Evaluating federal credit programs: Prepared testimony before the Joint Economic Committee of Congress

Thank you for the opportunity to testify today on the important topic of the federal credit programs, which now represent over $10 trillion of loans and guarantees to the private sector, but have received too little attention. I have tried to help remedy this over the years. In 2003, I founded the Center On Federal Financial Institutions, a non-partisan think tank devoted to analyzing the federal credit and insurance programs, which I ran for three years. (Douglas Holtz-Eakin was a board member, along with a number of other distinguished experts from both parties.) In 2011, I wrote a comprehensive book on the federal credit programs called “Uncle Sam in Pinstripes: Evaluating US Federal Credit Programs,” the first book of its nature in 25 years, since a previous Brookings publication by Barry Bosworth and co-authors.

Given the political nature of so much of the discussion, let me note that I am as close to a neutral expert as you will find in this area. I do not belong to a political party, have served in no administration, and am a moderate on the political spectrum. Further, my work in this area is as a technician. I do not intend to tell Congress which programs it should authorize or what their size should be; that is for you to decide by weighing the many trade-offs that exist. However, I do feel very strongly about ensuring you have the appropriate information to make those decisions and that the structure and operation of the programs maximizes their effectiveness and efficiency.

Well-designed federal credit programs are an appropriate response when there are imperfections in credit markets that cannot readily be solved in other ways. Lending to students is one of the clearest examples. Private lenders are simply unwilling to make very long-term, unsecured loans at any reasonable interest rate to individuals on the basis of their potential earnings capacity judged as of age 18. Yet we are clearly better off as a society if capable students are able to attend college even if their parents cannot, or will not, finance them. In this circumstance, it is difficult to see any alternative to a government program of loans, loan guarantees, or outright grants.

On the other hand, the government appropriately stays clear of most forms of business lending, where there are well-developed markets that function smoothly. The principal exception is for small business, where there is reason to believe that many worthy firms have trouble expanding because of difficulties in obtaining credit. This is a problem across the globe. I have been speaking with officials in Europe and China about the benefits of our Small Business Administration, which melds private sector credit underwriting with federal guarantees to increase credit availability and lower its cost. Not every aspect of our SBA program is right, but the general idea is an excellent one.

Note that I started this discussion with the phrase “well designed.” Just because there is a good case for a federal credit program in an area does not mean that aparticular program is worthwhile. Programs must be designed and operated so that the societal benefits outweigh the costs, which is not always the case in reality. Further, it would be good to increase benefits and reduce costs even for those programs that are already worthwhile.

In my book, I made a number of recommendations for improving the effectiveness and efficiency of the federal credit programs. I would like to repeat some of them here. Specifically, we should:

• • Target borrowers more carefully.
  • Take more account of the relative risks of different loans
  • Use the same budget rules for all federal credit programs
  • Use risk-based discount rates for federal budget purposes
  • Formalize the process of initiating new federal credit programs
  • Create a federal bank to administer all federal credit programs
  • Focus more on optimizing the allocation of money between programs
  • Spread best practices more effectively across the government.
  • Improve the compensation and training of federal financial workers

For this hearing, I understand the most interest may be in the point about risk-based discount rates, so let me expand at length upon that issue before reviewing the others.

Use Risk-Based Discount Rates for Federal Budget Purposes

Accounting systems, such as that used for the federal budget, are tools that should be designed to meet specific needs. The internal accounting systems used to make management decisions at companies, for example, often differ from those used for reporting to shareholders. Tax accounting rules, for their part, are designed to appropriately raise revenue and are not intended to be the same as managerial accounting methods or Generally Accepted Accounting Principles.

Our current budgeting approach for federal credit programs ignores the variability of potential results. A program that could with equal probability earn the government $110 or lose the government $100, is treated the same as a program that with certainty will earn the government $5, since on average they will yield the same result. (In the first case, half the time we make $10 more than we lose the other half of the time, for an average of $5 of profit.) Given how strongly the budget numbers drive decision-making, we are effectively acting as if Congress and the taxpayers do not care about risk.

The question of whether to use risk-based discount rates in the federal budget is a pragmatic one – will it help Congress and the Executive Branch to make better decisions? I think it would. Currently, taxpayers are effectively taking risk without charging for it, from a budget point of view. I believe subsidy costs in the federal budget should instead reflect this uncertainty, for several reasons.

Structuring programs to minimize risk. It is important that federal credit programs be structured to minimize risk, where possible, while still achieving the overall objectives. This is particularly crucial since there appears to be at least a modest bias in initial subsidy estimates to understate total costs. This bias is almost certainly worse in cases of highly uncertain outcomes, because of political pressure for relative optimism.

Recognizing risk. The benefit to borrowers of government loans as compared with loans on the private market is higher for risky loans, all else equal, since these would be priced higher by private lenders but are generally not priced by the government to reflect their risk. Ignoring this fact for federal budgetary purposes has distorting effects on the choices politicians make. In particular, there will be a tendency to direct scarce federal lending dollars to sectors where there is more uncertainty in the outcomes, since the borrowers will find the federal loans most valuable. They will lobby harder for them and are more likely to apply for such loans and to choose them over private alternatives.

Implementing risk-based pricing. Risk-based pricing, which is one of my other recommendations, is considerably more likely to be implemented if the budget appropriately reflects risk as a cost factor. The situation today, in which a loan with a wide range of potential outcomes is treated as costing the same as a relatively certain loan, discourages political decisions that take account of such risk.

There are reasonable counter-arguments to moving to risk-based discount rates, although I do not find them compelling. The principal one is that the US government can spread any unexpected losses over a wide tax base and many years of time and therefore does not need to worry about variability in outcomes. However, I buy the argument by Professor Deborah Lucas at MIT, and others. She maintains that taxpayers are the ultimate bearers of this risk, just as shareholders bear the risk at companies, and they are risk-averse. Shareholders certainly prefer more certain returns to riskier ones, and I would submit that taxpayers do as well. The way in which federal credit losses are ultimately offset is by increasing taxes or decreasing federal expenditures. It seems very likely that taxpayers would prefer less risk of a big tax increase to more risk of one, even if there is an offsetting potential on the other side for unusually good performance and future tax reductions. This is especially likely because there is a strong correlation across credit losses, so that losses are concentrated in a few years when the economy is particularly bad and taxpayers are unlikely to feel capable of comfortably bearing the resulting tax increases.

There are also technical arguments about maintaining the consistency of federal credit programs with other programs in the budget and of dealing with swings in estimated costs as interest rates move from year to year. These are reasonable concerns, but they are outweighed by the fact that Congress uses the initial subsidy estimates in the federal budget as by far the most important figures on which to make decisions about federal credit programs. As long as these are the critical numbers, I believe it is important to incorporate risk appropriately into them in order to improve the quality of decisions.

Let me now expand somewhat on the other recommendations:

Target Borrowers More Carefully

Federal credit programs are generally intended to spur greater investment, output, and employment in certain sectors. Unfortunately, a significant portion of the subsidies appear to be taken up by borrowers who would have performed the desired activities without any federal help. That is, many borrowers would have made the same total investment by obtaining funds through private channels or using more of their own assets. William Gale concludes that “credit subsidies cost the government in excess of 50 cents per dollar of incremental target-group investment.”¹ Other researchers are unable to nail down the portion of borrowings that represents additional activity but conclude that it is only a fraction of the total borrowings supported by the government programs.

It is politically and bureaucratically easiest to continue providing funds for everyone who meets the group characteristics, is able to repay, and wishes to borrow. However, the cost to the taxpayer would be considerably lower if we could eliminate subsidies for at least a portion of those borrowers who would not change their investment behavior because of the federal aid or would do so only marginally. In many cases, this may involve a greater degree of means testing, such as in the area of student loans. A lower-income student is more likely to enroll in or continue to attend college as a direct result of loan availability than is a higher-income student who would attend regardless. Similarly, mortgage aid to a lower-income homebuyer is more likely to result in a new home purchase than is aid to a wealthy family. In the case of housing, more targeting would only partially counteract the greater benefit to wealthy individuals of the deductibility of mortgage interest.

Take More Account of the Relative Risk of Different Loans

A key societal role of financial institutions and markets is to direct resources from savers to investors. The economy performs best when the financial sector chooses the optimal set of projects to fund, based on the trade-off of risks and potential returns. The role of the federal credit programs is largely the same, although the allocation decisions are intended to reflect additional societal factors that are not easily encompassed by the markets, either because of market flaws or economic externalities.

Unfortunately, government programs find it difficult to do two key things that are critical to optimal resource allocation. First, the programs are generally required to provide credit to anyone who meets certain simple criteria, such as being a college student, or being a veteran and meeting some simple down payment and income tests for a mortgage. In contrast, private financial institutions in normal times require more detailed information to make credit judgments. Second, federal programs often have no variation in their loan pricing to reflect risk. If they do, it is usually quite simple, with two or three price levels depending on simple criteria. The loan pricing practices of private institutions are much more sensitive to the particular level of risk.

It would be better if the government incorporated more underwriting and risk-based pricing into its credit programs. This would be fairer to the taxpayers and the various types of borrowers, since there would be lower loan default rates and a more even distribution of subsidies between borrowers representing different levels of risk. Default rates would decline because some weak borrowers would be turned down but also because risk-based pricing would encourage the more creditworthy borrowers to participate and discourage the worst risks from borrowing. (Using a single price for all potential borrowers creates a relative bargain for the riskiest borrowers.) A fairer distribution of subsidies would result because the gap between the private market rate and the federal rate would be more uniform, as federal pricing followed the private pattern more closely by adjusting for risk. (Everyone might still receive a significant subsidy, but there would be less variance in the subsidy).

The big obstacle to credit underwriting is the difficulty in having a government employee exercise discretion in evaluating loan eligibility, since every such choice can produce a backlash from the borrower, with repercussions for the politicians and bureaucrats involved and, in the extreme, the threat of a lawsuit. However, information technology is now so advanced that it would be possible to have significantly more refined automatic criteria for turning down the worst risks and for deciding how much to vary the interest rate on the loan.

The obstacles are both harder and easier when the program is implemented as a guarantee program, as is true for small-business loans. Private lenders can exercise credit judgment using the best information available, eliminating the problem of discretion by government employees. However, it can be difficult to provide sufficient incentives to persuade those lenders to invest in the resources necessary to gather the information and make the credit decisions carefully. The Small Business Administration provides less than a 100 percent guarantee of loan repayment in order to encourage such credit underwriting. However, being shielded from, say, 80 percent of any loss commensurately reduces the economic incentive to underwrite, which can be a costly activity for small-business loans. To counteract this, the federal government has provided various incentives and disincentives in its guarantee programs. For example, student lenders with low default rates among their borrowers used to receive a higher guarantee percentage and some other advantages, before the abolition of the guarantee program. Those with the worst records were knocked out of the program over time.

There is considerably more room to design or expand mechanisms in both the direct and guaranteed loan programs to reflect the relative risk of various potential borrowers. For example, Congress should allow risk-based pricing more frequently, indeed it should push for it. As noted, advances in information technology provide a sounder and more consistent basis than previously existed to make distinctions among borrowers.

Use the Same Budget Rules for All Federal Credit Programs

Budgets assist in resource allocation decisions, such as determining which federal credit programs should be expanded and which contracted. This process is distorted, however, if budget calculations are made under different rules for different programs, as is the case today. The traditional credit programs are currently subject to the rules of the Federal Credit Reform Act, which broadly represent best practice in this area. However, the emergency programs under TARP used a variant of those rules with a different discount rate, an approach that the author believes is superior. The Fed programs, for their part, were under another set of rules, and the FDIC programs under yet another.

It would be best to use the rules of the Federal Credit Reform Act, as amended for the TARP programs to include risk-based discount rates, for all federal credit programs. The consistency would remove a considerable temptation to house programs where they would appear cheapest on the federal budget, rather than where they would be administered best. In addition, the consistency would aid in allocating money between the programs.

Formalize the Process of Initiating New Credit Programs

It would be useful for Congress to impose on the executive branch and on itself the discipline of a formal proposal process for new credit programs. The Congressional Budget Office should review each proposal and comment on the consistency and achievability of the elements of the proposal.

Ideally, any proposal would be required to include the following elements:

• • a statement of the purposes of the program, with prioritization among objectives if possible, including an explanation of any perceived market failures or externalities meriting federal intervention or a statement that the intent is redistributive
  • a definition of the target groups to be aided
  • an explanation of the mechanisms to be used
  • a target initial subsidy level and explanation of how the level was arrived at

if a new agency is proposed, an explanation of why an existing federal credit provider should not run the program

A strategic plan is not a panacea, and Congress could choose to pass legislation that only minimally met these proposal requirements. However, some of the worst potential problems could be weeded out by encouraging a reasonable explanation of the new program, its merits, how it would work, and what it would be likely to cost over time. A coherent and comprehensive explanation of the key points might also provide helpful guidance to the administrators of the new credit program going forward.

This reform would force a discipline on proposals that create a credit program only as a by-product of larger legislation. For example, the broadcasting spectrum auction that took place some years back ended up including a requirement that the government accept deferred payments from minority bidders, to encourage a more diverse range of bidders.² Unfortunately, this effectively meant that the government in a number of cases lent speculative investors the funds to make their bid, taking what was effectively equity risk, not credit risk. When the tech bubble burst before those investors could finance themselves, the government ended up taking large losses. It is likely that the mechanisms for this aspect of the spectrum auction would have been given considerably more thought if a formal process for credit program proposals had been in place.

The reform would also have pushed Congress and the administration toward a greater consistency across the federal credit programs. By highlighting comparisons between a new proposal and existing credit programs, the Congressional Budget Office review could encourage standardization in cases where there was not a strong reason to follow a different procedure.

Create a Federal Bank to Administer All Credit Programs

Cabinet departments, such as Agriculture or Education, have no comparative advantage in evaluating and administering credit programs and having the programs spread out among so many different entities loses many potential advantages of economies of scale and expertise. It would be better to bring all the credit programs under one roof.

Admittedly, creating a federal bank would be a stretch politically, since moving existing programs means changing the relative power of different congressional committees and other groups. However, many commonalities across the federal credit programs could be handled more effectively by a centralized bank.

All of the direct loan programs have to perform the standard functions of

• informing prospective borrowers of the availability of credit and the terms on offer,
• evaluating loan applications,
• extending loans,
• collecting principal and interest payments,
• monitoring the financial health of borrowers, or at least their payment record,
• tracking collateral, for secured loans,
• intervening when borrowers run into trouble or fail to make required payments,
• managing and liquidating assets taken over as a result of defaults,
• monitoring potential fraudulent activity,
• performing strategic planning, and
• accounting for their present and future costs according to federal budget rules.

All guarantee programs have to ensure that the same steps are performed, although many of these would be delegated in the first instance to private lenders. In addition, the federal programs have to monitor their relationships with private lenders and the economic terms relevant to that relationship.

There could be significant efficiencies in standardizing the government’s approach to these functions and having common administration of each step, to the extent possible. Having a common agency that performed all these functions would also enhance the ability to apply best practices. One way to reduce the political problems, although it would hardly eliminate them, would be to continue with separate agencies to provide high-level management of each program but to move many of the standard services to the Federal Bank as a kind of “back office”, in industry parlance.

Focus More on Optimizing the Allocation of Money between Programs

Many federal credit programs are intended to serve the same broad purpose. For example, myriad programs were in place to deal with the credit crunch during the Great Recession, some administered through TARP and many at the Fed, as well as in other parts of the government. There does not appear to have been a systematic analytical approach to determining whether a dollar spent in one program would have more benefit than one spent in another. Similarly, the government has a host of housing-related programs, including Fannie Mae and Freddie Mac, the FHA, Ginnie Mae, housing loans at Department of Veterans Affairs, housing loans at the Department of Agriculture, and so on. They share the same underlying goal of making it easier for the middle and working classes to afford home ownership, yet there is little analysis done to determine which programs and subprograms across these agencies work most effectively. The allocation question ties into the related question of how budget costs should be calculated. The existence of differing methods across programs biases resource allocation in ways that create inefficiencies.

Implementation of the recommendation to create a single federal bank, would likely go a long way toward achieving such coordination. Failing that, the recommendation to force a formal proposal process, would to help to some degree. Perhaps beefing up the roles of the Congressional Budget Office and the Office of Management and Budget in reviewing the coordination of credit programs could assist further.

Spread Best Practices More Effectively Across the Government

Much of what each program does consists of basic banking activities. There are better and worse ways to do them in a federal government context. Unfortunately, little effort has been made to bring good ideas from one department to another or to spread the word of approaches that do not work. During the Clinton administration, the Office of Management and Budget created the Federal Credit Policy Working Group, which brought together representatives of the different programs to discuss common issues. The main focus at the time was implementation of the new Federal Credit Reform Act, but other issues were also hashed out. Perhaps there even needs to be a senior official with sole responsibility, and some authority, to coordinate at least the technical aspects of these programs.

A series of smaller steps could aid in emphasizing learning from other programs. For example, there could be special bonuses or other recognition for the five federal employees who have most effectively encouraged the spread of best practices. Encouraging movement of personnel across credit programs, perhaps tied to the enhanced training suggested below, would also help in this regard.

Improve the Compensation and Training of Federal Financial Workers

Many government departments face the problems of recruiting and retaining good workers in an environment in which the private sector can pay more. However, the disparity is far sharper in the financial area than in any other government activity, because of the generous sums paid by private sector financial firms for people with useful skills in this area. For this reason, the Federal Reserve and some of the regulatory bodies have the legal ability to pay more for certain positions than would be possible in other government departments. That flexibility should be extended to federal financial officials in the credit programs.

In addition, there ought to be an increased effort to improve the training of employees in these programs. To assist in this, it would be worthwhile to create a certification program for these employees that would be tied to higher pay and better career prospects.

The views expressed in this piece are those of the author and should not be attributed to the staff, officers or trustees of the Brookings Institution

---

¹ Gale (1991, p. 134; emphasis in original).

² This was an auction where the winning bidders were given the right to broadcast on specific parts of the electronic spectrum, often for mobile phone or other wireless communication uses.