Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Notes onLabor Share, Concentration, Markup
Investment, Intangibles
Vu T. Chau
Harvard University
May 7, 2019
1 / 20
Outline
Declining Labor Share
• Karabarbounis and Neiman (2014), Autor et al. (2018)
Markup and Concentration
• de Loecker and Eeckhout (2018), Gutierrez and Philippon
(2018)
Intangible Capital
• Crouzet and Eberly (2019)
2 / 20
Labor Share
3 / 20
Labor Share
Cost-minimization:
C (r ,w ,Y ) ≡ minK ,L
{wL+ rK } s.t. F (K , L) ≥ Y
First-order condition:
r = λFK , w = λFL
Note: The Lagrange multiplier is marginal cost:
λ = ∂C∂Y = MC (Envelope Theorem).
Labor share:
sL ≡wL
pY=λFLL
pY=
λ
p︸︷︷︸=1/µ
FLL
Y=
1
µ
FLL
Y
For Cobb-Douglas: FLL/Y = (1 − α), so sL = (1 − α)/µ
(constant).
4 / 20
Elasticity of substitution
If the production function is CES:
Y = F (K , L) =[(AKK )1−
1σ + (ALL)
1− 1σ
] σσ−1
Optimal factors demand:
lnrK
wL= (1 − σ) ln
r
w
where r ≡ r/AK , w ≡ w/AL.
Elasticity:d ln (rK/(wL))
d ln(r/w)= 1 − σ
Suppose capital got relatively cheaper (r/w ↓). The laborshare would decline (rK/(wL) ↑) if and only if σ > 1.
• Note: rKwL = rK/(pY )
wL/(pY ) = sKsL
. We have sK + sL + sΠ = 1. To
make the argument that sK/sL ↑ implies sL ↓, implicitly
assumed constant sΠ. (True under monopolistic competition.)
5 / 20
Is σ > 1?
Antras (2004): σ ∈ [0.5, 0.9], even when controlling for biased
technological change.
Oberfield and Raval (2018, WP): σ ≈ 0.7 for manufacturing.
• Estimate plant-level elasticity:
lnrKni
wLni= (σn − 1) lnwMSA
ni + controls + εni
• Assume r roughly constant across MSA (capital is perfectly
mobile).• Potential endogeneity: AL/AK correlates with w at MSA level→ solve by Bartik instruments.
• Aggregating plant-level elasticities:
σAGG = (1 − χ)σ+ χε, χ ≡∑ (αi − α)
2
α(1 − α)θi
6 / 20
Karabarbounis and Neiman (2014)
Identify σ from variation of relative price of investment goods
(ξ) in the cross-section of countries:
sL,j1 − sL,j
sL,j = γ+ (σ− 1)ξj + ui
ξ influences the rental rate of capital R: ξ = R.
7 / 20
Autor et al. (2017)
Document:1 The rise of “superstar firms” and rising concentration.2 Sectors with highest rise in concentration see largest decline in
labor share.3 Aggregate decline in labor share mostly due to re-allocation
between firms, not a general decline within.
Organizing framework: model with fixed overhead labor
sL =1
µ
FLL
Y+
wF
piYi=
1 − α
µ+
wF
piYi
A superstar firm can have lower sL because it has:
• higher markup µ.• high productivity Ai → high piYi → lower F/(piYi ).
8 / 20
Detour: Measuring Concentration
Measure 1: Herfindahl - Hirschman Index (HHI)
HHI ≡∑i
s2i
where si ≡ salei∑j salej
is the sale share of firm i .
• Value between 0 to 1 (or 0 to 10000, if share = 1 for 1%)• HHI = 1 corresponds to monopoly. HHI ≈ 0 corresponds to
perfect competition.
Measure 2: sale shares of top 4 / 8 / 20 firms (CR4, CR8,CR20).
• Often correlates with HHI, but not always.• Statistics readily available on Census’ website.
9 / 20
Detour: Concentration and Data Availability
Using Census data vs. Compustat:
Census has universe of firms. Compustat only publicly listed
firms.
Census only provides HHI for manufacturing (last time I
check), problematic if you do not have firm-level data.
Census only available for census years (1997, 2002, 2007 etc.)
• So it depends on at which frequency you want to run your
analysis.
10 / 20
Autor et al. (2017): Rising Concentration
11 / 20
Autor et al. (2017): Aggregate decline causedby reallocation
S =∑i
ωiSi = S +∑
(ωi −ω)(Si − S)
∆S = ∆S︸︷︷︸within
+∆[∑
(ωi −ω)(Si − S)]
︸ ︷︷ ︸between
12 / 20
Markup
13 / 20
De Loecker, Eeckhout, Unger (2018)
Recall FOC from cost-minimization for any variable input:
PVit = λit
∂Yit
∂Vit⇒ PV
it Vit
PitYit︸ ︷︷ ︸sVit
=λit
Pit︸︷︷︸=1/µit
∂Yit
∂Vit
Vit
Yit︸ ︷︷ ︸≡θVit
Thus, markup is given by:
µit =θit
sVit
Estimate θit a la Olley - Pakes (1996)
Calculate sVit using Compustat’s Cost of Goods Sold (COGS).
• Controversial
14 / 20
De Loecker, Eeckhout, Unger (2018)
15 / 20
Gutierrez and Philippon (2018)
Investment has been weak despite high Q.
Potential explanations:
• Financial frictions: external finance, bank dependence• Measurement errors: intangibles, globalization• Lack of competition: regulation, concentration due to other
factors• Tighter governance: ownership and shareholder activism.
Find evidence in support of declining competition hypothesis.
16 / 20
Gutierrez and Philippon (2018): Q
17 / 20
Gutierrez and Philippon (2018): BusinessDynamism
18 / 20
Gutierrez and Philippon (2018): RegressionResults
19 / 20
Eberly and Crouzet (2018)
20 / 20