Why Government Fails So Often: And How It Can Do Better (62 page)

The single best way to resist this tendency is to counter it with systematic and well-publicized analyses of the consequences of existing and proposed policies, which can then become a factor in the ensuing policy debates.
Chapter 2
described the methodologies of
cost-benefit and cost-effectiveness analyses, defended them against the standard criticisms, and noted that they are required for major executive branch agency regulations under a succession of executive orders going back almost forty years. Congress should endorse this system, lodged in an important Office of Management and Budget (OMB) unit, and extend it to the independent regulatory agencies as well. (Legislation to this effect has been introduced in several past Congresses.) It should also expand the definition of “major” regulations subject to the CBA requirement, while tailoring the rigor of each CBA to the likely costs of the regulation in question. According to Cass Sunstein, the regulation expert who administered this analytical process for President Obama, his agency attempted to include consideration of how regulations affect dignity and income distribution. It also began to use CBA to assess the effectiveness of
existing
regulations (“look-backs”), not just proposed ones.
⁴⁶
Unfortunately, his office has only about fifty staffers and its budget is only 0.0001 of those of the regulatory agencies it needs to review.
⁴⁷

Another source of irrational policies is Congress’s growing use of omnibus appropriations legislation, a notoriously crude policy-making vehicle that combines in a long, often unreadable bill many unrelated provisions on highly diverse subjects, each complex in its own right. As noted in
chapter 6
, this impedes the kind of deliberate, problem-focused analysis and debate that sound policy making requires. In principle, this technique is limited by internal congressional rules, but in practice, congressional majorities override these limits and use omnibus legislation for political and tactical reasons. Congress should make it more difficult to breach these limits.
⁴⁸

Policies are likely to be more rational to the extent that those who receive the benefits also bear the costs, and that both are made visible rather than (in the case of costs) being obscured. This unity of cost bearing and benefit receiving enables them to focus on the true value of the benefit, use the resources most efficiently, and be better informed about whether to support the policy. It also helps policy makers to target assistance to those who want and need it most. Unfortunately, as discussed in
chapter 2
, skilled politicians tend to abhor this
unity, preferring to exaggerate the size and distribution of the benefits and hide and diffuse the costs.
⁴⁹

This unity is less feasible in programs targeted to poor people who cannot bear the costs, but even here the same goal can be met by giving low-income consumers vouchers with which to purchase the service. Policy makers should more extensively employ user fees, such as those charged to national park visitors, which should also reflect the full cost of the service, albeit with appropriate means-tested reductions for the poor.

INFORMATION

In
chapter 1
, I noted that two top budget and policy officials in the Bush and Obama administrations agree that less than 1 percent of government spending is backed by even the most basic evidence of cost-effectiveness; that less than 0.1 percent of government
health care
spending goes to evaluating the effectiveness of the other 99.9 percent; and that the government has largely ignored the “moneyball” revolution in which private-sector decisions are increasingly based on hard data.
⁵⁰
In
chapter 6
, I gave systemic reasons for this official ignorance, some of them tactical in nature.

The good news is that policy information can be improved in many ways, and relatively cheaply. (The bad news is that policy makers must be induced to
act
on the improved information.) One approach, discussed above in the “Incentives” section, is to empower consumers to allocate their program benefits among competing providers rather than vice versa. Here are some others.

In
chapter 11
, I noted that the largely successful 1996 welfare reform legislation was enacted only after years of state-level experiments with different types of welfare-to-work programs. The Department of Health and Human Services facilitated such experiments by exercising its legal authority under Section 1115 of the Social Security Act to waive certain restrictions in order to promote and then assess policy innovations. The information gleaned from those programs had at least two positive effects. First, it increased Congress’s confidence
that reforms could succeed in reducing the welfare rolls while improving the prospects for women and children through work readiness and job training programs and wage supplements. Then Congress and the department used this experience-based information in the actual design and implementation of the new law. Today, states are seeking similar waivers from the Affordable Care Act of 2010 in order to determine how best to implement the new law. Carefully designed, large-scale policy experiments of this kind should be designed and funded. Devoting even a tiny percentage of program funding to effectiveness assessment would likely repay itself—according to estimates by Kennedy School of Government economist Jeffrey Liebman—a hundred times over.
⁵¹

Most proposals for program change are based on theoretical hunches about which policies will have which effects, and why. Unfortunately, even inspired, informed guesswork is problematic. As Michael Abramowicz, Ian Ayres, and Yair Listokin note,

theory alone cannot resolve many policy issues because different theories point in different directions. Scholars attempt to inform these debates by parsing historical data, but regression analysis of policy is fraught with complications. There is little policy variation on many topics of national importance, and the variation that does exist is correlated with many other factors. Empirical policy evaluation often resembles a drug study in which the experimental population does not receive an assigned treatment and instead gets to choose whether to take the medicine or the placebo.
⁵²

They propose instead to “randomize” law by having agencies conduct randomized, controlled experiments of proposed policy innovations before adopting them in more permanent form.
⁵³
Designing such experiments outside the laboratory can be difficult, but Jim Manzi, whose work on policy effectiveness was discussed in
chapter 11
, shows that they are now routine in business, especially online—Google claims to have run some twelve thousand experiments in 2009, with about 10 percent leading to business changes—and are also used by political strategists and political scientists to test possible reforms.
⁵⁴
Experiments’ validity, of course, depends on how rigorously
they are designed, but since companies’ profits will be affected by their accuracy they have strong incentives to design them well. The Office of Information and Regulatory Affairs (OIRA) should be tasked with these research designs.

Policy makers, in contrast, shy away from doing so. They may cite the costs and delays occasioned by such experiments, even though the costs of misguided policies are infinitely larger and delays in instituting them may therefore be salutary. They may also fear that experimental results may cast doubt on their proposals, as most do.
⁵⁵
Yet precedents do exist. Starting in 1968, the Department of Health, Education, and Welfare contracted with RAND to conduct the income maintenance and national health insurance experiments. Forty-five years later, the findings from these experiments continue to influence a variety of social policy debates, especially the design of health insurance reforms and cash assistance to low-income families. Such experiments are highly cost-effective, and their use should be expanded.
⁵⁶
Indeed, Manzi’s proposal to create a federal experimental agency to design and run these experiments is worth serious consideration.

Policy makers should also experiment with the kind of “social impact bond” (SIB) that New York City’s mayor Michael Bloomberg has developed, based on early efforts in Britain, to test new policy approaches to reducing the high recidivism rates among incarcerated teenagers. With SIBs, a government contracts with private investors (here, Goldman Sachs) to fund the new approach at no initial cost to taxpayers. Under New York City’s SIB, if the approach causes recidivism to drop by 10 percent, Goldman Sachs gets its money back. If it falls more than 10 percent, the firm receives a profit. But if recidivism does not decline by the target amount specified in the bond, it will lose some or all of its investment. With skin in the game, the firm’s incentive is to rigorously assess the proposed innovation in advance, and if it buys the SIB to try to make the program work, which would also save money for the taxpayers due to lower crime and incarceration costs.
⁵⁷
In June 2013, the firm announced that it will use the SIB form to finance early education programs.
⁵⁸
The experience
with SIBs should itself be rigorously assessed with randomized experimentation.

In their recent book
Big Data
, Viktor Mayer-Schonberger & Kenneth Cukier present many examples of how our capacities to gather, analyze, and perceive otherwise obscure patterns in large but often highly decentralized data sources can improve both the identification of, and solution to, problems by private and governmental actors in areas like public health, consumer protection, law enforcement, and product design.
⁵⁹
Cass Sunstein, the Obama administration’s former regulatory policy chief and a leading scholar on regulation, emphasizes the importance of officials accessing and using policy information that is widely dispersed in the private sector.
⁶⁰
Policy makers in all branches of government should exploit these opportunities at least as eagerly as private interests are doing, while protecting the legitimate privacy interests that this avalanche of information may threaten. Because information is power, however, the pursuit, analysis, and use of these new and more broadly based sources of information are less straightforward and more politically tactical in the policy realm than in the private sector. Indeed, “big data” is already enabling private firms to influence diet, smoking, high blood pressure, medication compliance, and other public health challenges often hobbled by bureaucratic failure. A promising example is scientists’ effective use of internet data-mining techniques to detect prescription drug side effects more quickly and accurately than does the Food and Drug Administration, which relies on its relatively cumbersome and flawed Adverse Event Reporting System.
⁶¹

Intellectual property law, expressly authorized in the Constitution, is a vital protection and spur for innovation and information sharing. Experts in this field are sharply divided over the normative, empirical, and policy issues that surround intellectual property rights and processes, with traditionalists emphasizing the legal protections that are needed to incentivize innovation, and other groups, such as Creative Commons and advocates for a new open-source economics, emphasizing the innovation-chilling and innovation-blocking risks of overprotection and the promise of more collaborative arrangements.
⁶²
With the pace of new “blockbuster” drugs slowing in recent years, the stakes in getting intellectual property rules right are soaring.

The federal government should make better use of private sources of credible information for consumers. In a number of policy areas, industries have established programs to independently assess and certify the safety and quality of important products in ways that are faster, cheaper, and more reliable than the government could manage itself.
⁶³
Some of these private certification programs relate to risks that state law primarily regulates (e.g., fire hazards, kosher food, and construction materials), but others, such as the Joint Committee on Accreditation of Health Organizations and bond rating agencies, operate in conjunction with federal programs (Medicare and securities regulation, respectively). Both of these examples (especially bond ratings in the wake of the financial crisis) have been severely criticized for conflicts of interest, lack of competition, and flawed procedures, and the Federal Trade Commission is investigating claims that programs for certifying forest products as “green” are being used to stifle competition.
⁶⁴
(Leadership in Energy and Environmental Design certification of buildings as green has also proved misleading; its highest mark was given to the Bank of America Tower, one of the greatest “energy hogs” among New York skyscrapers.
⁶⁵
) Policy makers should search for other policy areas in which qualified private certification systems—for non-gene-altered food, say
⁶⁶
—can supplant or supplement direct government regulation. At the same time, we need to recognize that private regulatory systems (the National Collegiate Athletic Association, for example) often exhibit many of the pathologies of public ones,
⁶⁷
and must assure that these systems actually protect the public values they are supposed to serve.