Transcript Slide 1
Today • Chapter 4 extensions 4.6 - 4.8 • Chapter 5 Krugman & Venables (1995-1996) • intermediate inputs • labor mobile between sectors but not between regions • firms use M-products (µ) and M-labor (1-µ) • also known as the Vertical Linkages (VL) model • base model of Chapter 3 is usually named the CP model (Core-Periphery) The VL model in writing • U = F1-δ Mδ N r M = ci i=1 1 r (1- e ) I = pi i =1 N • cj = pj -ε Iε-1 E 1 (1- e ) • total spending on M-products in stead of δY is now E = δY + μ npx value of all varieties produced Supply side • mark-up pricing in core model was p = βW/ρ normalization β = ρ -> p = W • now becomes p = Iµ W(1-µ) • zero profit condition px = Iµ W(1-µ) ( α + βx) • x = α(ε-1)/β = αε with • Food sector: – CRS: – DRS: F(1-λ) = 1-λ F'(1-λ) >0 ; F''(1-λ) < 0 • Consumer income = (M) wage income + output food sector: Y = Wλ + F(1-λ) Intersector mobility • price index is the same for F workers and Mworkers • for mobility between sectors only the nominal wage matters • dλ/λ = η [W - F'(1-λ) ] (4.14) • same as in core model: equal demand and supply leads to Regional wages in VL model Supply x1 = demand in region 1 + demand in region 2 + extra production melted away • α (ε -1)/β = (E1 p1 -ε I1ε-1 + E2 p1 -ε T 1-ε I1ε-1 ) leads to • W1 = (1- β)/α)1/ε(1-µ) I1-µ/1-µ (E1I1ε-1 + E2 T1-ε I2ε-1)1/ε (1-µ) • W2 = (1- β)/α)1/ε(1-µ) I2-µ/1-µ (E2I2ε-1 + E1 T1-ε I1ε-1)1/ε (1-µ) • simplifies to core model when µ = 0 • main differences: – E in stead of Y – Extra term I-µ/1-µ : supplier access effect : closer to suppliers lowers price index and can give higher nominal wages. The four forces in the LV model • extent of competition effect: a higher λ lowers the price index of all other products (-) • market size or home market effect: a higher λ increases the market (+) • (new) access supplier effect: a higher λ increases nominal wages (+) • (new) marginal productivity effect in food sector (-) – only with DRS: a higher λ increases food wages The VL model with DRS in the food sector T=1.5 W1/W2 region 2 B B stable equilibrium region 1 share of M-workers in region 1 The VL model with DRS in the food sector T=1.3 W1/W2 region 1 B B unstable equilibrium region 2 share of M-workers in region 1 The VL model with DRS in the food sector T=1.1 W1/W2 region 2 B B stable equilibrium region 1 share of M-workers in region 1 The VL model with DRS in the food sector Fig 4.10 The bell-shaped cirve 1 λ1 0,5 0 Unstable equilibria T Stable equilibria VL model: with lowering T from dispersion to agglomeration to dispersion Fig 4.3 The Tomahawk diagram 1 λ1 0,5 0 Unstable equilibria T Stable equilibria CP model: no (increasing) spreading force when T becomes low Next chapter: Helpman(1998) also gets a bell-shaped curve by introducing the housing market as a spreading force The generalized model • Puga (1999) not discussed here in detail • CP model plus µ (intermediate production) ηs (intersector migration) and ηr (interregional migration) • Only the model with ηr =0 (no interregional migration) gives the bell-shape curve The Footloose Entrepreneur (FE) model • two labor production factors in stead of one: • skilled/unskilled ; human capital/labour ; R&D/production; headquarters/plants • Skilled labor is mobile, unskilled labor immobile • In production function: – α: skilled labor as fixed costs – β: unskilled labor as variable costs • makes the model solvable because the mobile skilled labor demand is not a function of x • equation (4.25) for skilled labor wage rate r1/r2 • discussion on the FE model will come back later Chapter 5 Agglomeration, the home market effect and spatial wages Terminology (confusing!) • Concentration: – industry is concentrated in some regions (xri /xni ) / ( xr /xn ) >1 • Specialization: – region is specialized in some industries (xri /xr ) / ( xin /xn ) > 1 – (xri /xr ) / ( xin /xn ) also know as the location coeffcient • is the same : (xri /xni ) / ( xr /xn ) = (xri /xr ) / ( xni /xn ) Concentration=Specialization • The distinction between concentration and specialization is not relevant for one spatial level. It is only done to be consistent with trade theory terminology: – specialization=concentration at the country level • Agglomeration: – concentration of more than one industry Wrong terminology: • There is more car production in Germany than in The Netherlands or: concentration -> Eir / Ein ≠ Eis / Ein • concentration is relative not absolute No concentration, no agglomeration Concentration, no agglomeration No concentration, agglomeration concentration, agglomeration Industry 1 Industry 2 a. Neither specialization, concentration nor agglomeration b. Specialization (=country concentration), no agglomeration Country A Country B Industry 1 Industry 2 c. regional concentration, specialization, no agglomeration (??) in terms of ch 3-4 this was called agglomeration d. Concentration and agglomeration, no specialization here agglomeration is about more industries Country A Country B Industry 1 Industry 2 d. Concentration and agglomeration, no specialization e. Concentration, agglomeration and specialization Country A Country B Concentration manufacturing <1 Below national average > 1 Above national average (Eir / Er)/( Ein / En ) Absolute size manufacturing Eir Concentration bussiness services < 1 Below national average > 1Above national average (Eir / Er)/( Ein / En ) Absolute size bussiness services Eir Convergence • • • • Increase/decline of gdp/cap differences Barro & Sala-i-Martin and others: Global no;within EU yes,but Results for EU: – 1980-1990 convergence – Later: divergence – Depends on level of region disaggregation EU 1995-2001 Nuts2 1995 1 Gini coefficient: 0.1561 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 EU 1995-2001 Nuts2 2001 1 Gini coefficient: 0.1539 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 EU 1995-2001 Nuts3 1995 1.0 Gini coefficient: 0.2109 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 EU 1995-2001 Nuts 3 2001 1.0 Gini coefficient: 0.2118 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Figure 5.2 Krugman specialization index Ireland Greece Finland Denmark Portugal Netherlands Sweden 1994-1997 Average 1980-1983 1970-1973 Belgium Italy Germany Austria Spain UK France 0 0,2 0,4 0,6 0,8 1980-2000: Increasing specialization Figure 5.3 Agglomeration of manufacturing in the EU* Germany France Italy UK Moderate changes Spain Netherlands Belgium 1994-1997 Sweden 1970-1973 Austria Finland Denmark Portugal Ireland Greece 0 10 20 30 G. Ellison & E. L. Gleaser (1997)/(1999) • Concentration is the rule, not the exception • Geography accounts for 20% of economic concentration • Concentration itself does not imply the existence of spill-overs • Natural advantages (first nature) may have similar effects • -> no real support for GE D. Black & J. Vernon Henderson (1999) ‘Spatial Evolution of Population and Industry in the United States’, American Economic Review Vol. 89, No. 2, May 1999, pp321-327 • evolution US urban growth 1900-1990 • Scale economies and agglomeration • • • • distribution remains remarkably stable big cities stay big little downward mobility more upward mobility “Geography matters?” • Market potential • mpj = ∑ i ≠ j ( Ni /dij) Five hypotheses to be tested 1. 2. 3. 4. 5. The home market effect: large home market leads to net exporters Large market potential raises local factor prices Large market potential induces factor inflows (Chapter 9) Shock sensitivity Reductions in trade costs induce agglomeration 1. Home market effect • an increase in a country's demand for cars will lead to a more than proportional increase of the production of cars • if yes: support for new trade theory with transport costs and geographical economics • if no: support for new trade theory without transport costs or neoclassical theory Davis & Weinstein (1996-2003) • • • • • • • • • Distinguish between trade theory and geographical economics Measuring the home-market effect Xgnr = κgnr + κ1SHAREgnr + κ2IDIODEMgnr + END + errgnr SHARE = share of output good g in industy n for country r IDIODEM = difference between demand gn in r and demand gn in other countries END = endowments for gn + (neo-classical theory) if κ2 >1 home market effect (geographical economics) IDIODEM no geographical content (no distance) Test on Japanese regions Table 5.1 Home market effect for Japanese regions IDIODEM SHARE 1.416 (0.025) yes 0.888 (0.070) no 1.033 (0.007) -1.7441 (0.211) END included? No Yes # Observations 760 760 Source: Davis and Weinstein (1999); Standard errors between brackets, estimation method: Seemingly Unrelated Regressions Problems END is in fact endogenous according to GE theory Home market effect <-> lack of labor supply elasticity -> higher wages in agglomerations 2) Spatial wage structure • Neoclassical trade theory: factor price equalization -> no spatial wage structure • New trade theory: some varieties produced in country A and others in country B, no endogeous agglomeration towards A or B -> no spatial wage structure (unless A and B are different in size from the start) 2) Spatial wage structure: distance to centres • • • • • Hanson (1998) study on Mexico Hypothesis 1: regional wages lower at higher distances from Mexico City and USA Hypothesis 2: trade liberalization has lead to a decline of regional wage differences finds strong support for H1 and weak support for H2 H1: (H2 with time dummy) ln (Wit /Wct ) = k0 + k1 ln(tit ) + k2 ln(tfit ) + errit (5.2) (k1 and k2 negative) remember Wr = ( Σs Ys Trs1-ε Isε-1 )1/ε 2) Spatial wage structure: market potential • Log (Wj) = κ0 + κ1 log(Σk Yk e-κ 2 Dij) + erri (5.4) Table 5.3 EU regions 1992-2000 Coefficient Standard error k1 0.898 0.020 k2 0.013 0.001 R2 0.61 remember Wr = ( Σs Ys Trs1-ε Isε-1 )1/ε 2) Spatial wage structure: real market potential • Hanson (1996) • Log (Wj) = κ0 + ε-1log(Σk Yk ε+(1- ε)/δ Hk(1-δ)(ε-1)/δ Wk(ε-1)/δ T(1-ε)Djk) + errj (5.5) • assumption: agriculture replaced by the housing market as a spreading force of non-tradables. If local demand increases due to agglomeration prices will go up -> additional spreading force Structural wage equation: (5.5) 1970-80 1980-90 δ 0.962 (0.015) 0.956 (0.013) ε 7.597 (1.250) 6.562 (0.838) Log(T) 1.970 (0.328) 3.219 (0.416) Adjusted R2 0.256 0.347 Observations 3075 3075 ε/(ε-1) 1.152 1.180 ρ 0.868 0.847 Significant, but δ very high