- Separable Taste Shocks. Shocks to (current or future) housing tastes that do not affect substitution patterns in the non-residential sector will have no aggregate wealth effect. As Buiter (2008) explains, persons long housing will gain when tastes shift toward housing (and thereby increase housing prices), and persons short housing will lose. [non-technical version: if your taste for housing also affects your willingness to work, then the non-residential sector will be affected, although we would not call the effect a "wealth effect".]
- Shocks to the Production Set. A larger production set can increase housing prices, and is an aggregate wealth effect by any definition. The wealth effect calculations I made with Luke Threinen are most easily interpreted in such a model, although they can also be interpreted as the result of redistribution created by separable taste shocks.
- "Distortion" shocks that move the economy closer to the frontier of the production set. Increasing housing subsidies can both increase housing prices and look like an adverse wealth effect from the perspective of household behavior in the non-residential sector. Nonseparable taste shocks may have similar effects.
Changes in the production set or changes in the distance between the market allocation and the production frontier involve wealth effects by any definition. An example: the arrival of information about the technology for producing housing services. People may learn that the technology will advance, which causes a boom in housing prices and a favorable wealth effect. The market may initially believe that banks can streamline the mortgage process and thereby facilitate home-ownership for small home-owners. Or it may believe that the real estate brokerage industry will become more productive. Good news like this boosts housing prices and is an aggregate wealth effect. If the market later learns that ultimately the technological advance will not work (e.g., learn that small homeowners are really more expensive to service than initially thought), housing prices crash and there is an adverse wealth effect.
This conclusion might appear to disagree with Buiter (2008), which argues that changes in housing prices are not wealth effects in the sense that they do not enter the demand for non-housing goods because housing is both an asset and a consumption good. Buiter emphasizes that someone will consume the housing stock – even if it isn’t the current owners – and those prospective consumers gain from a housing price crash even while the current owners lose. I agree that housing prices changes that reflect changes in the current or expected future (separable) preferences for housing are ultimately redistributive. Demand changes are not changes in the production set or changes in the distance between the market allocation and the production frontier. But it is quite possible that the shocks that created the housing boom prior to 2006 and the housing price crash that followed were technological or public policy driven. A technology shock will have aggregate wealth effects, and a public policy shock might too.
 Another question is whether the persons short housing are represented in today’s economy (maybe they are unborn, or live abroad, or only exist in the imagination of exuberant housing investors), but my purpose here is not to reclassify a redistributive wealth effect as an aggregate one.