Inequality, Bargaining Power, and Startups
Earlier this year, Paul Graham wrote an apologia for the inequality produced by startups:
The reason he [Mark Zuckerberg] and most 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….Yes, there are a lot of people who get rich through rent-seeking of various forms, and a lot who get rich by playing games that though not crooked are zero-sum. But there are also a significant number who get rich by creating wealth.
The crucial assumption behind his defense is that startup founders like Mark Zuckerberg acquired their riches by creating wealth, not by winning zero-sum games. But to what extent is this actually true?
My answer: this is very messy. In reality, highly profitable businesses result from some degree of genuine wealth creation and some degree of capturing critical nodes in the web of market exchanges. Ownership of these critical nodes allows the business to extract lucrative rents.
To understand the messiness, let us use an example. Imagine that a rapid succession of technological advances has just made canal building much more viable. There exist certain geographic pathways which are optimal locations at which to build a heavily-traveled and very lucrative canal.
Now imagine five different scenarios:
First scenario: Imagine a landowner already owns all the land on a certain isthmus that is the prime location for a great canal. This landowner sells the land for a windfall of a billion dollars.
Second scenario: In this case a savvy real estate speculator is the first to see the opportunity. He buys up all the land in a critical isthmus from a collection of naive farmers for pennies on the dollar. Then he flips this land to a canal construction company for a billion dollars in profit.
Third scenario: A venture capitalist seizes the opportunity. He forms a canal company. He buys up the prime land at the critical isthmus, hires a bunch of laborers and engineers. He buys a bunch of machines. He organizes the digging of the canal. The total cost of building the canal is $250 million. The canal is a big success, customers love it. The canal is also wildly profitable. The company goes public for a market cap of $1.25 billion dollars, netting the capitalist $1 billion in pure profit.
Fourth scenario: In this scenario, there are several great paths for building a canal, and three different companies build a canal. Prices fall as the three compete each other. But one of the canals is run by an exceptional entrepreneur. He works 70 hour weeks making his company awesome. He is fantastic at managing an engineering team, maximizing efficiency, motivating workers, solving problems, and holding people accountable. Because of this, his canal never has breakdowns, and is extremely efficient. Ships choose to pay a substantial premium to use his canal, because it is so efficient. As a result, while the other two owners barely break even after expenses, his canal company is worth a billion dollars.
Fifth scenario: Same as scenario #4, except the entrepreneur executes so well, he drives the other canals out of business. Once they are out of business, he can charge monopoly prices. His profits triple.
It is clear that the first two capitalists have created no wealth. They just won a zero-sum game. Capitalist #4 has a much greater claim on being a true creator of wealth – his wealth comes from providing better service to his customers. But with capitalist #3 and #5, the situation gets very complicated. These guys both created wealth, but also won a zero-sum game.
Entrepreneurs such as Zuckerberg or Larry Page are like entrepreneur #5. They won because they out-executed their competition, and in doing so, created great value for consumers. Yet, in winning, they established ownership of a crucial node on the internet. They both own a near monopoly, and in doing so, are able to extract profits disproportionate to the amount of investment into their products. Are most of their riches due to out-executing and providing value? Or due to capturing this monopoly opportunity? It is impossible to say.
A lot of this debate comes down to either side taking an absolutist view. Silicon Valley’s detractors claim that most startups are much closer to the speculator of scenario #2. Detractors accuse founders and VC’s of being well-connected opportunists whose only skill is in the quickest to pounce on and capture monopolies. Paul Graham seems to view startups as being much closer #4 – as pure engines of wealth creation. But the truth is just that it is messy. Most companies are akin to scenario #3 and #5. Their profits are a very messy mix between earning returns on speculation and earning returns on superior execution.
The messiness also exists when we look at venture capital firms, not just individual startups:
Scenario one: A venture firm funds dozens of companies. Most fail. Perhaps the market opportunity just wasn’t right, perhaps the technology could never be perfected. But a few are wildly successful and earn massive profits. On net, the venture firm earns a modest return, just beating the S&P 500.
Note, that even though the individual success of one investment might be an example of winning a zero-sum monopoly, the venture firm as a whole is not making rent-seeking profits. If you taxed away the great profits from any individual investment, the venture firm would be net unprofitable, investment would dry up, and society might not get any more ground-breaking products.
Scenario two: Imagine a second venture firm. This firm got really lucky with a big bet early on. As a result, it gained fame and prestige. Now the best entrepreneurs seek out the firm first, so they can brag, “We’re funded by the same guys that funded Google.” The firm exploits this to build a network of successful companies. As a result of their brand and their network, the firm can give a 10% better chance to any company that chooses it over a competitor. Thus, any entrepreneur would make this firm their first choice. Thus, whenever there is a new opportunity for a monopoly, based on new technology or a new platform, this VC firm always gets to fund the best team that is attacking that space. As a result, the VC firm takes very total little risk, and is wildly profitable. The firm can simply coast on the existence of its brand and network, perpetuating great profits simply because it was profitable in the past.
Which model of venture funding is closer to the truth? Again, in reality, the truth is somewhere in the middle. Successful venture capital firms make money in part because of skill in allocating capital, and in part because of risk. But successful firms also have established market positions where they have a pipeline to exploit new market opportunities before anyone else.
Let us switch gears to view inequality from a philosophical basis. What is a “just” distribution of wealth?
Thinkers on the left have posed a number of different models. Marx believed in the labor theory of value and any profits beyond labor were unjust. Rawls believed that society should have a distribution scheme that would be chosen if we were to be born randomly.
In my opinion, these views are not helpful. They fail to incorporate the fact that some people are more skilled than others, that becoming skilled requires hard work to develop talent, that people take different risks, and that such hard work and risk taking has to be rewarded or else people will not do it.
On the other side, there is the conservative and libertarian view is that any outcome produced by voluntary transactions is fair. Wikipedia summarizes an argument made by libertarian Robert Nozick:
Wilt Chamberlain is an extremely popular basketball player in this society, and Nozick further assumes 1 million people are willing to freely give Chamberlain 25 cents each to watch him play basketball over the course of a season (we assume no other transactions occur). Chamberlain now has $250,000, a much larger sum than any of the other people in the society. This new distribution in society, call it D2, obviously is no longer ordered by our favored pattern that ordered D1. However Nozick argues that D2 is just. For if each agent freely exchanges some of his D1 share with the basketball player and D1 was a just distribution (we know D1 was just, because it was ordered according to your favorite patterned principle of distribution), how can D2 fail to be a just distribution? Thus Nozick argues that what the Wilt Chamberlain example shows is that no patterned principle of just distribution will be compatible with liberty. In order to preserve the pattern, which arranged D1, the state will have to continually interfere with people’s ability to freely exchange their D1 shares, for any exchange of D1 shares explicitly involves violating the pattern that originally ordered it.
But there is a problem with this theory. Basketball players and basketball team owners make a lot of money because they bargain collectively. Team owners negotiate TV contracts and other contracts as a league. Basketball players join a union to collectively negotiate a share of the league revenue. So if the owners and the players collectively bargain, why cannot consumers collectively bargain? Why cannot consumers bargain collectively for a revenue share?
Laborers bargaining have a special problem – the employer can hire scabs from outside the union. Thus to bargain effectively, a union would want to expand, and to encompase all laborers. When negotiating contract, the union would furthermore want to prevent the company from hiring any non-union labor, in order to prevent the union’s bargaining power from being undermined.
Furthermore the union needs to be coercive. The famous game theorist, Thomas Schelling, pointed out that it can be to one’s benefit to submit to coercion as part of a bargaining process. If all the union members vote to pass an edict that says that labor union muscle will break the knees of any defector who crosses the picket line, there will be no defectors, the union’s bargaining position is better, and the union is more likely to win a pay raise for all workers. Collective bargaining requires punishing defectors. The conservative argument that union coercion is anti-worker, is a bogus, concern troll argument that ignores game theory.
Thus for all workers in general to bargain effectively, you need a coercive union, encompassing the entire labor force, that demands a bigger share of total revenue – that sounds sort of like a democratic government instituting progressive taxation and redistributing money to the workers.
And of course, with collective bargaining, there is no hard-and-fast rule to determine what is a fair split. Hold out for too much and you might destroy the entire business. Surrender too early and you leave money on the table. And making sure the union bosses actual bargain in your interests is tough. For instance, arguably in the 1970s British labor acted as a giant union that bargained for too much, and almost brought the entire country to economic collapse.
So much for figuring out a “just” distribution from first principles.
So let us ditch abstract theory and look at this from the standpoint of naked self interest. Let us imagine the United States government as a giant union. How much should the people bargain for? What share of the national revenue should it tax and distribute?
Right now, in the United States, for the most successful startup founders, ~50% of their profits will end up in the hands of a government. This includes the corporate income tax, capital gains taxes, state income tax, and the Medicare millionaires tax. (There are many complaints that millionaires are under-taxed because capital gains taxes are lower than income taxes. But this ignores that capital gains are on top of corporate income tax, when you add two together, the total tax rate is higher than the highest personal income tax rate.)1
If we raised the tax burden to be 100% on all income over a million dollars, then I think it is pretty obvious that the entire start up world would be annihilated. Even if the income limit was $10 million, startups would likely cease to be a thing. Angel Investors would have far less reason to invest, since it would be 9 out of 10 chance they lose money, and in the 1 in 10 chance they win big, their win is capped by taxes. Entrepreneurs would always want to sell out early to big companies rather than try to actually win a market, since they would gain no more by trying to grow huge. Big companies would just buy off dangerous startups, and their products would die. Thus startups would be a thing of the past. In such a world, products created by startups, ranging from Dropbox to the Mac to Google, likely would never have happened. We would all be running crappy IBM computers, and every upstart would be gobbled up, paid off, and shut down.
So if logically we know that at 100% income tax there would be very few startups, we know that for every increment between then we’ll lose some portion of the startups. The rate of decreasing startups with money might not be linear, but for lack of any better guesstimate, we can model it that way. If you raised taxes on income over a million dollars by another 10%, that would net around $75 billion dollars. 2 Would you trade killing 20% of all future startups for the government having an extra 1.3% in total government revenue? Would you give up out of every five products you use, perhaps your Macbook, perhaps Dropbox, perhaps your Zipcar – so that the government would have 1.3% more funds?
Now as a startup entrepreneur I am completely biased about this issue. But still – I don’t think think this seems like a good trade.
Of course there could be more creative ways to tax inequality. Perhaps there could be a millionaires tax only on zero-sum professions – such as lawyers, professional athletes, and certain branches of finance. And absolutely there should be an elimination of all the shady means by which Wall Street plays “heads we win, tails the public bails us out.” But taxing startups does not seem wise to me.
Of course there is a final problem: the United States Government is neither a giant consumers union nor a giant labor union. It pretends to be a consumers union, but it is not. It is a leviathan, a hydra, a massive organism, largely unaccountable, unmanageable, driven hither and tither by a thousand different political factions. It has thousands of factions, all trying to gorge off its teat. If an additional $75 billion was going in a universal basic income check to all Americans, that might be useful. If it is going to be filtered into housing projects that wreck entire neighborhoods, or more subsidies for bloated college administration, then that is much less appealing.
Paul Graham noted in the original version of his essay:
So when I hear people saying that economic inequality is bad and should be eliminated, I feel rather like a wild animal overhearing a conversation between hunters. But the thing that strikes me most about the conversations I overhear is how confused they are. They don’t even seem clear whether they want to kill me or not.
Commenters called out this paragraph for attacking a “straw man.” Paul Graham has now rewritten the essay without this paragraph. But I think his original quote was exactly right. The conversation is in fact confused, and one of the most terrifying aspects is that there is no organized body with agency that has a specific set of demands that can be reasoned with and bargained with.
What is really happening in this debate is that the political left comprises a thousand different factions, each with their own pet issue. Some want student loan debt relief. Some want universal basic income. Some want a $15 minimum wage. Some want more money for schools. The Silicon Valley late-comers, who face long hours and absurd housing prices, envy the Silicon Valley winners and royalty, like Paul Graham, who have lucrative, entrenched positions. The only point these disparate groups can all agree on is to attack inequality. Attacking inequality is the lowest common denominator in left-wing politics, and perhaps any politics. So when anyone from these disparate groups denounces inequality, they get re-tweeted and re-blogged by the entire complex. That is also why the discussion is confused and why it is not clear what they actually want. There is no single, organized movement making these demands. And that is unfortunate, because instead of getting reasonable compromises that could make life better for everyone, we’ll probably get more messy kludges that distort incentives and that favor expanded bureaucracy over helping working folk.
There are also many complaints that corporations do not really pay the face value of the tax rate. But often it is really that the profits are fake paper profits, or that the money is being sequestered away and will be taxed when it is actually distributed to shareholders. If you divide total corporate profits by total cash returned to shareholders, the amount of underpaying is not really that large.↩
Total income from taxpayers making over a million was 1.1 billion dollars. Subtract the first million out of this for a total of around $750 billion -- source. In reality, the returns might not increase that much, as people might respond by spreading capital gains over multiple years instead.↩