The Skyscraper Curse: And How Austrian Economists Predicted Every Major Economic Crisis of the Last Century

The notion that a record-breaking skyscraper can cause economic crises sounds ridiculous, and it is very much absurd. There is no causal relations between skyscraper construction and the skyscraper curse.

The causality that does exist is between artificially low interest rates causing both record-breaking skyscrapers and economic crises. Artificially low rates also cause distortions throughout the economy. Very low rates over extended periods of time are what bring about the record skyscrapers and the economic crises. The skyscraper itself is merely an identifiable manifestation of what is happening throughout the economy.

There are distortions, also known as Cantillon effects, directly tied to the skyscraper, such as what happened with the new lightweight elevator cable. Resources had to be diverted to research and development of the new elevator cable from other investment possibilities. A new production facility and production process had to be designed to produce the cable in a profitable fashion. Distribution could probably take place with existing company facilities, but certainly the marketing aspect of the product would have to be built from scratch. When the economic crisis comes, all these resources could have very low value. What happened to the elevator cable company is taking place in all areas of the economy, although it is not universal.

This type of distortion is occurring throughout the economy as entrepreneurs succumb to the lure of artificially low interest rates and embark on investments in more roundabout and advanced production techniques. These investments will later be discovered to be malinvestments in what has been described as a cluster of entrepreneurial errors, a phrase first used by British economist Lionel Robbins1 in his description of the Great Depression of the 1930s.

Much of the interest in the Skyscraper Index can be linked to its ability to forecast the business cycle and to predict the business cycle. In my view, its primary and best use is not to be able to predict the future, but to be able to describe the types of real changes that occur in an economy exposed to artificially low interest rates. Those changes can then be linked to the troubles we experience in the economic crises that follow. Thus, it helps us to understand the business cycle. Unfortunately, many economists ignore the business cycle or do not believe in economic causes of the business cycle. I am afraid that if this situation is not rectified soon, Karl Marx might turn out to be right about business cycles. He argued that business cycles will intensify over time and bring about the demise of capitalism.

In what follows I review an editorial from The Economist2 that was published March 28, 2015, under the title “Towers of Babel.” The editorial was based on an academic article published by three Rutgers University economists. Unfortunately, the editorial staff of The Economist accepted the wrong, naïve understanding of cause and effect when it comes to skyscrapers. They did not refer to me by name in the editorial, but they did reference my 2005 article as a reference for what, in their minds, is the wrong point of view.

The editorial begins by noting that the world is in a major skyscraper boom and that such booms have often been an ominous signal of tough economic times ahead — the skyscraper curse. The Economist had long reported on and agreed with the Skyscraper Index,4 but they were no longer sure:

Does this frenzy of building augur badly for the world economy? Various academics and pundits, many of them cited by The Economist, have long argued as much, but new research casts doubt on it.3

They then explain the economics of skyscrapers, noting that taller buildings mean more potential revenues. However, they correctly note that the marginal costs of construction also increase with taller buildings. This part of the editorial is a great capsule summary of my 2005 paper, although the role of the interest rate is not introduced. They then mention Jason Barr’s 2010 article, which seems to provide some support for the Skyscraper Index.

Then they turn to the paper, “The Skyscraper Curse: Separating Myth from Reality.”5 Two sets of evidence from that paper are presented. The first set examined the question of why the biggest towers are built near the peak of the business cycle and whether that relationship could help you predict changes in Gross Domestic Product (GDP). They found that the time between announcement date of record setters and business cycle peaks is very long and that only half of the skyscraper opening dates occurred during a downward phase of the business cycle: “In other words, you cannot accurately forecast a recession or financial panic by looking at either the announcement date or the completion date of the world’s tallest building.”

The problem with this is that no one familiar with the Skyscraper Index would use the announcement dates and completion dates as a consistent forecasting tool. The World Trade Center towers were announced in the early 1960s, nearly a decade before the first tower opened. Also, many announced record-breaking buildings never get off the drawing board or off the ground, or are not built as planned. The better dating method for identifying the existence of a bubble and future trouble would be to look at groundbreaking ceremonies. Such ceremonies are an indication that plans have been approved, financing and permits have been obtained, land has been purchased, and any necessary testing has been started or completed.

When looking for signs of trouble — that is, the skyscraper curse — a better date would be when the project has actually beaten the old record, or is approaching that point. The Burj Khalifa tower broke the old record in the summer of 2007, when economic conditions seemed good, but it was two and half years before it was completed and opened to the public. As a word of caution, none of these dating processes are some kind of exact, precise, or magical process; they are just rules of thumb based on experience. However, by using announcement and completion dates, the Barr, Mizrach, and Mundra study exaggerated the amount of error that actually exists. Plus, as The Economist notes, it is based on a very small sample size of fourteen. The role of clusters or cycles of record-breaking skyscrapers should also not be ignored.

To rectify the small sample size, the study’s authors turned to a second set of data that includes the tallest building completed each year in four countries, which expanded the number of data points to 311. They compared data on tall but not necessarily record-breaking buildings to changes in local per capita GDP, not severe economic crises. As a result, they found that skyscraper construction and per capita GDP were cointegrated. When two time-series data sets are cointegrated it means that they move, in general, in the same direction and are thought to be the result of the same causal forces. For example, national income and national consumption will tend to move in the same direction, with small variations. You can think of a dog owner walking the dog on a leash as being cointegrated. The dog might be out front and then move behind the owner, but they are both following the same basic path. The fact that Barr, Mizrach, and Mundra found that skyscrapers and GDP are cointegrated means the two data sets move in the same general direction and implies, in other words, that skyscraper construction does not cause the business cycle, and that both statistics are caused by some other factor or factors.

One major problem with this data is that the Skyscraper Index is not based on general skyscraper construction, but instead on record-breaking skyscrapers. As we have seen before, because of the technology requirements and economic constraints, building two, hundred-story buildings is not the same thing as building one 200-story building. Another problem is that the skyscraper curse involves an economic crisis, not the ordinary ebbs and flows of the typical business cycle.

But let us ignore these fundamental problems with their evidence. The evidence that skyscraper construction and per capita GDP are cointegrated and move together with a common cause is exactly what is predicted by the Skyscraper Index! Both statistics move together and have the common cause of artificially low interest rates. The Skyscraper Index tells us that artificially low interest rates cause record-breaking skyscrapers, usually in clusters, as well as a bubble in the economy and eventually an economic crisis — the skyscraper curse. In other words, none of their evidence undermines the Skyscraper Index; it supports it.

Stunned by the editorial, I wrote The Economist a letter to the editor to try to clarify the meaning and status of the Skyscraper Index. That letter of March 30, 2015, is reprinted here verbatim:

Dear Editor of The Economist:

Thank you for discussing my research and referencing my journal article from the Quarterly Journal of Austrian Economics. (“Is there such a thing as a skyscraper curse?” March 28th) I would note that Mr. Barr, Bruce Mizrach and Kusum Mundra’s research actually supports my thesis that misaligned interest rates cause both record setting skyscrapers and economic crisis. The fact that skyscraper height and GDP are cointegrated is no surprise and actually supports the case for a skyscraper curse. Also, I claim no precision with respect to the exact timing of events, especially with respect to “announcements” and “completions.” Groundbreaking and record achieving dates are actually more relevant, yet are still imprecise. They say a picture is worth a 1000 words and I think you’re graphic of the timing of record setting skyscrapers and economic crisis says it all.

Mark Thornton, PhD.
Senior Fellow (economist)
Ludwig von Mises Institute
Auburn, AL 36830 (USA)

Unfortunately, they did not print my letter. I was contacted more than three months later and they explained that my letter had been misplaced.

Based on discussions with Lucas Engelhardt, we decided to go back to the original academic journal article and reexamine their findings. Based on that examination, we determined that a comment should be written on the article. Once a common practice, the comment is not nearly as common today, but it still exists at many academic economic journals, including Applied Economics, where the original Barr, Mizrach, and Mundra article was published. Originally, we thought the title of the comment should be “Skyscraper Height and the Business Cycle: Separating Data from Reality,” but we chose instead to go for the conventional approach. The comment is reproduced below.

Skyscraper Height and the Business Cycle: Separating Myth from Reality, a Comment

In a recent paper in this journal (Applied Economics), Jason Barr, Bruce Mizrach and Kusum Mundra test for the existence of a Skyscraper Curse, which Lawrence (1999) states is the “eerie correlation” between the building of record-breaking skyscrapers and economic crisis. Thornton (2005) shows the theoretical connections between record-breaking skyscrapers and economic crisis. However, the evidence that Barr et al. (2015) presents brings into doubt the existence of the Skyscraper Curse. Based on their evidence the Economist declared: “you cannot accurately forecast a recession or financial panic by looking at either the announcement or the completion dates of the world’s tallest building.”

Here we reexamine Barr et al. (2015) and come to a completely different conclusion. Their evidence does not refute the Skyscraper Curse and most of the more rigorous evidence actually supports it. With the Skyscraper Curse, output and height should be cointegrated and output should Granger cause height. Their evidence here is not only strong, but is more broadly applicable beyond the more narrow issue of record-breaking skyscrapers and once-in-a-lifetime economic crises.

Barr et al. (2015) use Granger causality and cointegration tests to analyze the relationship between skyscraper height and output. They use annual time series data for the tallest building completed each year and real per capita GDP for the United States, Canada, China, and Hong Kong as their measure for output. Their evidence shows that both height and output have a common trend indicating a cointegrated relationship. Granger causality tests show that output causes height, but height does not cause output.

The evidence from Granger causality and cointegration tests actually supports the theory of the Skyscraper Curse. No one believes that simply building a record-breaking skyscraper actually causes an economic crisis. The record-breaking skyscraper is more of an illustration of the types of microeconomic and technical changes to the overall structures of production that take place throughout the economy in response to artificially low interest rates.

Thornton (2005) clearly describes the Skyscraper Curse theory in terms of a third causal factor, artificially low interest rates, that cause both record-setting skyscrapers and unsustainable economic booms and, eventually, economic crisis. There have been several studies such as Barr (2012) that have suggested that such a third factor is responsible for the building of record-setting skyscrapers, such as builder competition, social status, and ego. However, in contrast to these psychological factors, artificially low interest rates provide an economic explanation for 1. Record-breaking skyscrapers, 2. The boom-bust cycle, and 3. Changes in social psychology. Therefore the results of the Granger causality and cointegration tests are completely in line with the expectations of Thornton’s (2005) model.

In contrast to their Granger causality and cointegration test results, the evidence in Table 1 of Barr et al. (2015) does strongly bring into question the existence of the Skyscraper Curse. They use the dates that record-setting skyscrapers were publically announced and the dates that those buildings were opened to the public and find little correlation with either of these dates and the business cycle.

However, there are multiple problems with their evidence. First, neither of these dates would be expected to be well correlated with the business cycle and especially with major economic crises, except in one sense. Announcement dates should generally occur during the boom phase of the cycle. They did find that 10 of 14 announcements did occur correctly in an expansion and 1 occurred at the very peak of the cycle. The 3 remainders occur because the “nearest US peak” is arbitrarily used, placing the 3 announcement dates after a previous peak. Additionally, using NBER peak and trough dates is not a true test of the Skyscraper Curse which is restricted to major economic crises.

Thornton (2014) suggests that announcement dates should be ignored and that instead, ground-breaking dates should be considered “skyscraper alerts” indicating that bubble-related investment opportunities exist, but danger is ahead. Furthermore, the date of record-completion, in the sense that the record-breaking height has been achieved, is a “skyscraper signal” suggesting that economic danger is imminent. Opening dates may be many months or even years in the future from record-completion dates and record-breakers often open in the midst of an economic crisis.

Barr et al. (2015) also downplay the Skyscraper Curse by noting that “the range of months between the announcement and peak is tremendous, varying from 0 to 45 months.” However, this variation is the result of using the announcement date so that, for example, the World Trade Towers was announced in January 1964, but the ground-breaking date for construction began in August 1968. They also use US cycle dating for two foreign records, Petronas Towers and Taipei 101. These record-breakers are normally connected to the Asian Financial Crisis of 1997–98 and also to the Tech Bubble-Bust (1997–2001), but these events bear little relationship with either the announcement or opening dates.

There remain several anomalies in Table 1. The Woolworth Building was announced in July of 1910 and opened in April 1913, but there is no economic crisis of note connected to it. However, the economy did peak and began contracting in the first quarter of 1913 and continued to contract until the fourth quarter of 1914. This contraction included the third worst quarterly decline in real GNP between 1875 and 1918, and was worse than any quarterly performance between 1946 and 1983. The founding of the Federal Reserve System in 1913 and the coming of World War I in Europe in 1914 provided stabilization for the American economy as exports to Europe soared. These 2 exogenous factors prevented the Woolworth Building from being associated with the Skyscraper Curse because the intervention of World War I reversed the deepening economic slump and prevented a historical label (e.g., “Depression of 1913–15”) from being created.

Table 1 also lists the Pulitzer (1890) and Manhattan Life (1894) buildings, which along with the Masonic Temple in Chicago (1892) and Auditorium Building (1889) represent a wave of record-breaking skyscrapers that preceded the beginning of the largest contraction in US history, culminating in the largest quarterly decline in real GNP in US history, which was then followed by the Panic of 1893 and 6 years of double-digit unemployment.

With these clarifications, 13 of the 14 buildings listed by Barr et al. (2015) come into agreement with the Skyscraper Curse model. The Park Row Building, which was announced in 1896 and opened in 1899 does not seem to fit the model, but is in synch with the emergence of the then new steel frame construction technology. If you take skyscraper waves and historical context into account, record-breaking skyscrapers are indeed associated with major economic crises and the Skyscraper Curse does add to our ability to foresee macroeconomic risks, even if the complexities of history prevent predictions of timing from being precise. We agree with Barr et al. (2015) and the Economist that the Skyscraper Index and its Curse is of little value in forecasting the normal ebb and flow of the macroeconomy.

We were quite surprised to learn many weeks later that our comment had been rejected by Applied Economics. The editor sent us two referee reports. Neither of the reports dealt directly with our primary comment, and both were defensive of the Barr, Mizrach, and Mundra paper. We noticed that in one of the reports, the referee identifies himself as one of the authors of the Barr, Mizrach, and Mundra paper, writing, “It is hard to reject a comment that agrees with your paper.” However, he managed to fight that urge and did reject our comment. It is not unheard of to send an author of an article a comment on their paper to referee, but it does seem odd to give them veto rights without the editor having read the paper and comment, which seems obvious in this case.

It is no embarrassment for a journal to publish a flawed paper. It happens on a regular basis. It is part of the academic process. For example, new econometric techniques have brought into question many early empirical papers. Hundreds of papers have been written on the Phillips Curve, and no doubt many are mistaken and now irrelevant. In the case of Barr, Mizrach, and Mundra, their paper is actually not wrong per se; they just came to the wrong conclusions based on their evidence. Even their secondary evidence could be salvageable. This experience provides a clear window into the messy world of academic publishing.

We expanded the comment into a paper with additional empirical evidence, and this paper was accepted for publication at the Quarterly Journal of Austrian Economics. There are other important developing lines of research on the Skyscraper Index, two of which I will report on next. One study looks at the Skyscraper Curse at the state level, the other examines the microeconomics of the Curse at the city level and helps explain the old real estate adage that what matters is “location, location, location.”

• 1Lionel Robbins, The Great Depression (London: Macmillan, 1934).
• 2“Towers of Babel: Is There Such a Thing as the Skyscraper Curse?” March 28, 2015.
• 4Jason Barr, “Skyscrapers and the Skyline: Manhattan, 1865–2004,” Real Estate Economics 38, no. 3 (2010): 567–97.
• 3Ibid.
• 5Jason Barr, Bruce Mizrach, and Kusam Mundra, “Skyscraper Height and the Business Cycle: Separating Myth from Reality,” Applied Economics 47, no. 2 (January 2015): 148–60.