sur 37
Affichage actuel
Competition and pass-through on international markets:
Firm-level evidence from VAT shocks
Philippe Andrade
Banque de France & CREM
Martine Carr ́e
LEDa-SDFi Universit ́e de Dauphine
Agn`es B ́enassy-Qu ́er ́e
CEPII
November 2010
Abstract
We exploit French customs’ firm-level data to study how exporters pass a change in the
VAT rate of one importing country through their prices. Looking at this reaction is of interest
beyond a mere fiscal incidence analysis: VAT shocks provide instruments to identifying the
extent of mark-up adjustment in exporting firms’ reaction to macroeconomic shocks. Indeed,
and by contrast with more usual shocks, typically exchange rate fluctuations, variations in VAT
rates leave the marginal cost of exporters unchanged. For a sample covering exportations to the
Euro-area over the 1996-2005 period our estimates show that, on average, markups contract by
50% of the VAT hike one year after the shock, implying a pass-through to final prices of 50%
of the shock as well. We furthermore exploit the individual dimension of the data to show that
the extent of mark-up adjustment varies across firms and markets. More precisely it depends
on the market power of an exporter, as measured by its relative market share, and competition
on a given market, as measured by standard concentration indexes. These evidences are in line
with a model featuring variable elasticity of demand and heterogeneous firms.
Keywords:
Export prices, markup adjustment, market share, market concentration, exten-
sive margin, Value-Added Tax.
JEL classification:
We thank Cl ́ement Carbonnier, Matthieu Crozet, Isabelle M ́ejean, Marc M ́elitz and Cyril Schwellnus as partic-
ipants in seminars at the IMF, U-Cergy, UCLA, UCSD, U-Cyprus and in the EEA2009, LAGV2010 and SED2010
conferences for useful comments. This paper does not necessarily reflect the views of the Banque de France. e-mails:
philippe.andrade@banque-france.fr
,
agnes.benassy@cepii.fr
,
martine.carre@dauphine.fr
.
1
1 Introduction
How do exporters pass macroeconomic shocks through their prices? The question is fundamental
to understand the international transmission of business cycles, or to characterize the effects of
macroeconomic policies in open economies. It has mostly been answered by looking at the reaction
of international prices to exchange rate fluctuations. This so called exchange rate pass-through
literature typically emphasizes small and long-lasting adjustments to these shocks
1
suggesting ex-
porters smooth the shock by adjusting their markup. However, it is not clear how much of these
results are driven by exchange rate peculiarities and therefore whether these conclusions extend to
other macroeconomic shocks. Indeed, exchange rates fluctuate a lot and it may be optimal not to
react to their transitory components. Moreover, exchange rates may affect the marginal cost of
an exporter through imported inputs so that the incomplete pass-through can occur even without
markup changes.
In this paper, we study whether incomplete pass-through holds for other shocks than exchange rate
fluctuations and to which extent it affects and relates to imperfect competition on international
markets. More precisely, we rely on a rich micro data set to look at exporter’s price reaction to
variations in the value-added tax (VAT) rates of the importing countries.
Focusing on the VAT rather than exchange rate bears three attractive features. First, a change
in VAT rates in a foreign country affects the consumer-price of a firm but leaves its marginal
cost constant. Thus incomplete pass-through can hardly be attributed to marginal cost adjustment
associated with imported inputs. Second VAT-rate changes are unfrequent, exogenous policy events
that are generally long-lasting (and perceived as very persistent). Any finding of incomplete pass-
through thus is unlikely to be related to perceptions that the shock will die out rapidly and to
transitory price rigidity issues. Finally, VAT applies in a symmetric way to imported and local
goods. Thus, incomplete pass-through cannot be attributed to any distortion created by the shock
in favor or against foreign suppliers and in particular non-traded local distribution services. In brief,
focusing on VAT shocks allows us to derive a measure of pass-through based on the sole markup
1
In an authoritative survey, Goldberg and Knetter (1997)) report a median (across studies) pass-through of 50%
one year after the shock. Goldberg and Campa (2005), Goldberg and Campa (2008) and Bouakez and Rebei (2008)
provide more recent evidence of incomplete pass-through.
2
adjustment and therefore this measure bears general implications for other types of macroeconomic
shocks.
We rely on the firm-level, French customs’ database that tracks volumes and prices of exported
goods at the firm’s level according to an 8-digit classification of goods (around 10,000 products),
between 1996 and 2005. We aim at identifying how each exporter adjusted its pricing decisions, for
each different good to each destination country, in reaction to VAT-rate shocks that occurred in the
eleven EMU12 partner countries of France over this period. Our identification strategy is derived
from the analysis of a stylized model of imperfect competition on international markets that features
heterogenous exporters and variable elasticity of demand. In this setup, the reaction of a firm’s
markup in response to a VAT shock depends on how its perceived demand-elasticity varies with
price, which in turn depends on the impact of the shock on the number of competitors on the shocked
market. Aside to the price reaction to a VAT shock at the level of a firm, the use of firm-level data
precisely allows us to estimate the change in competition on the export market, as measured by entry
and exit rates, hence the evolution of market shares. Subsequently, we are able to disentangle what
in the price/mark-up reaction to VAT shocks comes from changes in the strength of competition
and what comes from others sources in demand-curve elasticity variations. Furthermore, relying on
micro data, we can investigate how prices and markup adjustments vary across firms with different
market power and across different market with different degree of concentration.
We find that, at the firm level, the pass-through of VAT shocks to consumption prices is far from
complete. On our sample, we get an average (across sectors) estimate for the one-year horizon
pass-through of 50%. This incomplete degree of pass-through confirms the consensus view of
the literature, which also reports long-lasting, incomplete adjustment of international prices to
macroeconomic shocks. This result also stresses that incomplete pass-through due to may be a
very general pattern of international pricing strategies rather than just related to exchange rate
fluctuation features. Moreover as VAT shocks ties incomplete pass-through to markup adjustments,
this finding supports models featuring variable markups.
In addition to, and in relation with, incomplete pass-through and markup adjustments, we also
find that a VAT hike in a destination country modifies the conditions of competition on export
markets. We show that the VAT rate lowers the entry rate and increases the exit rate of exporters,
3
The rest of the paper is organized as follows. In Section 2 we describe the imperfect-competition
model on export markets and derive our pass-through identification strategy. Section 3 presents
the database. We provide and discuss the empirical results in Section 4. Section 5 concludes.
2 Optimal pass-through of VAT rate changes to export prices
2.1 Exporters’ optimal prices
Consider a firm
f
selling goods
k
= 1
, . . . , K
on several destination markets
l
= 1
, . . . , L
. Let
τ
l
be the VAT rate on destination market
l
,
Q
fkl
the quantity of good
k
sold by firm
f
on market
l
,
e
P
fkl
(
Q
) the (inverse) demand function for product
k
on market
l
and
C
fk
(
·
) the variable production
cost (including the VAT of the production country). The optimal price can be written as a markup
over marginal cost,
MC
fk
, namely
e
P
fkl
1 +
τ
l
1
1
fkl
=
MC
fk
(1)
where
MC
fk
∂C
fk
∂Q
fkl
and
fkl
e
P
fkl
∂Q
fkl
Q
fkl
e
P
fkl
the price-demand elasticity for good
k
produced by firm
f
and sold on market
l
, that we assume strictly greater than 1. It is important to remark that,
in contrast with the final price of the goods, the cost of production is not a function of the VAT
rate abroad. As a consequence, VAT rate changes abroad affect the marginal product but not the
marginal cost of a domestic exporter.
For notational convenience, let us index by
i
≡{
f, k, l
}
a good of type
k
produced by firm
f
and
sold on market
l
, and let
P
i
e
P
i
1+
τ
l
denote the corresponding optimal price net of VAT. As shown
in equation (1), this optimal price is the product of a markup over marginal cost. In our model,
the markup is specific to each unit
i
because the price-elasticity of demand depends on the target
market, on the firm and on the type of good. Let us denote by
μ
i
i
i
1
this markup. By
contrast, the marginal cost is specific to a firm and to a particular type of good but does not vary
across destinations. Using the log transformation, the optimal net price can be split into two terms,
namely:
log
P
i
= log
μ
i
+ log
MC
fk
Note that the fact that the marginal cost
MC
fk
does not depend on the destination market means
6








Figure 1: Distribution of sectoral elasticities
 
0
.05
.1
.15
.2
.25
Density
0
10
20
30
sigma_hs3
Source: Broda and Weinstein (2006).
32
Table 3: VAT rate changes between 1996 and 2005 – Euro-area
Countries
Standard rates
Reduced rates
Nb. of changes
Dates of changes
Nb. of changes
Dates of changes
Austria
0
0
Finland
0
1
1 Jul 1998
Germany
1
1 Apr 1998
0
Greece
1
1 Apr 2005
1
1 Apr 2005
Ireland
2
1 Jan 2001 and 1 Mar 2002
1
1 Jan 2003
Italy
1
1 Oct 1997
0
Luxembourg
0
0
Netherlands
1
1 Jan 2001
0
Portugal
4
5 Jun 2002 and 1 Jul 2005
1
Spain
0
0
Source
: European Commission (2008).
33
Table 4: Markup adjustment
(1) (2) (3)
VARIABLES ∆
p
it
p
it
p
it
VAT shocks
τ
slt
-1.824*** -2.138*** -1.989***
(0.385) (0.503) (0.498)
controls
p
it
1
-0.254*** -0.240*** -0.242***
(0.002) (0.003) (0.003)
log
s
it
1
-0.070*** -0.067*** 0.025***
(0.001) (0.001) (0.002)
log
s
klt
1
-0.003*** -0.004*** -0.004***
(0.001) (0.001) (0.001)
log
s
klt
1
0.032*** 0.031*** -0.009***
(0.002) (0.003) (0.003)
∆log
s
it
0.043***
(0.002)
∆log
s
klt
0.072***
(0.001)
y
lt
0.528*** 0.648*** 0.656***
(0.143) (0.162) (0.160)
π
lt
-0.138 -0.118 -0.176
(0.181) (0.201) (0.199)
y
c
lt
0.006*** 0.004*** -0.003***
(0.001) (0.001) (0.001)
n
l
ft
0.001*** 0.001*** 0.000
(0.000) (0.000) (0.000)
n
l
kt
0.001*** 0.001*** 0.000
(0.000) (0.000) (0.000)
.
.
.
Observations 648680 465035 465035
Number of id 181092 134135 134135
Notes
: ∆
p
it
: log price variation for firm
f
, product
k
and destination
l
; ∆
τ
slt
: VAT-rate variation in the
destination country, normalized by sector-demand elasticity in that destination (
b
sl
1)
1
∆log(1+
τ
lt
); controls
for destination country
l
:
y
lt
, real GDP growth rate,
π
lt
, consumer price inflation;
y
c
lt
, log of real GDP per
capita; controls for firm
f
:
n
l
ft
, number of destination countries;
n
k
ft
, number of exported products; controls
for market (
k,l
):
s
klt
, average market share among French exporters in the specific country
l
for the specific
product
k
;
s
klt
, minimum market share on market (
k,l
); controls for firm
f
on market (
k,l
):
s
it
, market share
of exporter
f
on market (
k,l
) relative to the minimum market share on that market; all regressions include
individual fixed effects and time dummies and are estimated using the Arellano & Bond procedure; numbers in
brackets are robust std errors of estimates;
∗∗∗
,
∗∗
,
indicates significance at the 1, 5 and 10% levels respectively.
34



Table 5: Entry and exit rates
(1) (2)
log[
r
(en
klt
)] log[
r
(ex
klt
)]
VAT shocks
τ
slt
-3.797*** 3.506***
(1.420) (1.255)
controls
log[
r
(ex
klt
1
)] 0.053***
(0.006)
log[
r
(ex
klt
1
)] 0.063***
(0.006)
s
klt
1
0.138*** 0.010***
(0.004) (0.004)
y
lt
-0.237 0.252
(0.379) (0.410)
π
lt
0.725 0.472
(0.462) (0.487)
y
c
lt
-0.209* 0.025
(0.120) (0.128)
n
l
ft
-0.041*** -0.058***
(0.003) (0.003)
n
k
ft
-0.002*** -0.001***
(0.000) (0.000)
.
.
.
Observations 89749 87036
Number of id 14679 14467
Notes
: en
klt
: entry rate on market (
k
,
l
); ex
klt
: exit rate on market (
k
,
l
) ∆
τ
slt
: VAT-rate variation in the
destination country, normalized by sector-demand elasticity in that destination (
b
sl
1)
1
∆log(1+
τ
lt
); controls
for destination country
l
:
y
lt
, real GDP growth rate,
π
lt
, consumer price inflation;
y
c
lt
, log of real GDP per
capita; controls for firm
f
:
n
l
ft
, number of destination countries;
n
k
ft
, number of exported products; controls
for market (
k,l
):
s
klt
, average market share among French exporters in the specific country
l
for the specific
product
k
;
s
klt
, minimum market share on market (
k,l
); all regressions include individual fixed effects and time
dummies; columns (1) to (4) are estimated using a within procedure; columns (5) and (6) are estimated using
the Arellano & Bond procedure; numbers in brackets are robust std errors of estimates;
∗∗∗
,
∗∗
,
indicates
significance at the 1, 5 and 10% levels respectively.
35



Table 7: Heterogeneous pass-through across sub-groups of exporters
(1) (2) (3) (4) (5) (6)
p
it
p
it
p
it
p
it
p
it
p
it
VAT shocks
τ
slt
-1.985***
(0.500)
τ
slt
×
Q
s
4
-1.165* 3.712** -0.163 -1.214 -3.844***
(0.629) (1.707) (1.346) (1.202) (1.042)
τ
slt
×
Q
s
3
-0.758 1.918 1.206 -3.839** -0.741
(0.821) (2.384) (1.875) (1.559) (1.300)
τ
slt
×
Q
s
2
-1.119 3.991 -0.606 -0.655 -3.446**
(0.984) (2.903) (2.255) (1.854) (1.528)
τ
slt
×
Q
s
1
-7.992*** -9.165* -13.607*** -4.903 -5.228**
(2.123) (4.883) (4.255) (3.249) (2.656)
controls
p
it
1
-0.241*** -0.241*** -0.246*** -0.241*** -0.237*** -0.240***
(0.003) (0.003) (0.006) (0.006) (0.006) (0.005)
log
s
it
1
-0.017*** -0.017*** -0.014*** -0.017*** -0.018*** -0.015***
(0.001) (0.001) (0.003) (0.003) (0.003) (0.003)
log
s
klt
1
-0.061*** -0.061*** -0.036*** -0.059*** -0.076*** -0.067***
(0.002) (0.002) (0.003) (0.003) (0.004) (0.003)
log
s
klt
1
0.106*** 0.105*** 0.061*** 0.102*** 0.129*** 0.117***
(0.004) (0.004) (0.007) (0.008) (0.010) (0.013)
∆log
s
it
0.039*** 0.039*** 0.024*** 0.035*** 0.047*** 0.046***
(0.001) (0.001) (0.002) (0.002) (0.002) (0.002)
∆log
s
klt
0.052*** 0.052*** 0.035*** 0.043*** 0.056*** 0.067***
(0.002) (0.002) (0.004) (0.005) (0.006) (0.008)
.
.
.
Observations 465035 465035 95806 108402 114343 143071
Number of id 134135 134135 29439 32160 33385 38251
Notes
: ∆
p
it
: log price variation for firm
f
, product
k
and destination
l
; ∆
τ
slt
: VAT-rate variation in the
destination country, normalized by sector-demand elasticity in that destination (
b
sl
1)
1
∆log(1 +
τ
lt
);
Q
s
1
to
Q
s
4
: dummies identifying which quartiles of relative market share (from lowest to highest) firms belong to;
controls for destination country
l
:
y
lt
, real GDP growth rate,
π
lt
, consumer price inflation;
y
c
lt
, log of real GDP
per capita; controls for firm
f
:
n
l
ft
, number of destination countries;
n
k
ft
, number of exported products; controls
for market (
k,l
):
s
klt
, average market share among French exporters in the specific country
l
for the specific
product
k
;
s
klt
, minimum market share on market (
k,l
); controls for firm
f
on market (
k,l
):
s
it
, market share
of exporter
f
on market (
k,l
) relative to the minimum market share on that market; all regressions include
individual fixed effects and time dummies and are estimated using the Arellano & Bond procedure; numbers in
brackets are robust std errors of estimates;
∗∗∗
,
∗∗
,
indicates significance at the 1, 5 and 10% levels respectively.
37