On the New Productivity Paradox

Well, the latest round of revised productivity numbers certainly puts the productivity question firmly back on the agenda, and the surprise jump in job losses in August only adds to the connundrum. Interestingly Stephen Roach has a strange take on this today: US workers are both working more and less hours. The are implicity working more because 24/7 connectivity means you now carry your work round with you (does Edward never sleep?), but they are working less since repeated crashes and outages lower the available working time:

I just blew another productivity opportunity. I spent the first 45 minutes of my workday fighting the mindless edicts of the technology gods. I always remind myself not to take it personally. After all, I am only one of America’s some 83 million IT-dependent knowledge workers. I guess that’s just the point: To the extent others feel the same pain, there may well be good reason to raise serious questions about America’s widely celebrated white-collar productivity miracle................

According to the BLS, the average workweek in the financial activities sector was 35.4 hours in July 2003 - essentially unchanged from the level a decade earlier (35.6 hours in July 1993). I find that most difficult to fathom. Over the past decade, the IT-enabled knowledge worker has seen a radical transformation of work schedules. Courtesy of miniaturized and portable information appliances, together with near-ubiquitous connectivity, workdays have been extended as never before. Yet in this increasingly "24 x 7" mindset, the official data speak of unbelievably short and unchanged work weeks. What a disconnect! To me, this smacks of a classic measurement problem.

I fully realize all this is heresy. But I fear that in our rush to rationalize the wisdom of our ways, we have lost sight of the pitfalls of the Information Age.

Now come on Steven which is it? Because clearly it can't be both, or at least if it is both, then these two factors probably balance out and the argument about measurement problems has less force. The truth is, I suspect, Steven's card-carrying 'skepticism'. As he says it is heresy, he is an information age non-believer. Brad - against whom one could never ever hold this - had an interesting post recently about Robert Gordon - about whom one certainly could - and an article of his in the FT on productivity. Browsing over to Gordon's homepage I discovered that the article is based on a longer piece of research that he has recently published as an NBER working paper. Since the productivity phenomenon is really one of the most striking things about the recent US performance, and since this paper is by and large non-technical (ie I think it could in the main be read by the average economically literate blog enthusiast) it may be worth looking quite closely at the argument and the conclusions. Two other papers of interest are: Yang and Brynjolffsson, "Intangible Assets and Growth Accounting: Evidence from Computer Investments", which is cited extensively in Gordon's paper but is fairly technical ( here ), and Bart Van Ark's "Is there a productivity puzzle?" which is a comparison of the EU and the US and is pretty readable since it is in the form of a presentation ( here ). Incidentally Gordon, who previously attracted some notoriety - and in a sense became seen as a key representative of the new economy cynics - by asking whether the internet and computational revolution measured up to the big technological revolutions of the 19th century, appears to be answering his own question: "In short, the 'marriage' of computer hardware with software and communications hardware in the 1990s was as important to the development of the computer as was the development of paved roads and then superhighways to the full exploitation of the internal combustion engine." However having ventured so boldly forth he does try to redeem his former self at the end, by pointing out that he considers the ICT-internet fushion a 'first rate' but not a 'mega' invention since it has, apparently, a one time only component.

In the paper Gordon presents five 'puzzles'. Here I will focus only on two: the problem of the cyclical component in productivity, and the question of the post-2000 acceleration. His conclusion on the first point is - following Yang and Brynjolffsson - that little of the 1995 - 2000 improvement was cyclical, since it is more than likely the result of a massive long term investment in 'intangible human capital' which was being measured as a cost earlier, but which was not showing up as a benefit. Post 1995 this 'asset' has been increasingly revealing its worth, and, as the figures show, it continues to do so. If we go back to Roach's skepticism for a moment, it is worth noting that according to the BLS the US services sector lost 67,000 jobs in August alone. Yet sevices output continues to grow significantly. The question really is - and this is a complete unknown - how much more of this is there still in the pipeline? Because if there is a lot, and I suspect that there may be, and if we continue to have 'lukewarm growth', then this is going to have serious implications for the deflation problem in the US. Put simply, this rise in intangible human capital is the result of learning by doing. This learning by doing has two components: you get to do the things your were doing better, and you find new things to do. It is a dynamic process, and this brings us to Gordon's conclusion on the second puzzle. His conclusion is that the 1995 Netscape-ICT 'alchemical wedding' had a one-time-only component, hence we may be near the end of the ride. His examples for this are trivial, as incidentally were his skeptical comments about word-processor improvements in his earlier 'does the internet measure up' piece. For an example of one of the potential new productivity enhancing innovations on the horizon you could take something like grid computing (see my earlier post here ). Many many more possibilities could be mentioned, but that's what they are, possibilities. The whole point about this learning-by-doing creative-machine which we've set in motion is that we won't know till we get there. However what we could notice is that imminent-intimate fushion of Robert Gordon (who has an excellent 'productivity analysing' mind) and Ray Kurzweil (whose imagination and insight as a practising technologist are invaluable) to create a new breed of 'spiritual economists' might be just what we're in need of right now. Trouble is I'm not sure whether we should load Robert's mind into Ray's computer, or up-load Rays hard disk contents into Robert's body.

(Puzzle 1) Whatever happened to the cyclical effect? Skeptics were justified on the basis of data through the end of 1999 in their claim that part of the post-1995 productivity growth revival reflected the normal cyclical correlation between productivity and output growth. In contrast data through mid-2003 reveal only a negligible cyclical effect for 1995- 99 but rather a temporary "bubble" in 2002 that repeats similar temporary blips in three previous business cycles.


(Puzzle 2) Why did productivity growth accelerate after 2000 when the ICT investment boom was collapsing? The most persuasive argument points to "hidden" intangible investments in the late 1990s that required labor input but were not counted in measured output; after 2000 the delayed benefits of intangible investments boosted output, while much of the labor input that created them was laid off. In short, productivity growth was understated in the late 1990s but overstated since then.

Our verdict on Puzzle 1 is that little of the initial 1995-2000 productivity growth revival was cyclical, and almost all of it it represented a fundamental shift in the trend. However, the 2001-2002 "bubble" contains a temporary element that has occurred in the early stages of the recovery in previous business cycles and deserves to be called "cyclical" because of its periodic repetition. The 2001-2003 economic recovery has been substantially more "jobless" than the 1991-92 recovery that gave rise to that label, and productivity in the first few quarters after the four quarter "bubble" period has been considerably healthier than in 1993-94 relative to a much more robust trend. The U. S. economy is on track to achieving a rate of productivity growth over the decade 1995-2005 of almost three percent per year, raising deep questions about why this has occurred and why these causes have not been equally relevant in Europe.


The period 2000-2003 has been marked by a sharp downturn in ICT investment, particularly in computer hardware but also in software and communications equipment, and a rapid decline in employment. Output can grow despite a continuing decline in labor hours, because the benefits of the previous hidden investment in improved business processes and better trained employees are transmitted to production, while the workers that produced the hidden output in the late 1990s (programmers, consultants, trainers) have been laid off and are walking the streets. In a sense the U. S. economy of 2000-2003 has been getting a "free ride" from the 1995-2000 wave of investment in hidden capital.


At least one obvious question is raised...........and this is whyintangible capital did not produce a productivity growth upsurge during previous periods when the share of spending on computer hardware was growing rapidly, particularly 1972-87, the interval that led Robert Solow to utter his famous quip that later became known as the Solow "computer paradox," that "we see the computer age everywhere except in the productivity statistics". One possible answer is that the 1972-87 increase in the share of computer spending in GDP was slow and gradual, while the post-1995 upsurge was sudden and hence created a greater imbalance between measured and unmeasured ICT investment. A second possibility is that the nature of ICT innovation in the 1990s was more disruptive and required a more substantial investment in intangible capital than did earlier waves of computer innovation.


Just as complementary investments in roads and suburbs were necessary to provide the full benefits of motorcars and motor transport, so complementary innovations in software and communication technology were necessary to provide the full potential benefits of the personal computer. Windows 95 and 98 provided an intuitive interface that instantly replaced DOS, with its command lines and DOS-based programs with their arcane codes. While the replacement of DOS programs with Windows-based programs may have been little more than an annoyance for experienced DOS users in the business world, they made it possible for business firms to reduce training expenses, and also for the personal computer to penetrate the household.

But the "killer application" that powered the post-1995 productivity revival was the marriage of computer hardware and Windows-type software to communications technology that made possible the WWW. Equally important were developments in hardware power and software that made it trivial to send documents as e-mail attachments, thus eliminating the need to print out many preliminary documents and spreadsheets and to send the printed versions via fax or courier service. Cheap communications caused a revolution in business practice...........In short, the "marriage" of computer hardware with software and communications hardware in the 1990s was as important to the development of the computer as was the development of paved roads and then superhighways to the full exploitation of the internal combustion engine.


One way of assessing the likely productivity impact of near-term ICT innovation is to ask whether such innovations can break through the inherent impediments to the replacement of human beings by computers.


The further acceleration of productivity growth in 2000-03 has laid the cyclical argument to rest insofar as it applies to the 1995-99 period. But another cyclical phenomenon has emerged more recently, the "early-recovery productivity bubble" that pushed up productivity growth in 2002 to incredible levels. This phenomenon is cyclical in the sense that it is periodic; similar "bubbles" occurred in 1975-76, 1982-83, and 1991-92, and in each case were followed by two or more years of productivity growth below trend. Data on productivity growth rates during 2000-03 are pushed up by the bubble phenomenon, as are estimated Hodrick-Prescott productivity trends that respond relatively rapidly to the evolution of the actual data.

Puzzle 2 was suggested by the paradox that productivity growth accelerated after the year 2000 despite the collapse in the ICT investment boom, suggesting that standard studies of growth accounting may exaggerate the causal role of ICT in achieving the first 1995-99 phase of the productivity revival. Three factors support the case for exaggeration. First, the growth accounting methodology unrealistically assumes that the full productivity benefits of ICT hardware and software are achieved at the instant of production, with no allowance for reorganization or training effects. Second, independent evidence for the retail trade sector finds that all of the rapid productivity growth in the 1990s was achieved by new establishments and none by old establishments, even though ICT investment has been made in both. Third, and most important, the boom in measured ICT investment in the late 1990s was accompanied by a boom of perhaps equal size in unmeasured or "hidden" improvements in intangible and human capital, as suggested by Yang and Brynjolfsson. The numerator of productivity omitted the creation of the intangible capital but the denominator included the labor input, artificially holding down the magnitude of the productivity growth revival. Then after 2000 productivity growth was exaggerated, because output was supported by intangible capital input that had been created before 2000, but the labor input that had created the intangible capital had declined, as programmers, consultants, and trainers were laid off. The cyclical analysis of the 2002 productivity growth "bubble" and the intangible capital argument both suggest that observed productivity growth in 2002-2003 may represent a high water mark and cannot be expected to continue.

That major source of productivity growth, capital-deepening investment, cannot occur forever without a continuous flow of innovations, and so the post-2000 crash in ICT investment raises the question as to whether the wave of innovation in the 1990s had a one-time-only component, and whether a new wave of innovations will emerge over the next few years to create a repetition of the investment boom. We classify innovations as "mega," "first-rate", "second-rate," and beyond and argue that the marriage of computer and communications hardware with software in the 1990s was a first-rate invention, but that it had a one-time-only component because the web could only be invented once, because part of the boom consisted of demand from dot.com firms which promptly went bust, because of the mismatch between hardware and software innovation, because of the timing of Y2K, and because of the overbuilding of telecom infrastructure. The main areas of ICT investment in the near future are innovations that look distinctly second-rate, the further move toward mobility with internet-enabled mobile phones and wi-fi enabled laptops that will allow e-mail, web access, Word, Excel, and Powerpoint to be accessed more conveniently, but the functions to be accessed will be the same as five years ago.