banque de france's workshop on granularity: claire lelarge's discussion, june 2016
TRANSCRIPT
Discussion of“Large Firm Dynamics and the Business Cycle”
by Vasco Carvalho and Basile Grassi
Claire Lelarge
RITM and CREST
A new quantification frameworkfor the granularity hypothesis
Paper motivation :I The main empirical contributions until now start from rather static
frameworks :I featuring firm level heterogeneity/ concentration of output among firmsI in which shocks are incorporated
I This paper takes a different route :I Starts from intrinsically dynamic framework : Hopenhayn 1992
Makes clear that the growth process of firms (“shocks”) and the firm sizedistribution (concentration of output among firms) are related.Also their response to shocks...(These aspects are lost in the existing empirical frameworks).
I also featuring firm level heterogeneityI and incorporating granularity (instead of the “continuum” assumption)
Content of the paper :I How does this model behave ?
I No stationary (firm size) distribution when granularity :Endogenous fluctuations in equilibrium.
I Under Gibrat’s law : gets complete characterization of aggregate dynamics
I Quantification exercises
Pushing Hopenhayn’s model to its limits
1. Assumption of competitive firms more difficult to sustain withgranularityboth on the output market and input market (no strategic interaction)"A mathematical model appropriate to the intuitive notion of perfectcompetition must contain infinitely many participants. We submit that themost natural for this purpose contains a continuum of participants"Nobel Prize laureate R. Aumann (Ecta, 1964)
2. Standard aspects of the Hopenhayn model not documented :granularity and overall industry structure ?e.g. comparative statics for entry cost, driving (long term) marketconcentration, productivity (size) dispersion, industry turnover.
3. Large labor demand elasticityThe Cobb-Douglas specification implies :
Ld (wt , ϕs) =(α.ϕs
wt
) 11−α
=⇒ ∂Ld∂wt
= −5Does it matter ?(generates fat tails but increases GE smoothing of shocks via wages)
Elasticity of Labor Demand
Fig. 1. Distribution of Labor Demand Elasticities.
Lichter, Andreas, Andreas Peichl and Sebastian Siegloch,
The Own-Wage Elasticity of Labor Demand: A Meta-Regression Analysis,
IZA DP No. 7958.
Quantification exercises
I Measurement aspects : what definition of firms ?I Analysis of the firm size distribution
I You seem to predict too few large firms, and too high FSD fluctuation ?I Argument that FSD is more stable for small firms hindered by too coarse
data binning ? (minor point)I Distinguish between finite number of draws from a stable (stationary)
distribution (e.g. measurement error)and fluctuation of the underlying distribution
I Impulse response functionsI Clarify what dynamics ? (convergence to the stationary equilibrium ?)I Minor point : why “average” marginal productivity of labor ? (p.34)
I Tail indices and dispersion of FSDI More loosely connected to your modelI Variance of sales very similar to herfindahl index
I Distributional dynamics over the business cycleI More like a test of specification (production function) ?
Measurement Aspects :What definition (“theory”) of a firm ?
Census data: domestic, public and private
Compustat data: international? public
Measurement Issues (1) : Double counts ?
Headquarter Listed units Simple public affiliates Complex public affiliates Private affiliates
Measurement Issues (2) : What concept of firm ?(What shocks ?)
Headquarter Listed units Simple public affiliates Complex public affiliates Private affiliates
Quantification exercises
I Measurement aspects : what definition of firms ?I Analysis of the firm size distribution
I You seem to predict too few large firms, and too high FSD fluctuation ?I Argument that FSD is more stable for small firms hindered by too coarse
data binning ? (minor point)I Distinguish between finite number of draws from a stable (stationary)
distribution (e.g. measurement error)and fluctuation of the underlying distribution
I Impulse response functionsI Clarify what dynamics ? (convergence to the stationary equilibrium ?)I Minor point : why “average” marginal productivity of labor ? (p.34)
I Tail indices and dispersion of FSDI More loosely connected to your modelI Variance of sales very similar to herfindahl index
I Distributional dynamics over the business cycleI More like a test of specification (production function) ?
Discussion of“Large Firm Dynamics and the Business Cycle”
by Vasco Carvalho and Basile Grassi
Claire Lelarge
RITM and CREST