She crafted a number of stories that highlighted the big picture issues: imbalance in bargaining power between city governments and giant multinational corporations, the blatant conflicts of interest on display in Texas subsidy procurement, and a border war between Kansas and Missouri involving multimillion dollar incentives to move existing facilities across the state line, with no net benefit for the Kansas City metropolitan area, let alone for the U.S. as a whole.
The last few days have given me time to absorb the articles and the database Story created, as well as surveying the commentary on the web from well-known experts on subsidies. Several tentative conclusions seem in order.
First, as I pointed out in my last post, and backed up by Timothy Bartik's detailed analysis of Michigan, 5/8 of the national total is in the form of sales tax breaks, and probably the overwhelming majority of those sales tax reductions should not be considered subsidies. Here is what Bartik says about Michigan:
For example, in my own state of Michigan, the New York Times database
identifies $6.65 billion in annual state and local business incentives.
Of this total, $4.83 billion is in “sales tax refund, exemptions, or
other sales tax discounts”. Of this $4.83 billion, almost all of these
refunds come from two provisions of Michigan tax law. First, Michigan
does not apply the sales tax to most services, including business
services, which saves businesses $3.88 billion annually. Second, for
manufacturing, Michigan does not apply the sales tax to goods used as
inputs to the manufacturing process, which saves manufacturers about
$0.92 billion in sales tax.
For those keeping score at home, that means that $4.80 billion of the $4.83 billion in sales tax breaks should not be considered subsidies, unless you consider manufacturing "specific" enough that this aid constitutes a subsidy, in which case only 80% of the sales tax breaks should be excluded from the subsidy tally.
Second, changes of this magnitude mean that the Times estimates are not sufficiently accurate to use in a statistical analysis, as Richard Florida attempts in The Atlantic Cities. Finding out if incentives affect outcomes like wages, employment, or poverty is precisely the type of analysis we would like to do, but the fragility of the data makes this premature. The good news is that since the data on these state programs are all in one place, it should be possible to get a better handle on state incentives by cutting out those programs which should not be considered subsidies. Different analysts will no doubt have different judgments about what should be counted as a subsidy, but since the database is so inclusive, it should be useful no matter what your definition of subsidy is.
Third, there are some smaller errors in the program database as well. The one I have identified so far is that it counts net operating loss (NOL) tax provisions as subsidies in Illinois and New Hampshire, but not in other states, even though all states with a corporate income tax will have an NOL provision. In any event, this should not be considered a subsidy at all, but a part of a state's basic macroeconomic framework. In addition, Timothy Bartik pointed out to me in correspondence that the program database does not include single sales factor apportionment (only counting what percentage of a multi-state firm's sales take place in a given state, rather than standard three-factor apportionment that uses percentages of payroll and property as well) as a subsidy, which it should.
Fourth, the program database does not distinguish between investment incentives (subsidies to affect the location of investment) and subsidies more generally, which may or may not require an investment to obtain them. This is an important distinction I have tried to make clear by providing separate estimates in Investment Incentives and the Global Competition for Capital: $46.8 billion in incentives, and $65 or $70 billion in subsidies, depending on whether or not you count non-specific accelerated depreciation as a subsidy.
Finally, as Phil Mattera at Good Jobs First points out, the deals database misses a number of large awards, leaving out Tennessee's $450 million (present value) subsidy to Volkswagen and an even bigger package for ThyssenKrupp in Alabama. It also underestimates other awards, including Apple in North Carolina and Boeing in South Carolina. I also found that it underestimated subsidies to Dell and Google in North Carolina by omitting the local subsidy portion of the awards, a problem Ms. Story is aware of, as I noted in my last post.
The Times series has been great for the spotlight it has put on state and local subsidies and the sometimes vulgar politics surrounding the process of awarding them, and for compiling a great database of programs all in one place. However, its interpretation of the sales tax breaks, which are 5/8 of the national total but largely not subsidies, confuses the issue of total impact on state and local budgets and makes statistical analysis premature. This will require some work to fix, but it appears like most of the raw material is there to do it.
Cross-posted at Angry Bear.