Remarks by Chairman Ben S. Bernanke At the Federal Reserve Bank of
Kansas City's Economic Symposium, Jackson Hole, Wyoming August 31, 2007
Housing, Housing Finance, and Monetary Policy
Over the years, Tom Hoenig and his colleagues at the Federal Reserve
Bank of Kansas City have done an excellent job of selecting interesting
and relevant topics for this annual symposium. I think I can safely say
that this year they have outdone themselves. Recently, the subject of
housing finance has preoccupied financial-market participants and
observers in the United States and around the world. The financial
turbulence we have seen had its immediate origins in the problems in the
subprime mortgage market, but the effects have been felt in the broader
mortgage market and in financial markets more generally, with potential
consequences for the performance of the overall economy.
In my remarks this morning, I will begin with some observations about
recent market developments and their economic implications. I will then
try to place recent events in a broader historical context by discussing
the evolution of housing markets and housing finance in the United
States. In particular, I will argue that, over the years, institutional
changes in U.S. housing and mortgage markets have significantly
influenced both the transmission of monetary policy and the economy's
cyclical dynamics. As our system of housing finance continues to evolve,
understanding these linkages not only provides useful insights into the
past but also holds the promise of helping us better cope with the
implications of future developments.
Recent Developments in Financial Markets and the Economy I will begin
my review of recent developments by discussing the housing situation. As
you know, the downturn in the housing market, which began in the summer
of 2005, has been sharp. Sales of new and existing homes have declined
significantly from their mid-2005 peaks and have remained slow in recent
months. As demand has weakened, house prices have decelerated or even
declined by some measures, and homebuilders have scaled back their
construction of new homes. The cutback in residential construction has
directly reduced the annual rate of U.S. economic growth about 3/4
percentage point on average over the past year and a half. Despite the
slowdown in construction, the stock of unsold new homes remains quite
elevated relative to sales, suggesting that further declines in
homebuilding are likely.
The outlook for home sales and construction will also depend on
unfolding developments in mortgage markets. A substantial increase in
lending to nonprime borrowers contributed to the bulge in residential
investment in 2004 and 2005, and the tightening of credit conditions for
these borrowers likely accounts for some of the continued softening in
demand we have seen this year. As I will discuss, recent market
developments have resulted in additional tightening of rates and terms
for nonprime borrowers as well as for potential borrowers through
"jumbo" mortgages. Obviously, if current conditions persist in mortgage
markets, the demand for homes could weaken further, with possible
implications for the broader economy. We are following these
developments closely.
As house prices have softened, and as interest rates have risen from
the low levels of a couple of years ago, we have seen a marked
deterioration in the performance of nonprime mortgages. The problems
have been most severe for subprime mortgages with adjustable rates: the
proportion of those loans with serious delinquencies rose to about
13-1/2 percent in June, more than double the recent low seen in
mid-2005.1 The adjustable-rate subprime mortgages originated in late
2005 and in 2006 have performed the worst, in part because of slippage
in underwriting standards, reflected for example in high loan-to-value
ratios and incomplete documentation. With many of these borrowers facing
their first interest rate resets in coming quarters, and with softness
in house prices expected to continue to impede refinancing,
delinquencies among this class of mortgages are likely to rise further.
Apart from adjustable-rate subprime mortgages, however, the
deterioration in performance has been less pronounced, at least to this
point. For subprime mortgages with fixed rather than variable rates, for
example, serious delinquencies have been fairly stable at about 5-1/2
percent. The rate of serious delinquencies on alt-A securitized pools
rose to nearly 3 percent in June, from a low of less than 1 percent in
mid-2005. Delinquency rates on prime jumbo mortgages have also risen,
though they are lower than those for prime conforming loans, and both
rates are below 1 percent.
Investors' concerns about mortgage credit performance have
intensified sharply in recent weeks, reflecting, among other factors,
worries about the housing market and the effects of impending
interest-rate resets on borrowers' ability to remain current. Credit
spreads on new securities backed by subprime mortgages, which had jumped
earlier this year, rose significantly more in July. Issuance of such
securities has been negligible since then, as dealers have faced
difficulties placing even the AAA-rated tranches. Issuance of securities
backed by alt-A and prime jumbo mortgages also has fallen sharply, as
investors have evidently become concerned that the losses associated
with these types of mortgages may be higher than had been expected.
With securitization impaired, some major lenders have announced the
cancellation of their adjustable-rate subprime lending programs. A
number of others that specialize in nontraditional mortgages have been
forced by funding pressures to scale back or close down. Some lenders
that sponsor asset-backed commercial paper conduits as bridge financing
for their mortgage originations have been unable to "roll" the maturing
paper, forcing them to draw on back-up liquidity facilities or to
exercise options to extend the maturity of their paper. As a result of
these developments, borrowers face noticeably tighter terms and
standards for all but conforming mortgages.
As you know, the financial stress has not been confined to mortgage
markets. The markets for asset-backed commercial paper and for
lower-rated unsecured commercial paper market also have suffered from
pronounced declines in investor demand, and the associated flight to
quality has contributed to surges in the demand for short-dated Treasury
bills, pushing T-bill rates down sharply on some days. Swings in stock
prices have been sharp, with implied price volatilities rising to about
twice the levels seen in the spring. Credit spreads for a range of
financial instruments have widened, notably for lower-rated corporate
credits. Diminished demand for loans and bonds to finance highly
leveraged transactions has increased some banks' concerns that they may
have to bring significant quantities of these instruments onto their
balance sheets. These banks, as well as those that have committed to
serve as back-up facilities to commercial paper programs, have become
more protective of their liquidity and balance-sheet capacity.
Although this episode appears to have been triggered largely by
heightened concerns about subprime mortgages, global financial losses
have far exceeded even the most pessimistic projections of credit losses
on those loans. In part, these wider losses likely reflect concerns that
weakness in U.S. housing will restrain overall economic growth. But
other factors are also at work. Investor uncertainty has increased
significantly, as the difficulty of evaluating the risks of structured
products that can be opaque or have complex payoffs has become more
evident. Also, as in many episodes of financial stress, uncertainty
about possible forced sales by leveraged participants and a higher cost
of risk capital seem to have made investors hesitant to take advantage
of possible buying opportunities. More generally, investors may have
become less willing to assume risk. Some increase in the premiums that
investors require to take risk is probably a healthy development on the
whole, as these premiums have been exceptionally low for some time.
However, in this episode, the shift in risk attitudes has interacted
with heightened concerns about credit risks and uncertainty about how to
evaluate those risks to create significant market stress. On the
positive side of the ledger, we should recognize that past efforts to
strengthen capital positions and the financial infrastructure place the
global financial system in a relatively strong position to work through
this process.
In the statement following its August 7 meeting, the Federal Open
Market Committee (FOMC) recognized that the rise in financial volatility
and the tightening of credit conditions for some households and
businesses had increased the downside risks to growth somewhat but
reiterated that inflation risks remained its predominant policy concern.
In subsequent days, however, following several events that led investors
to believe that credit risks might be larger and more pervasive than
previously thought, the functioning of financial markets became
increasingly impaired. Liquidity dried up and spreads widened as many
market participants sought to retreat from certain types of asset
exposures altogether.
Well-functioning financial markets are essential for a prosperous
economy. As the nation's central bank, the Federal Reserve seeks to
promote general financial stability and to help to ensure that financial
markets function in an orderly manner. In response to the developments
in the financial markets in the period following the FOMC meeting, the
Federal Reserve provided reserves to address unusual strains in money
markets. On August 17, the Federal Reserve Board announced a cut in the
discount rate of 50 basis points and adjustments in the Reserve Banks'
usual discount window practices to facilitate the provision of term
financing for as long as thirty days, renewable by the borrower. The
Federal Reserve also took a number of supplemental actions, such as
cutting the fee charged for lending Treasury securities. The purpose of
the discount window actions was to assure depositories of the ready
availability of a backstop source of liquidity. Even if banks find that
borrowing from the discount window is not immediately necessary, the
knowledge that liquidity is available should help alleviate concerns
about funding that might otherwise constrain depositories from extending
credit or making markets. The Federal Reserve stands ready to take
additional actions as needed to provide liquidity and promote the
orderly functioning of markets.
It is not the responsibility of the Federal Reserve--nor would it be
appropriate--to protect lenders and investors from the consequences of
their financial decisions. But developments in financial markets can
have broad economic effects felt by many outside the markets, and the
Federal Reserve must take those effects into account when determining
policy. In a statement issued simultaneously with the discount window
announcement, the FOMC indicated that the deterioration in financial
market conditions and the tightening of credit since its August 7
meeting had appreciably increased the downside risks to growth. In
particular, the further tightening of credit conditions, if sustained,
would increase the risk that the current weakness in housing could be
deeper or more prolonged than previously expected, with possible adverse
effects on consumer spending and the economy more generally.
The incoming data indicate that the economy continued to expand at a
moderate pace into the summer, despite the sharp correction in the
housing sector. However, in light of recent financial developments,
economic data bearing on past months or quarters may be less useful than
usual for our forecasts of economic activity and inflation.
Consequently, we will pay particularly close attention to the timeliest
indicators, as well as information gleaned from our business and banking
contacts around the country. Inevitably, the uncertainty surrounding the
outlook will be greater than normal, presenting a challenge to
policymakers to manage the risks to their growth and price stability
objectives. The Committee continues to monitor the situation and will
act as needed to limit the adverse effects on the broader economy that
may arise from the disruptions in financial markets.
Beginnings: Mortgage Markets in the Early Twentieth Century Like us,
our predecessors grappled with the economic and policy implications of
innovations and institutional changes in housing finance. In the
remainder of my remarks, I will try to set the stage for this weekend's
conference by discussing the historical evolution of the mortgage market
and some of the implications of that evolution for monetary policy and
the economy.
The early decades of the twentieth century are a good starting point
for this review, as urbanization and the exceptionally rapid population
growth of that period created a strong demand for new housing. Between
1890 and 1930, the number of housing units in the United States grew
from about 10 million to about 30 million; the pace of homebuilding was
particularly brisk during the economic boom of the 1920s.
Remarkably, this rapid expansion of the housing stock took place
despite limited sources of mortgage financing and typical lending terms
that were far less attractive than those to which we are accustomed
today. Required down payments, usually about half of the home's purchase
price, excluded many households from the market. Also, by comparison
with today's standards, the duration of mortgage loans was short,
usually ten years or less. A "balloon" payment at the end of the loan
often created problems for borrowers.2
High interest rates on loans reflected the illiquidity and the
essentially unhedgeable interest rate risk and default risk associated
with mortgages. Nationwide, the average spread between mortgage rates
and high-grade corporate bond yields during the 1920s was about 200
basis points, compared with about 50 basis points on average since the
mid-1980s. The absence of a national capital market also produced
significant regional disparities in borrowing costs. Hard as it may be
to conceive today, rates on mortgage loans before World War I were at
times as much as 2 to 4 percentage points higher in some parts of the
country than in others, and even in 1930, regional differences in rates
could be more than a full percentage point.3
Despite the underdevelopment of the mortgage market, homeownership
rates rose steadily after the turn of the century. As would often be the
case in the future, government policy provided some inducement for
homebuilding. When the federal income tax was introduced in 1913, it
included an exemption for mortgage interest payments, a provision that
is a powerful stimulus to housing demand even today. By 1930, about 46
percent of nonfarm households owned their own homes, up from about 37
percent in 1890.
The limited availability of data prior to 1929 makes it hard to
quantify the role of housing in the monetary policy transmission
mechanism during the early twentieth century. Comparisons are also
complicated by great differences between then and now in monetary policy
frameworks and tools. Still, then as now, periods of tight money were
reflected in higher interest rates and a greater reluctance of banks to
lend, which affected conditions in mortgage markets. Moreover, students
of the business cycle, such as Arthur Burns and Wesley Mitchell, have
observed that residential construction was highly cyclical and
contributed significantly to fluctuations in the overall economy (Burns
and Mitchell, 1946). Indeed, if we take the somewhat less reliable data
for 1901 to 1929 at face value, real housing investment was about three
times as volatile during that era as it has been over the past
half-century.
During the past century we have seen two great sea changes in the
market for housing finance. The first of these was the product of the
New Deal. The second arose from financial innovation and a series of
crises from the 1960s to the mid-1980s in depository funding of
mortgages. I will turn first to the New Deal period.
The New Deal and the Housing Market The housing sector, like the rest
of the economy, was profoundly affected by the Great Depression. When
Franklin Roosevelt took office in 1933, almost 10 percent of all homes
were in foreclosure (Green and Wachter, 2005), construction employment
had fallen by half from its late 1920s peak, and a banking system near
collapse was providing little new credit. As in other sectors, New Deal
reforms in housing and housing finance aimed to foster economic revival
through government programs that either provided financing directly or
strengthened the institutional and regulatory structure of private
credit markets.
Actually, one of the first steps in this direction was taken not by
Roosevelt but by his predecessor, Herbert Hoover, who oversaw the
creation of the Federal Home Loan Banking System in 1932. This measure
reorganized the thrift industry (savings and loans and mutual savings
banks) under federally chartered associations and established a credit
reserve system modeled after the Federal Reserve. The Roosevelt
administration pushed this and other programs affecting housing finance
much further. In 1934, his administration oversaw the creation of the
Federal Housing Administration (FHA). By providing a federally backed
insurance system for mortgage lenders, the FHA was designed to encourage
lenders to offer mortgages on more attractive terms. This intervention
appears to have worked in that, by the 1950s, most new mortgages were
for thirty years at fixed rates, and down payment requirements had
fallen to about 20 percent. In 1938, the Congress chartered the Federal
National Mortgage Association, or Fannie Mae, as it came to be known.
The new institution was authorized to issue bonds and use the proceeds
to purchase FHA mortgages from lenders, with the objectives of
increasing the supply of mortgage credit and reducing variations in the
terms and supply of credit across regions.4
Shaped to a considerable extent by New Deal reforms and regulations,
the postwar mortgage market took on the form that would last for several
decades. The market had two main sectors. One, the descendant of the
pre-Depression market sector, consisted of savings and loan
associations, mutual savings banks, and, to a lesser extent, commercial
banks. With financing from short-term deposits, these institutions made
conventional fixed-rate long-term loans to homebuyers. Notably, federal
and state regulations limited geographical diversification for these
lenders, restricting interstate banking and obliging thrifts to make
mortgage loans in small local areas--within 50 miles of the home office
until 1964, and within 100 miles after that. In the other sector, the
product of New Deal programs, private mortgage brokers and other lenders
originated standardized loans backed by the FHA and the Veterans'
Administration (VA). These guaranteed loans could be held in portfolio
or sold to institutional investors through a nationwide secondary
market.
No discussion of the New Deal's effect on the housing market and the
monetary transmission mechanism would be complete without reference to
Regulation Q--which was eventually to exemplify the law of unintended
consequences. The Banking Acts of 1933 and 1935 gave the Federal Reserve
the authority to impose deposit-rate ceilings on banks, an authority
that was later expanded to cover thrift institutions. The Fed used this
authority in establishing its Regulation Q. The so-called Reg Q ceilings
remained in place in one form or another until the mid-1980s.5
The original rationale for deposit ceilings was to reduce "excessive"
competition for bank deposits, which some blamed as a cause of bank
failures in the early 1930s. In retrospect, of course, this was a
dubious bit of economic analysis. In any case, the principal effects of
the ceilings were not on bank competition but on the supply of credit.
With the ceilings in place, banks and thrifts experienced what came to
be known as disintermediation--an outflow of funds from depositories
that occurred whenever short-term money-market rates rose above the
maximum that these institutions could pay. In the absence of alternative
funding sources, the loss of deposits prevented banks and thrifts from
extending mortgage credit to new customers.
The Transmission Mechanism and the New Deal Reforms Under the New
Deal system, housing construction soared after World War II, driven by
the removal of wartime building restrictions, the need to replace an
aging housing stock, rapid family formation that accompanied the
beginning of the baby boom, and large-scale internal migration. The
stock of housing units grew 20 percent between 1940 and 1950, with most
of the new construction occurring after 1945.
In 1951, the Treasury-Federal Reserve Accord freed the Fed from the
obligation to support Treasury bond prices. Monetary policy began to
focus on influencing short-term money markets as a means of affecting
economic activity and inflation, foreshadowing the Federal Reserve's
current use of the federal funds rate as a policy instrument. Over the
next few decades, housing assumed a leading role in the monetary
transmission mechanism, largely for two reasons: Reg Q and the advent of
high inflation.
The Reg Q ceilings were seldom binding before the mid-1960s, but
disintermediation induced by the ceilings occurred episodically from the
mid-1960s until Reg Q began to be phased out aggressively in the early
1980s. The impact of disintermediation on the housing market could be
quite significant; for example, a moderate tightening of monetary policy
in 1966 contributed to a 23 percent decline in residential construction
between the first quarter of 1966 and the first quarter of 1967. State
usury laws and branching restrictions worsened the episodes of
disintermediation by placing ceilings on lending rates and limiting the
flow of funds between local markets. For the period 1960 to 1982, when
Reg Q assumed its greatest importance, statistical analysis shows a high
correlation between single-family housing starts and the growth of small
time deposits at thrifts, suggesting that disintermediation effects were
powerful; in contrast, since 1983 this correlation is essentially zero.6
Economists at the time were well aware of the importance of the
disintermediation phenomenon for monetary policy. Frank de Leeuw and
Edward Gramlich highlighted this particular channel in their description
of an early version of the MPS macroeconometric model, a joint product
of researchers at the Federal Reserve, MIT, and the University of
Pennsylvania (de Leeuw and Gramlich, 1969). The model attributed almost
one-half of the direct first-year effects of monetary policy on the real
economy--which were estimated to be substantial--to disintermediation
and other housing-related factors, despite the fact that residential
construction accounted for only 4 percent of nominal gross domestic
product (GDP) at the time.
As time went on, however, monetary policy mistakes and weaknesses in
the structure of the mortgage market combined to create deeper economic
problems. For reasons that have been much analyzed, in the late 1960s
and the 1970s the Federal Reserve allowed inflation to rise, which led
to corresponding increases in nominal interest rates. Increases in
short-term nominal rates not matched by contractually set rates on
existing mortgages exposed a fundamental weakness in the system of
housing finance, namely, the maturity mismatch between long-term
mortgage credit and the short-term deposits that commercial banks and
thrifts used to finance mortgage lending. This mismatch led to a series
of liquidity crises and, ultimately, to a rash of insolvencies among
mortgage lenders. High inflation was also ultimately reflected in high
nominal long-term rates on new mortgages, which had the effect of "front
loading" the real payments made by holders of long-term, fixed-rate
mortgages. This front-loading reduced affordability and further limited
the extension of mortgage credit, thereby restraining construction
activity. Reflecting these factors, housing construction experienced a
series of pronounced boom and bust cycles from the early 1960s through
the mid-1980s, which contributed in turn to substantial swings in
overall economic growth.
The Emergence of Capital Markets as a Source of Housing Finance The
manifest problems associated with relying on short-term deposits to fund
long-term mortgage lending set in train major changes in financial
markets and financial instruments, which collectively served to link
mortgage lending more closely to the broader capital markets. The shift
from reliance on specialized portfolio lenders financed by deposits to a
greater use of capital markets represented the second great sea change
in mortgage finance, equaled in importance only by the events of the New
Deal.
Government actions had considerable influence in shaping this second
revolution. In 1968, Fannie Mae was split into two agencies: the
Government National Mortgage Association (Ginnie Mae) and the
re-chartered Fannie Mae, which became a privately owned
government-sponsored enterprise (GSE), authorized to operate in the
secondary market for conventional as well as guaranteed mortgage loans.
In 1970, to compete with Fannie Mae in the secondary market, another GSE
was created--the Federal Home Loan Mortgage Corporation, or Freddie Mac.
Also in 1970, Ginnie Mae issued the first mortgage pass-through
security, followed soon after by Freddie Mac. In the early 1980s,
Freddie Mac introduced collateralized mortgage obligations (CMOs), which
separated the payments from a pooled set of mortgages into "strips"
carrying different effective maturities and credit risks. Since 1980,
the outstanding volume of GSE mortgage-backed securities has risen from
less than $200 billion to more than $4 trillion today. Alongside these
developments came the establishment of private mortgage insurers, which
competed with the FHA, and private mortgage pools, which bundled loans
not handled by the GSEs, including loans that did not meet GSE
eligibility criteria--so-called nonconforming loans. Today, these
private pools account for around $2 trillion in residential mortgage
debt.
These developments did not occur in time to prevent a large fraction
of the thrift industry from becoming effectively insolvent by the early
1980s in the wake of the late-1970s surge in inflation.7 In this
instance, the government abandoned attempts to patch up the system and
instead undertook sweeping deregulation. Reg Q was phased out during the
1980s; state usury laws capping mortgage rates were abolished;
restrictions on interstate banking were lifted by the mid-1990s; and
lenders were permitted to offer adjustable-rate mortgages as well as
mortgages that did not fully amortize and which therefore involved
balloon payments at the end of the loan period. Critically, the savings
and loan crisis of the late 1980s ended the dominance of deposit-taking
portfolio lenders in the mortgage market. By the 1990s, increased
reliance on securitization led to a greater separation between mortgage
lending and mortgage investing even as the mortgage and capital markets
became more closely integrated. About 56 percent of the home mortgage
market is now securitized, compared with only 10 percent in 1980 and
less than 1 percent in 1970.
In some ways, the new mortgage market came to look more like a
textbook financial market, with fewer institutional "frictions" to
impede trading and pricing of event-contingent securities.
Securitization and the development of deep and liquid derivatives
markets eased the spreading and trading of risk. New types of mortgage
products were created. Recent developments notwithstanding, mortgages
became more liquid instruments, for both lenders and borrowers.
Technological advances facilitated these changes; for example,
computerization and innovations such as credit scores reduced the costs
of making loans and led to a "commoditization" of mortgages. Access to
mortgage credit also widened; notably, loans to subprime borrowers
accounted for about 13 percent of outstanding mortgages in 2006.
I suggested that the mortgage market has become more like the
frictionless financial market of the textbook, with fewer institutional
or regulatory barriers to efficient operation. In one important respect,
however, that characterization is not entirely accurate. A key f
