Shanghai Univ meeting June 3 2011 The Economic

Shanghai Univ. meeting June 3, 2011 The Economic Complexity of Countries and Products Luciano Pietronero, Andrea Tacchella Guido Caldarelli, Matthieu Cristelli, Andrea Gabrielli, Antonio Scala Istituto dei Sistemi Complessi, CNR Roma Dip. di Fisica, Università di Roma La Sapienza (WEB page: http: //pil. phys. uniroma 1. it)

Space of Products: two products are close if many counties produce both Complex structure of the Network from which we can get new information In the spirit of Google Page Rank (Hidalgo&Hausmann 2007 -2010)

The Building Blocks of Economic Complexity C. Hidalgo (Medialab MIT), R. Hausmann (Harvard, Economics) PNAS 2009 Standard Economic Theory: Richness and development of a country are linked to specialization and efficiency Global market should lead to work division on world basis with a raise of GDP for all countries … but instead the differencies in GDPL pro-capite Have been exploding in the last century

What matters for economic development are the non exportable resources and the flexibility and adaptability to a dynamical enviroment. Analogy with bio-systems: efficiency vs robustness New method to detect information on the potentiality of a country and the quality of products different from the usual GDP fromglobal export data. (inspired to Google page rank problem)

The Building Blocks of Economic Complexity (C. Hidalgo, R. Hausmann) The basic idea is to consider diversification as an element of fitness (beyond GDP) Model – Tripartite Network Available Data – Bipartite Network

The Building Blocks of Economic Complexity (C. Hidalgo, R. Hausmann) Bipartite network Definition of the variables of the Method of Reflections: Adjacency Matrix

The Building Blocks of Economic Complexity (C. Hidalgo, R. Hausmann) Starting information Ubiquity of exports Not differentiated countries export ubiquitous products Differentiated countries export less available products Differentiation of exports

The Building Blocks of Economic Complexity (C. Hidalgo, R. Hausmann) Results of iterations Correlations with per capita GDP. The deviations are interesting Good predictors of future growth

Summary: The method permits to extract new information from the complex structure of the C-P network and introduces new information odn the countries capabilities. Deviations from the GDP correlations identify a productive potential not fully expressed and represent indications of growth.

Problem 1: Interpretation of the variables The Building Blocks of Economic Complexity (C. Hidalgo, R. Hausmann) Definition Name Description kc, 0 Diversification How many products does “c” export? kp, 0 Ubiquity How many countries are exporting “p”? kc, 1 - How ubiquitous are the products exported by “c”? kp, 1 - How much are the countries exporting “p” diversified? Kc, 2 … how much are the countries exporting the same products that “c” exports diversified? Kp, 2 … ? ? ?

Problem 2: A simple model shows that, even rescaling the variables, the correlation with the “capabilities” is lost when iterating The distribution has no asymptotic meaning: it’s converging to a trivial value Kc, N Year 2000 Iteration Order (N) Rescaled variable: at N=28 the information is lost

Different Economic Principles: H&H 2009: If a product is made by a competitive country it is of high quality even if it is made also by a non-competitive country Linear iteration <---> Narrow distributions (delta funct. ) New Approach: If a product is made by a non-competitive country it is of low quality even if it is made also by a competitive country Nonlinear iteration <---> Broad ditrib. Power Law (Pareto)

New Approach: Fitness & Quality Two variables which are improved by iterations, while keeping the same meaning High Fitness: a country exporting many products of high quality High Quality: a product which is not exported by any low fitness country

Tests on the minimal model show that the correlation is kept Fc, N N=0 N=6 N=1 N=7 Capabilities di c Correlation is constant and not decaying with N

Fitness & Quality A testing model: countries Mcp= Products Fc, N kc, N Qp, N kp, N

Fitness & Quality The new variables and iterations lead to a broadening of the Fitness distribution which appears closer to real data (Laws of Pareto; Zipf) Fitness as a function of the iteration N for 172 countries in the year 2000 Fitness ranking (yellow) Rescaled GDP (blu) Zipf law (violet)

GDP per capita Power laws Fitness Exp(-0. 20) Exp(-0. 83) Exp(-2. 3) kc, 20 (green; 1 st=10) Rescaled kc, 20 (Black; 1 st=100) Fitness seem to correctly catch the exponential form of the decay of the GDP per capita. (For richer countries also the exponent is correct)

Country Ranking Method of Reflections Fitness Kc, 18 Country Fitness 18 Country 154, 8145439 Japan 19, 71 United. States 154, 8141902 Germany 7, 74 Germany 154, 8137368 Switzerland 6, 67 United. Kingdom 154, 8135273 Sweden 6, 10 Netherlands 154, 8132638 United. Kingdom 5, 77 Italy 154, 8129979 United. States 5, 11 France 154, 8129358 Finland 5, 03 Austria 154, 8125976 Austria 4, 84 Belgium 154, 8125543 Taiwan 4, 66 Czech. Republic 154, 8124380 France 4, 51 Spain 154, 8122669 Czech. Republic 4, 35 Japan 154, 8122427 Belgium 4, 24 Switzerland 154, 8122402 Ireland 3, 87 Sweden 154, 8121700 Italy 3, 66 Poland 154, 8119516 Denmark 3, 03 Hungary 154, 8119025 Slovenia 3, 02 Slovenia 154, 8118911 Netherlands 2, 88 Denmark 154, 8117618 Slovakia 2, 74 China 154, 8113951 Spain 2, 72 Slovakia 154, 8112456 Singapore 2, 72 Taiwan 154, 8112375 Israel 2, 37 Ireland 154, 8112137 South. Korea 2, 23 Finland 154, 8110124 Hungary 2, 18 Brazil 154, 8109313 Poland 2, 14 India 154, 8105402 Barbados 2, 06 Portugal 154, 8103882 Saint. Helena 2, 03 Hong. Kong 154, 8101889 Canada 1, 91 Canada

Quality of Products Ranking

TOP PRODUCTS OF ITALY (Fitness=5. 39) Cod Quality e Competitors Description 7851 12, 10 JPN, ITA, ESP MOTORCYCLES 5843 8, 73 JPN, ITA, MEX CELLULOSE ACETATES 6541 7, 61 ITA, CHN, IND FABRICS, WOVEN, OF SILK OR NOIL 7284 7, 11 DEU, JPN, ITA, AU T, SWE MACH. & APPLIANCES FOR SPEZIALIZED PARTICULAR IND. 5825 6, 53 DEU, ITA, NLD, ES P, SVN POLYURETHANES

TOP PRODUCTS OF SWEDEN (Fitness=3. 33) Code Qualit y Competitors Description 7284 7, 11 DEU, JPN, ITA, AUT, SWE MACH. & APPLIANCES FOR SPEZIALIZED PARTICULAR IND. 7451 5, 73 DEU, JPN, ITA, SWE, MEX TOOLS FOR WORKING IN THE HAND, PNEUMATIC, PARTS 7251 4, 11 DEU, ITA, AUT, SWE, POL, FIN MACHINERY FOR PAPER PRODUCTION 6954 3, 95 DEU, JPN, ITA, AUT, SWE, SVN, ISR INTERCHANGEABLE TOOLS FOR HAND & MACHINE TOOLS 7784 3, 81 DEU, JPN, AUT, CHN, SW TOOLS FOR WORKING IN THE HAND WITH E, SVN, MEX ELECT. MOTOR

TOP PRODUCTS OF MOLDOVA (Fitness=0. 51) Code Qualit y Competitors Description 6591 2, 62 DEU, ITA, NLD, AUT, MDA LINOLEUM 7751 1, 33 DEU, ITA, ESP, SVN, POL, T HOUSEHOLD TYPE LAUNDRY EQUIPMENT HA, SVK, FIN, TUR, MDA 5839 1, 26 DEU, JPN, ITA, NLD, AUT, POLYMERIZATION AND COPOLIMERIZATION ESP, FIN, GRC, COL, MDA PRODUCTS 7915 1, 15 AUT, SVN, POL, HUN, CA N, MEX, SVK, ROM, UKR, MDA 7412 1, 13 DEU, ITA, AUT, SWE, DNK, FURNACE BURNERS FOR LIQUID FUEL AND HUN, FIN, GRC, EST, MDA PARTS RAIL&TRAMWAY FREIGHT AND MAINTENANCE CARS

TOP PRODUCTS OF FINLAND (Fitness=1. 87) Code Qualit y 7251 4, 11 7281 3, 50 7138 3, 43 7742 3, 09 5123 2, 72 Description DEU, ITA, AUT, SWE, MACHINERY FOR PAPER PRODUCTION POL, FIN DEU, JPN, ITA, AUT, S MACH. TOOLS FOR SPECIALIZED VN, ISR, FIN PARTICULAR INDUSTRIES DEU, JPN, SWE, MEX, INT. COMB. PISTON ENGINES IND, FIN DEU, JPN, NLD, SWE, APP. BASED ON THE USE OF X-RAYS OR OF IND, ISR, FIN RADIATIONS DEU, JPN, NLD, ESP, PHENOLS& THEIR POL, SVK, ISR, FIN HALOGENAT. DERIVATIVES … … NLD, CHN, SWE, THA, ELECT. LINE TELEPHONIC & TELEGRAPHIC 1, 62 CAN, SGP, MEX, ISR, APPARATUS FIN, IRL … … 7643 … … 7641 Competitors DEU, CHN, SWE, HUN, RADIOTELEGRAPHIC & RADIOTELEPHONIC 1, 13 SGP, BRA, MEX, ISR, TRANSMITTERS

Dynamics Evolution between 1992 and 2000 The cloud has internal dynamics but it’s shape remains unaltered: we can infer about the growth or contraction of the outmost countries

Growth Expectations Year 2000 Log[GDPpc ] CONTRACTION Log[fitness] GROWTH

Growth Expectations Realized 1 year return of GDPpc Distance from equilibrium

Weighted Links Iterations can be refined by introducing the exchanged volumes as weights FQw First step: balancing the contribution of each country to the quality of products (Version 2. 0) FQw 2 Second step: weigthing the effect of each product’s quality to the fitness of countries (Version 2. 1) Weights are defined as:

Possible Developments: • More recent and accurate data 2001 – 2010 • Alternative and more realistic definitions of the variables F and Q • Include geographyc elements • Financial elements (debt, cost of labor) • Application to other bipartite networks • Banks, Companies, Providers …