Do Taxes Matter for Firm Location? Evidence

Plant Scale and Exchange-RateInduced Productivity Growth
Loretta Fung
Department of Economics, University of Alberta
Jen Baggs
School of Business, University of Victoria
Eugene Beaulieu
Department of Economics, University of Calgary
Motivation

Question: how do exchange rate movements
affect shipments and productivity growth?

Exchange rate fluctuations affect the competitive
environment in which firms operate:



Home currency appreciation: home firms face a more
competitive environment.
Home currency depreciation: home firms face a less
competitive environment.
How does this affect firm-level productivity?
2
Antecedents


Growing empirical research on impact of exchange
rate movements on real variables:

Investment: Campa and Goldberg (1999), and
Harchaoui, Tarkhani and Yuan (2005).

Firm survival and size: Campbell and Lapham (2004)
and Baggs, Beaulieu and Fung (2008).

However, the exchange rate effects on firm
productivity is rarely explored.
Theory:


Bernard, Eaton, Jensen and Kortum (BEJK 2003)
Fung (2008)
3
Theoretical Background
Productivity Growth of Continuing Firms

Bernard, Eaton, Jensen and Kortum (2003): an
appreciation of the home currency causes factor
substitution and increases productivity.

Fung (2008): where the exit rates are low, an
appreciation of the home currency reduces firm
scale of production and lowers the productivity of
continuing firms (the scale effect).


Scale effect: Effect of the Canada-US FTA:
Prior to the FTA: Cox and Harris (1985) suggested that the
scale expansion caused by the FTA may improve
productivity.
Head and Ries (1999) and Trefler (2004): Little scale effect.
4
Canadian Context
In this paper, we study the exchange rate effects
on Canadian manufacturing plants:

Exchange Rates:
A large appreciation against the US dollar in the late 1980s and
early 1990s followed by a large depreciation in the rest of
the1990s.

Canadian manufacturers are both highly export
oriented and import competing.

Detailed, longitudinal, plant level data.
5
Hypotheses:
Two empirically testable hypotheses for surviving
plants:

An appreciation of the Canadian dollar reduces
plant shipments and this effect is more
pronounced for exporters.

In the presence of increasing returns to scale,
labour productivity should decrease with the
reduction of shipments.
6
Data: Plant-Level Data
Annual Survey of Manufactures (ASM) Database,
1987-1997.

Annual data on industry affiliation, location, plant affiliation to multiplant firms, shipments and value-added, materials, number of
production and non-production workers and total payments to them,
hours paid and hours worked for production workers, and ownership.

Export data are available in years 1990, 1993, 1996 and 1997 for
plants that filled out the long form (larger plants).

We only use the plants that have export data available in at least
one of the years (including non-exporters – i.e. zero exports).

This data set is ideal for examining productivity because it consists
of detailed data on plant production activity.

However, there is no capital measures in this data set.
7
Data: Exchange Rates

Owing to differences in the composition of trading
partners, it is possible that exchange rate changes may
affect industries to differing degrees.

Trade weighted real exchange rates (TWRER):
Weighted average of normalized real exchange rates of
Canada’s top 10 trading partners for each industry at the
4-digit SIC level.
9
160
140
120
100
80
60
(mean) twrer
Figure 1: Trade-Weighted Real Exchange Rates, 1984-1997
1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997
10
Table 2: Descriptive Statistics
Nonexporter
Significantly
Diff’t at the
5% level?
All Plants
Exporter
Number of Observations
189,653
112,399
77,254
Mean TWR exchange rate
100.57
100.07
101.30
YES
Mean Plant Age
20.74
22.16
18.67
YES
Mean Number of Employees
76.41
107.53
31.12
YES
Mean Skill Intensity
11.95
19.49
0.73
NO
Mean Labour Productivity
92,878
102,288
78,766
YES
Mean Canadian Tariff Reduction
(in year t)
0.66%
0.62%
0.70%
YES
Mean US Tariff Reduction
(in year t)
0.30%
0.29%
0.31%
YES
Mean Foreign Ownership
Dummy
0.16
0.21
0.08
YES
Mean Multi-Plant Dummy
0.29
0.36
0.18
YES
1.07%
1.43%
0.54%
Mean Annual Industry Shipments
Growth
YES 11
Empirical Analysis: Plant Scale

Plant Shipments:
 ln shipments ft   ln employment f 0  ( 1   k int eractions ft 1 ) ln ERit
 1tariffitCAN  2 tariffitUS   x ft 1   yit 1   ft .

f: plants; i: industry; t: year.
∆lnshipmentsft: change in log real manufacturing shipments.
lnemploymentf0: plant initial size.
∆lnERit: Change in log trade weighted real exchange rates.
Interactions: ∆lnER*Exportf and ∆lnER*Foreignf-1

∆tariffit = tariffit-1 – tariffit: tariff changes

xft-1: age, labour productivity,, export status (= 1 if exported at least once), ownership,
and affiliation with multiplant firms.

yit-1: industry concentration (CR4), industry sales growth, and real interest rates
(skipped in the tables in presentation).

Also include industry fixed effects (at the 3-digit SIC level) and trend.




12
Table 3: Change in Shipments, dependent variable: ln(real shipments)
Initial Employment
= ln(employees)0
Age
Years in the sample
Exchange Rate
= ln(twrer)
Exchange Rate*Export Dummy
(1)
-0.016**
(0.001)
-0.002**
(0.000)
-0.060*
(0.026)
Exchange Rate*Foreign Dummy
Reduction in Canadian Tariffs
Reduction in US Tariffs
Labour Productivity Growth
= ln(vam/pdworker)
Foreign Dummy
= 1 if foreign owned in year t-1
Multi-Plant Firm Dummy
= 1 if multi-plant firm in year t-1
Export Dummy
= 1 if the plant have exported in 1990, 93, 96 or 97
Estimation Method
Observations
R-squared
-0.336**
(0.070)
-0.924**
(0.175)
0.158**
(0.003)
0.003
(0.002)
0.008**
(0.002)
0.045**
(0.002)
OLS
150575
0.08
(4)
-0.016**
(0.001)
-0.002**
(0.000)
0.123**
(0.034)
-0.241**
(0.036)
-0.230**
(0.047)
-0.305**
(0.070)
-0.897**
(0.175)
-0.023**
(0.001)
0.158**
(0.003)
0.001
(0.002)
0.007**
(0.002)
OLS
150575
0.08
Robust standard errors (corrected for clustering at the plant level) in parentheses, + significant at 10%; * significant at 5%; ** significant at 1%
13
Empirical Analysis: Plant Scale
Summary of Results
 A real appreciation of the Canadian dollar reduces
plant shipments on average.

Plant Heterogeneity:


The negative average effect is driven by the shipment reduction
among exporters and foreign owned non-exporters.
The shipments of domestic non-exporters are positively affected
by appreciations of the Canadian dollar.
14
Empirical Analysis: Productivity


Exchange rate  Plant scale  Productivity?
Use 2SLS model:
First stage: shipments are instrumented using the exchange rate, export
status, industry sales growth, real interest rate (and terms interacting
exchange rate with export dummy and with foreign ownership dummy).
Second stage: estimating equation is specified as:
 ln productivity ft   ln( total shipments / employees) f 0   1  ln shipments ft   ln skill ft 
1 tariff itCAN   2 tariff itUS  x ft 1  y it 1   ft ,





∆lnproductivityft: change in labour productivity (log of real
manufacturing value-added per production worker).
ln(shipments/employees)f0: log of initial labour productivity
∆lnskillft: skill intensity (based on relative total payments).
xft-1: age, ownership, and multi-plant dummy.
yit-1: CR4.
15
TABLE 5 (Continued): Second Stage Regressions
Dependent Variable: ln(Labour Productivity)
(1)
(3)
0.200**
0.232**
ln(shipments)
(0.047)
(0.044)
Age
-0.000
0.000
Years in the sample
(0.000)
(0.000)
Initial Total Shipments per Employee
-0.051**
-0.050**
= ln(vst/employee)0
(0.003)
(0.003)
0.216**
0.218**
Skill Intensity
(0.005)
(0.005)
= ln(payments to non-production
workers/payments to production workers)
Reduction in Canadian Tariffs
0.117
0.133
= ctart-1 - ctart
(0.111)
(0.110)
Reduction in US Tariffs
1.141**
1.171**
= utart-1 - utart
(0.277)
(0.275)
Foreign Dummy
0.015**
0.015**
= 1 if foreign owned in year t-1
(0.003)
(0.003)
Multi-Plant Firm Dummy
0.021**
0.021**
= 1 if multi-plant firm in year t-1
(0.002)
(0.002)
Estimation Method
2SLS
2SLS
Observations
149905
149905
R-squared
0.109
0.115
Standard errors in parentheses.
Robust standard errors (corrected for clustering at the plant level) are reported at the second stage.
+ significant at 10%; * significant at 5%; ** significant at 1%
 denotes changes from year t-1 to t
17
Empirical Analysis: Productivity
Summary of Results
 Our findings suggest that, overall, a real appreciation of
the Canadian dollar reduces plant shipments and it leads
to a decrease in labour productivity. A depreciation has
the opposite effect.
 The negative average scale effect on productivity is
driven by exporters and foreign owned non-exporters.
 Our findings support the scale effect hypothesis.
18
Economic Significance

How large is the economic magnitude?



Average effect.
Effects on plants with different ownership and export market
participation status.
Method:


Benchmark: A hypothetical plant with all variables at means.
Calculate the changes in shipment growth rate and resultant
changes in productivity growth rate when large appreciation or
depreciation of the Canadian dollar is introduced.
19
Economic Significance

Average effect (all plants):
TABLE 7: Predicted Economic Effects -- Average Effect, All Plants
∆ln(Shipments)
∆ln(Labour Productivity)
At Mean Values
0.0164
0.0033
Changes in Exchange Rate
Large Appreciation (∆lnTWRER = 0.0681)
Large Depreciation (∆lnTWRER = -0.0710)
0.0082
0.0231
0.0017
0.0043
20
Economic Significance

Plant reaction to exchange rate movements
depend on their ownership and participation in
the export market.

Four groups of plants:
1.
2.
3.
4.
Domestic non-exporters,
Domestic exporters,
Foreign non-exporters, and
Foreign exporters.
21
TABLE 8: Predicted Economic Effects by Plant Groups
∆ln(Shipments)
∆ln(Labour Productivity)
Domestic Non-Exporters Only
At Mean Values
-0.0074
-0.0034
Changes in Exchange Rate
Large Appreciation (∆lnTWRER = 0.0681)
0.0021
-0.0017
Large Depreciation (∆lnTWRER = -0.0710)
Domestic Exporters Only
At Mean Values
Changes in Exchange Rate
Large Appreciation (∆lnTWRER = 0.0681)
Large Depreciation (∆lnTWRER = -0.0710)
Foreign Non-Exporters Only
At Mean Values
Changes in Exchange Rate
Large Appreciation (∆lnTWRER = 0.0681)
Large Depreciation (∆lnTWRER = -0.0710)
Foreign Exporters Only
At Mean Values
Changes in Exchange Rate
Large Appreciation (∆lnTWRER = 0.0681)
Large Depreciation (∆lnTWRER = -0.0710)
-0.0152
-0.0047
0.0328
0.0036
0.0205
0.0429
0.0015
0.0054
-0.0032
0.0118
-0.0145
0.0061
0.0099
0.0134
0.0370
0.0188
0.0039
0.0642
0.0131
0.0235
22
Conclusions

This paper investigates the effect of exchange rate
on productivity via plant scale of production.

Findings:

A real appreciation reduces the shipments of surviving plants
while a real depreciation has the opposite effect.


This effect is driven by adjustment of shipments by exporters.
The scale reduction induced by real appreciations has significant
impact on productivity.


The scale effect is economically significant.
The direction and magnitude of it depend on plant ownership and
export market participation.
23
Conclusions

Three elements of aggregate productivity growth:
1.
2.
3.

A real appreciation of the Canadian dollar affects
productivity in two opposite directions:



Growth of continuing plants,
Replacement of exits by entrants, and
Market share reallocations.
Surviving plants suffer from a reduction in the scale of production and
productivity, causing a reduction in industry-level productivity.
The exit of less productive firms shifts up the productivity distribution
within an industry, leading to an increase in industry-level productivity
(Baggs, Beaulieu and Fung, 2008).
Directions of future research:


To investigate the direction and magnitude of exchange rate effects on
aggregate productivity growth.
To assess the relative importance of the three elements.
24