Transatlantic Technologies: The Role of ICT in the Evolution of U.S. and European Productivity Growth
- Date Posted:
- Is Database:
- Database
- Is Important:
- Important
US productivity gains btw 1995-2005 were largely attributed to ICT advancements, leading to a 17% annual productivity growth in its ICT-hardware industry.
Robert Gordon and Hassan Sayed, "Transatlantic Technologies: The Role of ICT in the Evolution of U.S. and European Productivity Growth," National Bureau Of Economic Research, June 2020, https://www.nber.org/papers/w27425
“…. Stepping back and viewing the postwar growth experience more broadly, the EU-10 started out in 1950-72 with rapid productivity growth of 4.86 percent per year, which we have previously interpreted as reflecting a process of catching up to the benefits of innovations that had buoyed U.S. productivity growth during the interwar and wartime periods…. Then the EU-10 transitioned to a slower productivity growth path of 2.31 percent in 1972-95, mimicking the 2.54 percent growth rate that the U.S. had previously achieved during 1950-72. Skipping over the very different experiences of 1995-2005, the EU-10 wound up in 2005-2015 with a productivity growth rate of a mere 0.63 percent, little different from the U.S. rate for the same interval of 0.87 percent. The remarkable similarity of the U.S. slowdown to 2005-15 from 1950-72 with the EU-10 slowdown to the same late interval from 1972-95 is, we think, more than a coincidence. The same process of the diminishing potency of ongoing innovation was occurring on both sides of the Atlantic. In short the postwar transatlantic productivity experience can be boiled down to three issues - the causes of the overall joint early-to-late slowdown, the sources of the temporary U.S. 1995-2005 acceleration, and the factors that held Europe back from enjoying a similar 1995-2005 revival. Our diagnosis of the first is the diminished impact of innovation over the postwar period that operated on both sides of the Atlantic, of the second is the U.S. success during 1995- 2005 in achieving a one-time boost in the level of efficiency in the production and use of ICT, and of the third is the multi-faceted failure of the EU-10 to mimic the U.S. achievement in producing ICT hardware, in making a similar level of investment in ICT, and in capturing the efficiency gains of the lower level of investment that actually occurred.….”
Looking at the two growth experiences overall Gordon argues they broadly mirror each other and support his thesis of the declining impact of innovation in contrast with previous waves of innovation:
“…Why did the EU-10 fail to benefit from the ICT revolution and instead why did it experience a two-step slowdown in labor productivity and TFP growth? The diagnosis has four components. First, despite all the U.S.-led innovation that drove its ICT-hardware EM industry to register a 17 per cent annual rate of productivity growth in 1995-2005, that same industry in the EU-10 actually experienced a productivity growth slowdown during the same decade. Second, the EU-10 had substantially lower values of the contribution-based ICT-use-intensity indicator variable, indicating less rapid growth of ICT investment. Third, the regressions reveal virtually no difference in the extent of the EU growth slowdown experienced by ICT-intensive versus non-intensive industries, indicating a failure of EU-10 industries to exploit the efficiency opportunities provided by the limited ICT investment that did occur. And fourth, the EU-10 shortfall in productivity growth during 1995-2005 can be traced to particular industries in which performance fell far short of the same industries in the U.S.These outlier industries include not just ICT-producing electric machinery but also agriculture, petroleum refining, and the large wholesale and retail sector where for many reasons EU nations lagged behind the U.S in adopting the big-box retail format which exploited the opportunities provided by the ICT revolution….”
They also examine why the EU10 failed to benefit from the diffusion of ICT:
We find that additionalproductivity growth in ICT intensive industries drove almost all of the post-1995 revival in U.S. productivity growth. However,this change in growth occurred almost entirely in services-producing industries rather than commodities-producing industries except for the electric machinery industry that produces computer hardware. This makes sense because the most intensive users of ICT were industries in the services sector. The EU-10 story is quite different. Productivity growth in producing computer hardware in the EU-10 actually slowed after 1995 in contrast to its explosive growth in the U.S. Further in the EU-10 there was little difference between the productivity growth slowdown after 1995 and after 2005 in industries that were intensive in ICT use versus the non-ICT industries. Europe not only invested less in ICT hardware but failed to reap its benefits even in industries that were heavy ICT users….”
He then looks the drivers of the decline in terms of sectors:
“…A retardation in the growth of labor productivity and of total factor productivity (TFP) has characterized both the United States and western Europe, in the sense that on both sides of the Atlantic growth has been slower since 2005 than it was before 1995. The notable difference in performance occurred in the middle interval of 1995-2005, when a sharp acceleration of growth in the U.S. contrasted to a growth slowdown in western Europe. As a result, the story of U.S. productivity growth since the mid-1970s has been one of slow-fast-slow over the three intervals divided at 1995 and 2005, in contrast to a two-step deceleration in Europe.…”
His research finds the decline in productivity growth was a three step process in the US (with an uptick btw 1995-2005) versus a steady decline in the EU10:
New Gordon argues that the ICT Revolution (Information-Communication Technology) is best understood as a temporary technological shock due to the lag in productivity growth after 1995. The uptick in productivity growth btw 1995-2005 was driven by the electronics manufacturing industry and ICT using service sector. Europe is a disaster, the E10 was unable to reap benefits (in terms of productivity growth) from ICT due to lack of response to ICT investment, as opposed to sheer absence of ICT investment.
Ed Comment 1/3:The thing he misses is what I identify in my chapter. High-skilled productivity seems to have grown faster than low-skilled and the US has a lot more low-skilled relative to Europe (more than the north, same as the south) so 1) our high-skilled seems to have grown faster and 2) our low-skilled productivity grew at least as fast but with substantially less supervision. What at the dual columns in table 5?
Steve Comment: Re: The dual columns“….Table 5….with results for the U.S. on the left and for the EU-10 on the right. We repeat the results for labor productivity growth in the total economy from Table 4 for ease of comparison with our new results explaining TFP growth. Thus, in Table 5, columns (A) and (B) repeat the U.S. results shown in Table 4 for the total economy, columns (A) and (D). Likewise columns (E) and (F) repeat the EU-10 results shown in Table 5, columns (E) and (H);these differ only in the choice of the ICT indicator….”
Ed Comment2/3 Ps I wonder how this effect the Gordon’s calculations on contributors productivity. This says intangible investment is mismeasured. Add this note to the Gordon entry with a link to this paper.
“…The first of these is the ICT “share indicator,” which is formulated as in Stiroh (2002). We compute the average ICT share of investment, which is the annual nominal expenditure on computing equipment, communications equipment and computer software and databases, all divided by the annual nominal expenditure on total capital investment.Initially we examine the actual values of this ratio for individual industries and subsequently in the regressions we convert the share indicator into a share dummy variable equal to unity for industries which are ranked above the median value in the 1991-95 period and zero otherwise. The other KLEMS ICT variable is the “contribution indicator,” equal to the contribution of ICT capital to real value-added growth. This is available at the industry level only beginning in 1999….”
Those are the “share indicator” and the “contribution indicator” defined earlier:
Ed Comment 3/3IT might be having a greater impact on productivity than Gordon acknowledges. It’s possible, for example, that as manufacturing productivity growth eliminates jobs, the only jobs left are low productivity jobs afflicted by Baumol’s cost disease. If low-skilled productive is declining (as high-skilled productivity is rising due to IT) IT’s contribution would be understated. It might also be the case that the capability of IT is growing but also growing more complex. With a shortage of properly trained talent, it may be the case that only the largest and most successful companies have the wherewithal to capitalize on the increased capability. That would explain why those companies are investing more in IT, growing their productivity faster, gaining share, and increasing their profitability. So it might be the case that productivity isn’t growing faster, not because IT isn’t delivering results, but for other reasons, namely constraints on the economy’s capacity to capitalize on it and therefore an increase need for economies of scale maximize the value of the constrained resources. Arguably, that can all be lumped in with “IT.” But lumping it might understating our ability to produce valuable innovation. Perhaps we’re still capable of producing it, but we are bumping into other constraints because of it