Economists across the political spectrum almost unanimously oppose the home mortgage interest tax deduction. While I'm sure there are a few whom I've missed, I could only find a single professional economist who supports it. Greg Landsburg believes that capital income should be untaxed, based on the Chamley-Judd model and its subsequent refinements. He argues that the home mortgage interest tax deduction effectively cancels the distortionary tax on home mortgage interest capital income, and so represents a step in the right direction, although he would prefer a complete elimination of all capital income taxation.

But does housing really counts as "capital" in the Chamley-Judd sense? A textbook example of capital gains is a lender lending money to a borrower who uses it to buy sewing machines, tractors, etc. and start up a new company, which then turns a profit and allows the borrower to repay the loan with interest. In this example, everybody wins (the borrow, the lender, and the new company's customers), and the Chamley-Judd result that these kind of loans should not be taxed is very reasonable and intuitive.

But as Greg Mankiw's Intro to Economics textbook points out, housing is somewhat of an edge case between capital and consumption, and the fact that "capital" is traditionally defined to include housing is a bit of a historical accident. Morever, there is debate among economists about whether or not housing should be properly classified as capital. (Mankiw also wrote a New York Times column on this issue.)

I'm not familiar with the precise assumptions of the Chamley-Judd model, but does the main result that capital income should be untaxed include income from housing appreciation? People typically buy first homes (the only kind to which the mortgage interest deduction applies) for their own personal use, not primarily as investment opportunities. It's not obvious to me why housing purchases would lead to the creation of new economic wealth as more traditional capital investments do. In contrast to Landsburg, Uwe Reinhardt claims that

If pressed, the proponents of the capital-gains preference might concede that a tax preference for gains on transactions in residential real estate is hard to defend.

Under the assumptions of the Chamley-Judd model (or looser, less formal economic reasoning), is housing naturally classified as capital whose appreciation should not be taxed?


3 Answers 3


I think there are a few different ways one can go about examining the topic and general question you bring up. For that reason, I'd like to highlight that this isn't my primary field, and so my answer might not align with a more expert view from that field. However, to get the ball rolling...

First, we should probably take a short digression and consider the intuition behind what their results suggest; why exactly is taxation on capital income so problematic? Generally, it's because it creates a comparatively large distortion in future consumption. As Banks and Diamond (2010) shows starting on page 31, capital taxation leads to increasingly large wedges in consumption as the number of periods increases. Unlike labor taxation, which creates large, but contemporary, distortions in consumption, capital taxation creates intertemporal distortions.

From that point, I think there's a growing literature questioning whether the traditional Chamley-Judd result is actually as substantial in application (as opposed to theory) as some consider it to be. For example, there are several, different papers which get non-zero optimal taxation rates on capital if there's endogeneity in wages, or if wages from different types of labor cannot be taxed at their individually optimal rates. In fact, one of these potentially problematic areas is entrepreneurship: as Reis' dissertation (2007) points out in the third chapter, "when entrepreneurial labor income cannot be observed separately from capital income, then it is optimal to have positive capital taxation in the long run." A pretty comprehensive analysis by Straub and Werning (2015) does a wonderful job of thoroughly analyzing both models presented by Chamley (1986) and Judd (1985) and highlighting which assumptions end up leading to their landmark result. Perhaps a useful workthrough of the math can be seen in Moll's lecture notes, where it's relatively straightforward to get a positive optimal taxation rate, depending on the intertempoiral elasticity of substitution considered in utility.

Second, I'd point out that there are reasons to be skeptical, even inside either Chamley or Judd's frameworks, that housing interest payments should be considered as capital. Most prominently, both models (I believe) assume no financial constraints (they only require no-Ponzi scheme conditions). Therefore, any agent would be able to purchase a house "out of pocket" (or through costless single-period bonds where the only interest payment is the one imposed by discounting). This in and of itself suggests that mortgages (as we observe them in real life) shouldn't ever be used, since you never ought to pay for liquidity beyond the natural rate of interest. Since mortgages are effectively assumed away in the model, I'd be cautious of applying the model to the optimal taxation of mortgage interest in the first place.

This, I hope, is a bit instructive for the third point- I agree with your skepticism that housing should be viewed as a capital good in this context. Not only because it isn't clear in the general debate of whether housing is capital or not, but also because it doesn't seem to create the same distortionary effects that lead to the zero-capital taxation result even if we accept the Chamley-Judd result. Given that the interest payments aren't an infinite horizon stream of payments (you'd hope), the wedge wouldn't necessarily lead to the same "intertemporal snowballing" that we observe in both models (and that serves as the basis of their zero-capital tax rate finding to begin with).

If I think of any more, I'll be sure to come back and update this answer. And I hope that this just gets the ball rolling for others (which more specialized knowledge in either public economics or optimal taxation theory) to provide more insightful comments. And let me know if this answer was just totally non-responsive to the question you were really interested in to begin with- it's pretty early in the morning here, so I can't guarantee my mind is working at capacity yet!

  • $\begingroup$ Consumption is intertemporal in most macro models. $\endgroup$
    – nathanwww
    Jun 7, 2018 at 21:53
  • $\begingroup$ @nathanwww Right, hence why there's a wedge between current period consumption, and future period consumption. $\endgroup$
    – AndrewC
    Jun 8, 2018 at 14:02

Should housing be considered a form of capital for the purpose of capital gains taxation?

Chamley-Judd's proposition aims at avoid[ing] unlimited growth in tax compounding as the horizon extends.

Criticism of the Chamley-Judd model is centered on the "assumption regarding infinite lives". But I think the distinction between short-term and long-term favors the idea of treating housing as a form of capital. From the standpoint of temporariness, housing qualifies as capital in that it largely meets the characterizations of a fixed factor; whereas it lacks both the flexibility of a variable factor (like labor), and liquidity.

Moreover, the fact that a house is the worker's property (whereas, say, an industrial factory is the investor's property) does not automatically negate that the house may be equivalent to worker's fixed factor of production. Hence, taxation should not discriminate between uses of a property.

does the main result that capital income should be untaxed include income from housing appreciation?

I don't see why not. At first glance, that may seem counter-intuitive because we are used to the notion of depreciation and/or obsolescence of capital. However, appreciation of capital may occur independently of what use (business/industrial vs. residential) is made of that asset. Two scenarios where the market value of a facility or residential property could increase are:

(1) corporations/industries compete to establish their respective headquarters/factories in that facility's location; and

(2) a catastrophe that renders large, neighboring areas inoperative, too hazardous for continued operation, or for which replacement/restoration is too expensive.

In such instances, it is hard to justify why taxation should discriminate against some uses of a type of fixed assets.

  • $\begingroup$ While you give an interesting response, I'm not sure your analysis entirely fits with the Chamley-Judd findings, though. The issue isn't really "short vs long" run of the life of the object in question, it's the complementarity between current and future consumption. Given that housing appreciation is really only realized in lump sums (when the house is sold), any income derived from it would be lump sum, meaning lump sum taxation (by taxing the income generated by the sale) would be non-distortionary, and avoid the main finding of Chamley Judd. The justification for "why taxation should (cont $\endgroup$
    – AndrewC
    Jun 14, 2018 at 1:01
  • $\begingroup$ discriminate against some uses of an asset" is precisely because of the nature of how the income stream is realized. The only thing that might result would be to spend less on housing, but the central finding of Chamley and Judd is that taxing capital income creates a wedge on future streams of earnings, which leads to distortions due to the complementarity of all consumption at every period, derived from the infinite horizon of agent's lives. (end) $\endgroup$
    – AndrewC
    Jun 14, 2018 at 1:06
  • $\begingroup$ @AndrewC You are right. And I should have emphasized that my argument was not based on Chamley-Judd's model, but a digression on the temporariness distinction of housing as a long-term / fixed asset and other aspects. Specific to Chamley-Judd's framework, if policy-maker's intent is traceable to (or explicitly premised on) the rationale about future streams of earnings, then consistency dictates that capital gains taxation should substitute for lump-sum [income] taxation at least in those cases where profit from housing sales is reinvested (I don't know whether or not this implemented at all). $\endgroup$ Jun 14, 2018 at 11:35

The mortgage deduction is generally opposed because it's a subsidy to take on debt -- among other things, this may additionally inflate housing prices and thereby increase the size of fallout from a bubble burst. This can negatively impact systemic stability.

Housing can be considered as capital in the sense of producing a flow of returns. For example, it could cost $3000 a month to rent a similar house. "Capital" need not produce a manufactured output to be "capital", rather, it need only produce a flow of revenues (e.g., after correcting for depreciation and/or maintenance).

Also, I'm not sure that the question of whether physical capital should be taxed at low, moderate or high rates (as contrasted with the question of whether human capital embedded in labour should be taxed at low, moderate or high rates) should be considered as well-informed by the question of whether housing is fundamentally "capital" -- instead, housing would be a special case, which is not like "capital" which produces manufactured goods. Housing is "consumed" as the equivalent rental value in any period, but is also "capital" in the sense of producing a flow of benefits.

  • $\begingroup$ So is your answer than Chamley-Judd-type reasoning applies equally to housing as to manufactured capital? $\endgroup$
    – tparker
    Jun 7, 2018 at 22:36
  • $\begingroup$ Not sure, but I think not because the framing seems to assume that a tax on capital implies that the tax on capital is higher than on other things (taking an economywide 'non-distortionary tax' as a point of reference). However, capital gains taxes are generally lower than income taxes (suggesting the logic would apply the in the opposite direction in terms of which tax is relatively 'too high' or 'too low'). Also, taxes on capital gains are different from taxes on profits earned on revenues from sales of goods and services. $\endgroup$
    – nathanwww
    Jun 8, 2018 at 3:21
  • $\begingroup$ Landsburg argues that indeed "a tax on capital implies that the tax on capital is higher than on other things", regardless of how the marginal rates compare: "Even if capital gains taxes were capped at one percent, income subject to those taxes would be taxed at a higher rate than straight compensation. That’s because capital gains taxes (like all other taxes on capital income) are surtaxes, assessed over and above the tax on compensation." $\endgroup$
    – tparker
    Jun 8, 2018 at 20:21
  • $\begingroup$ The fact of whether one tax is calculated before or after the other does not determine whether the overall tax rate is higher or lower. However, it can lend itself to obfuscation about actual tax rates. So, for example, if tax B of 15% is calculated on top of a previously calculated tax A at 10%, tax B should be understood as a 16.5% tax and not a 15% tax. However, the fact that 16.5 > 15 does not imply that 16.5 > any other number. For example, 16.5 is not greater than 20, 30, 40 or 50. $\endgroup$
    – nathanwww
    Jun 9, 2018 at 16:27
  • $\begingroup$ For an illustration of how the logic you cite could lead to gross error in macro-informed policy, consider the following scenario: An individual considers two investments, investment A being 100k into a zero-dividend asset based on speculative anticipation, and investment B being 100k for human capital development. If the income flows from the first are taxed at 15% and the second at 40%, it can be asserted that 15 < 40, and by which logic the preferential treatment for capital gains would lead to suboptimal outcomes at the macro level. $\endgroup$
    – nathanwww
    Jun 9, 2018 at 16:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.