Economia
The Case of Missing Trade and Other Mysteries Daniel Trefler The American Economic Review (Dec. 1995)
Assumptions of HO Model
2x2x2 model with homogeneous goods and factors; Identical CRS technologies with no FIRs; Perfect competition, full employment, absence of distortions and externalities; Identical homotheticpreferences. Above are sufficient for FPE and prediction of direction of commodity trade. Problem: with multiple goods the direction of commodity trade is indeterminate.
HOV Generalization
Same assumptions but general number of factors and goods, with goods at least equal to factors. FPE is assured. HOV states that a capital-abundant country net exports capital services. Generally, a ranking of countriesby factor net exports scaled by factor use reveals factor endowments (Maskus, JIE 1985; BLS, AER 1987).
Structure of Trefler paper
Specify general HOV theorem; Test HOV theorem against statistical alternatives; Test modifications of HOV, allowing home bias in consumption and international technology differences.
Specifying HOV theorem
Notation c = 1, …….., C index countries f = 1,…….., F index factors Vfc = endowment of factor f in country c Vfw = cVfc = world factor endowment Ffc = factor content of net export: (F1c , …… , FFc)’ = A*Tc Tc = the vector of net commodity exports A = “technology matrix” ie the amount of each factor needed to produce one unit of each commodity Bc = the trade balance in country c = p’Tc Yc = GDP in country c = p’Qc Yw = cYc sc = (Yc - Bc)/Yw =consumption share of country c “HOV equation” (Leamer, JPE 1980) (1) Ffc = Vfc - sc*Vfw LHS gives the factor content of net exports; RHS gives measured net factor endowments, or domestic minus consumption-adjusted global endowments (if positive, the economy is abundant in factor f). This system of equations (f x c) could be "tested" by considering the sign matches, rank correlations, and actual fit.Trefler wants to test the fit against alternatives.
Data
All data are from 1983. There are 33 countries. They account for 76% of world trade and 79% of world GDP. There are nine factors: capital, cropland, pasture, and six categories of labor.
Basic HOV Tests
“Sign HOV” test represents the percentage of observations for which Ffc and Vfc - sc*Vfw have the same sign. Weighted “sign HOV” .Weight sign statistics by factor contents (large flows get greater weight): | Ffc| / f,c|Ffc|.
Results:
“HOV equation” Cor(Ffc, (Vfc - sc*Vfw)) = 0.28 “Sign HOV” 48.9% Weighted “sign HOV” 71%
Missing Trade
(2) fc = Ffc – (Vfc - sc*Vfw) If HOV holds, these residuals should fluctuate around zero. But see Figure 1, which shows that fc = – (Vfc -sc*Vfw), or Ffc = 0, approximately. This is"missing trade", meaning that computed factor contents of trade are far less than net endowments.
Endowment Paradox
A further initial finding is that deviations from HOV are not random. Figure 2 shows that poor countries have large numbers of abundant factors and rich countries have large numbers of scarce factors. Poor countries "underperform" their endowments (in the sense that factorconsumption is below endowments) and rich countries "overperform".Suggests a technology-based explanation.
Economically Meaningful Alternative Hypotheses
1 Modify Technology assumption
2 Modify Consumption assumption
Modify Technology assumption
FUSfc = fc*Vfc - scjfj*Vfj fc is the productivity of factor f in country c relative to US productivity. FUSc = (FUS1c , ……….. , FUSFc)' =Aus*Tc is factor content of c's net trade using US matrix. First approach: Hicks-Neutral across factors; varies by country (T1) fc = c for all f and c. c is Hicks-neutral factor augmenting productivity measure.US = 1. FUSfc = c*Vfc - scjc *Vfj Second approach: Factor-biased across two country groups; varies by factor (rejected by data) fc = 1 for rich; = f for developing;
Estimate...
Regístrate para leer el documento completo.