The Future of Innovation

Robert J. Gordon, Northwestern University and NBERConference on The Future of U.S.

Economic GrowthCATO Institute, December 4, 2014

Today’s Question: Is the Pace of Innovation Speeding Up?

• The Second Machine Age, p. 44, “we’re at an inflection point” leading to faster growth

• But there’s no hint:– Faster growth in what measure of the aggregate

economy? – Compared to what? Last 10 years? Last 50?

• Their argument proceeds by example: exponential growth in big data, small robots, 3-D printing, driverless cars

My Position in a Nutshell

• Blog: “Taking Robert Solow Seriously”• Solow 1957: The history of TFP growth is the best

guide to the importance of innovation– Fast and slow TFP growth provides a guide to the

importance of different innovations• The most important innovations happened long ago• Over the past two decades the digital revolution caused

TFP growth to spike, but not for long• Flagging TFP growth suggests the most important

contributions of the digital revolution have already happened

To Understand TFP History, We NeedDefinitions of the Three IR’s

• The 1st IR occurred 1770-1840, continued impact through 1900• Steam engine, railroad, steamships, wood=>steel

– The 2nd IR occurred 1870-1920, continued impact through 1970 along at least 5 dimensions• Electricity, light, elevators, machines, air conditioning• Internal combustion engine, vehicles, air transport• EICT: Telephone, phonograph, movies, radio, TV• Chemicals, plastics, antibiotics, modern medicine• Utter change in working conditions, job & home

Third Industrial Revolution

• Since 1960 the “EICT” Revolution– Entertainment: TV – color, cable, time shifting, HDTV,

streaming– Information Tech – mainframes, minis, PCs, web

browsers, e-commerce– Communications: mobile phones, smart phones– Productivity enhancers: ATM, bar-code scanning,

lightning-fast credit card authorization– Lightning fast and free information, both public and

proprietary inside the firm• Everything on this list has already happened

IR #3 Has Failed the TFP Test

• Failure #1: TFP growth post-1970 barely 1/3 of 1920-70

• Failure #2: IR #3 boosted TFP growth only briefly 1996-2004

• STARTLING CONCLUSION: MIGHT THE PRODUCTIVITY IMPACT OF THE THIRD INDUSTRIAL REVOLUTION BE ALMOST OVER?

Real GDP Per Capita Is Not the Same As TFP And Does Not Measure

Innovation

Per-capita Real GDP Growth Now IsBelow Pessimistic Trend

The Standard of Living Is Not the Same as Productivity Growth

• Total Output or GDP (Y). • Total Hours of Work (H). • Total Population (N). Y&H refer to total economy• The Productivity Identity

Y/N Ξ Y/H * H/N

Debut of Data on TFP Growth Since 1890

• Forthcoming Book: Beyond the Rainbow: The American Standard of Living Since the Civil War• Real GDP, BEA back to 1929, Kendrick

1889-1929• Aggregate Hours BLS back to 1948, then

Kendrick 1889-1948• Capital Input, BEA back to 1925, then

Kendrick

Adjustments to Capital

• Both private and government capital, except weapons • Variable retirement (1929-65)• Changing weights on structures

vs. equipment to reflect shorter service lives of equipment (makes BEA = BLS)

Why Was Growth of Real GDP Per-capita So Steady at 2.1% Per Year?

Per-Capita Income GrowthDoes Not Equal Productivity Growth

The Same History, Justfor Productivity (Y/H) Growth

The Second Industrial Revolutionvs. the Third Industrial Revolution

The Effect of Education and Capital Deepening

The Second Industrial Revolutionvs. the Third Industrial Revolution

Are the TFP Effects of IR #3Almost Over?

The Second Industrial Revolutionvs. the Third Industrial Revolution

Productivity Impact of IR #3Is Largely Over (1960-2014)

• What’s old?– Amazon (1994), Google (1998), Wiki (2001), itunes

(2001), Facebook (2004)– Digitalization of library and parts catalogues is finished– Office work was revolutionized by 2004– Bar-code scanning, the ATM machine, instant credit card

authorization? Complete by 15 years ago• What’s new?• Iphone (2007), ipad (2010)

Stasis Everywhere You Look

• Offices use desktop computers and proprietary information as they did 10-15 years ago– The Northwestern econ department staff, 1998 vs. 2014

• Retail stasis. Shelves stocked by humans, meat sliced at service counters, checkout bar-code scanning. Maybe card authorization a bit faster

• Medicine: electronic medical records largely rolled out,

little or no change in what nurses and doctors do• Higher Education: cost inflation comes from rising

ratio of administrative staff to instructional staff

Stasis in Consumer Electronics

• NYT on Consumer Electronics Show, January 2014• This show was a far cry from the shows of old . . .

Over the years it has been the place to spot some real innovations (VCR 1970, CD 1981, HDTV 1998)

• This year’s crop of products seemed a bit underwhelming by comparison

• Editor of gadget website: “This industry that employs all of these engineers . . . Needs you to throw out your old stuff and buy new stuff – even if that new stuff is only slightly upgraded.

Quantitative Reasons 1999-2004Will Not Be Repeated

Growth in Manufacturing Capacity Has Ceased

The Most Dynamic Part of Manufacturing Has Disappeared

1972 1975-07 1979-01 1982-07 1986-01 1989-07 1993-01 1996-07 2000-01 2003-07 2007 2010-070

1

2

3

4

5

6

7

8

9

Share of ICT Manufacturing Value-Added in Total Manufacturing Sector, 1972-2013

Share as a Percentage

The ICT Deflator andThe Death of Moore’s Law

Optimistic to Conclude that FutureWill Resemble the Past Decade

• TFP Growth? The best guide is 2004-14, about 0.6• Capital deepening currently zero, will it gradually

return to 0.6?• Little further contribution from education• Conclusion? Depending on capital deepening,

upper bound on labor productivity growth of 1.2• Slower than 1.38 of 1972-96, but faster than the

0.80 of 2009-2014• Back to Erik, we are indeed at a “point of inflection”

we are heading down, not up