The Wealth of Humans, Part IV: Social Capital

<Part I, Part II, Part III>

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

<Part I, Part II>

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

<Part I>

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.

The Wealth of Humans, Part I

One of my most anticipated books of the year has arrived: The Wealth of Humans, by Ryan Avent. Avent is a writer and editor at The Economist, and author of The Gated City, a Kindle Single from a few years back that is absolutely worth reading. The stated mission of The Wealth of Humans is “to examine the challenge of ordering our lives and our labour in a world of technological abundance”.

Besides being a nod to Adam Smith’s foundational work, The Wealth of Nations, The Wealth of Humans has a double meaning. In one sense, it is about the wealth humans possess in the form of social capital, which is a key concept in the book. In another sense, it refers to the overabundance of human labor in a world of increasing automation.

From the outset, I like Avent’s approach. While I’ve read a fair amount on the future of work, Avent is the first to challenge me to think about work as a fundamentally social enterprise (rather than as an economic given), and the challenges of reorganizing society in the face of the (potential) end of work. In his words, “the promise of the digital revolution is an end to work…Creating mass digital prosperity is not about building institutions which ensure all workers benefit from economic growth; it is about building institutions which provide for people who do not work because their work is not necessary to generate economic growth” (pp.23-24).

At the moment I’m about halfway through the book, and will go into further detail once I’m finished. In the meantime, here is a lecture Avent recently gave at the London School of Economics, which is a nice quick summary of the ideas in the book (though I still highly recommend reading it!):

Political Order and Political Decay, Part I.5

Before getting into the examples of political development Fukuyama goes through in the book, I thought it might be useful to quickly clarify what a strong state is not.

A strong state is not necessarily a large state: there is an important distinction. A strong state has high quality of governance, whereas the size of a state refers to the scope of its functions. The ideal scope of the state depends on the society; a strong state is always preferred to a weak state. To illustrate the point that quality of government is far more important than scope, below is a graph of Govt Revenue as % of GDP (as a proxy for the size of the state) vs Govt Effectiveness (as measured by the World Bank’s World Governance Indicators), with the size of text scaled to the country’s GDP per capita.


Note that while rich states can exist across a span of government sizes, no country can be rich with poor governance. Any outliers, such as Kuwait and Qatar, are of course petro states .

To further illustrate the point, below is a graph of Government Effectiveness vs log GDP per capita that demonstrates the very strong correlation between the two.

This does leave open the possibility of causation running in the other direction. Perhaps it’s that only rich countries can afford good governance?  While I can imagine that causality can run both ways to a degree, I think again the case of petro states indicates that wealth is a not a free ticket to good governance.

Political Order and Political Decay, Part I

I’m currently making my way through Francis Fukuyama’s Political Order and Political Decay, the second half of the project that began with his earlier book, The Origins of Political Order. I’ll be using this space to work through my thoughts on the book, try and synthesize things, etc.


So to begin (as does the author), a short recap of the arguments from that first book, which addresses the question of, well, the Origins of Political Order. Namely, how did human beings ‘get to Denmark’ (Denmark being a stand-in for a “prosperous, democratic, secure, and well governed” state [p. 25]) from the natural state of decentralized bands and tribes? Fukuyama argues that the modern liberal democracy has three constituent political institutions that developed in different times and places throughout history, which are: the state, the rule of law, and democratic accountability.


(As a quick aside, what I especially appreciate about his approach, as a holder of a degree in evolutionary biology, is that he looks at human organization through an animal behavior lens. Political institutions of all stripes have to deal with the natural human instincts towards kin selection and reciprocal altruism, and indeed the biggest struggle in the development of the modern state seems to be this need to overcome our biological impulses.)


The first of these three to develop was the state, defined as that body with a legitimate monopoly on coercion over a defined territory. These earliest states, the ones we all learn about during the history of civilization, (the Egyptians, Mesopotamians, Chinese, etc) were patrimonial in nature: the state apparatus was composed of networks of family and friends of the rulers. The first modern style state, in which there is a bureaucracy selected by merit rather than connections with the ruler, and a citizen’s relationship with the state is impersonal, developed in China amidst a backdrop of constant warfare. In such survival conditions, the most efficient and effective state apparatus will outcompete the others.


The second institution to develop was the rule of law, which has its origins in religion. The idea of law as separate from the ruler developed in Europe in the wake of the collapse of the Roman Empire; as the Roman state receded, the religious hierarchy remained in place and led to constraints on the actions of subsequent political rulers. This was probably the most fascinating part of the first book for me. If you’re at all familiar with the basics of Catholic theology, it has a very legal character to it. In part, I suppose, because that’s where the law came from.


The final institution to develop was democratic accountability. In Europe, this originated in the power struggle between monarchs and the upper nobility and landed gentry. As the monarchs of these nascent states tried to centralize a scattered system in which the nobility owned land, and thus the livelihoods of the laborers in these agricultural societies. For example, to raise taxes on the populace, kings would have to first go through the nobility, which often objected. In response, the monarch would often then try to skip the nobility appeal directly to the masses In places where the nobility and was evenly matched, like England, this resulted in a shared balance of power between the monarch and the elites that became Parliament. Okay, but how does become democratic accountability? After all, early suffrage was limited only to the upper class. In Fukuyama’s telling, the rhetoric that was used by Enlightenment thinkers in justifying these systems, particularly in the American Revolution, took on a life of its own and led to what were perhaps unintended consequences.


What I mean is, one can certainly make the case that for a large number of participants in the American Revolution, while the rhetoric of ‘all men are created equal with inalienable rights’ was certainly useful, they probably didn’t actually believe it; even ignoring the obvious instance of slavery, the Founders clearly designed a political system around elites making the decisions. Some were downright horrified by the prospects of ‘universal’ (white male) suffrage. And yet, within just a generation or two, that’s precisely what happened during the Jacksonian era in large part due to these ideas being baked into the American project from the beginning.


Which brings me to what I take to be one of Fukuyama’s main points in this work: ideas have power. The history of political development is not simply a tale of materialist consequences of environment, geography, resources, etc. Ideas themselves are major players in this history. But ideas are also highly contingent. To go back to the example of rule of law: this was able to happen in Europe because Christianity happened to be a transcendental religion. It potentially could have been the case that hundreds of years earlier,if theologians had decided differently on seemingly obscure matters of theology, the rule of law would not have developed as it did.


So having set the stage, the focus of the book is on political development since the Industrial Revolution. In the next part, I’ll summarize his four particular examples of paths political development can take: Prussia, Greece, Italy, and the United States.