Tuesday, August 14, 2012

Analyzing the Ryan Budget: Part III

It's time to resume my review of Paul Ryan's budget proposal. If you're just tuning in, see my previous two posts. In the first, I discussed what the Ryan plan means for security and defense spending; in the second, I examined the impact on the Departments of Energy, Transportation, and Heath and Human Services along with an array of federal regulatory agencies.

Today, I'll look at safety-net spending in the Ryan budget -- see also here for the "blueprint" summary -- in particular, the major reforms to the program structures of Medicaid and the Supplemental Nutrition Assistance Program. We'll also examine changes to federal student lending, Pell Grants, and job training programs.

The most important point about the Ryan plan for Medicaid is the devolution of federal administration to the states using block grants. That is, instead of the federal government managing the program, the benefits, the participants, the spending, etc. for all 50 states, the federal government would allot funds to the states, who would run the program at that stratum of government.

I think this makes sense. The fundamental insight of federalism that multi-level government allows for the division of government functions according to the comparative advantages of each respective level. Federalism thus generates its own sort of "gains from trade" for constituents. As a citizen of the United States, I get public services of defense and international representation; as a resident of the State of New Jersey, I get public goods like highways and public services like state troopers; as a resident in my county and borough, I get public services like local police and public goods like local roads. Each of my examples I believe to be most efficiently delivered at such a level of government -- it would be disastrously suboptimal for the national government to manage every back road in my town, just as it would be if my town tried to collaborate with other towns to provide defense.

From these principles, I conclude that the finance for most safety-net programs is best conducted from the federal level. In times of economic crisis, the federal government can print money and run up deficits and debts, whereas states are forced to cut back heavily on public services when their finances face strain. That means federal safety-net services -- take food stamps for our example -- can be there when people need them most. That fiscal flexibility gives the federal government a comparative advantage in countercyclical public spending.

Yet I do not think that most safety-net programs are best administered from the federal level. Rather, I think state governments have a comparative advantage in their administration for several reasons: (1) there is more room for cost-reducing, quality- and access-enhancing, innovation in programs, with states serving as "laboratories of democracy," as Justice Brandeis memorably phrased it; (2) the smaller the constituency, the more accountable the government; (3) diseconomies of scale are severe for programs with large bureaucracies and millions of participants spread across the nation.

I don't think the level of program administration has to be a normative issue. (Democrats, I'm talking to you.) The belief that the federal government should be bigger and do considerably more should not impact your judgement as to what level of government should handle Medicaid administration. This is a positive question about efficiency.

The success of block grants is, of course, conditional on the formula which determines the size of the block grants. Here I think Ryan errs. Josh Barro explains how Ryan's proposed Medicaid block grant formula needs to changed here far better than I could:
First...the block grant must grow fast enough. Ryan hopes to save money by growing the Medicaid block grants in line with population and the Consumer Price Index, but this would lag the growth of medical spending per capita by 3 or 4 percentage points per year...As part of an integrated plan to reduce health care inflation, it makes sense to set this growth rate a bit below current expectations for cost growth—say, GDP plus 1 percent—but CPI is too low...Second, the block grant should move with economic cycles...t should be a measure that tracks the Medicaid-eligible population...Finally, there should be a feature so that the federal share of Medicaid expenditure actually grows in recessions, and then shrinks in recoveries.
Ryan's formula leads him to discover, as Barro put in it a related piece in City Journal, savings which are "magical" -- that is, fraudulent. I think his estimates for Medicaid costs, then, are no good: Ryan "assumes $810 billion in savings over 10 years" on page 93 of his budget. Most of those savings come from effectively sheer cuts in public services rather than incentive structures which produce productivity gains. That is, at least to me, a significant defect in the Ryan plan. What would strike me as more intellectually honest on Ryan's part is to cost out his Medicaid plan based on the assumption that health care costs will continue to rise according to their historical rate of growth, and to track the block grants along with them. That way, the states could enjoy the savings of the program if it worked as he expected by bending the cost curve, and eventually the federal government could adjust down the rate of payment growth. But simply assuming success is the way to failure.

All of the above applies to Ryan's plan on SNAP. I think block granting it for purposes of program administration is the right way to go, but I fault the block grant formula for its failure to include higher cost inflation for poor Americans and countercyclical cushioning.

Now we will move onto Ryan's plans for federal involvement in student lending. As of 2010, the federal government now makes direct loans to students to finance education and discounts their value according to the federal government's cost of borrowing. This is a bad idea: the government should discount the loans by more than its cost of borrowing; it must include expected cost of defaults and of administration and collection on a per-dollar-lent basis. Ryan's reform would require all federal lending to meet the fair-value accounting standards of private lenders by setting "market-based risk premiums."

Ryan also scales back the rate of growth of Pell Grants by tightening the means-testing and eligibility requirements and reducing the maximum grant size. The cost of that program will double on a nominal basis from $16.1 billion in 2008 to $36.4b in 2013. This is mainly a structural (not cyclical) change: Pell Grants were a nominal $8 billion in 2000. I don't have an informed opinion on whether the shift to financing higher education through taxation is desirable, as that would justify the expansion of the program. In my opinion, given that direct aid for higher ed now has the budget footprint comparable to a Cabinet-level department, it may be time for a national conversation about this issue. I see a case for extensive public finance of higher education, but I also recognize that scarce resources means that governments should not spend beyond the point where the marginal benefit of it exceeds marginal costs, given crowding-out and incentive effects.

Ryan also wants to consolidate all job training programs spread among federal agencies. The GAO, Ryan's budget says, found 82 different programs across 10 agencies and dozens of overlaps. This is a straightforward place to find efficiencies.

2 comments:

  1. It's worth pointing out that Canadian Medicare is funded by the federal government and organization is delegated to the provinces. (The bill is 26 pages long, and it's only that long because half is in French.)

    I think your assessment of the benefit of federalism here hits the nail on the head, although I worry a little bit about whether state demographics would make a block grant approach unfair: FL has a different demographic profile than NY, though they both have ~19 million people. Should they each get the same size block grant? (In the interest of keeping the computation fast and relatively free of political influence, probably yes.)

    ReplyDelete
    Replies
    1. Thanks for letting me know about Canada on this.

      Again, it all comes down to the block grant formula. One needs to be sure that the size of the grant is proportional to the right variables. In this case, I think that one would have to take into account not just population, but the size of the elderly population, and perhaps also the elderly population as a fraction of the state total (so as to assess that state's means of bearing the health costs).

      Delete