Workshop productivity Bern, Swit Does competition stimulate innovation and productivity in Dutch retail trade? Henry van der Wiel CPB Netherlands Bureau for Economic Policy Analysis & CentER OECD Workshop on Productivity Analysis and Measurement, Bern, October 2006
Workshop productivity Bern, Swit Outline Introduction/background Relation competition and innovation 2 research questions Model for innovation and productivity Data and empirical results Concluding remarks
Workshop productivity Bern, Swit Background (I) Dutch retail trade on its return? ► Missed strong productivity growth of US since mid 1990s ► Gradually losing its strong position in EU since end of 1980s Labour productivity (per hours worked) relative to EU-average (EU=100), Source: GGDC,2005
Workshop productivity Bern, Swit Background (II): lack of competition and innovation? General belief that competition may stimulate productivity => s tatic efficiency ► lower price margins ► efficient production (less X-inefficiencies) Likewise, innovation enhances productivity (growth) => dynamic efficiency Dutch policy measures in 1990s focussed on more competition ► New Competition Act in 1998 ► Regulatory reforms in retail trade => longer opening hours (1996)
Workshop productivity Bern, Swit Competition and innovation: negative or positive? Negative relationship: ► Standard IO literature and most (early) endogenous growth models ► Schumpeter model: competition reduces monopoly rents and thus the expected pay off from innovation Positive relationship: mostly based on empirics ► Paper of Nickell (1996): competition is good for innovation New development => inverted U-shaped curve ► Combining both theoretical ideas ► Empirically supported by Aghion et al. (2005, QJE) for UK
Workshop productivity Bern, Swit Competition and innovation: inverted U-curve? Inverted U: composition effect ► Weak competition: –industry is relatively often in a level state => –increase in competition stimulates innovation by the "escape" effect ► Intense competition: –industry is often unlevelled=> –increase in competition reduces innovation because there is little incentive for laggards to catch up
Workshop productivity Bern, Swit Two questions I.Did competition affect innovation in Dutch retail trade? II.Did competition and innovation contribute to productivity growth in this industry? Conclusion: more competition in Dutch retail trade stimulates both innovation and productivity growth
Workshop productivity Bern, Swit Outline Introduction/background Relation competition and innovation 2 research questions Model for innovation and productivity Data and empirical results Concluding remarks
Workshop productivity Bern, Swit Basic idea of CIP-model Assume no feedback from P to C or from I to C Competition (C) Innovation(I) Productivity (P) Static efficiency Dynamic efficiency Inverted U-curve?
Workshop productivity Bern, Swit CIP-model Presentation only focuses on results for innovation and productivity Skip model for explaining competition (in paper!), but not how to measure competition See also Creusen, Minne and Van der Wiel, 2006, in De Economist, September
Workshop productivity Bern, Swit How to measure competition We introduce a new measure, relative profits measure (RPM): ► based on intuition that in a more competitive market, firms are punished more harshly for being inefficient Firms differ in efficiency in terms of marginal costs (or productivity level). ► Cost advantages lead up to higher profits We estimate for an industry the following elasticity: percentage increase in profits due to a 1 percent increase in efficiency
Workshop productivity Bern, Swit Cons traditional measures competition Conventional ways of measuring competition (concentration (H) and price cost margin (PCM)) are not robust from a theoretical point of view Problem with H is that more aggressive conduct forces inefficient firms out of the market thereby increasing concentration ► It incorrectly suggests that competition is reduced As conduct becomes more aggressive, market share is reallocated from inefficient firms (with low PCM) to efficient firms (with high PCM) which tends to raise industry wide PCM ► It incorrectly suggests that competition is reduced
Workshop productivity Bern, Swit Innovation: model Explanation of innovation: inn = α 0 + α 1 RPM + α 2 RPM 2 + βms withinnlog innovation rate (firm level) RPMcompetition indicator (5-digit industry level) mslog market share (firm level) Expectations ► If inverted U: α 1 > 0 and α 2 < 0 ► Scale effect: β > 0
Workshop productivity Bern, Swit Productivity growth: model Simple Cobb Douglas function: ► Split TFP-growth in contribution of competition and innovation Explanation of labour productivity growth: Δp = γ 0 + γ 1 ΔRPM + γ 2 INN -1 + γ 3 (Δk - Δl) + γ 4 Δl ───────┬──────── TFP-growth withΔp labour productivity growth (firm level) (Δk - Δl) capital intensity (firm level) Δl labour (economies of scale, firm level)
Workshop productivity Bern, Swit Outline Introduction/background Relation competition and innovation 2 research questions Model for innovation and productivity Data and empirical results Concluding remarks
Workshop productivity Bern, Swit Data and method Firm-level data ► Two sources of Statistics Netherlands –CIS-innovation surveys: 1996,1998 and 2000 –Annual surveys ‘Production Census’: ► Matched both sources –Number of observations ≈1150 Regression methods: ► Innovation based on TOBIT I-method –Innovation outlays left censored: no innovation in 75% of firms ► Productivity based on OLS
Workshop productivity Bern, Swit Innovation results (I) Estimation results of quadratic model (Tobit-I model) DeterminantEstimateT-value Intercept−0.07−1.76 Competition−0.05−1.50 Competition Market share Scale parameter a Number of observations1147 Left-censored observations864 Log-likelihood Source: own calculations based on PS- and CIS-data a Scale parameter in the distribution used to normalize the underlying variable No inverted U-relation !!
Workshop productivity Bern, Swit Innovation results (II): simplified model Estimation results of linear model innovation (Tobit-I model) DeterminantEstimatet-value Intercept Competition Market share Scale parameter a 0.10 Number of observations1147 Left-censored observations864 Log-likelihood 72.9 Source: own calculations based on PS- and CIS-data a Scale parameter in the distribution used to normalize the underlying variable
Workshop productivity Bern, Swit Productivity growth results Estimation results productivity, a DeterminantEstimateT-value Competition Lagged innovation Capital intensity Labour (economies of scale) 0.00 0.45 Intercept 0.02 0.61 R-squared0.17 Number of observations883 Source: own calculations based on PS- and CIS-data a Incorporated years: 1997, 1999, and 2001, due to limited availability of innovation data
Workshop productivity Bern, Swit Concluding remarks No inverted U-relationship in Dutch retail trade! ► positive relation between competition and innovation Both competition and innovation have a positive impact on productivity growth So more competition in Dutch retail trade may stimulate productivity growth ► in the short term by reductions in X-inefficiency ► in the longer term by innovation