The Difference between Productivity and Innovation – and the Implications for Growth and Equity Markets



Official productivity measurements have been falling for decades – a puzzling development given the rapid pace of high-tech innovations. But rather than functioning as productivity-enhancers, many of these innovations disrupt economies and markets. This could have major implications for jobs and productivity, and trigger unforeseen amounts of social and political change.


Technological innovations are taking place at a rapid and accelerating pace. Apps on our mobile phones enable us to search and find information to an extent that was unimaginable only a few years ago. The digital footprints we leave on the internet allow companies to use vast amounts of data to provide tailor-made advertisements. Self-driving cars will probably become commonplace in the not-too-distant future, while self-learning machines are already taking over functions that were once carried out by highly trained specialists. And these are just a few examples. Innovations in the digital world, notably artificial intelligence (AI), are likely to have an even more profound effect on a range of industries in the future – energy, manufacturing, retail, finance, legal and even healthcare are all expected to change significantly.

In fact, the scale and scope of these technological changes have been described as a new industrial revolution – one that may be taking place right under our noses.

  • The First Industrial Revolution – from the late 18th to mid-19th centuries – was all about the introduction of mechanical production processes, steam power and railways.
  • The Second Industrial Revolution – from the late 19th century to the early 1930s – was characterised by major breakthroughs in the use of electricity, and by the invention of telephones, automobiles, the radio and plastics.
  • The Third Industrial Revolution – from around 1950 to the early 2000s – saw the development of mainframe computers, personal computers (PCs), mobile phones and the internet.
  • Since the beginning of this decade, some argue, we have been in the midst of a Fourth Industrial Revolution – one in which we can observe the fusing of the digital, physical and biological worlds.

One of the earliest champions of this theory was Klaus Schwab, the founder of the Davos World Economic Forum. He has described the changes presented by the Fourth Industrial Revolution as “so profound … there has never been a time of greater promise or potential peril”.


The promise and perils of the Fourth Industrial Revolution can be applied first-hand to the concept of productivity. In other words, while today’s technological innovations may lead to significantly higher productivity – that is, much higher output per unit of input, often measured as output per hour worked – many of these technologies could also disrupt existing business processes. This could have major implications on labour markets, and trigger unforeseen amounts of social and political change.

When we think about productivity in a traditional sense, we expect it to respond positively to the major technological innovations of the last few years. However, when analysing official productivity numbers, both in developed markets and in the emerging-market world, this does not seem to be the case. One is even tempted to repeat the famous Robert Solow quote from 1987: “You can see the computer age everywhere but in the productivity statistics”.

According to Robert Gordon, one of the godfathers of productivity research, US labour productivity was around 2 1/3 per cent per year from the late 19th century onwards – that is, until the early 1970s. Since then, labour productivity has fallen to around 1 1/3 per cent per year, with the exception of a few years around 2000. Data provided by the Conference Board confirm these conclusions (see Figure A). Labour-productivity growth has been trending down in the developed markets for many years already – a trend that started well before the global financial crisis of 2007/2008. Even in China, the biggest emerging economy in the world, labour-productivity growth peaked in 2006.


Arbeitsproduktivität, Veränderung gegenüber Vorjahr in %

Source: Allianz Global Investors Global Economics & Strategy, The Conference Board. Data as at 2017.

So how can we explain this perceived disconnect between technological innovation and low productivity numbers – and where has all the productivity gone? Unfortunately, as is often the case in the field of economics, there is no definite answer. Optimists would argue that we are still early in the multi-phase technological revolution cycle. During the current instalment phase – which is marked by the introduction of new innovations – aggregate productivity numbers still tend to be rather low. However, once we enter the deployment phase, new technologies are likely to be widely used. And once the labour force learns how to best use and apply these technologies, the argument goes, productivity is likely to pick up.

In other words, optimists believe that it is only a question of when – and not if – the economy-wide productivity numbers will ultimately rise. The sectors that are likely to contribute most to productivity growth should be the most intensive digital-using sectors, rather than the tech sectors themselves. Seen through this lens, the problems presented by the potential mismeasurement of productivity are of negligible relevance, as various studies have shown.

Robert Gordon1 and others take a much more pessimistic view of the long-term trend in productivity. To those in this camp, it is the lack of a “new general-purpose technology” that explains the structural decline we are seeing in productivity growth. Case in point: electricity, the major technological advance of the Second Industrial Revolution, is a technology sine qua non – one that drastically altered production processes across the board and had a massive impact on consumer behaviour. For evidence of how dependent we are on electricity, consider what happens during power outages: only a few hours after the lights go out, the world seems to come to a standstill. On the contrary, today’s technological innovations are potentially much less far-reaching – or so the argument goes. Moreover, Robert Gordon points to the overall declining trend of educational attainment – which is not only happening in the US – as an additional explanation for relatively weak productivity growth.

There is another argument – and a very different one – for why the productivity trend has been declining, and it relates to monetary policy. According to this view, easy monetary conditions from central banks have lasted for too long, depressing the costs of capital and contributing to the misallocation of resources. With an artificially reduced hurdle rate for investment, there is an increasing likelihood that investments that deliver little or no boost to productivity growth will be implemented. Overinvestment in real estate during the “noughties” – which directly followed the popping of the tech bubble in 2000 – provides a classic example. The low interest rates in effect at that time pushed capital into the real-estate market, which is not a sector that has traditionally contributed a significant amount to productivity growth. In this context, one could even make the argument that monetary policy in the US and Europe since the mid-1980s has been asymmetric: expansionary in times of actual or expected economic downturns and crisis, but not sufficiently restrictive in boom times. As a result, this asymmetric monetary policy has contributed to the secular decline in productivity growth rates. The long-term solution is for central banks to normalise monetary policy, rather than to maintain or extend their ultra-easy stance.

The jury is still out on which argument – the optimistic or pessimistic view – will answer the question of why productivity is falling, even as technological advancement is growing. Yet the facts are clear: the most recent wave of technological innovations has not yet led to a more efficient economy.

It also seems that today’s technological revolution is likely to exacerbate labour-market trends that have been in place since the 1980s, as a consequence of the move towards automation and globalisation:

  • In the middle-income segment of the population, there has been a decline in the demand for routine jobs such as clerking and manufacturing (seeFigure B).
  • In the high-wage bracket, the share of labour has been growing in all major developed economies, due to rising demand for highly skilled employees.
  • Interestingly, in the low-wage segment, the share of labour in most economies has been stable to rising – albeit accompanied by falling real wages.



Anteil von Routinejobs in den USA

Source: C. B. Frey, T. Berger, C.Che, Political Machinery: Automation Anxiety and the 2016 U. S. Presidential Election, 2017

This trend seems likely to continue as new technologies make routine and manual jobs increasingly obsolete, while making highly skilled jobs harder to fill. For example, consider the skills that are in high demand in a technology-driven work environment:

  • Strong IT and analytical skills (eg, data scientists)
  • Soft skills such as creativity and communication (eg, public relations)
  • Complex perception and refined manipulation tasks (eg, physicians)

Contrast these professions with, for example, the taxi or lorry drivers whose jobs might eventually be replaced by driverless vehicles. While there is a wide range of estimates on how many jobs will be affected by the introduction of new technologies – from less than 10 per cent to almost 50 per cent – the majority of recent studies seem to agree that highly skilled, highly paid labour will be in rising demand, and low-skilled labour carrying out routine jobs will see a marked decline. This shift in the labour share increases the risk of greater wage inequality – which we believe is a significant issue.

Since the 1980s, and particularly since the end of the global financial crisis, rising inequality has fuelled the rise of populist parties and politicians in Europe and the US. We believe that if these populist trends turn into policy, we could see an increasing number of negative economic and market implications – which we detail more extensively in our paper entitled “The Economics of Populism” (2017). The problem with this development is that the kind of economic nationalism that populists stand for results in less economic integration, less international trade and less international migration, all of which tend to hurt productivity growth.

Free trade, on the contrary, enhances productivity because it contributes to a better division of labour on an international level, and encourages the transfer of know-how. Migration also has productivity-boosting effects, in particular when it results in a foreign labour force with different skills being added to a local workforce.


Clearly, even if today’s technological advances were to generate higher productivity growth, their disruptive impact on the labour market could set in motion political dynamics that are – quite literally – counter-productive.

However, if technological innovations were to lift productivity, then potential economic growth would move higher, and inflation would move lower. At least in theory, equities should – in such an ideal world – benefit from high-tech innovations over the long run. Yet this theory fails to stand the empirical test.

Our research (see Figure C) demonstrates that periods of superior long-term equity returns occasionally, but not systematically, coincided with major technological innovations.2

In this example, superior long-term equity returns would have been produced on a few occasions if one were to have invested at the right time – for instance:

  • In the 1950s, when mainframe computers and nuclear energy were introduced.
  • In the mid-1970s, shortly after the development of the PC.
  • In the early 1990s, after the introduction of the world wide web and a new generation of mobile phones.

However, our research also shows that long-term equity returns were below average after several major technological innovations – for example:

  • The invention of the telephone (1876).
  • The roll-out of electricity in Western cities (1882).
  • The invention of autos (1886), radios (1920) and plastics (early 1930s).

In the same vein, the invention of the steam engine (1781), the introduction of railways (1825) and the advent of modern steel production (1850s) only opened up short windows of opportunity during which long-term equity investors could benefit. Those who came in a bit late experienced low, and sometimes even negative, real returns – similar to the late 1990s, when technology investors learned their lessons the hard way. Despite the evidence, it may still seem surprising that periods of major technological innovation don’t consistently coincide with superior long-term equity returns. So how can this be explained?

First, it is important to reiterate that innovation and productivity are not identical. Any new technological breakthrough can lead to higher productivity, but it might be accompanied by a substantial time lag. This was Robert Solow’s experience when he delivered his famous quote in the 1980s – just as personal computers were beginning to move into the mainstream.

Secondly, other, negative factors can more than offset the positive impact of a new technology on productivity. For example, it took until the beginning of the 20th century for the global economy to feel a sustainable rise in productivity from the major innovations of the late-19th-century Second Industrial Revolution. Several factors contributed to this lag during the period often called the “Long Depression” (1873-1896), which was characterised by relatively moderate economic growth: productivity and economic growth were dragged down following overinvestment in the 1860s and early 1870s; the economy experienced a build-up of private-sector debt; and in 1873, the United States switched to a gold standard, which implied a de facto tightening of monetary conditions. Moreover, World War I and World War II were major global events that held back productivity in the 1910s and 1940s, respectively.

Third, and most important from an investment perspective, valuations matter. The S&P 500 index’s cyclically adjusted price-toearnings (CAPE) ratio was particularly high during many previous periods of high-tech innovation:

  • At the beginning of the 20th century, which came after many years of major late-19th-century technological innovations.
  • At the end of the “Roaring Twenties”, which were accompanied by a dizzying array of technological advancements (especially radio).
  • During the “go-go years” of the mid-1960s, a period that lifted technology stocks in particular.
  • In the late 1990s and early 2000s, a time characterised by the blowing and bursting of the “dot-com” or “TMT” (technology, media and telecom) bubble.
  • In the mid 2000s, a period that culminated in the global financial crisis in 2007.

During all of these time frames, valuations were high because investors simply overpaid for the chance to enjoy future growth potential. Consequently, ensuing market returns were low – frequently (but not always) exacerbated by the appearance of major market shocks or financial crises soon thereafter, such as in 1901, 1929 and 2000.

It is important to note that today the S&P 500’s CAPE is again at a stretched level of around two times the long-term average – similar to the level seen in 1929 – which suggests that overall US equity returns in the coming decade are likely to be low compared to their long-term average. Some of the biggest drivers of the recent stockmarket boom have been high-tech innovators such as the FANG stocks: Facebook, Amazon, Netflix and Google. Moreover, our research highlighted several periods – such as the 1940s and 1980s – that were not characterised by major technological innovations, yet that still resulted in strong equity-market returns, as contrarian investors benefited from the low valuations seen at the time.



Arbeitsproduktivität, Veränderung gegenüber Vorjahr in %

Source3: Allianz Global Investors, Bloomberg, Wikipedia, Federal Reserve Bank of St. Louis. Òscar Jordà, Moritz Schularick, and Alan M. Taylor (2017). “Macrofinancial History and the New Business Cycle Facts.” NBER Macroeconomics Annual 2016, volume 31, edited by Martin Eichenbaum and Jonathan A. Parker. Chicago: University of Chicago Press. Macrohistory Lab, University of Bonn; S.Nairn (2002) “Engines that Move Markets”.
Legend: Year end data used, except for 2017 (21/9/2017). All other data as at 9/2017. Global returns: simple average of AUS, BEL, CAN, CHE, DEU, ESP, FIN, FRA, GBR, ITA, JPN, NLD, PRT, SWE, USA. Major innovations (approximation or year of first commercial use): steam engine (1775), railway (1812), modern steelmaking (1855), telephone (1876), electric illumination (1879), autos (1886), aircafts (1903), radio (1920), plastics (early 1930s), mainframe computer (1950s), nuclear energy (1954), PC (1974), digital mobile phone (1991), internet (1991), AI (~2010).
Financial crisis or major market downturns: UK, US 1796/97, US 1819, UK 1825, US 1837, UK 1847, US 1857, US 1873 and Long Depression 1873-1896, Paris 1882, Norway 1899, US 1901, US 1907, US 1929 and Great Depression, US 1937/38, Japan 1989, Nordics ~ 1990, Asia/ Russia 1997/98, World IT bubble 2000, global financial crisis 2007.

Even though technological innovations by themselves do not necessarily bode well for broad equity market returns, investors can still capitalise on innovation in a broader sense. Our research4 (see Figure D) demonstrates that “disrupting” sectors – those representing technological innovations – tended to outperform the broad market during the ensuing three, five, 10 and even 20-year periods after the technological innovation was made.

Clearly, this rather positive conclusion does not hold for every sector, nor for every sub-sector – and in fact our research uncovered a wide range of results. For instance:

  • After the development of the first commercial PC in the mid- 1970s, software stocks outperformed by a wide margin, while hardware producers did not.
  • Likewise, information-technology (IT) stocks performed well after the development of the world wide web and digital telephony in 1991, while telecom stocks only outperformed for a short time – notably during the years of the TMT bubble.

Moreover, the outperformance we uncovered in our research never occurred in a straight line. In particular, investors’ expectations about future earnings growth often became exaggerated within a few years of the debut of a particular innovation. Subsequently, returns suffered in “hot sectors” where lofty initial valuations were followed by stock-price setbacks.

Nevertheless, if the past is any guide to the future, sectors and companies in businesses related to certain high-tech innovations – particularly artificial intelligence – have a good chance to continue outperforming, on average, as long as valuations are not excessive. We are thinking in particular of stocks related to AI infrastructure (big data, the “internet of things” and cloud computing), AI applications (robotics and deep learning) and AI-enabled industries (the health care, transportation and automotive sectors are just a few).

It is true that with technology stocks in the US currently trading at an average P/E of around 30, valuations may no longer look cheap, but they don’t look overly stretched compared with historical levels. And today’s rich US equity valuations can primarily be explained by other sectors, notably staples and industrials.

There is a second market implication we want to highlight – namely the speed at which today’s changes are unfolding.



Source: AllianzGI, Fama, Cowles, Datastream. Data as at 12/ 2017
Legend: Chart shows median, average, minimum and maximum annual nominal excess returns of sectors representing major technological innovation relative to the broad US equity markets in 1,3, 5, 10, 20 years after the year of a major technological breakthrough. Overall, 13 technological innovations since the late 19th century were analysed. Data as at 12/2017.3

This is a critical factor for investors to consider. As Figure E shows, the innovation cycle is getting ever faster: it took around 100 years until railways were used by half of the countries worldwide, but less than 10 years for the internet to be adopted by a similar number of nations. The speed of these changes shows that investors have increasingly less time to assess the impact of innovations and to identify winning sectors, let alone the most attractive individual companies within those sectors. Joseph Schumpeter’s “creative force of destruction” – which these days is called “disruption” – seems to be growing ever stronger. This clearly strengthens the case for active investment management.

Only time will tell whether new technological innovations will help the global economy become structurally more productive. Nevertheless, innovation clearly matters for investors, and we advise them to be active in identifying the future winners, to not overpay for growth potential and, most importantly, to be patient.



Source: D. Comin and B. Hobijn (2010): “An Exploration of Technology Diffusion” (2010).
Legend: The y-axis shows the amount of time that it took in years for the new technology to be introduced in 50% of all countries.

Topic in focus:
Answers at a glance

Das Innovationstempo wird immer schneller: Ersteinsatz neuer Technologien

Adrian Jones Stefan Hofrichter,
Global Head of Economics and Strategy,
Allianz Global Investors

1) See Robert Gordon’s Rise and Fall of American Growth: The U.S. Standard of Living since the Civil War.
2) We looked at rolling ten-year real UK equity returns since the late 18th century and at rolling ten-year simple average returns of 15 global equity markets from 1870 onwards until 2017. We chose the UK because it was the world’s dominant economy and super-power until World War I.
3) Past performance is not a reliable indicator of future results.
4) We analysed the performance of US sectors representing major technological innovations since the late 19th century: telephone, electricity, autos, assembly line, planes, radio, plastics, TV, computers, PC, digital mobile telephony, the world wide web and artificial intelligence. Special thanks for running the numbers to my colleague Katharina Sänger.

Literature and data used:
Allianz Global Investors (2017): “The Economics of Populism”
D. Comin and B. Hobijn (2010): “An Exploration of Technology Diffusion”
The Conference Board (2014): “Prioritizing Productivity”
The Conference Board (2016): “Navigating the Digital Economy”
Robert Gordon (2012): “Is US Economic Growth Over? Faltering Innovation Confronts the Six Headwinds”
E. Brynjolfsson, A. McAfee (2016): The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies
C. B. Frey, T. Berger, C. Chen (2017): “Political Machinery: Automation Anxiety and the 2016 U.S. Presidential Election”
C. B. Frey & M. Osborne (2017): “The Future of Employment”
Òscar Jordà, Moritz Schularick, and Alan M. Taylor (2017): “Macrofinancial History and the New Business Cycle Facts”, published in NBER
Macroeconomics Annual 2016, volume 31, edited by Martin Eichenbaum and Jonathan A. Parker (Chicago: University of Chicago Press. Macrohistory Lab, University of Bonn)
S. Nairn (2002): “Engines that Move Markets”
The Conference Board productivity database
Cowles Data, Yale School of Management: Common stock indices
Fama/French data library

Investing involves risk. The statements contained herein may include statements of future expectations and other forward-looking statements that are based on management's current views and assumptions and involve known and unknown risks and uncertainties that could cause actual results, performance or events to differ materially from those expressed or implied in such statements. We assume no obligation to update any forward-looking statement. The value of an investment and the income from it may fall as well as rise and investors may not get back the full amount invested. There is no guarantee that the strategy will succeed and losses cannot be ruled out. Investors may not get back the full amount invested.

The volatility of fund unit prices may be increased or even strongly increased. Past performance is not a reliable indicator of future results. If the currency in which the past performance is displayed differs from the currency of the country in which the investor resides, then the investor should be aware that due to the exchange rate fluctuations the performance shown may be higher or lower if converted into the investor’s local currency.

This is for information only and not to be construed as a solicitation or an invitation to make an offer, to conclude a contract, or to buy or sell any securities. The products or securities described herein may not be available for sale in all jurisdictions or to certain categories of investors. This is for distribution only as permitted by applicable law and in particular not available to residents and/or nationals of the USA. The investment opportunities described herein do not take into account the specific investment objectives, financial situation, knowledge, experience or particular needs of any particular person and are not guaranteed. The views and opinions expressed herein, which are subject to change without notice, are those of the issuer and/or its affiliated companies at the time of publication. The data used is derived from various sources, and assumed to be correct and reliable, but it has not been independently verified; its accuracy or completeness is not guaranteed and no liability is assumed for any direct or consequential losses arising from its use, unless caused by gross negligence or willful misconduct. The conditions of any underlying offer or contract that may have been, or will be, made or concluded, shall prevail.

Contact the issuer electronically or via mail at the address indicated below for a free copy of the sales prospectus, the incorporation documents, the latest annual and semi-annual financial reports and the key investor information document in English. Please read these documents – which are solely binding – carefully before investing.

This is a marketing communication issued by Allianz Global Investors GmbH,, an investment company with limited liability, incorporated in Germany, with its registered office at Bockenheimer Landstrasse 42–44, 60323 Frankfurt/M, registered with the local court Frankfurt/M under HRB 9340, authorized by Bundesanstalt für Finanzdienstleistungsaufsicht ( This report does not satisfy all legal requirements on the guarantee of impartiality in investment recommendations and investment strategy recommendations and is not subject to any trade restrictions prior to the publication of such recommendations. The duplication, publication, or transmission of the contents, irrespective of the form, is not permitted.

Infrastructure Debt and ESG: the importance of strategic prioritisation



Infrastructure debt stands out due to its usually low risk profile and positive ESG impact. But for institutional investors, it can be worthwhile to invest in infrastructure projects with potentially higher ESG risks.