the impact of digital transformation on productivity … · the impact of digital transformation on...

www.conferenceboard.orgCarol Corrado, The Outlook for the US Economy, Brookings, Mary 31, 20171

The Impact of Digital Transformation on Productivity

Carol CorradoThe Conference Board and

Center for Business and Public Policy, McDonough School of Business,

Georgetown University

OECD-BEIS Conference

BEIS Conference Centre, London, UK

November 8, 2018

http://www.conference-board.org/


Why did productivity growth slow?

IT revolution over

Non-technological forces (secular stagnation, demographic

headwinds)

Mismeasurement (within GDP, beyond GDP)

Misunderstanding of history’s lessons



• Focus of my remarks:

• Diffusion, Digitization, (GDP) Dissatisfaction

• Summary:

• Better measurement of things digital clear part of

the puzzle



Compared with bubble years, weak US labor productivity

reflects slow growth in both capital deepening and TFP

Source. Elaboration of estimates issued by the BLS.

.6

1.4

.5

.4

.2

.3

.7

1.1

.4

.1

.1

.1

0

0.5

1

1.5

2

2.5

3

1987 to 1997 1997 to 2007 2007 to 2017

R&D Capital Deepening

Capital Deepening excl R&D

Labor composition

TFP

2.9

1.2

TFP: Diffusion

of commercial

and scientific

knowledge

Contributions to labor productivity growth, nonfarm business sectorPercentage points

1.7



“Old normal”

* ICT Manufacturing: Information services, IT design consulting


0

0.3

0.6

0.9

1.2

1.5

1987 to 1997 1997 to 2007 2007 to 2017

NFB TFP growth

ICT Producers* Contrib.

Total factor productivity, nonfarm business sector, and contribution of tech

producersPercentage points



“IT revolution” Diffusion of ICT



0

0.3

0.6

0.9

1.2

1.5

1987 to 1997 1997 to 2007 2007 to 2017

NFB TFP growth






Hard to interpret the ”new normal”



0

0.3

0.6

0.9

1.2

1.5

1987 to 1997 1997 to 2007 2007 to 2017

NFB TFP growth






…except, perhaps, when corrected for mismeasurement

of consumer IT services (within GDP)



0

0.3

0.6

0.9

1.2

1.5

1987 to 1997 1997 to 2007 2007 to 2017

NFB TFP growth


After correcting

for mismeasurement

of consumer

digital access

services





What accounts for the apparent lack of diffusion?

. . . there’s a consumer story and a business story

Consumer story -- an understatement of digital services owing to

many innovations in the delivery of content to consumers

“Free” services from apps and home use/consumption are uncounted

Internet/network access services measure payments for plans, not real

consumption of services

Both of these grow very fast from 2007 to 2017 (mismeasurement)*

Business story – has many tentacles

Investments in business models, IT systems, data analytics are

uncounted. Not a new story, but returns/spillovers may be diminished.

New service modes, new delivery systems. Are these being missed?

New product introductions. Is the value of variety being missed?

*See “Accounting for Innovation Consumer Digital Services: Implications for GDP and Consumer Welfare” by David

Byrne and Carol Corrado, November 2017 (preliminary)



Financial performance of intangibles-intensive large cap

firms outstrips its benchmark, especially in recent years

Notes:

• CNBC IQ100 is an index

comprised of 100 firms

(chosen from the Russell

1000) with greatest

forward-looking

profitability based on their

intangible capital

• Not tech-heavy

• Developed by M-Cam

using its global IP

database

0

50

100

150

200

250

300

350

6/1/07 6/1/09 6/1/11 6/1/13 6/1/15 6/1/17

Jan. 2010 = 100

CNBC IQ100 Russell 1000

Sept. 2018



5%

7%

9%

11%

13%

15%

17%

19%

1977 1982 1987 1992 1997 2002 2007 2012 2017

Tangibleinvestment rate

Intangibleinvestment rate

Investment rates, Private industries, 1977 to 2017

(investment relative to private industry value added)

…but the pace of increase in overall intangible capital

has slowed in recent years

Note: Non-national accounts Intangibles estimates are preliminary and do not include own-account expenditures on brand and customer

development as reported in Corrado and Hao (2013).. Data available at www.intaninvest.net



No payoffs in terms of productivity in industries where firms

increased use of ICT services (IT systems design, cloud)

Oil and gas

Other mining

Utilities

Wholesale

Retail

Air transport

Water transportation

Other transportation

Motion picture

Banks

Funds

Legal services

Waste management

Hospitals

Accomodation

Other services

Chemical

Metal productionMachinery

-14%

-10%

-6%

-2%

2%

6%

10%

-0.8% -0.6% -0.4% -0.2% 0.0% 0.2% 0.4% 0.6% 0.8% 1.0%

Accele

rati

on

in

TF

P g

row

th, 2009 t

o

2013 l

(an

nu

al

rate

) ess 2

001 t

o 2

007

(an

nu

al

rate

)

Relative intensity of ICT services use, 2008–2013



... this is not because they aren’t investing in intangibles!

Oil and gas

Utilities

WholesaleRetail

Water transportation

Pipeline transportation

Warehousing

Motion picturesBanks

Insurance

Funds

Hospitals

Chemical

Metal productionMachinery

Transport equipment

-8%

-6%

-4%

-2%

0%

2%

4%

6%

-0.8% -0.6% -0.4% -0.2% 0.0% 0.2% 0.4% 0.6% 0.8% 1.0%

Re

lati

ve

in

ten

sit

y o

f in

tan

gib

le

investm

en

t in

desig

n a

nd

eco

no

mic

co

mp

ete

nc

ies

, 20

08-2

01

3

Relative intensity of ICT services use, 2008–2013



Qualifications

If investments in artificial intelligence are not fully represented in

existing intangibles measures, then productivity is mismeasured

Pattern of impact is called the ”AI innovation J-curve" in a recent

paper **

Investments in AI should be counted as increases in intangible

capital

** Brynjolfsson, Rock, and Syverson, “Artificial Intelligence and the Modern Productivity Paradox: A Clash of

Expectations and Statistics,” NBER Working paper (October 2018).



Available indicators suggest a surge of investments in AI

.0%

.2%

.4%

.6%

.8%

1.0%

1.2%

1.4%

1.6%

1.8%

2.0%

1987 1992 1997 2002 2007 2012 2017

Pe

rce

nt o

f G

DP

ICT R&D and computer design servicescontinue a strong upward trend

Computer design services

IT Equip./Software/Data-related R&D

Source: Blog by Dario Amodei and Danny Hernandex,

“AI and Compute”, May 16, 2018, at

https://blog.openai.com/ai-and-compute/

Computing power used in AI training runs



.0

.1

.2

.3

.4

.5

.6

.7

.8

.9

1.0

1959 1966 1973 1980 1987 1994 2001 2008 2015

ICT

Other

Pharma and related

Business R&D, funder basis (1959 to 2016)Percent of GDP

But R&D investment rates in other broad business sub-

sector are flat

Note. ICT R&D is R&D in NAICS 334, 51pt, and 5415 (i.e., includes software development). ”Related” in “Pharma and related” refers to

medical instruments R&D and biotech R&D services. ”Other” is R&D in all other industries.

Source. Elaboration of data issued by BEA. The sector split in 2016 is estimated.



Summary

Part of the slowdown in productivity may be due to fewer spillovers

from nonR&D intangibles, i.e., a slowdown in diffusion of commercial

knowledge

All things digital suggest optimism for future productivity growth

R&D in ICT/AI/computing is very robust

Payoffs to measured and unmeasured investments in data analytics by

businesses not evident in productivity (yet)

Digital mismeasurement can also explain part of the slowdown

Productivity of telecommunication services (internet, mobile, video) is

understated, and the understatement is larger after 2007



Thankyou.

[email protected]


mailto:[email protected]

the impact of digital transformation on productivity … · the impact of digital transformation on...

Documents