EU harmonization of regulations, competition and rm-level productivity growth in the Dutch food industry

Transcription

1 EU harmonization of regulations, competition and rm-level productivity growth in the Dutch food industry Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan 2 Hasselt University, Université catholique Louvain and Statistics Netherlands December 2009 ark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level productivity December 2009 growth in1 the / 16 Dut

2 Motivation Previous research has identi ed the actual trade gains from the EU harmonization of food regulations. Examples include the CEC report, 1997; Henry de Frahan and Vancauteren, 2006; Chevassus-Lozza et al., 2008 However little is known on how these particular trade liberalization measures may impact productivity? Empirical rm/plant level studies on the direct relationship between trade liberalization and productivity o ers con icting answers. Only robust evidence is a positive relationship between export participation and productivity of rms (self-selection). Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level productivity December 2009 growth in2 the / 16 Dut

3 to empirically examine the direct and indirect impacts of TBT harmonization on TFPG. the general presumption is that trade liberalization has a positive impact on productivity through the e ect of competitive pressures to which domestic rms are exposed direct: resource allocation e ect indirect: resource allocation e ect via a mechanism of e.g., market structure provide more insights into the empirical tractability of this indirect link Given: empirical work points towards substantial industry/ rm and time variance of markups: e.g., Diewert and Fox, 2009; De Loecker, 2007, 2009; Botasso et al. 2001; Balk et al., 2009; Kort et al. (2009) Extensions: Bernard et al. (2003); Melitz and Ottaviano (2008) and Baldwin and Forselid (2006); Chen, Imbs and Schott (2009) Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level productivity December 2009 growth in3 the / 16 Dut

4 propose using production theory, a theoretically consistent framework a correct estimate of market power based on variable markups and adjusted for variable scale economies (Corsetti et al., 2007) Consistent TFP measure adjusted for markups Data: use an unbalanced panel production statistics of observations spanning over and 15 Dutch food sectors from the production statistics food TBT database further extension/update to a purpose-built database on TBTs at the level of CN 8-digits inclusion of Commission regulations assign weights to the technical importance by using forward citations (up to 2009) on counts (number of directives per product) Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level productivity December 2009 growth in4 the / 16 Dut

5 Markups, scale economies and TFP growth Markups We let each rm (industry) i 2 f1,...n) face the following production function for period t: y it = a it f i (x it ) i = 1, 2,... N ; t = 0,..., T, (1) where y it measures rm i s gross output, x it (x i1t, x i2t,..., x iji t) 0 denotes the vector of J i non-negative factor inputs, f i () is the core of the (di erentiable) production function and a it is TFP measured as Hicks-neutral disembodied technical change. We can write rm i s pro t optimization problem as follows: max p it (y it ) y it w y it, x it 0 x it : y it = a it f i (x it ) (2) it where w it (w i1t, w i2t,..., w iji t) 0 is the sector s vector of J i input prices, and p it (y it ) is the inverse demand function which represents the market price as a function of rm s output. Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level productivity December 2009 growth in5 the / 16 Dut

6 The rst order conditions implied by (2) yield the following equations: where c it are the marginal costs. p it (y it ) + p it(y it ) y it = c it and (3) y it p it (y it ) yit p it (y it ) + y it = w it. y it x it Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level productivity December 2009 growth in6 the / 16 Dut

7 For rm i the rst order condition (F.O.C.) implied by the solution of (3) can be rewritten as: p it (y it ) p it (y it ) c it = p it(y it ) y it y it p it, (4) or the Lerner index as a measure of a monopolist s market power is inversely related to the price elasticity of market demand, where ε it y it p it (y it ) p it (y it ) y it c it p it (y it ) price 1 and L it p it (y it ) (PCM). L it = 1 ε it, (5) is rm i s elasticity of demand with respect to is rm i s Lerner index or Price Cost Margin 1 The larger the elasticity of demand in absolute terms, the smaller the monopolistic rm s market power. ark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level productivity December 2009 growth in7 the / 16 Dut

8 From (4) and (5) we obtain that the second F.O.C. in (3) can be rewritten as: w it = y it x it p it (y it )[1 L it ], or for any individual input factor k ɛ J i : ln y it y it p it (y it )[1 L it ] = w ikt : k = 1, 2,..., J i ; t = 0,..., T (6) ln x ikt x ikt ln y it 1 w = ikt x ikt ln x ikt [1 L it ] y it p it (y it ) = µ it s ikt, where s ikt denotes the share of input k in the total production value of rm i, or s ikt w ikt x ikt / [y it p it (y it )], so that rm i s factor input share can be written as s it = J i s ikt = wit 0 x it/ [y it p it (y it )]. (7) k=1 Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level productivity December 2009 growth in8 the / 16 Dut

9 The time-varying markup in (??) can be rewritten as, where θ it J i k=1 y it x ikt x ikt y it µ it = y itp it (y it ) wit 0 x θ it (8) it = J i θ ikt. k=1 For simplicity, we assume that µ it = µ t ; equivalent to assuming that the rm i s elasticity of demand with respect to price do not vary substantially across rms within the same sector. Basus (2002); Diewert and Fox (2009), De Loecker (2007,2009), Botasso et al. (2001) Alternatively, interactions with other rm speci c variables (e.g., Konings et al. 2005; Bellone et al. 2005) In line with Basu and Fernald (2002), the correlation between economies of scale and markups, which is obviously also implied by (8), simply state that rms with increasing returns to scale, θ it > 1, must charge markups to cover their costs; on the hand, if pro ts are small, y it p it (y it ) wit 0 x 1, µ it it and θ it are similar. Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level productivity December 2009 growth in9 the / 16 Dut

10 Total Factor Productivity Growth We now have the full ingredients to de ne productivity, relating di erences in production between period t and some base period 0 as ln y it ln y it ln y i0 (9) ln f i (x it ) ln f i (x i0 ) + ln a it = 1 2 ln fi (x it ) + ln f i (x i0 ) [ln x ikt ln x ik0 ] ln x k ikt ln x ikt + ln a it Substituting the last equality of equation (6) in equation (9) yields: ln y it ln y it ln y i0 = 1 2 k [µ it s ikt + µ i0 s ik0 ] [ln x ikt ln x ik0 ] + ln a it, where the markup ratio, µ it, may vary over time. (10) Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level December productivity 2009 growth in 10 the / 16 Dut

11 Estimation of Scale Elasticities The translog assumption of the production form ts the index number approach developed in Nakajima et al. (1998, 2002, 2007) and Diewert and Fox (2009) for estimating the returns to scale parameter θ t. The authors derive the following simple equation, ln y it = tc t + θ t ln Q T (w i0, w it, x i0, x it ) + ε it, i = 1, 2,... N; t = 1, 2,.. (11) where Q T () is the Tornqvist aggregated input quantity index (k it, l it, and m it ), θ t is the estimate for the returns of scale parameter, tc t measures technical change, and ε it is the error term. Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level December productivity 2009 growth in 11 the / 16 Dut

12 Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level December productivity 2009 growth in 12 the / 16 Dut

13 Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level December productivity 2009 growth in 13 the / 16 Dut

14 EU harmonization of food regulations and productivity TFP it = Baseline z } { α 0 + T δ=t S t + σ=s δ t D TIME +ɛd EXP it +κµ st + π(µ st HARM it ) + ν it σ s D SECTOR s + τharm it (12) with i, s and t indexing rm, sector and period respectively. estimating parameters δ t, σ s, τ, ɛ, κ, π D TIME = time dummies, D SECTOR = sector dummies, Dit EXP =dummy if rm i = exports HARM it = export-citation weighted coverage ratio of TBT harmonization counts µ st = markup Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level December productivity 2009 growth in 14 the / 16 Dut

15 we propose a xed-e ect adjustment of the number of forward citation per regulation/directive. we do so by dividing the number of forward citations of each individual regulation/directive with the average number of forward citations of the same publication year. For each rm we than compute the average of the xed e ect adjusted number of forward citations of the regulations/directives In order to minimize further the problem of truncation, we restrict our analysis to directives/regulations issued up to the year 2005 with forward citations of the most recent directives/regulations Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level December productivity 2009 growth in 15 the / 16 Dut

16 Conclusion developed a framework where we integrate the impact of TBT liberalization on TFPG acounting for markups. Applying this analysis to thirteen industries in Dutch manufacturing and services for the period , we found that: A majority of Dutch food industries are characterized by an oligopolistic structure. The variation of market power within sectors over time is present When the adjustment of market power is made in addition to scale economies, the new TFPG rates vary according to both the competition measure and the industry Extension: nd an augmented measure of competition cleaned from a set of other alternatives (pro t elasticity; Her ndahl) link to an EU productivity database? Mark Vancauteren 1,2,3 ; Bruno Henry de Frahan EU 2 harmonization (Institute) of regulations, competition and rm-level December productivity 2009 growth in 16 the / 16 Dut