Oct012008

The Housing Bubble – Part 1

The Housing Bubble

Prices went up a large amount during the Great Housing Bubble, but what makes this price increase a bubble? To answer this question it is necessary to accurately measure price levels and review historic measures of affordability to establish these price levels are not sustainable. [1] Measuring house prices is not a simple task, and there are many methods market watchers use to evaluate market prices. These include the median, the average cost per square foot, and the S&P/Case-Shiller indices. Price levels in financial markets represent the collective result of individual actions. There are techniques to measure the actions of the individual market participants and their impact on house prices. These measures are debt-to-income ratios and price-to-income ratios.  The amount of debt people are willing to take on compared to the income they have available is their debt-to-income ratio. The amount of money people are able to put toward the purchase of residential real estate compared to their income is their price-to-income ratio. These ratios are important because they show how much people are borrowing and spending from their earnings to acquire real estate. When these ratios break with historic patterns, they signify a housing bubble.

There is a point where people are not able to bid up prices any higher because they do not have the savings or the borrowing power to pay more. This affordability limit determines where bubble rallies end; however, this limit is not predetermined or in a fixed location. The purpose of exotic financing programs is to expand this limit and bring more customers to the market and generate fees for the lenders. Unfortunately, these products have continually proven to be unstable, and the high default rates and lender losses inevitably lead to a contraction of credit known as a credit crunch. Interest-only and negative amortization loans created the housing rally and their elimination due to borrower default created the housing crash.  As mentioned previously, the housing bubble was a credit bubble.

Price Measurements

There is no perfect measure for any broad financial market activity. Markets for stocks, bonds and other securities are the most widely reported and measured financial markets. It is relatively easy to measure activity in these markets because all sales are recorded at a few central exchanges and the “products” are uniform (one share of stock is equal to another). In contrast, real estate markets are much more difficult to evaluate. [ii] Real estate transactions are recorded into the public record in thousands of locations across the country. Keeping an organized database of these records is such a daunting task that the title insurance industry has taken this responsibility as part of its business model, and many people are devoted to the arduous task of obtaining and organizing these records on a daily basis. Real estate does not have the uniformity of stocks or other financial instruments. Each property has unique qualities that differentiate it from all other properties making like-kind comparisons very difficult. Geographical location is a major influence on the value of real estate. Even if two properties could be found with identical physical characteristics, the values of these properties could vary considerably based on where they are located. Ideally, a market measure would record the changes in sales prices of identical assets or in the case of an index, a group of similar assets. The unique nature of real estate assets makes it difficult to use standard measures of reporting utilized in other financial markets.

Due to the problems of asset uniformity and variability based on location, real estate markets are typically measured using some form of median pricing over a specified geographic area. The median is a statistical measure of central tendency where half the data points are above and half the data points are below. For instance, in a list of 5 numbers sorted by size ($100,000, $200,000, $300,000, $500,000, $900,000,) the third number in the list ($300,000) would be the median because it has two numbers that are larger and two numbers that are smaller. The median ($300,000) is used rather than an average ($400,000) because a few very expensive properties can increase the average significantly, and the resulting number does not represent the bulk of the price activity in the market.

One of the problems with a median as a measure of house prices is a lag between when a top or a bottom actually occurs and when this top or bottom is reflected in the index. During the beginning of a market decline, the lower end of the market has a more dramatic drop in volume than the top of the market. This causes the median to stay at artificially high levels not reflective of pricing of individual properties in the market. In other words, for a time things look better than they are. At the beginning of a market rally, transaction volume picks up at the bottom of the market at first restarting the chain of move ups. During this time, the prices of individual properties can be moving higher, but since the heavy transaction volume is at the low end, the median will actually move lower.

The median is a good measure of general price activity in the market, but it does have a significant weakness: it does not indicate the value buyers are obtaining in the market. The houses or structures built on the land compose the most significant portion of real estate value in most markets. These structures deteriorate over time and require routine maintenance that is often deferred. During times of prosperity, many people renovate homes to add value and improve their living conditions. The impact of deterioration and renovation of individual properties is not reflected in the median resale value. Also, at the time of sale, there are often buyer incentives which inflate the recorded sales price relative to the actual cost to the buyer. These buyer incentives also distort the median sales price as a measure of value.

Many data reporting services measure, record, and report the average sales cost on a per-square-foot basis to address the problem of evaluating what buyers are getting for their money. For instance, in a declining market if people start buying much larger homes at the limit of affordability, the generic median sales price would remain unchanged, but since buyers are getting much larger homes for the same money, the average cost per-square-foot would decline accordingly. This makes the average cost per-square-foot a superior measure for capturing qualitative changes in house prices; however, this method of measurement does not capture the relative quality of the square footage purchased, only the price paid for it. High quality finishes may justify a higher price per square foot. There is no way to objectively evaluate the impact finish quality has on home prices. The main problems with using the average cost per-square-foot to measure price is that it does not provide a number comparable to sales prices since it has been divided by square feet, and it is not widely measured and reported.

Figure 15: National S&P/Case-Shiller Home Price Index, 1987-2007

S&P-Case-Shiller Home Price Index, National 1987-2007

 

To address some of the weaknesses of the generic median sales price as a measure of market value, Karl Case and Robert Shiller developed the Case-Shiller indices for measuring market trends. [iii] This index measures the change in price of repeat sales. It solves the dilemma of pricing like-kind properties–almost. Although these indices capture the price movements of individual properties far better than the generic median sales price, it does not take into account value added through renovation and improvement. To address this issue, the index gives less weight to extreme price changes assuming the outlier is a significant renovation. However, if there is a market-wide renovation of properties, as was the case in many markets during the Great Housing Bubble; this will cause a distortion in the index. The other weaknesses of the Case Shiller indices concern how and where it is reported. Since it is an index of relative price change rather than a direct measure of price, the index is reported as an arbitrary number based on a baseline date; therefore, the numbers are not useful for evaluating current pricing. The index is also confined to 20 large metropolitan areas around the United States. The large geographical coverage areas are required to obtain enough repeat sales to construct a smooth index. The broad yet limited geographical coverage fails to capture price changes in smaller markets. Also, since the Case-Shiller index is a measure of changes in prices of sales of the same home, it does not include any newly constructed homes. No measure is perfect, but the Case-Shiller index is the best at measuring historic movements in pricing because its methodology is focused on repeat sales of the same property.

Figure 16: Los Angeles S&P/Case-Shiller Index, 1987-2007

S&PCase-Shiller Index, Los Angeles 1987-2007

 

The examples from this work will use the median sales price, not because it is the best method, but because it is the most widely used and best understood of the common measures. Also, since it gives a number reflective of sales values in the marketplace, it is the easiest to understand and interpret. This measure has weaknesses, but over time it does a reasonable job of documenting overall prices and trends in the marketplace.

Figure 17: Median Home Prices, 1968-2006

Median Home Prices 1968-2006

 

The Great Housing Bubble was an asset bubble of unprecedented proportions. Between 2000 and 2006, home prices increased 45% nationally, and in California home prices increased 135%. [iv] Had this amazing price increase coincided with a period of high inflation, it may not have been indicative of a price bubble, merely the general increase in prices of all goods and services; however, inflation was low during this period. The inflation adjusted price increases nationwide were 23% and in California it was 100%. There was no great improvement in the quality of houses justifying the higher prices. Although some homeowners made cosmetic improvements, the vast majority of homes were unchanged during this period, and many deteriorated with age. Resale homes did not undergo any form of manufacturing process where value was added to the final product. There was little real wealth created during the bubble, just a temporary exaggeration of value.

Price-To-Income Ratios

Price-to-income ratios represent the amount borrowed relative to the incomes of the borrower. There are many variables that impact house prices, and some of the variability in prices over time can be attributed to changes in these variables; however, since most houses are purchased with lender financing, and since lender financing is linked to income, the price-to-income ratio is the best metric for evaluating long-term housing price trends. The price-to-income ratio does not need to be adjusted for inflation as both prices and income will rise with the general level of inflation. Most of the fluctuations in the ratio are based on changes in financing terms, in particular interest rates, and of course, irrational exuberance.

The Great Housing Bubble saw unprecedented price-to-income ratios because interest rates were at historic lows and the use of exotic financing including negative amortization loans were at historic highs. When measured against historic norms of house price to income, the degree of price inflation was staggering. [v] In markets where bubble behavior is not prevalent, price to income ratios hover between 2.3 and 2.8. In bubble markets there is a tendency to maintain higher ratios, and the range over time is much greater. Any ratio less than 3 is generally considered affordable.

Figure 18: National Ratio of House Price to Income, 1986-2006

National Ratio of House Price to Income 1986-2006

 

In bubble markets ratios of 3 to 4 are as affordable as they get. Anything greater than 4 is a strain on family budgets and generally a sign of an inflated market. Ratios greater than 5 are considered very unaffordable and prone to high rates of default because they tend to be characterized by exotic financing. Price-to-income ratios in the bubble of the early 90s in California did not exceed 6 because interest rates were higher and because negative amortization loans were not widely available. During the Great Housing Bubble, the national ratio of house price to income increased 30% from 4.0 to 5.2. This means 30% more debt is serviced by the same income. Some of this increased ability to service debt is explained by lower interest rates and exotic loan terms, and some of this increase came from people choosing to take on larger debt loads due to the irrational expectation of ever increasing house prices coupled with loose lending standards which enabled the populace to take on these debts. The national trends were small compared to the frenzied activities of bubble markets in California where most markets saw their house price to income ratio double.

Figure 19: Price-To-Income Ratio in California, OC and Irvine, 1986-2006

Price-To-Income Ratio in California, Orange County and Irvine 1986-2006

 

Buyers were never forced to buy; it was always a choice. During the market rally, greedy buyers motivated by rising prices and fueled by loose lending standards were able to bid prices up to ridiculous levels. The exotic financing was not a result of high prices; it was the cause of high prices. Lenders were keen to offer these products because they were not taking on the risk, and it allowed them to keep transaction volumes high. The lenders profits came from transaction volume. By late 2007, the market balance had shifted from favoring sellers to favoring buyers. The once greedy buyers were becoming desperate sellers: their dreams of riches from perpetual appreciation were in tatters. Many were forced to sell due to their inability to make their mortgage payments. Those that hung on were homeowners with 50% or more of their income going toward paying off an asset which was declining in value. It was not a set of circumstances to be envied.

Price-To-Rent Ratios

Price-to-rent ratios represent the cost of a dwelling unit relative to the cost of a comparable dwelling unit. This ratio is also subject to the same variability exhibited by the price to income ratio. [vi] This is not surprising considering rent is generally paid out of current income, so incomes and rents tend to track one another fairly closely. The ratio of rent to income has stayed within a range from 13.6% to 16.5% from 1988 to 2006. This demonstrates renters have been putting roughly the same percentage of their incomes toward housing for the 18 years period of data examined. The evidence from the sudden and dramatic changes in the price-to-income ratio and the price-to-rent ratio points to a housing bubble. [vii] If these two measures of value had been supported by a rise in the rent-to-income ratio, the increase in prices might have been explainable by a shortage in dwelling units causing all consumers of housing to see an increase in the percentage of their income going toward housing. Evidence from the rent-to-income ratio is to the contrary.

Figure 20: National Price-to-Rent Ratio, 1988-2007

National Price-to-Rent Ratio 1988-2007

 

Debt-To-Income Ratios

There was a significant price bubble in residential real estate in the late 1980s crashing in the early 1990s. This coastal bubble was concentrated in California and in some major metropolitan areas in other states, and it did not spread to housing markets nationwide. When comparing this previous bubble to the Great Housing Bubble, the macroeconomic circumstances were different: prices and wages were lower in the last bubble, interest rates were higher, the economies were different, and other factors were also unique; however,  the evaluation of personal circumstances each buyer goes through when contemplating a purchase is constant. The cumulative impact of the decisions of buyers is represented in the debt-to-income ratios–how much each household pays to borrow versus how much they make. Comparing the trends in debt-to-income ratios provides a great tool for elucidating the behavior of buyers.

Typically debt-to-income ratios track interest rates. As interest rates decline, it becomes less expensive to borrow money so borrowers have to put less of their income toward debt service. The inverse is also true. On a national level from 1997 to 2006 interest rates trended lower due to low inflation and a low federal funds rate. During this same period people were increasing the amount of money they were putting toward home mortgage debt service. If the cost of money is declining and the amount of money people are putting toward debt service is increasing, the total amount borrowed increases dramatically. Since most residential real estate is financed, this increased borrowing drove prices up and helped inflate the Great Housing Bubble.

Figure 21: Debt-To-Income Ratio and Mortgage Interest Rates, 1997-2006

Debt-To-Income Ratio and Mortgage Interest Rates, National 1997-2006

 

The figure on the following page shows the historic debt-to-income ratios for California, Orange County and Irvine from 1986 to 2006. It is calculated based on historic interest rates, median home prices and median incomes. Lenders have traditionally limited a mortgage debt payment to 28% and a total debt service to 36% of a borrower’s gross income. The figure shows these standard affordability levels. During price rallies, these standards are loosened in response to demand from customers when prices are very high. Debt service ratios above traditional standards are prone to high default rates once prices stop increasing. In 1987, 1988 and 1989 people believed they would be “priced out forever,” so they bought in a fear-frenzy creating an obvious bubble. Mostly people stretched with conventional mortgages, but other mortgage programs were used. This helped propel the bubble to a low level of affordability. Basically, prices could not get pushed up any higher because lenders would not loan any more money.

Figure 22: Debt-To-Income Ratio, California 1986-2006

Debt-To-Income Ratio, California 1986-2006

 

Changes in debt-to-income ratios are not a passive phenomenon only responding to changes in price. The psychology of buyers reflected in debt-to-income ratio is the facilitator of price action. In market rallies people put larger and larger percentages of their income toward purchasing houses because they are appreciating assets. People are not passively responding to market prices, they are actively choosing to bid prices higher out of greed and the desire to capture the appreciation their buying activity is creating. This will go on as long as there are sufficient buyers to push prices higher. The Great Housing Bubble proved that as long as credit is available there is no rational price level where people choose not to buy due to prices that are perceived to be expensive. No price is too high as long as they are ever increasing.

In market busts, people put smaller and smaller percentages of their income toward house purchases because the value is declining. In fact, it is possible for house prices to decline so quickly that no mortgage program can reduce the cost of ownership to be less than renting. The only thing justifying a DTI greater than 50% is the belief in high rates of appreciation. Why would anyone pay double the cost of rental to “own” unless ownership provided a return on that investment? Once it is obvious that prices are not increasing and even beginning to decrease, the party is over. Why would anyone stretch to buy a house when prices are dropping? Prices decline at least until house payments reach affordable levels approximating their rental equivalent value. At the bottom, it makes sense to buy because it is cheaper than renting. In a bubble market when the market debt-to-income ratio falls below 30%, the bottom is near.



[1] There were some valiant attempts during the bubble to determine if the price increases really were a bubble. The literature of the time almost universally missed it despite the obvious signs in the data. In the paper Assessing High House Prices: Bubbles, Fundamentals, and Misperceptions (Himmelberg, Mayer, & Sinai, 2005) the authors used almost the same approach to the analysis presented in this writing and reached the opposite conclusion, “As of the end of 2004, our analysis reveals little evidence of a housing bubble. In high appreciation markets like San Francisco, Boston, and New York, current housing prices are not cheap, but our calculations do not reveal large price increases in excess of fundamentals.” By the end of 2004, the data was unambiguously in support of a financial bubble. One of the few authors who recognized the problems early on was John Krainer an economist with the Federal Reserve Board of San Francisco (Krainer, House Price Bubbles, 2003).

[ii] Jordan Rappaport provides an overview of the various methods of house price measurement in A Guide to Aggregate House Price Measures (Rappaport, 2006).

[iii]  In the paper A Note on the Differences between the OFHEO and S&P/Case-Shiller House Price Indexes by Andrew Leventis (Leventis, 2007), the author makes the following observation: “OFHEO’s House Price Indexes (the “HPI”) and home price indexes produced by S&P/Case-Shiller are constructed using the same basic methodology. Both use the repeat-valuations framework initially proposed in the 1960s and later enhanced by Karl Case and Robert Shiller. Important differences between the indexes remain, however. The two models use different data sources and implement the mechanics of the basic algorithm in distinct ways.”

[iv] Praveen Kujal and Vernon L. Smith noticed an interesting phenomenon in the studies of perceptions of fairness in their paper (Kujal & Smith, Fairness and Short Run Price Adjustment in Posted Offer Markets, 2003), “perceptions of fairness cause people to resist price increases following abrupt changes in conditions with no cost justification. Fairness is thus interpreted as being a result of expectations that are not sustainable.” This implies that people have an intuitive sense that nothing is justifying the dramatic increase in prices during a bubble rally. There is no perception of fairness because houses are not any better, nor are houses any more expensive to build. The increase in prices has no justification in cost. Carl Case and Robert Shiller also noticed the same behavior among sellers in financial manias who felt guilty that the buyer paid so much (Case & Shiller, The Behavior of Home Buyers in Boom and Post-Boom Markets, 1988).

[v]  In the paper Are House Prices the Next “Bubble?” (McCarthy & Peach, 2004) the authors completely missed the implications of the rising price-to-income ratio. Some amount of the increase in price (less than 50%) nationally can be attributed to lower interest rates. The authors make the statement, “The marked upturn in home prices is largely attributable to strong market fundamentals: Home prices have essentially moved in line with increases in family income and declines in nominal mortgage interest rates.” An analysis of the impact on lower interest rates on actual payments and debt-to-income ratios would have revealed their conclusion erroneous, but no such analysis was undertaken. In the paper (Gallin, The Long-Run Relationship between House Prices and Income: Evidence from Local Housing Markets, 2003) Joshua Gallin reaches the following, completely erroneous conclusion, “More formally, many in the housing literature argue that house prices and income are cointegrated. In this paper, I show that the data do not support this view. Standard tests using 27 years of national-level data do not find evidence of cointegration.”

[vi] The paper for the Federal Reserve Board by Joshua Gallin, (Gallin, The Long-Run Relationship between House Prices and Rents, 2004) demonstrates there is a relationship between these variables long term. What is interesting is the Mr. Gallin did not reach the same conclusion with respects to the relationship between house prices and income (Gallin, The Long-Run Relationship between House Prices and Income: Evidence from Local Housing Markets., 2006) despite the fact that rents and income track each other very closely.

[vii] In the paper Housing: Boom or Bubble (Schiller, 2007), author Tim Schiller shows a chart on page 17 that looks very similar to the one in this work (He used a different data source, but the results were almost the same.) The chart shows the obvious sign of a massive housing bubble with prices showing a deviation in the price-to-rent relationship 5 times the previous high of the coastal bubble of the early 1990s. Despite the visual appearance of the chart, he goes on to say the rally in prices was supported by fundamentals. Obviously, he was proven wrong. There is a history of scholarly papers on the price-to-income ratio that completely missed the housing bubble.