Substitutability between U.S. Domestic and Imported Forest Products: The Armington Approach

1
Substitutability between U.S. Domestic and
Imported Forest ProductsThe Armington Approach

• Jianbang Gan
• Departments of Forest Science and Agricultural
  Economics
• Texas AM University

U.S.-Canada Forest Products Trade Symposium March
8, 2005, Michigan State University
2
Presentation Outline

• Motivation
• Armington elasticity
• Scope of this study
• Estimation methods
• Results
• Conclusions and implications

3
Motivation

• Armington elasticity reflects demand substitution
  between domestic and imported goods under the
  assumption that products are differentiated by
  country of origin.
• The product-differentiation model has been widely
  used in trade policy analysis. Yet its
  applications in forestry are limited.
• Armington elasticities for disaggregated forest
  product sectors/groups are unavailable. The
  elasticity estimates may provide new implications
  for international trade of forest products
  between the U.S. and other countries.
• Armington elasticity is an important parameter
  used in CGE modeling.
• Armington elasticity can be used to analyze
  border effects and homebias.

4
What is Armington Elasticity?

• The CES demand function for a composite good (C)
• where
• M demand for imported goods
• D demand for domestic goods
• s Armington elasticity (CES between domestic
  and imported goods, 0ltslt8)
• a, ß parameters

5
Scope of This Study

• Estimating Armington elasticities for U.S. wood
  and paper product sectors/groups
• Comparing short-run and long-run elasticities,
  and the elasticities at different product
  aggregation levels and
• Testing if the elasticity changes over time.

6
The CES Assumption
(a) CES consumption structure
(b) CES production structure
7
Estimating the Elasticity

• Based on both CES consumption and production
  structures, utility maximization or cost
  minimization gives rise to
• Logarithmic transformation of (2) leads to
• where
• ? s lnß/(1-ß)

8
Estimating the Elasticity (contd)

• Let y M/D and xpd/pm.
• Case 1 If x and y are both stationary, i.e.
  I(0), then s can be estimated directly using (3).
  
• To derive both short- and long-run elasticities,
  the following parsimonious lag model is used
• yt a0 a1xt a2yt-1 et
  (4)
• The short-run elasticity a1
• The long-run elasticity a1/(1- a2) if 0lt a2 lt1

9
Estimating the Elasticity (contd)

• Case 2 If x and y are I(1) and cointegrated,
  the following single-equation error-correction
  model is used
• ?yt a0 a1?xt a2yt-1 a3xt-1 et
  (5)
• The short-run elasticity a1
• The long-run elasticity - a3/a2

10
Estimating the Elasticity (contd)

• Case 3 If x and y are I(1) but are not
  cointegrated, or if either x or y is stationary,
  the following model is estimated
• ?yt a0 a1?xt et (6)
• The short-run elasticity a1
• The long-run elasticity cannot be estimated.

11
Testing for Elasticity Changes over Time

• Varying parameter models are used to test if the
  elasticity changes over time.
• A maximum likelihood method is used to identify
  the change point where the elasticity was most
  likely to change.
• The logarithm of the maximum likelihood
  function
• where , the standard deviation of
  regression residuals for sub-periods
• 1 and 2, respectively
• T1 K1, K2, , T-K-1 with K the number
  of regressors
• T the number of observations (years)

12
Data

• Yearly data covering the period from 1961-2002
• Variable measurements
• Imports (M)
• Domestic sales (D) Domestic Output Exports
• Import price (pm) Value of Imports/Volume of
  Imports
• Domestic price (pd) proxied by PPI.
• Data sources
• FAO FORSTAT
• Howard (2001) FPL-RP-595
• USDC
• USDA Foreign Agricultural Service
• US Department of Labor

13
Results

• Import shares in U.S. markets
• Estimates of Armington elasticities
• Testing results for possible elasticity changes
  over time

14
Import Shares for Logs
15
Import Shares for Lumber
16
Import Shares for Wood Panels
17
Import Shares for Pulp
18
Import Shares for Paper and Board
19
Estimated Armington Elasticities
Note Significant _at_ 1, 5, and 10.
20
Estimated Armington Elasticities (contd)
Notes Significant _at_ 1, 5, and 10. The
figures inside parentheses are elasticity
estimates for the first and second sub-periods.
21
Estimated Armington Elasticities (contd)
Notes Significant _at_ 1, 5, and 10. The
figures inside parentheses are elasticity
estimates for the first and second sub-periods.
22
Estimated Armington Elasticities (contd)
Notes Significant _at_ 1, 5, and 10. The
figures inside parentheses are elasticity
estimates for the first and second sub-periods.
23
Conclusions

• Imports are far from perfect substitutes for
  domestic wood and pulp/paper products.
• Aggregation matters.
• Long-run elasticities are larger than short-run
  elasticities.
• In general, imports have become more
  substitutable for domestic forest products over
  time.

24
Implications for U.S.-Canada Softwood Lumber Trade

• U.S. domestic and imported softwood lumbers, in
  general, are not perfect substitutes. (The
  estimated Armington elasticity is 0.153 for the
  short run and 0.622 for the long run)
• Trade policies that are intended to alter the
  relative price between U.S. domestic and imported
  softwood lumbers would have a limited impact on
  import share, particularly in the short run.
• Analyses that assume perfect substitution between
  domestic and imported lumbers or the law of one
  price may overestimate the impact.
• Different product group aggregations may lead to
  different results and arguments.

25
Thanks

• For additional information, please contact
• Jianbang Gan
• Department of Forest Science
• 2135 TAMU
• Texas AM University
• College Station, Texas 77843-2135, USA
• Tel. [phone]
• Fax [phone]
• Email j-gan_at_silva.tamu.edu