Solyndra to Amtrak, why government picks losers when it picks winners

Folks, I have a business proposition for you. A man with a fine suit and a stack of slides walks into your office and says he can see the future. He knows which industry will boom, which company will lead it and which technology will win. He needs only your money. You ask what he has done with other people's money before, and he says: "Oh, I have never risked any of mine. But the committee believes in me."

You would show him the door. You would show him the door quickly. Yet that is the pitch Washington makes every few years, and for some reason a large part of the country keeps buying it. The pitch is called industrial policy: the idea that government officials, using your tax dollars, are better placed than the people whose own money is at stake to decide where capital should go.

I want to give that idea its fair day in court, because its defenders have better cases than the headlines suggest. Then I want to show you why, over and over, the verdict comes back the same way.

The definition, so nobody can wriggle

Let us define the terms. Picking winners means a government choosing particular firms or technologies for special help: a loan, a guarantee, a subsidy, a purchase mandate, a tax break aimed at one thing and not another. It is different from setting general rules that apply to everybody, like contract law or a tax code that treats each firm alike. Rules are the referee. Picking winners is the referee putting on a jersey.

The claim I am making is not that every government investment loses money. Some do not. The claim has two parts, and I will argue each:

1. When government picks, it lacks the knowledge and the discipline that private investors face, so its picks fail more often.
2. When a government pick fails, the loss does not close the program, so the failure repeats.

Both halves matter. The first explains why the bets are bad. The second explains why the bad bets keep coming.

Exhibit A: Solyndra

Start with the case that became a household word. Solyndra was a California maker of solar panels with an unusual design, cylindrical tubes instead of flat panels. In 2009 the Department of Energy, under a loan guarantee program authorized by the Energy Policy Act of 2005 and expanded by the 2009 stimulus law, backed a $535 million loan guarantee for the company. The idea was that taxpayer support would help a promising American manufacturer scale up and compete.

It did not compete. The price of the competing product, panels made from conventional silicon, fell sharply as Chinese manufacturers expanded and as the cost of silicon dropped. Solyndra's design, which had looked clever when silicon was expensive, was suddenly an expensive answer to a question the market had stopped asking. On September 6, 2011, the company filed for bankruptcy. The guaranteed money was largely lost.

Now notice what happened in the weeks and months before. In 2011 the loan was restructured, and under that restructuring, as reported at the time and in later inspector general work, new private money was placed ahead of the government's claim. Put plainly, the taxpayer agreed to stand behind someone else in line. That is not how a careful lender treats its own money. A bank that did that would lose its examiners' goodwill. A government agency doing it lost nothing but public patience. (I would check the inspector general's report on the program for the specifics before you repeat that point at dinner.)

Is Solyndra fair to use? Its defenders say no, and here is the steelman. The loan guarantee program was built to take risks that private lenders would not, and a portfolio of risky bets will include losses. The program, they argue, should be judged as a whole, and across the full portfolio the losses were modest compared with the total lent. That is a respectable argument. I will answer it in a moment. For now, hold on to the point that a loss like Solyndra's was not a bug. It was the model working as designed. A program that exists to back the ventures the market refused will back some ventures the market was right to refuse.

Exhibit B: Amtrak

Now go back a generation. Passenger trains in America were losing money by the late 1960s. Private railroads had been run for decades under federal regulation that kept them running passenger service they would have dropped if allowed. Congress's answer, in the Rail Passenger Service Act of 1970, was to create a government-chartered corporation, the National Railroad Passenger Corporation, which took over intercity service and began operating on May 1, 1971. We know it as Amtrak.

The promise at the time was that this would be a limited effort, a reorganization that would put passenger rail on a sounder footing. Look at the sequel. Amtrak has relied on federal money, in grants for operations and for capital, in essentially every year since its founding. Whenever it appears that the company is near to covering its operating costs, there is a new set of capital needs, a new set of routes demanded by members of Congress, and a new request for funds. Whether a route earns its keep is a question the market would answer in one quarter. In a government-chartered railroad it is a question that Congress answers by geography. A route that crosses an influential district survives a long time.

That is my second point at work. A private company that keeps losing money on a line either fixes it or closes it, because its own owners will not keep writing checks. A government program has a different owner, and the owner is the public, which is not in the room.

Why the picks fail: the knowledge problem

Let me turn to the theory, because the failures are not accidents. The economist Friedrich Hayek explained the reason in a 1945 article, "The Use of Knowledge in Society," published in the American Economic Review. His point was subtle and enormously important. The knowledge needed to run an economy is not collected in one place. It is scattered, in fragments, among millions of people: the shop owner who knows what his customers want this month, the engineer who knows that a part is failing, the buyer who knows a supplier is about to cut prices. Much of this knowledge is local, fleeting and impossible to write down in a report.

Prices, Hayek argued, are how that scattered knowledge gets coordinated. When a material becomes scarce, its price rises, and everyone who uses it adjusts without needing to know why. The price system is a way of communicating what no committee could gather.

Now apply this to a panel of officials choosing which energy company to back. They have the same sort of information any investor has: a business plan, projections, a technology that looks promising. What they lack is what the market delivers every day, the verdict of thousands of buyers and sellers deciding with their own money. They cannot see the Chinese factory that is about to cut the price of silicon in half. They do not know which competitor is two quarters from a breakthrough. And even the smartest official is working from a report, not from the grain of the market.

That is the knowledge problem, and no amount of intelligence cures it. Hayek was not saying that planners are stupid. He was saying that the knowledge they would need does not exist in any form that a planner can use.

Why the picks keep coming: the discipline problem

The second half of the argument concerns incentives. A venture investor who backs a failing firm loses money. He loses reputation. His investors leave. The loss is a lesson, and the lesson changes his behavior. The market's discipline is the loss. It is a brutal teacher and an effective one.

A government official who backs a failing firm loses nothing. The money was not his. The agency's budget rarely shrinks because of a failed bet, and there is a good chance the official has moved on before the verdict arrives. Worse, an agency that wants to defend its record has every reason to keep a failing project alive, to extend the loan or add a guarantee, because shutting it down is an admission of error. Economists call this the sunk-cost trap. Political scientists call it Tuesday.

And the winners that government picks are chosen under a different pressure than a lender's. A lender asks: will this firm repay me? A political process asks: will this win votes, make a good announcement and create jobs in a district I care about? Those are not the same question. A project that is attractive on camera is not always sound on paper.

Let me be funny about it for a moment, because the absurdity is useful. Imagine a venture capitalist who answered to voters. Every time he made a poor bet, he would put out a press release about the jobs "saved." Every time a company he backed failed, he would call for a new round of funding to "protect the investment." If it worked in his own firm, he would be bankrupt within a year. In government it is called a program.

The best case for government as investor

Let me do what a fair argument demands and bring the other side's best examples forward, because they are not nothing.

The internet and GPS. The research that led to the internet began with a Defense Department agency, and the Global Positioning System was built and operated by the military. Government spending on basic research has produced technologies that private firms later turned into industries. Defenders say this proves government can pick, and pick big.

Tesla. The electric carmaker received a Department of Energy loan of 465 million dollars, awarded around 2010, and repaid it in full in May 2013, years ahead of schedule. If government loans are doomed, how do you explain that one?

The interstate highways. A massive public investment that most Americans would say paid off.

These are real. Here is my answer.

1. Basic research is not picking winners. Funding a laboratory to explore what is possible is different from selecting a single company to build a particular product and guaranteeing its debt. The government has a legitimate role in some basic science, and the military's need for GPS was a defense need, not a bet on the market. The internet's growth into commerce came from thousands of private firms that nobody in Washington chose. The state funded the seed and the market grew the forest.

1. One success does not rescue the method. Tesla repaid its loan, and good for the company. But a program has to be judged by what happened to its whole list, and by what would have happened without it. Tesla was a venture that private investors were also funding, and the company has since raised vast sums in public markets. A single winner that would likely have found money anyway is a weak defense of a program that also backs Solyndras. In the stock market, you do not defend a trader by pointing at his best trade.

1. Infrastructure is a different category. Roads and bridges are public goods that the Constitution contemplates in the Post Roads clause (Article I, Section 8) and that no private market supplies on its own. Whether a particular highway is worth building is a fair debate. But it is not the same as choosing between two solar-panel companies. When government picks a firm, it is picking a competitor and handing it a tax-funded advantage over its rivals.

1. The portfolio defense confuses the average with the loss. Yes, a venture fund expects some failures. But a venture fund's partners eat their losses and the fund closes when it fails. A government portfolio is never closed by its failures, and its losses are borne by people who did not choose to invest. The comparison with venture capital is flattering only because it leaves out the part where the venture capitalist is risking his own money.

The moral and constitutional angle

Let me get serious for a minute, because there is a principle under the economics. Government is entitled to take your money for public purposes, and the Founders were careful about which. The Constitution gives Congress the power to lay taxes to "pay the Debts and provide for the common Defence and general Welfare of the United States" (Article I, Section 8, Clause 1). It does not say that Congress may take money from taxpayers to place bets on private companies. Madison put the point bluntly in the 1790s, when he argued in the debate over the Bank and other measures that the general welfare language could not be read to swallow the enumeration of powers that follows it. A government of enumerated powers has to ask not only "would this work?" but "who gave us the authority?"

There is also an equal-treatment problem. When government favors one firm with a loan on terms no competitor can get, it is not neutral. It has told every other company in the field that the rules apply to them and not to the favored one. That is a kind of privilege, and privilege is what the Founders thought they were rid of. The conservative objection to crony capitalism is not that businesses succeed. It is that political connection becomes more valuable than serving customers. That is the logic behind big companies love the regulations their small rivals cannot afford, and it is the same disease in another form.

The alternative: rules, not picks

What should government do instead? Not nothing. It should do the job that only government can do, which is to set the rules of the game and keep them stable. Protect property. Enforce contracts. Keep the tax code simple and the same for everybody. Keep regulation predictable. Fund the genuine public goods: defense, courts, basic infrastructure, and a modest amount of fundamental research.

Hayek's own warning about planners is worth a full read in Why the Worst Get on Top, Hayek's Most Uncomfortable Chapter, and the cash-for-clunkers episode in Cash for Clunkers destroyed working cars in 2009 to stimulate the economy shows a smaller version of the same mistake. Then let capital find its own level. If solar panels are a great business, investors will fund them, and they will lose their own money, not yours, when they are wrong. If passenger rail makes sense on a route, a business will run it. If it does not, a subsidy to run it anyway is just a decision to spend your money on something you were never asked about.

I would add a pointed suggestion. If a program of government investment is truly sound, it should be able to pass a sunset test: it expires on a fixed date unless Congress votes, in public and by name, to renew it, with an honest report of what it cost and what it returned. Programs that cannot survive that test were never about investing. They were about spending. (For the same logic applied to budget rules, see From Gramm-Rudman to PAYGO, a timeline of fiscal rules that bent.)

Why this should make us optimistic

Here is the part the pessimists skip. The argument against picking winners is not a counsel of despair. It is a statement of confidence in ordinary Americans. A free economy works because millions of people, each with a piece of knowledge and a stake in the outcome, keep trying things, and the ones that work grow. The knowledge is out there. It is in the garage, the shop floor, the small company that nobody in Washington has heard of. The thing to do is to get out of its way.

Every time a government program fails, the people who made it will say it needed more money. Every time one succeeds, they will say it proves the method. A conservative keeps both ledgers and reads the totals. The totals say that the best economic policy is a level field, not a thumb on the scale.

The case in brief

1. Picking winners means choosing particular firms for special help, which differs from setting neutral rules for everyone.
2. Solyndra's $535 million loan guarantee ended in bankruptcy on September 6, 2011, and Amtrak, founded May 1, 1971, has depended on federal money ever since, showing both the bad bet and the program that never closes.
3. Hayek's knowledge problem explains why officials pick badly: the information needed is scattered and revealed only by prices and by private risk.
4. The government's best examples, such as basic research, Tesla and the interstates, are either not firm-picking or are single successes that do not vindicate the method.
5. The conservative alternative is stable rules, open sunset reviews and a level field, trusting ordinary Americans with the knowledge that no committee has.

What you can do: ask your member of Congress to support a fixed sunset date and a public cost report for any federal loan, guarantee or subsidy that benefits a named company.