Lecture 9 Notes: The Heckscher-Ohlin Model (Cont.) - Theory, Part II

14.581 International Trade Class notes on 3/6/2013 1 1 1.1 Two-by-two-by-two Heckscher-Ohlin model Basic environment Results derived in previous lecture hold for small open economies – relative good prices were taken as exogenously given We now turn world economy with two countries, North and South We maintain the two-by-two HO assumptions: – there are two goods, g = 1,2, and two factors, k and l – identical technology around the world, yg = fg (kg ; lg ) – identical homothetic preferences around the world, dcg = g (p)I c Question What is the pattern of trade in this environment? 1.2 Strategy Start from Integrated Equilibrium competitive equilibrium that would prevail if both goods and factors were freely traded Consider Free Trade Equilibrium competitive equilibrium that prevails if goods are freely traded, but factors are not Ask: Can free trade equilibrium reproduce integrated equilibrium? If factor prices are equalized through trade, the answer is yes In this situation, one can then use homotheticity to go from di¤erences in factor endowments to pattern of trade 1 The notes are based on lecture slides with inclusion of important insights emphasized during the class. 1 1.3 Integrated equilibrium Integrated equilibrium corresponds to (p; !; y) such that: p = A0 (!) ! y = (p) (! 0 v) v = A (!) y (ZP ) : (GM ) : (F M ) : (1) (2) (3) where: –p (p1 ; p2 ), ! (w; r), A (!) [af g (!)], y (p) [ 1 (p) ; 2 (p)] – A (!) derives from cost-minimization – (p) derives from utility-maximization 1.4 (y1 ; y2 ), v (l; k), Free trade equilibrium Free trade equilibrium corresponds to (pt ; ! n ; ! s ; y n ; y s ) such that: (ZP ) : (GM ) : (F M ) : pt A0 (! c ) ! c for c = n; s yn + ys = pt (! n0 v n + ! s0 v s ) v c = A (! c ) y c for c = n; s (4) (5) (6) where (4) holds with equality if good is produced in country c De…nition Free trade equilibrium replicates integrated equilibrium if 9 (y n ; y s ) 0 such that (p; !; !; y n ; y s ) satisfy conditions (4)-(6) Two-by-two-by-two Heckscher-Ohlin model Factor Price Equalization (FPE) Set De…nition (v n ; v s ) are in the FPE set if 9 (y n ; y s ) tion (6) holds for ! n = ! s = !. 0 such that condi- Lemma If (v n ; v s ) is in the FPE set, then free trade equilibrium replicates integrated equilibrium Proof: By de…nition of the FPE set, 9 (y n ; y s ) c v = A (!) y 0 such that c So Condition (6) holds. Since v = v n + v s , this implies v = A (!) (y n + y s ) Combining this expression with condition (3), we obtain y n + y s = y. Since ! n0 v n + ! s0 v s = ! 0 v, Condition (5) holds as well. Finally, Condition (1) directly implies (4) holds. 2 1.5 Integrated equilibrium: graphical analysis Factor market clearing in the integrated equilibrium: v y2a2(ω) k a2(ω) y1a1(ω) a1(ω) O l 1.6 The “Parallelogram” FPE set (v n ; v s ) inside the parallelogram ls Os v y2a2(ω) ks vs kn a2(ω) vn y1a1(ω) a1(ω) On ln When v n and v s are inside the parallelogram, we say that they belong to the same diversi…cation cone This is a very di¤erent way of approaching FPE than FPE Theorem – Here, we have shown that there can be FPE i¤ factor endowments are not too dissimilar, whether or not there are no FIR – Instead of taking prices as given— whether or not they are consistent with integrated equilibrium— we take factor endowments as primitives 3 1.7 Heckscher-Ohlin Theorem: graphical analysis Suppose that (v n ; v s ) is in the FPE set HO Theorem In the free trade equilibrium, each country will export the good that uses its abundant factor intensively ls Os v ks s v kn vn C Slope = w/r On ln Outside the FPE set, additional technological and demand considerations matter (e.g. FIR or no FIR) Heckscher-Ohlin Theorem: alternative proof HO Theorem can also be derived using Rybczynski e¤ect: 1. Rybczynski theorem ) y2n =y1n > y2s =y1s for any p 2. Homotheticity ) cn2 =cn1 = cs2 =cs1 for any p 3. This implies pn2 =pn1 < p2s =p1s under autarky 4. Law of comparative advantage ) HO Theorem 1.8 Trade and inequality Predictions of HO and SS Theorems are often combined: – HO Theorem ) pn2 =pn1 < p2 =p1 < ps2 =ps1 – SS Theorem ) Moving from autarky to free trade, real return of abundant factor increases, whereas real return of scarce factor decreases – If North is skill-abundant relative to South, inequality increases in the North and decreases in the South So why may we observe a rise in inequality in the South in practice? – Southern countries are not moving from autarky to free trade 4 – Technology is not identical around the world – Preferences are not homothetic and identical around the world – There are more than two goods and two countries in the world 1.9 Trade volumes Let us de…ne trade volumes as the sum of exports plus imports Inside FPE set, iso-volume lines are parallel to diagonal (HKa p.23) – the further away from the diagonal, the larger the trade volumes – factor abundance rather than country size determines trade volume volumes ls Os ks y2a2(ω) kn a2(ω) y1a1(ω) a1(ω) On 4 :pdf ln If country size a¤ects trade volumes in practice, what should we infer? 2 2.1 High-Dimensional Predictions FPE (I): More factors than goods Suppose now that there are F factors and G goods By de…nition, (v n ; v s ) is in the FPE set if 9 (y n ; y s ) for c = n; s 0 s.t. v c = A (!) y c If F = G (“even case”), the situation is qualitatively similar If F > G, the FPE set will be “measure zero”: fvjv = A (!) y c for y c is a G-dimensional cone in F -dimensional space 5 0g Example: “Macro” model with 1 good and 2 factors ls Os ks vs kn vn a(ω) n O 2.2 ln FPE (II): More goods than factors If F < G, there will be indeterminacies in production, (y n ; y s ), and so, trade patterns, but FPE set will still have positive measure Example: 3 goods and 2 factors ls Os y1a1(ω) y2a2(ω) v ks s v y3a3(ω) kn a3(ω) vn a (ω) 2 a1(ω) n O ln By the way, are there more goods than factors in the world? 2.3 Stolper-Samuelson-type results (I): “Friends and Enemies” SS Theorem was derived by di¤erentiating zero-pro…t condition With an arbitrary number of goods and factors, we still have P pbg = f f g w bf where wf is the price of factor f and fg wf af g (!) =cg (!) Now suppose that pbg0 > 0, whereas pbg = 0 for all g 6= g0 Equation (7) immediately implies the existence of f1 and f2 s.t. w bf1 w bf2 pbg0 > pbg = 0 for all g 6= g0 , < pbg = 0 < pbg0 for all g = 6 g0 . 6 (7) So every good is “friend”to some factor and “enemy”to some other (Jones and Scheinkman 1977) 2.4 Stolper-Samuelson-type results (II): Correlations Ethier (1984) also provides the following variation of SS Theorem If good prices change from p to p0 , then the associated change in factor prices, ! 0 !, must satisfy (! 0 !) A (! 0 ) (p0 p) > 0, for some ! 0 between ! and ! 0 Proof: De…ne f (!) = !A (!) (p0 p). Mean value theorem implies f (! 0 ) = !A (!) (p0 p) + (! 0 !) [A (! 0 ) + ! 0 dA (! 0 )] (p0 p) for some ! 0 between ! and ! 0 . Cost-minimization at ! 0 requires ! 0 dA (! 0 ) = 0 Combining the two previous expressions, we obtain f (! 0 ) f (!) = (! 0 !) A (! 0 ) (p0 p) From zero pro…t condition, we know that p = !A (!) and p0 = ! 0 A (! 0 ). Thus f (! 0 ) f (!) = (p0 p) (p0 p) > 0 The last two expressions imply (! 0 !) A (! 0 ) (p0 p) > 0 Interpretation: Tendency for changes in good prices to be accompanied by raises in prices of factors used intensively in goods whose prices have gone up What is ! 0 ? 7 2.5 Rybczynski-type results Rybczynski Theorem was derived by di¤erentiating the factor market clearing condition If G = F > 2, same logic implies that increase in endowment of one factor decreases output of one good and increases output of another (Jones and Scheinkman 1977) If G < F , increase in endowment of one factor may increase output of all goods (Ricardo-Viner) In this case, we still have the following correlation (Ethier 1984) (v 0 v) A (!) (y 0 y) = (v 0 v) (v 0 v) > 0 If G > F , inderteminacies in production imply that we cannot predict changes in output vectors 2.6 Heckscher-Ohlin-type results Since HO Theorem derives from Rybczynski e¤ect + homotheticity, problems of generalization in the case G < F and F > G carry over to the Heckscher-Ohlin Theorem If G = F > 2, we can invert the factor market clearing condition yc = A 1 (!) v c By homotheticity, the vector of consumption in country c satis…es dc = sc d where sc c’s share of world income, and d world consumption Good and factor market clearing requires 1 d=y=A (!) v Combining the previous expressions, we get net exports tc yc dc = A 8 1 (!) (v c sc v) 2.7 Heckscher-Ohlin-Vanek Theorem Without assuming that G = F , we can still derive sharp predictions if we focus on the factor content of trade rather than commodity trade We de…ne the net exports of factor f by country c as P c c f = g af g (!) tg In matrix terms, this can be rearranged as c = A (!) tc HOV Theorem In any country c, net exports of factors satisfy c = vc sc v So countries should export the factors in which they are abundant compared to the world: vfc > sc vf Assumptions of HOV Theorem are extremely strong: identical technology, FPE, homotheticity – One shouldn’t be too surprised if it performs miserably in practice... 3 3.1 Quantitative Issues Basic Idea Stolper-Samuelson o¤ers sharp insights about distributional consequences of international trade, but... – Theoretical insights are only qualitative – Theoretical insights crucially rely on 2 2 assumptions Alternatively one may want to know the quantitative importance of international trade: – Given the amount of trade that we actually observe in the data, how large are the e¤ects of international trade on the skill premium? – In a country like the United States, how much higher or smaller would the skill premium be in the absence of trade? 9 3.2 Eaton and Kortum (2002) Revisited Eaton and Kortum (2002)— as well as other gravity models— o¤er a simple starting point to think about these issues Consider multi-sector-multi-factor EK (e.g. Chor JIE 2010) – many varieties with di¤erent productivity levels z (!) in each sector s – same factor intensity across varieties within sectors – di¤erent factor intensities across sectors Unit costs of production in country i and sector s are proportional to: ci;s = where: – wiH , wiL – 3.3 h H s wiH 1 + L s wiL 1 i1=(1 ) (8) wages of skilled and unskilled workers. elasticity of substitution between skilled and unskilled Dekle, Eaton, and Kortum (2008) Revisited Suppose, like in EK, that productivity draws across varieties within sectors are independently drawn from a Fréchet Then one can show that the following gravity equation holds: where Ej;s Ti ( ij;s ci;s ) s Xij;s = Pn l=1 Tl ( lj;s cl;s ) s Ej;s , (9) total expenditure on goods from sector s in country j Two key equations, (8) and (9), are CES: – One can use DEK’s strategy to do welfare and counterfactual analysis – But one can also discuss the consequences of changes in variable trade costs, lj;s , or technology, Ti , on skill premium – How large are GT compared to distributional consequences? 10