Banque nationale de Belgique SA · EURONEXT:BNB
Banque nationale de Belgique SA — NBB Colloquium 2018
21 slides from Banque nationale de Belgique SA's conference presentation, Oct 25, 2018
Slide 1: Price Updating in Production Networks
Price Updating in Production Networks Cedric Duprez Glenn Magerman NBB ECARES, ULB NBB Colloquium October 25, 2018 Disclaimer: The opinions expressed are strictly those of the authors and do not necessarily reflect the views of the National Bank of Belgium. 1 / 21
Slide 2: How do firms change their prices?
How do firms change their prices? 8 6 Density 4 2 0 −1 −.5 0 .5 1 Output price changes I Yearly producer price changes, log differences (Belgium, 2002-2014) I Symmetric, centered around zero I Highly dispersed (p1/p99: ± .75) 2 / 21
Slide 3: How do firms change their prices? (demeaned)
How do firms change their prices? 1 Output price changes, demeaned .5 0 −.5 −1 −1 −.5 0 .5 1 Output price changes I Co-movement vs idiosyncrasies within 8-digit products I Pure co-movement: horizontal line, pure idiosync: 45 degree line I Large amounts of within-product heterogeneity 3 / 21
Slide 4: Motivation
Motivation I Why do firms adjust output prices? d ln p = d ln mc + d ln µ If d ln µ = 0, markups are constant or zero =⇒ complete cost pass-through I Empirical evidence for incomplete pass-through I Implying markup variability, d ln µ 6= 0 I Exchange rate pass-through, sector studies I Why is it imp
Slide 5: This paper
This paper I Questions 1. How do firms update prices? 2. How do producer price changes propagate and aggregate? I Contributions 1. Develop general, non-parametric framework of price changes 2. Construct change in firm-level input price index from observables 3. Estimation of cost pass-through in net
Slide 6: Related literature
Related literature I Theory on variable markups, incomplete pass-through Atkeson and Burstein (2008), Melitz and Ottaviano (2008), Weyl and Fabinger (2013), Atkin and Donaldson (2015), Edmond et al. (2015), Amiti et al. (2016), Parenti et al. (2017), Arkolakis and Morlacco (2018) I Empirical markups
Slide 7: Today
Today General framework of price updating Data on production networks and prices Empirical pass-through Propagation and aggregation 7 / 21
Slide 8: Pricing and markups
Pricing and markups I General pricing equation under cost minimization ln pjt = ln cjt ((1 + τ1j ) p1t , ..., (1 + τnj ) pnt , zjt ) + ln µjt (pjt , P−jt ) I τij : bilateral wedge (tariffs, transaction tax, transport costs) I zjt : efficiency I P −jt : price index of j’s environment (e.g. aggregate,
Slide 9: Price updating
Price updating I Total differentiation of pricing equation X ∂ ln cjt ∂ ln cjt d ln pjt = d ln pit + d ln zjt ∂ ln pit ∂ ln zjt i∈Sjt | {z } productivity shock | {z } total input price shock ∂ ln µjt ∂ ln µjt + d ln pjt + d ln P−jt ∂ ln pjt ∂ ln P−jt | {z } | {z } own price markup effect environment
Slide 10: Towards estimation equation
Towards estimation equation X d ln pjt = βjt ωijt−1 d ln pit +γjt d ln zjt + δjt d ln P−jt i∈Sjt | {z } ≡d ln Pjt where the coefficients have a clear interpretation 1 βjt = ∂ ln µjt 1− ∂ ln pjt ∂ ln y γjt = − ∂1ln µjt ∂ ln zjtjt 1− ∂ ln p jt 1 ∂ ln µjt δ jt = ∂ ln µjt ∂
Slide 11: Identification
Identification I Estimation equation to obtain parameters β, γ and δ d ln pjt = βd ln Pjt + γd ln zjt + δd ln P−jt I OLS estimates are biased I d ln pjt simultaneously determined with d ln Pjt (e.g. common price shocks, cyclicality of network) I d ln p simultaneously determined with d ln P jt −jt (e
Slide 12: Data construction
Data construction I Datasets (2002-2014) I Prodcom: firm, product (PC8), year, value, quantity, unit I Int’l trade: firm, product (CN8), country, year, value, quantity, unit I NBB B2B Transactions dataset: seller, buyer, year, value in Belgium I Change in domestic output prices d ln pjt I Obtained a
Slide 13: Data construction
Data construction I Observe all input shares mijt I Input share ωijt = P with mijt value of shipment i to j i∈Sj mijt I Correct for capital goods and re-exports 13 / 21
Slide 14: Cost pass-through
Cost pass-through OLS IV (i) (ii) (iii) (iv) (v) (vi) Dep. var. d ln pjt d ln pjt d ln pjt d ln pjt d ln pjt d ln pjt d ln Pjt 0.229*** 0.199*** 0.192*** 0.440*** 0.388*** 0.379*** (0.032) (0.029) (0.029) (0.056) (0.058) (0.047) d ln zjt -0.091*** -0.094*** -0.095*** -0.089*** -0.092*** -0.092*** (0
Slide 15: Propagation
Propagation I Price updating in production networks d ln p = [I − βΩ − δΘ]−1 dM where I β is the N × N diagonal matrix of pass-through rates I Ω is the N × N matrix of cost-based expenditure weights I δ is the N × N diagonal matrix of reaction to other prices I Θ is the N × N matrix of weights for t
Slide 16: Aggregation
Aggregation I Change in producer price index due to shock X d ln P = νj d ln pj (β, Ω, δ, Θ, dM) j I Ultimately a function of I Network structure of production at the firm level (heterogeneity) I Incomplete pass-through of cost shocks through production network I Responses to environment’s prices I
Slide 17: Conclusions
Conclusions I Takeaways I General, non-parametric model of price updating in production net- works I Cost pass-through is incomplete I Implications for change in aggregate price index I From there I GE counterfactuals (imports, TFP) I Heterogeneity in pass-through rates 17 / 21
Slide 18: Robustness – weighted regressions
Robustness – weighted regressions WLS IV (weighted) (i) (ii) (iii) (iv) (v) (vi) Dep. var. d ln pjkt d ln pjkt d ln pjkt d ln pjkt d ln pjkt d ln pjkt d ln Pjkt . 0 312*** . 0 280*** . 0 259*** . 0 529*** . 0 499*** . 0 462*** . (0 070) . (0 067) . (0 065) . (0 093) . (0 083) . (0 105) d ln zjkt -0.
Slide 19: Robustness – pass-through by sector
Robustness – pass-through by sector OLS IV NACE Rev.2 sectors N β γ δ N β γ δ 8-9 – Mining and quarrying 435 .363 -.75* .001 398 .933* -.050 .387 (.269) (.033) (.100) (.367) (.038) (.242) 10-12 – Manufacture of food products and beverages 6242 .360*** -.055*** .362*** 6,023 .340*** -.059*** .512***
Slide 20: Extension – multi-product firms
Extension – multi-product firms I Extension 1 – Model at firm-product level ln pjkt = ln cjkt ((1 + τ1j ) p1t , ..., (1 + τnj ) pnt , zjt )+ln µjkt (pjkt , P−jkt ) I Additional assumptions I A1: No physical synergies across products within producers I A2: Proportionality of inputs to outputs I Exten
Slide 21: Results – multi-product firms (firm-level)
Results – multi-product firms (firm-level) OLS IV (i) (ii) (iii) (iv) (v) (vi) Dep. var. d ln P̃jt d ln P̃jt d ln P̃jt d ln P̃jt d ln P̃jt d ln P̃jt d ln Pjkt 0.236*** 0.202*** 0.197*** 0.436*** 0.374*** 0.368*** (0.032) (0.030) (0.029) (0.055) (0.056) (.045) d ln zjt -0.087*** -0.090*** -0.092*** -