Policy and Economic Geography: What is new? World Bank, Wash. DC, February 16 2006 Philippe Martin University Paris 1 Panthéon Sorbonne and CEPR Two main questions on economic geography: 1) What are the causes of spatial inequality? 2) What should and what can public policies do? 1) What are the causes of spatial inequality? Main elements of answer from New Economic Geography models: little emphasis on “first nature” geography but circular causality mechanisms economies of scale and transport costs Non linear effects of lower transport costs (easier trade between regions): a first phase of concentration, possibly followed by a phase of deconcentration Simple example of location choice for a firm (Krugman) with lessons for emerging markets central region: large market (or easy access to foreign markets) but high wages and production costs periphery region: small market but low wages and production costs economies of scale: less costly to have only one plant than two (fixed costs) cost to trade (barriers to trade and transport costs) choice of location: minimize production + trade costs production trade costs medium low costs high produce in -2 regions - central region - periphery region 0 0 0 10 3 1.5 0.75 8 8 4 2 12 Example suggests: transport infrastructures that At lowtrade trade costs: firms exploit low production AtAt Example high suggests: costs, economic spatial concentration geography is may be medium trade costs, economic geography lower transport costs between poor and rich costs in periphery dispersed natural phase (firms for want emerging to be close markets to all markets) becomes concentrated because firms exploit region may favor rich region (liberalization) economies of scale and trade costs still important Some implications: 1) Necessary (but not sufficient) condition for periphery region to develop: compensate geographical disadvantage by lower costs (wages): important to allow regional differences in wages (no centralized bargaining) 2) New Economic Geography models tell us that periphery status is endogenous Emphasize self-reinforcing or cumulative agglomeration mechanisms: firms follow other firms which follow workers who are also consumers Are these mechanisms empirically relevant ? Empirical results for emerging markets (Brazil): 1) Industries with increasing returns to scale are mainly located in regions with large markets + good market access 2) Cost and demand linkages matter (industries locate close to customers and input providers) 3) Labor intensive industries locate in low wage regions Why should policy makers care about economic geography? The usual suspects: Equity Efficiency Four issues on equity and geography: First issue: Regional inequality and migration double effect of labor mobility: more agglomeration but regional inequalities may have less effects on welfare agglomeration with empty places but less regional inequalities in GDP/cap workers follow firms: both labor supply and labor demand decrease: effect of spatial concentration on wages and unemployment rates inequalities are lower With low mobility of workers: Agglomeration has negative welfare effects on immobile agents: hurt by departure of mobile factors as both workers and consumers One implication: fostering inter-regional mobility is part of regional policy Some public policies make it more difficult: - public housing (workers who move have to get at end of line in the other region) - some labor laws reduce inter-sectoral + inter-regional mobility - some unemployment benefits - legal hurdles (China) But not a miracle solution: - Cultural, ethnical differences: mobility will never be perfect - firms move faster than workers: transition problem - The most mobile workers are (usually) the most skilled , those with high level of “positive externalities”, lost for those staying behind: Increase in mobility for certain agents may worsen effects of spatial inequalities on most fragile Suggests (valid for emerging markets with basic functioning markets, not necessarily for poorest countries) - concentrate mobility policy on low skilled workers - to be consistent: concentrate public resources on poor urban areas Second issue: Regional and individual inequalities Usual view of policy makers: fight regional inequalities (regional subsidies) is a way to fight individual inequalities Is this true? Do wage differences across individuals depend on geographic factors and not only individual factors (education, age, sex…)? Are wages in poor regions lower because? - 1) poor regions have many low skilled - 2) firms pay lower wages for given skill due to lower revenues (far from markets) and higher costs (far from suppliers) Evidence (need more for developing countries) Studies using individual data suggest (also in emerging markets , Brazil) - most of explained differences in individual wages is due to individual characteristics (education, age, sex) : 2/3 - suggests priority is education in poor regions as a regional policy - but still 1/3 explained by economic geography in particular market potential: i.e. size of market for firms - market potential difference across regions explains twice as much of wage variance as discrimination against women - economic geography still matters for inter-personal inequality Third issue: Do spatial inequalities lead to individual inequalities or is it the reverse? - In Poland for example: evidence going in both directions the increase in individual inequalities (higher return to education and skills) may have led to an increase in spatial inequalities: those in the big cities are those with human capital (composition effect) FDI has been concentrated in richer regions and has led to increase in relative labor demand (geography effect) Fourth issue: Do subsidies to firms investing in poor regions take from the poor of rich regions to give to the rich of poor regions? subsidies to firms relocating in poor regions are often subsidies to capital: capital is mobile so if return to capital increases in poor region should increase also in rich region reducing regional inequalities (through subsidies to capital and not labor) may increase inequalities between workers and capital Efficiency argument for regional policies Is spatial agglomeration of economic activities efficient or “an underutilization of economic and social potentials”? if economies of scale exist: some agglomeration must be efficient - internal economies of scale: save on transport costs and fixed costs for firms - external economies of scale: localized spillovers are maximized with spatial concentration (many empirical studies on this, also in developing countries: not only high tech) at what level does congestion makes agglomeration inefficient? - congestion depends on public policies (urban infrastructure) Is there a efficiency-equity trade-off at the spatial level? If growth poles or clusters are necessary to trigger the growth of a country: difficult economic choice what should be the priority? - External convergence of the country -Internal convergence between regions inside the country - if trade-off exists, public policies that have a spatial dimension (public infrastructure, regional subsidies…) have to explicitly take it into account What is the evidence for this tradeoff? Evidence for trade-off: European regions (NUTS1) GDP/cap and regional inequality (Crozet and Koenig, 2005) Does this trade-off hold for developing countries? Short answer: we don’t know ! Need for research on this Reasons for which the trade-off may be less important (or even reversed) : -1) in poor countries with little manufacturing, mechanisms based on economies of scale will matter less - 2) under-utilization of capital and labor in poor regions (capital and labor market failures) - 3) congestion problems (cities) due to insufficient public resources may be more important - 4) inter-regional, ethnic conflicts may be more important in countries with more regional inequalities Reasons for which the trade-off may more important in developing countries: - 1) in emerging markets, growth is based on manufacturing: mechanisms based on economies of scale will matter a lot (more than in service based economies) - 2) in emerging markets, industrial clusters are in formation: important not to counter them - 3) export-led growth favours coastal cities (China) : most international trade is made between cities; and regions close to export market (Mexico) Regional policies are difficult to evaluate - economic geography is a cumulative process: firms follow other firms which are their consumers and their input producers implication: very non linear effects of regional policies -a policy that gives subsidies to a poor region may have no effect at all if does not put into motion a cumulative process (agglomeration rent) -rarely, a small public policy may have very large effects when economic geography is not settled (anecdotal evidence is biased!) Infrastructure policies: demand and supply effects may be of opposite sign Example: transport infrastructure between poor and rich region - short term (keynesian) positive effect : local politicians are likely to only care about these - more complex supply long term effect Lower transport costs between rich and poor region: can make it easier for firms in sectors with economies of scale to concentrate production in the large market and re-export in the poor region: increases agglomeration (but may be efficient) Lower transport costs inside poor region (local roads) increases effective market size: relocation towards poor region: efficient? empirical studies show mixed results of regional policies - non surprisingly: positive short-term effect - controversial on the long term basic infrastructure (education, health, transport, etc.) in the poorest regions of these countries may contribute to reduce labor mobility: problem if do not attract firms too Again: equity-efficiency trade-off growth GG -lower inter-regional transport costs - liberalization of trade (inter-regional and international) AA agglomeration growth - education policy GG AA agglomeration Conclusion: some debatable implications (for emerging markets): 1) market forces may lead to economic geography that is not concentrated enough (from a pure efficiency point of view) - 2) agents in poor regions can gain from agglomeration in richer regions if increases growth of country (if enough mobility) - 3) transport infrastructure between rich and poor regions may lead to more concentration rather than less but this may be good - 4) Education policy in poor regions may help both reduce interregional inequalities and increase mobility - 5) public policies should facilitate migration to growth poles which implies more resources for public services in those regions (congestion)