Economists often talk of a ‘wealth’ effect in terms of private consumption. In economies such as the US, where private consumption is around 70% of GDP, determining the size and significance of this wealth effect is rather important. An important question, however, is whether or not all wealth effects are the same, or are some forms of private wealth more influential than others in encouraging consumption?
In a paper that revisits their earlier work on this topic, Karl E. Case, John M. Quigley, and Robert J. Shiller have provided some rigorous and highly informative analysis to answer this question. To preview the conclusion of the paper, the authors find that housing wealth has more of an impact on consumption than stock market wealth. Furthermore, this impact is symmetric, meaning that losses in housing wealth matter just as much as increases in housing wealth. This is especially relevant to the US economy, where households have seen a significant drop in housing wealth since the great recession.
How they did it
Case et al used a range of multivariate techniques to assess this question. Rather than proposing a theoretical framework for how these factors are related to consumption, the authors tested various specifications of the econometric models to ensure the robustness of the findings. Setting a high standard for other econometrician’s to follow, the various models showed the relationships to be robust to different specifications.
The four main series used were housing wealth, financial wealth, retail sales (as a proxy for consumption) and income. The data was at the state level, transformed using logarithms, deflated so they are ‘real’ and presented in per capita terms. As well as the standard ‘ordinary least squares’ model, Case et al also looked at ‘first difference’ and ‘error correction’ models.
What they found
Case et al, across these different specifications, consistently found that housing wealth matters for consumption with its effect being larger than that of stock market wealth. The results showed elasticities of housing wealth to consumption in the range of 0.03 to 0.18, implying for example, that a 1% increase in housing wealth would increase consumption in the range of 0.03% and 0.18%.
There were also additional findings that are worth mentioning. Unlike their previous study, Case et al found evidence that declines in housing wealth are also significant. The elasticity in the down market was around 0.1%. Moreover, as the error correction models are a real interest of mine, it was important to see that even when the long run relationship between income and consumption was accounted for, the short run impacts of changes in housing wealth still had an influence in the range cited above.
Why is this important
This work from Case et al is very important in at least two respects. Firstly, by quantifying these effects, it has really helped our understanding of the impact of the housing market on consumption. Given what has happened in the US housing market since 2007/2008, this enables some economic perspective on the housing crash. As Case et al note, with consumption about $10 trillion, a 35% decrease in housing wealth would reduce consumption by $350 billion.
Secondly, this work matters for economic policymaking. One of the targets of QE, or quantitative easing, is the stock market. The logic is that by reducing the yield on bonds, this encourages investors to buy equities, which pushes up the stock market. This increase in the stock market makes people feel wealthier and encourages spending. But as Case et al show, the housing market is much more influential in encouraging private spending. The US Federal Reserve is aware of this, as they are also buying ‘mortgage backed securities’, with the aim of driving down mortgage rates. While this is important, it also seems that policy towards housing debt would have also been useful in the Great Recession, as outlined by the IMF .
Finally, given my own research interests, the concept of deleveraging must have some relevance here. There is little doubt that housing wealth matters, but perhaps what also matters is housing and stock market wealth relative to factors like income or debt. It would be highly interesting, therefore, to explore these same specifications with a control variable of household debt and possibly an interaction term of household debt with household wealth or income. This research would build on the notion that housing wealth matters, but it would begin to shed light on deleveraging: we know that the private sector is deleveraging, but where is it deleveraging to?