Paul Graham (left) and Tim O’Reilly (right) in a virtual discussion

What Paul Graham Is Missing About Inequality

Before he published his essay on inequality, Paul Graham sent it to me for comment. I sent him some quick impressions and promised a more detailed critique later. I didn’t provide those more detailed comments before Paul published his piece, which I regret. It is good that Paul is wrestling with the question of income inequality, as Silicon Valley as a whole should be.

I think Paul’s analysis is wrong on several levels, though, and it’s important to understand why. The growth of income inequality is such an important conversation to frame properly! We have to understand what’s wrong with the world as it is, because only then can we envision the world we want to create, and think about how to get there.

There are many people who blame technology for increasing income inequality. Paul’s defense of technology (which he frames as “startups”) is that it creates wealth for all, even as (he claims) it creates inequality. I agree with him that technology can make us all richer, but I disagree that it necessarily creates greater inequality, even if some startup founders become very rich. It only does that if companies don’t create real value in return for that wealth.

In addition, I think we need to take a closer look at how one of Silicon Valley’s most treasured tools for creating wealth for employees — the stock option — has played an unexpected role in increasing income inequality.

On a long flight back from Europe, I decided to intersperse more detailed comments into Paul’s essay, as I might have sent detailed feedback and suggested emendations to an author writing a book for O’Reilly Media. I realize it’s an unusual way to respond — to actually take someone else’s words and extend them with my own, but it seemed preferable to the usual quote and respond style.

Paul, I hope you take no offense, and consider this a useful exercise in public discussion. We should get on stage together somewhere and have the discussion in real time!

In the text below, Paul’s original words are in normal type, and my suggested changes and additions are in italic, and sometimes, for additional clarity, with editorial asides in brackets.

January 2016

Since the 1970s, economic inequality in the US has increased dramatically. And in particular, the rich have gotten a lot richer. Nearly everyone who writes about it says that [too much] economic inequality is bad, and should be decreased. [Note: suggest changing this to read “too much economic inequality” because there is no serious economist arguing that there should be no economic inequality. That idea was discredited within a few years of the Russian Revolution!]

I’m interested in this topic because I was one of the founders of a company called Y Combinator that helps people start startups. Almost by definition, if a startup succeeds its founders become rich. [I suggest inserting:You might think this”] means that by helping startup founders I’ve been helping to increase economic inequality. If economic inequality is bad and should be decreased, I shouldn’t be helping founders. No one should be. [Suggest adding:] But that would only be true if those startups extracted more value for themselves than they created for society as a whole. In his book, Capital in the 21st Century, which brought the topic of income inequality to the center stage, even Thomas Piketty argues that increased productivity and better diffusion of knowledge create more wealth for society and are among the forces that reduce income inequality.

Many startups do produce enormous new value for society, and the best of them create far more value than they capture, even as they disrupt or even destroy older, less efficient industries.

[In any event, Paul, it seems that you are conflating the idea of some people being rich while others are poor with the worry about increasing income inequality, which is that, on average, one group of people is becoming significantly richer, while another is becoming significantly poorer. It’s a straw man to pretend that the argument is simply that some people are too rich.]

So have we just shown, by reductio ad absurdum, that it’s false that economic inequality is bad and should be decreased? That doesn’t sound right either. How can economic inequality not be bad? Surely it’s bad that some people are born practically locked into poverty, while at the other extreme fund managers exploit loopholes to cut their income taxes in half.

[You’re absolutely right that the growth of the financial industry is central to the inequality discussion. But it isn’t just a matter of hedge fund managers or loopholes. Financial markets have been extracting a larger and larger slice of the entire economy. You should read this post about a presentation by economist Alan Blinder: How the Financial Sector Consumed America’s Economic Growth. Around the turn of the century, financial markets provided capital to business and consumers at a cost of about 2% of the total economy. By 2013, that cost was up to 9%! (By contrast, the entire internet sector is about 5% of GDP!)

While markets play an important role in financing innovation — and even speculative bubbles are part of the innovation process — the size of the financial sector has increased at the same time as the role it plays in financing innovation and productive investment has decreased! And as I’ll outline below, I think you have to ask yourself how much this “financialization of the economy” is also a major contributor to the Silicon Valley wealth that you celebrate in this piece.]

The solution to this puzzle is to realize that economic inequality is not just one thing. It consists of some things that are very bad, like kids with no chance of reaching their potential, and others that are good, like Larry Page and Sergey Brin starting the company you use to find things online.

[Suggest inserting:] “Google’s founders created enormous wealth for themselves — Larry Page and Sergey Brin are each worth somewhere north of $35 billion — and through stock options distributed to every employee, they have also created wealth for everyone who has worked for Google, as well as for those who invested in the company.

But more importantly, they also have created enormous value for other businesses, and for society as a whole. Every year, Google chief economist Hal Varian and his team publish an economic impact report. They estimate that Google increased US economic activity in 2014 by $131 billion dollars. That means that value created for other businesses in 2014 was more than double Google’s own $61 billion in annual 2014 revenue. (Note that the value creation number from Varian’s study is US only, while the revenue figure is worldwide.) Given that Larry and Sergey founded Google in 1998, you can count the cumulative economic impact in the trillions of dollars. And the consumer surplus provided by free access to vast amounts of online information has to be far larger.

You can see from this example that there is no problem when a company’s founders and investors reap enormous value and become members of the very top 0.01% of the wealth distribution. As long as the startup creates more value for society than it takes out for itself, it can actually decrease inequality, rather than increasing it.

But if you want to understand economic inequality — and more importantly, if you actually want to fix the bad aspects of it — you have to tease apart the components.

[This is where things get interesting! There are many contributors to income inequality, but there’s one big one that ought to be the elephant in the room in Silicon Valley, especially in today’s overheated financing environment, a game of musical chairs where, once again, it looks like the music may be about to stop.

Here’s what I’d write in your place: ]

“Let’s start by teasing apart the value that a company like Google creates. Google’s revenues are in return for services that other businesses purchased from Google. Presumably, they paid for these services because they found them valuable. And if Varian’s analysis is correct, those services did indeed result in increased revenues for those 1.8 million Google business customers. As for the users of the Google search engine, we also participated in an exchange of real value, receiving free search services, navigation, office applications, and much more, in exchange for clicking on some of the advertisements that those paying Google customers placed via the service.

But the value that Google’s founders, investors, and employees received is of an entirely different kind. That $35 billion+ that Larry and Sergey are each worth, and the tens of billions more received by early investors and other employees are the value placed on ownership of shares in Google by financial markets. Financial markets are very different than the market of goods and services I described above in a number of ways.

The price of a stock is fundamentally a bet on the future. Will the company’s earnings from the real market of goods and services be greater in the future? If so, it is worthwhile to own a share of that company. But if they will be less in the future, so too, the company will be worth less.

In short, the wealth of Larry and Sergey is the net effect of millions of bets placed by others that the company they created will continue to create value into the future.

When a company fails to deliver value that lives up to that bet, but still cashes in through IPO or acquisition, the wealth that is gained by startup founders and early investors is taken from public market investors. This is a risk that both sides of the bet willingly take, and it has provided enormous fuel for innovation as it encourages innovators to take risks in hope of future rewards. But in over-excited markets, it’s too easy for many startups to aim to cash out with “dumb money” while the getting is good with no real plan for ever delivering real revenues or profits.

In an earlier era, Larry and Sergey would still have been quite rich, because they built a company that has produced enormous profits. As owners, they could have laid claim to a significant share of Google’s net income (nearly $14.5 billion for 2014.) In order to benefit from that wealth, they would have had to decide how much of that net income to reinvest in the business to help its continued growth, and how much to take out for themselves and their employees in the form of wages or dividends. And they would have paid taxes on those wages or dividends.

As it turns out, the value that Larry, Sergey, and other early insiders have realized from Google through financial markets roughly matches the share of Google’s profits they could have claimed as owners of a private company. But that isn’t always the case. Some companies (including many startups) have stock market valuations far in excess of the actual profits that the companies generate. That’s because financial markets have increasingly gone from being a source of capital for companies to a kind of giant betting pool, in which winning and losing is much less correlated with underlying economic activity.

This is a contributor to economic inequality that not enough people are talking about. In an economy where financial instruments are increasingly unmoored from the real market of goods and services and profits derived from those services, it’s possible for many people to reap rewards that weren’t actually earned. I’m not just talking about bubble-inflated stock-based compensation that has made many people in Silicon Valley so rich, or the excesses of Wall Street banks which nearly wrecked the economy in 2008, but the entire structure of executive compensation.

It turns out that stock options, which have paid such a large role in Silicon Valley wealth, have been misused, and have become a key part of the problem of income inequality.”

[Note: I shared some of my comments with Bill Janeway, formerly Vice Chairman of investment firm Warburg Pincus, and author of the book Doing Capitalism. He wrote:

“When [options] began to be deployed by start-ups they were tickets to a lottery in which, you are quite correct, most holders would receive nothing. Take a look at this chart from an article published in the American Economic Review by two very distinguished economists, Bob Hall and Susan Woodward: R. E. Hall and Woodward, S., “The Burden of the Nondiversifiable Risk of Entrepreneurship,” American Economic Review 100 (2010), pp. 1163–1194. Basically, it says that in 75% of VC-backed start-ups, the entrepreneur gets zero. If becoming rich in Graham’s world means making $100 million (pre-tax), then 0.4% of entrepreneurs make the grade.
But then this innovation in compensation, mobilized to lure executives out of the safe harbors of HP and IBM, was hi-jacked. The established companies began using stock options when there was essentially no risk of company failure. It reached the destructive extreme when the CEOs of banks, whose liabilities are guaranteed by taxpayers, began getting most of their compensation in options.

[So, if I were writing this in your place, I’d add something like this: “Starting in the 1970s and accelerating in the 1980s, most CEOs and other top executives began to receive the bulk of their compensation in stock options, rather than as ordinary income. And in 1993, a well-intentioned law pushed by President Clinton limited the ordinary income that could be paid to top management, with the unintended consequence that even more of the compensation moved to stock options. The intention was to align the interests of management with the interests of shareholders, but is often the case, those good intentions were derailed by poor implementation.

Congress allowed a huge loophole in the accounting treatment of stock options — unlike ordinary income paid to employees, value paid through options did not need to be charged against company earnings. It was thus a kind of “free money” for companies, invisibly paid for by dilution of public market shareholders (of whom a large percentage are pension funds and other institutional shareholders representing ordinary people) rather than out of the profits of the company. As a result, executive compensation soared, to the point that in one outrageous 1999 case, the CEO of retailer Abercrombie and Fitch received $120 million in option pay (not charged to the P&L), while in the same year, the company had only $150 million in earnings. Even now, when the rules have been tightened up, much of this stock based compensation is hard to trace in company financials.

Meanwhile, there is an incentive to cut income for ordinary workers, because that still shows as an obvious expense on the P&L. Cutting wages drives up net income and thus the price of the stock in which executives are increasingly paid.

In addition, it has become increasingly clear that there are strong incentives for financial maneuvers like stock buybacks, which too often replace productive re-investment in the underlying business

At any rate, the use of stock options and other financial instruments led to a widening gap between the pay of executives and ordinary workers. In the 1960s, CEO pay was 20x that of the average worker. Now, it is 300x that of the average worker. This is a major driver of inequality.

Silicon Valley companies are actually better than many other companies, because they offer options to virtually every employee, but even there, those options are overwhelmingly weighted towards top management, with each lower rank of workers typically receiving a full order of magnitude less in value.

The net net is that a huge proportion of the productivity gains we’ve seen in the past decades have increasingly gone to a small group of managers rather than to all workers.

In the case of companies like Walmart and Amazon, productivity gains may also be given to consumers in the form of lower prices, as a way to expand the market share of a business. Or in the case of Google and Facebook, given to consumers as free services, paid for by advertisers. To the extent that workers are also consumers, this has an offsetting effect. But as Nick Hanauer points out, in general we have forgotten the hard fought lessons of the 20th century, that workers are also customers, and that unless they receive a fair share of the proceeds, they will one day be unable to afford our products.

Even worse, when we saw that many members of society were no longer able to afford the products our companies have on offer, we encouraged workers to borrow money at high credit card interest rates so that they could maintain the illusion of middle class wages, and even encouraged them to take on student debt so that they could retrain themselves for the jobs of the future that we were busy stripping away. And of course, when the government stepped in with a safety net to support those who could no longer afford even the bare necessities of life, businesses gratefully accepted the economic boost from their spending of government benefits, but claimed that those we no longer paid enough to live without that assistance were freeloaders.”]

And yet despite all these factors, the trend in nearly everything written about the subject, [including this paper!], is to do the opposite: to squash together all the aspects of economic inequality as if it were a single phenomenon.

Sometimes this is done for ideological reasons. Sometimes it’s because the writer only has very high-level data and so draws conclusions from that, like the proverbial drunk who looks for his keys under the lamppost, instead of where he dropped them, because the light is better there. Sometimes it’s because the writer doesn’t understand critical aspects of inequality, like [suggest you insert: the way that we allocate profits from increased productivity to top managers versus ordinary workers, or the way that our tax system favors capital gains over income earned by labor.] Much of the time, perhaps most of the time, writing about economic inequality combines all three.


The most common mistake people make about economic inequality is to treat it as a single phenomenon. The most naive version of which is the one based on the pie fallacy: that the rich get rich by taking money from the poor.

Usually this is an assumption people start from rather than a conclusion they arrive at by examining the evidence. Sometimes the pie fallacy is stated explicitly:

…those at the top are grabbing an increasing fraction of the nation’s income — so much of a larger share that what’s left over for the rest is diminished…. [1]

Other times it’s more unconscious. But the unconscious form is very widespread. I think because we grow up in a world where the pie fallacy is actually true. To kids, wealth is a fixed pie that’s shared out, and if one person gets more it’s at the expense of another. It takes a conscious effort to remind oneself that the real world doesn’t work that way.

In the real world you can create wealth as well as taking it from others. A woodworker creates wealth. He makes a chair, and you willingly give him money in return for it. A high-frequency trader does not. He makes a dollar only when someone on the other end of a trade loses a dollar.

You have to ask yourself, though, whether a Silicon Valley startup is more like the woodworker who made five chairs, or more like the high frequency trader. It’s clear that many startups are like the woodworker, but in bubble times like these, there are a lot of startups that are more like the high frequency trader.

If all the rich people in a society got that way by taking wealth from the poor, then you have the degenerate case of economic inequality where the cause of poverty is the same as the cause of wealth. But instances of inequality don’t have to be instances of the degenerate case. If one woodworker makes 5 chairs and another makes none, the second woodworker will have less money, but not because anyone took anything from him.

[This analogy is way too simplistic. What’s happening is actually a combination of several factors. First, we’ve figured out how to have machines make many of the products that we used to pay people to make. And we’ve figured out how get people on the other side of the world, in countries that are much poorer than ours, to accept much lower wages and much worse working conditions as a way to increase the profits of our businesses.

Doing this has actually increased the wealth of people in those other countries — income inequality has decreased worldwide, with fewer and fewer people living in abject poverty, and hundreds of millions of others beginning to enter the middle class. But in formerly rich countries, many people who used to be paid well for their work now have to compete for lower-paid jobs, while those who already own meaningful capital take a larger and larger share of the pie.

This is the real “pie fallacy” — the idea that as long as the pie is getting bigger, everyone is better off. It’s true that through technology, trade, and the spread of knowledge, we have made a bigger pie. But that doesn’t mean that some people aren’t getting far more of the benefit, while others are losing out.

All it takes to increase income inequality is for there to be more companies practicing the “degenerate case” than practicing the virtuous kind of capitalism that creates more value than it extracts.]

Even people sophisticated enough to know about the pie fallacy are led toward it by the custom of describing economic inequality as a ratio of one quantile’s income or wealth to another’s. It’s so easy to slip from talking about income shifting from one quantile to another, as a figure of speech, that we can forget that is literally what’s happening.

Except in the degenerate case, economic inequality can’t be described by a ratio or even a curve. [This doesn’t make any sense. Of course it can be described that way. If you couldn’t successfully analyze a situation with statistics, half of the internet startups you celebrate wouldn’t exist!] In the general case it consists of multiple ways people become poor, and multiple ways people become rich. [Yes, and in aggregate, they are subject to the law of large numbers, which means that statistical analysis can be surprisingly insightful!] Which means to understand economic inequality in a country, you have to go find individual people who are poor or rich and figure out why. [2]

[I agree, but you haven’t done that in this essay. Paul, you’ve always encouraged the startups you’ve coached at Y-Combinator to focus on their users, to get out of the office and talk to people. Yet here in this piece, you assume that what’s true in Silicon Valley is true everywhere. If you’d spent as much time talking to people who work at today’s low-wage jobs, you might have a very different perspective.]

If you want to understand change in economic inequality, you should ask what those people would have done when it was different. This is one way I know the rich aren’t all getting richer simply from some new system for transferring wealth to them from everyone else. When you use the would-have method with startup founders, you find what most would have done back in 1960, when economic inequality was lower, was to join big companies or become professors. [Surely you jest. What they would have done was to start new companies. After all, when folks like Bob Noyce, Gordon Moore, Bill Hewlett and David Packard started what became Silicon Valley, it was harder to start new companies than it is today, but that didn’t stop them! Today’s Silicon Valley stands on the shoulders of giants. It’s ridiculous to claim that today’s generation is creating more value.]

Before Mark Zuckerberg started Facebook, his default expectation was that he’d end up working at Microsoft. The reason he and many [not most! most startups fail] other startup founders are richer than they would have been in the mid 20th century is not because of some right turn the country took during the Reagan administration, but because progress in technology has made it much easier to start a new company that grows fast. [I think you underestimate the increased role of financial markets in how entrepreneurs are compensated, so that even many companies without real profits can reap enormous stock gains. You have to realize that Zuck, like Larry and Sergey, is an outlier, in that he created an extraordinarily profitable business.

When a startup doesn’t have an underlying business model that will eventually produce real revenues and profits, and the only way for its founders to get rich is to sell to another company or to investors, you have to ask yourself whether that startup is really just a financial instrument, not that dissimilar to the CDOs of the 2008 financial crisis — a way of extracting value from the economy without actually creating it.]

Traditional economists seem strangely averse to studying individual humans. It seems to be a rule with them that everything has to start with statistics. So they give you very precise numbers about variation in wealth and income, then follow it with the most naive speculation about the underlying causes. [Unfortunately, this closing paragraph seems like a good summary of what you’ve done in this essay.]

Subscribe to the Next:Economy newsletter for news, ideas, and insights on how technology is transforming the nature of work, life, and society.

I call today’s economy “The WTF Economy.” It fills us with wonder, and it fills us with dismay. I’ve been working hard to understand what we need to change if we want to create a Next Economy that will preserve the wonders of innovation but also addresses the dark futures that we face unless we put people first. Let’s work together to harness the power of technology to decrease income inequality rather than increase it! — Tim O’Reilly