Real Exchange Rate and Export Behavior:
Firm Level Evidence
Anıl Dönmez
Refik Erzan
Tolga Umut Kuzubaş
Bogaziçi University
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
1 / 23
Motivation
Real exchange rate elasticity of exports for Turkey is estimates using
bilateral aggregate trade flows ranging from 0.2 to 1.1. (OECD
Economic Surveys-2013)
Asymmetric effects of real exchange rate movements at the firm level
Different sectors
Exporting to a number of destinations/single destination
Imported inputs
Different sizes/productivity
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
2 / 23
This Paper
Decompose aggregate export supply of Turkey to extensive and
intensive margins using firm level data
Extensive margin: Changes in export volume due to entry/exit
Intensive margin: Changes in the export volume due to existing
exporters
Estimate the effects of real exchange rate movements on these
margins
Determine the components of aggregate elasticity of exports relative
to real exchange rates
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
3 / 23
Theoretical Framework
Models the micro-decision of export market participation by
decomposing it to entry-exit and quantity decisions (Roberts and
Tybout (1997), Campa (2004))
Firms are heterogeneous and in each period choose to supply to the
export market given the fixed cost of entry to maximize discounted
value of profits
Conditional on being an exporter firm chooses the export quantity
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
4 / 23
Theoretical Framework
Vit = max Et
Iit ,Qit
X
∞
δ
j−t
Rij (Iij )
j=t
where the revenue function Rit is given by:
Rit (Iit ) = Iit πit (Qit , eit ) − Fi (1 − Ii,t−1 )
(
Iit = 0
Iit = 1
if do not export
if export
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
5 / 23
Theoretical Framework
Iit satisfies:
Vit = max Rit (Iit ) + δE Vi,t+1 | Iit
Iit
Firm chooses to export if:
π(Qit , eit ) + E Vi,t+1 | Iit = 1 − E Vi,t+1 | Iit = 0 > Fi − Fi Ii,t−1
Conditional on Iit = 1, firm chooses the export quantity
max π(Qit , eit ) | Iit = 1)
Qit
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
6 / 23
Theoretical Framework
Firms are heterogeneous, conditional on this heterogeneity
characterize the export market participation decision and aggregate
For a given level of exchange rate, expected export supply is given by:
Et (Qit ) = Q̂it Pr (Ii,t = 1)
Elasticity of export supply relative to the real exchange rate:
ηe =
Dönmez, Erzan, Kuzubaş
∂Pr (Ii,t = 1)
∂ Q̂it
∂Et (Qit )
=
Pr (Ii,t = 1) + Q̂it
∂eit
∂eit
∂eit
()
Real Exchange Rate and Export Behavior
7 / 23
Empirical Methodology
Reduced form estimation
Estimate exchange rate elasticity and decompose to intensive and
extensive margins
Estimate the export participation decision using a dynamic probit
model
Conditional on being an exporter, estimate the relationship between
real exchange rate and export volume
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
8 / 23
Data
Data is taken from Turkish Statistical Institute
Monthly bilateral trade data and annualy industry data covering the
period 2003-2013
Trade data includes:
Firm ID, year and month
Amount of import and export in terms of relevant measures (kg,
meters etc.)
Import and export value in TL, US Dollar and Euro terms
Destination country
Industry data includes:
Firm ID and year
NACE2 classification of products
Firm specific information (number of workers, weekly working hours
etc.)
Revenue and cost items in TL terms
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
9 / 23
Data
Real exchange rate (RER) data is taken from OECD database.
For exporting firms, individual RER is calculated by multiplying
destination country’s export share in that firm’s overall export value
with destination country’s RER to get the weighted RER.
For non-exporting firms, RER is the weighted average of the five
largest destination markets (Germany, France, Italy, England and
USA)
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
10 / 23
Firm Distribution
Table 3: Number of Total Firms by Industry and Size 2013 (Number of Employees=x)
Industry
Food
Beverage
Tobacco
Textile
Clothing
Leather
Wood Products
Paper Products
Press Printing
Refined Petroleum
Chemicals
Medicines
Plastics
Minerals
Base Metal
Fabrication Metal
Computer-Optic
Electrical Equipment
Unclassified Machineries
Motor Land Vehicles
Other Vehicles
Furniture
Other Manufacturing
Machine Setup
Dönmez, Erzan, Kuzubaş
()
x<100
3925
319
17
3136
5641
1125
1152
837
1105
166
1123
167
2325
2841
1080
4463
417
1606
3019
1062
403
2284
1240
1393
100≤x<500
416
18
9
579
604
52
48
93
34
4
85
37
246
368
156
365
38
151
208
161
31
162
54
47
Real Exchange Rate and Export Behavior
x>500
68
4
2
96
70
4
2
7
2
1
8
12
20
41
36
19
3
27
24
53
4
18
3
3
11 / 23
Firm Distribution
Table 4: Number of Exporters by Industry and Size 2013 (Number of Employees=x)
Industry
Food
Beverage
Tobacco
Textile
Clothing
Leather
Wood Products
Paper Products
Press Printing
Refined Petroleum
Chemicals
Medicines
Plastics
Minerals
Base Metal
Fabrication Metal
Computer-Optic
Electrical Equipment
Unclassified Machineries
Motor Land Vehicles
Other Vehicles
Furniture
Other Manufacturing
Machine Setup
Dönmez, Erzan, Kuzubaş
()
x<100
46
1
0
19
58
4
0
0
0
0
10
1
6
8
14
11
2
10
13
7
2
0
17
1
100≤x<500
54
0
7
31
93
4
1
7
1
0
11
2
19
15
45
31
1
27
20
18
20
0
10
1
Real Exchange Rate and Export Behavior
x>500
14
0
2
34
19
2
1
3
0
1
4
9
15
10
22
10
2
20
17
30
2
1
1
1
12 / 23
Exporting Firms
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
13 / 23
Total Sales: Exporting and Non-Exporting Firms
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
14 / 23
Average Productivity: Exporting and Non-Exporting Firms
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
15 / 23
Export Shares: Five Largest Destinations
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
16 / 23
Export Volume and Real Exchange Rate
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
17 / 23
Export Market Participation: Transition Rates
Table 1: Firm Transition Rates in Export Market
Exporter
export
no export
Non-Exporter
export
no export
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
-
0.847
0.153
0.791
0.209
0.799
0.201
0.805
0.195
0.808
0.192
0.680
0.320
0.815
0.285
0.834
0.166
0.711
0.289
0.747
0.253
-
0.019
0.981
0.019
0.981
0.019
0.981
0.021
0.979
0.021
0.979
0.008
0.992
0.013
0.987
0.016
0.984
0.004
0.996
0.005
0.995
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
18 / 23
Real Exchange Rate and Entry/Exit
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
19 / 23
Estimation
Estimate a dynamic probit panel model for participation using
Wooldridge(2002) or Heckman(1981) approaches to correct for the
endogeneity of initial conditions
Compute inverse Mills ratios for each time period
Estimate export volume equation including inverse Mills ratios to
correct for selection using pooled or panel regressions
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
20 / 23
Table 2: Export Volume Regressions
Variables
Real Exchange Rate
Productivity
Destination and year fixed effects are included
R-squared
Observations
Dönmez, Erzan, Kuzubaş
()
Pooled
0.8229***
(0.1231)
0.5704***
(0.0487)
Panel FE
0.6931***
(0.1435)
0.4843***
(0.0409)
0.19
12298
0.22
12298
Real Exchange Rate and Export Behavior
21 / 23
Table 3: Estimates of Export Market Participation
Variables
Lag of export status
Initial export status
Real Exchange Rate
Lag of Real Exchange Rate
Destination and year fixed effects are included
R-Squared
Observations
Pooled Probit
3.5877***
(0.0375)
0.9877***
(0.0526)
-9.0642***
(0.4118)
4.575***
Dynamic Probit
2.7952***
(0.0443)
3.3604***
(0.1079)
-11.1346***
(0.2502)
2.6799***
0.12
172948
0.09
172948
Table 4: Marginal Effects
Variables
Real Exchange Rate
Lag of Real Exchange Rate
Dönmez, Erzan, Kuzubaş
()
Pooled Probit
-0.16
0.17
Dynamic Probit
-0.09
0.03
Real Exchange Rate and Export Behavior
22 / 23
Conclusions and Future Work
The effect of real exchange rate movements on the export volume
through entry/exit is relatively small (need robustness checks)
Destination specific effects/the effect of diversification/exchange rate
volatility
The effect of imported inputs and relaxation of financial
constraints/international borrowing
Dönmez, Erzan, Kuzubaş
()
Real Exchange Rate and Export Behavior
23 / 23