(FORTUNE Magazine) – If Hollywood gave an Oscar for economics, the person who figured out how to measure productivity more accurately would be a sure winner. It's a tough role -- no one really knows what the new economy is doing to productivity. How, for instance, do we measure the value that increased quality gives to the goods we produce? The productivity equation from the old industrial economy -- output divided by input -- doesn't take into account much of what has changed about output. Today, quality matters more than ever, with a more powerful car or computer representing more output than one that's less powerful. But measuring quality is a vexing problem because that measure should include especially hard to quantify factors that buyers value highly, factors like timeliness, variety, and customer service. With that difficulty in mind, MIT economist Erik Brynjolfsson measures productivity by letting customers be the judge of how much output they are getting. Accordingly, on the theory that customers buy more of what they want, he investigated what effect information technology investments had on sales gains and on return on capital at 400 FORTUNE 500 industrial and service firms from 1987 to 1991. ''Returns on info tech,'' he reports, ''were orders of magnitude larger than returns on other types of capital investments.'' (See chart.) Also, info tech accounted for half of all sales gains. But so hard to measure is info tech's contribution to productivity that some experts, particularly the authors of a recent National Research Council report on the topic, believe that the long years of sagging productivity may represent only a failure of metrics, not a failure of technology to have boosted productivity. Says Dartmouth business professor James Brian Quinn, chairman of the report's panel: ''Productivity figures understate significantly the benefits which have been achieved.'' It's not only that the national accounts data the BLS uses to calculate productivity cover only 42% of service-sector employment, leaving more than half the economy uncounted. Beyond that, pricing often skews the equation. If a product is greatly improved and its price drops -- phone service being one obvious example -- the productivity increase can be understated, because the price decrease may not fully reflect the improvements in the product. Nor do the data capture the alternative costs of all this investment -- the costs that would have been incurred had the investments not been made. What would be the cost in lost brokerage fees and higher capital costs for business, for instance, without the New York Stock Exchange's electronic trading system, which handles over 200 million shares a day, sometimes even 600 million, when in the pre-automation era a 12 million-share day could jam the system. According to these arguments, the recent numbers that show rising productivity are just as faulty as the past decades' disappointing numbers and just as likely to understate real productivity growth. Says Quinn: ''The cumulative effect of these investments seems to be picked up by the measures now, but we don't have the slightest idea whether performance improved by 5% last year or by 500%. We just don't know.'' The only thing we do know for sure, he says, is that ''total U.S. productivity is a heck of a lot higher than any of us has been willing to say.''

CHART: NOT AVAILABLE CREDIT: FORTUNE CHART/SOURCE: MIT SLOAN SCHOOL OF MANAGMENT CAPTION: RETURN ON CAPITAL INVESTMENT