Employment Effects of Innovation at the Firm Level
description
Transcript of Employment Effects of Innovation at the Firm Level
ifo Institute for Economic Research at the University of Munich
Employment Effects of Innovation at the Firm Level
Stefan Lachenmaier*, Horst Rottmann♦
* Ifo Institute for Economic Research at the University of Munich♦ University of Applied Sciences Amberg-Weiden and Ifo Institute
3. Konferenz für Sozial- und Wirtschaftsdaten, Mai 2006, Wiesbaden
ifo Institute for Economic Research at the University of Munich
The Research Question
• Do innovations have a significant effect on employment?
• Concentrate on the analysis of long-term effects
ifo Institute for Economic Research at the University of Munich
Motivation
• Theoretical contributions show different results– Product innovations: increase demand
increase employment level decrease competition increase market power reduce output decrease employment
– Process innovations: increase labour productivity decrease employment level lower costs lower prices higher demand stimulate emplyoment
Overall effect is depending on elasticity of demand
• Empirical evidence is necessary• Panel studies are rare due to the lack of appropriate data
ifo Institute for Economic Research at the University of Munich
Main Idea
• Exploit long innovation panel data set
• Distinguish between product and process innovations
• Introduce different innovation categories
ifo Institute for Economic Research at the University of Munich
Related Literature
• Theoretical Contributions:– Petit (1995)– Stoneman (1984), Hamermesh (1993)
• Empirical Contributions:– Chennels / Van Reenen (1999)– Cross-Sectional Analyses: Zimmermann (1991), König et al.
(1995)– Employment Growth Analyses: Brouwer et al. (1993),
Blanchflower/ Burgess (1999), Blechinger et al. (1998)– Panel Analyses: Smolny (1998), van Reenen (1997),
Rottmann/Ruschinski (1998)
ifo Institute for Economic Research at the University of Munich
Database
• Ifo Innovation Survey• Panel Structure: 1982-2003 (unbalanced)• German Manufacturing Sector• ~1300 observations per year• Contains information on:
– Innovation: Product and Process Innovation, Innovations introduced, Innovation expenditure
– Firm characteristics: firm size, NACE, German states, turnover
• Control variables added on sector level (2digit NACE)
ifo Institute for Economic Research at the University of Munich
Empirical Model
• Modelling employment adjustment process is complex, esp. for small firms (e.g. Hamermesh / Pfann 1996)
• Labour demand reacts slowly to changes in innovation behaviour
• Estimating long-term effects: Following Blanchard / Wolfers (2000), Nickell (1997, 2003)
– Calculating averages for 4-(and 5-year periods)
– Use period averages for panel analysis (time index t indicates period)
ifo Institute for Economic Research at the University of Munich
Estimation
),,( XQTfL
L: Labour demand T: TechnologyQ: Product quality X: Controls
Level Equation:
Linear Equation in differenced log values:
3210 xqtl
- transformed into growth rates
- allows to introduce innovation variables
- Eliminates unobservable firm effect
ifo Institute for Economic Research at the University of Munich
itititititPd
itPc
it uegwIIl 543210
w: Growth of Real Hourly Wage Rate (sectoral)g: Growth of Real Gross Value Added (sectoral)eit: log of employment start level
Estimation Equation:
Pc: Process Innovation (proxy for t) Pd: Product Innovation (proxy for q)
Remember:
-Variables are expressed in averages over periods
ifo Institute for Economic Research at the University of Munich
The problem of endogeneity
• Potential contemporaneous correlation of innovation and error term resulting from a shock simultaneously affecting employment and innovation
• IV Strategy– So far we tested Innovation Impulses, Innovation Obstacles,
lagged values– No robust results: Either instruments are not good (low
significance in first stage) or not valid (Sargan test)
ifo Institute for Economic Research at the University of Munich
Descriptive Statistics
Until 1990: Former West Germany, since 1991: GermanyUnbalanced panel: 9142 „observations“, 4567 different firms, 5 time categories
Descriptive Statistics Mean Std. Dev. Min MaxEmpoloyment Growth (log) -0.016 0.261 -2.708 2.996Innovation 0.497 0 1 Product Innovation 0.406 0 1 Process Innovation 0.317 0 1Employment Start Level (log) 4.682 1.506 0 11.513Sectoral GVA Growth 0.005 0.046 -0.265 0.283Sectoral Real Wage Growth 0.018 0.026 -0.231 0.428
n=9142, N=4567
ifo Institute for Economic Research at the University of Munich
Regressions I
Dependent Variable: Average Yearly Employment Growth(1) (2) (3)
Estimated OLS standard Heteroskedasticity Covariance robustCoefficients errors robust s.e. standard errors
Employment Start Level -0.034 (0.002)*** (0.003)*** (0.003)***Real Wage Growth -0.437 (0.132)*** (0.162)*** (0.161)***Real GVA Growth 0.257 (0.081)*** (0.102)** (0.102)**Product Innovation 0.033 (0.008)*** (0.009)*** (0.009)***Process Innovation 0.057 (0.009)*** (0.009)*** (0.009)***Year incl.Sector incl.States incl.Constant 0.112 (0.026)*** (0.024)*** (0.026)***Observations 9142Adj. R-squared 0.039Standard errors in parentheses* significant at 10%; ** significant at 5%; *** significant at 1%
ifo Institute for Economic Research at the University of Munich
Regressions II
Dependent Variable: Average Yearly Employment Growth
Employment Start Level -0.039*** (0.004) -0.034*** (0.004) -0.060*** (0.008)Real Wage Growth -0.500** (0.220) -0.572** (0.255) -0.471 (0.433)Real GVA Growth 0.282** (0.124) 0.341** (0.137) 0.134 (0.244)Product Innovation 0.053*** (0.012) 0.049*** (0.014) 0.066*** (0.024)Process Innovation 0.052*** (0.013) 0.062*** (0.014) 0.036 (0.028)Year incl. incl. incl.Sector incl. incl. incl.States incl. incl. incl.Constant 0.099** (0.039) 0.076* (0.039) 0.200*** (0.048)Observations 5485 4136 1349Adj. R-squared 0.038 0.031 0.087Robust standard errors in parentheses
* significant at 10%; ** significant at 5%; *** significant at 1%
1991-2003 West 1991-2003 East 1991-2003(4) (5) (6)
ifo Institute for Economic Research at the University of Munich
Regressions III
Dependent Variable: Average Yearly Employment Growth
Employment Start Level -0.044*** (0.005) -0.025*** (0.006)Real Wage Growth -0.498** (0.215) -0.399 (0.250)Real GVA Growth 0.157 (0.145) 0.405*** (0.140)Product Innovation 0.044*** (0.012) 0.018 (0.013)Process Innovation 0.064*** (0.012) 0.044*** (0.013)Year incl. incl.Sector incl. incl.States incl. incl.Constant 0.142*** (0.030) 0.067 (0.060)Observations 6062 3080Adj. R-squared 0.035 0.031Robust standard errors in parentheses
* significant at 10%; ** significant at 5%; *** significant at 1%
less than 200 employees
equal or more than 200 employees
(7) (8)
ifo Institute for Economic Research at the University of Munich
Regressions IVDependent Variable: Average Yearly Employment Growth
Employment Start Level -0.034*** (0.003) -0.035*** (0.003)Real Wage Growth -0.439*** (0.161) -0.444*** (0.162)Real GVA Growth 0.260** (0.102) 0.256** (0.102)Innovation 0.063*** (0.014) ---Innovation (R&D) -0.007 (0.014) ---Innovation (Patents) 0.026** (0.011) ---Product Innovation --- 0.044*** (0.017)Process Innovation --- 0.050*** (0.013)Product Innovation (R&D) --- -0.027* (0.016)Process Innovation (R&D) --- 0.006 (0.014)Product Innovation (Patents) --- 0.026** (0.012)Process Innovation (Patents) --- 0.031 (0.025)Year incl. incl.Sector incl. incl.States incl. incl.Constant 0.107*** (0.026) 0.119*** (0.026)Observations 9140 9096Adj. R-squared 0.038 0.039Robust standard errors in parentheses
* significant at 10%; ** significant at 5%; *** significant at 1%
(9) (10)
ifo Institute for Economic Research at the University of Munich
Summary
• Innovations show positive effects on employment growth• True for product as well as process innovations. Process
innovations show even higher effect• No additional effect for R&D based innovations• Additional effect for product innovations which involved
patent applications• Further Research: Dynamics of adjustment process