but does not deduct the shade and the beauty that are lost. GDP values only those things priced and marketed.

This matters because growth is always measured in terms of GDP. The year 2004, when TFP growth, after jump-starting in 1995, slowed down again, is when Facebook began to occupy the outsized role it currently plays in our lives. Twitter would join in 2006 and Instagram in 2010. What is common to all these platforms is the fact that they are nominally free, cheap to run, and wildly popular. When, as is now done in GDP calculations, we judge the value of watching videos or updating online profiles by the price people pay, which is often zero, or even by what it costs to set up and operate Facebook, we might grossly underestimate its contribution to well-being. Of course, if you are convinced that waiting anxiously for someone to like your latest post is no fun at all, but you are unable to kick the Facebook habit because all your friends are on it, GDP could also be overestimating well-being.

Either way, the cost of running Facebook, which is how it is counted in GDP, has very little to do with the well-being (or ill-being) it generates. That the recent slowdown in measured productivity growth coincides with the explosion of social media poses a problem, because it is entirely conceivable that the gap between what gets counted as GDP and what should be counted in well-being widened exactly at this time. Could it be there was real productivity growth, in the sense that true well-being increased, but our GDP statistics are missing this entire story?

Robert Gordon is entirely dismissive of this possibility. In fact, he reckons Facebook is probably responsible for part of the productivity slowdown—too many people are wasting time updating their status at work. This seems largely beside the point, however. If people are actually much happier now than they were before, who are we to pass judgment on whether it is a worthwhile use of their time and therefore whether it should be included in well-being calculations?19

Can the missing value of social media compensate for the apparent productivity growth slowdown in rich countries? The difficulty of course is that we have no idea how much value to assign to these free products. But we can try to estimate what people would be willing to pay. There are attempts to do this by looking at, for example, how much time people spend browsing on the internet as a proxy for how much they value it. The idea is that people could be working and earning money instead. If we follow this approach, the average annual value of the internet for an American went from $3000 in 2004 to $3,900 in 2015.20 If we were to add this missing bit to the 2015 GDP, one could explain about one-third of the $3 trillion of “lost output” in that year (compared to what the GDP would have been if the post-2004 slowdown had not happened).21

One problem with this way of getting at the consequences of the internet is that it assumes people have the option of working longer hours for more money instead of spending time on the internet. But this is not true for most people with nine-to-five jobs; instead they need to find ways to keep themselves amused (or at least out of trouble) for another eight hours or so every day. If they spend time on the internet, all this means is they like it more than reading a book or hanging out with friends or family. If they are not particularly sociable and don’t like books, this is hardly a ringing endorsement; it may be worth much less than $3,900.

However, there is also the opposite problem. Take someone who cannot imagine life without the internet, who needs an hour of Twitter fix every morning. That first hour brings almost infinite joy. But by the end of that hour all the enemies have been nailed, and every clever twist of phrase has been processed and passed on. What is left for the second hour is much more ho-hum, so much so that there is never a third hour. Compare that person with someone who also spends two hours desultorily responding to Facebook posts by or about friends half-forgotten and “friends” they would like to forget. In the data both will show up at the same place, valuing the internet at the price of two hours of time. But obviously they are different, and treating them the same may lead us to vastly underestimate the value of the internet.

Faced with the possibility that we could be either massively overvaluing the internet or the other way around, scholars looked for other ways to measure its value to consumers. In particular, there were several randomized control trials of what happened when the experimenter (with the permission of the participant) blocked access to Facebook (or social media more generally) for a random group of individuals for some relatively short period of time. The biggest of these experiments, which involved more than two thousand participants paid to deactivate Facebook for a month, found that those who stopped using Facebook were happier across a range of self-reported measures of happiness and well-being and, interestingly, no more bored (perhaps less). They seemed to have found other ways to keep themselves amused, including spending more time with friends and family.22

When Facebook access was restored after the experiment, those who spent a month without it were slow to return to their Facebook habit, and after several weeks were spending 23 percent less time on the app than they had before the experiment. Consistent with this, the estimate of how much they would need to be paid to give up Facebook for a second month was substantially lower at the end of the first month (after experiencing life without Facebook) than before.

All of this seems very consistent with the view that Facebook is addictive in the sense that it is hard to imagine life without it, but when you do give it up, things are not obviously worse. However, it is interesting that after the month of abstinence, the experimental subjects still wanted to be paid to give up Facebook; they did not simply feel grateful to be rid of it. The researchers assumed this was because they actually missed it, if less than they had expected, and therefore concluded Facebook generates over $2,000 of well-being per user.

How does this square with the fact that getting cut off made people happier on average? In part of course, like all averages, it hides the fact that some people really enjoy Facebook. Moreover, it is likely that what was costly for the participants was in part being the only one among their friends who was now off Facebook, and this inconvenience probably got worse the longer one was absent (it is okay to take a sabbatical from your social connections, but checking out totally is costly). If Facebook did not exist, the problem would not be there.

Where does that leave us? Not quite at a resolution. What we can say with some confidence is that Facebook is not the obvious win for all mankind as its devotees would have it, though people still value it more than they pay for it, at least in the current configuration where all their friends are on Facebook, Instagram, and/or Twitter. Could it be that if we valued these new technologies at their “real value,” growth would appear to be much faster? Probably not, based on the evidence at hand.

What we can say with some confidence is that there is nothing in the available evidence promising a return to the kind of fast growth in measured GDP that characterized the Trente Glorieuses in Europe and the golden years in the United States.

This should not come as a complete surprise. Remarkably, at the height of postwar growth, in 1956 Robert Solow wrote a paper suggesting growth would eventually slow down.23 His basic point was that as per capita GDP goes up, people save more, and therefore there is more money to invest, and more capital available per worker. This makes capital less productive; if there are now two machines in a factory where there was only one, the same workers will have to operate both at the same time. Of course, a single factory can hire more workers if it gets more machines. But the whole economy cannot (assuming migration remains unchanged), once its reserve of underused workers is exhausted. Therefore, the extra machines bought with the additional savings will have to be worked with fewer workers. Each new machine and as a consequence each additional unit of capital will contribute less and less to GDP. Growth will slow down. Furthermore, the lower productivity of capital lowers its financial return, which in turn discourages savings. So eventually people will stop saving and growth will slow down.

This logic operates in both directions. Capital-scarce economies grow faster because new investment is highly productive. Rich economies, which are, in general, capital abundant, tend to grow more slowly because new investment is not as productive. One implication of this is that any large imbalance between labor and capital should get corrected. Economies overabundant in labor grow faster, and since incomes grow faster, savings do as well. So these economies accumulate capital faster and become more capital abundant. By the reverse argument, economies with too much capital relative to labor accumulate capital more slowly.

As a result, a sharp divergence between the rates of growth of capital and the labor force is not sustainable over the long haul because if, say, capital grows faster than the labor force, then the economy will have too much capital relative to labor, which will slow down growth. There can be imbalances in the short run (as we are witnessing today in the United States where the share of the GDP paid to the labor force is falling24), but in the long run there is a natural tendency for economies to stay close to a balanced growth path, where labor and capital grow at roughly the same rate, and so does human capital—the part of capital embodied in the skills of the workers, for very much the same reason. Solow argued that GDP (which is after all the product of labor, skills, and capital) would also grow at the same rate as well.

Now, the growth of the effective labor force is determined by past fertility and how much people want to work, both factors that seemed to Solow to be more driven by demography than economics, and therefore more related to a country’s history and culture than to the current state of its economy or economic policy. However, there is also the improvement of TFP—if one worker becomes so productive that he can do the work of two, because of improvements in technology, then the effective labor force would have doubled. Solow assumed such transformations were also unrelated to contemporary economics and policies of the country, in effect placing the growth rate of the effective labor force outside the realm of economics. This is why he called it the “natural rate of growth,” and from his theory, we know that GDP must also grow at the same rate as the effective labor force in the long run; that is, at the natural rate.

A number of implications follow from Solow’s theory. First, growth is likely to slow down after a phase of fast growth that follows a dramatic transformation, once the economy is back on the balanced growth path. This is clearly consistent with what happened to Europe after 1973. After the wartime destructions, capital was scarce and Europe had a lot of catching up to do; by 1973 the era of catch-up growth was over. In the United States, the kind of investment-driven growth Solow had in mind clearly slowed down after the war, but conveniently its place was taken by rapid TFP growth until 1973. Since then, as we already discussed, there has been a slowing trend even in the United States. Interest rates have been falling throughout the West, reflecting, it seems, an abundance of capital, exactly as in the Solow model.

The second implication of Solow’s theory, and perhaps the most striking, is what economists call convergence. Countries scarce in capital and relatively abundant in labor, like most poor countries, will grow faster because they have not yet reached their balanced growth path. They can still grow by improving the balance between their labor and capital. As a result, we would expect the difference in GDP per worker across countries to be reduced over time. All else being the same, poorer countries will catch up with their richer counterparts.

Solow himself was careful to stop well short of promising this. If a country has a lot of labor and very little capital, which is how many poor countries start out, then only a fraction of the labor force will be employable at a wage sufficient to ensure their subsistence (there may be nothing for the others to do), and as a result the country will not benefit much from its labor abundance. Convergence, if it happens at all, may be very slow.

Notwithstanding Solow’s warnings, this vision of an orderly transition from dire poverty to relative wealth as the countries catch up and then go on to the nirvana of balanced growth, combined with the promise of global convergence in living standards, provided such a comforting narrative for progress under capitalism that it took some thirty years before economists started noticing the model did not fit reality all that well.

To start with, it is not true that poor countries as a rule grow faster than richer ones. The correlation between GDP per capita in 1960 and subsequent growth is very close to zero.25 How does this square with the fact that after the war Western Europe caught up with the United States? Solow had a possible answer. What his model actually says is that countries that are otherwise identical will head toward each other. This could be why Western Europe and the United States, which are very similar in many ways, converged toward each other. On the other hand, in Solow’s world countries that are naturally thriftier than others and invest more of their output will be richer in the long run. Moreover, for a while, before settling down to grow at the natural rate, initially poor countries that invest more will also grow faster as they converge toward this higher level of GDP per capita.

Could the lack of investment be the one reason the developing world differs from Western Europe and the United States? As we will see, the answer seems to be no.

The third and most radical prediction from Solow’s model is that the growth rate of GDP per head among the relatively rich countries, once the economy reaches balanced growth, may not be very different. Essentially, in Solow’s world these differences must come from differences in TFP growth, and Solow believed that, at least for these rich countries, TFP growth should be more or less the same.

In Solow’s view, as mentioned above, TFP growth just happens—policymakers don’t have very much control over it. This was something many economists were not entirely happy about. Given that growth rates are the language in which the league tables of international competition are written, there was something rather off-putting about Solow’s refusal to offer some assurance that TFP would be higher for countries that pursue “good” economic policies. Was he just being deliberately quixotic? After all, don’t we see many more of the latest technologies being deployed in the richer countries?

This resistance to the idea that a country’s balanced growth rate is not easily influenced by policy is perhaps to be expected. But it misses the subtlety of Solow’s thinking, in multiple ways. First, Solow is asking what drives technological upgrading in countries already at the cutting edge. Presumably the flow of new ideas is a big part of growth for these countries, and it is not clear why ideas should stop at the border. A new product invented in Germany could be simultaneously developed for production in several other countries, possibly by local subsidiaries of the mother company. Productivity would then go up more or less equally in all these countries, even though the invention came from only one of them.

Second, he is talking about growth after countries get to their balanced growth path, and while this might have already happened for some of the richer countries, it is probably a long way away for the ones where capital is still scarce. By the time Kenya or India gets to Solow’s balanced growth path, they necessarily would be much richer and be using many or all of the latest technologies. Their current technological backwardness could just be a symptom of their lack of capital.

Finally, and this might be the hardest piece to wrap one’s head around, countries on the way to the balanced growth path could actually be upgrading their technologies faster than those already there. Of course, the most showy breakthroughs, the self-driving cars and 3D printers of the day, will always be in the more advanced countries, but most technology upgrading is just moving from day-before-yesterday’s technology to yesterday’s. This is typically easier than pushing the frontier, precisely because it has already been done and we know exactly how to do it. It is a matter of pulling things off the shelf rather than coming up with something new.

For all these good reasons, Solow deliberately opted to punt on what drives differences between the balanced growth rates of different countries. He simply assumed the rate of improvement in TFP was a product of mysterious forces that had nothing to do with the countries, their culture, the nature of the policy regime, and so on. This meant he had very little to say about what we can do about long-run growth once the process of accumulation of capital has run its writ and the return on capital is low enough. Solow’s was what economists call an exogenous growth model, where the word “exogenous,” meaning driven by outside effects or forces, acknowledges our inability to do anything about the long-run growth rate. Growth, in short, is beyond our control.

It was a combination of the evidence that many poor countries were not growing and the Solow model’s inability to say something useful about how to affect long-term growth that eventually made economists look elsewhere. They desperately wanted to be able to say something about what could help countries grow. As Robert Lucas, one of the doyens of the Chicago school of anti-Keynesian macroeconomics and one of the most influential economists of our times, confessed in his much quoted Marshall lecture in 1985, he would like to know “if there is some action a government of India could take that would lead the Indian economy to grow like Indonesia’s or Egypt’s? If so, what exactly? If not, what is it about the ‘nature of India’ that makes it so? The consequences for human welfare involved in questions like these are simply staggering: once one starts to think about them, it is hard to think about anything else.”27

But Lucas had more than just an aspiration to offer. He was also arguing that we are missing something important, and that the reason why India was poor could not all be because of a shortage of skills and capital. He recognized that India had less capital and skills than the United States, maybe because of its colonial history or the caste system. But to explain the enormous difference in GDP per capita between two countries based solely on lack of resources, those resources would have to be extraordinarily scarce. And if they were so scarce they should be very valuable. For example, the one tractor available would be used very intensively on hundreds of fields prepared by thousands of workers; the rental rate on this tractor would be extremely high. Based on this logic, Lucas computed that if the difference in GDP between the United States and India was to be explained by the scarcity of capital in India and nothing else, capital would have to be so scarce that its price (what is paid to the owner of the resources that finance the machines in the economy) would have to be fifty-eight times higher in India than it was in the United States.28 But in that case why wouldn’t all the capital in the United States move to India, he wondered. Since it evidently did not, he concluded the price could not in fact be that high. In other words, the intrinsic productivity of capital must be less in India than in the United States to explain why, despite its obvious scarcity, capital in India does not earn the kinds of astronomical returns that Lucas’s computation would predict—or to put it in Solow’s terms, TFP must be much lower in India.

Lucas was, perhaps unsurprisingly, being too optimistic about the functioning of markets. We now know that we live in a sticky economy where nothing moves very fast, and certainly not from the United States all the way to India. Nonetheless, some version of his basic insight has been rediscovered by many others who keep hitting up against the TFP puzzle. For one, if you simply try to explain the cross-country variation in GDP by the amount of resources in different countries, you will quickly realize that even though poor countries are indeed desperately short of skills and capital, their GDP per capita is even lower than this lack of resources would predict.29 In other words, poor countries are poor in substantial part because they make less good use of the resources they have, and even within poor countries some do better than others with the same resources. The question is why?

Paul Romer, a PhD student of Lucas’s, was one of the people inspired to respond to Lucas’s passionate plea that we have to find a better way to explain growth. What made it a challenge was that Solow’s answer rested on perhaps the two most basic ideas in economics. First, that capitalists invest in the pursuit of high returns; when and where returns go down, capital accumulation tends to go down as well. Second, that as capitalists as a class accumulate more and more capital, the productivity of capital becomes lower because there are not enough workers to work with it. In economics this is known as diminishing returns. It has a long pedigree. French economist Anne Robert Jacques Turgot, who was briefly France’s finance minister and one of the many experts who tried unsuccessfully to head off France’s headlong descent into the economic chaos that eventually precipitated the French Revolution, wrote about it in 1767.30 Karl Marx took it as a premise. As he saw it, this was why capitalism was doomed: the insatiable greed of the capitalist class in the pursuit of more and more capital will drive the return on capital into the ground (in Marxist parlance this is called the “falling rate of profit”) and precipitate the crises that eventually end capitalism.31

The assumption of diminishing returns makes a certain amount of intuitive sense. What is the point of acquiring new machines if there are no workers to operate them (or new engineers to program them, or salesmen to sell the products)? Of course, there are also counterexamples. Amazon clearly derives a lot of its ability to cut costs from the volume of its sales. Setting up the kind of storage and delivery systems it is famous for would not make sense if there were not a constant flow of demand for everything it sells, and to finance that it needs lots of capital. Amazon at a hundredth of its size could not possibly make money. In fact, Amazon made little or no money until it grew very large, and then profits soared. In July 2018, Amazon’s profit reached 2.5 billion dollars.32

Economists of Solow’s generation were aware of the possibility of increasing returns, which is how economists describe the idea that bigger is better (and the source of Amazon’s present dominance). But one obvious implication of increasing returns is that the biggest firms should be the most profitable, and therefore the best situated to undercut the others and push them out of the market. Such markets are doomed to end up with monopolies. This is indeed what is happening with the online retail sector. But while we do see some industries where there are also a small number of dominant players (social networks and hardware stores are both in this category), most important markets—cars, clothes, and chocolate, for example—have many firms. It is for this reason that economists have tended to shy away from theories that rely too heavily on increasing returns.

Romer wanted to stick with the idea that a single firm was still subject to the law of diminishing returns. His insight was that all we need to undo the Solow effect is to be able to assume that as a whole an economy with more capital also has a more productive capital stock. This could be true even if every firm faced diminishing returns and there was therefore no tendency for firms to become monopolistic behemoths. To explain how this might happen, Romer invited us to think of the production of new ideas in a place like Silicon Valley, though his paper was written years before Silicon Valley achieved its iconic status.33 Firms in Silicon Valley are very similar to the firms in Solow’s world except in one important way: they use less of what we usually think of as capital (machines, buildings) and more of what economists call human capital, essentially specialized skills of different kinds. Many Silicon Valley companies invest in clever people in the hope they will come up with some brilliant and marketable idea, and sometimes this indeed happens.

The usual forces of diminishing returns are present in these companies as well. Too many temperamental geniuses and not enough drudges to manage the cash and make sure the gaming during work hours remains in check, and you have a disaster on your hands. What is different, Romer argues, is the overall environment of the Valley. Ideas can be heard and overheard everywhere, in the coffee shops and wheatgrass bars, in parties and public transport. One stray thought expressed by someone you will never meet again might prompt another, and all of it cumulates into a set of ideas that have remade the world. What matters is not just how many smart people you work with, but also how many smart people you are competing with, or just happen to be around in the Valley as a whole. Silicon Valley, in Romer’s theory, is what it is because it brings together the best minds of the world in an environment where they can cross-pollinate each other. The increasing returns here are at the level of the industry, the city, or even the area. Even if every firm faces diminishing returns, doubling the number of high-skilled people in the Valley makes all of them more productive.

Romer argues that the same goes for all successful industrial cities: Manchester in the middle of the eighteenth century, New York and London during various periods of financial innovation, Shenzhen or the Bay Area today. In all of these places, he would claim, the force of diminishing returns that comes from the scarcity of land and labor (labor becomes scarce in part because land is scarce and therefore living in these places is so expensive) was defeated by the exuberant energy that comes out of learning from each other and coming up with new ideas. As a result, high growth can keep going forever as more and more high-skilled people come together, even without help from Solow’s mysterious exogenous productivity growth.

Getting rid of diminishing returns at the level of an entire national economy also helps us explain why capital does not flow to India. In Romer’s world, capital earns roughly the same return in India and in the United States, even though there is much less capital in India, because the standard law of diminishing returns helping India in Solow’s model is compensated for by the faster flow of ideas in richer economies. The question is whether this is just a clever intellectual maneuver, a comforting story we tell ourselves, or whether the force Romer emphasizes looms large in the world.

Before we get to that, it is worth pointing out something the careful reader might have already noted: as soon as we started talking about the theory of economic growth, the conversation just got a whole lot more abstract. Both Solow and Romer are telling stories about what happens to entire economies over long periods of time. To do so, they are telescoping an incredible amount of real-world complexity into as few building blocks as possible. Solow, for example, gives a central role to the idea of economy-wide diminishing returns. Romer, for his part, puts his money on the flows of ideas between firms, but we never get to see the ideas themselves, just their supposed benefits at the level of the entire economy. Given the sheer diversity of occupations, enterprises, and skills that constitute an economy, it is very hard to get a feel (let alone an empirical counterpart) for any of these very broad concepts. Solow wants us to think of what happens in an economy when the total capital available to it goes up. But economies typically don’t accumulate capital; individuals do. Then they decide what to do with that capital: whether to lend it out, start a new bakery, buy a new house, and so on. Each such decision changes many things; house prices may go up, bread prices may come down, good pastry chefs may become harder to come by. Solow wants to reduce all that complexity to one change: the change in the availability of labor relative to capital. Likewise, when a city gets an influx of tech people, many things change—you get better espresso, for one, and many low-income residents get pushed out—but Romer highlights just one key thing: the exchange of ideas. Both Romer and Solow may well be right in their guesses about what really matters, but it is difficult to map their abstractions into the real world.

To make matters worse, the data, which has been our main recourse so far, cannot help us very much here. Because the theories operate at the level of entire economies, our tests will need to compare different economies (countries or, at best, cities) rather than individual firms or people. As we discussed in the chapter on trade, this is always a challenge since economies tend to be different from each other in any number of ways, making them hard to compare.

Moreover, even if we were willing to draw conclusions from the comparison of entire economies to each other, it is not clear what we would learn. Take the idea of diminishing returns at the level of the economy. We want to test whether capital is less productive in a country that ends up with some extra capital. The problem once again is that countries don’t accumulate capital, individuals do. Those individuals may then invest that capital in firms. Those firms buy machines and buildings and so on, and then try to hire workers to make use of their newly installed capital. This increases competition in the labor market, forcing the firms to settle for fewer workers than they would want, which is what depresses productivity of capital. Now suppose we do observe that an inflow of capital made capital less productive. How can we be sure that the reason this happened is the one Solow has in mind? After all, it could be that the capital was invested in the wrong place and that is what made it unproductive. Or that it was never invested at all. Perhaps if it were invested properly, the return on capital would actually go up (and not down as Solow would have it).

Finally, a lot of the claims in growth economics are about what happens in the long run. In the long run, growth slows down in Solow’s world; it does not in Romer’s. But how long is long enough? Is it enough to observe a slowdown? Or could that just be a temporary blip, a piece of bad luck to be reversed soon enough?

So at the end of the day, although we will try to stitch together the best evidence for these theories, the result will be tentative. We have already seen that growth is hard to measure. It is even harder to know what drives it, and therefore to make policy to make it happen. Given that, we will argue, it may be time to abandon our profession’s obsession with growth. The most important question we can usefully answer in rich countries is not how to make them grow even richer, but how to improve the quality of life of their average citizen. It is in the developing world, where growth is sometimes held back by an egregious abuse of economic logic, that we may have something useful to say, though, as we will see, even that is very limited.

The key ingredient of Romer’s happy narrative was the spillovers: the idea that skills build on each other and that putting skilled people together in one place makes a difference. Clearly, this is something people in Silicon Valley believe. There are many parts of California prettier than Silicon Valley, and most are cheaper. Why do companies still want to locate there? States and cities in the United States and elsewhere offer large subsidies to attract firms. In September 2017, Wisconsin gave at least $3 billion in fiscal advantages to Foxconn to have it invest $10 billion in an LCD manufacturing plant.34 This is $200,000 for every job they promised to create. Similarly, Panasonic received more than $100 million to move its North American headquarters to Newark, New Jersey ($125,000 per job), and Electrolux was given $180 million in tax abatements to start a new plant in Memphis, Tennessee ($150,000 per job).35 The most recent example of this competition was the very visible scramble to attract Amazon’s second headquarters, HQ2. Amazon received 238 proposals from different locations before choosing Arlington, Virginia, and New York City.36 These 237 or 238 cities (depending on whether New York finally withdraws or not) clearly believe in spillovers.

Apparently, Amazon does too. In choosing the location for HQ2, Amazon listed a preference for (among other things) “metropolitan areas with more than one million people” or “urban or suburban locations with the potential to attract and retain strong technical talent.”37

Amazon’s theory seems to be that being in a “thick” market, a market where there are lots of sellers, in this case of skilled labor, is valuable, presumably because it is easier to find, retain, and replace workers.

Romer’s theory, you may recall, was more about informal conversations that occur when many people working on related topics are together. There is some evidence for such spillovers. We know, for example, that inventors are more likely to cite patents from other inventors in the same city, suggesting they were more likely to be aware of them.38

A variant of Romer’s hypothesis that is less specific to Silicon Valley and its imitators is that the presence of more educated people makes everyone else more productive. It turns out, however, that the evidence that we are all becoming more productive as a result of having more educated people around us is not overwhelming. We do observe that everyone earns more in cities where there are more educated people, but this could be for a variety of reasons. Cities with more educated people may also attract more high-paying firms (high-tech firms, more profitable firms, firms that care more about the quality of work, etc.), drawn in by the prospect of being able to find the right kind of workers. The problem is finding instances where the level of education in the population at large goes up significantly without other things (policies, investments, etc.) changing at the same time.

There is clear evidence, however, that cities as a whole can benefit from a large investment. Michael Greenstone, Rick Hornbeck, and Enrico Moretti (who is the author of The New Geography of Jobs,39 which argues that spillovers are the reason why cities are growing and rural areas are not) ask whether cities as a whole benefit from attracting a high-profile plant, much like Amazon’s HQ2.40 To answer this question, their study compared the winners of bidding wars to attract companies to the first runners-up. They find that TFP of the plants already present in the winning county surged, consistent with there being large spillovers—TFP five years after the plants were set up was on average 12 percent higher in places that received the plant than the ones that just missed out, translating into $430 million per year more in earnings for the county. Both wages and employment went up. In many cases, we do not know how much the average state or city spent to attract the plant, but we have some examples. For instance, in the case of the BMW plant that eventually went to Greenville-Spartanburg, South Carolina, over Omaha, Nebraska, the subsidy on offer was $115 million. If they got the average 12 percent benefit, the investment clearly paid off handsomely. This was the argument made in New York City in support of the subsidies to Amazon: that as an investment they were well worth it.41

An alternative way to attract businesses to a particular location is to build infrastructure. This is what the Tennessee Valley Authority (TVA) did for Tennessee and its neighboring states over the period 1930–1960, using public funds to build roads, dams, hydroelectric plants, etc. The idea was that infrastructure would attract firms, firms would attract other firms, and so on. Jane Jacobs, one of the most influential American urbanists of the twentieth century, was skeptical. She wrote a piece about it in 1984, called, quite simply, “Why TVA Failed.”42

But it did not fail. Enrico Moretti and a colleague compared the TVA region with six other areas initially supposed to receive the same type of investment but where, for various political reasons, nothing happened. They found that between 1930 and 1960, the TVA counties generated gains both in agricultural and manufacturing employment relative to this comparison group. It is true that once outside funding for the program stopped in 1960, the gains in agriculture vanished, but the gains in manufacturing persisted and actually continued to intensify all the way until 2000, consistent with a widely held view that spillovers are more important in manufacturing than in agriculture. The effects are substantial; the authors estimate that over the long run the income gains as a result of TVA in the region will be $6.5 billion more than what it cost to set it up.43

Does this mean countries can create the conditions for permanently faster economic growth by promoting regional development, perhaps in multiple regions at the same time? There are two reasons why this does not follow. First, it is not enough that the firms gain from the initial investment. They have to gain enough to overcome the usual forces that slow down growth: shortages of land, labor, and skills. Moretti estimates that a 10 percent change in employment today will increase employment in the future by 2 percent, which is not big enough to generate sustained growth over the long term; pretty rapidly the original boost will peter out.44

Second, growth in one region is different from national growth because it can happen in part by cannibalizing growth in the rest of the economy, drawing capital, skills, and labor away from other areas. The cities where Amazon eventually locates will grow, but partly that will be at a cost to other American cities. Moretti estimates the two effects might actually net out, with the result that national growth will be more or less unaffected.45

Moretti concludes from his reading of this entire literature that regional development is unlikely to be the lever that will help us avoid the end of growth.46 It is possible his assessment is slightly too pessimistic, but the note of warning is certainly valid. While it may make sense for an individual city to try to lure jobs away from another, this is unlikely to be a large win for a country as a whole, unless it is a very small country (the city state of Singapore, for example) that can grow at the expense of others.

It is worth emphasizing, however, that this evidence mainly comes from the United States or Europe. It could be that the developing world is quite different in this respect. Certainly, high-quality urban infrastructure is much more concentrated in a few cities in most of these countries, and a case could be made both for building more “high quality” cities and for making the few existing big cities more livable in order to promote economic growth. This is a key policy focus of the World Bank. For example, a 2016 report on urbanization in India47 highlights “messy” and “hidden” urbanization, dominated by slums and sprawl. In essence, cities grow horizontally, by outgrowing their formal boundaries, rather than vertically through taller and better-quality buildings. In total, 130 million people in South Asia (more than the population of Mexico) live in informal urban settlements. Distances are long, traffic is impossible, and the pollution levels are extraordinary. This makes it more difficult to attract talent to cities, and also limits the effectiveness of cities as places of production and exchange. Better cities could potentially generate entirely new growth opportunities for the countries, without taking any growth away from elsewhere.

Romer’s own focus for several years (even before his short and rocky tenure as the World Bank’s Chief Economist) was on the cities of the third world. It continues to be a priority of his. He wants these countries to build cities where creative people would want to come together and new ideas would be born out of the cross-pollination. Cities that would be business friendly but also genuinely livable—Shenzhen without the pollution and the traffic. Unusually for a successful academic, he believed and cared enough in his message to set up a nonprofit think tank to help in the creation of what he called “charter cities.” These would be giant protected enclaves (Romer wants hundreds of them around the world, each of them hosting eventually at least a million people) that live by Romerian rules within nations that do not. There would be a contract by which the national government agreed that a third-party government, from a developed country, would enforce those rules. So far, there has been just one taker, the government of Honduras, which had plans to set up as many as twenty zones for employment and economic development (ZEDEs). Unfortunately, though it claimed inspiration from Romer’s ideas, the Honduran vision seemed closer to the banana enclaves the United Fruit Company and its competitors ran in the first part of the last century, where the company’s writ was law. They deviated from the project from the get-go when they decided not to use the oversight of a third-party government. It eventually turned out that the Honduran government was more interested in Romer’s name and fame than his counsel, and when it signed a deal with an American entrepreneur with a strong taste for totally unregulated capitalism to develop the ZEDEs, Romer walked out. This story suggests charter cities are unlikely to hold the key to sustained growth in developing countries for the very good reason that the internal political compulsions the charter is intended to hold at bay often have a way of biting back.

To summarize the previous sections, regional spillovers seem real, but based on the limited evidence we have, probably not powerful enough for the task of keeping growth going at the national level. Perhaps anticipating this, Romer had a second story up his sleeve; in that story, growth is driven by firms developing new ideas, which turn into more productive technologies.48

Romer was describing a force that ensured technologies would constantly keep improving, and more so in countries pursuing pro-innovation policies. Unlike in Solow’s world, technological progress would no longer be some mysterious force we have no control over.

To build a model where there is ongoing innovation and unbridled growth, Romer needed a force to counterbalance what every scientist and engineer knows: the more things have already been invented in the past, the harder it is to find an original idea. To get there, Romer assumed that once produced, new ideas become freely available for others to build on. Knowledge spills over. The advantage of building on previous ideas is that the new inventor is standing on the shoulders of giants. The inventor just needs to tweak the previous invention, not invent something entirely novel. In this way, the growth process can continue unabated.

Romer is a true optimist, as is perhaps evident from his faith that he would be able to entirely ring-fence his charter city project from the notorious politics of Honduras. The same optimism inspires his vision of the innovation process. In his world, new ideas just waft in like the smell of roses on a summer breeze.

In the real world, it seems, the production of new ideas is a much more fraught affair. Many marketable ideas are produced by firms, and firms tend to be possessive of their discoveries. Pharmaceutical companies and software firms, for example, do many things, legal and sometimes not so legal, to acquire and retain control over new ideas. Industrial espionage is a major global industry today, and so is its foil, patent law. A classic paper by Philippe Aghion and Peter Howitt, published a couple of years after Romer’s, argued that innovation-led growth was possible even in that much more cutthroat environment.49 In their world, firms innovate less out of a desire for knowledge than to make sure they get there before the competition. Nevertheless, new ideas do continue to get produced, as long as patent protection does not entirely preclude building on past ideas.

This shift in perspectives is not without its consequences. In Romer’s world, innovation is a boon innovators offer the world. They do make some money, but what the economy gets in return is incomparably more valuable because future generations of innovators get to build on it, for free. As a result, Romer in particular wants us to bend over backward to make the world as friendly as possible for innovators—low taxes on profits and capital gains, incubators and innovation cells, patents that protect the innovators’ rights as long as possible, and so on.

Aghion and Howitt have a much less romantic view of innovators. Interestingly, Aghion is the rare economist who had a chance to observe the innovative process close at hand. His mother, who was from a French-speaking Jewish family, founded the well-known designer brand Chloé when she moved to France, after being forced to leave her home in Egypt in the early 1950s. The years when Chloé went from being a dressmaker to a global brand were exactly the years of Philippe’s growing up. Nevertheless, inspired by Joseph Schumpeter (the Harvard economist of the mid–twentieth century and braggart extraordinaire50), Aghion sees innovation as a process of creative destruction, in which each innovation involves both creation of the new and destruction of the old.51 In his world, sometimes the creative dominates, but at other times the destructive holds sway; novelties get created not because they are useful but because they defeat someone’s existing patent. Making it more rewarding to innovate might backfire as a result. Innovators may worry that the time interval between the moment they displace the previous incumbent patent holder and the less happy moment they lose their own patent to someone else could be frustratingly short. Patent protection is important to get people to innovate, but it is easy to get too much of it, permitting the incumbents to rest on their laurels. Instead, there needs to be a balance between greenfield innovation and the possibility of adopting other people’s ideas.

You’ll recall that one of the reasons why economists like Lucas were dissatisfied with the Solow model is that it did not provide any direction to an eager policy maker. Romer’s model does. Conveniently, the advice is not exactly revolutionary. In particular, for Romer the government needs to get out of the way of stifling incentives to work hard and invent the new technologies that will make everyone more productive. In other words, cut taxes.

Romer is a Democrat in the United States. Or at least that’s what the economics rumor mill tells us. His father was a Democrat who was the governor of Colorado. But the idea that low tax rates can affect long-term growth by encouraging innovation is one that US Republicans have come to dearly love. From Reagan to Trump, Republican politicians have consistently promised to cut taxes, and the perennial justification is that they promote growth. Low tax rates are necessary at the top, because the likes of Bill Gates need to be given the incentive to work hard, be creative, and invent the next Microsoft to make us all more productive.

It was not always like that. Top tax rates were above 77 percent for the period 1936–1964, and above 90 percent for about half of that period, mostly in the 1950s under a solidly right of center Republican administration. The top tax rate was brought down to 70 percent in 1965 by a more left-wing Democratic administration, and since then it has drifted down to mid 30 percent. Every Republican administration has tried to cut it down further and every Democratic administration has tried to raise it a little, though always with great trepidation. Interestingly, for the first time in over fifty years, the idea of a top marginal tax rate above 70 percent has gained some traction among Democrats in 2018.

Yet, looking at growth rates since the 1960s, it is evident the low tax rate era ushered in by Reagan did not deliver faster growth. There was a recession in the beginning of the Reagan administration, followed by a catch-up phase when the growth rate went back to normal. Growth rates were a little higher during the Clinton years and declined afterward. Overall, if we take the long-run view (the ten-year moving average, which averages the ups and downs of the business cycle), economic growth has been relatively stable since 1974, remaining between 3 and 4 percent over the entire period. There is no evidence the Reagan tax cuts, or the Clinton top marginal rate increase, or the Bush tax cuts, did anything to change the long-run growth rate.52

Of course, as the Republican Paul Ryan, former Speaker of the House of Representatives, pointed out, there is no evidence that they did not. Many other things were happening at the same time. Ryan painstakingly explained to a journalist why all of these things lined up to make tax increases look good and tax decreases look bad:

Paul Ryan is right about one thing. Just looking at the variations over time, it is hard to conclude whether there is any causal effect of tax rates on growth. It is indeed possible there is a true relationship, but it is obscured by the many other things that are happening. The same lack of correlation between growth rates and tax rates remains true, however, when we look at changes in taxes across countries. There is absolutely no relationship between the depth of the cut between the 1960s and 2000s in a country and the change in growth rate in that country during the same period.54

Within the United States, the experience of individual states is also telling. In 2012, Republican leaders in Kansas passed deep tax cuts, with the promise this would spur the economy. Nothing like that happened. Instead the state went broke and had to cut back on its education budget, the school week was cut to four days, and teachers went on strike.55

A recent study from the University of Chicago’s Booth School of Business (not a place known for its socialist tendencies) uses a clever trick to answer whether tax cuts that benefit the rich have more or less of a growth effect than tax cuts that benefit the rest of the economy. Different states have very different income distributions, and therefore tax cuts for the rich should have very different consequences in different states. Connecticut, for example, has many more rich people than Maine. Using the thirty-one tax reforms since the war, the study shows that tax cuts benefitting the top 10 percent produce no significant growth in employment and income, whereas tax cuts for the bottom 90 percent do.56

One can also directly look at the question of whether high-income earners slack off when taxes are higher. This question can be answered much more precisely than the effects on overall growth, because tax reforms affect different people differently, so it is possible to compare the changes in behavior for people who are more or less affected. The key conclusion from a very large literature, summarized by two of its most respected experts, Emmanuel Saez and Joel Slemrod, is that “there is no compelling evidence to date of real economic responses to tax rates at the top of the income distribution.”57

By now, there seems to be a consensus among a large majority of economists that low taxes on high earners are not guaranteed to, on their own, bring about economic growth. This was reflected in the response of the IGM Booth panel of top economists to the Trump tax cut of 2017. The tax cut provides deep and durable tax cuts for businesses, including a cut in the corporate tax rate from 35 percent to 21 percent. The bill also includes a new top tax rate of 37 percent for the wealthiest Americans (down from 39.6 percent), raises the threshold for top earners, and eliminates the estate tax. It has much smaller tax cuts for the rest of the population, and most of these are meant to be temporary. To the question “If the US enacts a tax bill similar to those currently moving through the House and Senate—and assuming no other changes in tax or spending policy—US GDP will be substantially higher a decade from now than under the status quo” only one person agreed with the statement and 52 percent either disagreed or strongly disagreed (the rest were uncertain or did not answer).58

Despite this consensus, a memo from the government’s treasury department on the fiscal impact of the bill assumed (without any stated justification) an increase in 0.7 percent in annual growth rates from reducing taxation.59 How could they get away with a statement that had nothing to do with what anybody seriously believes? One answer, of course, is that it was not the only instance where the administration asserted a non-truth to support its decision. But we suspect that part of the reason the public so easily bought into the idea that tax cuts for the wealthy lead to economic growth is that they have heard this particular message for so many years, from so many prominent economists of a previous era. In those days, evidence was scarce and it was normal to argue from “first principles” based on intuition and no data. The repetition of this mantra by generations of serious economists has given it the soothing familiarity of a lullaby. We still hear it every day from a gaggle of business experts, who even today feel unconstrained by the data. It is now part of the “common sense.” When we asked respondents in our survey the question similar to the one asked by the IGM booth panel, 42 percent of respondents agreed or strongly agreed with the proposition the tax cut would increase growth within five years (only one economist did). Twenty percent of our respondents disagreed or strongly disagreed.

It did not help that nine conservative academic economists, mostly with solid reputations but also part of this older generation, wrote a supporting letter to the administration arguing that growth would go up and “the gain in the long-run level of GDP would be just over 3 percent, or 0.3 percent per year for a decade.”60 It was immediately pointed out that this letter was based, once again, on first principles and a very selective reading of the empirical literature.61 But it was so much in line with what the public and the press expect from economists that it sounded perfectly legitimate.

Once again, this underscores the urgent need to set ideology aside and advocate for the things most economists agree on, based on the recent research. In a policy world that has mostly abandoned reason, if we do not intervene we risk becoming irrelevant, so let’s be clear. Tax cuts for the wealthy do not produce economic growth.

While the tax changes at least are happening in the public eye, there is another very major transformation in the US economy that could have a direct bearing on growth: the increasing concentration of economic activity. The driver of long-run growth, in the Solow and the Romer models, is technological innovation. It is because people constantly invest in new products or new better ways of doing things that TFP grows, and the economy grows with it. But, as Aghion and Howitt reminded us, innovation does not come out of nowhere; someone needs to have a financial incentive to invent something new.

Companies that innovate need access to markets to sell their products. And some evidence suggests this is becoming increasingly difficult for new entrants. At the national level, most sectors (including technology, but not only) are increasingly dominated by a few companies. A 2016 report by the Council of Economic Advisers, for instance, finds that the share of the top fifty corporations in the national revenue of each of their sectors increased across most sectors between 1997 and 2012.62 This concentration is largely accounted for by a growing share of the “superstars,” partly the result of a fairly liberal attitude on mergers in the United States.63 For example, the share of the top four companies in a sector’s revenues has increased in every sector. In manufacturing, the top four accounted for 38 percent of revenues in 1980 and 43 percent in 2012. In retail trade, the share more than doubled, moving from 14 percent to 30 percent.64

It is not entirely clear that this increased concentration has been bad for consumers. Depending on the data source and computation methods, some economists find huge increases in markups65 (the difference between what a firm charges and its costs) but others do not.One thing that has protected consumers is that in the retail sector there has been concentration at the national level but not at the local level. When Walmart or other superstores come to town, they displace some mom-and-pop operations. But this does not make the market less competitive for the final customers and superstores offer more varieties, often at cheaper prices.66 And Amazon has actually fostered intense competition among sellers on its platform.67

But the problem with the increased concentration at the national level is that to the extent it reflects a decline in the competition faced by these behemoths, it may actually lead to reduced innovation because it creates higher barriers for new entrants to disrupt an industry. In the logic of Aghion and Howitt, the promise of (temporary) monopoly power, through a patent, spurs innovation, and this innovation in turn results in the new technologies everyone will eventually be able to use. This is what causes growth. But if monopoly is guaranteed forever anyway, innovation and growth may slow down; a monopolist can sit on their hands and never invent anything new. Some evidence suggests something like this is happening now. In particular, a study found that when a large planned merger and acquisition in a sector narrowly fails to happen for some unpredictable reason (the judge was not lenient enough or the deal fell through), the sector remains more competitive for several years afterward. These sectors with “near misses” see the entry of more new firms, more investment, and more innovation. This result does suggest that the relatively low growth in TFP may in part be explained by the increase in concentration.68

Even if the increase in industry concentration is partly responsible for the slowdown of growth in the United States, it would be unreasonable to conclude that breaking up monopolies will single-handedly restore fast growth. After all, growth has also been sluggish in Europe, and European regulators have been much more aggressive against monopolies. This illustrates, once again, the only clear lesson of the last few decades. We don’t understand very well what can deliver permanently faster growth. It just happens (or not).

But if growth in rich countries is not about to explode, what will these countries (and, soon enough, middle-income countries like China or Chile) do with their increasingly abundant capital? The business community, which is sometimes smart enough not to buy into the ideological messaging it offers the rest of us, has been for some years focused on another way out for the abundant capital in its hands. We noticed this about twenty years ago, when, all of a sudden, businesspeople, perhaps sensing they could not count on reliable economic growth in the West, started to quiz us about the countries we knew best, which are all in the developing world. We had become inured to the slightly uncomfortable expression that appeared on the faces of most businesspeople as soon they found out what we do, which is study poor countries—they clearly wanted to find someone else who knew something more useful to them, and were trying to figure out how quickly they could dump us without causing offence. But, suddenly, a couple of decades ago, poor countries became interesting.

They were interesting because some of them were growing fast, and any place growing fast needs investment, and that investment was a potential antidote to the specter of diminishing returns haunting the rich countries’ financiers. One way to prevent growth from slowing down is to send capital to the countries where productivity is high. That won’t help workers in rich countries, since the production won’t take place in their country, but at least national income will keep growing because capital owners will be paid well for their investment abroad.

Of course, for most economists and many businessmen, growth in poor countries is also important because of its implications for human welfare. The last few decades have been rather good for the world’s poor. Between 1980 and 2016, incomes for the bottom 50 percent of the world’s population grew much faster than the next 49 percent, which includes almost everybody in Europe and the United States. The one group that did even better was the top 1 percent, the rich in the already rich countries (plus an increasing number of superrich in the developing world), who collectively captured an amazing 27 percent of total growth in the world GDP. For comparison, the bottom 50 percent received only 13 percent of global growth.69

Nevertheless, perhaps fooled by the fact that they only see the rich getting richer, nineteen out of twenty Americans think world poverty has increased or stayed the same over this period.70 In fact, absolute poverty rates (the fraction of those living under $1.90 a day at PPP) have been halved since 1990.71

This is undoubtedly in part due to economic growth. When people are extremely poor, it takes very little growth in their incomes to lift them up. Thus, even though they often got only the crumbs, those crumbs were enough to push them above $1.90 per person per day.

This might be because the particular definition of extreme poverty we have been using sets too low a bar. But the story of the last three decades is not just one of poverty going down; we also see large and important improvements in the quality of life of the poor. Since 1990, the infant mortality rate and the maternal mortality rate were cut in half;72 as a result, more than a hundred million child deaths have been averted since 1990.73 Today, barring major social disruption, nearly everyone, boys and girls, has access to primary education.74 Eighty-six percent of adults are literate.75 Even deaths from HIV-AIDS have been declining since their peak in the early 2000s.76 The gains in income for the poor have not just been paper gains.

The new “sustainable development goals” propose to end extreme poverty (those living under $1.25 a day) by 2030, and it is quite conceivable this target will be met, or at least we will get close if the world continues to grow anywhere near the way it has been growing.

This shows how important economic growth remains for the very poor countries. For those who believe in either the Solow model or the Romer model, extreme poverty of the kind we still see in the world is a tragic waste, because there is an easy way out. In the Solow model, poor countries have the scope to accelerate their growth by saving and investing. And to the extent poor countries do not in fact grow faster than the richer ones, the Romer model tells us this has to be a consequence of their bad policies.

As Romer wrote in 2008: “The knowledge needed to provide citizens of the poorest countries with a vastly improved standard of living already exists in the advanced countries.”

He goes on to offer his growth masala:

This sounds like the usual right-wing mantra: low taxes, less regulation, less government involvement in general, except perhaps in education and in protecting private property. And by 2008, when Romer wrote this passage, this was familiar ground and we already knew enough to be skeptical.

During the 1980s and the 1990s, one of growth economists’ favorite empirical exercises became cross-country growth regressions. The game is to use the data to predict growth based on everything from education and investment to corruption and inequality, culture and religion, the distance to the sea or to the equator. The idea was to find what in a country’s policies could help predict (and hopefully affect) its economic growth. But that literature eventually hit a brick wall.

There were two problems. First, as Bill Easterly, a vocal skeptic of the ability of “experts” to give any recipe for economic growth, has convincingly shown, growth rates for the same country change drastically from decade to decade without much apparent change in anything else.78 In the 1960s and the 1970s, Brazil was a front-runner in the world growth tables; but starting in 1980, it essentially stopped growing for two decades, before resuming in the 2000s, and stopping again after 2010. Lucas’s poster child for a country that failed to grow, India, started to grow faster more or less exactly when Lucas wrote the famous piece we quoted above, where he was puzzling over why growth in India was so low. For the last thirty years, India has been one of the growth stars of the world. Growth in the countries Lucas wanted India to emulate, Indonesia and Egypt, on the other hand, tanked. Bangladesh, famously described by Henry Kissinger as a “basket case” in the 1970s, has grown at a rate of 5 percent per year or more for most years in the 1990s and 2000s, and at above 7 percent in 2016 and 2017, which puts it among the twenty fastest growers in the world.

Second, perhaps more fundamentally, these efforts to discover what predicts growth make very little sense. Almost everything at the country level is partly a product of something else. Take education, for example, one factor emphasized in the early cross-country growth literature. Clearly education is in part a product of the effectiveness of the government in running schools and funding education. A government good at delivering education is probably good at other things as well; maybe the roads are better in the same countries where teachers show up to work. If we find growth is faster where education is higher, it could be due to these other policies it tends to be bundled with. And of course it is likely that people feel more committed to educating their children when the economy is doing well, so perhaps growth causes education, and not just the other way around.

More generally, both countries and country policies differ in so many different ways that in effect we are trying to explain growth with more factors than the number of countries, including many we may not have thought of or cannot measure.79 Consequently, the value of these exercises depends very much on how much faith we have in our exact choice of what we put in them. Given that we have very little to justify any of these choices, we think the only reasonable position is to forget the entire project.

That does not mean we have not learned anything. Some of the most surprising results came from efforts to cleanly separate cause and effect. A classic pair of papers by Daron Acemoglu, Simon Johnson, and Jim Robinson (affectionately known as “AJR”) contains the most striking of these.80 They showed that countries where, in the initial years of European colonization, mortality among the early settlers was high still tend to do badly today. AJR argue that is because Europeans preferred not to settle there; instead they set up exploitative colonies where the institutions were designed to allow a small number of Europeans to lord it over vast numbers of natives who labored to grow sugarcane or cotton or to mine diamonds that the Europeans would then sell. By contrast, the places that were relatively empty to start with (think of New Zealand and Australia, for example) and where settler mortality from malaria and other such diseases was low, were the places where Europeans settled in large numbers. As a result, these places got the institutions the Europeans were then developing and that would eventually provide the basis of modern capitalism. AJR show that settler mortality several hundred years ago is an excellent predictor of, say, how business friendly contemporary institutions are in a particular country. And the countries that had low settler mortality once upon a time and are business friendly today tend to be substantially richer.

While this does not prove being business friendly causes growth (it could be the culture the Europeans brought, or the political traditions, for example, or something else entirely), it does imply that some very long-run factors have a lot to do with economic success. This broad insight has been confirmed by a number of other studies, and indeed it is in some ways what historians have always insisted on.

But what does all this tell us about what countries can actually do here and now? We learn that if you want high growth in the modern era, it is useful to have been largely empty and have had less malaria in the period between 1600 and 1900, and to have had large numbers of Europeans settle in your country (though that may have been cold comfort if you happened to be a native resident of the country at the time). Does it mean countries should try to attract European settlers in today’s very different world? Almost certainly not. The brutal indifference to local custom and lives that allowed settlers to promulgate their institutions in the pre-modern period is not likely to be available today (thank God for that).

What this also does not tell us is whether it would help to set up a particular set of institutions today, because the evidence emphasizes institutional differences that have their roots in events that took place several hundreds of years ago. Does it mean institutions need to be developed over several hundred years for them to be effective? (After all, the US Constitution of today is a very different document than when it was written, enriched by two hundred years of jurisprudence, public debate, and popular involvement.) If so, must the citizens of Kenya or Venezuela just wait?

Moreover, it turns out that among countries at roughly the same level of business friendliness, none of the conventional measures of good macroeconomic policy (such as openness to trade, low inflation, etc.—the kinds of things Romer wanted countries to adhere to) seem to predict GDP per capita.81 Conversely, while it is true that countries with “bad” policies grow slower, they are also more likely to have “worse” institutions by the measures used in this literature (less business friendly, for example), and therefore it is not clear if they are doing poorly because of policies, or because of some other side effects of their poor institutions. There is little evidence of policies having independent traction, over and above the effects of institutional quality.

What does that leave us with? It seems relatively clear there are things to avoid: hyperinflation; extremely overvalued fixed exchange rates; communism in its Soviet, Maoist, or North Korean varieties; or even the kind of total government chokehold on private enterprise India had in the 1970s with state-ownership of everything from ships to shoes. This does not help us with the kinds of questions most countries have today, given that no one, except perhaps the Venezuelan madmen, seem to be very keen on any of these extreme options. What Vietnam or Myanmar want to know, for example, is whether they should aim to emulate China’s economic model, given its stunning success, not whether to follow North Korea.

The problem is that while China is very much a market economy, as are Vietnam and Myanmar, China’s approach to capitalism is quite far from the classic Anglo-Saxon model and even its European variant. Seventy-five of the ninety-five Chinese firms on the 2014 Fortune Global 500 list were state owned, though organized like private corporations.82

Most banks in China are owned by the state. The government at both the local and the national level has played a central role in deciding how land and credit should be allocated. It also decides who gets to move where and with them the supply of labor to various industries. The exchange rate was kept undervalued for some twenty-five years, at the cost of lending billions of dollars to the United States at almost zero interest rates. In agriculture, the local governments decide who gets the right to use the land, since all land belongs to the state. If this is capitalism, it is surely with very Chinese colors.

Indeed, for all the excitement generated by the Chinese miracle these days, very few economists in 1980 or even 1990 predicted it. Often, at the end of one of our talks someone rises and asks why whatever country we are talking about doesn’t just emulate China. Except it is never clear what part of the Chinese experience we are supposed to emulate. Should we start with Deng’s China, a dirt-poor economy with comparatively excellent education and healthcare systems and a very flat income distribution? Or with the Cultural Revolution, a valiant attempt to wipe out all cultural advantages of the erstwhile elites and place everyone on an even playing field? Or with the Japanese invasion in the 1930s and its insult to Chinese pride? Or with five thousand years of Chinese history?

A similar puzzle arises in the cases of Japan and South Korea, where the governments initially pursued an active industrial policy (and to some extent still do), deciding what products to push for eventual export and more generally where investments should be made. And Singapore, where everyone had to put a large part of their earnings in a central provident fund, so the state could use their savings to build a housing infrastructure.

In all of these cases, the debate among economists has been whether growth happened because of particular unconventional policy choices, or in spite of them. And in each case, predictably, the discussion has been inconclusive. Did East Asian countries just luck out, or is there actually a lesson to be learned from their successes? Those countries were also devastated by war before they started growing fast, so a part of the fast growth might have been just the natural bounce-back. Those who herald the experience of the East Asian countries to prove the virtue of one approach or the other are dreaming; there is no way to prove any such thing.

The bottom line is that, much as in rich countries, we have no accepted recipe for how to make growth happen in poor countries. Even the experts seem to have accepted this. In 2006, the World Bank asked the Nobel laureate Michael Spence to lead the Commission on Growth and Development (informally known as the Growth Commission). Spence initially refused, but convinced by the enthusiasm of his would-be fellow panelists, a highly distinguished group that included Robert Solow, he finally agreed. But their report ultimately recognized that there are no general principles, and no two growth episodes seem alike. Bill Easterly, not very charitably perhaps, but quite accurately, described their conclusion: “After two years of work by the commission of 21 world leaders and experts, an 11-member working group, 300 academic experts, 12 workshops, 13 consultations, and a budget of $4m, the experts’ answer to the question of how to attain high growth was roughly: we do not know, but trust experts to figure it out.”83

The young social entrepreneurs basking in Silicon Valley’s enthusiastic glow have probably not read the Spence report. According to them, we do know what will get the developing world to grow—they just need to adopt the latest technologies, chief among them the internet. Mark Zuckerberg, CEO of Facebook, is a strong proponent that internet connectivity will have a huge positive impact, a sentiment echoed in a hundred reports and position papers. One report from Dalberg (a consulting firm) tells us that “the internet is a tremendous, undisputed force for economic growth and social change [italics added]” in Africa.84

The fact is evidently so obvious that the report does not bother to cite much solid evidence, which is sensible since there is no such evidence to cite. After all, in developed countries there is no evidence that the advent of the internet ushered in a new era of growth. The World Bank’s flagship publication, the World Development Report, in its 2016 edition on digital dividends, after much hemming and hawing, concluded that on the impact of the internet, the jury was still very much out.85

The internet is just one of the technologies tech enthusiasts believe can be both a commercial success and an engine of growth for poor countries. The list of “bottom of the pyramid” innovations that are supposed to change the life of the poor and power growth from the bottom up is long: clean(er) cookstoves, telemedicine, crank-powered computers, and rapid testing kits for arsenic in water, to name a few.

One common feature a lot of these technologies (though not the internet) share is that they were developed by “frugal” engineers, such as the students at MIT’s D-Lab or the entrepreneurs funded by Acumen Fund, a prominent “social” venture capital fund. Behind this and other similar funds is the believable idea that one reason why developing countries are poor is that the technologies developed in the North are not appropriate for them. They use too much energy, too many educated workers, too expensive machines, etc. In addition, they are often developed by monopolies in the North, and the South has to pay a premium to get them. The South needs its own technologies, and for that it needs capital not available from the markets. This may be why growth does not happen on its own in many countries and it’s the gap that Acumen Fund tries to fill.

While the Acumen Fund sees itself as an entirely new type of organization, not an aid organization but a venture fund for the poor countries, in a sense its technology-oriented view of growth harks back to the 1960s, when engineers dominated the aid world and went bust trying to bridge the “infrastructure gap,” giving large loans to poor countries for building dams and train lines that would allow them to catch up with rich countries. Despite the lack of evidence that this has helped those countries to grow, the fascination for electricity as the source of growth and development has never really gone away. Ecuador is currently under severe financial strain thanks to a loan from China to build a massive dam that was never fully operational. Acumen loans are smaller and they are given to private actors rather than to governments, but the dream is still one where engineers will fix the world’s problems. One of Acumen Fund’s key sectors is electricity. The ideal source of energy has changed from large dams to power from grain husks, or the sun, and the latest “cool” idea is that it is possible to develop cheaper “off the grid” solutions to reach poor communities; but the focus on electricity goes back fifty years.

It turns out, however, that it is not easy to invent appropriate technologies that are also profitable in a poor country. A good part of what Acumen funds fails. A rule of thumb in the social investing world is that 10 percent of the ventures work out (the rest fold) and only 1 percent reach significant scale. The issue is more that it is difficult to identify those supposedly life-changing new products and services, and efforts to do so often meet a frustrating lack of interest from the people whose lives are supposed to be changed.

Electricity is a case in point. In a recent randomized controlled trial in Kenya, researchers partnered with Kenya Rural Electrification Authority to offer electricity connection at different prices in different communities. The demand fell very sharply as price rose, and villagers were not willing to pay anywhere near what would have been sufficient to cover the cost of connecting to the grid (not to mention building the grid).86

The frugal engineering world is littered with many similar disasters, from the $100 laptop to educate the world (which actually costs $200 and has been shown to have no impact on what children actually learn),87 to cleaner cookstoves that nobody wanted,88 to various water-filter technologies89 and innovative latrines.90 A lot of the problem seems to be that these innovations take place in a void, insufficiently connected to the lives they wish to change. The core ideas are often clever, and it remains possible that one day they will click, but it is hard to place a lot of faith in this prospect.

A central tenet of all the growth theories we have discussed is that resources are smoothly delivered to their most productive use. This is a natural hypothesis as long as markets work perfectly. The best companies should attract the best workers. The most fertile plots of land should be farmed most intensively, while the least productive will be used for industry. People who have money to lend should lend to the best entrepreneurs. This assumption is what allows macroeconomists to speak of the stock of “capital” or “human capital” of an economy, despite the obvious reality that the economy is not one giant machine: as long as resources flow to their best use, each separate enterprise is like one cog in a smoothly operating machine, which spans the entire economy.

But this is often not true. In a given economy, productive and nonproductive firms coexist, and resources do not always flow to their best use.

Lack of adoption of available technologies is not just a problem for poor households; it seems to also be a problem in industrial settings in developing countries. In many cases, the best firms in an industry use the latest worldwide technology but other firms do not, even when it seems it would make sense economically.91 Often, this is because the scale of their production is too small. For example, until recently the typical clothing manufacturer in India was a tailor who made made-to-measure clothes in his one-man workshop, rather than a firm that mass produces. TFP is low not because the tailors are using the wrong technology, but because tailoring firms are too small to benefit from the best technology. In a sense, the puzzle is why these firms exist.

So the problem with technology in developing countries is not so much that profitable technologies are not available and accessible, but that the economy does not appear to make the best use of available resources. And this is true not only of technologies but also of land, capital, and talents. Some firms have more employees than they need while others are unable to hire. Some entrepreneurs with great ideas may not be able to finance them, while others who are not particularly good at what they are doing continue operating: this is what macroeconomists call misallocation.

A vivid instance of misallocation comes from the impact of the introduction of cell phones on fishing in the state of Kerala in India. Fishermen in Kerala would go out to fish early in the morning and return to shore midmorning to sell their catch. Before the cell phone, they would land at the nearest beach, where their customers would meet them. The market would run until there were no customers left or the fish ran out. Since the catch varied quite a bit from day to day, there were a lot of wasted fish at some beaches, while at the same time there were often disappointed customers at others. This is a stark example of misallocation. When cell phone connectivity became available, fishermen started to call ahead to decide where to land; they would go where there were lots of customers waiting and not a lot of boats. As a result, waste essentially vanished, prices stabilized, and both customers and sellers were better off.92

This first story spawned a second one. The main tool of trade for a fisherman is his boat, and good boats last much longer than bad boats. The technology of making a fishing boat is always the same, but some craftsmen are much better at it than others. Before cell phones, fishermen used to purchase their boats from the nearest boat makers. But when they started to travel to different beaches to sell their fish, they often discovered there were better boat makers elsewhere, and they started to ask them to build their new boats. The result was that the better boat makers got more work and the worst went out of business. The quality of the average boat improved and in addition, because the better boat makers got more work and therefore got to use their existing boat-making infrastructure more effectively, they could lower the price of the boats. Misallocation went down: the workers making boats, the equipment, the wood, the nails, and the ropes that went into a boat were all used more effectively.93

What is common to these two stories is that a communication barrier led to misallocation. When communication improved, the same resources were better used, resulting in higher TFP, since more was done with the same inputs.

Misallocation is pervasive in developing economies. Take the city of Tirupur in South India, the T-shirt capital of the country, which we have already encountered in chapter 3.94 There are two kinds of entrepreneurs in Tirupur: those who come from outside to start a T-shirt-making business, and those born and brought up in the area. The latter are almost uniformly the children of affluent local farming families, the Gounders, looking to do something different with their lives. Those who go there to make T-shirts are generally better at T-shirt making than the locals; many have family connections in the T-shirt business, and perhaps as a result firms run by outsiders make the same number of T-shirts with many fewer machines and their firms grow a lot faster.

But despite being more productive, Abhijit found in a study with Kaivan Munchi, the firms run by the immigrants were smaller in size and had less equipment than the firms run by the locals. The Gounders poured money into the firms run by their children instead of doing the “efficient” thing: lending money to migrants and passing the interest income so earned to their sons. As a result, efficient and inefficient firms could persist in the very same town.95

When Abhijit asked them why they preferred to sponsor their sons rather than lend money to the more talented outsiders and live off the proceeds, the Gounders explained they could not be sure of getting their money back. In the absence of a well-functioning financial market, they preferred to give money to their inept sons and get lower but relatively safe returns. It is also probably the case that they felt they had a duty to give their sons not only some hard cash, but also a means to earn a decent living.

Family firms are common all over the world (from small farms to large family groups), and they do not always fully adapt to “economic” incentives. Firms are passed on to sons even when daughters would be better at managing them,96 all the fertilizer in the family goes to one (male) person’s plot when it would make sense to use a little bit in all the fields.97 That is of course true not just of small farms in Burkina Faso or family concerns in India and Thailand, but of the United States as well. Out of 335 CEO successions at family firms a researcher investigated, 122 were “family successions” where the new CEO was a child or a spouse of the current CEO (often a founder or the child of a founder). On the day of the succession, the stock market returns of the companies that appointed an outside CEO went sharply up, while the returns of the companies that appointed an inside CEO did not. The market was rewarding the appointment of an outsider. And apparently the market was onto something. Firms that appointed family CEOs experienced large declines in performance in the subsequent three years, compared to firms that promoted unrelated CEOs: their return on assets fell by 14 percent.98

What all of this tells us is that we cannot take it for granted that resources will flow to their best use. If they do not within a single family, or within a town, we clearly should not expect them to do so across an entire country. Misallocated resources will in turn lower overall productivity. Part of the reason poor countries are poor is they are less good at allocating resources. The flip side is that it is possible to grow just by allocating the existing resources to more appropriate uses. In the last few years, macroeconomists have spent a lot of effort trying to quantify just how much growth could come from better allocation. This is hard to do perfectly, but the results have been very encouraging. One very prominent estimate suggests that, in 1990, just the reallocation of factors within narrowly defined industries could have increased Indian TFP by 40 percent to 60 percent and Chinese TFP by 30 percent to 50 percent. If we allowed reallocations across broader categories, the estimates would surely be even larger.99

And then there is the misallocation we do not see, the great ideas that never see the light of day. Given that venture capital is so much more active in scouting out new ideas in the United States than in India, it is plausible that India is also missing more of these unsung geniuses.

Where does misallocation come from? Indian firms grow much more slowly than US firms, but are also much less likely to shut down.100 In other words, the United States is an “up or out” economy, where people try something new and either succeed and make it big or fail after a few years. By contrast, the Indian economy is exceedingly sticky: good firms do not grow and bad firms do not die.

These two facts are probably closely related: the fact that good firms cannot grow fast enough also helps explain why bad firms can survive. If the best firms were to grow fast, they would drive down the price of whatever they sold and therefore force out everyone except those efficient enough to make money even when the prices were low. By the same token, they would drive up wages and the cost of raw materials, further discouraging bad firms. In contrast, if they remain small and service only the local demand, a less efficient firm can easily survive in the market next door.

One natural culprit is the capital market. It clearly plays a role in the Tirupur example, where the most productive entrepreneurs in the most productive T-shirt cluster in India cannot borrow enough to catch up in size with the less productive local firms. In India and China, estimates imply that simply reallocating capital across firms would erase most of the TFP gap created by misallocation.101

This interpretation dovetails with a generally shared sense that the banking sectors in both China and India have serious problems. Indian banks are famous for trying to avoid lending to anyone except blue-chip borrowers (usually without recognizing that yesterday’s blue-chip firms are often today’s disaster waiting to happen). Chinese banks have undergone significant reforms since the 1990s, with the goal of allowing for entry of different actors and improving the governance of the state-owned banks, but the “big four” state-owned banks still tend to be all too willing to lend to dubious projects with good political connections.102 Finding money remains difficult for a young and ambitious entrepreneur with a good idea but no powerful friends.

Indian banks have very much the same problem, and in addition they are reputedly extremely overstaffed. Overstaffing means they need to put a large wedge between the rate at which they lend to firms and the savings rate they offer to depositors if they want to break even. As a result, bank lending rates in India are high relative to the rest of the world,103 even though depositors earn very little interest.104 This also discourages investment by those who need to borrow to do so and favors those with a rich relative to support them, like the Gounders of Tirupur. Bad banks hurt efficiency from both ends; because of them, savings rates are lower than they could be and savings are poorly managed.

In addition, companies need risk capital, funding that unlike bank funding protects them when they are hit by bad luck. Stock markets do this, but the Chinese stock market is yet to be widely trusted and the Indian one, while older and better run, is still very blue-chip dominated.

Poorly developed land markets are another reason why companies do not grow. In order to grow, a productive firm will need to acquire more land and buildings to make room to accommodate new machines and employees. In addition, land and buildings can be used as collateral for loans. This becomes a huge problem when land markets function poorly. To take a very common example, in many countries ownership of land and property is often disputed. A claims B’s land, the land gets placed under court authority, and it typically takes years to settle the dispute. A recent study suggests that in India land and buildings play a big role in misallocation.105 In fact, in about half the districts in India, more productive firms tend to have less land and buildings than the least productive ones! This is likely to be a large problem in many countries where property rights on land are not very clearly defined.

But there are other, more psychological, reasons why the best firms are not taking over India, Nigeria, or Mexico. Perhaps the owners like the idea of leaving their son a running business and prefer to avoid the risk of outside control that comes with outside financing; raising money on the stock market, for example, requires setting up an independent board of directors who might get in the way of the succession plans.

And perhaps ultimately the owners do not care enough about growth to put all they have behind that agenda. If no one else is growing fast, they are not at risk of being pushed out. They have a reasonable living and a place to work. Why make it more stressful by trying to grow? A very interesting recent study looks at management gaps in Indian firms.106 By the norms of what the United States calls good management, firms in developing countries are terribly managed. One might dismiss this as prejudice against other ways of managing. Indians in particular are very proud of their way of doing business on a shoestring, what they call jugaad.107 This requires being inventive in using what you have, and perhaps this is what the managers are doing. But managers are failing in ways that could not possibly make sense for them. For example, trash is allowed to accumulate on the shop floor, to the point that it becomes a fire hazard. Or unused materials are bagged and thrown into an inventory room, but nobody labels or lists them so it becomes virtually impossible to reuse them. When the researchers, one of them a former management consultant, sent (for free) a team of highly paid consultants to work for five months with the managers of a randomly chosen set of these firms, profits went up by $300,000 per firm, which even for these relatively large firms was not chicken feed. Moreover, most of the changes that made this happen were relatively simple things, like labeling inventories and removing trash. It is hard to see why the managers, if they wanted to raise profits, would need this rather expensive external help (the consulting would have cost them $250,000 had they paid for it). They undertake obvious changes if someone points them out and shames them into doing it, but not when left to themselves. It has to be that the owners ultimately don’t feel strongly about doing the best they can possibly do.

Companies also need labor. One might imagine this at least would not be a problem in a labor-abundant poor country, but it is actually not true. Even unskilled laborers in Odisha, one of India’s poorest states, hold out for what they think is a fair wage, even if the alternative is not getting a job; workers who accept a lower wage are punished by others.108

According to the nationally representative National Sample Survey, in 2009 and 2010, 26 percent of all Indian males between the ages of twenty and thirty with at least ten years of education were not working. This is not because there were no jobs: the fraction of those under thirty with less than eight years of education who were not working was 1.3 percent. And, in fact, the fraction of those with ten years of education above thirty who were not working was about 2 percent.109 We see the same pattern in 1987, 1999, and 2009, so this is not because the young of today are less employable.110

There are plenty of jobs, just not jobs these young men want. They will eventually accept jobs they refused to take when they were younger, probably because the economic compulsions become stronger as they age (their parents, who feed and house them now, will retire or pass on; they will want to get married), and the job options shrink (government jobs, in particular, have an age cut-off that is often close to thirty).

Esther found something very similar in Ghana. A little over ten years ago, about two thousand adolescents were identified as having passed the (hard) exam necessary to qualify for higher secondary school in Ghana (corresponding roughly to grades ten to twelve) but had not enrolled in the first trimester for lack of funds.111 A third of them were randomly selected and offered a full scholarship for their entire time in secondary school. Before they were chosen for the scholarship, Esther and her co-authors asked their parents what they thought the economic benefit of enrolling in secondary school would be. The parents were generally optimistic. On average, they thought a person like their son or daughter could earn almost four times as much if they completed secondary school than if they did not start it. Moreover, they believed these gains would come because of greater access to government jobs, such as teaching and nursing. Not surprisingly, given these beliefs, three-quarters of the kids offered a scholarship jumped at the opportunity and completed secondary school, compared to only about half of the kids who did not get a scholarship. Esther and her colleagues have been following the progress of these adolescents ever since, interviewing them about once a year. They find many positives: the students learned useful things in school and it changed their lives in many ways; they all performed better on a test that measures their ability to apply knowledge to concrete situations; girls waited longer before starting a family and had fewer children.

The not so good news is that the impact on their average earnings was not very large, except for the few who got a government job. The parents were right about one thing: secondary education is indeed essential to get access to the college degrees that allow graduates to get coveted jobs. Secondary school graduates were indeed more likely to be teachers, to have other government jobs, or to have private jobs with benefits and fixed salaries. But where they got it wrong is that although secondary education is necessary, it is not sufficient. Secondary school scholarship winners (especially the girls) were more likely to go on to college, but the probability was still quite low (16 percent among scholarship winners as against 12 percent in the comparison group). And only a few of them managed to get a government job. The scholarship doubled this probability, but it went from 3 percent to 6 percent; that is, from really, really, small to really small.

Meanwhile, though they were already twenty-five or twenty-six, most of those who had gone to secondary school were still waiting for something better. A substantial fraction were not working at all: only 70 percent of the kids in the sample (treatment and control combined) had earned anything in the last month.

Intrigued by what these young people could be doing instead of working, we visited several of them. Steve, a young, affable, well-spoken man, received us in his home. He had graduated from secondary school over two years before but had not worked since then. He was hoping to go to college and study politics, with the aim of being a radio anchor one day, but his grades on the admission test had been too low so far. He kept retaking it. In the meantime, he was living off of his grandmother’s pension. He saw no reason to let go of his dreams yet. He probably will eventually, but as he sees it, he’s still young.

The flip side of this is that even in countries with frighteningly high unemployment rates, like South Africa (where 54 percent of those between the ages of fifteen and twenty-four say they are unemployed112), companies complain they cannot get the workers they want: workers with some education, a good attitude toward work, and a willingness to accept the wages on offer. In India, the government has invested an enormous amount of public resources on getting workers ready for the jobs the economy is generating. A couple of years ago, Abhijit collaborated with one of these businesses that does vocational training and job placement for the service sector. The company was worried they were not doing particularly well at placing their students. The data confirmed this. Out of 538 young men and women who signed up for a course, 450 completed it. Of those, 179 got job offers and 99 accepted their offers, but after six months only 58 were in the jobs the company had found for them, a hit rate of just over 10 percent. Another 12 were working elsewhere.113 What were they doing instead, we asked a group of those who had been offered a job but had either never taken it or quit more or less immediately. They were either taking what they called “competitive exams” (to get a government job or a job in a quasi-governmental organization, like a public-sector bank) or studying to complete their bachelor’s degree and then apply for a government job. Or just sitting at home, despite the fact that their families could ill-afford that.

Why did they not want the jobs they had been offered? We heard many answers, but it all came down to their not liking them—too much work, too long hours, too much time spent standing, too much going from one place to another, too little pay.

Part of the problem is a mismatch of expectations. The young men and women we interviewed in India grew up in families where post-primary education was still often a novelty; their fathers had on average eight years of schooling, their mothers less than four. They were told that if they studied hard they would get a good job, meaning mostly a desk job or a teaching job. This was closer to the truth in their parents’ generation than it is today (especially for historically disadvantaged populations like the lower castes who benefitted from affirmative action). The growth in government jobs slowed and eventually stopped in the face of budgetary pressures,114 but the population of the educated, even among the historically disadvantaged, kept growing.115 In other words, the goalposts have moved.

Something similar happened in countries like South Africa, and also in Egypt and other countries of the Middle East and North Africa, which were more developed than India to start with. There, it was not enough to have completed secondary school, but for a while having a bachelor’s degree served the same screening function: if you could show your BA degree you would walk into a government job. That is no longer true, but these countries are still producing millions of BAs in subjects like Arabic and political science, for which there is no market anymore. That today’s graduates do not have the skills employers want is of course a constant complaint the world over, including in the United States. But the situation is quite extreme in those countries.

The mismatch between reality and expectation is reinforced by the lack of exposure to the real labor market. With Sandra Sequeira, Abhijit evaluated a program in South Africa providing young workers in the townships (the erstwhile black ghettos of the apartheid era) with free transportation to look for jobs far from home. Those randomly chosen to get the transportation subsidy did travel a lot more, but there was no effect on employment. What did change, however, was their perception of the labor market. Almost everyone was too optimistic to start with; the salaries they expected to earn were 1.7 times higher than the actual salaries reported by employed workers similar to them. Being exposed to the actual labor market put a dampener on their expectations, and their wage expectation became closer to the truth.116

Labor markets frozen by this kind of radical mismatch are wasting resources. These young people are mostly waiting for jobs they will not get. In India, newspapers frequently write about the mad rush for government jobs; for example, that twenty-eight million people applied for ninety thousand low-level jobs in the government-owned railways.117

From the perspective of developing countries, some of these problems are purely self-inflicted. Part of the problem is that there are a small fraction of jobs that are much more attractive than the rest, for reasons having nothing to do with productivity. The best examples are government jobs. In the poorest countries, there is a large gap between the wages of public- and private-sector employees. In the poorest countries, public-sector workers earn more than double the average wage in the private sector. And this is not counting generous health and pension benefits.118

This kind of difference can throw the entire labor market into a tailspin. If government-sector jobs are so much more valuable than private-sector jobs, but also very scarce, it is worthwhile for everybody to wait around and queue for those jobs. If the process of queuing and screening entails, as it often does, taking some exams, people may spend most of their working lives (or as much as they are allowed to by their families, anyway) studying for those exams. If the government jobs stopped being quite so desirable, the economy would gain many years of productive labor, wasted in the pursuit of the mostly unattainable. Of course, government jobs are attractive in other countries as well, particularly because they often come with job security. But the wage gap is not quite as large and the queue not nearly as long.

Cutting wages in government jobs would probably be a battle, but it would not be so difficult, for example, to limit the number of times people could apply for government jobs, or to make the age cut-off more stringent. This would avoid the massive waste of everyone waiting around. It could add an element of luck to the job allocation process, but it is not obvious that the resulting allocation would be worse than under the current system, which favors those who can afford to wait. In Ghana, while Steve was twiddling his thumbs, some other young graduates had had to find something to do because they had no one to subsidize their lifestyle. They did not lack imagination: we met a nut farmer, a DJ who specialized in funerals, a preacher in training, and two footballers on a minor league team.

The labor market problems in developing countries are not, however, limited to the outsized attractiveness of the government sector. In Ghana, secondary school graduates are also attracted by a class of private jobs that offers benefits, high wages, and a measure of employment protection. In many developing countries, the labor markets feature this duality: there is a large informal sector without any protection, with many people who are self-employed for lack of better options, and a formal sector where employees are not only pampered but also strongly protected. Some employment protection is of course necessary; workers cannot be at the whim of their employer. But labor market regulations are so stringent that they really put a chokehold on any efficient reallocation of resources.

Where does all of this leave us in our understanding of economic growth? Well, Robert Solow was right. Growth seems to slow down as countries get to a certain level of per capita income. At the technological frontier, that is to say in the rich countries, TFP growth is largely a mystery. We do not know what propels it.

And Robert Lucas and Paul Romer were right too. For the poorer countries, convergence is not automatic. This is probably not mainly because of spillovers. It is more that TFP is much lower in poorer countries, to a significant extent because of market failures. And therefore to the extent that business-friendly institutions have something to do with fixing market failures, Acemoglu, Johnson, and Robinson were right too.

And yet all of these economists were also wrong, because they thought of economic growth and of a country’s resources as aggregate things (the “labor force,” the “capital,” the “GDP”), and in doing so they probably missed the key point. Everything we have learned about misallocation tells us we have to step beyond the models and think of how the resources are used. If a country starts by using its resources very badly, like China did under communism or India did in its days of extreme dirigisme, then the first benefits of reform may come from moving resources to their best uses. Perhaps the reason why some countries, like China, can grow so fast for so long is that they start with a lot of poorly used talent and resources that can then be harnessed. This is neither Solow’s nor Romer’s world, in which a country would need either new resources or new ideas to grow. It might also suggest the growth could slow down rapidly, once those wasted resources have all been put to good use, and growth becomes dependent on additional resources. Much is being written about the economic slowdown in China; growth is definitely slowing down and that is probably to be expected. This trend will almost surely continue, whatever Chinese leaders do now. China accumulated resources rapidly, as it had plenty of room to catch up; in the process, the most blatant sources of misallocation were eliminated, which means there is less room to improve now. The Chinese economy relied on exports to provide know-how, investment, and endless (for a while) global demand. But now they are the largest exporter in the world, so they cannot possibly continue to grow their exports much faster than the world is growing. China (and the rest of the world) will have to come to terms with the reality that their era of breathtaking growth is likely coming to an end.

In terms of what is to come, it looks like the United States can relax a bit. In 1979, Harvard professor Ezra Vogel published a book, Japan as Number One, that predicted Japan would soon overtake all other countries to become the number one economic superpower. Western countries, he argued, needed to learn from the Japanese model. Good labor relations, low crime, excellent schools, and elite bureaucrats with long time horizons was the new recipe Vogel identified for permanently faster growth.119

Indeed, had it continued to grow at its average growth rate over the decade 1963–1973, Japan would have overtaken the United States in terms of GDP per capita by 1985, and in overall GDP by 1998. It did not happen. What happened instead is enough to make one superstitious. The growth rate crashed in 1980, the year after Vogel’s book came out. And it never really recovered.

The Solow model suggests a simple reason. Due to a low fertility rate and the near complete absence of immigration, Japan was (and still is) aging rapidly. The working-age population peaked in the late 1990s and has been declining. This means TFP must grow all the more rapidly to keep fast growth going. Another way to say this is that Japan would have to find some miracle for its existing labor force to become more productive, since we still have no reliable way to boost TFP.

In the euphoria of the 1970s, some believed this to be possible, which may explain why people continued to save and invest in Japan in the 1980s, despite the slowdown. Too much good money chased too few good projects in the so-called bubble economy of the 1980s, with the consequence that banks ended up with many bad loans and a huge crisis in the 1990s.

China faces some of the same problems. It is aging fast, partly as a result of the one-child policy, which has proven difficult to reverse. It might still eventually catch up with the United States in per capita terms, but the slowing growth means it will take quite a while. If China slows to 5 percent per year, which is not implausible, and stays there, which is perhaps optimistic, and the United States continues to bounce around 1.5 percent, it will take at least thirty-five years for China to catch up with the US in terms of per capita income. Meanwhile, the Chinese authorities may also want to relax and accept the writ of Solow. Growth will slow.

They are aware of it, and have made a conscious attempt to alert the Chinese people to this fact, but the growth targets they have set may still be too high. The danger is that it could put the leadership in a bind and lead them to make bad decisions in an effort to make growth come back, as Japan did before them.

If a fundamental driver of economic growth is resource misallocation, it opens the door to various unorthodox strategies to make growth happen. Such strategies are meant to respond to the particular way in which resource use in a country is distorted. The Chinese and the South Korean governments did a good job of identifying sectors that were too small and therefore not meeting the economies’ needs (they tended to be heavy industry providing basic raw materials to other industries, like steel and chemicals) and directed capital toward them through state investments and other interventions. This might have sped up the transition to efficient resource use.120

That it worked in those two countries does not necessarily mean it is something every country should emulate. Economists tend to be very wary of industrial policy, for good reasons. The history of state-directed investments is not one that inspires confidence; judgments are frequently bad even when they are not actually deliberately distorted to benefit someone or some group, which is often. These are “government” failures just as there are market failures, and there are so many instances of these that it would be very dangerous to blindly rely on governments to pick the winners. But there are also so many market failures that it makes no sense either to rely on the market alone to allocate resources to the right use; we need an industrial policy designed that keeps in mind these political constraints.

Another implication of the idea that growth is slowed down by misallocation is that countries like India that are growing fast right now should fear complacency. It is relatively easy to grow fast, starting from a spectacularly messed-up economy, because of the gains from better resource use. In Indian manufacturing there was a sharp acceleration in technology upgrading at the plant level, and some reallocation toward the best firms within each industry after 2002. This appears to be unrelated to any economic policy, and is described as “India’s mysterious manufacturing miracle.”121 But it is no miracle. At its root, it is a modest improvement from a dismal starting point, and one can imagine various reasons it happened. Perhaps a generational shift, as control passed from the parents to their children, often educated abroad, more ambitious, and savvier about technology and world markets. Or the effect of the accumulation of modest profits that eventually made it possible to pay for the shift to bigger and better plants.

But as the economy sheds its worst plants and firms, the space for further improvement naturally shrinks. Growth in India, like that in China, will slow. And there is no guarantee it will slow when India has reached the same level of per capita income as China. When China was at the same level of per capita GDP as India is today, it was growing at 12 percent per year, whereas India thinks of 8 percent as something to aspire to. If we were to extrapolate from that, India will plateau at a much lower level of per capita GDP than China. The growth tide does raise all boats, but it doesn’t lift all boats to the same level—many economists worry that there may be such a thing as the middle-income trap, an intermediate-level GDP where countries get stuck or nearly stuck. According to the World Bank, of 101 middle-income economies in 1960, only 13 had become high income by 2008.122 Malaysia, Thailand, Egypt, Mexico, and Peru all seem to have trouble moving up.

Of course, there are many pitfalls in any such extrapolation, and India should treat it as what it is: no more than a warning. It is quite possible that India’s growth, in spite of all of its problems, has very little to do with some special Indian genius. Instead, it has a lot to do with the flip side of misallocation: the opportunities of being an economy with a large pool of potential entrepreneurs to draw upon and lots of unexploited opportunities.

If this is the right story, India should start to worry about what happens when those opportunities begin to run out. Unfortunately, just as we don’t know much about how to make growth happen, we know very little about why some countries get stuck but others don’t—why South Korea kept growing but Mexico did not—or how one gets out. One very real danger is that in trying to hold on to fast growth, India (and other countries facing sharply slowing growth) will veer toward policies that hurt the poor now in the name of future growth. The need to be “business friendly” to preserve growth may be interpreted, as it was in the US and UK in the Reagan-Thatcher era, as open season for all kinds of anti-poor, pro-rich policies (such as bailouts for overindebted corporations and wealthy individuals) that enrich the top earners at the cost of everyone else, and do nothing for growth.

If the US and UK experience is any guide, asking the poor to tighten their belts, in the hope that giveaways to the rich will eventually trickle down, does nothing for growth and even less for the poor. If anything, the explosion of inequality in an economy no longer growing has the risk of being very bad news for growth, because the political backlash leads to the election of populist leaders touting miracle solutions that rarely work and often lead to Venezuela-style disasters.

Interestingly, even the IMF, so long the bastion of growth-first orthodoxy, now recognizes that sacrificing the poor to promote growth was bad policy. It now requires its country teams to include inequality in factors to take into consideration when providing policy guidance to countries and outlining conditions under which they can receive IMF assistance.123

The key ultimately is to not lose sight of the fact that GDP is a means and not an end. A useful means, no doubt, especially when it creates jobs or raises wages or plumps the government budget so it can redistribute more. But the ultimate goal remains one of raising the quality of life of the average person, and especially the worst-off person. And quality of life means more than just consumption. As we saw in the previous chapter, most human beings care about feeling worthy and respected; they suffer when they feel they are failing themselves and their families. While better lives are indeed partly about being able to consume more, even very poor people also care about the health of their parents, about educating their children, about having their voices heard, and about being able to pursue their dreams. A higher GDP may be one way in which this can be given to the poor, but it is only one of the ways, and there is no presumption that it is always the best one. In fact, the quality of life varies enormously between middle-income countries. For example, Sri Lanka has more or less the same GDP per capita as Guatemala but maternal, infant, and child mortality are much lower in Sri Lanka (and are comparable with those in the United States).124

More generally, looking back, it is quite clear that many of the important successes of the last few decades were the direct result of a policy focus on those particular outcomes, even in some countries that were and have remained very poor. For example, a massive reduction in under-five mortality took place even in some very poor countries that were not growing particularly fast, largely thanks to a focus on newborn care, vaccination, and malaria prevention.125 And it is no different with many of the other levers for fighting poverty, be it education, skills, entrepreneurship, or health.We need a focus on the key problems and an understanding of what works to address them.

This is patient work; spending money by itself does not necessarily deliver real education or good health. But the good news is that by contrast to growth we know how to make progress here. One big advantage of focusing on clearly defined interventions is that these policies have measurable objectives and therefore can be directly evaluated. We can experiment with them, abandon the ones that do not work, and improve the ones with potential.

The recent history of malaria is a good example. Malaria is one of the biggest killers of small children and a disease preventable by avoiding mosquito bites. Since the 1980s, the number of malaria deaths had been rising every year. At the peak in 2004 there were 1.8 million deaths from malaria. Then in 2005 there was a dramatic turning point. Between 2005 and 2016, the number of deaths from malaria declined by 75 percent.126

Many factors probably contributed to the decrease in the number of malaria deaths, but the widespread distribution of insecticide-treated bed nets almost surely played a key role. Overall, the benefits of nets are well established. In 2004, a review of the evidence from twenty-two carefully done randomized controlled trials found that, on average, one thousand more nets distributed contributed to a reduction of 5.5 deaths per year.127 As we described in Poor Economics, however, there was a big debate at the time on whether nets should be sold to beneficiaries (at a subsidized price) or given for free.128 But an RCT by Pascaline Dupas and Jessica Cohen,129 replicated since then by several other studies, established that free nets are in fact used just as much as nets that are paid for, and free distribution achieves a much higher effective coverage than cost sharing. Since Poor Economics was published in 2011, this evidence eventually convinced the key players that massive distribution was the most effective way to fight malaria. Between 2014 and 2016, a total of 582 million insecticide-treated mosquito nets were delivered globally. Of these, 505 million were delivered in Sub-Saharan Africa and 75 percent were distributed through mass distribution campaigns of free bed nets.130 The magazine Nature concluded that insecticide-treated net distributions averted 450 million malaria deaths between 2000 and 2015.131

The accumulation of evidence took some time, but it worked. Even the skeptics were convinced. Bill Easterly who in 2011 was an outspoken critic of free bed net distribution, gracefully acknowledged in a tweet that his nemesis Jeff Sachs was more right than he was on this particular issue.132 The right policy choices were made, leading to tremendous progress against a terrible scourge.

The bottom line is that despite the best efforts of generations of economists, the deep mechanisms of persistent economic growth remain elusive. No one knows if growth will pick up again in rich countries, or what to do to make it more likely. The good news is that we do have things to do in the meantime; there is a lot that both poor and rich countries could do to get rid of the most egregious sources of waste in their economies. While these things may not propel countries to permanently faster growth, they could dramatically improve the welfare of their citizens. Moreover, while we do not know when the growth locomotive will start, if and when it does, the poor will be more likely to hop onto that train if they are in decent health, can read and write, and can think beyond their immediate circumstances. It may not be an accident that many of the winners of globalization were ex-communist countries that had invested heavily in the human capital of their populations in the communist years (China, Vietnam) or countries threatened with communism that had pursued similar policies for that reason (Taiwan, South Korea). The best bet, therefore, for a country like India is to attempt to do things that can make the quality of life better for its citizens with the resources it already has: improving education, health, and the functioning of the courts and the banks, and building better infrastructure (better roads and more livable cities, for example).

For the world of policy makers, this perspective suggests that a clear focus on the well-being of the poorest offers the possibility of transforming millions of lives much more profoundly than we could by finding the recipe to increase growth from 2 percent to 2.3 percent in the rich countries. In the coming chapters, we will go one step further and argue that it may even be better for the world if we did not find that recipe.