63330 A p r i l 2 0 11 Wo r l d B a n k B ra z i l Quarterly Knowledge Report INTERVIEWS BRINGING BRAZIL TO WASHINGTON 7 Minas Gerais’ Governor EDITORIAL BY MAKHTAR DIOP Antonio Anasta- sia on the next wave of public B razil’s experience shows us that middle income country clients are looking to the World Bank knowledge work not only to provide the traditional diagnostics, but to help them devise their own ways to fix development problems, export and adapt their solutions, and upgrade their own analytical sector reforms. capacity. 15 Daniel Lederman looks into Bra- This requires bringing in more than leading-edge knowledge and experience, but also using the find- ings from rigorous impact evaluation to inform policy making, facilitate south-south links, and strate- gic use of the Bank’s seal of approval and convening power. zil’s options in industrial and The Brazil Quarterly Knowledge Report is a part of this effort, but this edition export policies. (Continued on page 20 ) INSIDE THIS ISSUE ACHIEVING WORLD CLASS EDUCATION Editorial 1 IN BRAZIL: THE NEXT AGENDA Fernando Bizerra Jr./BG Press World Class Education 1 By Barbara Bruns Brazil Knowledge Day 6 T he 2009 results for PISA (the OECD’s test of high school student learning levels) con- firmed Brazil’s impressive progress in raising Antonio Anastasia 7 education performance over the past decade. Brazil’s 52 point increase in math since 2000 Wage Convergence 9 implies that students gained a full academic year of math mastery over the decade, and TFP and Growth 11 Brazil’s overall score increase – from 368 to 401 – is the third largest on record. The coun- Daniel Lederman 15 try’s 2009 scores still trail the OECD average and East Asian countries by a large margin and Jobs and Exports 17 are no grounds for complacency, but few coun- tries have made faster or more sustained pro- Missing Middle 19 gress. The PISA secretariat is sending a film crew to Brazil to document the sources of this In the Loop 20 education advance. How Brazil moved from one of the worst- KEY DATES performing education systems of any middle- • Executive Directors Visit income country to one with strong and sus- tained improvement, not only in learning but to Brazil—April 27-May 3 also in primary and secondary school coverage, sustained reform in education over the past 15 • President Zoellick’s Visit is one of the central questions addressed in the years. to Brazil — May 31-June 3 new World Bank report: Achieving World Class Education in Brazil: the next agenda. By bench- A six-year-old Brazilian child born today into the • Financial Literacy Con- marking the country’s current education per- bottom quintile of the income distribution will ference — Rio, May 9-10 formance and identifying key issues, the book is complete more than twice as many years of • Accelerate Brazil Fo- intended as a resource for the new Federal schooling as her parents. The average educa- rum—Rio, May 25-27 government in setting education priorities for tional attainment of the labor force since 1995 the next four years. But it also tells the story of has been one of the fastest on record – faster Brazil’s remarkable run of policy continuity and than China’s, a global leader in schooling ex- Printed on recycled paper Page 2 Ap r i l 2 0 11 pansion in prior decades. Major perform- grams into Bolsa Família, and ance gaps with middle-income countries coverage grew from 4.9 million Global Impact in LAC and elsewhere are closing, such as families in 2002 to 12 million in in primary school completion and pre- 2009, with transfers increasing The Achieving World Class Education in Brazil school coverage. And in key areas such as from 3.4 to 11.9 billion reais report was featured by “The Economist” maga- assessing student learning and education (about US$ 7 billion in 2009 zine in December 2010. The article, titled “No performance monitoring more generally, prices). The testing of a small na- Longer Bottom of the Class” highlighted the Brazil today is a best practice example not tional sample of students under report’s assessments of the promise of Brazil’s only in LAC, but globally. SAEB every two years was ex- tended to a nation-wide test of myriad education experiences, especially to The “managed revolution” of Brazilian math and Portuguese called Prova improve teacher quality and effectiveness, and education began when the Brasil and applied to the country’s struggle to close the education gap Cardoso government as- all 4th, 8th and 11th with OECD and other middle income countries. sumed critical normative grade students. functions at the federal Combined with data level that had previously “While Brazil is on student enrollment, support includes programs such as Mais gone unfilled: equalizing Educação, and expansion of the federal repetition and graduation funding across regions, moving up, the rest of rates, a comprehensive technical schools. The Federal govern- states and municipalities ment has also strongly supported innova- index of school performance with the FUNDEF reform; tion – whether school-level planning un- the World is not was generated, called IDEB der FUNDESCOLA, multi-grade teaching measuring the learning of (Índice de Desenvolvimento all children on a common under Escola Ativa, or capacity building da Educação Básica). With national yardstick (SAEB); standing still” an IDEB score for every one for municipal education managers, with and protecting the educa- the Plano de Ações Articuladas (PAR). And of Brazil’s 175,000 primary tional opportunity of stu- the report acknowledges an even larger and secondary schools, dents from poor families wave of innovative reforms at the state 5,000-plus municipal school systems, 26 (Bolsa Escola). With those reforms, plus and municipal levels – which have core state and federal district systems, every the first comprehensive legal framework responsibility for the delivery of pre- single segment of the Brazilian education for basic education (the Lei de Diretrizes e primary, primary and secondary education system can benchmark how well their Bases in 1996) and the first national cur- students are learning and how efficiently in Brazil. riculum guidelines, the Ministry of Educa- their school or school system is perform- But while Brazil is moving up, the rest of tion got the core elements of a national ing. No other large federal country in the the world is not standing still. The core of education policy profoundly right. world has achieved this. the report examines how Brazilian educa- But what happened next was equally im- tion today stacks up against OECD and Federal, state and municipal-level policies portant. The Lula administration elected LAC countries on three paramount func- in education have been progressive and in 2002 not only retained these core poli- tions: i) developing the labor force skills innovative in many other areas as well. cies, but expanded and strengthened for sustained economic growth; ii) contrib- The Ministry of Education’s strong norma- them. FUNDEF financing equalization was uting to poverty and inequality reduction tive role has included new standards for extended to secondary school and pre- by providing educational opportunity to all; teachers, federally-supported, higher qual- school and called FUNDEB. Bolsa Escola and iii) transforming education spending ity teacher training programs, textbook was consolidated with other transfer pro- into education results – above all, student screening and production. Investment learning. It concludes that while progress Figure 1: Brazil VS. Strongest Improvers—PISA Math Performance 2000-2006 has been substantial, the agenda ahead is crucial. Labor force skills are improving, but still lag behind. Brazil is still quite far from the average learning levels, secondary educa- tion completion rates, and student flow efficiency of OECD and other middle- income countries. While Brazil improved more than any other LAC country between 2000 and 2006 on PISA, both Poland (the non-OECD country with the largest gains) and Mexico, the country with the second fastest improvement in LAC, performed better in absolute terms. While the 2009 results are expected to close Brazil’s gap with Mexico, it will still lag OECD, East Q u a r t e r l y K no w l e d g e R e po r t Page 3 Figure 2: Brazil: Change in Education Attainment, 1993-2009 in the share of workers cies such as Bolsa Família. In 1993, the with secondary educa- child of a father with no formal education tion has been accom- would complete only 4 years of schooling, panied by a decline in on average; today Brazilian students com- the real wage for sec- plete between 9 and 11 years of school- ondary education ing, regardless of their parents’ education. graduates in recent Learning outcomes for students from the years. At the same bottom income quintile have also im- time, there has been proved. But a gap still remains. an increase in the wage premium for The issues underlying the low learning higher education attainment of children from poor families graduates, which is become more complex over time. Physical consistent with a access to schools and household budget global pattern of de- constraints recede in importance, while mand for workers with social issues (teen pregnancy, gang and Source: PNAD, various years strong analytical skills drug involvement), family instability [Editors note: see also the article by Ro- (unemployment, domestic violence, home- Asian and Eastern European countries lessness), and learning issues and devel- cha and Firmo in this edition]. Labor mar- by a substantial margin. The implica- opmental deficits stemming from chil- ket data in Brazil are signaling that “21st tions are serious, as researchers over dren’s earliest years become more promi- century skills” are important for the next the past decade have generated com- nent. Since the core equity issue in Brazil- generation of workers in Brazil, and pro- pelling evidence that what students ian education has shifted from equalizing ducing these will be a critical challenge for actually learn – measured on globally access to equalizing learning attainment, the education system over the next dec- benchmarked tests such as PISA – and secretaries of education across Brazil are ade: graduates with the ability to think not how many years of schooling they increasingly focused on two major strate- analytically, ask critical questions, learn complete, is what counts for economic gies to address this, which are both con- new skills, and operate with high level growth1. communications/interpersonal skills, sistent with global best practice: preven- including foreign language mastery and tive interventions (expanding Early Child- The massive expansion of schooling in hood Development services for low in- Brazil over the past 15 years has had the ability to work effectively in teams. For the basic education system, the overriding come families) and remedial interventions dramatic effects on the labor force. In (tutoring, accelerated learning programs 1993, close to 70 percent of the labor implication is the urgency of raising stu- dent learning. and other programs targeted to children force had not completed secondary with special needs). school. Today that number is 40 per- More equal education access is contribut- cent. The biggest change is not access Education spending is outpacing results. ing to lower inequality and poverty, but to primary school, but the much higher The report raises concerns about a mis- gaps in learning achievement still remain. share of children who stay in school match between Brazil’s current level of There has been a striking equalization in through secondary education. The rise spending on public education and results, schooling attainment in just one genera- tion in Brazil, as a result of aggressive especially in light of the government’s 1. Hanushek and Woessman, 2009, 2010. proposed target of 7 percent of GDP for expansion of schooling coverage and poli- education. First, in 2007, Brazil’s spend- Figure 3: Average Schooling Completed and PISA Learning Outcomes, by Income Quintile Sources: PNAD 1993 and 2009, PISA data 2000-2006 Page 4 Ap r i l 2 0 11 Figure 4: Consolidated Education Spending Figure 5: Public Expenditure on Education as a Percent of GDP in 2007, OECD and Brazil in Brazil, 2000-2009 (Constant 2009 Reais) Source: Brazilian National Treasury Source: OECD (2010). Education at a Glance. Table B2.4 ing on public education (5.2 percent of not detract attention from the areas Brazilian “education action lab.” GDP) was already above the OECD aver- where comparative data show that Bra- age of 4.8 percent of GDP (total spending zil’s current level of spending should be Raising teacher quality. In Brazil, teaching on education in OECD countries, including producing better results. Among these has become a low-status profession that private spending, was higher, but compa- are: anomalies in the allocation of public does not attract high academic perform- rable data on private education spending funds across different levels of education ers. Data show that teachers are recruited in Brazil are not available). As the school- – while OECD countries spend on average from the bottom third of high-school stu- aged share of the population is larger in 2 times as much per student in tertiary dents – in contrast to Singapore, Korea Brazil than in most OECD countries, this is education as at the primary level, Brazil and Finland, where they come from the to be expected. However, Brazil also spends almost 6 times as much; persis- top third. Raising teacher quality in Brazil spends more than Mexico, China, India tent high repetition and high costs per will require recruiting higher-capacity indi- and Indonesia, which have similar demo- graduate; rising teacher costs (several viduals, supporting continuous improve- graphic profiles. Second, Brazilian GDP is policies over the past decade have raised ment in practice, and rewarding perform- growing rapidly, faster than the school-age teacher costs with little evidence – either ance. Both the federal government and population. But third and most impor- in Brazil or elsewhere – that this contrib- some state and local government have tantly, Brazil is experiencing a demo- utes to improved results); innovation, already begun reforms in these areas, graphic transition that will have a dra- privately-supported programs and sub- such as with teacher bonus (pay for per- matic impact on the school-aged popula- stantial investments in new technologies formance) programs in Minas Gerais, tion over the next decade. The projected with little cost-effectiveness research; and Pernambuco and São Paulo states and 23 percent drop in the number of primary mismanagement of education funds. Rio de Janeiro municipality. With support school students will mean almost 7 mil- from the Bank’s education team, school lion empty seats in primary schools across Four key challenges for Brazilian educa- systems are also using standardized the country. Were Brazil to follow the Ko- tion from 2010-2020. classroom observation methods devel- rean example and hold class size con- oped in OECD countries to look “inside the stant over this period, the primary school Grounded in this comparative assessment black box” of the classroom and identify teaching force would decline by over of Brazilian education vis a vis OECD and examples of excellent teacher practice 200,000 (from 840,000) by 2025. This LAC comparators, the report zeroes in on that can anchor their professional devel- transition is a bonus for the education four critical challenges for the coming opment programs. In Minas, Pernambuco system and would permit current spend- decade: raising teacher quality; protecting and Rio municipality, data showed that ing levels to finance a large increase in the early development of the most vulner- while the OECD benchmark is 85 percent able children; building a world-class sec- of each instructional hour effectively schooling quality. ondary education system; and maximizing spent on learning activities, in no Brazilian In this context, the report suggests that the impact of federal policy on basic edu- system to date has this exceeded 66 per- the pursuit of spending increases should cation – especially by capitalizing on the cent. A high share of teachers also failed Q u a r t e r l y K no w l e d g e R e po r t Page 5 to use available learning materials and recommends introducing curricula tai- 60 percent of their students. between 43-64 percent of the time, lored to each educational level; more in- students were visibly not engaged (in tense training and supervision of caretak- The challenges are extreme, but a number OECD countries, the benchmark for ers and educators; and strengthened of states are already working on compre- students “off task” is 6 percent of time monitoring and evaluation of ongoing hensive strategies. Some, such as Minas or less). Instead of theory-oriented programs. In each of these areas, there is Gerais, are developing important new courses, training programs designed a role for the Ministry of Education – in approaches to a key issue for secondary from classroom observation evidence providing guiding materials and oversight education: the balance between aca- use videos and practical exercises to – as well as the state or municipality – in demic and vocational content. Promising impart effective techniques for manag- implementing programs. directions which can contribute to im- ing the classroom, using learning mate- provement in secondary education in Bra- rials and keeping students engaged and Building world class secondary education. zil can be loosely grouped as: system-wide on-task. This practice-oriented training No segment of the Brazilian education strategies (curriculum and training reform, is the new direction in OECD countries system crystallizes the quality gap with major infrastructure investments to sup- and key states in Brazil are ahead of the OECD countries as clearly as secondary port a longer school day and eliminate curve. school. Fully 42 percent of secondary evening instruction, improving teacher students are enrolled in night shifts, quality); demonstration schools (full-day, Protecting early child development which deliver only 4 hours of instruction a highly resourced secondary schools that (ECD). Global research points to ECD day – compared with 7 hours or more in both test out innovations and demon- interventions as the most powerful strat- most OECD countries. Infrastructure is strate that high quality secondary schools egy for reducing inequality and leveling deficient – lacking the libraries, science are achievable); and public-private part- the education playing field. Over the labs, computer and language facilities nerships for technical and vocational edu- past 15 years, Brazil has made progress most OECD students enjoy. The curricu- cation (to ensure a smooth transition to in raising crèche enrollments from 8 to lum is overloaded and memorization- work for secondary education graduates 18 percent of the 0-3 age group and oriented, and virtually every state secon- who do not go on to higher education, by pre-school enrollments from 49 to 80 dary school system faces severe short- orienting the vocational content of the percent. Priorities for the next decade ages of qualified math and science teach- curriculum to skills that are in local de- are improving the targeting of services ers. As a result, 40 percent of all Brazilian mand and supporting more results-driven to the lowest income, most vulnerable secondary schools are considered school management). children, and raising quality. The report “dropout factories” – failing to graduate Figure 6: Spending per Student at Different Education Levels Relative to Unit Costs in Primary Education (2007) Primary Education = 100 Note: A ratio of 300 for tertiary education means that expenditure by educational institutions per tertiary student is three times the expenditure by educa- tional institutions per primary student. A ratio of 50 for pre-primary education means that expenditure by educational institutions per pre-primary student is half the expenditure by educational institutions per primary student. 1. Public institutions only (for Italy, except in tertiary education). 2. Some levels of education are included with others. Countries are ranked in descending order of expenditure by educational institutions per student in tertiary education relative to primary education. Page 6 Ap r i l 2 0 11 Maximizing federal impact and capitaliz- tem; at this moment across Brazil by dynamic, ing on the Brazilian “education action results-oriented secretaries of education. create incentives for state-wide im- lab.” Given the progressive, effective edu- Few other countries in the world have the provement to reward states for closer cation policies pursued by successive scale, scope and creativity of policy action integration of state and municipal Federal government administrations over that can be seen today in Brazil. Even school systems, rather than directing the past 15 years, the report states that more unique is the large number of cut- direct federal support to small mu- “in this context, it is not trivial to identify ting-edge policy areas in which different policies that could substantially speed nicipal education systems; states and municipalities are experiment- Brazil’s progress toward world class basic capitalize on the Brazilian “education ing with similar programs with slightly education.” But the analysis points to four action lab” by supporting systematic different design features – such as the recommendations: evaluation of innovative state and pay for performance programs in Minas municipal programs. Gerais, São Paulo, Pernambuco and Rio stay the course on the core policies Of these four recommendations, perhaps de Janeiro municipality. The chance to of the last 15 years (FUNDEB funding the last holds the most potential impact. study these systematically makes Brazil equalization; IDEB results measure- The long-term work of improving primary one of the world’s best laboratories for ment; and Bolsa Família conditional and secondary school performance is the generating global evidence on “what cash transfers); works” in education. A concerted federal responsibility of over 5,500 state and focus on spending efficiency rather municipal schools systems in Brazil. Liter- effort to mine this rich experience more than targets for higher spending, ally tens of thousands of creative new effectively through rigorous impact which can worsen the risks of leak- programs and policies are being tried out evaluation might be the single fastest age and mismanagement in the sys- route to world class education BRAZIL KNOWLEDGE DAY: A “S O UT H TO B A N K ” EXCH A NGE By Mauro Azeredo themes of interest to the country dialogue. Canuto. Pablo Fajnzylber, Brazil PREM Reports have close supervision for quality Sector Leader, and Laura Chioda LAC Re- K nowledge is a central dimension of the Bank’s work in Brazil. It repre- sents much of our value-added and is assurance and tailored partnership and dissemination strategies. search Economist, presented the reports, which were commented by the World Bank’s Chief Economist for the Africa Re- what clients ultimately seek, especially The result are highly influential and inno- gion, Shanta Devarajan and Mauricio when it is one with operations. The country vative products, some of which were show- Cardenas (Brookings Institution). can be though as a very special case. It is cased in the Brazil Knowledge Day. The full an über MIC, with very sophisticated chal- day discussion was divided in three broad Finally, the Bank’s analytical program in an lenges and capacity, and fundamental sessions, focusing on the human capital, environmental superpower such as Brazil contributions to international development opportunities and the green agendas. could not lack a strong sustainability com- and the global policy debates. ponent. Sector Director Laura Tuck and Tamar Atinc, World Bank Vice-President for Brazil Sector Leader Mark Lundell pre- Bringing this knowledge and experience to Human Development, chaired the first sented two seminal reports, which are other teams in the Bank could be a signifi- session, which highlighted four closely influencing public policy in Brazil: the Brazil cant contribution to our effectiveness and related studies, from early child develop- Low Carbon and Amazon Dieback assess- innovation. ment to Brazil’s aging population, passing ments. Marianne Fay, Chief Economist of through the country’s challenges in educa- the Sustainable Development Network It was with this in mind that the Country tion and job market development. Discuss- Office organized the “Brazil Knowledge chaired the session, with comments from ants Ariel Fiszbein (Chief Economist of the Andrew Steer, Special World Bank Envoy Day” in December 2010 in DC. The event Human Development Network) and Nancy was opened by the World Bank’s Vice- for Climate Change, and Joelle Chassard, Birdsall (Center for Global Development) President for Latin America and the Carib- Manager of the Carbon Finance Unit. commented on the presentations by Mi- bean, Pamela Cox, and Brazil Country Di- chelle Gragnolati (Brazil HD Sector Leader) The Bank’s knowledge work is geared to rector Makhtar Diop and drew strong me- and Barbara Bruns (Lead Education Econo- have maximum usefulness and impact on dia attention. Several authorities, among government policy, and our clients and mist). which the Brazilian Ambassador to the stakeholders in Brazil are appreciative of U.S., participated. Brazil’s drive to provide opportunities for this contribution. Brazil generates much all its children, a central goal of the knowledge and South-South exchanges, Transparency and Competition Rousseff administration, and its long strug- and it is natural that the broader experi- The Brazil Analytical and Advisory program, gle with crime and violence were the focus ence behind the Bank’ analytical engage- under the leadership of Lead Economist of the second session, chaired by the ment also benefit the institution as a Tito Cordella, is based on a transparent Bank’s Poverty Reduction and Economic whole, in what could be accurately termed competitive process with a focus on Management Vice-President Otaviano a “South to Bank” exchange Q u a r t e r l y K no w l e d g e R e po r t Page 7 MINAS GERAIS: R EADY FOR THIRD GENERATION OF PUBLIC SECTOR REFORMS Interview with Minas Gerais Governor Minas Gerais is the reduction of Antonio Anastasia these inequalities through the im- provement of social, economic, and By Denise Marinho physical infrastructure, as well as F the creation of opportunities. or two decades, Antonio Anastasia held increasingly important techno- In 2002, when Governor Aécio Ne- cratic positions in state and federal ves was first elected, he invited me administrations, becoming an expert in to serve as coordinator of the Gov- public policy management. ernment’s plan and, in the post- transition period, as Secretary of In 2003, as secretary of Planning for the Planning. State of Minas Gerais, Anastasia was responsible for implementing a highly This effort therefore began in 2003, effective and innovative public sector and involved a momentous process, management program, widely known as with the assistance of many partners “Choque de Gestão” (Management including the World Bank and the Shock). The idea was to improve public Dom Cabral Foundation. When our administration through the adoption of partnership with the World Bank targets and indicators, to make public started in 2003, we did not have the investments more efficient and improve tools, let alone the technical exper- public services to the population. The tise, to take the first step. program was expanded and strength- ened in the second term of Governor But we knew that the public admini- Aécio Neves, in which Anastasia was stration in the State was in dire need Vice-Governor. of improvement, professionalism had to be introduced, goals and The program's success has enabled the results established, procedures Governor Anastasia: Minas’ famous Management Shock Government of Minas Gerais to put its streamlined, and the mechanisms program will give way for “Management for Citizenship”. finances in order, pay civil servants sala- that had led to private sector success ries on time and foster public invest- needed to be identified and adapted to So, since 2003, during the first term of ments, especially in public safety and the public sector. The overarching idea Governor Aécio Neves, we developed a education areas. It was only natural that was one of “management shock.” large and ambitious project called “Pró- Anastasia was chosen as Governor Ne- Acesso”, with the Bank’s participation ves’ chosen successor, and he won the What was our major objective? To demon- from the onset. 2010 ballot with 63 percent of the strate that public administration should votes. not be handled in an amateur fashion. It Now, nearly 200 municipalities of those had to be professional and, to achieve 225 have paved roads, something that In an exclusive interview with the Quar- this, results and goals had to be set, and nobody imagined that could be done. terly, Governor Antonio Anastasia ex- individuals and managers had to be ac- Paved roads mean much more than the plains how the State will trig- countable. The situation mere physical access. They represent ger the "third wave" of this had to change. Unfortu- opportunities and connection to a broader model of public manage- “Our objective was to nately, in our culture, context of services and development for ment, which he calls public administration was these small towns. Because now they “Management for Citizen- have the possibility to attract businesses show that public ad- like a no man’s land. to generate jobs and income for local ship." There was no owner and, for this reason, there was people. Brazil Economic Team: Minas ministration should no accountability. BET: What effects this new management Gerais was the pioneer Brazil- ian State in developing a not be handled in an BET: Could you mention approach had on the role of government? management model based some results of the new on results and you are one of Anastasia: We achieved spectacular re- the main mentors behind this amateur fashion” management model? sults, first in terms of motivation of civil program. What led Minas Anastasia: For instance, servants, who were initially fearful of lay- Gerais to adopt this management para- Minas Gerais has 853 municipalities and offs. However, the reverse happened. digm? 225 of them had no paved road. What is We’ve have hired 40,000 new civil ser- the paved road to a city? It is an entry vants through competitive examination, Antonio Anastasia: Minas Gerais is a route; is the way to access opportunities, and salaries began to be paid on time. very complex state. Unfortunately, many jobs, businesses and the conditions of There were salary increases and we were inequalities, both regional and social, development and progress. able to adopt productivity goals to grant still persist. The main challenge facing bonuses known as the “14th salary,” a Page 8 Ap r i l 2 0 11 Minas Gerais’ Public Sector Management Revolution to be active citizens. To accomplish this, we developed the idea Minas Gerais emerged from the 1990s facing a very difficult fiscal situation. In of a government network, launched in my 1999, the state government declared a default on its state debt with the federal government platform. We are now making government. preparations for the next four years. Here again, our goal is for citizens to play an Since 2003, the Minas Gerais state government has promoted a strong fiscal adjust- active rather than passive role in public policies, since we know that to achieve ment strategy. Under a program called Choque de Gestão (Management Shock), the effectiveness, citizens must engage in state government defined a set of measures directed to curb the increasing trend of such policies as education, health, and expenditures through tight control of personnel and operating costs. security, which are the most important. Results were immediate and impressive. Since 2003, the state government has BET: And how this will be achieved? obtained vigorous primary surpluses. As a result, the state consolidated debt fell Anastasia: This participation will be markedly, increasing investment capacity. achieved through a government network system. For example, we will launch a The Minas Gerais Partnership for Development, supported by over $1.5 billion in project called “Family Teacher” (Professor loans from the Bank, has moved the reforms further, creating Brazil’s pioneer results da Família). What is the rationale behind -oriented management environment, in which public servants commit to targets and this? In the case of education, we know receive incentives to meet them. that we can improve schools and teach- ers. However, if a family does not encour- The successful program is now being emulated in other states ands municipalities. age a child or adolescent to study at The Bank was also instrumental in sustaining Minas’ broad reorganization and mod- home, he or she will fail. ernization of the state public administration and helping the state boost sustainable As part of this program, we will go to economic development through promotion of private sector investments. homes to visit students with the worst grades and to encourage family participa- cate of good governance. While sound tion. novel concept for the government. administrative management is not the BET: Have other Brazilian States ex- What was the effect on the average citi- end goal of the State or public authority, it pressed interest in learning more about zen? Positive. As we improved education, is the bedrock of all public policies. health, and security indicators, works and this management model? investment expanded and Minas Gerais We will never have high-quality education, Anastasia: First of all, we are the product joined the ranks of Brazilian states with high-quality public health, and sound pub- of a successful undertaking, given that we greatly improved indicators. lic security in Brazil if we do not take into received major incentives from external account that efficient governance under- sources, including the World Bank, for Minas became the second major em- lies and supports all these policies. On the ployer in the country. Given our great het- which we are very grateful. contrary, even if resources, good will, and erogeneity and regional inequalities, it is a good technical team were available, it Of course, we've managed to “tropicalize” wonderful to see that Minas Gerais cur- would not be possible to achieve the de- this experience, and make it more from rently ranks first among states in basic sired results without sound governance. Minas Gerais, and it became a successful education according to the Brazilian Index case. We're always willing to share this of Basic Education Quality (IDEB), and BET: What are Minas’ next steps? experience. Minas Gerais is open to re- third in sanitation. We are fortunate to ceive visiting mission and also send our have security indicators that make us the Anastasia: During former Governor Aécio technicians, because we believe that this safest state in Brazil in terms of per capita Neves’ first term, the “Management fosters improvements and is a require- homicides. Shock” program was introduced. The goal was to put the house in order. During the ment of solidarity. Naturally, while the results for citizens are second term, the notion of the “results- We've done that, receiving several mis- positive, much remains to be done. How- based State” was introduced to provide sions from other states and municipali- ever, progress has been impressive. citizens with concrete governance results. ties. We also visit other states to learn from their good ideas. So there is already BET: Is it possible to improve the effective- Now, the third wave is the so-called a Brazilian collaboration spirit. On the ness and efficiency of the Government “Management for Citizenship” (Gestão same way, we have welcomed developing through this management model? How? para a Cidadania), which seeks to stimu- countries that can benefit from this man- late citizens to increase their interaction Anastasia: I have always been an advo- with public authorities, to participate and agement model Q u a r t e r l y K no w l e d g e R e po r t Page 9 FORMAL-INFORMAL WAGE CONVERGENCE By Romero Rocha and Figure 1: Full Time Employees Average Real Monthly Wages, by Sector Márcio Firmo T his article describes and analyzes the wage convergence between formal (com carteira) and informal (sem carteira) employees in Brazilian metro- politan areas from 2002 to 2010. Social policies resulting in higher reservation wages have been pointed as the main factor associated with this wage conver- gence. However, monthly employment survey data show that what drives the phenomenon is the fact that the wage premium for high levels of schooling has decreased in the formal sector but not in the informal one. Therefore, the con- vergence is actually observed in the upper level of education (and conse- Barbosa1 attributed the growth in informal However, the evidence presented in this quently, income) distribution, and not in sector wages to the growth in the mini- article points in another direction. Al- the lower ones as the social policies- mum wage and the reservation wage in- though some increase can be seen in the reservation wages channel would have crease created by social programs. Ac- wages of formal employees in the low suggested. cording to him, poor people in the infor- levels of the education distribution, the mal sector request salary raises when the average real wage has been fairly stable Brazilian labor market has been respon- minimum wage is increased, so that the from 2002 to 2010 for formal employees sible for the most important part of the minimum wage acts as a reference index with a complete secondary education; and impressive reduction in the country’s for informal sector adjustments. A similar actually decreased for workers with uni- inequality over the last decade. Unem- increase of the minimum wage and of versity degrees, from R$ 4,567 to R$ ployment rates have fallen systemati- informal wages reduces the informal/ 3,620. That is: the more educated the cally and recently reached the lowest formal wage ratio. In addition, programs worker, the less it has benefited from the recorded level. Surprisingly, however, like Bolsa Família tend to increase the growth in real wages over the last eight informal employees’ wages are growing reservation wage of the poorer, which years. faster than those of formal employees. would make workers refuse bad jobs, This is reflected by a 36 percent in- In the informal sector, the dynamics has crease in the ratio between the average pushing informal wages up. been quite the opposite: while a small monthly wage of full time informal work- increase is noticed for the low levels of ers and full time formal workers, from 1. Referenced in O Estado de S. Paulo, Febru- education, wages increased very fast 2002 to 2010, going from 60 percent to ary 21, 2011 “Mínimo afetará o salário dos among higher education levels for the 81.7 percent in the period (Figure 1). informais” http://www.estadao.com.br/estadaodehoje/2 informal employees, from R$ 3,080 to R$ 3,565 on average for workers with univer- A recent study by Fernando Holanda 0110221/not_imp682221,0.php Figure 2: Wage Premium (Full Time Employees Monthly Wages) Page 10 Ap r i l 2 0 11 Figure 3: Wage Premium (Full Time Employees Monthly Wages in the Industry Sector) sity degrees. Consequently, the ratio be- difference of returns to education be- In summary, the well-documented fall of tween informal and formal wages in- tween formal and informal employees). wage premiums for high skilled workers creased 15 percent for workers with a can be observed only in the formal sector. complete primary education, 17 percent Figure 2 shows returns of education levels The same pattern is not observed in the for workers at the secondary level, and an for the formal and informal sectors. The informal sector. This disparity is even impressive 46 percent for university education levels used as reference were more prominent in the industrial sector, graduates. Convergence has thus been incomplete primary and no education. It is where the wage premium has increased higher for workers with more education. observed that the wage premium for high for informal employees. Moreover, the This evidence contradicts the interpreta- skill levels of education (complete univer- changes in observable characteristics tion that the increase in informal wages sity and complete secondary) has de- (including education distribution) are re- comes from the reservation wages of creased in the formal sector from 2002 to sponsible for only 29.7 percent of the some social policies, since the faster con- 2010. The same thing is not observed for wage convergence. The other 70.3 per- vergence is noticed in the upper part of informal employees. For them, the com- cent are explained by changes in returns education (and consequently wage) distri- plete university wage premium has fluctu- (observable and unobservable) and bution. ated around the initial level in the period, changes in unobservable characteristics. with no observable trend, and the com- A decomposition was made to check if the plete secondary wage premium, after a It seems fair to conclude that social poli- changes in the composition of observed decrease from 2002 to 2003, has been cies, such as the increase in minimum characteristics (such as years of study, almost stable from 2003 to 2010. wages and the conditional cash transfers experience, gender, race, region, house- like Bolsa Família, are not driving the hold position and training) play an impor- An interesting fact comes out when we wage convergence between formal and tant role in the reduction of the gap. We look at the returns to education by eco- informal sectors. If social policies are not find that these changes explain only 29.7 nomic activity sectors. For industry, the driving the formal-informal wage conver- percent of the wage convergence. Another wage premium for informal workers with a gence, natural candidates appear to be 20.3 percent is explained by common complete university education has in- the dynamics of the labor market itself, in changes in returns of the characteristics creased from 2002 to 2010 (Figure 3). At a context of higher steady-state economic (for example, changes in returns to educa- the same time, the share of informal growth. It might be the case that formal tion will affect more formal workers, as workers in industry with a complete uni- market rigidities prevent formal employ- they are still more educated). Finally, 50 versity education, although increasing, is ees to fully capture economic benefits of percent of the reduction of the gap is ex- still much below the share of formal work- this enhanced growth, while informal plained by changes in unobservable char- ers with that education level, hinting that workers are better able to adjust their acteristics and by changes in the way this could be one of the reasons why the wages to the growing value of their these unobservable characteristics are wage premium is still increasing for this work paid (for example, the reduction in the education category. Q u a r t e r l y K no w l e d g e R e po r t Page 11 T O TA L F A C T O R P R O D U C T I V I T Y A N D E C O N O M I C G R O W T H : S TA T I C A N D D Y N A M I C L I N K S By Guilherme Lichand Figure 1: Total Factor Productivity (Share of São Paulo’s TFP), for Selected States W hat are the determinants of pro- ductivity in Brazil? Is it true that richer states are also the most produc- tive ones? Or that most productive states are those filing more patents, or with the higher share or personnel in R&D activities? In order to shed some light on these important questions, this piece investigates how Total Factor Pro- ductivity (TFP) at the state level in Brazil over the last 30 years covariates with measures of economic activity (such as investment in physical capital, access to basic services and exports), innovation outputs (e.g., patents filed) and innova- tion effort (e.g., personnel in R&D activi- ties). The productivity debate has a long story in economic literature. With decreasing returns to the accumulation of factors of production in the long-run, when at least one input is fixed (e.g., land), only tech- nical change could drive economic growth. Notwithstanding its relevance, it is generally agreed that there is a lack some reason, this would prevent output It is apparent that if any convergence in of understanding about what exactly from adjusting along the equilibrium path TFP levels among States has taken place productivity stands for and what are its towards high-TFP firms that face a lower over time, it is a “convergence to the bot- growth drivers. price of capital. In addition, it can also be tom”: while the distance between São the case that high-TFP firms face dispro- Paulo and the other states only increased Total Factor Productivity, its most com- portionately high capital prices. This pre- over the last 25 years, initial dispersion is mon empirical measure, can be best vents those firms from growing by accu- reduced, and in 2008 no other state had understood as the “measure of our igno- mulating more capital, whereas low-TFP more than 20 percent of SP’s TFP. None- rance” (Abramovitz, 1956) about what firms accumulate ‘too much’ capital in theless, there is a strong association be- drives economic growth besides factor countries like China or India. This would tween the rankings of TFP and GDP, with accumulation and returns to scale. Sur- negatively affect growth as long as invest- richer states displaying higher TFP. prisingly, maybe, the contribution of TFP ment and innovation do matter for growth. to growth exceeds by far that of factor Figure 2 on page 12 highlights a distin- accumulation (over 50 percent of GDP As such, focusing on TFP as a public pol- guishing feature of the data: TFP current per worker growth even by more recent icy benchmark would be useful as long as levels are not correlated with their subse- accounts, see for instance Abramovitz this measure is indeed correlated with quent variation. In fact, the previous fig- and David, 2000). That makes it even innovation and investment. This piece ure also does not back ‘club conver- more important to deepen our knowl- investigates whether that is the case at gence’ -- where high-TFP states have edge about the nature of TFP, how it is the level of the Brazilian states. Although higher TFP growth while the opposite is linked to GDP growth and how policy the ideal level of analysis would be the true for the lagging ones. could affect it. firm, this preliminary exercise can possibly shed some light on the correlations be- This pattern has important implications This piece looks at the links from TFP to tween the variables of interest, pointing for policy: units of analysis with higher TFP growth. Recent research (see for in- out the relevant questions for future re- are not those experiencing higher TFP stance Hsieh and Klenow, 2009) search under a more precise identification growth. If TFP growth (as opposed to its suggests that one of its most prominent setting. levels) is thought of as the result of invest- links to economic growth might be mal- ment and innovation, then the fact that functioning financial markets, operating Data for TFP at the state level (for 24 wedges are disproportionately higher though capital misallocation. More pre- states, excluding only MS, RO and RR), among high-TFP firms is of little interest to cisely, if firms facing high wedges from 1981 to 2008, draws upon Jivago understanding the impact of capital mar- (distortions to the market price of capi- Ximenes’ work (to be published), and is ket imperfections to economic growth. tal under a perfectly functioning finan- displayed in Figure 1 above as the share cial market) do not exit the market for to São Paulo’s TFP, which leads the TFP We proceed to look at the correlation of ranking. TFP levels and growth to more concrete Page 12 Ap r i l 2 0 11 Figure 2: Total Factor Productivity Level vs. Growth The descriptive statis- gression of TFP second-order change on tics for two variables the other variable’s change. All specifica- of innovative effort tions include year fixed-effects to control presented for se- for non-observable factors that vary along lected states in fig- the time. ures 3A and 3B are surprising, with poorer As such, the idea is to capture the rele- states’ firms allocat- vant reduced-form dynamic between TFP ing resources to inno- and the variables of interest. Moreover, vation to a greater we run the four specifications for five dif- extent. This is particu- ferent definitions of TFP: (i) contempora- larly interesting in neous to the covariates, (ii) leaded on face of our previous year, (iii) leaded 3 years, (iv) lagged one finding concerning the year, and (v) lagged three years, since association between there might be a non-linear association TFP and GDP levels: it between these variables that would not must be the case that be captured otherwise. Variables are used states with higher in natural logarithmic form so that our innovative efforts are estimates correspond to (semi-) elastic- precisely those with ities. For the sake of expositional simplic- lower initial TFP. Of course, this is merely ity, we display empirical results only quali- variables usually regarded as sources of tatively, so that (+) indicates a pair-wise economic growth. Since we deal with suggestive evidence, it would be interest- ing to assess whether there is a consis- correlation positive and significant at the states as opposed to firms, we also con- 10% level, (-) indicates a pair-wise correla- sider indicators of urban services and tent pattern over time and if any correla- tion exists between TFP, innovation and tion negative and significant at the 10% activity composition. Data for access to level, and (0), a pair-wise correlation non- adequate water and sanitation, exports, GDP, whether this association if found to be in levels or changes of the variables. significant at the 10% level. Results for investment and schooling, and for the economic activity, access to services and share of manufacturing and services in With that in mind we moved on to empiri- physical investment are omitted from the GDP are drawn from IPEADATA1. Data for cal exercises. We run four specifications, tables so that we can focus on the innova- our nine measures of innovative effort always looking at the pairing correlation tion variables; the complete set of results come from the Brazilian Institute of Geog- between TFP and another variable of in- is available upon request. raphy and Statistics’ industrial surveys on terest: (1) with both variables in levels, (2) innovation (PINTEC) for 2000, 2003, with state fixed-effects, implying that our Results are condensed in Tables 1A, 1B, 2005 and 2008, available only for 13 coefficient mimics that of a regression 2A and 2B – each cell of the tables States. with both variables in changes, (3) with stands for a different regression. In Table only TFP in changes, and (4) with TFP in 1, the first set of coefficients of interest is changes and state fixed-effects, implying that of levels-on-levels regressions. 1. http://www.ipeadata.gov.br that our coefficient mimics that of a re- Figure 3A: Percentage of Innovative Firms for Selected Figure 3B: Skill Profile (Percent of R&D Personnel with University Degree) for States Selected States Q u a r t e r l y K no w l e d g e R e po r t Page 13 In conformity with the evidence sug- Table 1A: Levels Regressions Contemporaneous 1-year-ahead TFP 3-year-ahead TFP gested by descriptive statistics, TFP is (Contemporaneous TFP and positively correlated with GDP, exports Leads) (1) (2) (1) (2) (1) (2) and investments, but negatively and Level- Level- Level- statistically significantly correlated with Δ—Δ Δ—Δ Δ—Δ level level level several measures of innovative effort at the state level such as percent of inno- % of innovative firms - 0 - 0 - 0 vative firms (and its adjusted measure % of innovative firms (innovations that account for that accounts only for higher impact over 10% of revenues) - 0 - 0 - 0 innovation), percent of innovative firms that implemented advanced manage- R&D expenditures (% of revenues) 0 0 0 + 0 0 ment techniques; and is negative Skill profile (% of R&D personnel with university though not statistically different from 0 0 0 0 0 0 degree) zero with almost every other measure from PINTEC. These findings are very % of innovative firms that deposited patents 0 0 0 0 0 0 homogenous across definitions of TFP: analogously to contemporaneous TFP % of innovative firms that implemented advanced management techniques - 0 - 0 - 0 and its leads, regressions with lagged TFP present the same patterns showed % of innovative firms that implemented change in in table 1. The evidence, hence, is not organizational structure 0 0 0 0 0 - consistent with targeting high-TFP states % of innovative firms that implemented new as the sources of innovative effort, marketing strategies 0 0 0 0 0 0 though these states indeed invest in physical capital and grow at faster rates. % of innovative firms that implemented changes in products design 0 0 0 + 0 0 When we move to the changes-on- changes regressions (Table 1), while Table 1B: Levels Regressions 1-year-behind TFP 3-year-behind TFP almost no coefficient is significant, the (TFP Lags) sign of the coefficients of the regres- (1) (2) (1) (2) sions with the innovation variables are basically all reversed to positive. The Level-level Δ—Δ Level-level Δ—Δ dynamic link is more clearly seen on the changes-on-levels regressions (Table 2): % of innovative firms - 0 - 0 contemporaneous TFP variation is posi- tively and statistically significantly corre- % of innovative firms (innovations that account for over 10% - 0 - 0 lated with percent of innovative firms of revenues) and percent of innovative firms that R&D expenditures (% of revenues) 0 0 0 0 implemented advanced management techniques. However, the dynamics are not homogenous when one considers Skill profile (% of R&D personnel with university degree) 0 0 0 0 lagged- or leaded-changes: an increase in innovative effort increases current % of innovative firms that deposited patents 0 0 0 0 TFP and boosts TFP variation one year % of innovative firms that implemented advanced manage- ahead, but the effect is not significant ment techniques - 0 - 0 for TFP 3 years ahead (Table 2A). Con- % of innovative firms that implemented change in organiza- versely, states with previous increases tional structure 0 0 0 0 in TFP display a significant decrease in innovative effort (Table 2B). % of innovative firms that implemented new marketing strate- gies 0 0 0 0 These results indicate non-linear dy- namics of TFP and innovation over time, % of innovative firms that implemented changes in products design 0 0 0 0 with innovative effort cycles matching fluctuations in productivity. As such, tion propensity or capital expenditures; growth are not (or are even negatively) these are at least suggestive of a new since current TFP does not predict TFP correlated with current TFP, but rather reading of productivity-driven public growth, the potential reallocation gains with TFP growth, and in a rather non- policies: if one is concerned about the from a static perspective might be dy- linear, cyclical fashion over time. effects of capital market imperfections namically misleading, either because of on economic growth, he/she should In this sense, the literature that investi- adjustments along the equilibrium path or care about the distribution of capital gates how R&D expenditures follow a because innovation variables traditionally price wedges in what comes to innova- suboptimal pattern due to economic cy- understood as sources of economic Page 14 Ap r i l 2 0 11 cles and credit constraints (see, for in- capital markets (see Visaria, 2009, variables correlated both with innovation stance, Aghion et al., 2005) or that which among others), might be better to inform and TFP – although this is a minor prob- assesses to what extent extensive- and public policy on how government can best lem when we included state fixed-effects intensive-margin investments of entrepre- contribute to productivity growth. (unless those unobserved variables also neurs respond to improvements in the vary over time), it limits the interpretation institutional environment, particularly While these are interesting findings, some of the levels-on-levels regressions. More- those conducive to better functioning caveats are in order. First and foremost, over, without firm-level data, intra-state we do not control for potentially important heterogeneity -- such as different innova- tive behavior conditional on firms’ attrib- Table 2A: Changes Regressions Contemporaneous 1-year-ahead TFP 3-year-ahead TFP utes such as age or size – could bias our (Contemporaneous TFP and estimates if the distribution of such attrib- Leads) (3) (4) (3) (4) (3) (4) utes is correlated with state-level TFP. Alternative interpretations, such as that Δ-level Δ2—Δ Δ-level Δ2—Δ Δ-level Δ2—Δ high-TFP states do not innovate to a greater extent precisely because firms in % of innovative firms + 0 0 0 0 0 those states face higher credit constraints % of innovative firms (innovations that account for – despite the fact that these states dis- over 10% of revenues) 0 0 0 0 0 0 play higher investment in physical capital –, cannot be completely dismissed and R&D expenditures (% of revenues) 0 0 0 0 0 0 deserve further investigation Skill profile (% of R&D personnel with university degree) 0 0 0 - 0 - References % of innovative firms that deposited patents 0 0 0 0 0 0 ABRAMOVITZ, M. (1956) ‘‘Resource and % of innovative firms that implemented advanced Output Trends in the United States since + 0 + 0 0 0 management techniques 1870”, American Economic Review, Pa- % of innovative firms that implemented change in pers and Proceedings, 46, pp. 5–23; organizational structure 0 0 0 0 0 0 ABRAMOVITZ, M. and DAVID, P. (2000) % of innovative firms that implemented new ‘‘American Macroeconomic Growth in the 0 0 0 0 0 0 marketing strategies Era of Knowledge-Based Progress: The % of innovative firms that implemented changes Long-Run Perspective’’ in The Cambridge in products design 0 0 0 + 0 0 Economic History of the United States, vol. 2, The Twentieth Century, edited by Stanley L. Engerman and Robert E. Table 2B: Changes Regressions 1-year-behind TFP 3-year-behind TFP Gallman. Cambridge University Press, (TFP Lags) Cambridge, 2000, pp. 1–92; (3) (4) (3) (4) AGHION, P., ANGELETOS, G., BANERJEE, Δ-level Δ2—Δ Δ-level Δ2—Δ A. and MANOVA, K. (2005) “Volatility and Growth: Credit Constraints and Productiv- % of innovative firms - 0 - 0 ity-Enhancing Investment”. NBER Working % of innovative firms (innovations that account for over 10% Paper No. 11349; of revenues) - 0 - 0 HSIEH, C. and KLENOW, P. (2009) R&D expenditures (% of revenues) 0 0 0 0 “Misallocation and manufacturing TFP in China and India”, The Quarterly Journal of Skill profile (% of R&D personnel with university degree) 0 0 0 0 Economics, 124 (4), pp. 1403-1448; MIDRIGAN, V. and XU, D. (2010) “Finance % of innovative firms that deposited patents 0 0 0 0 and Misallocation: Evidence from Plant- % of innovative firms that implemented advanced manage- Level Data”, NBER Working Paper No. ment techniques 0 0 - 0 15647; % of innovative firms that implemented change in organiza- 0 0 0 0 VISARIA, S. (2009): “Legal reform and tional structure loan repayment: the microeconomic im- % of innovative firms that implemented new marketing strate- pact of debt recovery tribunals in India,” gies 0 0 0 0 American Economic Journal: Applied Eco- nomics, 3, 59–81. % of innovative firms that implemented changes in products design 0 0 0 0 Q u a r t e r l y K no w l e d g e R e po r t Page 15 D O E S W H A T Y O U E X P O RT M A T T E R ? Interview with World Bank Senior might be concerns about microeconomic Economist Daniel Lederman volatility associated with mineral exports because export concentration tends to By Denise Marinho be higher in economies that depend on O il, gas or precious metals... Devel- mineral exports. oping countries with natural re- On the other hand, smart goods are in- source abundance and little export dustries or products associated with diversification face the risk of having higher return to skills or to schooling, blessing turned into a plague. However, meaning that skilled workers employed on the path to long-term development, in the production of smart goods tend to natural resources can be associated have higher wages than unskilled work- with cursed goods only if an economy ers, and that that relationship is higher concentrates its exports on a few com- when compared to other industries or modities, becoming more vulnerable to products. So, it is the relationship be- macroeconomic volatility in the market- tween the wages of skilled workers and place, explains Daniel Lederman, World unskilled workers. Bank Senior Economist in the Develop- ment, Trade and International Integra- BET: Which criteria a middle-income tion Research Group (DECTI). country like Brazil should take into ac- count when choosing exporting sectors to In the paper "Does What You Export support? Matter? In Search of Empirical Guid- ance for Industrial Policies", Lederman Lederman: There are four types of crite- Lederman: smart goods are industries or products and his colleague William Maloney, ria that one should consider. One is the associated with higher return to skills or to schooling World Bank Lead Economist in LAC, level of income; another is the size of the explain that the main challenge for economy or the labor force; another one sign of computers can be high-tech in the countries like Brazil is to maintain a is the degree of dependency on mining sense that the industry is focused in in- very diversified export structure and and energy products. The fourth could be vesting in research and development and foster a knowledge intensive produc- the level of risk in the product composi- patenting ideas, whereas in another tion process. Here are the highlights of tion of exports. In general, poor coun- country, the same product might be pro- his interview to the Quarterly Knowl- tries, smaller countries and countries duced and exported abroad in an indus- edge Report. that depend on mining and petroleum for trial organization characterized by an energy exports tend to have a more con- assembly operation and not by a knowl- Brazil Economic Team: The study refers centrated export structure or portfolios, edge intensive production process. to cursed and smart goods. Could you which are then associated with higher explain what they are? macroeconomic volatility. Therefore, we think it’s much safer to target the act of innovation rather than Daniel Lederman: In our In those types of economies, the product itself. What a government report, we included a “The evidence for the it is useful to think about should do is stimulate the private sec- chapter called “Cursed innovative ways that could tor’s investment in innovation – R&D or goods” and there is a long-standing academic supposed curse on help the economy diversify product development, upgrading mana- its exports. We would not gerial practices, introduction of new pro- and popular literature on suggest, however, particular duction methods that provide informa- the role that natural re- natural resources , products or industries to be tion to all companies and can be useful sources play in develop- subsidized. for commercial purposes in other indus- ment. There is a so-called especially in mining, tries. curse on natural re- BET: Focusing on high-tech sources that poses that is not robust” exports is a good strategy to BET: Do Latin American countries stead- countries with a high develop a country’s indus- ily invest in R&D? dependency for their try? exports on natural resources, espe- Lederman: In a previous work from the cially mineral and energy products, Lederman: We believe that the notion of World Bank’s Latin America region, we tend to grow more slowly because of a high technology good is a little bit mis- have benchmarked almost every major various reasons. What we find is that leading, because we’ve observed that economy in terms of their intensity of the evidence for the supposed curse on countries that export the same product investment in R&D. Without exception, natural resources, especially mining can have dramatically different industrial we believe that most countries in Latin products, is not robust. organization in the production process. American are lagging behind global stan- dards, even after controlling for the in- We do believe, however, that there In one economy, the production and de- Page 16 Ap r i l 2 0 11 dustrial structure of the economy. For But as the country begins to exploit and Government receives proposals for inno- example, Chile, which is a mineral based produce more mineral products, includ- vative investments and it creates a com- economy, invests less than 0.8 percent ing energy products, it is likely that its mission to blindly evaluate each individ- of GDP in R&D. Norway, which is an en- export structure will become more con- ual proposal and then runs them by the ergy rich economy, invests close to 2 centrated. The challenge is developing a potential value in business, from the percent of its GDP. It knowledge economy point of view of society, funding 50 per- could be that Chile does out of a mineral rich cent of the investment. The private sec- just fine because cop- “Brazil’s needs to economy, that’s why tor really needs to be committed to that per does not require a we do believe there’s a idea. lot of R&D. But after rational justification for taking that into account, develop a knowledge the public sector to be BET: Brazil has a long history of IPs and we still believe that proactive in the inno- is considering changes. How would you Chile under-invests in economy out of a mineral vation and technology assess the country's record and options R&D. agenda, because the in the area? rich economy. For this, the private sector by itself Lederman: Embrapa, for instance, has a In the case of Brazil, might not make the what we’ve observed in good record. The international literature optimal levels of in- the data coming from public sector needs to be vestment from the shows that the ratio of return for R&D UNESCO is that the and knowledge diffusion in agriculture point of view of the country in recent years proactive in innovation and country as a whole. tends to have very high rates of return, has steadily increased which also dispels the notion that innova- its R&D investment as a technology” BET: How to ensure tion is only a phenomenon that occurs in share of GDP, and is that industrial policies manufacturing industries. However, one approaching 1 percent are not captured by of the reasons why it is so difficult to of its GDP. To give you a better sense, powerful interests? have a very rigorous impact evaluation of the Government of the United States the Brazilian industrial policy system as a alone invests about 1 percent of GDP in Lederman: There is no cookie-cutter solu- whole is because there are so many play- R&D. The private sector in the U.S. in- tion to these institutional challenges. ers. vests another 1.5 percent of the GDP in Surely, it requires a political leadership to prevent the establishment of an overly In Brazil, we have all the ministries, R&D. So given that the structure of the close relationship between the private BNDES, Embrapa, universities, the state economy in Brazil is different from that in sector and the Government. But institu- governments and their own agencies as the U.S. one would have to do the com- tional mechanisms can also be set up. players. So to do a systemic evaluation putation. But even without having done For example, suppose that a Government becomes almost impossible because we that type of structure of analysis of the wants to subsidize pri- cannot disentangle the composition of R&D across industries, vate investments in effects of the different we tend to believe that Brazil can do better. R&D or in product de- “One of Brazil’s main programs and how they velopment across a end up affecting the It is important to say that Brazil is a wide set of industries. outcomes in the mar- challenges for the long ketplace. We look for- global leader in some types of R&D, for That policy would be example in the agricultural sector unlikely to be captured ward to strike a dia- through Embrapa. And the World Bank by one interest, be- term is to maintain a logue with the capable has learned a lot from the practices of cause it would be fo- staff in the ministries Embrapa. So we are not saying that Bra- cused on all possible much diversified export and the regional State zil has an underdeveloped innovation products and sectors governments to make system, but it may be interesting to raise that entrepreneurs can structure” an assessment of their the national level of investment in R&D, think of. system in a qualitative and in innovation more generally, both in way and also to under- terms of the Government and private So just having programs that are focused stand how to design new and innovative sector. on diversification of innovation rather programs such as matching grant pro- than on a sector already helps counter- grams, in a way that we can also evalu- What we do not know is whether this balance possible interest groups. Sec- ate them overtime and have rigorous and kind of effort should be concentrated in a ond, one can establish mechanisms credible impact evaluation data that can few sectors, or much more likely, diversi- where the public funding for private sec- help improve results and impact. It is fied across the whole economy. One of tor investment in innovation across a very important to design these policies in Brazil’s main challenges for the long term broad set of industries is not 100 per- a way that they can be evaluated is to maintain, as it has right now, a cent of the investment. These would be much diversified export structure. matching grant programs, where the Q u a r t e r l y K no w l e d g e R e po r t Page 17 HOW MANY MORE JOBS WILL BETTER ACCESS TO FOREIGN MARKETS BRING? By Abdoulaye Sy and erages and Tobacco, Wood Products and eign markets as the number of destina- Barbara Farinelli Manufacturing and Assembling of Motor tions (countries) a particular industry ex- Vehicles). Our analysis implies that differ- ports to3. Pooling different rounds of the O ver the last decade, Brazil’s access to foreign markets improved con- siderably. In 1998 the average manu- ences in access to foreign markets ex- plain a great deal of the evolution in in- PNAD we compute employment and aver- age monthly labor earnings for the 26 dustry structure in Brazil during 1998- two-digit export industries in Brazil during facturing industry in Brazil exported to 2009. We also find that while improved 1998-2009.4 87 countries, by 2009 this reached 132 access to foreign markets led to similar countries1. However, there are large increases in employment levels of high- Our first set of results is displayed in Ta- differences in access to foreign markets skill and low-skill workers, it widened the ble 1 (next page), where each row corre- across export industries and, more im- labor earning gap between high-skill and sponds to a different regression. Column portantly, within export industries over low-skill workers. (1) shows the results where the depend- time (Figure 1). Changes in the number ent variable is the logarithm of employ- of export destinations within an industry As recent evidence shows, access to for- ment, and row 1 shows results pooling all reflect evolutions in the competitiveness eign markets leads firms to export, inno- industries. The point estimate in column 1 of that industry or industry-specific sup- vate and invest in productivity-enhancing and row 1 is .06 and is precisely esti- ply and demand shocks. Such shocks technologies2. This suggests that a larger mated. This result suggests that an addi- may result from changes in world de- market size will raise productivity of firms, tional export destination raises industry mand for goods and services of an in- their demand for labor and market equi- employment by 6 per cent or about dustry or the differential impact of do- librium wages. We define access to for- 58,000 new jobs. Row 2 to 8 repeats this mestic and international macroeco- analysis for seven groups of industries. All 2. Alla Lileeva, Daniel Trefler: “Improved Ac- nomic fluctuations across industries. cess to Foreign Markets raises Plant Productiv- point estimates are of the same magni- Given recent periods of large volatility in ity..for Some Plants”, The Quarterly Journal of tude and are significant except for two world economic activity, it is important Economics(2010) 125(3):1051-1099; and industry groups (Manufacturing and As- to understand how these fluctuations Paula Bustos: “Trade Liberalization, Exports sembly of Motor Vehicles). and Technology Upgrading: Evidence on the may have affected Brazilian export in- Impact of Mercosur on Argentinian Firms”, The dustries through a reduction in the num- An important question is whether some American Economic Review(2011) 304-340. ber of countries they export to. In this 3. All regressions are weighted by the total industries are likely to grow or shrink due note we analyze how changes in access value of trade to account for the fact that some to differences in access to foreign mar- export destinations are large and have a higher kets changes. To respond to this question to foreign markets affect employment share of total export. we compute simple estimates of the and wages in different export industries. 4. We use a fixed effect model to estimate the model (including both an industry fixed effect change in employment resulting from Before turning to the detailed descrip- and a year fixed effect). We also estimated differences in export destinations across tion of the analysis, we summarize our models using the exchange rate as an instru- industries. We then examine and discuss main findings. Our results show that ment for the number of export destinations the implication of these estimates in (the results were similar). Standard errors are changes in export competitiveness had terms of different industry’s relative ex- clustered at the two-digit industry level. different effects on Brazil’s export indus- tries. We find that differences in the Figure 1: Average Number of Export Destinations by Industry Groups number of export destinations explain the relative expansion of some sectors as measured by their share in total em- ployment (Textile, Apparel and Leather, Wood Products, Paper and Publishing and Manufacturing of Machinery and Equipment industries) while other sec- tors contracted (e.g. Mining, Food; Bev- 1. The composition of export destinations changed slightly during this period. The share of high income export destinations (resp. middle income and low income) went from 45 percent in 1998 to 38 percent in 2009 (resp. from 30 percent to 25 percent for middle income export destinations, and from 20 to 24 percent for middle income export destinations). Page 18 Ap r i l 2 0 11 pansion and contraction as measured by Table 1: The Effect of Access to Foreign Markets on Employment by Industry their share in total employment. We imple- Effect of an Addi- Ratio Between ment this analysis as follows. For each of Estimated Actual Change tional Export Des- Estimated and Change in log in log Employ- the 26 export industries, we estimate the tination on Em- Employment ment 1998- Actual Change in change in the logarithm of employment ployment log Employment 1998-2009 2009 (Standard Error) 1998-2009 between 1998 and 2009 as the product between the marginal effects reported in (1) (2) (3) (4) column (1) and the increase in the num- ber of export destinations during the 1. All Industries 0.06 (.018) 0.056 0.051 1.1 same period. We then aggregate these 2. Mining (Extraction) 0.08 (.015) -0.11 -0.68 0.16 estimates across all industries (row 1) and across the seven broad industry 3. Food; Beverages and 0.06 (.039) -0.17 -1.25 0.13 groups (row 2 to 8)5. In column (3) we Tobacco report the actual change in the logarithm 4. Textile; Apparel and of employment and in column (4) we show 0.07 (.018) 0.07 -0.61 0.11 Leather the ratio between the estimates and the data. The results show that our model 5. Wood Products, Paper 0.06 (.015) 0.12 0.58 0.21 and Publishing explains a great deal of the pattern of relative expansion and contraction of in- 6. Manufacturing of Min- 0.07 (.04) -0.20 -0.47 -0.42 dustries as measured by their share in eral Products total employment. In particular, our model 7. Machinery and Equip- explains 8 to 21 percent of the expansion 0.08 (.03) 0.13 1.68 0.08 ment of the Machinery and Equipment; Textile; Apparel and Leather and Wood Products, 8. Manufacturing and Assembling of Motor 0.03 (.02) -0.01 -1.43 0.01 Paper and Publishing industries. For in- Vehicles dustries that contracted in relative terms, our model also explains a significant amount of the evolution, except for the Table 2: The Effect of Access to Foreign Markets on Employment and Wages7 Machinery and Equipment industry. High-skill Work- Low-skill Work- Ratio of High-skill The results presented above suggest Bra- All Workers ers ers to Low-skill zil’s improved access to foreign markets is accompanied by significant changes in (1) (2) (3) (4) the relative size of different export indus- tries in Brazil. Which workers are more 1. Employment 0.069 0.068 0.063 -0.002 likely to gain from these changes? Do (0.016) (0.016) (0.015) (0.002) these changes affect inequality in earn- ings? With these questions in mind we 2. Earnings 0.018 0.024 0.002 (0.021) analyze how changes in the number of export destinations affect industry level (0.008) (0.009) (0.005) (0.071) demand for high-skill and low-skill workers and their labor earnings6. The results from this analysis are reported in Table 2. Col- 2.1 percent increase in the earnings dif- umn 1 shows that an additional export ference between high-skill and low-skill destination raises overall employment by workers. 5. We have experimented using both un- weighted and weighted means using as 6.9 percent and labor earnings by 1.8 weights an industry’s share of total employ- percent. We find that this leads to similar The results from our analysis suggest that ment in 1998. The results remained overall increases in the employment of high-skill higher skilled workers are more likely to very similar. and low-skill workers and no change in reap the benefits of improved perform- 6. We define high-skill and low-skill workers the skill mix of industries (ratio of high- ance of Brazilian export industries. The using the distribution of monthly labor income. evidence points to the fact that educa- High-skill (resp. low-skill) workers are defined skill to low-skill workers). However, while the earnings of high-skill workers increase tional policies and programs that promote as those workers earning above (resp. below) the median monthly labor income in a the by 2.4 percent for each additional export human capital accumulation and skill industry where they work for a given year. destination, the earnings of low-skill work- upgrading will not only increase productiv- 7. Each row and column correspond to a differ- ers remain unchanged, which leads to a ity in Brazilian firms but also respond to ent regression. rising demand for skills Q u a r t e r l y K no w l e d g e R e po r t Page 19 BRAZIL’S MISSING MIDDLE BY MAKHTAR DIOP FOLHA DE S. PAULO (SUNDAY, JANUARY 2) The best translation for the “missing vate investment capacity. In this regard, versity education is stifling opportunities middle” is Brazil’s own middle class – the country is rising to the challenge. for millions of young people. Similarly, having grown substantially, its full po- while Brazil has developed a highly pro- At the same time, however, while the tential is yet to be tapped ductive private sector, there is virtually no enormous social and economic gains upward movement of small and medium A few days ago, a front page article in [Brazilian newspaper] Folha de S. Paulo noted that the “C” class is already made in recent years have largely benefit- ted the poor, they have not yet managed to eliminate Brazil’s fundamental inequali- enterprises to allow for competition with big businesses. The former are unable to boost productivity and gain access to fi- buying more household electrical appli- ties between north and south and be- nancing. Many promising initiatives end ances than the “A” and “B” classes, a tween the poor outskirts and the wealthy up falling by the wayside. situation that marks an important inflec- neighborhoods of the big cities. The in- tion point for Brazil, particularly at a come disparity between the wealthiest 10 The best illustration for the lack of those time when the country is actively looking percent and the poorest 10 percent re- “middles” is the middle class itself. While to a future of growth and development. mains vast – on the order of 4,400 per- its ranks have grown substantially, its full cent. potential is yet to be developed, owing in In many respects, Presi- large measure to the services dent Dilma Rousseff will to which it lacks access. face a development land- scape that is starkly differ- As Brazil continues to lift ent from her predecessors. individuals out of poverty, it needs to redouble its efforts Over the past two decades, to create the institutions and Brazil managed to over- opportunities to the new come many of its most middle class in areas that basic social and economic transcend income. Viewed challenges. It achieved from this angle, the country’s universal basic education, divide between rich and poor strengthened the fiscal remains firmly in place. base needed for growth, investment, and job crea- Brazil has succeeded in com- tion, and lifted tens of mil- bining economic and social lions of people out of pov- progress, making domestic erty. demand and poverty reduc- tion the drivers of the country’s growth. These are remarkable feats. However, Consequently, Folha’s headline serves to paradoxically, the sheer scope of these underscore both the intractable nature of This is an essentially homegrown agenda, achievements makes the road ahead these challenges and the progress being and represents another step on the road more difficult. An analogy can be made made, a situation that reflects Brazil’s to becoming a developed country. Ad- with a building – the higher up you go, “missing middles” — gaps between two dressing the variety of “missing middles” the more difficult it becomes to build extremes. Brazil will have to tackle these will be an effective way to overcome new the next floor, and the more important distortions, which pose a threat to the challenges and ensure that new opportu- the foundations become. country’s potential. nities are within everyone’s reach. Brazil will now face new challenges in Now that basic education is being pro- Our hope is that the “class C” will con- the areas of fully eliminating poverty, vided to virtually all children, secondary tinue to buy more household appliances, providing opportunities for all children, education access and quality are begin- cars, houses, education, and health care and expanding infrastructure in a sus- ning to loom large in the context of Bra- for their children, and that this becomes tainable and planned manner. These zil’s competitiveness. the norm and does not justify headlines challenges call for ongoing institution any more The obvious gap between basic and uni- building and the strengthening of pri- EDITORIAL (CONTINUED FROM PAGE 1) highlights another important aspect of generating relevant knowledge for a MIC: the need to bring it back to the benefit of the whole Bank. This goes beyond the knowledge itself, to include the means by which it was generated and made useful to the client. The Bank has a comparative advantage in embodying and sharing lessons of international experi- ence, and is also in a unique position to be the central hub for South-South exchanges. But to be effective in both, especially in a decentralized structure, it is imperative to incentive the flow back of ideas to Washington. The Quarterly Knowledge This is what we tried to do in the First Brazil Knowledge Day. Report is a publication of the World Bank With it, we hoped to contribute to the leveraging of knowledge work in other MICs, so that the Brazil event was really about much more than just one country. Chairs and commentators at he event, from outside the Bank and across its sectors, were unanimous that these two dimensions – the knowledge generated in the Brazil context, and the strategy and processes behind this generation – are very useful for the institution as a whole, and consequently to our clients. I hope you enjoy Mauro Azeredo’s reporting on this important event, and also the rich selection of analytical pieces in this edition: from Romero Rocha’s counter intuitive results on closing the wage gap and Barbara Bruns’ deep analysis of Brazil’s education sector, to Abdoulaye Sy’s and Barbara Sign up to receive BET Farinelli’s linking of export markets to total factor productivity and Guilherme Lichand’s dense analy- reports: sis of state TFP dynamics. Denise Marinho’s excellent interviews with Governor Antonio Anastasia, the man behind Minas Gerais’ famous Choque de Gestão, and Daniel Lederman, on the challenges betmail@worldbank.org of an effective industrial strategy in Brazil, top this edition. Enjoy your reading. Contacts: Makhtar Diop Tito Cordella Brazil Country Director Mauro Azeredo WB Link: 5761-1000 IN THE LOOP Ph.: +55 61 3329-1000 Some of the quarter’s noteworthy events BET Publications Events Projects Approved BET Daily Report: the most Brazil Disaster Risk Management High Level Rio de Janeiro Metropolitan Urban and important daily economic Workshop with Cities Minister—April 11, Brasília Housing Project —$ 485 million, March news and indicators about the Becoming Old in an Older Brazil, seminar and 15—Project will benefit 2 million poor report launch with BNDES President Luciano Brazilian economy [internal]. Coutinho–April 6, Rio de Janeiro people who live in informal low income settlements in Rio de Janeiro. The fi- Brazil Monthly Report : Bloomberg Brazil Economic Summit, Makhtar Diop presentation—March 29, São Paulo nancing, a request of President Dilma updated information on Rousseff, follows the record January economic developments, Brazil BBL: “Does What you Export Matter?” with policy challenges, and Daniel Lederman—March 2, Brasília flooding that killed over 900 in the State. economic trends [internal]. Brazil BBL: “Trade Liberalization, Firm Heteroge- Federal University Hospital Moderniza- neity and Wages” with Pravin Krishna—February tion Project—$ 150 million, March 31— Quarterly Knowledge Report: 28, Brasília will support modernizing management information about the Bank’s Land Use Low Carbon Seminar with Embrapa— and infrastructure of 46 federal univer- analytical work and short February 21, Brasília sity hospitals throughout Brazil, helping knowledge pieces on policy Transport Low Carbon Seminars —February to improve the quality of services deliv- relevant questions. 17,18, São Paulo and Brasília ered to millions of people in the country. Brazil BBL: “Eradicating Extreme Poverty” with ---- Serguei Soares—February 1, Brasília Waste Picker Inclusion Initiative—$ 2.7 The Quarterly, Monthly and Brazil BBL: “Using IBGE Censuses Through million Japan Social Development Fund Daily Brazil reports are in the MySQL” with Daniel Simplicio and Guilherme Grant, associated with the IBRD Inte- Lichand—January 19, Brasília grated Solid Waste Management and Brazil Internet and Intranet: Press launch of the Brazil Education Report with Carbon Finance Project, January 28—first http://www.worldbank.org/br Education Minister—December 13, Brasília phase will benefit approximately 5,000 http://go.worldbank.org/EX2K94L7V0 Brazil BBL: “Treasury Briefing” with Miguel people who depend on the informal recy- Navarro-Martin—November 24, Brasília cling of solid waste in urban landfills.