Wednesday, May 9, 2012

A Large Barrier to Achieving Privatization Savings

Last week Dru Stevenson linked a study which concluded Arizona pays more for private prisons than it would cost the state if it built and ran the prisons themselves.  In a similar vein, John Stossel and John Pistole of the TSA have been going back and forth over public vs. private screening costs:
TSA administrator John Pistole wrote a letter to the Wall Street Journal claiming there were "inaccuracies" in my recent reporting on the TSA.
Pistole claims that private screening at San Francisco International Airport is no better than TSA and comes "at a higher cost than federal screening."
As I note in today's WSJ, the opposite is true. A Congressional Transportation Committee Report (2011) found that if Los Angeles International Airport switched to private screeners similar to those at San Francisco, screening costs would fall by 42%....
Pistole was likely referring to an internal TSA study. In 2007, the TSA determined that private screeners were 17% more expensive.
But the GAO looked into that study and found 10 different problems with it, noting that the TSA had simply ignored many costs, including "workers' compensation, general liability insurance, certain retirement costs..."
After the GAO report, the TSA came out with a revised study last year that found that private screeners were 3% more expensive.
The GAO says that the TSA's revised study is better, but that it still fails to address four concerns. "TSA needs to take additional actions... to address the remaining four limitations," the GAO said, noting that the TSA makes assumptions about some costs in their study that they were unable to justify.
Despite Stossel's later implication, I don't think the TSA is being disingenuous here.  The heart of the problem is that public accounting systems are simply not set up to do this kind of analysis, and the public officials who use them simply don't have the relevant experience to even spot these limitations.

Back when I was in the corporate world, "Make-Buy" decisions -- decisions as to whether the company should do some task itself or outsource it to companies with particular expertise or low costs in that area -- were quite routine.  Even in the corporate world, though, where accounting systems are built to produce product line profitability statements and to do activity-based costing, this kind of analysis is easy to get wrong (in particular, practitioners frequently confuse average versus marginal costs).

But if these analyses are tricky in the private world, they are almost impossible to do well in the public sphere.  Grady Gammage, a senior and highly respected research fellow at Arizona State University's Morrison Institute, has as much experience with public policy analysis as anyone in the state.  Several years ago, he spent months digging into the financial numbers of Arizona State Parks, with the full cooperation of that agency.  A critical question of the study was how much it actually cost to operate a park, vs. do all the other resource and grant management tasks the agency is asked to perform.  Despite a lot of effort by Gammage and his staff, he told me once that the best he could do was make an educated guess --plus or minus several million dollars -- as to how much of the Agency's budget is spent actually operating parks vs. performing other tasks.

The reasons that this is so hard is that the parks agency's budgeting process was not set up to determine true net operating gains and losses at parks.  It was set up, like most public accounting systems, to enforce accountability to different pools of money that have been allocated by the legislature for certain tasks.  This tends to lead to three classes of problems that cause public make-buy decisions, as well as ex post facto third-party analyses, so difficult.  Since I am most familiar with the parks world, I will discuss these three issues in the context of parks:

  1.  Costs are spread among multiple funds.  A parks agency typically has numerous sources of funds -- gate fees, donations, general revenue funds, capital funds, special grants, heritage funds, lottery funds, etc.  The budgeting process is set up to make sure money from, say, a special heritage fund is tracked, is being spent within its legislated limits, and that its money is used only for allowable purposes for this particular fund.  But this means that to find all the costs associated with a particular park, one must look across numerous budgets.  The local operating budget will have some of the costs, but it will almost never include the cost of equipment and capital assets, which tend to be in an entirely different account; often short term park maintenance is in a different budget item from long-term capital maintenance; and certain activities like local educational programs may be charged to specific grants.
  2. Many costs are not allocated to the field at all, but show up only in headquarters budgets.  For example, Arizona State Parks charges park employee wages to the individual park budgets, but the health insurance contributions and pension contributions or park employees are charged to headquarters.  In addition, costs for numerous headquarters services from HR to IT are not allocated to the parks.  To make analysis even harder, some of these headquarters costs would go away if the park operations were privatized (e.g. lease on a T1 line or maintenance of the park-specific web site) while some costs would not be reduced at all (e.g. cost of maintaining an IT help desk).
  3. A number of large operating costs don't show up anywhere in public accounts.  The best examples of these are liability and workers compensation costs.  Most public agencies are self-insured, so they appear to have no liability insurance costs.  But in reality this self-insurance has real costs (private park operators pay about 6% of revenues for liability insurance), but since states generally pay out liability claims by special legislative acts, these costs never hit park budgets.  In states with pay-as-you-go retirement benefits systems, the cost of future retirement liabilities don't show up on budgets either.
For those who are not ideologues, one way or another, about privatization, the whole point is to save taxpayer money.  But how do we know privatization of an activity is saving money if the government does not know how much money is being spent on the activity?  It's a tremendous problem, and one which leads, inevitably, to a lot of disputes over privatization results, not to mention a serious lack of accountability.

Postscript:  These accounting issues also lead to at least two typical games played by private companies and public agencies:
  • Private companies will often claim savings of allocated costs that may not actually go away if the services are privatized.  Public agency employees tend to have little experience with marginal analysis and zero-based budgeting, and struggle to counter these arguments.
  • Public agencies frequently play a hide-the-pea game to avoid privatization.  Here in Phoenix, we have a rule that if private contractors are allowed to bid on providing a service to the city, the city's departments may bid as well.  These departments have gotten good at cost-shifting to budgets outside of their area of control to submit unrealistically low bids.

1 comment:

  1. This is a really great post! Very good discussion.