Megan McArdle has done consistently excellent reported pieces on the GM bailout, and her recent evaluation of its net effect on the U.S. Treasury is no exception. Her bottom line is that the deal caused U.S. taxpayers to:
burn $10-20 billion in order to give the company another shot at life. To put that in perspective, GM had about 75,000 hourly workers before the bankruptcy. We could have given each of them a cool $250,000 and still come out well ahead compared to the ultimate cost of the bailout including the tax breaks
This is in line with the Obama administration’s $14 billion estimate of the net cost to the Treasury, as reported in the Wall Street Journal. If anything, I think this understates the case on the direct costs, because it does not consider other direct transfers of economic value like the government support for Delphi that inflated the value of the asset that GM sold to create a big chunk of their headline profits this past quarter, green-car development subsidies, and uncompensated interest costs on the government investment.
But no matter what realistic direct bailout costs you estimate, the objection of bailout defenders is that it is dwarfed by the other receipts or avoided expenditures created by the bailout. According to the Wall Street Journal, this is exactly the defense offered by the Obama administration:
The White House report said the money invested in GM and Chrysler ultimately saved the government tens of billions of dollars in direct and indirect costs, including the cost of unemployment insurance and lost tax receipts that the government would have incurred had the big Detroit auto makers collapsed.
There is a lot to this point, but it’s not really so simple. You can’t compare all of these net tax receipts (or more broadly, economic activity) to what would happen in “the world as it is today, minus GM.”
First, in the event of a bankruptcy, you don’t burn down the factories, erase all the source code on all the hard disks, make it illegal to use the brand name Chevrolet, and execute all of the employees. Others take ownership of the assets, and the employees go on with their lives. Some of these assets will be put to use generating revenues, profits, and taxes, and some of these former employees will get jobs or start businesses, and generate revenues, profits, and taxes. In order to measure the effect of the bailout over, say, five or ten years, you have to compare the actual taxes collected to what would happened over this same period in the counterfactual case where the bankruptcy was allowed to proceed. What owners would have bought the factories and IP assets, and what would they have done with them? What businesses would the former employees have started? Who would have moved to Arizona and retired? What new industry clusters will evolve in Arizona because of this transfer of people?
Second, some of the profit GM makes today would have been made by other companies that picked up some of the slack if the company lost market share after a bankruptcy. They would pay taxes on these profits, and as far as government receipts are concerned, money is money. How would auto industry structure evolve over time given whatever changes happened to the assets currently owned by the legal entity GM, or the employees currently paid by it?
Anybody who tells you they can answer all of these questions reliably is full of it.
And that doesn’t even start to get to the really long-run considerations of what effects this has on rule of law and moral hazard (or if you want to make the case for the bailout, social solidarity and degradation of the working class).
I hold the belief, quite strongly, that the net effect of the GM bailout will be negative. More precisely, I hold the belief that over a series of many such decisions, a mindset that would have been stringent enough not to have sanctioned the GM bailout is likely to lead to better overall economic outcomes for America. This belief is ideological — not in the sense that I just hold it for inexplicable reasons that cannot ever be changed by empirical analysis, but in the sense that I don’t believe that human beings currently have the capability to conduct the kind of analysis that should convince a rational observer to change his mind about the GM bailout in isolation from a more profound paradigm-shift-like change in his beliefs about the world.
The GM bailout is not an isolated case of this problem. And as I’ve argued many times, impressive-sounding empirical analysis is typically insufficient to measure the effect of important economic interventions like the stimulus program. If you can’t even measure what effect already-executed programs have had, how likely is it that you can predict the effects of future programs?
Acceptance of this degree of ignorance doesn’t cut equally against all ideological positions. It leads naturally to a call for decentralized decision-making, experiments, and entrepreneurial groping toward knowledge.