• "IT investments did not have an impact on productivity in 53 out of 59 economic sectors."

• "The relationship between IT and productivity improvement is murky."

• "Except in rare cases, IT did not produce dramatic increases in labor productivity."

Productivity is firm-specific, not sector-specific. McKinsey associates 99% of IT productivity growth with six "jumping" economic sectors (retail, wholesale, securities, telecommunications, semiconductors and computer manufacturing), then dismisses productivity gains in 53 other sectors. But actually, productivity gains from IT are company-specific and shouldn't be generalized to an economic sector. There are abysmal failures in the application of IT among retailers, for instance, yet there are also spectacular gains in the steel industry.

Even the favorable gains are smaller than claimed. McKinsey attributes 30% of the value of the economy to the jumping sectors, leaving the remaining 70% with only small productivity gains, offset by comparable losses. I found detailed sectoral data for 7,719 U.S. corporations with total 1999 profits of $417.3 billion. According to my data, the six jumping sectors accounted for only 16.4% of the economy. And the inclusion of three sectors is based on questionable measures. For instance, the securities sector's measure of output reflects rising stock prices based on investor psychology, not IT investments. If the questionable sectors are removed from the calculations, the jumping sectors account for only a very small share of the economy.

Reliance on federal statistics is questionable. The McKinsey conclusions are based entirely on contradictory government statistics regarding productivity gains, as evidenced by frequent after-the-fact revisions of published productivity data. Thus, McKinsey uses data similar to what Alan Greenspan and other economists have used to assert that IT was the engine that would assure remarkable further productivity gains [ Credit Greenspan, Not Computers, June 7, 1999]. Only corporate-level financial reports are sufficiently reliable and consistent for judging actual productivity gains. Government-issued statistics represent averages, whereas the effective uses of IT are highly concentrated in a small number of leading-edge firms. The report doesn't discriminate among individual patterns of success or failure to make the data useful.

Labor productivity isn't a measure of IT productivity. The McKinsey analysis uses an obsolete view of labor productivity: measured in labor hours per capita or in terms of people employed in production. Corporations measure IT's contributions based on how they affect profits. Management makes investment decisions based on cash flow that includes employee compensation as well as the cost of capital. Since salaries in sectors such as banking rose much faster than revenues during the past decade, head count-based ratios are meaningless. Also, a rising outsourcing of services renders the man-hour ratios unreliable.

Implications

If someone hits you with the McKinsey findings during budget deliberations, take the position that McKinsey's conclusions are irrelevant and can't be applied to your firm's conditions. Your budget proposals should be judged solely on how they contribute to your shareholder value.

You'll also have to produce credible numbers that show how your IT spending contributes to profits. You didn't need McKinsey to do that, but from now on, getting more money for IT will be hard. Accept the McKinsey study as a warning to prepare for coming budget battles.

Paul A. Strassmann (paul@strassmann.com) says that despite what he sees as flaws in the McKinsey report, he agrees with its broader conclusion that IT spending and productivity are unrelated.