Competition, Uncertainty, and Misallocation

Transcription

1 Competition, Uncertainty, and Misallocation Kaoru Hosono (Gakushuin University) Miho Takizawa (Toyo University) Kenta Yamanouchi (Keio University) Summer Workshop on Economic Theory August,

2 Composition 1. IntroducJon 2. Literature Review 3. Model and SimulaJon 4. Data and Methodology 5. EsJmaJon Results 6. Conclusion 2

3 1. Introduction CompeJJon Uncertainty MisallocaJon This paper invesjgates the effects of the degree of compejjon on the negajve impact of uncertainty on capital misallocajon across producers. Using a large dataset of manufacturing plants in Japan, we find that the impact of uncertainty is stronger when the 3 product market is compejjve.

4 Productivity Shock and Misallocation Under favorable producjvity shock, plants adjust their inputs to the opjmal amount. If adjustment is insufficient by some reasons, marginal revenue product of capital (MRPK) is deviated from rental rate (E to A, not B). If MRPK is dispersed across firms, reallocajon of capital from low MRPK firms to high would increase aggregate output. MRPK A Rental Rate E B Productivity shock K 4

5 -4-2 change in lnmrpk Changes of Productivity and MRPK change in productivity 1 2 If the amount of capital is adjusted immediately a]er the producjvity shock, MRPK is constant and equal to rental cost. But the changes of producjvity and MRPK are posijvely correlated, 5 so insufficient adjustment is implied.

6 Uncertainty and Misallocation Larger uncertainty deteriorates allocajve efficiency of capital across firms. 1. Time- to- build If capital adjustment requires Jme- to- build, uncertainty directly affects on the deviajon between actual and opjmal levels of capital. Asker, Collard- Wexler, and De Loecker (2014) 2. ReducJon of investment Many papers considers the effects of uncertainty on investment (e.g. Dixit and Pindyck (1994)). But all of those studies don t focus on the effects of allocajve efficiency. 6

7 Competition Matters The degree of compejjon is closely related to the effects of uncertainty on investment. While the fierce compejjon theorejcally mijgates the negajve effects of uncertainty, the results of empirical studies are mixed. In addijon, no studies link the effects of compejjon to uncertainty- misallocajon relajonship. CompeJJon macers for the effects of uncertainty on resource misallocajon. 7

8 Competition and Allocative Efficiency Some papers focus on the effects of compejjon on allocajve efficiency across producers. Olley and Pakes (1996) Collard- Wexler and De Loecker (2015) Edmond, Midrigan, and Xu (2015) Tougher compejjon promotes the reallocajon of resources from less producjve firms to producjve firms. In other words, producjvity determines the firm size strongly in a compejjve industry. Most of those papers assume immediate adjustment, but the effect of uncertainty is strong under costly adjustment and Jme- to build. The reducjon of uncertainty complements compejjon policies to improve allocajve efficiency. 8

9 This Paper This paper invesjgate the relajonship between the degree of product market compejjon and the adverse effects of uncertainty on resource misallocajon across plants. We conduct numerical simulajon for investment model and reduced form esjmajon, using a large panel dataset of manufacturing plants in Japan. The results show that the adverse effects of uncertainty are stronger for industries with tougher compejjon. The effects of uncertainty on investment The dependence of opjmal amount of capital on producjvity 9

10 Research Questions and Results 1. Does greater uncertainty worsen resource allocajon across plants in Japan? - Yes, uncertainty increases capital misallocajon in Japan. 2. How is the product market compejjon related to the adverse effects of uncertainty on allocajve efficiency? - The adverse effects of uncertainty are large for the industries with tougher compejjon. 3. How does the degree of compejjon affect on the adjustment process of capital? - The negajve effects of uncertainty on investment is stronger in compejjve industries along both extensive and intensive margin. 10

11 Composition 1. IntroducJon 2. Literature Review 3. Model and SimulaJon 4. Data and Methodology 5. EsJmaJon Results 6. Conclusion 11

12 1. Competition and uncertainty-investment relationship Uncertainty and investment A. Convexity of MRPK with respect to producjvity: posijve Uncertainty increases the expected return from investment by Jensen s inequality. Hartman (1972), Abel (1983) B. Real opjons theory: negajve It is becer to wait and avoid a costly disrupjon in a highly uncertain environment. McDonald and Siegel (1986), Pindyck (1988), Dixit and Pindyck (1994) C. Empirical studies: negajve Most but not all studies find negajve effects of uncertainty on investment. Bloom, Bond, and Van Reenen (2007), Bloom (2014) 12

13 1. Competition and uncertainty-investment relationship Theory A. Capital adjustment cost Both of imperfect compejjon and asymmetric capital adjustment costs are required for the delay of investment. Caballero (1991) B. OpJon exercise games OpJon values erode under fierce compejjon because compejtors may preempt the opportunity of investment. Williams (1993), KulaJlaka and Peroj (1998), Grenadier (1996, 2002) 13

14 1. Competition and uncertainty-investment relationship Evidence A. CompeJJon mijgates the negajve effects of uncertainty. Guiso and Parigi (1999): price- cost margin Bontempi, Golinelli, and Parigi (2010): price- cost margin Bulan (2005): industry concentrajon rajos Bulan, Mayer, and Somerville (2009): # of compejtors Akdogu and MacKay (2008): HHI B. NegaJve effect is stronger in compejjve industries. Ghosal and Loungani (1996, 2000): concentrajon rajo No studies focus on the effects of uncertainty on allocajve efficiency. 14

15 2. Resource Misallocation across producers Baseline framework A. Perfect compejjon Even in the U.S., the allocajon of inputs across plants are severely distorted. Restuccia and Rogerson (2008) B. MonopolisJc compejjon Compared with the U.S., the allocajve efficiency is very low in China and India. Hsieh and Klenow (2009) 15

16 2. Resource Misallocation across producers CompeJJon Increased compejjon reduces markup distorjon across plants. Edmond, Midrigan, and Xu (2015) Uncertainty Higher Jme- series volajlity contributes to larger cross- secjonal dispersion of MRPK. Asker, Collard- Wexler, and De Loecker (2014) The effects of compejjon and uncertainty are not simultaneously explored. 16

17 Composition 1. IntroducJon 2. Literature Review 3. Model and SimulaJon 4. Data and Methodology 5. EsJmaJon Results 6. Conclusion 17

18 Model Simple dynamic investment model Asker, Collard- Wexler, and De Loecker (2014) Demand ProducJon Sales Profit Q i = BP i ε Q i = A i K i α K L i α L M i α M, α K +α L +α M =1 β S i = Ω i K K β i L L β i M M # i, β X = 1 1 & % (α X $ ε ' # π (Ω it, K it ) = (β K +ε 1 ) β L % $ p L & ( ' β L # (β K +ε 1 ) β % M $ p M & ( ' β M (β K +ε 1 ) 1/(β K +ε 1) β Ωit K K /(β K +ε 1 ) it 18

19 Dynamic process Model K it = δk it 1 + I it 1 ω it = ρω it 1 +σκ it, κ it ~ N(0,1) Value funcjon V(Ω it, K it ) = maxπ (Ω it, K it ) C(I it, K it,ω it )+ β V(Ω it+1,δk it+1 + I it )ϕ(ω it+1 Ω it )d Ω it+1 I it Ω it+1 Capital adjustment cost C(I it, K it,ω it ) = I it +1 Iit >0 ( * )* { } C K " F+ π (Ω it, K it )+ C Q+ I K K it it $ # K it % ' & 2 + -,- +1 I it <0 ( * )* { } C K " F π (Ω it, K it )+ C Q K K it $ # I it K it % ' & 2 + -,- 19

20 Simulation MisallocaJon Misallocation = SD ln MRPK i ln MRPK i = ln(β K )+ s i k i Uncertainty Volatility = σ CompeJJon ε = 2, 4, 6 OpJmal amount of capital ( ) k it * = cons +εω it 20

21 Simulation result 1 Large uncertainty increases capital misallocajon. When the compejjon is tough, the effects of uncertainty are large. The amount of opjmal producjon strongly depends on producjvity in the case of tougher compejjon, so the volajlity of producjvity is largely reflected into large dispersion of MRPK. Symmetric costs Asymmetric costs SD(lnMRPK) SD(lnMRPK) volatility ε = 2 ε = 4 ε = volatility ε = 2 ε = 4 ε = 6 21

22 Simulation result 2 In the case of symmetric costs, the fracjon of posijve investment is high under large volajlity due to the Hartman- Abel effect. On the other hand, uncertainty reduces the fracjon of posijve investment when adjustment costs are asymmetric. Fraction of positive investment Symmetric costs Fraction of positive investment Asymmetric costs volatility ε = 2 ε = 4 ε = volatility ε = 2 ε = 4 ε = 6 22

23 Composition 1. IntroducJon 2. Literature Review 3. Model and SimulaJon 4. Data and Methodology 5. EsJmaJon Results 6. Conclusion 23

24 Data Source: Census of Manufacturing published by METI Coverage: All manufacturing plants with more than 30 employees located in Japan Period: Industry: 4 digit (491 industries) 24

25 Measurement (plant-level variables) Log- linearized producjon funcjon: s it = β Ks k it + β Ls l it + β Ms m it +η i + year t + ω it +ζ it ω it = ρ ω it 1 +ξ it ω it = η i + year t + ω it +ε it : TFPR Dynamic panel esjmajon: Blundell and Bond (1998, 2000) s it = π 1 k it + π 2 k it 1 + π 3 l it + π 4 l it 1 + π 5 m it + π 6 m it 1 + π 7 s it 1 +η i * + year t * +ω it ( ) = 0 E x it s Δω it E Δx it s ( η * i +ω it ) ( ) = 0 s 3 (Log of) MRPK: " MRPK it = ln S it $ # K it % ' = ln(β Ks )+ s it k it & 25

26 Measurement (industry-level variables) Uncertainty: Volatility st = SD st (ω it ω it 1 ) VolaJlity of producjvity shocks MisallocaJon: Misallocation st = SD st (MRPK it ) Dispersion in MRPK results in aggregate producjvity losses from the stajc view point. Hsieh and Klenow (2009) CompeJJon: Markup s = ε s ε s 1 = 1 β Ks + β Ls + β Ms Constant return to scale is imposed. We also check the method of De Loecker and Warzynski (2012). 26

27 Estimation method 1. Greater uncertainty reduces investment and results in stajc misallocajon. Misallocation st = βvolatility st + FE s + u st 2. The impact of uncertainty on misallocajon is weaker in the market where compejjon is severer. Misallocation st = γ 1 Volatility st +γ 2 Volatility st * Markup s + FE s + u st 3. The probability of investment and the investment rate is related to uncertainty and compejjon. 1 {I /K>0.05}it = λ 1 ln MRPK it + λ 2 Volatility st + λ 3 ln MRPK it *Volatility st + FE i + FE t + u it I it / K it = λ 1 ln MRPK it + λ 2 Volatility st + λ 3 ln MRPK it *Volatility st + FE i + FE t + u it 27

28 Summary statistics 28

29 Capital misallocation Capital misallocajon has increased within industry in Japan. 29

30 Change of misallocation by investment status MisallocaJon is reduced a]er investment among posijve investment plants. 30

31 Composition 1. IntroducJon 2. Literature Review 3. Model and SimulaJon 4. Data and Methodology 5. EsJmaJon Results 6. Conclusion 31

32 Volatility and misallocation The measures of volajlity and capital misallocajon are posijvely correlated. 32

33 Volatility and misallocation The effects of volajlity on misallocajon are stronger in small markup industries. 33

34 Estimation results 1 SD(MRPK) is posijvely correlated with volajlity measure. The effects of volajlity are strong in the industries with small markup. 34

35 Estimation results 2 The results are not changed even if the measures of volajlity and markup are changed. Volatility2 st = SD st (ω it ρω it 1 )! β Markup2 st = median# Ms " P Mst M it / S it $ & % 35

36 Estimation results 3 Large MRPK plants are more likely to conduct posijve investment. When uncertainty is large, the investment is less likely to be made even if MRPK is large. The negajve effects of uncertainty are not observed in the industries with tough compejjon. NegaJve investment are not affected by uncertainty. 36

37 Estimation results 4 Large MRPK plants tend to conduct more investment. Under large uncertainty, the investment rate is less affected by MRPK. The negajve effects of uncertainty are observed in the industries with tough compejjon. CompeJJon affects misallocajon both through the opjmal amount of capital and through the reducjon of investment. 37

38 Composition 1. IntroducJon 2. Literature Review 3. Model and SimulaJon 4. Data and Methodology 5. EsJmaJon Results 6. Conclusion 38

39 Summary This paper invesjgates the effect of compejjon to the uncertainty- misallocajon relajonship. SimulaJon results show that the negajve effects of uncertainty are severe in the industries with tougher compejjon. These results are confirmed in the esjmajon, using a large dataset of Japanese manufacturing plants. The negajve effects of uncertainty on investment are strong in the compejjve market. Some papers showed compejjon improves allocajve efficiency, but our results suggest compejjon induces fragility to uncertainty. 39

40 Future work While this study sheds new lights on the role of compejjon in the uncertainty- misallocajon relajonship, we have not yet explored the durability of uncertainty- driven misallocajon. If the major source of such misallocajon is Jme- to- build, then uncertainty- driven misallocajon may be short- lived. We find a relajvely small quanjtajve impact of volajlity, which may reflect the short- run effect of uncertainty. Mean(lnMRPK) Mean(lnMRPK) in next year Year Year zero positive zero positive negative negative 40