How Large Is the Housing Wealth Effect? A New Approach

I have had a quick read of Carroll, Otsuka, and Slacalek, How Large Is the Housing Wealth Effect? A New Approach (October 18, 2006). The model is essentially curve-fitting over historical data for the Marginal Propensity to Consume (MPC).

The conclusion of the paper is:

Housing price fluctuations apparently have substantial effects on consumer spending. The immediate (first-quarter) impact is likely to be relatively small (the immediate quarterly MPC in our preferred model is about 2 cents on the dollar), but over a time span of several years it probably accumulates to the 4–10 cent range. These figures are consistent with evidence from micro data and the experience across US states. Whether the housing wealth effect is substantially larger than the stock wealth effect is more uncertain; while the bulk of the evidence seems to point in that direction, the estimated size of the differences is not large enough (in US aggregate data) to yield confidence in the conclusion. For monetary policy purposes, these results suggest that it is important to keep a close eye on developments in housing markets separately from equity markets, since even the possibility of a significantly higher MPC out of housing wealth can shift the balance of risks in a macroeconomic forecast. Such a perspective, for example, could have helped in understanding and interpreting the surprising strength of the US consumption and residential investment spending in the early 2000s even as the stock market suffered a historic decline.

More importantly, the risks of the opposite experience are also worth noting. While in most places there seem to be good fundamental reasons for the rise in housing wealth over the last few years, the housing price dynamics in some areas might be driven by bubble components. If these components grow large and decline abruptly, our results suggest that consumption will be affected substantially.

Emphasis Mine

My ignorant opinion is that the choice of MPC as the model output and forms of wealth as the model input produces a substantial bias in the model.

MPC assumes smooth changes around a central value. That is, the actors rationally adjust their behaviours to changed conditions in a predictable. They would do more or less of the same things they had been doing before (habits). There is no edge to a MPC where the actors start changing their habits.

The use of forms of wealth as model input is unreliable because it ignores the reasons for consumption:

• Labour reproduction - immediate (food, accommodation, transport costs, rest, clothing, tools, medical, etc.)
• Labour reproduction - long-term (children, education, preventative health-care, etc.)
• Capital reproduction (investment, debt servicing, etc.)
• Luxuries (anything else)

The sources of income to satisfy these different consumptions is wages, interest payments, rents, and profits. These do not arise directly from the various forms of wealth. As Marx wrote, capital that is not involved in the production process, is dead capital. This paper only considers dead capital. It is the living capital that drives the economy.

I think the paper is a relic of its times and will not be useful over the coming economic crisis before the actors are going to have to change their habits drasticly in order to survive. And these forms of wealth (housing and stocks) will be exposed truly as dead capital.