3.2.1Mismeasurement

The mismeasurement hypothesis can be interpreted as the idea that gains in productivity were not visible because not measured properly by national agencies. Indeed, as stated in section 2.1.2, productivity is a major concept in economics, but is also a very difficult one to measure. This is particularly true in the “information economy” where inputs and outputs are more intangible than in the “industrial economy.” Brynjolfsson and Hitt (1998) interestingly noted that

‘While we have more raw data today on all sorts of inputs and outputs than ever before, productivity in the information economy has proven harder to measure than it ever was in the industrial economy.’

Table 3.1Classification of Theoretical Literature on the Productivity paradox by Hypothesized Explanation

Hypothesis	Authors
Mismeasurement	Griliches (1994)- Baily and Gordon (1988) - Ives (1994) - Brynjolfsson and Hitt (1993) - Brynjolfsson, Hitt (1998)
Long learning lags	David (1990) - Greenwood (1999) - Powell (2000) – Brynjolfsson and Hitt (1993) - Roach (1998)
Mismanagement	Chapman (1996) - Powell (2000) - Pentland (1989) – Roach (1998)
Complementarity	Bowen (1986) - Brynjolfsson & Hitt (1995) - Chapman (1996)
Redistribution	Jorgenson, Stiroh (1999) - Brynjolfsson and Hitt (1993) – Gordon (1999)
Small share	Oliner, Sichel (1994, 2000)

Ives (1994) reports the weaknesses of the economy-wide productivity data produced by the Bureau of Labor Statistics. Supporters of the paradox often relied upon these data. However, according to Ives (1994), U.S. government productivity data are not available for 58% of service industries, and are suspect in others. As a matter of fact, in education, health care, government and financial services productivity is often arbitrarily set to one (output to input) because of measurement difficulties.

Apart from the quality of the process measuring productivity is the questioning of the nature of productivity and its appropriateness to reflect gains from IT capital. Information technology deals with intangible materials such as knowledge and communication of knowledge. Thus, it might be difficult to measure the productivity benefits of intangible capital with tools made for older tangible inputs. In other words, the gains from IT might be represented as greater quality, convenience, reliability, timeliness, safety, flexibility, and variety, which is hard to measure. To illustrate this point, Ives noticed that a supplier’s order entry system is able to automatically replenish a retailer’s depleting shelves based on scanner data, but this quality improvement is not necessarily reported in productivity figures.

In the same fashion, Brynjolfsson and Hitt (1998) argued that productivity is becoming difficult to measure in the information economy, because of the change in the very nature of output and inputs. On one hand, output is becoming hard to measure because value depends today more and more on product quality, timeliness, customization, convenience, variety and other intangibles. On the other hand, as stated by Brynjolfsson and Hitt (1998), to measure properly inputs one should include

‘ (...) not only labor hours, but also the quantity and quality of capital equipment used, materials and other resources consumed, worker training and education, even the amount of organizational capital, such as supplier relationships cultivated and investments in new business processes’

Finally, new products such as software or ATMs have appeared with the booming of information technology. Until recently, these were not accounted for by national statistics agencies in their measurement of output, underestimating productivity growth. For instance, BEA started counting software as an output only in 1998, after considering it as an intermediate product for a long time. Another argument reinforcing the mismeasurement hypothesis is that roughly 80% of IT capital is located in service industries, where output is most difficult to measure because of its intangible aspect. Thus, new techniques for measuring productivity should be adopted, as stated by Brynjolfsson and Hitt (1993)

‘Just as some managers look beyond “productivity” for some of the benefits of IT, some researchers must be prepared to look beyond conventional productivity measurement techniques.’

Therefore, if benefits from IT capital are not measured properly, the return to IT capital would be understated, explaining in part the productivity paradox. Another argument deals with the long learning lags associated with the introduction of new technologies.