You wouldn’t know it from my writing, but I’ve been wading into real-estate mapping and assessment this year. For reasons I’ll discuss in future posts, I’m researching land-residual vs. building-residual assessment methodologies. Naturally, my initial Internet searches into building-residual assessments led to Georgist writings, and with good reason. Land-residual assessment, which subtracts land values from the total price, invariably leads to absurd results.
One source on the topic is Michael Hudson, whom long-time readers will recall as the economist who coined the phrase, “Debts that can’t be repaid won’t be.” Hudson gave a speech about the demerits of land-residual assessment in 2001, which he republished in 2010. He repeated a claim I first read in the superb collection The Losses of Nations: In 1993, the government valued nonfinancial corporate land at -$4 billion, and as a result, it stopped publishing economy-wide real-estate data.
1994 is the last year for which [the Fed] has estimated economy-wide land values. The problem was that the Fed discovered that its methodology produced nonsensical results – a negative value of $4 billion for all land owned by nonfinancial corporations for the year 1993. This number resulted from imputing land values by subtracting the estimated replacement cost of buildings from overall property market prices. This “land residual” method left little room for land value, for replacement values continue their rise even when overall market prices decline, as periodically occurs. In such downturns the replacement value absorbs nearly all the market value of corporately owned real estate.
It’s a damning accusation, but it’s also untrue. The Fed never stopped tracking nonfinancial corporate real-estate data. Perhaps it changed its source, but in the age of FRED, all this information is readily available. In fact, one can find three releases of the supplemental tables to the Fed’s Financial Accounts of the United States between 1997 and 1998 that indicate a residual nonfinancial corporate land value of -$4.6 billion, which appears to be what Hudson discovered. Starting with the June 1998 release, however, the residual land value for 1993 rose to $21.1 billion and ultimately to $25.5 billion when 1993 last appears (June 2000 release).
You can chalk Hudson’s mistake up to the migration of data to the Internet, or any other reason really. I don’t think it’s bad faith on his part, just bad luck. Obviously, though, there isn’t a clear conspiracy by Fed statisticians to cease reporting data that made it look bad.
Indeed, if anything, the reported data vindicate Hudson and make the financial accounts look worse: In 1996 and 2009-2010 the final residual nonfinancial corporate land value fell below zero—far lower than -$4.6 billion. More disturbingly, it’s skyrocketed since 2010.
See for yourself.
(Source: Federal Reserve (NCBEMVQ027S, RCSNNWMVBSNNCB, RCVSRNWMVBSNNCB), BEA fixed-investment price indexes (Table 1.1.9.), my calculations)
(I deflated residential and nonresidential structures by their respective BEA price indexes and then estimated the land value by keeping it in proportion to assessed fair-market real-estate values. It’s crude but “accurate,” I think.)
So at the beginning of 2010, the entire species missed out on the best real-estate deal ever: $566.83 billion (current dollars) to anyone willing to take all nonfinancial corporate land along with it. Just last week the Fed valued it at $4 trillion. And here you thought you missed out on speculating on Bitcoin.
So what should we make of this?
Are the Fed’s estimates merely imprecise or inaccurate? Imprecision just means that the land value is off by about a few trillion dollars one direction or the other. When the land value is negative, that’s just fuzzy math that ought to be improved.
By contrast, inaccuracy suggests government estimates of either the total real-estate values or the structures is systemically flawed. These properties might be worth far more than their assessed values. For instance although it’s a nonprofit, Brooklyn Law School’s 2 Pierrepont Street dormitory was ridiculously under-assessed at $3.88 million when the school sold it for $35 million. Alternatively, as I think Hudson tends to argue, the overall real-estate prices are correct but too much of their price increases are imputed to structures. He comments persuasively that “Building prices seem to be responsible for the rise in real estate prices, while land prices are held responsible for their decline.” The implication is that commercial land owners can depreciate land value.
I tend to think both inaccurate assessments of fair-market and structure values are at work, but the former is the bigger culprit. Regardless, Hudson is correct that encouraging local governments to adopt building-residual assessment methodologies would prevent absurd numbers from appearing in the financial accounts of the United States.
The Last Gen X American 
The value of a building is straightforward. It is replacement cost, less depreciation. Calculating replacement cost does require a degree of specific building-by-building analysis when historic preservation is involved in the cost of replication versus simple replacement to achieve functional similarity. Use of the income method to value a “property” that generates revenue gives one estimate of the total building+land value that is useful for determining the capital yields independent of imputed rent attached to the land parcel.