Transcript Slide 1
Does what you export matter? In search of empirical guidance for industrial policies D. Lederman & W. F. Maloney DECRG & LCRCE, World Bank April 2011 Pretoria, South Africa The question, the truth about industrial policies (IP), and this report Question: If we are “condemned to choose,” what empirical criteria can we use to select products or industries? The truth: Everybody does it What we do in this report Discipline our thinking about the desirability of IP in “theory” Review empirical indicators advocated by various literatures Some evidence from Latin America, but broader relevance Our motto: “Give IP a chance” (was it John Lennon?) Why might price signals be deceptive in choosing goods and warrant IP? Marshallian externalities related to goods Local externalities that raise productivity with size of the industry: e.g., local industry-level knowledge spillovers, input-output linkages, labor pooling Not infant industries Pushing the envelope: inter-industry externalities through price signals Industry growth and private returns to schooling Volatility externalities and export diversification Empirical Concerns for Policy Makers How can we measure these externalities? Does the world see the same benefits and drive prices down? Look for safe rents, too? e.g., natural resources Think of demand side (!) Do externalities necessarily come with a good, or does it matter how we produce it? Heterogeneity of experiences within industries What if externalities emanate from input use? e.g., knowledge, human capital What if externalities come through price co-movement? e.g., export diversification In practice, measurement of MEs is difficult, so the profession has taken shortcuts Natural resources Low productivity (Smith, Matsuyama, Sachs), few externalities Rent seeking Volatility High productivity goods Rich country goods (Rodrik, Hausmann) “High tech” (based on inputs, e.g., Lall) with high inter-industry ME Preview Introduction Part I: What Makes a Good Good? Conceptual issues (Marshallian externalities) Cursed goods Rich country, high-productivity goods Smart goods Part II: Beyond Goods Heterogeneity in production: how versus what Heterogeneity in quality growth and risk Goods or tasks? (domestic value added) Export portfolio diversification (cursed goods revisited) CURSED GOODS: NATURAL RESOURCES Heterogeneity in Natural-Resource Experiences: Net Exporters, 1980-2005 4 NOR SAU GAB 2 log Natural Resources Net Exports/ labor force 1980 - 2005 KWT ISL VEN DZA SUR COG 0 MYS CHL KAZ ARG RUS ECU IRN PNG NAM CRI CIV BOL URY COL NGA IDN ZAF CMR PER MEX MRT MNG AZEPRY ZMB SLB HND YEM BRA ZWE TJK VNM SYR BTN LVA MWI GTM GIN THA TGO GHA FJI MDG UGAKEN MOZ NER LBR -2 -4 TTO NZL AUS CAN FIN ARE DNK NLD IRL SWE y = 1.2584x - 12.102 R2 = 0.5593 MLI GNB CAF TZA -6 MDA ETH BDI -8 5 6 7 8 log GDP pc 2005 Source: Lederman & Maloney (2008) 9 10 11 HIGH PRODUCTIVITY GOODS Does It Matter What We Export? Hausmann, Hwang, & Rodrik (2007) Model: broadly inter-industry spillover Country should produce the highest productivity good within its comparative advantage (!) Empirics PRODY, EXPY Similar to Lall (2000) Find higher EXPY (partially) correlated with higher growth. Caveats General equilibrium critique again? Rents- higher for products already exported by rich countries? Not generally the case If easy to move into these goods, then barriers to entry and rents are low Empirical findings muddy Animals, electrical machinery same PRODY Finding of an impact on growth fragile Actually, no neat breakdown of rich/poor country goods 35000 30000 25000 20000 15000 10000 5000 0 PRODYs (with +/- 1 SD*) Empirically, some support for MODEL Growth Regressions Base: HHR Regressions Log ( initial gdp) Log (expy) Including the Export Herfindahl and the Investment Share IV GMM IV GMM IV GMM IV GMM -0.0382*** -0.0203** -0.0414* -0.0177 -0.0166* -0.0177 -0.028 0.0215 (0.01) (0.01) (0.02) (0.01) (0.01) (0.04) (0.02) (0.03) 0.0925*** 0.0532** 0.107 -0.00687 0.102*** 0.0504** 0.124 0.00275 (0.02) (0.02) (0.07) (0.03) (0.02) (0.02) (0.08) (0.03) -0.0577*** -0.00566 -0.0431 -0.119 (0.02) (0.10) (0.03) (0.08) Category Log (expy) Log (primary schooling) 0.00468* 0.00565 0.00271 0.0101 0.00394 0.00582 0.00207 0.00958 (0.00) (0.01) (0.00) (0.01) (0.00) (0.01) (0.00) (0.01) 0.0111* 0.0360** 0.00935 0.0566*** (0.01) (0.02) (0.01) (0.02) 0.0551 -0.0381 0.0615 -0.0283 (0.06) (0.04) (0.06) (0.04) Log (Investment Share) Root Herfindal Index Constant Observations With Income Average Value Including the Export Herfindahl and the Investment Share -0.426*** -0.250* -0.572 0.14 -0.186* -0.199 -0.449 0.699 (0.10) (0.13) (0.44) (0.18) (0.10) (0.47) (0.40) (0.46) 285 285 285 285 285 285 285 285 Number of wbgroup 75 75 75 75 Regressions include decade dummies Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 IVs: log population, log land area Income groups: world deciles GMM: Blundell-Bond System Estimator Note: EXPY is n.s. after including either export concentration or investment rate. SMART GOODS Spillovers and the Education-Expansion “Problem” Social returns to schooling can be higher than private returns (Krueger & Lindhal 2001 JEL) When supply of skilled workers increases, returns decline (common sense & own estimates) Prototypical spillover; economy-wide The “problem”: private incentives to invest in education decline … unless demand for skilled workers rises Do some industries provide higher returns to skills than others? If so, IP could help reduce gap between social & private returns by raising the latter Dispersion of skill premium across industries Malcolm Keswell & Laura Poswell (SAJE 2004): RTS (8-12) b/n 0.33 (quadratic) and 0.15 (cubic) What explains skill premiums? Countries versus industries Source: Brambilla, Dix Carneiro, Lederman & Porto (2010) … and exports … Source: Brambilla, Dix Carneiro, Lederman & Porto (2010) IS IT WHAT WE PRODUCE, OR HOW? BEYOND GOODS BRA MYS ROM CHL ESP JPN NLD CHE RUS BGR SWE ISRAUT AUSHRV KOR BEL CYP DNK MEXCHN NOR ZAF THA IRL NZL SAU HKG ARG GRC UKR PHLIND ROM CHL ISL COLLUX SVK VEN HUN TTO BGR HRV CYP KWT TKM BHS VNM ZWE PER ECU URY QAT CZE POL MAR BHS VNM ZWE PRY IRN GIN PRY BIH BIH TTOPHL IDN KWT CRI BHSBOL LKA LVA MAR KWT IRN QAT GTM JAM ISL MDA DZA UGA ATG NGA 76 1 83 Patents Ratio (ranked) World EGY PAK SYR 1 1 TWN USA JPN CHN NLD DEU GBR TWN MYSMEXHKG IRL KOR THA FRA CAN ITA BEL HUN ESP CHE AUS DNKAUT CZE SWE FIN ISR NOR BRA IND PRT NZL RUS LUX POL GRCZAF ROM ARG TUR UKR SVN HRV BGR CHL SAU COL PER LBN BLR VEN VEN PER PAN 1 1 PAN TWN KOR 1 IRN GIN FIN TKM RUS ISR Exports vs Patents in Computers (SIC 357) ECU URY MAR USA DEU FRA GBR ITACAN BRA MYS Exports Ratio (ranked) 174 COL Exports vs Patents in Aircraft (SIC 372) BOL IDN SAU PHLIND POL BOL IDN TUR PRT MEXCHN ZAF HKG ARG UKR THA TUR 181 171 Exports vs Patents in Aircraft, excl. High-Income OECD Countries Producing and exporting without generating knowledge? LAC Patents Ratio (ranked) Patents Ratio (ranked) World LAC World LAC 68 Domestic value added: Does China really export the iPOD? Table 2 China: 10 Exports with the Lowest Domestic Value Added Electronic computer Telecommunication equipment Cultural and office equipment Other computer peripheral equipment Electronic element and device Radio, television, and communication equipment Household electric appliances Plastic products Generators Instruments, meters and other measuring equipment China: 10 Exports with the Highest Domestic Value Added Agriculture, forestry, animal husbandry and fishing machinery Hemp textiles Metalworking machinery Steel pressing Pottery, china and earthenware Chemical fertilizers Fireproof materials Cement, lime and plaster Other non-metallic mineral products Coking Source: Koopmans, Wang, and Wei (2008). 4.6 14.9 19.1 19.7 22.2 35.5 37.2 37.4 39.6 42.2 “..the electronic components we make in Singapore require less skill than that required by barbers or cooks, involving mostly repetitive manual operations” Goh Keng Swee, Minister of Finance Singapore (1972) 81.8 82.7 83.4 83.4 83.4 84.0 84.7 86.4 86.4 91.6 HETEROGENEITY IN QUALITY OF EXPORTS (UNIT VALUES) Quality ladders by product and countries (relative unit values, standardized) But high growth is risky (Brazil on the edge of the cloud, p62) DIVERSIFICATION OF THE EXPORT BASKET Diversification Market failures inhibit diversification Export concentration leads to terms of trade volatility Spillovers in product innovation, which is correlated with diversification Correlation of prices and quantities across products are not internalized poor, small and mining-dependent economies have higher export-revenue concentration, and terms of trade volatility (Lederman & Xu 2010) Problems of diversification policy Big hits are rare and associated with high concentration of (manufacturing) exports (Easterly et al. 2009) Never really know where the next product comes from .4 Export concentration and terms of trade volatility, 1980-2005 .3 IRN RWA NIC .2 SDN BIH GEO CMR UGA CIV ZMB AZE GNB COM TJK KAZ SYR GHA RUS TZA MWI ETH CPV KGZ UKR GMB MOZ ECU VNM BEN IND PAK TGO MDG ARM BFA PER EGY NAM SLV DOM BLRLVA JORIDNKEN CAF BOL PNG CHL ERI SEN BRAARG LKA ZWE MEX COL LSO MLI CZE CRI ALB JPN GUY GTM BGR URYPHL HND ZAF ESP GIN THA MAR PRY SVK TUR MYS MDA ROM MKD MUS KOR POL CHN NZL ITA LTU LBN DEU CHE PRT USA FRA KHM FIN GRC SVN HUN IRL TUNISR PAN EST SWE SWZ GBR DNK BEL HKG AUT NLD HRV BTN DJI TKM 0 .1 YUG 0 BGD MRT BWA YEM .2 .4 .6 .8 Export-Revenue Concentration (Root of Herfindahl) 1 Where Is South Africa in Export Concentration? 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 ITA USA CHN BRA ARG MEX ZAF BWA SAU VEN South Africa: Output Composition Source: Ryan Hawthorne, Reena Das Nair & Keith Bowen, TIPS Trade and Industry Monitor, vol. 37, 2006, p. 101. South Africa: Composition of Manufactures Source: Ryan Hawthorne, Reena Das Nair & Keith Bowen, TIPS Trade and Industry Monitor, vol. 37, 2006, p. 102 (?). The three robust determinants of export concentration and volatility, ToT & GDP volatility Dependent Variable: (1) (2) (3) Export Concentration Terms-of-Trade Volatility GDP-per-Capita Growth Volatility 0.351** (0.000) Export concentration Net exports of energy and mining per worker 0.040** (0.000) 0.004 (0.170) -0.003 (0.154) Net exports of agriculture per worker -0.036* (0.022) -0.000 (0.941) -0.002 (0.456) Labor force (log, initial) -0.058** (0.000) 0.015** (0.000) -0.005** (0.000) GDP per capita (log, initial) -0.065** (0.000) Geographic trade over GDP -0.002* (0.030) 101 0.505 0.000 0.309 0.310** (0.000) 101 0.295 0.000 0.257 Terms-of-trade volatility Observations Pseudo R-squared F-stat (p-value) Adj. R-squared/First Stage 101 0.541 0.000 0.519 Notes: ** and * represent statistical significant at the 1 and 5 percent levels. Cross-equation error correlations are assumed to be unstructured. All explanatory variables, except the dependent variables (export concentration, terms of trade volatility, and GDP-per-capita growth volatility) are assumed to be exogenous. Volatility is measured by the standard deviation of the annual growth rate of each variable during 1980-2005. The “firststage” estimates are not reported. P-values appear inside parentheses and correspond to standard errors adjusted for degrees of freedom due to finite-sample assumptions. “Initial” means that the observation is from 1980; the results correspond to cross-sectional estimates for 1980-2005. Intercepts are not reported. Source: Lederman and Xu (2010). Doing IP blindfolded Little guidance on what goods are good Even whether we should be focusing on goods vs. tasks Leads us back to horizontalish policies that Resolve market failures related to innovation in old and new goods Other barriers to the emergence of new goods and improvement of old Strategic coordination policies Risk taking (entrepreneurship, finance) Fin / Einde