Upload
mcguinness-institute
View
1.005
Download
3
Embed Size (px)
Citation preview
Innova&on & the performance of New Zealand firms
Simon Wakeman NZAE conference 1 July 2016
1
Low investment in knowledge-‐based capital may explain up to 40% of the produc&vity gap rela&ve to the OECD average
2 Source: NZPC; de Serres et al. (2014)
Interna&onal comparisons show NZ near top of OECD rankings in genera&ng science but lower ranked in commercialising innova&on
NZ’S AVERAGE RANKING IN TOP-‐LEVEL CATEGORIES OF OECD STATISTICS
3
7
11 12
20 19
26
Science base Human resources Internet use for innovaCon
Business R&D and innovaCon
Knowledge flows and
commercialisaCon Entrepreneurship
Source: OECD (2012). Compara?ve performance of na?onal science and innova?on systems
Use Sta&s&cs NZ’s Longitudinal Business Database to study rela&onship between innova&on and produc&vity for New Zealand firms
DATA SOURCES
Innova&on ac&vity - Business Opera?ons Survey (BOS) - Intellectual Property Office (IPONZ)
Produc&vity es&mates (Fabling & Mare, 2015)
Financial data - Annual Enterprise Survey (AES) - IR10 tax returns
Access to the data presented was managed by Sta?s?cs New Zealand under strict micro-‐data access protocols and in accordance with the security and confiden?ality provisions of the Sta?s?c Act 1975. These findings are not Official Sta?s?cs. The opinions, findings, recommenda?ons, and conclusions expressed are those of the author/researcher, not Sta?s?cs New Zealand or the New Zealand Produc?vity Commission.
Longitudinal Business Database
R&D ac&vity - Business R&D Survey - Business Opera?ons Survey (BOS)
Government assistance - Payments from government programmes (incl. R&D grants)
LBD contains a range of measures reflec&ng different aspects of innova&on
MEASURES OF INNOVATIVE ACTIVITY
5
5
Sample contains responses to Business Opera?ons Survey (2005-‐2013).
0.0% 1.0% 2.0% 3.0% 4.0% 5.0% 6.0% 7.0% 8.0% 9.0% 10.0%
0% 5% 10% 15% 20% 25% 30% 35% 40% 45%
Engaged in R&D
Product development expenditure (R&D/design/markeCng/other)
Filed patent
Registered trademark
Any innovaCon
Product innovaCon (new to world/NZ/firm)
Sales from new products
Process innovaCon
OrganisaConal innovaCon
MarkeCng innovaCon
% total expenditure/sales
% firms engaged in ac&vity
R&D expenditure is not a good proxy for innova&on output in all cases
R&D ACTIVITY VS INNOVATION ACTIVITY TYPE OF INNOVATION EXPENDITURE BY SECTOR
6
Ever engaged in innovaCon
No Yes Ever engaged
in R&D No 27.5% 54.7% Yes 1.5% 16.3%
Sample contains firms responding to Business Opera?ons Survey Innova?on module in 2007, 2009 & 2011.
Sample contains firms responding to Business Opera?ons Survey (2005-‐2012).
0% 2% 4% 6% 8% 10% 12% 14%
Primary
Manufacturing
Services
R&D expenditure Design expenditure MarkeCng expenditure Other expenditure
Use approach developed by Fabling & Mare (2015) to es&mate firm produc&vity
MEASURING FIRM PERFORMANCE
7
Labour (L)
Raw materials/ inputs (M)
Capital (K)
ProducCon Output (GO)
( ) ( ) ( ) ( )ln ln ln ln
where , { , , }
r r sit j r it rs it i
r s rt i
r
GO X X X
r s L K M
α β δ γ ε≠
+= + + +
∈
∑ ∑∑
( )∑ ( ) ( ) ( )MFP:
Empl
ln(
oyment:
Valued added:
Labour productivity
ln ln ln ln
where , { , , }
:
)
it
it it
it it
it
r r sit it r it rs it it
rr s r
GO
GO
MFP GO X X X
r s L K M
L
M
ML
β δ≠
−
∈
−
−
−
= ∑∑ ∑
0
20
40
60
80% dif
Employment Value-added output Labour productivity MFPoutput type
R&Dactivity
Productinnovationnew to world
Productinnovation
Processinnovation
Organisationalinnovation
Marketinginnovation
RELATIVE OUTPUT LEVEL INNOVATORS VS. NON-‐INNOVATORS BY INNOVATION MEASURE
On average innova&ng firms larger but less produc&ve than non-‐innova&ng firms in year in which innova&on occurs
8
Chart shows difference in performance measure predicted from series of OLS regressions of output level in year 0 on innova?on in year 0 with controls for year and industry. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with performance measures up to 2012. Observa?ons weighted by BOS sampling weights.
0
10
20
30
%
1 2 3 4 5 6 7 8 9 10decile of mfp
Product innovation(new to world)
1 2 3 4 5 6 7 8 9 10decile of mfp
Product innovation(new to NZ)
1 2 3 4 5 6 7 8 9 10decile of mfp
Product innovation(new to the firm)
1 2 3 4 5 6 7 8 9 10decile of mfp
Process innovation
1 2 3 4 5 6 7 8 9 10decile of mfp
Organisational innovation
1 2 3 4 5 6 7 8 9 10decile of mfp
Marketing innovation
PROPORTION OF FIRMS ENGAGED IN INNOVATION BY DECILE OF PRODUCTIVITY IN YEAR 0
Most produc&ve firms least likely to innovate
9
Chart shows likelihood of firm innova?on predicted from probit regression of innova?on ac?vity in year 0 on produc?vity decile in year 0. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011. Observa?ons weighted by BOS sampling weights
Use difference-‐in-‐differences approach to measure impact of innova&on on produc&vity
EMPIRICAL APPROACH • Regress change in firm output/producCvity (lnYin -‐ lnYi0) on innovaCon in year 0 (I0)
10
0 0 0ln – ln in i I i X iY Y I X tα β β ε= + + + +
• Compare within-‐firm differences in output controls for unobserved firm characterisCcs affecCng output levels
• Control for observed firm characterisCcs (age, size, industry, etc.)
controls for observed firm characterisCcs affecCng changes in output
• For MFP, use 2-‐year moving average accounts for measurement error • Weight observaCons by BOS sampling weights Cmes predicted gross output
quanCCes corresponds to aggregate economic output/producCvity
• Do not instrument for innovaCon (cf Crepon, Duguet & Mairesse, 1998)
-6
-4
-2
0
2
4%
0 1 2 3 4years
change in employment
-6
-4
-2
0
2
4%
0 1 2 3 4years
change in value-added output
-6
-4
-2
0
2
4%
0 1 2 3 4years
change in labour productivity
-6
-4
-2
0
2
4%
0 1 2 3 4years
change in MFP
Yes No
-10
0
10
20%
0 1 2 3 4years
-10
0
10
20%
0 1 2 3 4years
-10
0
10
20%
0 1 2 3 4years
-10
0
10
20%
0 1 2 3 4years
RELATIVE GROWTH BY OUTPUT TYPE OF R&D ACTIVE VS NON-‐ACTIVE FIRMS
Firms engaged in R&D ac&vity have faster employment and output growth than non-‐R&D-‐ac&ve firms but produc&vity growth not different
Charts show difference in predic?ve margins from series of OLS regressions of change in output from year 0 to year t on innova?on in year 0. Regressions include controls for base year and firm characteris?cs. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output. 11
-4
-2
0
2
4%
0 1 2 3 4years
change in employment
-4
-2
0
2
4%
0 1 2 3 4years
change in value-added output
-4
-2
0
2
4%
0 1 2 3 4years
change in labour productivity
-4
-2
0
2
4%
0 1 2 3 4years
change in MFP
Yes No
-10
-5
0
5
10%
0 1 2 3 4years
-10
-5
0
5
10%
0 1 2 3 4years
-10
-5
0
5
10%
0 1 2 3 4years
-10
-5
0
5
10%
0 1 2 3 4years
RELATIVE GROWTH BY INNOVATION ACTIVITY OF INNOVATORS VS NON-‐INNOVATORS
Similarly firms engaged in innova&on grow faster in size but not produc&vity
Charts show difference in predic?ve margins from series of OLS regressions of change in output from year 0 to year t on innova?on in year 0. Regressions include controls for base year and firm characteris?cs. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output. 12
-6
-4
-2
0
2%
0 1 2 3 4years
Product innovation
-6
-4
-2
0
2%
0 1 2 3 4years
Process innovation
-6
-4
-2
0
2%
0 1 2 3 4years
Organisational innovation
-6
-4
-2
0
2%
0 1 2 3 4years
Marketing innovation
Yes No
-4
-2
0
2
4%
0 1 2 3 4years
-4
-2
0
2
4%
0 1 2 3 4years
-4
-2
0
2
4%
0 1 2 3 4years
-4
-2
0
2
4%
0 1 2 3 4years
RELATIVE MFP GROWTH BY INNOVATION TYPE OF INNOVATORS VS NON-‐INNOVATORS
Only firms engaged in marke&ng innova&on experience significantly higher produc&vity growth
Charts show coefficients from OLS regressions of change in 2-‐yr-‐MA of MFP from year 0 to t on innova?on in year 0. Regressions include controls for base year and firm characteris?cs. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
13
-6
-4
-2
0
2
4%
0 1 2 3 4years
2005
-6
-4
-2
0
2
4%
0 1 2 3 4years
2007
-6
-4
-2
0
2
4%
0 1 2 3years
2009
-6
-4
-2
0
2
4%
0 .2 .4 .6 .8 1years
2011
Yes No
-5
0
5
10%
0 1 2 3 4years
-5
0
5
10%
0 1 2 3 4years
-5
0
5
10%
0 1 2 3years
-5
0
5
10%
0 1years
RELATIVE MFP GROWTH BY YEAR OF PRODUCT INNOVATORS VS NON-‐INNOVATORS
Returns to innova&on vary drama&cally by year
Charts show coefficients from OLS regressions of change in 2-‐yr-‐MA of MFP from year 0 to t on innova?on in year 0. Regressions include controls for base year and firm characteris?cs. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
14
-6
-4
-2
0
2
4%
2005 2006 2007 2008 2009 2010 2011 2012year
all No Yes
RELATIVE MFP GROWTH OVER TIME OF PRODUCT INNOVATORS VS NON-‐INNOVATORS
Product innovators experience higher MFP growth during economic recovery but possibly lower growth during economic downtown
Charts show coefficients from OLS regressions of change in 2-‐yr-‐MA of MFP from year 0 to t on innova?on in year 0. Regressions include controls for base year and firm characteris?cs. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
15
RELATIVE MFP GROWTH BY YEAR OF PROCESS INNOVATORS VS NON-‐INNOVATORS
Process innovators appear not to do significantly becer or worse than non-‐innovators
Charts show coefficients from OLS regressions of change in 2-‐yr-‐MA of MFP from year 0 to t on innova?on in year 0. Regressions include controls for base year and firm characteris?cs. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
16
-6
-4
-2
0
2
4%
0 1 2 3 4years
2005
-6
-4
-2
0
2
4%
0 1 2 3 4years
2007
-6
-4
-2
0
2
4%
0 1 2 3years
2009
-6
-4
-2
0
2
4%
0 .2 .4 .6 .8 1years
2011
Yes No
-10
-5
0
5%
0 1 2 3 4years
-10
-5
0
5%
0 1 2 3 4years
-10
-5
0
5%
0 1 2 3years
-10
-5
0
5%
0 1years
-6
-4
-2
0
2
4%
2005 2006 2007 2008 2009 2010 2011 2012year
all No Yes
RELATIVE MFP GROWTH OVER TIME OF PROCESS INNOVATORS VS NON-‐INNOVATORS
Returns to process innova&on do not vary significantly by year
Charts show coefficients from OLS regressions of change in 2-‐yr-‐MA of MFP from year 0 to t on innova?on in year 0. Regressions include controls for base year and firm characteris?cs. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
17
RELATIVE MFP GROWTH BY YEAR OF ORGANISATIONAL INNOVATORS VS NON-‐INNOVATORS
Similarly returns to organisa&onal innova&on in 2009 significantly higher but not in other years
Charts show coefficients from OLS regressions of change in 2-‐yr-‐MA of MFP from year 0 to t on innova?on in year 0. Regressions include controls for base year and firm characteris?cs. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
18
-6
-4
-2
0
2
4%
0 1 2 3 4years
2005
-6
-4
-2
0
2
4%
0 1 2 3 4years
2007
-6
-4
-2
0
2
4%
0 1 2 3years
2009
-6
-4
-2
0
2
4%
0 .2 .4 .6 .8 1years
2011
Yes No
-4
-2
0
2
4
6%
0 1 2 3 4years
-4
-2
0
2
4
6%
0 1 2 3 4years
-4
-2
0
2
4
6%
0 1 2 3years
-4
-2
0
2
4
6%
0 1years
-6
-4
-2
0
2
4%
2005 2006 2007 2008 2009 2010 2011 2012year
all No Yes
RELATIVE MFP GROWTH OVER TIME OF ORGANISATIONAL INNOVATORS VS NON-‐INNOVATORS
Firms doing organisa&onal innova&on during economic downturn experience significantly higher returns during economic recovery
Charts show coefficients from OLS regressions of change in 2-‐yr-‐MA of MFP from year 0 to t on innova?on in year 0. Regressions include controls for base year and firm characteris?cs. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
19
RELATIVE MFP GROWTH BY YEAR OF MARKETING INNOVATORS VS NON-‐INNOVATORS
By contrast, marke&ng innovators in 2005 experienced lower produc&vity growth but in 2007 experienced higher growth
Charts show coefficients from OLS regressions of change in 2-‐yr-‐MA of MFP from year 0 to t on innova?on in year 0. Regressions include controls for base year and firm characteris?cs. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
20
-10
-5
0
5%
0 1 2 3 4years
2005
-10
-5
0
5%
0 1 2 3 4years
2007
-10
-5
0
5%
0 1 2 3years
2009
-10
-5
0
5%
0 .2 .4 .6 .8 1years
2011
Yes No
-10
-5
0
5
10
15%
0 1 2 3 4years
-10
-5
0
5
10
15%
0 1 2 3 4years
-10
-5
0
5
10
15%
0 1 2 3years
-10
-5
0
5
10
15%
0 1years
-10
-5
0
5%
2005 2006 2007 2008 2009 2010 2011 2012year
all No Yes
RELATIVE MFP GROWTH OVER TIME OF MARKETING INNOVATORS VS NON-‐INNOVATORS
Firms that had introduced marke&ng innova&on did significantly becer during economic downturn
Charts show coefficients from OLS regressions of change in 2-‐yr-‐MA of MFP from year 0 to t on innova?on in year 0. Regressions include controls for base year and firm characteris?cs. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
21
-20
-10
0
10%
0 1 2 3 4years
2005
0 1 2 3 4years
2007
0 1 2 3years
2009
0 .2 .4 .6 .8 1years
2011
new to world new only to NZ new only to firm not new
RELATIVE MFP GROWTH BY YEAR OF PRODUCT INNOVATORS BY NOVELTY OF PRODUCT
Innova&on followers did becer than innova&on leaders in 2005 & 2007 but innova&on leaders did becer in 2009
Charts show coefficients from OLS regressions of change in 2-‐yr-‐MA of MFP from year 0 to t on innova?on in year 0. Regressions include controls for base year and firm characteris?cs. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
22
-10
-5
0
5
10%
2005 2006 2007 2008 2009 2010 2011 2012year
all new to world new only to NZ new only to firm not new
RELATIVE MFP GROWTH OVER TIME OF PRODUCT INNOVATORS BY NOVELTY OF PRODUCT
Innova&on followers performed becer during economic downtown but innova&on leaders performed becer during the recovery
Charts show coefficients from OLS regressions of change in 2-‐yr-‐MA of MFP from year 0 to t on innova?on in year 0. Regressions include controls for base year and firm characteris?cs. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
23
-20
-10
0
10
20%
0-5 yr
s
5-10 y
rs
10-20
yrs
20-50
yrs
50+ y
rs
age
R&Dactivity
Productinnovation
Processinnovation
Organisationalinnovation
Marketinginnovation
Younger firms experience highest boost in produc&vity growth following innova&on
RELATIVE CHANGE IN MFP OVER 3 YEARS BY INNOVATION TYPE & AGE
24
Charts show difference in predic?ve margins from series of OLS regressions of change in 3-‐yr MA of MFP from year 0 to year t on innova?on in year 0 interacted with firm characteris?cs and controls for base year. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
-10
-5
0
5
10%
0-20 20-50 50-100 100+total employment
R&Dactivity
Productinnovationnew to world
Productinnovation
Processinnovation
Organisationalinnovation
Marketinginnovation
Small-‐to-‐medium size firms appear to do worse if they innovate
RELATIVE CHANGE IN MFP OVER 3 YEARS BY INNOVATION TYPE & FIRM SIZE
25
Charts show difference in predic?ve margins from series of OLS regressions of change in 3-‐yr MA of MFP from year 0 to year t on innova?on in year 0 interacted with firm characteris?cs and controls for base year. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
-20
-10
0
10
20%
Primary Manufacturing Servicessector
R&Dactivity
Productinnovationnew to world
Productinnovation
Processinnovation
Organisationalinnovation
Marketinginnovation
Firms in manufacturing sector experience strongest impact of innova&on on MFP growth
RELATIVE CHANGE IN MFP OVER 3 YEARS BY INNOVATION TYPE & SECTOR
26
Charts show difference in predic?ve margins from series of OLS regressions of change in 3-‐yr MA of MFP from year 0 to year t on innova?on in year 0 interacted with firm characteris?cs and controls for base year. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
-20
-10
0
10
20%
NZ-owned domestic foreign-owned domestic internationallevel of internationalisation
R&Dactivity
Productinnovationnew to world
Productinnovation
Processinnovation
Organisationalinnovation
Marketinginnovation
Interna&onally connected firms engaged in product or organisa&onal innova&on experience higher MFP growth, while domes&c firms have lower growth
RELATIVE CHANGE IN MFP OVER 3 YEARS BY INNOVATION TYPE & INTERNATIONAL FOCUS
27
Charts show difference in predic?ve margins from series of OLS regressions of change in 3-‐yr MA of MFP from year 0 to year t on innova?on in year 0 interacted with firm characteris?cs and controls for base year. Bars show 95% confidence intervals. Sample contains firms in BOS 2005-‐2011 with output measures un?l 2012. Observa?ons weighted by BOS sampling weights ?mes predicted output.
Innova&on macers for produc&vity, but more for some firms than others
FINDINGS • R&D and innovaCon associated with growth in size but not necessarily with producCvity • Firms doing markeCng innovaCon show clearest returns overall, especially in 2007-‐2009 (economic downturn)
• Firms doing product and organisaConal innovaCon performed beeer afer 2009 (economy recovery), but possibly worse before
• Younger firms, manufacturing firms, and firms with internaConal connecCons have highest returns to innovaCon
CONCLUSIONS • More to innovaCon than R&D • Non-‐technological-‐types of innovaCon maeer as much if not more • Market condiCons important • Some types of firms get more value out of innovaCon than others
For more informa&on www.produc&vity.govt.nz
@Nzprocom 28