How Rental Income Gets Lost
in United States National Income Accounts

Michael Hudson, Ph.D.


Real estate held by the nonprofit sector

The Fed's treatment of the land values for non-profit organizations undervalues land by a steeper discount than occurs anywhere else in the accounts published so far -- only 10 percent of the estimated real estate value. This is because the replacement cost of buildings in this sector absorbs about 90 percent of the overall value.

This figure looks suspicious for a number of reasons. For one thing, tax assessments on real estate held by non-profits tend to be notoriously low. Secondly, their buildings tend to be specially designed in ways that make it particularly difficult to shift them to normal household residential use or to commercial use.

For this reason, land should represent a higher proportion of the true market value of structures held by non-profits than is found for corporately held real estate (60 percent, using a historical cost basis to estimate building values). Estimating land at 60 percent of the value of nonprofit real estate would indicate a value of $0.5 trillion for 1996. Estimating land at 75 percent would raise this figure to $0.6 trillion.

These two adjustments -- for corporate and non-profit real estate -- shift some $2.5 trillion in value from structures to the land.

The household sector's residential real estate

The largest category of land and real estate value, of course, consists of residential housing held by the household sector (that is, owner-occupants). Unfortunately -- and unlike the case with corporate real estate -- the FRB does not estimate homes at their historical cost, but only on a replacement cost basis. Using this "land residual" technique leaves today's household sector with a land value of just under $2 trillion, virtually unchanged from 1987 (that is, for nearly a decade). This means that land has fallen from 30 percent to just 20 percent of overall household real estate value.

Is this reasonable? During these years the overall value of household real estate has grown from under $5.5 trillion to over $8.1 trillion, a gain of $2.6 trillion. Even if we were to accept the "land residual" method as being valid during the many decades of soaring land and site values, holding the land value total steady at 30 percent would increase land values by $0.5 trillion, to a total $2.4 trillion.

One of the two really big "sense of proportion" questions to be asked, of course, is whether household residences should be valued at their replacement cost or at their historical cost. The problem with valuing them at their steadily rising replacement cost is that this absorbs all the appreciation in property values -- an appreciation that seems more realistically the result of rising land-site values, to the extent that it is "environmental" (referring in this case to the business environment and its asset-price inflation). In the late 1980s, household and corporate replacement costs of buildings left land with only about 28 percent of total property values. If we made the same adjustment for residential housing that the Fed made for corporate buildings in valuing them at their historical cost, this would raise residential land values to 60 percent of the total (the 1990 figure) or at least 45 percent (the 1996 figure).

Taking as a rough order of magnitude a 50/50 split between residential land and buildings would produce a land value of over $4 trillion, double the Fed's estimate.

This addition of some $2 trillion of residential land value seems justified on pragmatic grounds by the notorious undervaluation of land appraisals by local tax assessors. As these assessments are the basis on which Census statistics are compiled, one must accordingly use common sense in making the above adjustments.

The other "sense of proportion" question is whether one should use a "building residual" rather than a "land residual" appraisal method. This method likewise would add about $2 trillion to the rise in land values over the past decade, confirming the above estimate.

So far, these adjustments have added $4.5 trillion to the nation's land values. And this does not include the noncorporate real estate sector, that is, the sector to which most commercial real estate belongs.

As matters turn out, there also is another reason to suspect that the Fed may be undervaluing household real estate. The Bureau of Labor Statistics (BLS) conducts its own surveys of what homeowners believe the market value of their residences to be.

The FRB's estimates of residential real estate and land values in light of BLS statistics

An alternative government source of statistics on residential home values is published by the Bureau of Labor Statistics (BLS). The BLS asks its statistical sample of homeowners to estimate the market value of their homes. Expanding this sample to a nationwide scale, owner-occupants reported a value of $7.4 trillion for their homes in 1995.

In evaluating the quality and reasonableness of the BLS statistics, a number of caveats apply. Ego may indeed play a role in the respondants' estimates of the market value of their homes. On the other hand, this probably does not distort their reporting as much as do the under-appraisals that appear in tax assessment records. The political sway of homeowners when it comes to contesting appraisals of their property for tax purposes may be stronger than the egos reflected in what they tell the BLS.

The BLS estimate for homeowners' real estate value is closely in line with the FRB estimate of $7.6 trillion. Indeed, over the past decade the BLS and Fed figures closely reflect each other. Yet one would have expected that the Fed's statistics would be somewhat higher, for its estimates for the value of residential real estate owned by households cover more categories than are surveyed by the BLS. For instance, it is well known that the wealthiest 5 or 10 percent of the population own a disproportional value of its real estate and other assets. Recognizing this fact, the Fed oversamples these categories. It also includes household ownership of vacation housing, whereas the BLS asks homeowners only to estimate the market value of their primary residences.

The relative absence of the very richest families from the BLS numbers suggests that its estimates miss an important element of the economy's total household real estate value. Yet the FRB statistics do not provide a higher estimate. This suggests that homeowners probably estimate the market value of their residences at a higher level than local assessors (on whose reports the Fed relies), but that this probably more realistic estimate may be more than offset by the more narrow scope of BLS surveys.

A comparison of the year-to-year changes in BLS and Fed series on household residential real estate holdings suggests that the BLS reporters have a good idea of the market shifts occurring. In 1984, the year when the BLS series began, housing prices were just beginning to take off. The BLS nationwide estimate of $4.265 trillion was slightly higher than the Fed's $4.018 trillion estimate. But in 1985 the Fed's estimate pulled ahead, and has remained higher ever since. In 1994, owner- occupants surveyed by both the BLS and Fed estimated that the value of their homes rose by about $0.3 trillion, and by the same amount again in 1995. These estimates seem reasonable.

It is notorious that in times of rapidly raising prices, local real estate appraisals fail to keep up with the exploding market. The reason is largely political: homeowners would protest against the sharply rising taxes that would accompany realistic annual property reappraisals.

In sum, the estimates made by respondents to the BLS Consumer Expenditure survey regarding the market value of homes to their owners seem to be based on realistic calculations by the homeowners making them. At least their estimates stand closely in line with those made by the Federal Reserve System in its compilation of nationwide balance sheets. But one would have expected the FRB statistics in Table B.100 to have been substantially higher. If the 5 percent wealthiest American families own 20 percent of the real estate, this would have added about $1.5 trillion to BLS statistics that exclude these families. This suggests that the Fed's real estate estimates for the household sector should be higher than those of the BLS -- unless the homeowners surveyed by the BLS tend chronically to overvalue their homes.

This higher value in turn would be deemed to generate a proportional flow of rent.

In sum, a comparison of the year-to-year changes in BLS and Fed series on household residential real estate holdings suggests that the BLS reporters have a good idea of the market shifts occurring. In 1984, the year when the BLS series began, housing prices were just beginning to take off. The BLS nationwide estimate of $4.265 trillion was slightly higher than the Fed's $5.6 trillion estimate. But in 1985 the Fed's estimate pulled ahead, and has remained higher ever since.

It is notorious that in times of rapidly raising prices, local real estate appraisals fail to keep up with the exploding market. The reason is largely political: homeowners would protest against the sharply rising taxes that would accompany realistic annual property reappraisals. Yet in 1994, the BLS and Fed both estimated that the value of their homes rose by about $0.3 trillion, and by the same amount again in 1995. These estimates seem reasonable.

The question that inevitably arises is, how much should the rise in home values be attributed to rising land appreciation, and how much to building costs. The BLS does not ask its respondents to answer this question. The Fed answers it in an arbitrary way. After estimating the value of residential real estate (and other real estate) held by households, non-profit institutions, corporate and non-corporate business, farms and other categories of holders, it then estimates the replacement cost of real estate structures. This leaves a residual, which formerly was called "land" and which now is not even listed separately, evidently because the Fed recognizes that its methodology produces nonsense results.

The noncorporate real estate sector

The most recent statistics published for noncorporate real estate remain those for 1994. In that year, land ($551 billion) accounted for less than 25 percent of the sector's overall assessed real estate value -- down from 42 percent as recently as 1989. Here we see the FRB once again charging land with the entire price decline that is derived by subtracting the replacement costs of buildings from their market values (which are scarcely recognized to have budged in any year since 1989!).

Using the land residual technique of appraisal, in 1989 the Fed derived a land value of over $1,024 trillion. But by 1994 this derivative had been cut nearly in half, to $551 billion. Yet inexorably, the replacement cost of buildings was calculated as rising.

Certainly this derived figure did not represent market values. As the economy shrunk, demand for office space contracted. This suggests that a building residual technique would have been more relevant to use.

Simply holding land values steady would have added some $0.5 trillion to the FRB's land estimates. This is virtually what we would get by holding land constant at its 1989 proportion to overal real estate value. And taking a ratio of 60 percent -- the proportion indicated by valuing corporate real estate at its historical cost basis -- produces a land value of $1.3 trillion, some $0.8 trillion over the reported 1994 land figure.

Adding similar adjustments for farmland and for financial corporation increases the land value by some $1.0 trillion. This figure, taken in conjunction with the $4.5 trillion added for residential, non-profit and corporate real estate more than doubles the estimated land value, from $4.4 trillion in 1994 to nearly $9 trillion.

Taken together, these adjustments suggest that in imputing the proportion of real estate rental income as between land and buildings, land should be credited with 60 percent of the total.

Should structures be valued at their historical cost or their replacement cost?

While the cost of construction always is rising (by about 3 percent annually), these buildings also are depreciating in value. Private firms within the real estate industry make estimates both of cost increases and depreciation rates. And in summer of 1997, the NIPA statisticians made their own critique of the depreciation numbers that had been based on IRS filings.

To some extent (but not entirely), depreciation and obsolescence offsets rising replacement costs of buildings. But building values in reality rarely decline by more than 25 percent from physical wear and tear, unless the building's purpose becomes obsolescent (as in the case of many industrial structures, primarily those in the corporate nonfinancial sector).

Of course, real estate values rise and fall in keeping with overall property prices. But these shifts are not properly attributed to the buildings themselves. Being "environmental" (in this case, the financial environment as well as the physical and economic environment), they fall into the category of land value.

If the purpose of statistics is to depict the economy in terms of the motivations of its actors (investors and consumers, as well as bankers and other intermediaries), it is necessary to have better calculations of land values for the household, non- corporate and corporate sectors, and to juxtapose these land values to overall real estate prices and building values. In such an analysis, it is the building values that would be the residual item, not land values.

Conclusion: What is needed for a satisfactory estimate of land and building values

The above discussion has explained that the Fed's appraisal methodology is not the only way in which this estimate could have been made. In fact, it is not the most reasonable way. Structures could have been taken at their historical cost, for instance. A comparison with corporately owned real estate shows that there are good reasons for using this means of evaluation. Furthermore, it is more plausible to use a "building residual" method of appraisal than to use a "land residual." This is because the real estate landscape is continually changing, and few builders would choose to replicate existing buildings.

In addition to the problem of overall household real estate assessments, the Fed statisticians face the further problem of estimating the value of residential structures at their replacement cost rather than on a historical cost basis.The results produced for the corporate nonfinancial sector suggest that it would be more logical to subtract the historical cost of structures from the property's current market value.

But an even preferable method exists -- the building residual method of valuation. This would begin with an independent appraisal of land values, and then credit (or in some cases, debit) the structures with the residual value.

Most desirable of all would be to create a land-map of the U.S. economy, and attribute the rise and fall of real estate values to the land, using the buildings as the residual item rather than the land. Such a land-value map would cost money to initiate. But it is the best way to track what turns out to be the economy's most valuable asset: the land.

Such a calculation would go beyond the FRB's resources, to be sure. The Fed is dependent primarily on Census data for its real estate values, and the task of creating a land map of the United States (presumably by states) is properly the task of the Bureau of the Census.

The job certainly is possible. The government can incorporate this into the Census routine. It might use this data for federal grants.

My estimates suggest that such a study would revalue land relative to structures by $1.5 trillion for corporate real estate, another $2 trillion for owner-occupied residential real estate, and $1 trillion more for noncorporate real estate partnerships. This would double the value of land from the amount estimated by the FRB.

The final question to be asked is thus why this is not being done.

Political factors supporting an underestimation of the land

In the physical sciences, quantitative measurement is sacrosanct. From objective quantification emerge the patterns that provide the clues to how nature's physical and chemical relationships are organized. Emulating the natural sciences, economists express their theories in mathematical garb, but not many professors venturefrom their classrooms to deal in concrete statistics. Indeed, some important theoretical categories appear muddy when tested empirically. Specifically, the disappearance of land from the economy's balance sheet statistics cast doubt on how empirical modern economics can reasonably claim to be.

Nonsensical results indicate that the methodology being used is seriously flawed. The acronym GIGO ("garbage in, garbage out") is appropriate. But it is axiomatic that when a wrongheaded methodology is maintained despite its evident defects, it is wise to suspect special interests of being at work. Almost invariably there is a political motive behind what appears to be innocent statistical madness. In the case of land valuation, the FIRE sector has a self-interest in not tracking land gains more closely, and government agencies have acquiesced in this industry bias. This undervaluation reflects modern economic theory's general disregard for land's role, despite the fact that more wealth is generated in land appreciation than anywhere else in the economy remains the best-kept secret of statistical portrayals of the American economy.

The FRB's emphasis on rising building replacement costs attributes the rise in property values to something other than land. It also tends to justify over-depreciation, as owners can claim that they would have to pay more to replace their structures.

An understandable fear exists that any statistics that estimate a higher value for rental or land services will lead the government to tax these assessments at some point. (I understand that this is the reason why national income statisticians have stopped trying to estimate the economic value of housewives.) There is no political constituency that derives a more realistic assessment of the economy's real estate rent (except perhaps for manufacturing), nor even to track land values, whose increase has provided the basis for countless family fortunes and corporate balance sheets as seen by prospective stock buyers but not by the IRS and local fiscal authorities.

The neglect of land values seems all the more remarkable in view of the fact that most of the annual appreciation in real estate stems from rising land values. Yet such capital gains are virtually neglected in the statistics, although they are a primary objective of commercial real estate investment.

It will be noted that the value of a site's location, its zoning , municipal improvements and the economic development occurring in its vicinity (including transport accessibility) are not the direct result of capital expenditures by owners, yet the Fed's statistics depict the remarkable growth in U.S. real estate values (and, in times of economic downturn, their shrinkage) as being the result exclusively of the inexorably rising cost of replacing buildings.