US Watch – The US productivity puzzle

by: Maritza Cabezas

152611-US-productivity-puzzle2.pdf (242 KB)
  • There is an ongoing debate of whether recent low productivity growth in the US is the new normal or whether it has been depressed temporarily.
  • We think that there is room for some improvement in productivity growth, but it is unlikely to reach the high levels seen before the crisis.
  • Given the slowdown in the growth of labour supply – partly  due to ageing – trend growth is likely to be around 2.5% going forward, compared to more than 3% pre-crisis.

US average GDP growth in the past decade amounted to around 1.5%, 2 percentage points lower than the average growth rate from 1994-2005. Slower labour productivity growth partly explains this slowdown. Productivity refers to output per hour of work. It is generally, decomposed into contributions from improvements in both the quality of labour (i.e. better skills) the availability of capital (i.e. investment in better equipment for workers), as well as technological innovation. In this note we assess whether the productivity slowdown will last and the implications for GDP growth in the coming years.

Impact of low productivity on growth

US productivity growth has been very low, at around 1.4%, after a period of fast productivity growth in 1994-2005 of 2.8%[1]. There are concerns that after the global recession, productivity growth has been so severely affected that potential growth  will remain low . The argument is that the impact of the slowdown in investment and the loss of labour skills will make it difficult for productivity to recover, while demographic factors not related to the crisis will also put downward pressure on labour supply in terms of hours worked and hence on future growth. For some researchers the rapid pace of expansion of productivity in the 1990s fueled by the ICT boom is unlikely to be restored. Some estimates of long run growth,  in fact the less pessimistic,  converge to 2% , while average long run growth was a bit higher than 3% in the 1980s and 1990s. Reduced prospects of long run growth in the US is often associated with lower equilibrium real interest rates, implying that the Fed will have to keep the federal funds rate low in the long run. Lower growth also makes it more difficult to reduce public and private debt ratios.


The debate about the slowdown in productivity

Those who forecast a longer lasting slowdown in labour productivity believe the IT revolution of the 1990s has run its course. Recent inventions are unable to match past technological breakthroughs and we are experiencing ‘technological exhaustion’. An ageing population, insufficient spending on education and inequality are creating even more headwinds (Gordon, 2014 and Cowen, 2012) [2].  Others are more optimistic and see the slowdown as only temporary. The impact of technological change comes with a lag. Knowledge spillovers from technological innovations are building and will become game-changers (Brynjolfsson and McAfee, 2014)[3]. Their optimism comes from the idea of exponential growth – coming from, for instance, digital technology. The amount of digital information that is being created and the amount of relative cheap devices that are increasingly becoming linked to each other and can store this information will have some ’wonderful properties’. From their perspective, it is only a matter of time for the combination of these innovations to show impressive results. On top of this, they argue that new technology is difficult to measure (see below). Some innovations cannot be priced in using traditional methods, resulting in an underestimation of productivity growth.

Productivity measurement issues

We think there is some validity to the argument that productivity is underestimated. Most productivity optimists argue that low productivity is partly explained by changes in the economy and quality improvements, which are not immediately captured in GDP measures. For instance, the value added of the service sector is much more difficult to capture than that of agriculture and manufacturing-related activities. The margin of error could be large given that the service sector has been growing at a fast pace in the past years and is much large than the industrial sector. Moreover, it can also be difficult to measure quality improvements. Many analysts point to Uber as a prime example of undercounted quality improvement, while the value of Google on improving productivity is also difficult to assess. These firms need to invest less in relative terms, but have an extraordinary market share. Similarly, recent technological changes are making it easier to work remotely and reduce start-up costs that are difficult to capture as productive activity. Statistical agencies try to measure these changes using various methods, which regularly leads to upward revisions in growth.

Alternative measures

One implication of the possible miscalculations is that national income statistics may be somewhat more reliable for reporting certain activities than GDP by expenditure. Since 2011, growth of Gross Domestic Income has been running 0.5 ppts above GDP growth. A study from the Council of Economic Advisors suggests that using a combination of GDP and Gross Domestic Income (GDI) is better than using one of the two. The BEA has started publishing Gross Domestic Output, which combines both measures. This measure has also been more consistent with profit margins, low inflation and high equity valuations in recent years. If the national income measure is now used, assuming a better indicator of GDP, productivity growth over recent years would be higher, at most 1.9%. Though that is still lower than the 2.8% seen pre-crisis.


Outlook for trend US GDP growth

We think that there is room for some improvement in productivity growth, but it is unlikely to reach the high levels seen before the crisis. That period looks to have been historically exceptional. There are signs that fixed investment growth is picking up. Indeed, as a consequence of the global financial crisis, financial conditions were more restricted, while a slow economic recovery and uncertainty resulted in firms increasing their savings (cash holdings).


In this sense there is some catching up to do. Investment growth will not likely increase to the rates of the 1990’s, but there is still room for a stronger recovery.  This supports our forecasts for GDP growth of around 2.5% in the long run. Overall, given the slowdown in the growth of labour supply – partly  due to ageing – trend growth is likely to be around 2.5% in the long run, compared to 2% in recent years and more than 3% pre-crisis.



[1] By definition GDP growth is equal to the sum of growth in productivity and growth in hours of work.

[2] Gordon R., The Demise of US Economic Growth: Restatement, Rebuttal, and Reflections, 2014, Cowen T. The Great Stagnation, 2011
[3] Brynjolfsson E and McAfee A, The Second Machine Age: Work, Progress and Prosperity in a Time of Brilliant Technologies, 2014