The Wealth of Humans, Part VI: Superstars

It’s tempting to ask, what’s really different about this time? Isn’t this the same story we’ve heard every time some new advance comes along? It’s the nature of technology to destroy some jobs, and enhance and create others; that’s creative destruction and it’s always been with us.

While I’m generally sympathetic to this sort of ‘Ecclesiastical’ thinking (nothing new under the sun), I do think there are signals that this time really is different in important ways.

I don’t believe Avent touches on this in the book, but what I see as one of the big differences is the prevalence of what’s dubbed The Superstar Effect (pdf). This is most plainly evident in certain industries, like music, film, and televised sports: industries that are based around the transmission of information, or where, to quote from the paper, ‘the costs of production do not rise in proportion with the size of the market’. In the case of information transmission, such as film, television, or recorded music, the marginal cost of distribution is virtually zero. Furthermore, these types of goods have the characteristic of being ‘non-rivalrous’: just because I purchased The Wealth of Humans doesn’t prevent you from buying it also (so you should!).

Another key feature is imperfect substitution in these markets. Buying 10 cheap sponges could be the equivalent of buying 1 good sponge: the good one lasts ten times as long as the cheap sponge, but beyond that there isn’t much difference; they function as perfect substitutes. But watching 10 episodes of Home Improvement doesn’t add up to one episode of Breaking Bad.  Consequently, demand increases more than is proportionate to the degree of quality. To quote again from the paper, ‘If a surgeon is 10 percent more successful in saving lives than his fellows, most people would be willing to pay more than a 10 percent premium for his services.’

When markets have both of these features, non-rivalry and imperfect substitution, the superstar effect comes into play: one or a few players can capture outsize shares of the market. Think of the acting abilities of the average Hollywood star; they aren’t thousands of times better at acting than the people you see on TV, but being just slightly preferable to the average consumer means those stars make huge gains.

Technology, via replication, is what allows non-rivalry. And again, historically this has been most evident in industries based around transmission or communication of information. So it’s quite likely the Digital Revolution will continue to exacerbate this problem by rendering more and more industries subject to replication. Furthermore, technology will continue to reduce the marginal costs of production in many other fields so that they also can also expand more than proportionally. One example could be programmers; in the deep learning market, the good being designed is multi purpose. Once you have your program, you can apply it to all sorts of problems at little additional cost; DeepMind didn’t need to be reworked from the ground up each Go match it played, nor will it have to be totally redesigned for its next endeavor. 

Or consider remote surgery. The best surgeons will be able to service vastly expanded markets, rendering the services of all but the most skilled surgeons largely unnecessary. So I could see superstar effects taking hold in many more industries.

This is one big reason why I think the changes brought by the Digital Revolution aren’t some temporary phase. The very structure of many job markets has the potential to be fundamentally transformed to dramatically favor only the best. And if this comes to pass, what happens to the rest of us?

The Wealth of Humans, Part V: The Employment Trilemma

What then does this combination of automation, an abundance of labor and the rise of social capital mean for the future? Avent suggests that we now face an employment trilemma between three factors, of which we can choose at most two: high productivity/wages; resistance to automation; and mass employment.

The reasoning is thus: Any job with the potential to have all three characteristics will become self limiting almost immediately. Jobs with high productivity inevitably limit the amount of employment, either because output quickly exceeds demand (lowering prices and thus productivity) or the high wages incentivize producers to automate the tasks (lowering employment).

We can see this in the data. Let’s look again at the BLS data on industry performance from 2004-2014. Again, for this data the measure for productivity is output per worker rather than the usual output per labor hour. I’m working on getting another data set from the BLS together that has the usual productivity measures in it, but this will have to do for now.

First, a look at the headline sectors. The size and color of each bubble are the number of employees and productivity of each sector, respectively. The x-axis is employment growth, y-axis is output growth, and the 45 degree line marks where the two are equal (above the line means output has grown faster than employment, and vice versa).


An industry that avoided the trilemma would be big and dark red, and we do see one of those: manufacturing. But the sector is also shrinking and has been for decades thanks to automation and globalization. The information sector is promising in having high productivity and resistance to automation, but again employment is shrinking. The sectors that are commonly touted as being the future of mass employment, education and health care, have low productivity. Indeed it is clear from this graph that all the large sectors and growing sectors have low productivity, as the trilemma suggests.

What about mining? That has high productivity and rapid employment growth, after all. Could this be a future source of mass employment? Well, consider again that as employment increases, so does output, which will push down oil and gas prices (the main driver of the job increases). This is already happening.

Let’s dive a little deeper into these industries:

Continue reading “The Wealth of Humans, Part V: The Employment Trilemma”

Urbanization and poverty reduction in Africa

Brookings had an interesting article a short while back with an interesting analysis on urbanization in Africa that’s forced me to rethink my priors on urbanization as a tool of poverty reduction.

It seems that there are important differences in the reasons for urbanization in the developed world vs much of Africa. While urbanization has a been a driver of enormous creativity and growth in the Western world, Africa has largely not seen the push/pull dynamic of advances in agricultural productivity (that would allow people to leave rural areas for the city) and industrial productivity (that would lure people to the city in the first place) that was the case here.

If these forces are absent, what is driving African urbanization? The authors point to three strands of research. One is that urbanization is driven by natural resource extraction, the exports of which create wealth. Remi Jedwab at George Washington University has a paper that suggests that in areas that derive their income from the exports of natural resources, the people tend to desire urban amenities with their newfound cash. What are these urban amenities? Well, they tend to be in the non-tradables sector; that is, services which can’t simply be bought from abroad. So this creates a labor pull, but without the commensurate increase in productivity. Instead, basically all economic activity is sourced from the rents from resource extraction. This leads to low productivity, service sector oriented cities, which Jedwab dubs ‘consumption cities’. (The whole paper is worth a read, it serves as a pretty good overview of this topic).

Another strand posits that politics is a driver of urbanization. Apparently, some research suggests that the capitals of autocratic governments are larger than they ‘should’ be, and this makes sense. In many of these countries, the formal sector is severely constrained or practically absent, and a government job is the closest guarantee to success there is. So, people will try to relocate to the source of political power in the hopes of making connections.

The final strand discussed is improvements in health. There seems to be evidence that, contra the history of industrialization in the West, new urban areas have far better health outcomes than rural areas, and the effect is greater the larger the city. Indeed, even growing up in the slums of a large urban area is still healthier than being raised in a rural area, which is pretty crazy. We could imagine that if people are aware of the health related possibilities in urban areas they would relocate even absent any other incentives.

The message here is that the reasons for urbanization matter, and it isn’t a given that the good things associated with urbanization in the developed world will follow just by throwing a bunch of people together in close proximity.

The Wealth of Humans, Part IV: Social Capital

The second meaning of the wealth of humans refers to social capital. What is that, exactly?

In the most basic economic terms, capital is a factor of production. It is a black box that takes labor as an input and creates value without being consumed in the process. Factories, trucks, computers, even money itself are examples of capital. In recent years the concept of human capital has become popular. We can think of our individual skills, knowledge, and talents as factors of production: those with a greater stock of human capital are able to produce more in certain sectors of the economy. Human capital can be increased through education or training.

The difference between classic capital and human capital is that human capital cannot be transferred: you can transfer ownership of a factory, but you can’t sell your degree to someone else. It is individual specific and context independent: when you’ve learned calculus, you can apply it to any context.

Social capital, on the other hand, is context dependent. It can be thought of as all of the embedded norms and behaviors at some level of group organization. The example Avent uses is his place of employment, The Economist. Having been there for some time, he is firmly invested in the social capital of The Economist. But if he were to jump ship to another newspaper, say the New York Times, that social capital would be lost. He would still have his accumulated human capital, but the types of behaviors that led to success at and for The Economist would likely be very different for the Times.

So from here, Avent makes the supposition that the most successful firms today are those that have high amounts of social capital. He points out that, in the 1970s, over 80% of the value of average firm on the S & P 500 was physical capital, buildings, personnel, etc; the rest was ‘dark matter’.  Nowadays, this dark matter makes up 80% of the value of the average firm, and he proposes that social capital makes up a fair amount of this.

Avent encourages us to think of firms as information processing organisms. Let’s take The Economist as an example again. The real product of The Economist is not the physical newspaper delivered to your door every week, it’s the information within. The workers at the firm have taken some subset of all the information in the world and processed and compressed it into a comprehensible form. This applies to all sorts of firms: even a retail chain is an information processing organism, just not a very high added value one.

Putting these ideas together, Avent writes “The social capital of successful firms is increasingly the most important component of their success: the shared understanding of how the firm does what it does is more valuable than the machines it uses or the patents it holds.” (p.119) If we look at the top firms by market capitalization, they are overwhelmingly companies that deal with information: Apple, Alphabet (formerly Google), Microsoft, Amazon, and Facebook are hugely successful technology firms, and most of the rest of the top 20 are in industries like telecommunications or financials. With this social capital framework, we should understand that the culture of these companies is as vital to their success as their accumulated human capital. You could have a massive stock of human capital available to your firm, but if you are organized in a way that stymies the development of social capital within the firm you won’t get very far.

Now here is the problem with social capital: how can workers reap the gains from it? Normally, we think of workers as having bargaining power through the threat of taking their labor and human capital elsewhere. But because social capital is context dependent and relies on some critical mass of people to be effective, no one individual can bargain with it. Thus, the returns to social capital will tend to go to the problem the owners of classic capital, such as those people with equity in the firm. So we have a growing situation where despite creating large amounts of value for a firm, workers may not able to command a fairer share of the gains. Imagine if we move to an economy where automation has resulted in the most basic tasks being completed by computers, and the real value added activities occur through social capital. This is potentially a world where workers have very little bargaining power.

While I like this framework, to be honest I think Avent oversells social capital a bit in the book. There’s a strain of logical positivism in me that’s wary of concepts that can’t be quantified, and at points social capital becomes kind of a hand wavy argument for explaining a great many things. Still, it’s a good starting point.

The Wealth of Humans, Part III: An Abundance of Labor

In what ways does technological progress lead to a ‘wealth of humans’, i.e. a surplus of labor? Avent identifies three. The first is automation: Robots and software have enabled some tasks to be completed without the aid of any humans, eliminating jobs. Think warehouse robots, bookkeeping programs, and soon to come, driverless cars. The second, related, way is that technology significantly augments the productivity of the highest performing workers, allowing them to perform the duties of several lower productivity workers, again reducing the demand for jobs by employers. Finally, globalization has added hundreds of millions of workers (if not billions) in the developing world to global supply chains. One need only consider the aftermath of the China Shock to get a sense of the impact.

In simple supply and demand terms, the labor supply curve has shifted out and the labor demand curve has shifted in, which reduces equilibrium wages and employment. But how does this explain increasing inequality of income? Well, we have to think of the labor market not as one giant pool, but several segmented markets that are difficult to move across. So in the market for low skill labor, supply has greatly increased and demand has fallen. But in certain professions (programmers, for example), demand has increased greatly and supply has not yet adjusted.

If this is the case, then one of the best ways to deal with the problem is through education. When workers raise their skill levels, they are able to move across segmented labor markets, shifting supply to where it is needed (computer science) and reducing it where it isn’t (manufacturing).

Indeed, in Avent’s telling, this is exactly what happened during the Industrial Revolution. Despite being considered low skill work by today’s standards, at the time factory work required a certain set of skills that weren’t in common supply. Apparently, the modern education system sprang forth precisely to ensure the populace would be prepared for the mass employment of the day. Education was a way to transfer out of the agricultural labor market and into a potentially more lucrative manufacturing job. And for several decades, education has served this purpose well as the economy has evolved.

But nowadays, according to Avent, the option to educate yourself into a higher labor market segment is running out of steam. Rates of educational attainment have started to stabilize in many advanced countries, while the wage premium to a college or advanced degree continues to rise. Avent supposes that we can only reasonably expect a certain level of educational attainment in a society. And if so, then as the required skill level of the most advanced jobs increases beyond what most people can achieve, fewer and fewer workers will be able to perform them, driving up wages for the lucky few and down for the rest.

What would lead education to basically stop working? I can think of a few reasons, all speculative. One could be that there are still gains to be made from education, but the way our system is set up is not setting people up for the types of jobs brought about by the Digital Revolution. As mentioned above, our education system was largely brought about during the Industrial era, and is in many ways set up for certain types of work: work that mostly involves showing up and following instructions. So, if the high value jobs of the future involve different types of skills, a rethinking of education could help shift people into those labor markets and reduce inequality.

Another could be that our education system works fine and we’ve just reached a hard limit of human ability. If an education can only supplement our natural cognitive ability so much, then as the technology frontier shifts out only a small set of gifted workers will actually be able to perform work with any substantive value add.


But maybe we should take a step back and think about how much of a problem an abundance of labor really is. In big picture terms, we are now able to create much more output with much less labor input. In theory, this should be a great thing; it means we’re able to provide for the whole of humanity with less labor. If we have a vision of a utopian future where work is no longer required, this seems to indicate we are on the path. Of course, these predictions have been made before. In 1930 John Maynard Keynes famously thought that by our time, productivity growth would allow the average person to work on the order of a dozen hours a week. Obviously, that hasn’t happened. Why?

Many people have attempted to answer this question (for some examples, see this Tyler Cowen lecture or this David Graeber article). Avent stresses the social and political elements. Yes, we have the technical means to drastically reduce work, but not the will. There are two main challenges. One is that work is a fundamentally social enterprise: it’s how we structure our lives, how many of us find meaning, and perhaps most importantly, how we’ve decided to allocate purchasing power across society. This leads to the second challenge: convincing those who are currently thriving in the digital economy to accept a massive redistribution of income. As happened during the Industrial Revolution, we may now be in the beginning stages of a “political battle over the spoils of economic growth”.


A Brief Digression: Employment and Output by Sector

I’d like to take a brief detour into some of the numbers around how employment and output have changed in the US over the past decade. This should provide some more concrete perspective on both meanings of ‘The Wealth of Humans’. All the following data from the the BLS.

The BLS uses the NAICS, which is hierarchical. To my limited knowledge, there’s no easy way to get R to play nicely with data coded in this manner, so what follows is intended to be more of a guide rather than anything definitive. Additionally, when I refer to productivity, I’ll be using the concept of output per worker, rather than the more traditional definition of output per labor hour. The reason is that the data I’m using doesn’t have information on hours worked, though apparently some people do prefer this conception. If there are major differences in average hours worked across sectors, this will lead to divergences between the two concepts of productivity. For now, this will have to do.

So to begin, I’ve plotted the compounded annual growth rate in output versus the compounded annual growth rate in employment for the major industry categories, sized by productivity. The dotted line is the 45 degree line, which signifies where output and employment growth have been equal. Industries above the 45 degree line have seen output grow faster than employment (high productivity growth), and those below have seen employment grow faster than productivity (low productivity growth). Using Cartesian notation, quadrant I has increasing output with increasing employment; II has decreasing employment and increasing output; III has decreasing employment and decreasing output; and IV has increasing employment and decreasing output. From an efficiency standpoint, quadrant IV is the worst place to be: jobs are being added that produce less output than before.


Next is the same graph, but broken down into subcategories for each industry.  It’s too messy to add all the labels for each industry, so it will be broken down subsequently.



The ideal world is one with big light blue bubbles in quadrant 1 above the diagonal:growing mass employment in highly productive sectors. The reality is one where employment is mostly growing in low-middling productivity sectors – like health care and retail. And the highest productivity sectors – manufacturing and information – have mostly declining workforces, despite increases in output. I’m not sure what it means, but it’s interesting that real estate has pretty much constant returns across the board despite large variations in employment change.

(As an aside, this graph also serves as a good illustration of what I see to be a problem with the NAICS system. Despite being updated every five years, it still seems clearly designed for another time: manufacturing has highly detailed subcategories, whereas information has only a few, despite being of increasing importance.)

Here are looks at each quadrant (with manufacturing removed,  it gets too messy and that picture is pretty clear from above):









The Wealth of Humans, Part II: Some Stylized Facts, and Technological Progress

Let’s start with some stylized facts about the economy, in particular some troubling trends over the past several decades that have become apparent in labor markets (the examples here are for the US, though broadly similar patterns persist across most advanced economies).

For one, median wage growth has been low:


And in real terms, median household income has still not reached the level it was at even two recessions ago:


Additionally, productivity growth has been low (and decreasing since 2000 excepting a moderate burst during the Great Recession):


Finally, the labor share of the economy has been decreasing, particularly since about 2000:



All of these suggest that there is a glut of labor available. In supply and demand terms, the idea is that the labor supply curve has shifted out, driving down wages.

So what’s behind this? Economists generally agree it has something to do with technological progress: either there is too little, or too much.

Probably the most prominent proponent of the ‘too little’ hypothesis is Robert Gordon, who argues as much in his recent book, The Rise and Fall of American Growth. The basic story is that for virtually all of human history, the modal rate of economic growth has been essentially zero. The span of time extending from the Industrial Revolution to the end of the postwar era, which saw remarkable and sustained technological and economic growth, was actually highly unusual. The rather placid growth of the past 60 years is merely a return to normal. Most of the technological advances during this period, steam power, electricity, flight, and so on, led to completely transformed lives.

If you think of the today’s average home kitchen, and imagine someone from 50 years ago being transported into it, they wouldn’t have much of a problem finding their way around. Now imagine someone from 1916 being transported into the average 1966 home kitchen. The refrigerator, microwave, dishwasher, would all be totally fantastical. Even running water may be a surprise. What do we have today that could similarly wow our unwitting time traveler?

The obvious answers are computers and the internet, but the techno-pessimists have a rejoinder to this: “You can see the computer age everywhere but in the productivity statistics”, to quote Robert Solow. Here’s another look at the productivity numbers:


We can see that other than a short lived burst in the early 2000s, this measure of productivity seems to have shifted to a lower growth regime sometime in the 70s. So the argument is that, sure, smartphones are nice, but they don’t really contribute to economic growth. Apparently Robert Gordon liked to pose the following question: given the choice between a life with all the available technology pre-2000, or a life with all present day technology except indoor plumbing, which would you choose?

The answer was perhaps once obvious, but with each passing year it becomes less and less clear. This is why Avent is on the side of too much technological growth being the cause of today’s problems. He’s more convinced by the arguments put forward in works like The Second Machine Age: Computing is itself a new general purpose technology like steam power, not just a fancy dishwasher, and will transform our lives like electricity did.

So why the low productivity? Well, many of the new technologies of the Industrial Revolution took several decades to show an effect as well. It takes time, not just for the technology to be adopted, but for us to change our lives to take advantage of the new possibilities.

Rather than being mired in a lengthy stagnation, we are instead on the precipice of a new Digital Revolution. And much like the Industrial Revolution, our lives and social structures will have to adapt to accommodate the changes it will bring.