Bubbles have been frequent in economic history, and they occur in the laboratories of experimental economics under conditions which — when first studied in the 1980s — were considered so transparent that bubbles would not be observed.
We economists were wrong: Even when traders in an asset market know the value of the asset, bubbles form dependably. Bubbles can arise when some agents buy not on fundamental value, but on price trend or momentum. If momentum traders have more liquidity, they can sustain a bubble longer.
But what sparks bubbles? Why does one large asset bubble — like our dot-com bubble — do no damage to the financial system while another one leads to its collapse? Key characteristics of housing markets — momentum trading, liquidity, price-tier movements, and high-margin purchases — combine to provide a fairly complete, simple description of the housing bubble collapse, and how it engulfed the financial system and then the wider economy.
In just the past 40 years there were two other housing bubbles, with peaks in 1979 and 1989, but the largest one in U.S. history started in 1997, probably sparked by rising household income that began in 1992 combined with the elimination in 1997 of taxes on residential capital gains up to $500,000. Rising values in an asset market draw investor attention; the early stages of the housing bubble had this usual, self-reinforcing feature.
The 2001 recession might have ended the bubble, but the Federal Reserve decided to pursue an unusually expansionary monetary policy in order to counteract the downturn. When the Fed increased liquidity, money naturally flowed to the fastest expanding sector. Both the Clinton and Bush administrations aggressively pursued the goal of expanding homeownership, so credit standards eroded. Lenders and the investment banks that securitized mortgages used rising home prices to justify loans to buyers with limited assets and income. Rating agencies accepted the hypothesis of ever rising home values, gave large portions of each security issue an investment-grade rating, and investors gobbled them up.
But housing expenditures in the U.S. and most of the developed world have historically taken about 30% of household income. If housing prices more than double in a seven-year period without a commensurate increase in income, eventually something has to give. When subprime lending, the interest-only adjustable-rate mortgage (ARM), and the negative-equity option ARM were no longer able to sustain the flow of new buyers, the inevitable crash could no longer be delayed.
The price decline started in 2006. Then policies designed to promote the American dream instead produced a nightmare. Trillions of dollars of mortgages, written to buyers with slender equity, started a wave of delinquencies and defaults. Borrowers’ losses were limited to their small down payments; hence, the lion’s share of the losses was transmitted into the financial system and it collapsed.
During the 1976-79 and 1986-89 housing price bubbles, the effective federal-funds interest rate was rising while housing prices rose: The Federal Reserve, “leaning against the wind,” helped mitigate the bubbles. In January 2001, however, after four years with average inflation-adjusted house price increases of 7.2% per year (about 6% above trend for the past 80 years), the Fed started to decrease the fed-funds rate. By December 2001, the rate had been reduced to its lowest level since 1962. In 2002 the average fed-funds rate was lower than in any year since the 1958 recession. In 2003 and 2004 the average fed-funds rates were lower than in any year since 1955 when the rate series began.
Monetary policy, mortgage finance, relaxed lending standards, and tax-free capital gains provided astonishing economic stimulus: Mortgage loan originations increased an average of 56% per year for three years — from $1.05 trillion in 2000 to $3.95 trillion in 2003!
By the time the Federal Reserve began to slowly raise the fed-funds rate in May 2004, the Case-Shiller 20-city composite index had increased 15.4% during the previous 12 months. Yet the housing portion of the CPI for those same 12 months rose only 2.4%.
How could this happen? In 1983, the Bureau of Labor Statistics began to use rental equivalence for homeowner-occupied units instead of direct home-ownership costs. Between 1983 and 1996, the price-to-rental ratio increased from 19.0 to 20.2, so the change had little effect on measured inflation: The CPI underestimated inflation by about 0.1 percentage point per year during this period. Between 1999 and 2006, the price-to-rent ratio shot up from 20.8 to 32.3.
With home price increases out of the CPI and the price-to-rent ratio rapidly increasing, an important component of inflation remained outside the index. In 2004 alone, the price-rent ratio increased 12.3%. Inflation for that year was underestimated by 2.9 percentage points (since “owners’ equivalent rent” is about 23% of the CPI). If home-ownership costs were included in the CPI, inflation would have been 6.2% instead of 3.3%.
With nominal interest rates around 6% and inflation around 6%, the real interest rate was near zero, so household borrowing took off. As measured by the Case-Shiller 10 city index, the accumulated inflation in home-ownership costs between January 1999 and June 2006 was 151%, but the CPI measured a mere 23% increase. As the Federal Reserve monitored inflation in the early part of this decade, home-price increases were no longer visible in the CPI, so the lax monetary policy continued. Even after the Fed began to slowly raise the fed-funds rate in May 2004, the average rate remained low and the bubble continued to inflate for two more years.
The unraveling of the bubble is in many ways the most fascinating part of the story, and the most painful reality we are now experiencing. The median price of existing homes had fallen from $230,000 in July to $217,300 in November 2006. By the beginning of 2007, in 17 of the 20 cities in the Case-Shiller index, prices were falling. Serious price declines had not yet begun, but the warning signs were there for alert observers.
Kate Kelly, writing in this newspaper (Dec. 14, 2007), tells the story of how Goldman Sachs avoided the fate of many of the other investment banks that packaged mortgages into securities. Goldman loaded up on the Markit ABX index of credit default swaps between early December 2006 and late February 2007, as their price dropped from 97.70 on Dec. 4 to under 64 by Feb. 27. But the market was not yet in free-fall: The insurance on AAA-rated parts of the mortgage-backed securities (MBS) remained inexpensive. By mid-summer 2007, concern spread to the AAA-rated tranches of MBS.
At the end of February 2007, the cost of $10 million of insurance on the AAA-rated portion of a mortgage-backed security was still only $68,000 plus a $9,000 annual premium. Housing-market conditions deteriorated further in the first half of 2007. Case-Shiller tiered price sequences in Los Angeles, San Francisco, San Diego and Miami all show serious declines by the summer of 2007. Prices in the low-price tier in San Francisco were down almost 13% from their peak by July 2007; in San Diego they were off 10% by July 2007. Startling developments began to unfold that month. Between July 9 and Aug. 3, 2007, the cost of insuring AAA MBS tranches went from $50,000 upfront plus a $9,000 annual premium for $10 million of insurance to over $900,000 upfront (plus the annual premium).
Once the cost of insuring new mortgage-backed securities skyrocketed, mortgage financing from MBS rapidly declined. Subprime originations plummeted from $160 billion in the third quarter of 2006 to $28 billion in the third quarter of 2007. Mortgage-backed security issuance fell comparably, from $483 billion in all of 2006 to only $30.7 billion in the third quarter of 2007. Other measures of new loan originations were falling at the same time. The liquidity that generated the housing market bubble was evaporating.
Trouble quickly spread from the cost of insuring mortgage-backed securities to problems with credit markets generally, as the spread between short-term U.S. Treasury debt and the LIBOR rate increased to 2.40% from 0.44% between Aug. 8 and Aug. 20, 2007. Since U.S. Treasury debt is generally considered secure, but a bank’s loans to another bank carry some risk of default, the spread between these rates serves as an indicator of perceived risk in financial markets.
In one city after another, prices of homes in the low-price tier appreciated the most and then fell the most; prices in the high-priced tier appreciated least and fell the least. The price index graphs for Los Angeles, San Francisco, San Diego and Miami show that in all of these cities, prices in the low-price tier have fallen between 50% and 57%. Moreover, housing prices have continually declined in every market in the Case-Shiller index. According to First American CoreLogic, 10.5 million households had negative or near negative equity in December 2008. When housing prices turned down, many borrowers with low income and few assets other than their slender home equity faced foreclosure. The remaining losses had to be absorbed by the financial system. Consequently, the financial system has suffered a blow unlike anything since the Great Depression, and the source is the weak financial position of the people holding declining assets.
Earlier, during the downturn in the equities market between December 1999 and September 2002, approximately $10 trillion of equity was erased. But a measure of financial system performance, the Keefe, Bruyette, & Woods BKX index of financial firms, fell less than 6% during that period. In the current downturn, the value of residential real estate has fallen by approximately $3 trillion, but the BKX index has now fallen 75% from its peak of January 2007. The financial sector has been devastated in this crisis, whereas it was almost completely unaffected by the downturn in the equities market early in this decade.
How can one crash that wipes out $10 trillion in assets cause no damage to the financial system and another that causes $3 trillion in losses devastate the financial system?
In the equities-market downturn early in this decade, declining assets were held by institutional and individual investors that either owned the assets outright, or held only a small fraction on margin, so losses were absorbed by their owners. In the current crisis, declining housing assets were often, in effect, purchased between 90% and 100% on margin. In some of the cities hit hardest, borrowers who purchased in the low-price tier at the peak of the bubble have seen their home value decline 50% or more. Over the past 18 months as housing prices have fallen, millions of homes became worth less than the loans on them, huge losses have been transmitted to lending institutions, investment banks, investors in mortgage-backed securities, sellers of credit default swaps, and the insurer of last resort, the U.S. Treasury.
In an important paper in 1983, Ben Bernanke argued that during the Depression, severe damage to the financial system impeded its ability to perform its economic role of lending to households for durable goods consumption and to firms for production and trade. We are seeing this process playing out now as loan funds for automobile purchases have withered. Auto sales fell 41% between February 2008 and February 2009. Retail and labor markets too are now part of the collateral damage from the housing debacle. Housing peaked in early 2006. Losses from the mortgage market began to infect the financial system in 2006; asset prices in that sector began to decline at the end of 2006. Meanwhile, equities and the broader economy were performing well, but as the financial sector deteriorated, its problems blindsided the rest of the economy.
The events of the past 10 years have an eerie similarity to the period leading up to the Great Depression. Total mortgage debt outstanding increased from $9.35 billion in 1920 to $29.44 billion in 1929. In 1920, residential mortgage debt was 10.2% of household wealth; by 1929, it was 27.2% of household wealth.
The Great Depression has been attributed to excessive speculation on Wall Street, especially between the spring of 1927 and the fall of 1929. Had the difficulties of the banking system been caused by losses on brokers’ loans for margin purchases in 1929, the results should have been felt in the banks immediately after the stock market crash. But the banking system did not show serious strains until the fall of 1930.
Bank earnings reached a record $729 million in 1929. Yet bank exposures to real estate were substantial; as the decline in real estate prices accelerated, foreclosures wiped out banks by the thousands. Had the mounting difficulties of the banks and the final collapse of the banking system in the “Bank Holiday” in March 1933 been caused by contraction of the money supply, as Milton Friedman and Anna Schwartz argued, then the massive injections of liquidity over the past 18 months should have averted the collapse of the financial market during this current crisis.
The causes of the Great Depression need more study, but the claims that losses on stock-market speculation and a monetary contraction caused the decline of the banking system both seem inadequate. It appears that both the Great Depression and the current crisis had their origins in excessive consumer debt — especially mortgage debt — that was transmitted into the financial sector during a sharp downturn.
What we’ve offered in our discussion of this crisis is the back story to Mr. Bernanke’s analysis of the Depression. Why does one crash cause minimal damage to the financial system, so that the economy can pick itself up quickly, while another crash leaves a devastated financial sector in the wreckage? The hypothesis we propose is that a financial crisis that originates in consumer debt, especially consumer debt concentrated at the low end of the wealth and income distribution, can be transmitted quickly and forcefully into the financial system. It appears that we’re witnessing the second great consumer debt crash, the end of a massive consumption binge.