The Effect of Refinancing Waves on the Borrowing Costs of Mortgage GSEs. January, 2005



Similar documents
Mortgage-Related Securities

HUD s AFFORDABLE LENDING GOALS FOR FANNIE MAE AND FREDDIE MAC

Regulation of the Housing Government-Sponsored Enterprises

P r i m e r : Agency Bonds

A PRIMER ON THE SECONDARY MORTGAGE MARKET

EXECUTIVE OFFICE OF THE PRESIDENT COUNCIL OF ECONOMIC ADVISERS CEA NOTES ON REFINANCING ACTIVITY AND MORTGAGE RATES

Mortgage-Backed Sector of the Bond Market

Chapter 45. Primary and Secondary Mortgage Markets INTRODUCTION

US TREASURY SECURITIES - Issued by the U.S. Treasury Department and guaranteed by the full faith and credit of the United States Government.

GOVERNMENT-SPONSORED ENTERPRISES

Mortgage-backed Securities

Managing the Investment Portfolio

Mortgage Backed Securities. Masaryk University Brno, CZ

Chapter 10. The Good Old Days. The New Way. Secondary Markets. Depository Lenders in the Primary Market. Nondepository Lenders in the Primary Market

GAO SMALL BUSINESS ADMINISTRATION. Secondary Market for Guaranteed Portions of 7(a) Loans

GOVERNMENT-SPONSORED ENTERPRISES

6/18/2015. Sources of Funds for Residential Mortgages

Secondary Mortgage Market Policy Fannie Mae to QRM. Kevin Park PLAN 761 September 19, 2012

Evergreen INSTITUTIONAL MONEY MARKET FUNDS. Prospectus July 1, 2009

Have the GSE Affordable Housing Goals Increased. the Supply of Mortgage Credit?

Overview of Mortgage Lending

MORTGAGE BACKED SECURITIES

Recourse vs. Nonrecourse: Commercial Real Estate Financing Which One is Right for You?

Opening Doors For Muslim Families In America

Answers to Chapter 7 Questions

THE BERWYN FUNDS. Shareholder Services Ultimus Fund Solutions, LLC P.O. Box Cincinnati, Ohio

Mortgages and Mortgage -Backed Securiti curi es ti Mortgage ort gage securitized mortgage- backed securities (MBSs) Primary Pri mary Mortgage Market

Important Information about Investing in Bonds

Lecture Notes on MONEY, BANKING, AND FINANCIAL MARKETS. Peter N. Ireland Department of Economics Boston College.

INTEREST RATES: WHAT GOES UP MUST COME DOWN

Northwest Bancshares, Inc. Announces Quarterly Earnings and Dividend Declaration

Assumable mortgage: A mortgage that can be transferred from a seller to a buyer. The buyer then takes over payment of an existing loan.

Residential Mortgage Finance. Early American Mortgages. Early Mortgage Lenders

The Mortgage Market. Concepts and Buzzwords. Readings. Tuckman, chapter 21.

RBC Money Market Funds Prospectus

Measuring the Benefits of Fannie Mae and Freddie Mac to Consumers: Between De Minimis and Small?

Structured Financial Products

Financing Residential Real Estate: SAFE Comprehensive 20 Hours

Mortgage Lending Basics

Rethinking Fixed Income:

Chapter 10. Fixed Income Markets. Fixed-Income Securities

Optimize Your Liquidity and Profitability. Create Investment Securities from Your Mortgage Pipeline

Unit 1 Overview of the Mortgage Markets

Testimony of Anthony P. Costa. On behalf of the. American Bankers Association. before the. Subcommittee on Oversight and Investigations.

Answer Outline. ECONOMICS 353 L. Tesfatsion/Fall 06 EXERCISE 6: Six Questions (8 Points Total) DUE: Tues, October 10, 2006, 2:10pm

Weekly Relative Value

ALLOCATION STRATEGIES A, C, & I SHARES PROSPECTUS August 1, 2015

Understanding Fixed Income

Private Mortgage Insurance (PMI)

Chapter 3 Fixed Income Securities

City of Bloomington, Minnesota Investment Policy

Housing markets and economic growth: lessons from the US refinancing boom 1

Timothy K. Zimmerman President and CEO Of Standard Bank Monroeville, PA

Prospectus Socially Responsible Funds

GLOSSARY OF TERMS. Amortization Repayment of a debt in regular installments of principal and interest, rather than interest only payments

Federated Quality Bond Fund II

DFA INVESTMENT DIMENSIONS GROUP INC.

October 20, Benefits of FRMs

Mortgage loans and mortgage-backed securities

Bond Mutual Funds. a guide to. A bond mutual fund is an investment company. that pools money from shareholders and invests

S Responsible Homeowner Refinancing Act of 2012

The GSEs Are Helping to Stabilize an Unstable Mortgage Market

SINGLE-FAMILY CREDIT RISK TRANSFER PROGRESS REPORT June Page Footer. Division of Housing Mission and Goals

Statement of. David Hehman President and CEO Federal Home Loan Bank of Cincinnati. Before the. House Financial Services Committee

How To Sell A Callable Bond

- Short term notes (bonds) Maturities of 1-4 years - Medium-term notes/bonds Maturities of 5-10 years - Long-term bonds Maturities of years

PRIMING THE PUMP: PROGRAMS TO ENCOURAGE LENDERS TO LEND. By: Sheelagh Allston, Esq.

Summary of the Housing and Economic Recovery Act of 2008

CIO WM Research 22 October 2014

INVESTMENTS HELD IN DISTRESSED FINANCIAL INSTITUTIONS AND STOCKBROKERS: A USER S GUIDE

CFPB Issues Much Anticipated Final Rules: Ability to Repay, Qualified Mortgages, Escrow Requirements and Homeownership Counseling

GLOSSARY OF INVESTMENT-RELATED TERMS FOR NATIONAL ELECTRICAL ANNUITY PLAN PARTICIPANTS

Enterprise Risk Management: Auto Insurers

Report of Audit OFFICE OF INSPECTOR GENERAL. Approval Process of Funding Corporation Debt Issuances A Tammy Rapp Auditor-in-Charge

Chapter 3. How Securities are Traded

5 Common Types of Home Loans

Mortgage Credit in the USA and Denmark

House Committee on Financial Services. November 29, 2012

Financial-Institutions Management. Solutions 6

PASS~THROUGH AND COLLATERALIZED

Sankaty Advisors, LLC

Small Balance Loans. July 2015

Community Investments Vol. 15, Issue 2 Ginnie Mae Project Loans Maintain Affordability

STAFF REPORT: ENHANCING DISCLOSURE IN THE MORTGAGE-BACKED SECURITIES MARKETS

Corporate System Resolution Cause of the Corporate System Crisis Frequently Asked Questions (FAQs)

MISSION OF THE FARM CREDIT ADMINISTRATION

Investment insight. Fixed income the what, when, where, why and how TABLE 1: DIFFERENT TYPES OF FIXED INCOME SECURITIES. What is fixed income?

Exit Strategies for Fixed Rate Financing. Comparing Yield Maintenance and Defeasance Alternatives. by Regan Campbell and Jehane Walsh

Glossary of Common Derivatives Terms

Catastrophic Mortgage Insurance and the Reform of Fannie Mae and Freddie Mac

Opportunities in Mortgage Derivatives

A LEADER IN MULTIFAMILY AND HEALTHCARE LENDING

MULTIPLE CHOICE. Choose the one alternative that best completes the statement or answers the question.

KBW Mortgage Finance Conference. June 2, 2015

F INANCIAL S TATEMENTS. Brazos Student Finance Corporation Year Ended June 30, 2014 With Independent Auditors Report

Fin 5413: Chapter 8 Mortgage Underwriting

GNMA Mortgage-Backed Securities: A Treasury Alternative Offering Quality and Yield

PROFESSIONAL FIXED-INCOME MANAGEMENT

Financing Residential Real Estate. Lesson 12: VA-Guaranteed Loans

MORTGAGE BANKING TERMS

Transcription:

The Effect of Refinancing Waves on the Borrowing Costs of Mortgage GSEs January, 2005 Timothy R. Burch Associate Professor of Finance University of Miami Dept. of Finance P.O. Box 248094 5250 University Dr., Jenkins Bldg. Coral Gables, FL 33124-6552 ph: (305) 284-1509 fax: (305) 284-4800 email: tburch@miami.edu web: http://www.bus.miami.edu/~tburch Andrea J. Heuson Associate Professor of Finance University of Miami Dept. of Finance P.O. Box 248094 5250 University Dr., Jenkins Bldg. Coral Gables, FL 33124-6552 ph: (305) 284-1866 fax: (305) 284-4800 email: aheuson@miami.edu

The Effect of Refinancing Waves on the Borrowing Costs of Mortgage GSEs Abstract In the past decade two important trends have impacted the trillion-plus dollar single family mortgage market. One is the debt-financed accumulation of large quantities of mortgage pass-through securities by Fannie Mae and Freddie Mac, two mortgage-oriented government-sponsored entities who also act as guarantors that the primary mortgage market servicer will deliver the payments received from the mortgage borrower. The second is the increased propensity of mortgage borrowers to prepay and refinance the underlying mortgage loans as soon as primary market rates fall sufficiently below existing loan rates. The prepayment phenomenon asymmetrically increases the interest risk faced by the ultimate lender because although fixed rate funding costs do not drop when the asset s return does, variable rate funding costs rise with market rates while loan rates are fixed. While both of these phenomena have received a great deal of academic and industry attention, one area that has not been studied is the impact of refinancing waves on the borrowing costs of the mortgage GSEs. Using a large sample of capital market deals, we show that mortgage refinancing has led to significant increases in the borrowing costs faced by Fannie Mae and Freddie Mac relative to those of other GSEs. The increase is concentrated in the intermediate term debt maturities that most closely match the expected duration of single family mortgage loans. Furthermore, we present evidence that mortgage refinancing waves are followed by periods where coupon rates on the debt of these enterprises are elevated. 2

The Effect of Refinancing Waves on the Borrowing Costs of Mortgage GSE 1 s Introduction The Federal National Mortgage Association ( Fannie Mae ) and the Federal Home Loan Mortgage Corporation ( Freddie Mac ), entities which we refer to as mortgage GSEs, are federally-sponsored financial institutions charged by Congress with improving efficiency and liquidity in the residential mortgage market and increasing access to mortgage credit for under-served segments of the population. In the past the two institutions acted primarily as guarantors on pools of single family mortgage loans that were sold to long-term institutional investors. In recent months Fannie Mae and Freddie Mac have made headlines for understating earnings due to a desire to smooth volatility and then overstating earnings due to accounting irregularities regarding their use of interest rate derivatives to hedge retained mortgage portfolios. These portfolios have been growing at a rapid, and some say alarming, rate since the mid 1990 s. The strategic decision to accumulate predominantly fixed rate mortgage assets has been scrutinized by Wall Street and the Office of Federal Housing Enterprise Oversight (OFHEO). 2 The increase in the retained mortgage portfolios of the mortgage GSEs has been funded by a concomitant increase in interest bearing liabilities. The two agencies have accessed the capital markets more frequently as a result. These developments have potentially affected the financial position of these borrowers in at least two distinct ways: 1) the amount of underwriting fees paid, and 2) the coupon rate attached to the debt issued. At the same time that Fannie Mae and Freddie Mac have been accumulating large portfolios of fixed-rate, single-family mortgage loans, declines in market interest rates have led to waves of mortgage refinancings. Refinancing volume in excess of 70% of all primary mortgage market originations occurred 1 Government Sponsored Enterprises, or GSEs are federally chartered agencies 2 OFHEO is a government sponsored overseer set up by Congress in 1992 to monitor the activities of the two publicly traded but quasi-governmental entities. 3

for prolonged periods in late 1995, 1998 and 2001. It is a well known fact that fixed rate mortgage lending financed by fixed rate long term debt is subject to significant, asymmetrical interest rate risk. Mortgage borrowers are free to refinance without penalty if interest rates drop, but increases in market rates must be absorbed by the lender. In this paper, we utilize a large, detailed data base containing information on issue and underwriting characteristics for borrowings by both financial firms and GSEs from 1994 to 2001 3. We show that the mortgage refinancing phenomenon led to relative increases in the underwriting spreads faced by Fannie Mae and Freddie Mac and, to a lesser extent, by the Federal Home Loan Bank Board (FHLB), a non-public entity that makes short-term and long-term funding advances to commercial banks (many of which are also involved in originating and holding fixed rate mortgage-backed securities). The fee increases, which were not experienced by the other four GSEs in the data set, were concentrated in the intermediate term (2 year to 10 year maturity) debt issues. Notably, this range also reflects the expected duration of single family mortgage loans originated during the sample period. In addition, we provide preliminary evidence that the coupon rates attached to the debt of Fannie Mae, Freddie Mac and the FHLB, set with the advice and consent of the banks underwriting the issues, were often elevated significantly in the months after the refinancing waves for these GSEs. Our results suggest that underwriters and investors reacted to the increased uncertainty about the debt issues by increasing both underwriting fees and borrowing costs. In the next section of the paper we provide a brief review of existing literature that analyzes the determinants of underwriting spreads in debt markets and the current financial position of the GSEs. Section 2 presents our empirical results on the determinants of underwriting spreads while Section 3 contains graphical evidence on the relationship between mortgage refinancing waves and coupon rates paid 3 The five additional GSEs studied are the Federal Home Loan Bank Board (FHLB), the Student Loan Marketing Association (Sallie Mae), the Farm Credit Bank (FarmCr), the Farmers Home Administration (FarmAg), and the Government National Mortgage Association (Ginne Mae). 4

by the three institutions on which we focus. To our knowledge, there is no research available that compares underwriting spreads faced by GSEs to those charged to other financial firms, or that focuses on the combined impact of the strategic decision to portfolio fixed rate loans during refinancing waves on the borrowing costs faced by these GSEs. Our results show that the net result of internal GSE decisions and market developments has been a significant increase in both the up-front and long-term financing fees paid by these particular entities. I. Determinants of Underwriting Spreads and the Single Family Mortgage Market Securities underwriters are investment and commercial bankers who help borrowers market their new debt and equity issues to private and institutional investors for a fee. 4 This fee, defined as the spread between the face amount issued and the proceeds actually received by the borrower, as a percentage of the face amount, ostensibly contains two components: distribution and information costs for publicizing the issue and the issuer and the risk contained in bringing the issue to market. The underwriter is also intimately involved in setting the coupon rate of the forthcoming issue a few days before it hits the market, with the goal of choosing the coupon so that the security floats at a price as close to par as possible. Several recent studies address the determinants of the underwriting spread faced by a given issuer. Burch, Nanda and Warther (2005) and Gande, Puri and Saunders (1999) demonstrate the importance of borrower risk factors, documenting a positive relationship between the underwriting spread and issue maturity and the credit rating of the issuer. In addition, Altinkilic and Hansen (2000) demonstrate that initial economies of scale in issue size are eventually outweighed by risk considerations so that spreads rise eventually with size for large debt issues. Burch et. al. address the distributional impact of offer size issue in another way and show that syndicated offerings, which are typically so large as to require more than one 4 The provisions of the Glass-Steagall Act of 1933 effectively prohibited commercial banks from underwriting corporate debt issues until 1987, when commercial banks were allowed establish subsidiaries to engage in a limited amount of underwriting activities. In 1996, the Federal Reserve raised the limit substantially and commercial bank participation in the market for underwriting corporate debt has increased substantially since that time. 5

underwriter, generate higher fees while shelf registrations, which are prepared when convenient for the borrower and floated as needed, also are associated with marginally inflated spreads. Furthermore, evidence in Gandi et. al. documents a secular decline in underwriter spreads on debt issues as commercial banks entered the markets in force in the mid 1990s and a general decline over time as technological developments have lowered information costs. Finally, Drucker and Puri (2004) demonstrate that the ability of both commercial and investment banks to loan money to the corporate borrowers whose debt securities they underwrite leads to discounts in floatation costs. These studies examine debt issues by non-financial corporations, however, and do not consider the unique relationship between Government Sponsored Entities (GSEs) and capital markets. There are seven major GSEs charged with improving efficiency and liquidity in a different area of the overall economy: 1) Freddie Mac- conventional loan residential mortgage market pass through guarantor, 2) Fannie Mae- conventional loan residential mortgage market pass through guarantor, 3) FHLB advances to commercial banks, 4) Ginnie Mae (Government National Mortgage Association) FHA/VA loan pass through guarantor, 5) Sallie Mae (Student Loan Marketing Association) student loans for college education, 6) Farmer Mac (Farm Credit Banks) agricultural lending support 7) Farmers Home Administration rural lending. Of these, only Fannie Mae and Freddie Mac are currently publicly traded companies, although there are plans are in place to privatize Sallie Mae. Although the debt of these entities is not formally guaranteed by the full faith and credit of the U.S. government, credit markets behave as if they expect the Treasury to act as a lender of last resort for all of these institutions. In the mid-1990s, administrators at Fannie Mae and Freddie Mac made a strategic decision to begin to hold substantial debt-financed portfolios of the pass-through securities for which they act as guarantors. As shown in Figure 1, the pace of this activity increased in 1997 and again in 2000. Most of these assets 6

are held in securities backed by fixed rate residential mortgage which subject the holder to welldocumented asymmetric interest rate exposure. Specifically, the loans can be refinanced by the underlying borrower at any time, which complicates the process of funding the assets by fixed rate debt that cannot be called without penalty or premium. On the other hand, funding the assets with variable rate or short-term debt is equally risky because the yield earned on the mortgage assets is fixed if market rates rise. As interest rates have dropped since the beginning of the 1990s, residential borrowers have shown an increasing propensity to refinance in reaction to smaller and smaller declines in primary mortgage rates. Figure 3 shows prolonged periods in late 1995, 1998 and 2001 where more than 70% of new loan originations were classified as refinancings. This increased prepayment elasticity is likely due to technological improvements which shorten and streamline the refinancing process and an aggressive television, print and internet marketing campaign supported by the mortgage banking industry that reaps significant up-front refinancing fees. Fannie Mae and Freddie Mac s acquisition and hedging activities have had such a dramatic impact on U.S. financial markets that in 1992 Congress established an independent regulator, the Office of Federal Housing Enterprise Oversight (OFHEO), to supervise them. Despite the extra scrutiny, both firms were accused of understating earnings in 1999 and 2000 in an apparent effort to smooth out volatility in their income streams. In addition, Fannie Mae was recently charged with overstating earnings and failing to account correctly for the interest rate risk of the retained mortgage portfolio. while OFHEO s administrators successfully pressured Fannie Mae s Chairman and CFO to resign in December 2004. Academic research has attempted to document the impact of Fannie Mae and Freddie Mac s acquisition activities on the residential mortgage market. Recent apolitical empirical analysis indicates that GSE activity has not dramatically lowered the mortgage yield faced by underlying borrowers. For example, Ambrose, Lacour-Little and Sanders (2004) find a differential of only about 10 basis points between loans that conform to Fannie Mae and Freddie Mac s secondary market sale guidelines and those that do not. On the other hand, studies of the liability side of the mortgage GSE s balance sheets suggest 7

that the entities enjoy the same significant funding advantage that benefits all of the other Federal agencies. Specifically, recent work by Ambrose and Warga (2002) finds that offering yields for Fannie Mae and Freddie Mac debt lie about 30 basis points below that on AA banking sector bonds. Two things are clear from the discussion above. One is that the assets held by the two entities are subject to significant asymmetric interest rate risk. Secondly, the increase in debt financing that goes along with the increase in mortgage assets has resulted in more frequent interactions with underwriters and capital markets. 2. Determinants of Underwriting Spreads We wish to determine whether or not the two phenomena discussed above, (i.e. the debt-financed acquisition of fixed rate mortgage backed assets by Fannie Mae and Freddie Mac and the increased propensity of mortgage borrowers to refinance loans in waves as market rates drop) has impacted either the flotation or borrowing costs of the mortgage GSEs. In order to implement the desired analysis we collect coupon rate, term to maturity, S & P credit rating and issue size for the debt issues of all financial firms and GSEs from the new issues section of the Thomson Financial SDC Platinum database (SDC) for the years 1994 2001, a total of more than 31,000 entries. 5 The SDC data also gives us important details about the offering such as whether or not it is syndicated or shelf registered, and how many lead underwriters are involved. That data is used as the input matrix for the following OLS regression: PERSPR i,t = α 0 + Σ j β j * ISSUE CONTROLS + Σ k β k * UNDERWRITER CONTROLS+ Σ m β m * GSE CONTROLS + Σ n β n * GSE REFINANCING INTERACTIONS + ε I,t (1) 5 We would prefer to examine the issue s yield to maturity but this information is missing for a large proportion of the observations. The sample contains 7926 syndicated offerings and 5392 shelf registrations. The number of lead underwriters is typically one but can be as high as six. Coupon rates ranges from zero to 17% and maturities range from one month to 99 years. Issue size averages $45 million with a low of $100,000 and a high of $6 billion. Underwriting fees average $300,000 with a low of $50 and a high of $60 million. In percentage terms, the average spread is.3% with a minimum of one tenth of one basis point and a maximum of.47%. The number of observations attributable to each of the GSEs are as follows: Fannie Mae 3745, Farm Credit Banks 933, Farmer s Home Administration 41, Freddie Mac 3895, Sallie Mae 256, Ginnie Mae 3 and Federal Home Loan Bank Board 16041. 8

where: PERSPR = the underwriter s fee as a percentage of the net proceeds of the issue, and the ISSUE CONTROLS include: LSIZE: LSIZESQ: MAT: COUP: SPDUM: SHELF: the log of the offer size, in millions the square of the log of the offer size the maturity of the issue, in years the coupon rate attached to the issue a linear representation of the issuer s credit rating which begins with 1 as the best rating a dichotomous variable that is one if the offering is a shelf registration TIME: a linear time trend that starts at one in 1994 and the UNDERWRITER CONTROLS include: NUMLD: SYN: the number of lead underwriters a dichotomous variable that is one if the offering in syndicated and the GSE CONTROLS include: FAN: FARMCR: FARMAG: FRED: SAL GIN: FHLB: a dichotomous variable that is one if Fannie Mae is the issuer a dichotomous variable that is one if Farmer Mac is the issuer a dichotomous variable that is one if the Farmer s Home Administration is the issuer dichotomous variable that is one if Freddie Mac is the issuer dichotomous variable that is one if Sallie Mae is the issuer dichotomous variable that is one if Ginnie Mae is the issuer dichotomous variable that is one if the Federal Home Loan Bank is the issuer. The test variables of interest are composed of a matrix of interaction terms that each measure the product of the individual GSE dichotomous control variable and the percentage of loan originations classified as refinancings, (See Figure 2). 9

Given the results of previous research we expect underwriting spreads to decline with issue size in the region where economies of scale are present but increase for large issues, (LSIZE < 0 and LSIZESQ > 0), and to increase with maturity, coupon rate and issuer credit rating because of borrower risk considerations (MAT > 0, COUP > 0 and SPDUM > 0). Shelf registrations will require higher fees if the convenience aspect of the arrangement is priced by underwriters while technological improvements in information processing costs across the sample period suggest a decreasing trend, (SHELF > 0 and TIME < 0). If issues that require multiple lead underwriters or syndicates carry increased marketing costs then NUMLD and SYN will be positive. Furthermore, if the implicit government guarantee of the creditworthiness of its GSEs lessens their risk relative to private firms, all of the GSE control dummies will be negative. Turning to the interaction terms, we hypothesize that increased uncertainty surrounding the duration of the mortgage GSE s retained loans and the increased pace of mortgage acquisitions across the sample period will combine to diminish the marketability of mortgage GSE debt relative to debt of other issuers. If so, the Fannie Mae and Freddie Mac interaction terms will be positive. The other interaction terms will be insignificant unless mortgage refinancing activity is a broader measure of economic uncertainty than we expect. OLS regression results for the full sample and sub-samples segmented by maturity appear in Table 1 and conform to our expectations in most cases. Underwriting spread declines with issue size throughout the sample but eventually increases for large issues in the shorter maturity segments. The spread increases with maturity for all but the shortest term and increases with coupon rate for three to ten year issues. These results combined with the significant positive coefficient on issuer credit rating (S&P Dum) demonstrate the importance of borrower risk characteristics on marketability. The negative time coefficient for the full sample and the one to five year segments indicates that technological improvements in information processing and transfer have decreased floatation costs to some extent. 10

In contrast to earlier work, we find that shelf registered issues carry lower underwriting fees in the full sample and the three to twenty year maturity range. We concur in finding that syndicated issues and those with multiple lead underwriters require higher underwriting fees, however. New results attributable to this research appear in the GSE controls section (Panel C of Table 1) where Fannie Mae, the Farm Credit Banks, Freddie Mac, Sallie Mae and the Federal Home Loan Bank Board are shown to pay significantly lower underwriting fees than either Ginnie Mae, the Farmer s Home Administration or other financial firms. The results for the maturity based sub-samples indicate that the floatation costs savings in concentrated in the one to twenty year maturities, as opposed to very short or very long term debt. Turning now to the test variables, coefficients on the series of interaction terms will determine whether or not the underwriting fees paid by the mortgage GSEs change in a relative sense as the volume of mortgage refinancing activity ebbs and flows. Interest rate volatility impacts the value of mortgage-backed assets asymmetrically, (the present value of expected cash flows drops when rates rise but does not rise when rates drop because the increased probability of a refinancing wave lowers expected duration) and underwriting fees reflect the impact of borrower risk on issue marketability. Therefore, we expect the coefficients on the interaction terms, (which are measured in basis points, as opposed to the percent format in the rest of the Table) on the Fannie Mae and Freddie Mac variables to be positive. Inspection of Panel D of Table 1 reveals this to be the case for the full sample and the five to ten and three to ten year maturity segments respectively. This range also represents the expected duration of the 15- and 30-year mortgage loans that dominate the fixed rate mortgage market. Another notable result appears in the last line of the table, which shows that borrowing by the Federal Home Loan Bank Board (the entity that accounts for over half of the issues in our sample) requires increased underwriter fees in the two to ten year maturity segments. The Bank Board advances short and long term funds to commercial banks and these banks hold a significant percentage of their long term assets in mortgage-backed securities. The finding that a non-mortgage GSE intimately connected to the banking 11

market is affected by the uncertainty caused by waves of refinancing illustrates the pervasiveness of the influence of the mortgage market. 3. The Impact of Refinancing on Coupon Rate The data also allows us to address the impact of mortgage refinancing on the coupon rate paid on borrowing by the mortgage GSEs and the Bank Board. To do so, we first alter the regression model of Equation (1) to make the coupon rate the independent variable. We retain all of the other issue controls except TIME, the underwriting controls and the refinancing interaction terms. In order to abstract from the general decline in coupon rates across the sample period, we segment the data by issue year and month and then capture the t-statistics on the Fannie Mae, Freddie Mac and Fhlb interaction terms to generate a graphical representation of the relative increases and decreases in coupon rate as refinancing activity rises and falls. The time trend appears in Figure 3, where the horizontal lines indicate positive and negative statistical significance. Extended periods of elevated coupon rates occur for all three enterprises and are especially pronounced in early 1996, mid 1999, mid 2000 and 2001. The first two of these periods follow obvious mortgage refinancing waves shown in Figure 2 while the last two correspond to the obvious increase in mortgage acquisition activity by Fannie Mae and Freddie Mac in Figure 1. We take these findings as preliminary evidence that the combination of refinancing waves and the accumulation of fixed rate mortgage assets by the mortgage GSEs and the banking sector have led to increased long-term borrowing costs for the Government Sponsored Enterprises most directly affected by the phenomena. Conclusions This paper compares underwriting spreads faced by GSEs to those charged to other financial firms. WE pay particular attention to the combined impact of the strategic decision to portfolio fixed rate loans made by the mortgage GSEs and the uncertainty in mortgage asset value caused by the potential for refinancing waves on the borrowing costs faced by these GSEs. Our results show that the net result of internal GSE decisions and market developments has been a significant increase in both the up-front and 12

long-term financing fees paid by these particular entities. Notably, the effects are concentrated in debt maturities that match the expected duration of the mortgage loans held as assets. In addition, there appear to be spillover effects in that borrowing by the Federal Home Loan Bank Board, an enterprise that is not directly involved in the mortgage market but lends to entities that are, is also subject to fees and coupon rates that rise and fall with the relative volume of mortgage refinancing activity. 13

References Altinkilic, O. and R. Hansen, 2000, Are there economies of scale in underwriting fees? Evidence of rising external financing costs, Review of Financial Studies 13, 191-218. Ambrose, B. and A.Warga, 2002, Measuring potential GSE funding advantages, The Journal of Real Estate Finance and Economics, 25. Ambrose, B., M. Lacour-Little and A. B. Sanders, The Effect of Conforming Loan Status on Mortgage Yield Spreads: A Loan Level Analysis, Real Estate Economics, 32, 541-570. Burch, T., V. Nanda and V. Warther, 2005, Does it pay to be loyal? An Empirical Analysis of Underwriting Relationships and Fees, Journal of Financial Economics, forthcoming. Drucker, S. and M. Puri, 2005, On the benefits of Concurrent Lending and Underwriting, The Journal of Finance, forthcoming. 14

Figure 1: Fannie Mae and Freddie Mac's Earning Assets and Interest Bearing Liabilities 900000 800000 700000 600000 $, InMillions 500000 400000 300000 200000 100000 0 1994 1995 1996 1997 1998 1999 2000 2001 1994-2001 Study Period fnast fnliab fhast fhliab Figure 2: Percentage of Loan Originations that are Classified as Mortgage Refinancings 100.00 90.00 80.00 70.00 Percent of 100 60.00 50.00 40.00 30.00 20.00 10.00 0.00 199401 199501 199601 199701 199801 199901 200001 200101 1994-2001 perrefi 15

Figure 3: T-Statistics for Monthly Coupon Rate Regressions 10 8 6 4 T-Statistic 2 0 199401 199501 199601 199701 199801 199901 200001 200101-2 -4-6 1994-2001 tfan tfrd tfhl lower upper 16

Table 1 Dependent Variable: Under- writer Spread in Percent Sample Segmented by Maturity in Years Variable All <1 1-2 2-3 3-5 5-7 7-10 10-20 >20 Adj. R- Square 0.195 0.060 0.114 0.179 0.083 0.213 0.310 0.299 0.206 N Obs. 31096 1332 4644 6550 10894 8559 4964 6079 1238 Controls: Panel A Issue Controls Intercept 1.054 0.174 0.106 0.165 1.32 0.735 0.589 1.316 0.191 31.0 1.2 4.6 9.1 16.2 14.1 8.0 17.5 6.9 Lsize -0.266-0.064-0.029-0.019-0.510-0.078-0.052-0.087-0.276-26.2-2.5-4.0-4.3-25.0-5.5-2.7-4.0-3.8 LsizeSq 0.024 0.006 0.003 0.001 0.055 0.003-0.003-0.005 0.014 19.7 2.4 3.1 2.0 22.5 1.9-1.3-1.9 1.6 Mat 0.019 0.106 0.050 0.030 0.026 0.034 0.018 0.019 0.014 41.8 0.8 12.7 12.6 4.5 8.2 5.3 9.4 3.9 Coup 0.001 0.000 0.003 0.003 0.015 0.025 0.070-0.042-0.133 0.4-0.1 1.6 1.5 2.8 4.9 10.1-8.3-0.5 S&P Dum 0.042 0.002 0.008 0.012 0.029 0.020 0.023 0.040 0.046 22.0 0.5 5.0 10.8 5.4 6.2 6.4 9.9 4.7 Shelf -0.300 0.057 0.019-0.014-0.204-0.449-0.472-0.341 0.019-18.3 1.8 1.8 1.4-4.6-18.6-15.8-10.8 0.2 Time -0.009-0.005-0.003-0.005-0.018-0.003 0.001 0.016-0.106-5.7-1.6-3.0-7.3-5.1-1.6 0.4 4.3-1.0 Panel B Under- writer Controls Numld 0.031-0.003-0.001 0.009 0.029 0.011 0.025 0.044-0.036 5.3-0.5-0.4 3.8 2.8 1.6 2.0 3.2-0.5 Syn 0.042 0.001 0.001 0.007 0.031 0.029 0.037 0.031 0.757 5.9 0.1 0.2 2.4 2.2 3.3 2.6 1.9 7.6 t-statistics appear below estimated coefficients. Bold face entries are significant at 95% level of confidence, one-tailed test. 17

Underwriter Spread Regressions: Full Sample and by Maturity All <1 1-2 2-3 3-5 5-7 7-10 10-20 >20 Panel C GSE Controls Fan -0.295-1.500-0.009-0.067-0.179-0.658-0.792-0.340 0.074-6.8-1.7-0.3-3.3-1.9-11.6-8.9-3.9 0.2 FarmCr -0.333 0.094 0.047-0.031-0.255-0.714-1.072-0.869-0.025-4.3 0.7 1.1-1.1-1.9-6.5-5.6-3.0-0.9 FarmAg -0.593 N/A -0.049-0.036-0.415-0.687-0.982-0.462 N/A -1.4-0.2-0.2-0.4-0.8-1.7-0.6 Fred -0.298-0.052-0.005-0.063-0.247-0.728-0.871-0.346 0.013-7.7-0.8-0.2-3.2-2.7-13.1-12.2-5.0 2.3 Sal -0.278-0.034-0.035-1.070-0.156-0.601 N/A -0.233 N/A -2.4-0.3-0.8-2.3-0.6-3.4-0.5 Gin 0.091 N/A N/A N/A N/A N/A N/A N/A N/A 0.04 Fhlb -0.339-0.052-0.040-0.075-0.283-0.695-0.995-0.477-0.570-12.0-1.1-2.3-5.0-4.0-17.3-16.7-7.1-0.6 Refinancing Interaction Terms 1 Fan*%Refi -0.117 0.118-0.004 0.039 0.148 0.150 0.260-0.192-0.060 2.3 1.1-0.1 1.8 1.5 2.3 2.5-1.8-0.1 Fred*%Refi 0.173-0.006-0.016 0.028 0.224 0.291 0.401-0.141-1.710 3.8 0.0-0.5 1.3 2.2 4.5 4.7-1.8-2.5 Gin*%Refi -0.477 N/A N/A N/A N/A N/A N/A N/A N/A -0.2 FarmCr*%Refi 0.023-0.150-0.077-0.019 0.134 0.096 0.420 0.166 3.467 0.2-0.9-1.5-0.5 0.8 0.7 1.7 0.4 0.7 FarmAg*%Refi 0.175 N/A -0.004-0.020 0.167 0.013 0.220-0.477 N/A 0.3 0.0-0.1 0.1 0.0 0.3-0.5 Sal*%Refi 0.014-0.012 0.025 1.400 0.000 0.040 N/A -0.704 N/A -0.1-0.1 0.4 1.4 0.0 0.0-0.8 Fhlb*%Refi 0.011 0.024 0.034 0.029 0.168 0.061 0.278-0.374 0.575 0.4 0.5 1.9 2.1 2.7 1.6 4.3-5.1 0.5 t-statistics appear below estimated coefficients. Bold face entries are significant at 95% level of confidence, one-tailed test. N/A: insufficient observations in maturity bucket N/A 1 : variables replaced by interaction terms 1 Coefficients on interaction terms in underwriter spread regressions in basis points, all others in percent 18