The United States government is starting to sell off its stake in General Motors, taken as part of the bailout which saved the company in 2009. It plans to take 15 months to completely disinvest, but in the meantime, that investment is doing so well that the total value of the bailout may be far smaller than was previously thought.
When the government intervened in July 2009, it spent $49.5bn to purchase most of the assets and trademarks of "old GM", through an intermediary called NGMCO Inc, ensuring the continued operation of most of the company's plants and continued employment of most of its workers.
Since then, the Treasury has already earned back $28.7bn of its money from "repayments, sales of stock, dividends, interest, and other income ". And with its first move towards disinvestment, it plans to sell 200m of its 500.1m shares in GM back to the company itself, for $27.50 a share, raising a further $5.5bn. So at the end of that sale, the government will be left with $14.8bn still in GM and a further 300.1m shares.
It's obviously unlikely that the state will make back its entire stake; Felix Salmon estimates that  the price would need to rise to $50 a share, considerably higher than the all-time peak of $39.48 early last year. But it is possible; and it's definitely the case that the state will lose a lot less than the $50bn figure which was causing such consternation when the bailout was announced.
Such is always the case with investment programmes like this one, though. The headline figure gets reported, and debated over, as though it were just the same as any other spending; the fact that that money comes back to the Treasury, either in actual cash, as with this sort of investment, or in kind, as with most infrastructure investments, is buried in the discussion.
If the government manages to sell the its remaining shares at today's face value, it will end up losing around $6.5bn from its four-year investment in GM. If the share price rises, that number will fall lower still. At the time, there was obvious uncertainty about how successful the bailout would be; and there was always a chance that the government would lose its whole stake.
But there was also a chance that, as with its similar stake in insurance company AIG, it would make a profit. And absent either of those, a $6.5bn programme which saved a company employing 202,000 people isn't that bad. But as Matt Yglesias points out , the problem may be that those jobs are, in the long run, not saveable at all:
The total collapse of the Michigan-centered auto industry would, for better or for worse, have opened up new market opportunities for other automaker with production facilities located elsewhere… On the other hand, either the total collapse of the midwestern auto industry or a huge wave of bank failures would have produced massive dislocations in people's lives and a lot of misery on the road to renewal. Those are the questions to think about, not how much money was made or lost in this or that investment.