Banque nationale de Belgique SA · EURONEXT:BNB
Banque nationale de Belgique SA — Investor Presentation (February 2015)
36 slides from Banque nationale de Belgique SA's investor presentation, Feb 4, 2015
Slide 1: Multinational Firms and International Business Cycle Transmission
Multinational Firms and International Business Cycle Transmission Javier Cravino and Andrei A. Levchenko University of Michigan February 4, 2015
Slide 2: Question
Question Do multinational firms contribute to international comovement? I International business cycle comovement: significant and not fully-understood I Potentially large role of multinationals I Produce about 0.25 of world output I Can transfer technology across countries I Vertical production lin
Slide 3: New data and facts on comovement within multinationals
I- New data and facts on comovement within multinationals Data: I Firm level data from ORBIS, cross- and within-border ownership I Parents and foreign affiliates observed in the same dataset I 8 Million firms, 34 countries, 2004-2012 Findings: 1. Firm-level: strong correlation between parents and affili
Slide 4: Quantitative model to evaluate aggregate implications
II- Quantitative model to evaluate aggregate implications 1. Embed observed comovements into quantitative MP framework I Multinationals transmit technology shocks across countries I Shocks originated in the source country are important for affiliates’ productivity ' 20 40% 2. Measure contribution of m
Slide 5: Literature
Literature I Multinationals and technology transfers [McGrattan and Prescott (2009, 2010); Burstein and Monge-Naranjo (2009); Keller and Yeaple (2013); Ramondo and Rodríguez-Clare (2013); Ramondo (2014); Alviarez (2013); Fons Rosen et al. (2013) etc.] I Contribution: Estimate importance of parents’
Slide 6: Data
Data I ORBIS (Bureau van Dijk) I Data from business registries and annual reports I Both publicly listed and private firms I Manufacturing and non-manufacturing I 2004-2012 I Subsample of 34 countries with good coverage, mostly in Europe I Cross-firm ownership data I Multinationals >50% ownership
Slide 7: Country sample
Country sample Country Number of Number of Correlation Ratio of ORBIS Country Number of Number of Correlation Ratio of ORBIS Firms Multinationals between revenue to Firms Multinationals between revenue to ORBIS growth total revenue ORBIS growth total revenue and GDP and GDP growth growth Austria 15,
Slide 8: The importance of multinationals
The importance of multinationals I Account for a large share of revenue (1/4 at the median) I Larger than domestic firms
Slide 9: Largest firms
Largest firms I Large share of revenues concentrated in a few multinationals
Slide 10: Affiliate-parent correlations
Affiliate-parent correlations gin,t (f ) = f gii,t (f ) + āinss 0 ,t + ein,t (f ) I gin,t (f ): revenue growth rate of firm f I source i, destination n I gii,t (f ): growth rate of parent in i I āinss 0 ,t : source⇥sector⇥destination⇥sector⇥year FE I Sample: run on affiliates only
Slide 11: Affiliate-parent correlations
Affiliate-parent correlations All Manufacturing Services f 0.278*** 0.228*** 0.402*** 0.299*** 0.233*** 0.213*** (0.00524) (0.0117) (0.0137) (0.0394) (0.00628) (0.0131) Obs. 181978 181978 19756 19756 105774 105774 N. mult. 18881 18881 2470 2470 12419 12419 R2 0.047 0.724 0.102 0.789 0.032 0.674 FE No
Slide 12: Bilateral comovements
Bilateral comovements gin,t = si,t + dn,t + ain,t I gin,t : growth rate of combined sales of firms from i operating in n I si,t : source effect, common to all sales of firms from i worldwide I dn,t : destination effect, common to all sales in n
Slide 13: Bilateral comovements
Bilateral comovements Source Destination Part. R 2 F -stat. p-val. Part. R 2 F -stat. p-val. year-by year (2005-2012) Mean 0.10 2.54 0.002 0.19 7.41 0.000 Median 0.09 2.28 0.000 0.19 7.57 0.000 Pooled + in FE 0.10 6.82 0.000 0.17 8.40 0.000
Slide 14: Model
Model I Multi-country structure I Homogeneous final good, produced with multiple intermediate goods I MP model of productivity spillovers I Productivity of multinationals responds to shocks in source and destination markets I Aggregate productivity driven by productivity of all firms within the coun
Slide 15: Technologies
Technologies I Output of firm f : f 1 f Qin,t (f ) = Zin,t (f ) Lin,t (f ) = Zi ,t (f ) Zn,t (f ) Lin,t (f ) I Composite intermediate from i: " # r r 1 r 1 Qin,t = Â Qin,t (f ) r f 2⌦ i I Final good: " # r 1 r 1 r 1 Qn,t = Â Ain,t Qin,t r r i I Final good is freely traded, and set its price to 1: Pt
Slide 16: Preferences
Preferences I Utility: ✓ ◆ y0 ȳ u (Cn,t , Ln,t ) = Â d n Cn,t t L t ȳ n,t I Labor supply: 1 Wn,t ȳ 1 Ln,t = PtW where Wn,t is the wage.
Slide 17: Equilibrium
Equilibrium I Real wage " # r 11 Wn,t r 1 PtW = r   Ain,t Zin,t (f ) r 1 i f 2⌦i I Aggregate revenues: r  Pin,t Qin,t = PtW Qn,t = r 1 Wn,t Ln,t i I Aggregate output: " # ry 1 Qn,t =   Ain,t Zin,t (f )r 1 i f 2⌦i where y ⌘ ȳȳ 1 > 1
Slide 18: Aggregate growth rate
Aggregate growth rate I Revenue growth: ain,t gn,t =y   win,t (f ) r 1 + f zi ,t (f ) + (1 f ) zn,t (f ) i f 2⌦ i where win,t (f ) is firm f ’s revenue share. I Special case: zn,t (f ) = zn,t ain,t gn,t = y  win,t + f zi ,t + y (1 f ) zn,t i r 1 win,t (f ) where, zi,t = Âf 2⌦i win,t zi,t (f )
Slide 19: Interpreting affiliate-parent comovements
Interpreting affiliate-parent comovements I Firm f revenue growth in destinations n and i can be written as: gin,t (f ) = āin,t + (r 1) f zi ,t (f ) + (r 1) (1 f ) zn,t (f ) gii ,t (f ) = āii ,t + (r 1) zi ,t (f ) I Substituting: gin,t (f ) = ãin,t + f gii ,t (f ) + ein,t (f ) where āin,t ⌘ ãin,t
Slide 20: Interpreting source and destination specific shocks
Interpreting source and destination specific shocks I Under zn,t (f ) = zn,t bilateral sales growth is: gin,t = si ,t + dn,t + ain,t With I si ,t = f (r 1) zi ,t y +1 r r 1 Âi win,t [ain,t + f (r 1) zi ,t ] + y (1 f ) zn,t I dn,t =
Slide 21: Estimating phi with bilateral data
Estimating f with bilateral data I f enters relationship between source and destination shocks y y 1 f dn,t = 1 Â win,t [ain,t + si ,t ] + sn,t r 1 i r 1 f I We can write: ss ,t f = ss ,t + s t I s t combines destination and GE effects. s t = sd ,t in special case of y = r 1 I Intuition: f is relat
Slide 22: Estimating phi with bilateral data
Estimating f with bilateral data y y y 2 Year r 1 =1 r 1 =2 r 1 = 3 2005 0.470 0.552 0.375 2006 0.449 0.531 0.373 2007 0.390 0.472 0.319 2008 0.373 0.482 0.286 2009 0.395 0.532 0.294 2010 0.400 0.518 0.308 2011 0.379 0.491 0.289 2012 0.357 0.444 0.289 Mean 0.401 0.503 0.317 Median 0.392 0.505 0.301
Slide 23: Quantitative exercises
Quantitative exercises 1. Elasticity of growth in n to a shock in i (“impulse response”) 2. Correlations due to multinational activity (“model correlations”) 3. Counterfactuals: I Change multinationals shares or spillovers I Feed estimated structural shocks (zi ,t and ain,t )
Slide 24: Transmission of shocks across countries
Transmission of shocks across countries I Q: How does the UK respond to a shock that increases US output by 1%? I Elasticity of growth in n to a shock in i ∂ gn = y [win f + (1 f ) Ii=n ] ∂ zi I Relative to i: ∂ gn ∂ gi win f / = n 6= i. ∂ zi ∂ zi wii f + (1 f)
Slide 25: Response to shock that increases source country GDP by 1%
Response to shock that increases source country GDP by 1% 0.18 0.16 0.14 0.12 0.1 ∂γd /∂γs 0.08 0.06 0.04 0.02 0 IRL NLD SVK SGP USA BEL DEU CZE GBR AUT FRA GBR NLD POL JPN EST SWE SRB ITA SWE AUT BGR KOR FIN ESP NOR AUS ESP LTU AUS HUN SVN MEX HUN DEU GRC ROU CZE NOR BEL LTU FRA HRV HRV EST SVN UKR
Slide 26: Combined impact of all multinational activity
Combined impact of all multinational activity I Q: what would be the combined impact of a shock to all multinationals operating in the country? I Change in productivity will be: f (1 wnn )
Slide 27: Combined impact of all multinational activity
Combined impact of all multinational activity IRL NLD SVK SGP CZE BEL AUT GBR POL SRB EST BGR SWE MEX HUN HRV ESP DEU UKR ROU LTU LVA FRA AUS SVN NOR ITA PRT GRC FIN TUR KOR JPN 0 .1 .2 .3 .4 φ(1−ωnn)
Slide 28: Firm level
Firm level I Q: How does Ireland respond to a firm-level shock to that increases US output by 1%? I Elasticity of growth in n to a shock in i ∂ gn = ywin (f ) [f + (1 f ) Ii=n ] ∂ zi (f ) I Relative to i: ∂ gn ∂ gi win (f )f / = n 6= i. ∂ zi (f ) ∂ zi (f ) wii (f ) [f + (1 f )]
Slide 29: Response to firm-shock that increases source country GDP by 1%
Response to firm-shock that increases source country GDP by 1% 1 0.8 0.6 ∂γd /∂γs 0.4 0.2 0 EXXON METLIFE EADS IRL SAMSUNG NLD SUMITOMO SVK NIPPON T&T SGP TOYOTA BEL TOSHIBA CZE NISSAN AUT GBR MITSUI POL MITSUBISHI EST PANASONIC SRB MARUBENI SWE BGR ITOCHU ESP HITACHI AUS ENI LTU ENEL HUN MEX GLENCO
Slide 30: Counterfactual correlations
Counterfactual correlations I Q: what would be the correlation in output under the current levels of multinational activity if productivity shocks were uncorrelated? I The correlation between any pair of countries would be: ⇥ ⇤h sz,z 0 i sz,z 0 f (1 f ) wn0 n,t + wnn0 ,t + f 2 Âi win,t win0 ,t 1 sz2
Slide 31: Counterfactual correlations
Counterfactual correlations rn,n0 Mean St.Dev. Min Max d rn,n0 /d w Data 0.18 0.35 -0.68 0.87 2.27 Uncorrelated shocks 0.01 0.02 0.00 0.25 0.54 Correlated shocks 0.18 0.03 0.15 0.40 0.50 I Uncorrelated shocks: I max: US-Ireland (0.25), US-UK (0.12), US-Netherlands (0.12) I 95% of country-pairs under
Slide 32: Counterfactual dispersion in growth rates
Counterfactual dispersion in growth rates I Changing multinational shares: 1. (i) “No multinationals:” win NM ,t = 1 if i = n, win,t = 0 if i 6= n NM 1 N 2. (ii) “Full Integration:” win FI ,t = w̄i ,t = N Ân win,t FI 1 Snc,t = Â winc ,t si ,t N Â Â winc ,t si ,t i n i 1 Anc,t = Â winc ,t ain,t N Â Â
Slide 33: Counterfactual dispersion in growth rates
Counterfactual dispersion in growth rates Cross-sectional standard deviation in gn,t Data Model No Full C1/Model C2/Model Multinationals Integration 2005-2012 Mean 0.039 0.058 0.064 0.039 1.094 0.673 Median 0.040 0.060 0.066 0.039 1.087 0.654
Slide 34: Counterfactual dispersion in growth rates
Counterfactual dispersion in growth rates I Changing correlation in parent-affiliate growth f 2 [0, 1] : cf fc Sn,t = Sn,t , f cf An,t = An,t h i cf y cf y 1 f c own Dn,t = 1 An,t + Sn,t + Sn,t r 1 r 1 f
Slide 35: Counterfactual variances: correlation in parent-affiliate growth
Counterfactual variances: correlation in parent-affiliate growth .065 .06 Median stv .055 .05 0 .2 .4 .6 .8 1 phi
Slide 36: Taking stock
Taking stock 1. Documented strong comovements between parent’s and their foreign affiliates ' 20 40% 2. Limited contribution of multinationals for observed comovements I Small bilateral MP shares I Important for some country pairs (i.e. involving the US) 3. Can become an important channel as MP shares
Banque Nationale De Belgique — Investor Presentation (February 2015) on 6ix