In the wake of the economic crisis that began in 2007, homeowners in many countries have faced substantial losses. Prices have fallen in both nominal and real terms. In the US, for example, house prices in the first quarter of 2012 were down more than 40% in real terms from their peak (Shiller 2012). Nevertheless, housing remains a popular investment1. This popularity is surprising because, over the post-war period, US house prices have been essentially flat in real terms while the US stock markets have risen more than fourfold in real terms over the same period2.
House prices and psychology
The change in market value of the asset is of course only part of the overall return on investment – housing provides housing services and shares typically pay dividends3. Nevertheless, the market value is important, particularly in psychological terms, because of the possibility of losses. It is well established that losses loom larger in people’s minds than corresponding gains (Kahneman and Tversky 1979), but the results of our study suggest a particular dislike of losing money – that is, an aversion to nominal losses.
When viewed in nominal rather than real terms, the capital gains from US housing look more appealing. From 1946 to 2012, nominal house prices showed a 12-fold increase. On an annual basis, housing investments have mostly resulted in gaining money (in 58 out of 66 years), while at the same time producing real losses more often than not (in 36 v 30 years)4.
Perceptions of investments
In a recent CEPR paper, we show that perceptions of housing transactions are shaped by gaining versus losing money, even when real losses are held constant. Our research builds on evidence from US housing markets showing that homeowners are reluctant to sell when facing nominal losses (Anenberg 2011, Engelhardt 2003, Genesove 2003, Genesove and Mayer 2001). Our survey experiment, with a large, heterogeneous sample, adds to these studies by relating evaluations to detailed information about decision makers and by clearly pinning down the role of nominal losses (as opposed to nominal changes more generally), using controlled variation of the environment.
The starting point of our paper is the fact that, in the presence of inflation, real and nominal losses need not coincide. To illustrate, imagine buying a house for $200,000 in cash, and selling it several years later for $170,000. Without inflation, the nominal and real losses will coincide at 15%, irrespective of the holding period. With even low, stable inflation, however, the nominal loss will rapidly disappear. If inflation is 2%, a real loss of 15% will become a nominal gain within nine years.
To measure the effect of nominal gains vs. losses on perceptions, we present subjects with hypothetical housing transactions involving the purchase and subsequent sale of a house, and ask them to evaluate the advantageousness of these transactions. None of the transactions are in fact advantageous: they all involve smaller or larger real losses. However, each real transaction is presented twice (on separate screens); once with low inflation, so that it involves losing money (a nominal loss), and once with high inflation, so that it involves gaining money (a nominal gain).
We then take differences between evaluations of a given real loss when presented as a nominal gain and nominal loss, and average them. This gives us an index of nominal loss aversion – a number indicating the strength of a subject's dislike of losing money – for each subject.
Figure 1 shows the distribution of the nominal loss aversion index. A subject concerned solely with real gains or losses would have an index of zero, as indeed do 17% of our subjects. The rest, however, are heavily skewed towards positive values, with 73% of the index values being positive – indicating a dislike of losing money – against only 10% that are negative. Treating the 10% as symmetrical noise, about 60% of our subjects prefer identical real losses when they gain rather than lose money.
Figure 1. Distribution of nominal loss aversion index
A unique advantage of our subject pool is the availability of detailed official socioeconomic data from the National Bureau of Statistics (Statistics Denmark), as well as the results of cognitive and personality tests. This data set allows us to identify which sorts of people are prone to nominal loss aversion.
Within our sample, we find that subjects with more education and higher incomes tend to be less likely to let monetary gains or losses influence their evaluations. At the same time, there is no significant difference between those who own property and those who do not, suggesting that this bias is not eliminated by experience5.
Cognitive measures are particularly strongly correlated with nominal loss aversion. Subjects with higher cognitive ability are far less likely to be influenced by purely nominal differences. The most important aspect of cognitive ability is not intelligence per se, but cognitive reflection (Frederick 2005) – a tendency to rely on slower, more deliberative cognitive processes rather than rapid, intuitive ones. Taken together, our findings on education and cognitive reflection suggest that there may be scope for reducing nominal loss aversion through improvements in financial education.
Gaining or losing money is key
To separate a dislike of losing money from simple nominal thinking, we ran a second survey experiment with transactions involving both real losses and real gains. We duplicated the first experiment, but added a second treatment in which all of the real losses were changed to equivalent real gains. Struck by the powerful effect of nominal loss aversion observed in our first experiment, we were also curious to learn if the effect of nominal loss aversion would persist if the scenarios were presented to subjects in a highly transparent way, on a single screen. We find that it does.
Figure 2 shows average evaluations of the housing transactions in the second experiment. The two leftmost bars show the average evaluations of real losses, and the two rightmost bars show the average evaluations of real gains. The light bars represent transactions with low inflation, and the dark bars transactions with high inflation.
Figure 2. Evaluations of housing transactions at high vs. low inflation, real losses (left) v real gains (right)
Note: Average evaluations by real and nominal treatments, with 97.5% CIs for means, and nominal GAIN or LOSS.
Comparing the two bars on the left with the two on the right, it is clear that real gains are viewed more favourably than real losses, as should be the case from the perspective of standard economics. At the same time, a comparison of the first two bars shows that evaluations of losses are shaped by gaining or losing money. Identical real losses are viewed more favourably when they involve gaining rather than losing money.
In contrast to real losses, higher inflation has essentially no effect on evaluations when holding real gains constant (compare the third and fourth bars). This contrast (the two bars on the left are different, the two on the right are not) shows that subjects are not simply thinking in nominal terms, but rather dislike losing money.
Many people view housing as an attractive investment with good potential, despite meagre real capital gains over the long run. We suggest nominal loss aversion as a psychological mechanism that can help to explain the surprising popularity of housing as an investment. Using data from a survey experiment, we find that evaluations of housing transactions are shaped by gaining or losing money. We find no evidence that property ownership reduces this bias, but do find strong evidence that more education and greater cognitive reflection do. These results suggest that better financial education may reduce this bias towards overinvesting in housing.
Anenberg, E (2011), “Loss Aversion, Equity Constraints and Seller Behavior in the Real Estate Market”, Regional Science and Urban Economics, 41(1), 67-76.
Engelhardt, G V (2003), “Nominal Loss Aversion, Housing Equity Constraints, and Household Mobility: Evidence from the United States”, Journal of Urban Economics, 53(1), 171-195.
Fannie Mae (2012), “Fannie Mae National Housing Survey”, First Quarter.
Frederick, S (2005), “Cognitive Reflection and Decision Making”, Journal of Economic Perspectives, 19(4), 25-42.
Genesove, D (2003), “The Nominal Rigidity of Apartment Rents”, Review of Economics and Statistics, 85(4), 844-853.
Genesove, D and C Mayer (2001), “Loss Aversion and Seller Behavior: Evidence from the Housing Market”, Quarterly Journal of Economics, 116(4), 1233-1260.
Hasanov, F and D C Dacy (2009), “Yet Another View on Why a Home Is One's Castle”, Real Estate Economics, 37(1), 23-41.
Kahneman, D and A Tversky (1979), “Prospect Theory: Analysis of Decision under Risk”, Econometrica, 47(2), 263-291.
Shiller, R J (2012), Online data, accessed 4 November, http://www.econ.yale.edu/~shiller/data.htm.
Stephens, T A and , J-R Tyran (2012), "’At Least I Didn't Lose Money’ – Nominal Loss Aversion Shapes Evaluations of Housing Transactions”, CEPR discussion paper, DP9198.
1 A survey by Fannie Mae (2012) in the first quarter of 2012 found that 58% of US citizens viewed housing as an investment with ‘a lot of potential’, and 65% viewed it as a ‘safe’ investment. By comparison, 55% viewed shares as an investment with ‘a lot of potential’, and only 15% viewed them as ‘safe’.
2 From 1946 to the start of 2012, US house prices increased in real terms by about 7% (Shiller 2012). The house price data refer to individual properties sold more than once, to control for changes in housing characteristics over time. The stock market figure is the Dow Jones Industrial Average (DJIA).
3 Other relevant differences include liquidity, transaction costs and taxation. See Hasanov and Dacy (2009) for a comparison of overall returns from 1952–2005.
4 Excluding the post-bubble price declines from 2006–2011, housing gained money, on average, in 57 of 60 years. The DJIA, in contrast, was more likely to provide positive than negative annual real capital gains (41 years v 25 years), but viewing these gains in nominal terms makes little difference (positive nominal gains in 47 years v negative in 19).
5 The lack of an experience effect is perhaps not surprising, since the number of transactions over a lifetime tends to be relatively low.