Why don't the statistics show the boom? Unmeasured investment in the 1990s

There's a large discrepancy between the number of per capita hours people actually worked in the 1990s and the number of hours predicted by the official statistics and the standard growth model. In their most recent paper on the topic, Nobel Laureate and W. P. Carey professor of economics Edward Prescott and his colleague Ellen McGrattan find that the discrepancy between the actual data and the official statistics comes from the fact that the official statistics don't account for investments in intangible capital.

To say that there was a big economic boom in the 1990s wouldn't surprise many people. We know that Americans worked many more hours between 1992 and 1999 (actual per capita hours worked rose 8 percent during that period). And we know, intuitively, that GDP, productivity, and investment growth must have been bigger than reported.

The puzzle, for Nobel Laureate and W. P. Carey professor of economics Edward Prescott and his colleague Ellen McGrattan, a senior economist at the Federal Reserve Bank of Minneapolis (where Prescott also services as senior monetary policy advisor), is why don't the official economic measures of output paint the same picture of the 1990s.

Based on the official statistics from the national income and product accounts (NIPA) and a standard growth model, the number of hours worked per capita should have fallen significantly during the 1990s — given the behavior of GDP, productivity, and investment reported by the official statistics.

In their most recent paper on the topic, "Unmeasured Investment and the 1990s U.S. Hours Boom" (a Federal Reserve Bank of Minneapolis Staff Report last revised in June, 2006), Prescott and McGrattan find that the discrepancy between the actual data and the official statistics comes from the fact that the official statistics don't account for investments in intangible capital.

Intangible capital investments are those that can't be touched — or easily defined — including patents, trademarks, business know-how, reputation, and investments in building organizations.

The missing link: Unmeasured intangible investment

"Unlike tangible capital investment that can be and is directly measured, intangible capital investment cannot be directly measured and is not counted as part of the country's outputs in the national accounts," Prescott says. Prescott and McGrattan created an "extended" growth model that included an estimate of intangible investment. That extended model predicts a large overall increase in per capita hours worked — exactly what happened in the 1990s.

The authors separate intangible investment into two categories: expensed investment and sweat investment. Expensed investment, they write, "is expenditures financed by stockholders of the publicly traded corporations which, by national accounting rules, are expensed rather than capitalized. Examples of this type include research and development (R&D), advertising, and investments in building organizations."

Sweat investment, then, is "financed by worker-owners who allocate effort and time to their business and receive compensation at less than their market rate. This type of investment is made with the expectation of realizing future profits or capital gains when the business goes public or is sold," write Prescott and McGrattan.

Examples of sweat investment are the entrepreneurs working out of their garages, living on credit cards, waiting for their ideas to take off. In a period when sweat and expensed investments are large — as they were in the 1990s — their absence in the national accounts can dramatically affect economic measures such as GDP and productivity.

"If the relative importance of unmeasured investment stayed constant, then abstracting from it studying the behavior of hours worked in the market sector is reasonable," says Prescott. But that relative importance was not constant in the 1990s — "unmeasured intangible investment was abnormally large during the last two-thirds of the 1990s period."

Indeed, the value of net intangible investment as a share of total output increased from about 3 percent in 1990 to nearly 8 percent in 2001. "But that's not the full answer," Prescott says. The real question, he says, is: Why was intangible investment so high?

The 1990s boom in intangible investment

"Investment in intangible capital was so high in the 1990s because technological changes during that period were biased toward industries that produce large amounts of intangible capital," Prescott says. "The important finding from our research is that productivity boomed in the production of intangible capital and not in the production of the goods and services that are measured and included in official measures of GDP."

Prescott cites the information technology industry — from dot com companies to software manufacturers — as one of the intangible capital intensive industries in which productivity took off in the 1990s. "Rapid technological advancements were being made in industries that are relatively intensive in producing intangible capital, such as those related to information technology," the authors write. And R&D investment, which is the expensed type of intangible investment, was particularly highly productive in the 1990s.

Sweat investment was also high in that period. "There were a lot more owner/workers in the 1990s making sweat investments in companies that they hoped would take off. A lot of entrepreneurs were going by the 'make hay when the sun shines' motto — MBAs dropping out of school to start up tech companies while the market was still hot," Prescott says.

Back to trend in 2002

But after 2001, the value of intangible investment as a share of total output returned to a pre-1990s level. Prescott says that happened because the special opportunities for highly productive intangible investments that were present in the 1990s had disappeared. "There's some evidence that changes in the regulatory system discouraged intangible capital investments," says Prescott.

"And the environment has become so legalistic," he adds.

"Companies are afraid of getting sued if they make a mistake, so they don't try anything new anymore." Or maybe all of the opportunities were exploited in the 1990s. "Perhaps all of the gold nuggets had been found," Prescott suggests.

"And another reason the level of intangible investment has returned to trend is that some investments in the 1990s did not turn out to be as productive as people thought," Prescott said. "Sure, there were the Googles, Yahoos, and Amazons, but the dot com bust demonstrated that a lot of the expensed and sweat investments were not well-founded."

The consequences of mismeasuring intangible investments

"Because the standard accounting measures and models do not take account of intangible investment, they do not accurately reflect what was going on in the U.S. economy during this period," write Prescott and McGrattan in "Unmeasured Investment and the 1990s U.S. Hours Boom." According to Prescott and McGrattan's research, leaving intangible investment out of the equation in the 1990s led to the gross undermeasurement of GDP, productivity, and total investment.

At its height in 1999, GDP was about 3 percent less by national accounts measures than by Prescott and McGrattan's measure (which includes intangible investment). The national accounts undermeasured productivity by about 5 percent at its height and total investment by nearly 12 percent. But, people knew there was a big boom — even if the national accounts didn't reflect it.

"LA cab drivers knew there was a boom when their tips went from $1 to $20," Prescott quips. "And they knew the boom was over when their tips went back down to $1. The Bureau of Economic Analysis [the agency that reports the national accounts] recognizes that intangible investment is important and they want to be able to measure it," Prescott says.

"But they can only include intangible investment in the national accounts if there is a good way to measure it. They need a direct measure — versus the estimate we used in our paper — and that direct measure is just not available. In my view, the national accounts are far from ideal, but they're getting better," says Prescott.

As for Prescott and McGrattan, having solved the actual-versus-predicted hours puzzle, they're now focusing on the current account deficit — using their extended model to answer questions there, too. "In the end, our work is about advancing economic theory," Prescott says. "So that we can better understand what's going on in the major economies."

Bottom line

There's a large discrepancy between the number of per capita hours people actually worked in the 1990s and the number of hours predicted by the official statistics and the standard growth model. Prescott and McGrattan's extended growth model — which includes intangible investment where national accounts measures don't — follows closely the number of hours actually worked.

The absence of intangible investment from the national accounts causes discrepancies in measured and actual data (in this case, a gross undermeasurement of GDP, productivity, and total investment) when intangible investment is relatively large — as it was in the 1990s.

Intangible investment was relatively large in the 1990s because of technological changes that were concentrated in industries that produce large amounts of intangible capital — such as the information technology industry. After 2001, the value of intangible investment as a share of total output returned to a pre-1990s level because the special opportunities for highly productive intangible investments that were present in the 1990s had disappeared.