Discussion: (0 comments)
There are no comments available.
A Cure That's Worse Than the Disease
View related content: Housing Finance
With the publication in early April of a Notice of Proposed Rulemaking (NPR) on the risk-retention requirement of the Dodd-Frank Act (DFA), we are beginning to see the outlines of the housing finance system the act envisions. If this proposed rule is adopted substantially as written, and there are no changes in the other provisions the act has added to the laws governing mortgage lending, the housing finance system of the future will place immense financial risks and regulatory costs on mortgage originators and securitizers, fail to prevent the growth of subprime and other low-quality lending, virtually ensure the continued existence of the government-sponsored enterprises (GSEs) Fannie Mae and Freddie Mac, impair the development of a robust -private-sector housing finance system in the United States, and provide insurmountable advantages for the largest banks in the limited private securitization system that might exist. These adverse consequences cannot be corrected through regulatory action, so the housing finance system envisioned by the DFA should be replaced by a plan that would define prime mortgages by statute.
Key points in this Outlook:
In late March, the bank and securities regulators charged with filling in the details of the DFA’s provisions on mortgage-market reform released a set of proposed rules outlining how the act’s broad terms will likely be implemented. The reaction in the market was something close to shock, as mortgage bankers, mortgage insurers, securitization specialists, and even the Center for Responsible Lending panned the rules in testimony before the House Financial Services Committee. Others not at the table were probably standing by in stunned silence. Given the many flaws in the hastily drafted and enacted DFA, this is no surprise. The wonder is why it took so long for the reality to set in.
The principal error of the act’s sponsors–in common with many on the left–was to see the decline in mortgage underwriting standards and the buildup of weak mortgages in the financial system as the result of greed among originators and securitizers, abetted by lax or nonexistent regulation. In this view, mortgage originators and securitizers–by making and distributing subprime and other risky loans–drove the growth of a housing bubble full of low-quality loans, eventually causing the 2008 financial crisis.
This view turns the market upside down; it fails to recognize that–in the mortgage market as in any market–suppliers respond to consumers. Loan originators, the suppliers, could not have made the weak and risky loans that caused the financial crisis unless the consumers–government or private-sector buyers and investors–wanted these instruments. With their aberrant view of how the housing finance market actually works, the framers of the DFA sought to impose controls on the suppliers, making it difficult or impossible for them to perform their regular market functions. The NPR makes things worse by adding complex and costly proposed regulations.
As I will show, all the major deficiencies of the DFA’s treatment of mortgage finance stem from the fundamental misperception that controlling the quality of mortgages requires controlling the process of origination and securitization through stringent regulation. If this regulatory system goes into full effect, it will not achieve the reforms its supporters have touted; instead, it will seriously impair the housing finance system and deprive many would-be homeowners of mortgage credit. As discussed below, because of the conceptual errors inherent in the DFA’s approach to housing finance reform, the sections of the DFA that address the mortgage market cannot be repaired by regulation; instead, they should be repealed. Mortgage quality can and should be controlled, but the most effective way to do this is to specify the constituents of a prime mortgage by statute. This is done in a plan outlined at the end of this Outlook.
The Influence of Demand on Mortgage Quality
The weak and risky mortgage loans that eventually brought down the US mortgage market and caused the financial crisis were the result of demand, not supply. Beginning with the affordable-housing requirements imposed on Fannie Mae and Freddie Mac in 1992, the US government’s housing policy created artificial demand for subprime and other risky mortgages that had previously been shunned by most investors. Before 1992, Fannie and Freddie were required by their charters to purchase only mortgages that were “acceptable to institutional lenders”–that is, prime mortgages as understood at the time. That traditional mortgage, the mortgage acceptable to institutional investors, had a down payment of 10 to 20 percent and was made to a borrower who had both a demonstrated record of meeting financial obligations and an income that promised a similar pattern after the mortgage obligation was assumed. Nontraditional mortgages (NTMs) included subprime loans (made to borrowers without good credit records) or otherwise deficient mortgages (for example, mortgages without substantial down payments or lacking income information).
Before 1992, NTM lending was a niche business, operated by relatively few lenders who bought and held NTM loans. These loans paid higher rates than traditional mortgages, and this compensated the lenders for the greater risks of default. In many cases, NTMs were sold to the Federal Housing Administration (FHA), the US government agency charged with providing mortgage credit to low-income borrowers. In 1992, this market was fundamentally changed when an affordable-housing mission was imposed on Fannie and Freddie, creating a substantial new source of demand. Although the FHA was a significant buyer of NTMs, it did not have access to as much financing as Fannie and Freddie. Because the GSEs were perceived in the market as backed by the government and had very low statutory capital requirements, they had almost unlimited access to funds.
Initially, the affordable-housing goals for Fannie and Freddie were modest. The statute required that 30 percent of all mortgages they bought from originators be “affordable”–defined generally as loans to borrowers at or below the median income in the areas where they lived. This was not a difficult goal to meet because about 30 percent of the loans that Fannie and Freddie routinely acquired were made to people at or below the median income in their neighborhoods. However, the Department of Housing and Urban Development (HUD) was given authority to administer the affordable-housing goals, and over time increased the goals substantially–to 42 percent in 1995, 50 percent in 2000, and 55 percent in 2007.
In other words, by 2000, at least 50 percent of all loans purchased by Fannie and Freddie had to be made to borrowers at or below the median income. In addition, HUD had created and increased various “subgoals”–so, for example, by 2005, 25 percent of the loans purchased by the GSEs had to be made, in part, to borrowers at or below 80 percent, and in part to borrowers at or below 60 percent, of the median income. Obviously, while it might be possible to find enough prime mortgages to meet a 30 percent affordable-housing goal, it was much more difficult to find prime mortgages among low- and moderate-income (LMI) borrowers, who often have weak credit histories and insufficient funds for down payments. When more than half of the mortgages Fannie and Freddie bought had to be made to LMI borrowers at or below–and in some cases well below–the median income where they lived, it was understandable that mortgage underwriting standards would deteriorate. The affordable-housing goals, in other words, created substantial demand for NTMs–loans that would allow Fannie and Freddie to comply with the goals.
But that was not all. The FHA was also looking for similar loans; insured banks and savings and loan associations were required to comply with the Community Reinvestment Act, which mandated lending to borrowers at or below 80 percent of the median income, and HUD started a program called the Best Practices Initiative, which attempted to get subprime and other lenders to reduce mortgage underwriting stand-ards so more low-income borrowers would have access to mortgage credit. Thus, by the early 2000s, these policies had created substantial demand for NTMs–demand that would not have existed but for government housing policies–and built a housing bubble of unprecedented size. All previous modern bubbles had lasted three or four years and increased real home prices by about 10 percent; by the early 2000s, however, this bubble was already five years old and eventually increased real housing prices by almost 90 percent.
The bubble created by this artificial government-mandated demand eventually became, itself, a significant source of demand. By their nature, bubbles suppress delinquencies and defaults. If a borrower cannot meet his or her mortgage obligations, the home can be refinanced or sold without a loss because it has increased in value. This means that by the early 2000s, when the massive housing bubble was already five years old, mortgage-backed securities (MBS) backed by NTMs looked like excellent investments. Because of their risk, NTMs were offering high yields, yet the data showed relatively few delinquencies and defaults. In other words, MBS backed by NTMs were offering very high risk-adjusted returns. This produced, for the first time, a securitization market based on NTMs–another market that would not have existed but for government housing policies–and a housing bubble of unprecedented size and duration.
This private-label NTM securitization market did not begin to grow significantly until the early 2000s; by 2002, the first year when it was larger than $100 billion, it amounted to only 4 percent of the market. Nevertheless, driven by the bubble, it grew to 15 percent of the market in 2004 and continued its explosive growth until flattening out and collapsing in 2007. At that point, the NTMs attributable directly to government housing policies, and held on the balance sheets of government agencies or firms required by government policies to purchase these loans, consisted of 19.2 million mortgages, with an aggregate value of approximately $2.7 trillion. At the same time, the NTMs attributable to the issuance of private MBS backed by NTMs amounted to 7.8 million loans with an aggregate value of $1.9 trillion. With the flattening of the bubble, NTMs on both government and private balance sheets began to fail in unprecedented numbers, driving down housing prices and collapsing the market for private MBS. This, in turn, coupled with the requirements of mark-to-market accounting, caused the weakness and instability among financial institutions, as large portions of the $1.9 trillion in private MBS had to be written down or became unusable for liquidity.
The lesson to be drawn from these developments is that demand–and not supply–is responsible for the quality of the mortgages that originators and securitizers produced. The NTMs that defaulted in the financial crisis would never have been made in such numbers without the demand created by government policies.
The DFA’s Approach: Misconceptions about the Mortgage Market
The drafters of the DFA failed to understand this process. They saw the housing finance market as driven not by buyers but by the mortgage originators–the sellers. In this narrative, mortgage originators, spurred by greed and not effectively regulated, produced the vast number of NTMs that eventually caused the financial crisis. It seems not to have occurred to the drafters that no mortgage, of any quality, would ever be made unless someone–an investor or lender–wanted to hold it. However, if one starts the analysis with the view that originators drove the process, then tighter regulation of mortgage origination and securitization becomes essential.
Thus, to address the poor quality of the mortgages in the financial system before the financial crisis, the DFA sponsors added the term “qualified mortgage” (QM) to the Truth in Lending Act and defined it in substantial part as a mortgage that a borrower–given all his or her financial circumstances–could afford. The DFA then provides that if an originator makes a mortgage loan that is not a QM, the borrower may assert this lapse as a defense to foreclosure against the originator and any subsequent holder of the mortgage. Although the Fed is authorized to make changes in this provision (which will eventually be interpreted and enforced by the Consumer Financial Protection Bureau), as long as the originator bears the onus for determining whether a mortgage is a QM, it will add substantially to the risks–and thus the costs–of holding mortgages or MBS. This provision, in other words, effectively poisons the well by infecting all mortgages with a deficiency that can never be effectively eliminated. This defense to foreclosure, based on the borrower’s financial circumstances years earlier, and involving the records of an originator who may no longer be in business, is a nightmare scenario. If the purpose of the DFA were to impair the housing finance process, it would be difficult to think of a more efficient way to accomplish it.
Another provision in the DFA that derives from the same distorted picture of the mortgage-financing process is the requirement that securitizers retain at least 5 percent of the risk of the MBS they securitize. The purpose of this requirement is to ensure that securitizers have “skin in the game” by taking a portion of the same risks they pass on to investors. The risk-retention idea arose from the notion that no one in the private mortgage securitization chain–characterized as the “originate to distribute model”–had an incentive to ensure that the mortgages underlying the MBS were good quality. This reflects a further misunderstanding of markets. Suppliers of goods and services do not need an incentive to distribute a quality product; every market participant knows that if he distributes an inferior product he will soon be out of business. In other words, securitizers had plenty of incentives–if they wanted to remain in business–to distribute quality MBS. (This is not to say there are no bad actors or fraudsters in the mortgage market; there are some in every market, but in general they are eventually discovered and driven out.) The reason that low-quality mortgages were distributed by securitizers before the financial crisis was that most securitizers–along with investors–did not understand the risks of MBS backed by NTMs because they did not understand how the growing bubble was suppressing delinquencies and defaults. This is shown by the fact that many financial institutions were taken down not because they sold MBS backed by low-quality NTMs but because they retained many of those MBS–believing they were good investments.
But the notion that securitizers were responsible for the low quality of the mortgages they distributed–a key tenet of the Left’s narrative about the financial crisis–led the framers of the DFA to focus their regulations on securitizers, rather than on the quality of the mortgages themselves. The animating idea was that if securitizers were required to take risks through the risk-retention device, they would seek out and securitize higher-quality mortgages. A corollary of this approach is that if very high-quality mortgages were securitized, no retention would be necessary–the MBS thus created would not involve significant risk. This was the logic behind creating an exemption from risk retention for the “qualified residential mortgage” (QRM), the terms of which were left to be defined by the bank and securities regulators.
The flaw in this approach is that a mortgage with essentially no risks is out of reach to most home buyers, so they will be left paying the higher costs associated with a mortgage that will carry a 5 percent risk-retention requirement. It was not a bad idea to try to control the quality of the mortgages that made their way into the securitization process, but the framers of the DFA–because of their distorted understanding of the financial crisis–attempted to apply the controls in the wrong place. (In the last section of this Outlook, I discuss a plan for housing market reform, which places controls directly on the mortgages entering the securitization system and thus avoids the problems associated with risk retention, the QRM, and the QM, all of which can then be repealed.)
Moreover, for the risk-retention idea to work, it has to be comprehensively applied–that is, there can be no way for a non-QRM mortgage to be securitized without the application of the 5 percent retainage. If not, subprime and other low-quality mortgages will always find their way into the securitization process. Unfortunately, this safeguard was not adopted. On the contrary, as discussed below, the DFA exempts the FHA from the risk-retention requirement and thus opens a loophole through which non-QRM mortgages–including subprime and other low-quality loans–can be securitized without any additional risk-retention burden. And the regulators, following the logic of the DFA, opened another even broader avenue for the securitization of low-quality loans through Fannie and Freddie. By using risk retention as the means to control mortgage quality, the act also created significant advantages for the largest banks, allowing them to eventually dominate whatever private securitization market develops in the future. Finally, maintaining the QRM standard required so much complex regulatory language in the NPR that there is virtually no way small originators can be expected to comply. They will be driven out of the origination business as effectively as small securitizers will be driven out of the securitization business.
The FHA Exempt from Risk Retention. The DFA exempts the FHA from the act’s risk-retention requirements. This is consistent with the underlying purpose of risk retention because the FHA is a government agency that is supposed to insure NTMs–its mission is to give LMI borrowers access to mortgage credit, even though they might not have access to the financial resources for a prime mortgage.
However, because Congress has never required the FHA to insure only mortgages to low-income borrowers–indeed, Congress has authorized the agency to insure mortgage loans up to $729,000–the FHA will ultimately become a preferred conduit for mortgages that do not meet the QRM standard. Over time, as always happens, Congress–responding to constituent and interest-group pressure–will reduce underwriting stand-ards for the FHA, allowing the agency to insure more non-QRM mortgages. Eventually, these lower-quality mortgages will default–as they did in triggering the 2008 financial crisis–and taxpayers will have to bear the costs. Thus, although it was attempting to increase the quality of mortgages through the risk-retention requirement, the DFA will have the (perhaps) unintended consequence of creating a new incentive for originating subprime mortgages and other NTMs.
Fannie and Freddie Likely Exempt from Risk Retention. In the NPR, regulators made the FHA problem considerably worse by exempting Fannie and Freddie from the risk-retention requirements. Again, however, this decision is perfectly consistent with the DFA’s rationale for imposing the risk-retention requirements in the first place; if the purpose of the risk-retention requirements is to encourage better-quality mortgages by requiring securitizers to take on a portion of the risk they are securitizing, Fannie and Freddie–which guarantee 100 percent of the promised payments of principal and interest on the pooled mortgages–were already taking 100 percent of the risk. As a policy matter, however, this decision is disastrous and inconsistent with the purported objectives of the act. It creates another even larger FHA-type incentive for originating low-quality, non-QRM mortgages. Again, as in the years leading up to the financial crisis, Fannie and Freddie could become a conduit for weaker and weaker mortgages and will be required by Congress to take on greater risk. Eventually, as they are today, they will become another source of bailout costs for taxpayers.
Even worse, the cost of the risk-retention requirement for privately securitized mortgages will widen the spread between taxpayer-subsidized GSE loans and nonsubsidized private-sector mortgages–this time by raising the costs of private-sector financing. As Tom Deutsch of the American Securitization Forum told Congress in recent testimony, “The capital markets need a clear signal from the government as to whether policy is, or is not, to encourage the return of a robust [residential] MBS market. These proposals . . . further [entrench] government involvement in the housing markets.” Direction from Congress is also necessary for the regulators who are designing the rules; the DFA does not declare whether Fannie and Freddie should be covered by the risk-retention requirements, so the regulators appear to have applied what they thought was Congress’s logic in deciding where risk retention was required. Without some further legislative direction from Congress on this subject, it is unlikely that the exemption of Fannie and Freddie from the risk-retention requirements will be changed in the final rule making, seriously impairing the chances that a robust private securitization market will develop.
Advantages for the Biggest Banks. The 5 percent risk-retention amount, which cannot be hedged or sold, can only be carried by a securitizer that has a substantial balance sheet. Thus, if there is a market of any size in mortgages that do not meet the QRM standard, it will likely be a hospitable environment only for the largest banks, which alone have the balance-sheet capacity to do a substantial business in securitization and still carry the risk-retention amount for an indefinite period.
As a matter of policy, this does not make sense. The securitization process should be open to as many players as wish to participate, so that competition will produce innovation, efficiency, and low prices. One of the attractive elements of securitization is that the securitizer need not be a large institution. The highly rated tranches in a securitization are not protected or guaranteed in any way by the securitizer but by the lower-rated tranches, which generally take the first loss. There are often buyers with a risk appetite for the lower-rated tranches, which generally offer high yields. In an unfettered market, this would open the business of securitization to many participants and widespread competition. The risk-retention requirement, however, ensures that only the largest banks will be securitizers.
Accordingly, if the DFA-designed system is allowed to go into effect, it will continue to permit the origination and securitization of subprime mortgages, impose substantial costs on the mortgage system through the QM requirement, make it more difficult–perhaps impossible–for a robust private market to develop, ensure the continued dominance of Fannie and Freddie, and provide substantial advantages for the largest banks in any non-QRM securitization market–including any jumbo market–that does develop in the future. Many of these deficiencies arise from the underlying concepts embodied in the DFA and cannot be fixed by regulation.
Burdens for Small Originators. As described above, the consequences of failing the new QM test–originating a loan that the borrower cannot afford–are dire. The borrower then has a defense against foreclosure at some point in the future. It appears, in this connection, that the failure to meet the requirements for a QRM will also be a failure to meet the requirements for a QM. For example, the DFA requires that “documentation and verification of the financial resources relied upon to qualify the mortgagor” are necessary to meet the QRM standards. It is also necessary, under the QM requirements, that the originator verify the income of the borrower to show that the mortgage was affordable. All well and good, except that the NPR contains thirty-five pages of rules on how to determine the income of a borrower. Obviously, without an experienced lawyer always at the originator’s side, there is a serious danger that the mortgage will not qualify for QRM treatment, and if it does not, it may also fail to qualify for QM treatment–thus giving the borrower a defense to foreclosure. Apart from that serious deficiency, it will be difficult for any small originator to satisfy a securitizer that a particular mortgage meets the test for either a QRM or a QM. The risk to the securitizer will be too great to rely on the ability of a small originator, with a small staff, to vet a borrower in the detail required by thirty-five pages of rules on income determination.
The Right Way to Control Mortgage Quality Deterioration
Experience tells us that mortgage lending and particularly securitization are susceptible to a gradual weakening in underwriting standards. This is because housing prices frequently become a source of speculative investment, causing booms and busts. As housing prices rise in bubbles or booms, they tend to obscure delinquencies and defaults, and both lenders and investors come to believe that “this time it’s different” or that continued rising prices will compensate for weak underwriting standards. Thus, the AEI white paper on housing finance, presented in March 2011, specifies that only prime mortgages (defined by statute) would be eligible for securitization. If this idea is adopted, it would ensure a stable mortgage market in the future, the privatization or other elimination of Fannie and Freddie, and the repeal of the risk-retention requirement and both the QM and QRM requirements, which would no longer be necessary.
By defining a prime mortgage by statute, the plan developed at AEI avoids the problem of mortgage underwriting deterioration as a bubble grows. Indeed, it will tend to suppress bubbles. The DFA approach, which focuses on the regulation of originators and securitizers, does not prevent the deterioration of mortgage underwriting standards–first because it opens a loophole through the FHA and the GSEs, and second because a 5 percent retention may not seem particularly risky if a future bubble is suppressing defaults.
Under this plan, a prime mortgage would have all the following elements, where applicable:
The role of mortgage insurance is important in our prime-mortgage criteria, so we would also define the required elements of mortgage insurance:
|Amortized LTV (%)||Suggested risk-to-capital|
ratio for thirty-year
|Current risk-to-capital ratio|
|92.51-95.00||8 to 1||25 to 1|
|90.01-92.50||10 to 1||25 to 1|
|85.01-90.00||13 to 1||25 to 1|
|80.01-85.00||16 to 1||25 to 1|
|75.01-80.00||29 to 1||25 to 1|
|70.01-75.00||31 to 1||25 to 1|
|65.01-70.00||38 to 1||25 to 1|
|60.01-65.00||41 to 1||25 to 1|
A mortgage that meets these standards would be competitive with a Fannie Mae mortgage. The discussions my coauthors and I had with institutional investors, securitizers, and mortgage insurers, using the foregoing criteria, indicate that a thirty-year fixed-rate freely prepayable conventional mortgage would be priced at 25 to 40 basis points above the equivalent Fannie Mae mortgage. If the Federal Housing Finance Agency raises the guarantee fees for Fannie and Freddie, this would further narrow the spread between the AEI white paper’s thirty-year fixed-rate example and its Fannie equivalent. Since the Obama administration itself, in its February 11 white paper, said that mortgage rates will have to increase with the elimination of Fannie and Freddie, we believe our proposal offers a superior and stable housing finance system, without any risk to taxpayers.
Other elements of the plan developed at AEI include:
The housing finance plan outlined in the DFA is largely unworkable. If it goes into effect, the risks it imposes on regulators will substantially raise mortgage costs, and the risk-retention and QRM provisions will not have the intended effect of reducing the origination of subprime or other weak and risky mortgages. Moreover, the act will impede the development of a robust private securitization market and, to the extent that a private market develops, ensure that the largest banks continue to dominate it. For all these reasons, the DFA’s housing market finance provisions should be repealed and replaced with a nongovernmental plan, along the lines of the plan summarized in this Outlook, that relies primarily on a statutorily defined prime mortgage to prevent the deterioration of mortgage underwriting standards in the future.
Peter J. Wallison ([email protected]) is the Arthur F. Burns Fellow in Financial Policy Studies at AEI.
1. Understanding the Implications and Consequences of the Proposed Rule on Risk Retention, Before the House Committee on Financial Services Subcommittee on Capital Markets and Government-Sponsored Enterprises, 112th Cong. (April 14, 2011), http://financialservices.house.gov/Hearings/hearingDetails.aspx?NewsID=1852 (accessed April 27, 2011).
2. For a full discussion of the affordable-housing requirements and the general effect of government housing policy on the growth of NTMs in the 1990s, see Peter J. Wallison, Dissent from the Majority Report of the Financial Crisis Inquiry Commission (Washington, DC: AEI, January 2011), www.aei.org/paper/100190.
3. Ibid., 16, figure 1.
4. Wall Street Reform and Consumer Protection Act, HR 4173, 111th Cong. (2010), §1413.
5. Understanding the Implications and Consequences of the Proposed Rule on Risk Retention, Before the House Committee on Financial Services Subcommittee on Capital Markets and Government-Sponsored Enterprises (statement of Tom Deutsch, Executive Director of the American Securitization Forum).
6. § 941(b)(4)(B)(i).
7. Peter J. Wallison, Edward Pinto, and Alex J. Pollock, “Taking the Government Out of Housing Finance: Principles for Reforming the Housing Finance Market” (AEI Policy White Paper, Washington, DC, March 24, 2011), www.aei.org/paper /100206.
8. In 1991, over 98 percent of Fannie’s loans met this standard. Data are from Fannie Mae’s random sample review covering single-family acquisitions for the period October 1988-January 1992, dated March 10, 1992 (document in files of Edward Pinto).
9. Ibid. In 1991, over 91 percent of Fannie’s home-purchase loans had LTVs of 90 percent or lower.
10. Ibid. In 1991, over 93 percent of Fannie’s loans had LTVs of 80 percent or lower.
11. Ibid. In 1991, over 92 percent of Fannie’s loans had LTVs of 75 percent or lower.
12. Ibid. In 1991, over 98 percent of Fannie’s loans had one or no mortgage late payments at origination and 85 percent had two or fewer nonmortgage late payments at origination.
13. Ibid. In 1991, over 90 percent of Fannie’s loans met this standard.
14. Fannie stopped acquiring low-documentation or no-documentation loans in 1990. Freddie followed in 1991. See “Haste Makes . . . Quick Home Loans Have Quickly Become Another Banking Mess,” Wall Street Journal, July 5, 1991.
15. Fixed-rate loans with shorter amortization periods pose a lower risk of default due to faster buildup of borrower equity and therefore would have somewhat higher risk-to-capital requirements (requirements that less capital be held). For example, fifteen-year term loans at an 80 percent LTV might have a 38:1 risk-to-capital ratio, the same as for a 70 percent LTV loan with a thirty-year term.
16. Coverage must be maintained until the original loan balance amortizes to 60 percent based on the lesser of original sales price (if applicable) or original appraised value.
There are no comments available.
1150 17th Street, N.W. Washington, D.C. 20036
© 2016 American Enterprise Institute for Public Policy Research