Economists have pointed out three contributing factors for the Great Recession. All of them closely related to the financial and banking system: Leverage, complexity, and liquidity (Blanchard, 2013). In this article, we try to advance an alternative interpretation of the origin of the Great Recession by looking at the housing market and to its links to mortgage borrower’s expectations. More precisely, we look at one reason why a Mortgage-backed loan (MBL) may go underwater: that is, a “Mortgage default Risk derived from housing building fatigue”. Here, we advance the first hypothesis of a series of three that compose the entire proposal. We assume that the Great Recession was largely an effect of a substantial decrease in consumer confidence. Likewise, we consider that the Great Recession has arguably many similarities with the U.S. Recession of 2001. The work cites those two crisis in order to determine stylized facts of U.S. Recessions.
On the third part of the article we look at the logic behind the boom of the housing prices. We aim at situating the discussion in the frame of capital inflows to the housing market. We see this as a reasonable movement since nominal interest rates in the financial markets were at record lows. After presenting the macroeconomic context based on the stylized facts, we propose a model of what we call “Mortgage default Risk derived from housing building fatigue”, followed by some linear regression to estimate the correlation of the first hypothesis. Finally we present some conclusion about the limits and the open windows for further research.
We propose the following hypothesis as a contributing factor to the origin of the Great Recession: default risk in mortgages payments are higher for low income Americans and old housing buildings mortgages exacerbate the risk. There is a correlation between low income Americans dwelling in aged housing buildings. There are reasons to believe that a mortgage loan goes underwater more often as the housing building gets older. This basically means that the older the building, the greater the chances for default of payments in the mortgage.
We put forth the following three hypothesis extending the contributing factors of the Great Recession:
Hypothesis number one: low income Americans tend to live in aged housing buildings.
Hypothesis number two: mortgage loans go underwater more often as the housing buildings get older.
Hypothesis number three: default in mortgages payments are higher for both low income Americans as for old housing buildings mortgages loans.
We use descriptive statistics to give an account of our hypothesis. We first take data from the American Housing Survey which is processed by the United States Census Bureau. This data inquire into the various aspects of housing buildings. Data on age of buildings and household income are described used such a survey. Then, some simple linear regression are run for establishing the correlation of the first hypothesis. We certainly acknowledge limitations of the data base and the conclusions we are making. Nonetheless, the aim of this paper is to point out some aspects of the hypothesis so that it can be tested in either larger samples or by using panel comparative data. Not enough numbers of observation may be the main source of statistical violations methodologically speaking.
Stylized facts of the Great Recession:
Stylized facts of the Great Recession relate to the analytical framework of IS-LM model. More precisely, stylized facts refer to the immediate effects of the financial crisis over the macro economy. It is arguably accepted that, in terms of the IS-LM model, the IS curve shifted sharply to the left thereby fostering the Great Recession (Graph # 1). The Great Recession was largely an effect of both, a substantial decrease in consumer confidence preceded by a large increase in interest rates. The United States Government reacted wisely but late. By the time the U.S. government realized there was a crisis in the economy, the Great Recession had spread throughout the economy. By 2008 Banks were reluctant to lend to each other, which is a signal of how pervasive Mortgage Backed Securities were considered to be (Blanchard, 2013). After Lehman Brothers declared bankruptcy in September 15th 2008, the Government of the United States decided to intervene by putting in place two programs: the Troubled Asset Relief Program (TARP); and later on, the American Recovery and Reinvestment Act (ARRA). TARP was intended to back toxic assets already in the market, whereas ARRA meant an increase in government deficit (Blanchard, 2013). Clearly, both ARRA and TARP consisted of an effort to use fiscal policy as a tool for pushing back the IS curve, since it is hard to assume TARP as a monetary tool.
A more detailed account of the events that fostered the Great Recession focus at the role of banks in the economy. Economists have pointed out onto three contributing aspects across the financial system: Leverage, complexity, and liquidity (Blanchard, 2013). First, bank leverage went at high level during the Great Recession. Through the usage of “innovative” financial instruments, Banks were able to hide off-balance-sheet leverage ratios (Kalemi-Ozcun, S. et al, 2011). Second, complexity. The growth of securitization of Mortgages-Baked loans (MBS) and the proliferation of Collateralized Debt Obligations (CDO) helped lending risk to hide too. Rating agencies were unable to detect and discriminate toxic assets grouped in large security bundles (Horton, B. 2013). Finally, liquidity related regulation was circumvent by Banks via Credit Default Swaps CDS. Through the issuance of security insurance, Banks were able to mask their liabilities with respect of the level of risk they took. In other words, riskier mortgages loans or Subprime loans were masked riskless throughout CDSs and bundled into MBSs (Blanchard, 2013). This structure helped the crisis to spread out deeper the financial system. These three aspects cogently support the Great Recession explanations in regards to the financial system.
Generally speaking, a fall in consumer confidence affects consumption negatively. Given that investment depends partially on sales, a fall in consumer spending will make non-residential investment to drop. Private savings will decrease since consumption depends on income. The overall effect of a decrease in consumer confidence would eventually lead to a deferral of household spending whereby hurting output through a deferral of non-residential investment.
Thus, housing prices increased from 2,000 to 2,006, which in some indexes such as the Case-Shiller that increase represents more than 200% change in less than a decade (Blanchard, 2103). This increase in demand for loans was outcome of low interest rates in mortgages. Economists have come into a consensus that during 2,000 and 2,006 Banks apparently were more willing to lend to borrowers, to the extent that by 2,006 20% of the U.S. loans were mortgages (Blanchard, 2013). Additionally, both borrowers and lender seemed to agree on the assumption that housing buildings do not depreciate over time.
U.S. Recession of 2001:
To some extent the Great Recession has many similarities with the U.S. Recession of 2001. The years preceding 2001 were characterized by a similar economic expansion. Though investment declined 4.5% in 2001 (Blanchard, 2013). Data at the time did not show negative growth nor signals of slowdown. Both Recession were triggered by a decrease in autonomous spending. The Fed pursued a very expansionary monetary policy after realizing there was a recession in the U.S. economy. The Fed chair of the time, Alan Greenspan, called the preceding expansion that led the economy to the slow down, a period of “irrational exuberance” (Blanchard, 2013). There were high levels of economic optimism that lasted for almost a decade before 2001. On the fiscal policy side, President George W. Bush pushed through legislation that lowered income taxes and increased spending in homeland security sector. On the monetary policy side, the Fed increased the money supply in 2001. Graph #2 depicts the effect of a decline in investment demand as well as the effects of the policies put forth by the government. Note that both graphs look similar.
Graph # 2.
Nonetheless, as Horton (2013) point out financial derivatives have always existed and performed well. Yet an explanation that considers only the financial crisis seems to be not sufficient. Perhaps, a deeper look into consumer expectations and into the history of the housing market may shed light onto some complementing conclusions. Thus, the model presented in this paper advances an emulation of the real interest rate calculation as the interest rate in terms of the housing prices. In the context of low nominal interest rates, analysis considering either the appreciation or depreciation of houses prices become more relevant. Investor’s decision on whether to hold bonds or equity on a house makes the difference for the real estate market, and particularly to the housing market. When interest rates are at low levels it is expected that investments seeking profits turn into the real sector. Even more, under the perceived reality that housing prices do not depreciate with time.
Capital inflows to the housing market are reasonable movements since nominal interest rates are at records lows. Additionally, expected inflation has been also around 1.1% (Blanchard, 2013). Both nominal interest rate and real interest rate in the American economy have been moving closely since 1995, with both fluctuating below 4% (Blanchard, 2013). This is an outcome of the so-called liquidity trap. Monetary institutions in the United States aim at stimulation economic growth by lowering nominal interest rates. The outcome has been no so much of an increase of household spending but a mobility of capital into more profitable markets.
The effect of changes in interest rate comes in two places. First the effect nominal interest rate has over money stock. The other effect, which is the one this paper is concerned with, is the effect over the goods market. The direct effect yields changes in investment.
This view main explain how expectations of the interaction between the nominal interest and the inflation lead investors to the housing market. Such movement of capital inflows helped create the increase in housing prices often cited by Case-Shiller index. More precise, the increase from 2000 to 2006. Some other indexes indicate somewhat similar. The index of monthly housing prices created by the United States Census Bureau has 1991 as year base, and also shows the housing price increase from 2000 to 2006. Graph number three shows the evolution of housing prices for the Census Bureau index. Perhaps the only difference with the Case-Shiller index is a couple of month lagging in the case of the Census Bureau index. Nonetheless, both indexes point out the sharp increase.
That increase in housing demand can also be seen in the number of housing units completed for the same period. Data taken from the United States Census Bureau indicates that (not seasonally adjusted) completion of housing units in the United States also went sharply up during the same period. From 1990 through 2006 completion of new housing buildings doubled. This reaffirm the statement on the housing price increase due to an increase in demand as well as financial capital seeking real sector to invest. Graph number four shows that such increase ranging from 1990 through 2006 was of about hundred percent. In 1991 the number of new housing built in the United States was roughly one million units. The same indicator for the year of 2006 grew up to roughly two million.
Total number statistics of household organized by income are in table #1. Graph #5 shows the aggregate housing building stock in the United States. 47 percent of Housing buildings in the United States are between 60years and 40 year aged. 33 percent of housing buildings are less than 35 years aged. And roughly 20 percent are more than 60 years aged.
Table # 1.
Total number of households by Income (numbers in thousands).
Sum of Less than $10,000 Sum of $10,000 to $19,999 Sum of $20,000 to $29,999 Sum of $30,000 to $39,999 Sum of $40,000 to $49,999 Sum of $50,000 to $59,999 Sum of $60,000 to $79,999 Sum of $80,000 to $99,999 Sum of $100,000 to $119,999 Sum of $120,000 or more
11401 12228 13882 11434 10212 8581 14330 10054 7195 16577
A bigger picture of the correlation between age of housing buildings and household income in the United States id depicted by graph number 6. That graph shows that the big chunk of buildings currently occupied in the United States were built during 1950’s through late 1970’s. Average Households in the United States tend to live in building built during those years. It is also reasonable to think that those same buildings –that same chunk- comprise a great percentage of the Mortgage-backed loans MBLs. If we assume those buildings require repairs and maintenance, we can assume Building Fatigue came into place as a contributing factor for the Great Recession.
Graph # 6.
The graph clearly indicates the distribution of occupied buildings by household income. Note that the great concentration of number of buildings are between 1950’s and late 1970’s. Those four columns may shed light into our hypothesis. Roughly 54 million housing buildings may be running progressively into building fatigue. This effect may be exacerbated by household income.
We look at the reasons why a Mortgage-backed loan MBL may go underwater. As defined by Blanchard, “underwater” means that the amount of the mortgage exceeds the market value of the house. Without trying to contest any of the current explanations about such phenomenon, it is important to consider how age of buildings combined with household income may have an effect on payments default. In the aggregate, we claim that the United States is currently in transition for renewing housing building stock. That is, more than half of the housing stock has aged over 40 years. Such condition makes structural aspects of the buildings deteriorating. Such effect may affect expectation of the mortgage-backed loans.
Therefore, we claim that there is a higher risk of defaulting based on buildings fatigue. We call that risk “Mortgage defaulting Risk derived from housing building fatigue”. We would argue that the defaulting risk from building fatigue is a function of Household Income (X); plus the age of the house (y) times a ratio of the house price relative to the amount of debt or Mortgage-Backed loan.
f(d)=βx+∂y(Avg Housing Price/Avg Mortgage loan)
If we consider a fact that low income people live in aged housing buildings, it is feasible to start to correlate some more aspects of such stylized fact. Our second hypothesis claims that mortgage loans go underwater more often as the housing building gets older, basically because aged buildings require more repairs than newer housing buildings. The empirical foundations for intuitively thinking this is the case are the following: 3% of the total housing building in the United States have moderate structural problem –mostly in kitchens. Data available in the survey for 2010 shows that roughly 3,939,000 building in the United Sates have problem of such type. A lower number, but in bigger trouble are 1,950,000 housing unit that count on severe physical problems. This are basically plumbing related deficiencies. It makes up to 1% of the total housing buildings. 1% may not sound alarming, but it certainly adds up, especially when such problems are coupled with signs of mice in buildings up to a 9%; and also signs of cockroaches of about ten percent of the total housing buildings in the United States.
This combination of factors is what we call the Housing building fatigue. Living condition after housing purchase may change rapidly creating investment depreciation. The economic activity known as housing flipping may be inflating expectation on the borrower’s side. Renovations performed in old building may increase temporarily the value of the property. In order to keep up with the purchase value, instead of depreciating, the borrower depends on neighbor’s own renovations. A kind of externality will have an effect on the initial value at which the borrower purchased the house. If the borrower bought a renewed house, which be older than say 40 year, its price will depends heavily on neighbors’ renovations. If neighbors renew their houses our borrower’s price will at least keep its value. If the neighbors do not renovate our borrower’s house may depreciate, and therefore the mortgage loan may go underwater.
The neighbors’ externality takes place in a housing market populated with housing buildings demanding renovations. That market segment happens to be concentrated at the low income brackets. The real interest rate is the analogues borrower’s property appreciation. Mortgage loans for many families represent the investment of their life. As such, investments are expected to give a return on capital. There is a borrower’s expectation of a future price increase that incentivize the purchase of the house. This is the same incentive that brings capital to the housing market. Thus, we can claim that there is a housing appreciation that mimics the interest rate in terms of the basket of goods –the real interest rate. We actually can call that the real-real-estate interest rate. We just need to change the level price for the average housing price in order to calculate this rate. Unlike the real interest rate, housing appreciation need not to follow investor’s decision on whether to hold bonds or equity, because the investment is made not necessarily for extracting profits. Nonetheless, this type of investment still requires to show some return, otherwise it goes underwater. As we said we just need to replace the price level (p) from the real interest rate formula.
(1+r_t )=((1+i ̇_t ) p_t)/(pe_(t+1) )
Replacing (P) with the average house price,
(1+r_t )=((1+i ̇_t ) 〖AVG House price〗_t)/(AVG House Price e_(t+1) )
Empirical testing of the first hypothesis:
Hypothesis number one: low income Americans tend to live in aged housing buildings.
We found not a strong correlation among the two variables. However there is a slight sign of such correlation that can get stronger as data allows for more observations and comparisons. That is, low income Americans tend to live in aged housing buildings.
We break down the data of income into tens brackets: less than $10,000; from $10,000 to $19,999; from $20,000 to $29,999; from $30,000 to $39,999; from $40,000 to $49,999; from $50,000 to $59,999; from $60,000 to $69,999; from $70,000 to $79,999; from $80,000 to $89,999; from $90,000 to $99,999; from $100,000 to $109,999; from $110,000 to $120,000 and over. We also break the housing age by fifteen brackets ranging from houses built from 1919 or earlier to houses built in 2013.
The first bracket data depicted in Graph # 1 of the annex shows a correlation between household income and year built of the building they live in. The correlation is a direct negative correlation telling us that it is unlikely to find household income than less than ten thousand dollars living in buildings built the last decade. This correlation gradually inverses as income rises. The highest bracket of household income which for this regression is $120,000 and over gives a positive correlation unlike the lower income bracket. This basically let us generalize –with some caveats- some of the results for the hypothesis. Such conclusion is that low income people tend to live in old houses, whereas high income people tend to live in newer building houses. More details can be seen in the following regressions. As income increase chances to find newer buildings occupied by those households rise either. The threshold where the correlation starts to be positive in at income level $80,000 and higher (Graph # 10 of the annex). Form that level of income it is likely to find such households living in newer building houses. The correlation and the r^2 become higher as income rises.
We proposed that a contributing factor of the origin of the Great Recession can be what we identify as: default risk in mortgages payments for low income Americans due to a housing building fatigue. In this first article we established empirically that there is a correlation between low income Americans dwelling in aged housing buildings. This correlation makes up just a first step in a series of hypothesis testing regressions in which the general hypothesis will be tested.
Second, by looking at the descriptive statistics we could state that the number of housing buildings in the U.S. were built between 1950’s and late 1970’s, which helps support intuitively our hypothesis. More precisely, 47 % of Housing buildings in the United States age between 60 years and 40 year; 33 % of housing buildings are less than 35 years aged. And roughly 20 % are more than 60 years aged.
Third, this article could also show data that confirm comparable findings of the Case-Shiller index. That is, the increase in housing demand could also be seen in the number of housing units completed for the period ranging from 2000-2006, which allowed us to reaffirm that housing price increased due to an increase in demand as well as financial capital seeking real sector to invest.
Fourth, leverage, complexity, and liquidity in the financial and banking sector may help explain how the crisis spread and how it was amplified. However, looking at the aspects related to housing age may help understand better how the housing sector affects consumption via housing repair needs. Furthermore, housing building conditions may shed light into an explanation of why mortgages go underwater since roughly 5% of the existing housing buildings in the United States have reported some sort of structure deficiency. Certainly, we cannot conclude so far that the age of the house and its neighbor’s effect may exacerbate the risk for mortgage payments defaults, but we can intuitively point out at that as a possible factor for increasing the risk.
Neither can we claim that renovations performed in old building may increase temporarily the value of the property, nor prices of these houses depend on neighbors’ renovations. We are certainly far from proving the risk of defaulting in a mortgage increases with the age of the house, and that housing prices depend on the pace at which old house buildings are renovated.
Graph # 1
Blanchard, Olivier, & Johnson, David (2013). Macroeconomics. Pearson. Sixth Edition.
Kalemi-Ozcon, S. & Sorensen, B. (2011). Leverage across firms, banks and countries. NBER, Working paper # 17,354.
Horton, B. (2013). Toward a more perfect substitute: how pervasive on the issuers of private-label mortgages-backed securities can improve the accuracy of rating. Boston University Review.