Policy Research Working Paper                          10897




         A Literature Review on Productivity
                and Growth in Brazil
                                Fernando Veloso
                                Gabriel Zaourak




Macroeconomics, Trade and Investment Global Practice
September 2024
Policy Research Working Paper 10897


  Abstract
  Productivity growth in Brazil has not kept pace with devel-                       evidence on misallocation and growth dynamics. Addition-
  oped and emerging economies, despite progress in achieving                        ally, the paper analyzes the direct and indirect effects of
  macroeconomic stability and implementing reforms in                               several reforms, such as trade liberalization and formaliza-
  product and input markets. This paper reviews the literature                      tion policies. Based on the main lessons from this review,
  on reforms and growth in Brazil to understand the factors                         the paper proposes a potential reform agenda to improve
  that have hindered productivity growth. The paper discusses                       improve productivity in Brazil in the coming years and
  the impact of competitive distortions on productivity, the                        narrow the income gap with developed countries.
  effect of structural change on productivity growth, and the




 This paper is a product of the Macroeconomics, Trade and Investment Global Practice. It is part of a larger effort by the
 World Bank to provide open access to its research and make a contribution to development policy discussions around the
 world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may
 be contacted at gzaourak@worldbank.org.




         The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
         issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
         names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
         of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
         its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.


                                                       Produced by the Research Support Team
   A Literature Review on Productivity and Growth in
                         Brazil
                                                *                           „…
                          Fernando Veloso and Gabriel Zaourak

                                            World Bank




JEL Classification:   D22, F17, O11, O30, O40, O47.

Keywords:      Growth, misallocation, Innovation, Empirical Studies of Economic Growth,

Cross-Country Convergence, T                .
                            rade, Informality




   * Email: fernando.veloso@fgv.br
   „ Thispaper was prepared as a background paper for the report Brazil 2040. The views expressed herein
are only my own and should not be attributed to the World Bank, its executive directors, or the countries
they represent. All errors are omissions are our own.
   … Email: gzaourak@worldbank.org
1    Introduction
Since the mid-1990s, Brazil has made progress in achieving macroeconomic stability and im-

plemented reforms in product and input markets. However, productivity growth has been

disappointing, lagging behind developed and emerging economies. Demographic dividends

are nearly exhausted, with Brazil being one of the fastest aging societies in the world. This

implies that in the next decades improvements in living standards will crucially depend on an

acceleration of productivity growth. At the heart of Brazil's low and stagnant productivity is

a business environment that discourages competition and induces misallocation of resources.

Alongside important reforms, such as trade liberalization in the 1990s and improvements in

credit collateral and bankruptcy procedures in the 2000s, several competitive distortions were

created in the last two decades. In particular, credit subsidies, tax exemptions and local con-

tent requirements were introduced to benet specic sectors and rms, distorting labor and

capital markets rather than fostering competition. As a result, the development model based

on state intervention and import-substitution that prevailed from the 1930s until the 1980s

is still present to a large extent. Instead of a level playing eld, the business environment

favors incumbents and hampers entry. The lack of competition in both domestic and exter-

nal markets in turn reduces eciency and undermines productivity growth. The objective of

this paper is to review the literature on reforms and growth in Brazil in order to understand

why there has been so little productivity growth in the last decades. We will discuss reform

mechanisms and how their eects depend on other determinants of productivity, such as

the schooling level of the labor force.   We will also analyze the evidence on the impact of

competitive distortions on productivity. The goal is not only to assess their direct eect, but

also to analyze how they may have dampened the potential impact of some reforms, such as

trade liberalization and formalization policies. Based on the main lessons from this review,

we will discuss a potential reform agenda to improve productivity in the coming years. The

paper is structured as follows. Section 2 will present the main stylized facts on aggregate and

sectoral productivity in Brazil and a discussion of the contribution of structural change in

the last decades. Section 3 will discuss the evidence on misallocation and growth dynamics.

Section 4 will analyze the eect of several competitive distortions on productivity. Section 5

will review the literature on reforms and productivity growth. Finally, section 6 will discuss

a potential reform agenda to improve productivity in Brazil in the next decades and reduce

the income gap with developed countries.




                                               2
2      Stylized facts on aggregate and sectoral productivity
       growth
This section documents the main productivity facts in Brazil at the aggregate and sectoral

level since the mid-1990s.



2.1 Aggregate productivity growth
Economic growth in Brazil since the mid-1990s has been driven largely by factor accumula-

tion, namely an expansion of the employed labor force, human and physical capital, with only
                                                                                      1
a small contribution from improvements in total factor productivity (TFP).                An analysis of

the decomposition of GDP growth shows that during the 1997-2005 period, factor accumu-

lation had an annual contribution of 3.5 percent to GDP growth, whereas TFP declined 1.1
                               2
percent per year (Figure 1).       The GDP growth acceleration from 2.4 percent to 4.3 percent

per year between 1997-2005 and 2006-2011 was entirely due to TFP, since the contribution of

factor accumulation remained around 3.5 percent. Between 2012 and 2019 there was a sharp

deceleration of annual GDP growth to 0.3 percent, reecting a decline in the contribution of

factor accumulation to 2.5 percent per year and a collapse of TFP (-2.2 percent per year).

The reduction of factor accumulation in the recent subperiod reects lower physical capital

accumulation and a signicant slowdown in labor force growth. The contribution of human

capital has been positive throughout the 1997-2019 period, averaging a little more than 1
                   3
percent per year.

     The negative contribution of TFP during the 2012-2019 period may in part reect the

fact that TFP is calculated as a residual, such that the it may appear that the recession of

2014-2016 was caused by a negative contribution of productivity, when in fact it may have

been driven by other unobserved factors such as an increase in spare capacity. Nevertheless,

the fact that TFP growth was still negative in the 2017-2019 period seems to suggest that

its decline in recent years has not only been due to cyclical factors (Figure 2).

     International comparisons conrm that Brazil's negative TFP growth stands out in com-

parison to other emerging and developed economies (Figure 3). Whereas TFP declined at

an annual rate of 1 percent per year between 1997 and 2019 in Brazil, it increased around

2 percent per year in China and 2.3 percent in India. Even though there was also a TFP

    1 See
        Bonelli et al. (2017)for a review and analysis of the productivity evidence in Brazil.
    2 Figures
            1to 5 are updated versions of gures presented in Qian et al. (2018)and Dutz (2018).
   3 The data used to perform the growth decomposition exercises was obtained from the Penn World Table
10.0. The human capital index is constructed using the average years of schooling for the population aged
15 and older and rates of return based on cross-country Mincerian wage regressions.


                                                   3
      Figure 1: Real GDP growth and contributions, percentage points




       Source: Penn World Tables 10.0. World Bank sta calculations.




Figure 2: Real GDP growth and contributions, percentage points (annual data)




       Source: Penn World Tables 10.0. World Bank sta calculations.




                                     4
          Figure 3: Real GDP growth contributions, percentage points, 1997-2019




              Source: Penn World Tables 10.0. World Bank sta calculations.



decline in the Latin America and the Caribbean (LAC) region, it was much smaller than in

Brazil (-0.2 percent per year). Average annual TFP growth in the OECD was modest but

positive (0.5 percent) and much stronger in East Asia (1.6 percent).

   Low TFP growth is reected in Brazil's labor productivity trajectory, which grew less

than advanced countries and other emerging countries since the mid-1990s. Between 1995

and 2019, value added per worker in Brazil has been growing at about the same rate as the rest

of Latin American and Caribbean (LAC) countries, but much below low and middle-income

countries (Figure 4).   After catching up with developed economies in the 2000s, Brazil´s

productivity trajectory diverged in the last decade.

   The lack of sustained productivity growth in Brazil is particularly worrisome because the

contribution of labor to per capita Value Added (VA) growth will probably decline in the

next decades due to the fast aging of the Brazilian population. Per capita annual VA growth

increased signicantly from 1995-2002 (0.7 percent) to 2003-2010 (3.4 percent), and turned

negative in 2011-2018 (1.8 percent) (Figure 5). Between 1995 and 2002, the change in labor

productivity was slightly positive (0.1 percent per year). There was a signicant acceleration

to 2.4 percent in 2003-2011, but it was followed by a decline in 2011-2018 (-0.4 percent per

year).

   Throughout the period, increases in employed labor force, as a result of demographic

growth and in labor force participation, played an important role. However, the contribution

of the increase in the share of the working age population declined from 0.7 percent per year

in 1995-2002 to 0.5 percent in 2003-2010 and only 0.2 percent in 2011-2018, reecting the

aging of the labor force. The increase in the participation rate also had a positive contribution

to per capita VA growth, varying between 0.1 percent per year (2003-2010) and 0.3 percent




                                               5
          Figure 4: Labor productivity, 1995=100




        Source: WDI. World Bank sta calculations.




Figure 5: Decomposition of Growth in Per Capita Value Added




        Source: WDI. World Bank sta calculations.




                            6
per year (1995-2002 and 2011-2018). On the other hand, the change in the employment rate

has been negative during most of the 1995-2018 period, with the exception of the 2003-2010

subperiod.



2.2 Sectoral labor productivity growth
In the last decades, agriculture was the only major sector that had signicant and sustained
                                            4
labor productivity growth (Table 1).             Whereas the industry sector had negative annual
                           5
growth (-0.4 percent)          in 1996-2020, services´ productivity was stagnant (0.1 percent per

year) throughout the period. Since services account for more than 70 percent of total em-

ployment in Brazil, the lack of productivity growth in this sector contributed signicantly

to the sluggish aggregate productivity growth. The subperiod 2003-2010 was characterized

by substantial productivity growth in several service activities, such as retail and nancial

intermediation.       However, this turned out to be an exception, since services had negative

productivity growth both in 1996-2002 and 2011-2020.



   Table 1: Sectoral Labor Productivity Growth, Brazil  1996-2020 ( percent per year)

             Years                         1996-2002       2003-2010   2011-2020    1996-2020

             Agriculture                        5.3           5.7         6.2           5.8
             Mining                             6.3           2.2         2.5           3.5
             Manufacturing                      -3.7          0.3         0.1          -0.9
             Construction                       -2.8          0.6        -1.4          -1.2
             Utilities                          3.7           2.4         3.2           3.1
             Retail                             -2.5         2/1         -0.2          -0.1
             Transportation                     -1.2          0.5        -2.4          -1.1
             Information Services               -3.1         -1.7         1.7          -0.7
             Financial Intermediation           0.6           5.0         0.3           1.9
             Other Services                     -1.2          0.4        -1.0          -0.6
             Real Estate                        2.3          -0.5         0.8           0.8
             Government Services                0.7          -0.2        -1.4          -0.4

             Agriculture                        5.3           5.7         6.2           5.8
             Industry                           -2.6          0.5         0.5          -0.4
             Services                           -0.5          1.1        -0.4           0.1

               Source: Observatório da Produtividade Regis Bonelli at FGV IBRE.




   4 Labor productivity is measure as value added per worker. The mining sector also experienced sustained
growth in the last decades but it accounts for a much smaller proportion of value added and employment.
   5 The industry sector includes mining, manufacturing, utilities and construction. Manufacturing had
negative productivity growth during 1996-2020 (-0.9 percent per year).

                                                       7
                  Figure 6: Labor shifts and productivity changes (1995-2020)




              Source: Observatório da Produtividade Regis Bonelli at FGV IBRE.



2.3 Contribution of structural change
Figure 6 shows changes in employment shares and the relative productivity of sectors, mea-

sured as the log of the ratio between sectoral productivity and average productivity between

1995 and 2020. For positive gains to occur through structural change, sectors would either

be located in the top-right corner (e.g. nancial intermediation) where labor shifts into rel-

atively high-productivity sectors or the lower-left quadrant (e.g.               agriculture) where labor

shifts out from low productivity sectors.

    A steady shift in employment from agriculture to services underpins the productivity

gains from structural change.         Despite its strong productivity growth in the last decades,

agriculture is still a low-productivity sector. Hence its 14 percent employment decline con-

tributed to aggregate productivity growth, as well as the employment expansion in higher-
                                                                                          6
productivity sectors such as other services, retail and government services.                  Computing the

counterfactual aggregate productivity using the 1995 employment distribution and actual

labor productivity dynamics and labor force increases, Figure 7 shows that aggregate labor

productivity increased purely due to structural change by 4.4 percent in 2020, which corre-

sponds to roughly one quarter of accumulated aggregate productivity growth between 1995

and 2020. Hence structural change had a positive but limited contribution to productivity

   6 Using data from Brazil, Bustos et al. (2019) show that, while positive in the short-run, improvements in
agricultural productivity can generate specialization in less-innovative industries and have negative eects on
manufacturing productivity in the long-run. Bustos et al. (2020) nd that agricultural productivity growth
in Brazil generated an increase in savings, and that capital was reallocated toward urban regions, where it
was invested in the industrial and service sectors.



                                                      8
                      Figure 7: Labor productivity: actual vs counterfactual




              Source: Observatório da Produtividade Regis Bonelli at FGV IBRE.


                                 7
growth in the last decades.

     Using productivity data for 35 sectors from the Social Economic Accounts, Veloso et al.

(2017) perform a counterfactual exercise which shows that if Brazil had the same sectoral

labor allocation as the U.S., aggregate productivity would increase by 68 percent. If instead

Brazil had the same productivity as the U.S. in all sectors, aggregate productivity would be
                       8
430 percent higher.        A similar comparison with a sample of developed countries suggests that

Brazil could increase aggregate productivity by 50 percent if it had the same sectoral labor

allocation and by 192 percent if it had the same sectoral productivity levels. This suggests

that the main reason for the low level of aggregate productivity in Brazil in comparison

to developed countries is the low productivity level in most sectors and not the sectoral

allocation of labor.




3      Misallocation and productivity dynamics
The previous evidence indicates that even though there are potential productivity gains

associated with changes in the sectoral allocation of labor, in order to reduce signicantly the

    7 PerezSebastian et al. (2020) analyze the relationship between grid electrication and structural trans-
formation in Brazil. In their model, increasing electricity availability induces a reallocation of inputs to more
productive activities by generating higher returns and lowering entry costs in sectors with greater infras-
tructure intensity. The results of modeling and econometric analysis based on Brazil's data over the period
1970-2006 show that the manufacturing sector beneted the most, followed by services and agriculture.
   8 The Social Economic Accounts contain sector-level data on employment, capital stocks, gross output and
value added at current and constant prices and are part of the World Input-Output Database. Veloso et al.
(2017) use data from 2009 in these counterfactual exercises. In order to make international comparisons at the
sectoral level, the authors use sectoral PPP indexes obtained from the Groningen Growth and Development
Centre Productivity Level Database, which are based on the methodology developed by Inklaar and Timmer
(2014).


                                                       9
                            Figure 8: Distribution of Labor Productivity




           Source: Barbosa Filho and Correa (2017). Reproduced from Dutz (2018).
                      Note: Labor productivity (in logs) is expressed in PPP.



productivity gap with developed countries, Brazil will have to improve productivity in most

sectors. This highlights the fact that factor misallocation in each sector may have important

implications for the low level of sectoral and aggregate productivity. In order to shed light on

these issues, this section analyzes the literature on misallocation and productivity dynamics

in Brazil.   Moreover, it will discuss the evidence on the relation between informality and

productivity.



3.1 Productivity dispersion and misallocation
Barbosa Filho and Correa (2017) use data for emerging economies from the World Bank

Enterprise Survey and compare the distribution of rm labor productivity in Brazil with
                                                                                    9
China, Chile, Colombia, Mexico, Peru and the Russian Federation.                        They show that the

dispersion of labor productivity in Brazil is higher than in the comparator countries. More-

over, Brazil has a large concentration of low-productivity rms in comparison to the other

emerging economies (Figure 8).

    The authors also analyze the productivity distribution for several sectors, including tex-

tiles, garments, food, machines and equipment, chemicals and pharmaceuticals, automobiles,

retail, hotel and restaurants, construction, transportation and other services. The results con-

   9 Due  to limitations in the available data, the information for Brazil is from 2009. For the other countries,
it varies between 2008 and 2010. The productivity measure is labor productivity expressed in purchasing
power parity to allow for international comparisons.


                                                       10
                        Figure 9: Distribution of Total Factor Productivity




                   Source: Vasconcelos (2017). Reproduced from Dutz (2018).



rm the patterns documented at the aggregate level. There is a large productivity dispersion

and high concentration of low-productivity rms in most sectors in Brazil in comparison

to the other countries. This indicates that these productivity facts do not result from the

sectoral composition of the Brazilian economy. It indicates instead that they are systemic

characteristics of the business environment.
                                                                                                           10
    This evidence suggests that misallocation is possibly high in many sectors in Brazil.

Some papers investigate this possibility for specic sectors. Vasconcelos (2017) documents

the existence of resource misallocation in the Brazilian formal manufacturing sector.                    Us-

ing the methodology developed by Hsieh and Klenow (2009) and rm-level data from the

National Survey of Industry (Pesquisa Anual da Indústria, PIA) for 1996-2011, the author

nds evidence of misallocation in the manufacturing sector during this period. First, Vas-

concelos shows that there is a large TFP dispersion among manufacturing rms and that the

proportion of rms with low TFP increased over time (Figure 9). He then calculates how

much manufacturing output in Brazil could increase if capital and labor were reallocated

to equalize marginal products across rms to the extent observed in the U.S. He nds that

moving to U.S. eciency would increase manufacturing output by 160-180 percent. He also

uses production factors parameterized using the Brazilian rm-level data. In this case, he

nds that moving to Brazilian Firm-Specic eciency would increase manufacturing output

by 110-130 percent. The results also show that misallocation has increased since 2005.

    Vasconcelos (2017) also shows that the allocative ineciency of capital in the manufactur-

  10 Itshould be noted, however, that productivity dispersion does not necessarily imply that there is misal-
location. For instance, it is possible that dispersion in observed measures of labor productivity (or TFP) are
associated with variable markups. See Restuccia and Rogerson (2017) and Cusolito and Maloney (2018).


                                                     11
ing sector is relatively high. The rm's capital wedge in the 90th percentile of the productivity

distribution is 4 (U.S. Industry Share) or 3 (Brazilian Firm-Specic Share) times greater than

the capital wedge for rms in the 10th percentile.          The rm's labor wedge is considerably

smaller, varying between 2 (U.S. Industry Share) and 2.75 (Brazilian Firm- Specic Share).

De Vries (2014) analyzes the existence of resource misallocation in the Brazilian formal retail

sector.    The author uses rm-level data from the National Retail Survey (Pesquisa Anual

de Comércio, PAC) constructed by Instituto Brasileiro de Geograa e Estatística (IBGE).

The author applies the Hsieh and Klenow (2009) methodology to study changes in resource

allocation in Brazil's retail sector during the period from 1996 to 2006. The ndings suggest

there are large potential output gains from the reallocation of resources to the most ecient

retailers. Improvements in resource reallocation may increase TFP levels by a factor of two.

However, he does not nd any allocative eciency improvements for the retail sector be-

tween 1996 and 2006. Brazil opened up its retail sector in the World Trade Organization's

1995 General Agreement on Trade in Services, and also within MERCOSUR. Furthermore,

the participation of foreign capital in Brazilian retail rms was freed from restrictions in

the Sixth Constitutional Amendment of 1995. It could thus be expected that these reforms

would result in a productive reallocation through the expansion of modern retail chains and

the growth of small successful retail businesses. According to De Vries (2008), however, the

Brazilian experience suggests a dierent pattern.          Retail chains did not replace mom-and-

pop stores during the period following reforms. The author argues that this may be partly

explained by business regulations, such as zoning laws, and diculties in setting up national

distribution systems because of the low quantity and quality of rail and road networks. The

limited role of reallocation in Brazil's retail sector may explain its low productivity growth.



3.2 Informality and productivity
The previous analysis provided evidence of signicant misallocation of factors of production

among Brazilian formal rms. However, in order to have a broader picture of misallocation
                                                                  11
in Brazil, we have to consider the large informal sector.              Meghir et al. (2015a) use data

from the Monthly Employment Survey (Pesquisa Mensal de Emprego, PME), which provides
                                                                                                       12
a rotating panel of individuals sampled from the six main metropolitan regions of Brazil.

They show that the lowest part of the productivity distribution is populated only by informal

rms. However, over a large segment of the support of the productivity distribution, formal

and informal rms coexist. The increased probability of detection for larger informal rms

  11 See Ulyssea (2020b) for a recent review of the literature.
  12 PME  has been discontinued in 2016. It was replaced by PNAD Contínua, which also has a rotating panel
but is representative at the national level.

                                                   12
together with the regulatory costs means that informal rms are much more prevalent in

lower levels of productivity and are much smaller. Ulyssea (2018) develops and estimates an

equilibrium model of informality using data from the Survey of the Urban Informal Sector

(Pesquisa de Economia Informal Urbana, ECINF), a survey of informal rms conducted

by IBGE in 2003.       He considers two margins of informality.           First, the decision of rms

to not register their business (extensive margin).          Second, the choice to hire workers o

the books (intensive margin).        The model integrates the three leading views of informal

rms La Porta and Shleifer (2014) in a unied setting.               The rst view argues that the

informal sector is a reservoir of potentially productive entrepreneurs who are kept out of

formality by high regulatory costs. The second sees informal rms as parasite rms that

are productive enough to survive in the formal sector but choose to remain informal due

to the cost advantages of not complying with taxes and regulations. The third argues that

informality is a survival strategy for low-skill entrepreneurs who are too unproductive to ever

become formal. The results show that the largest proportion of informal rms in Brazil (48.8

percent) correspond to the view that informal rms correspond to low-skill entrepreneurs

who are too unproductive to ever become formal.             The view that informal rms are to a

large extent parasite rms also corresponds to a signicant fraction of informal rms (41.9

percent). On the other hand, the view that informal rms are potentially productive rms

hindered by high regulatory costs corresponds to only 9.3 percent of informal rms in Brazil.

This evidence on heterogeneity of rm productivity in the Brazilian informal sector suggests

that the potential eects of formalization policies focused on the reduction of formal sector

entry costs are probably limited. On the other hand, human capital policies that increase

the skill of informal entrepreneurs may have signicant eects. We will return to these issues

when we discuss the evidence on formalization policies in section 5.



3.3 Productivity dynamics
Using data from the World Bank Enterprise Survey, Barbosa Filho and Correa (2017) show
                                                                                                       13
that the average size of Brazilian rms is small in comparison to other emerging economies.

The average number of workers per rm in Brazil is similar to the one found in Russia, but

considerably smaller than the average size in Chile, Colombia and China. The authors also

note that older rms are not much larger than young rms in Brazil. Moreover, productivity

does not increase with age. This suggests that the competitive selection mechanism is weak

in Brazil. As a result, more productive rms do not grow and low-productivity rms stay

  13 This evidence was obtained from the same dataset that Barbosa Filho and Correa (2017) used to compare
the distribution of labor productivity in Brazil to other emerging economies.


                                                   13
in the market, contributing to the sluggish growth of aggregate productivity.                    Cavalcanti

et al. (2021) present further evidence on rm dynamics using data from Relação Annual de

Informações Sociais (RAIS), a matched employer-employee administrative dataset covering

all formal rms in Brazil that follow rms and workers over time. They nd that rm size in

Brazil is increasing but concave in age over the rst 15 years of rm life, on average rising 50

percent relative to its entry size. This growth is signicantly less than the lifecycle growth for

manufacturing plants reported by Hsieh and Klenow (2014) for the U.S., which show 8-fold

average growth over 30 years, but greater than the roughly 1.25-fold increase reported for

India. Ulyssea (2020a) analyzes both formal and informal rm dynamics in Brazil, combining

rm-level data with a structural model. The paper documents that in, both sectors, rms

display an increasing and concave age-size prole. However, the growth in size is signicantly

higher for formal rms. The results show that after 10 years there is a 50 percent growth in

the sample that includes only formal rms and their formal workers in contrast with the 20

percent average growth that one obtains when using both formal and informal rms. The

author shows that the age-size prole using only formal rms and workers in Brazil is very

similar to the one documented for Mexico by Hsieh and Klenow (2014). However, once he

incorporates informal rms, the age-size prole of Brazilian rms becomes much closer to that

of Indian rms, which is remarkably atter. These facts show that dynamic selection takes

place in both sectors but is substantially weaker in the informal sector. They also suggest

that the lack of dynamism found in the Indian data might also be present in other high-

informality countries. Failing to incorporate informal rm dynamics may therefore lead to a
                                                                                                         14
substantial underestimation of the lack of dynamism among developing countries' rms.




4       Competitive distortions and productivity
There are signicant barriers to competition in Brazil, many of them policy-induced. This

section will discuss the evidence on the eect of several competitive distortions on produc-

tivity in Brazil.



4.1 Credit subsidies, nancial frictions and lack of competition in
    credit markets
In the presence of nancial market imperfections, the process of ecient allocation of re-

sources can break up. Financial frictions can reduce aggregate productivity via two chan-

    14 See
        Akcigit et al. (2021) for an analysis of the dierences in the competitive selection mechanism between
India and the United States and their aggregate implications.

                                                     14
nels. First, they may distort entry and technology adoption decisions and thus reduce the

productivity of individual rms. Second, nancial frictions may generate dierences in the

returns to capital across individual rms, thus generating eciency losses through misalloca-

tion (Banerjee and Duo (2005); Restuccia and Rogerson (2008); Hsieh and Klenow (2009)).

    One potential source of nancial frictions are policy-induced interventions. During several

decades the Brazilian government intervened heavily in credit markets. As will be discussed

later, several credit reforms contributed to a substantial increase in private credit in the 2000s.

However, after the 2008-2009 nancial crisis there was a signicant increase in the expansion

of subsidized credit provided by public banks.            The main source of long-term nancing to

rms in Brazil is the Brazilian Development Bank (Banco Nacional de Desenvolvimento

Econômico e Social, BNDES), a government-owned development bank.

    Several papers do not nd positive eects of BNDES nancing on productivity, especially
                 15
for large rms.       Even when there is evidence of a positive impact on productivity, they

do not seem to be strong enough to be cost-eective when the sizable subsidy costs are

considered. We will next briey discuss a few papers that illustrate the main issues.

    Bonomo et al. (2015) use a large repository of loan contracts between banks and rms

to investigate the eect of the expansion of BNDES credit expansion between 2004 and

2012.    Their results show that larger, older and less risky rms beneted most from the

government sponsored credit expansion.           Additionally, the eect on investment appears to

be insignicant for publicly traded rms. These results are consistent with the hypothesis

that those established public rms in part substituted more expensive credit with subsidized

loans.

    Calice et al. (2018) study the impact of BNDES credit on resource misallocation in Brazil,

using manufacturing rm-level data from 2003 until 2014. The paper rst estimates measures

of resource misallocation based on Hsieh and Klenow (2009), documenting high variation in

rms' capital and output distortions. It then estimates the eect of access to BNDES loans on

distortions and their dispersions. The analysis nds some preliminary evidence that the use

of BNDES credit is not associated with a more ecient allocation of resources. The lower

cost of BNDES loans reduces the marginal cost of capital and induces rms to reallocate

inputs from labor to capital. This eect is amplied for more nancially dependent rms.

The authors point out that their results only reect the eectiveness of a particular credit

line, FINAME, which nances the acquisition of machines and equipment in the domestic

  15 Bank  (2017b) reviews the costs and benets of business support policies in Brazil, including subsidized
credit policies by public banks. Bank (2017a) provides a detailed evaluation of the impact on productivity
of subsidized credit programs and other business support policies in Brazil. Calice et al. (2018) provide a
review of the literature on the eects of BNDES credit on several performance indicators and analyzes its
eects on misallocation.

                                                     15
market. They suggest that though relatively large rms are the main recipients of BNDES

nancing, the impact of BNDES on smaller rms may be dierent.

    In fact, Cavalcanti and Vaz (2017) use a variation in access to a targeted loan program

from BNDES to assess the causal eect of better credit conditions on investment and produc-

tivity of Brazilian manufacturing rms. The estimated causal eects point to positive eects

on investment and productivity for small and medium rms. The authors argue that their

results support the hypothesis that nancial constraints for long-term investment matter for

small and medium rm's productivity and investment decisions. In addition to signicant

government intervention in credit markets, Brazil also has very large private lending spreads

(the dierence between lending rates and the deposit rate). According to the International

Financial Statistics, the average interest rate spread is approximately 0.7 percent in Japan,
                                                                                                 16
3 percent in the United States, 10 percent in Uruguay and 40 percent in Brazil.                       High

spreads may arise due to several reasons, such as poor collateral, credit taxes, regulatory

costs and lack of competition. The banking sector is very concentrated in many countries,

but concentration in Brazil seems to be particularly high and increasing over time. Between

the mid-1990s and 2016, the share of assets held by the 5-largest banks in Brazil increased
                                                17
from 50 percent to more than 85 percent.

    Joaquim and Van Doornik (2019) use heterogeneous exposure to large bank mergers to

estimate the eect of bank competition on both nancial and real variables in local Brazilian

markets.    Specically, they use merger and acquisitions episodes of large Brazilian banks

between 2005 and 2015 as a source of exogenous variation of competition in local banking

markets to identify the causal eect of bank competition. The authors nd that a reduction

in bank competition results in a signicant increase in lending spreads and a decrease in loan

volume.    They also show that bank competition has real eects.               A 1 percent increase in

spreads leads to a 0.2 percent decline in employment. Moreover, they show that if Brazilian

spreads fell to world levels, output would increase by approximately 5 percent.

    In addition to a high average level, there is a large dispersion in credit spreads to Brazilian

rms. Cavalcanti et al. (2021) develop a quantitative dynamic general equilibrium model and

calibrate it to the Brazilian data for the period 2005-2016. They use data from the Brazilian

credit registry, a condential loan level data set covering all the credit operations in Brazil

and containing information on loan characteristics and interest rates.               They merge these

data with Brazil's linked employer-employee administrative dataset to examine how interest

  16 SeeCavalcanti et al. (2021).
  17 SeeJoaquim and Van Doornik (2019). They report that during this same time span the share of assets
held by the 5-largest banks in the U.S. increased from 30 percent to more than 45 percent. Averaging across
countries, the share of assets held by the 5 largest banks is 78 percent.



                                                     16
rates and loan size vary with rm characteristics. The results show that, even controlling for

several loan characteristics, the loan interest rate and the volume of credit vary considerably

with rm characteristics, such as rm size and age.           In particular, young and small rms

pay higher interest rates.    The calibrated nancial frictions lower output per capita by 31

percent relative to a frictionless credit benchmark. TFP is 25 percent lower and capital is

26 percent lower relative to the frictionless credit benchmark.           Counterfactual simulations

reveal that spread variation coming from direct intermediation costs drive the vast majority

of impacts. The evidence thus suggests that the high average level and large dispersion of

spreads in Brazil has signicant negative eects on output and productivity. There is also

some evidence that lack of competition and high intermediation costs may contribute to high

spreads. In Section 5 we will examine the potential role of poor collateral in also accounting

for these patterns.



4.2 Labor market frictions
4.2.1 Labor informality
The previous discussion about informality emphasized the decision of rms to not register

their business (extensive margin) and their implications for aggregate productivity. Next we

will focus on distortions that may aect the rm´s choice to hire workers o the books

(intensive margin).

    Meghir et al. (2015b) show that search frictions reduce the competition for workers and

make it harder for workers to locate to higher productivity rms.              In their model, search

frictions increase the prevalence of low-productivity rms and reduces the probability of

workers matching with higher productivity rms, thus reducing output.

    One of the most important stylized facts of the Brazilian labor market is the decline in

labor informality in the 2000s. Between 2003 and 2012 the informality rate among salaried

workers fell by 10.5 percentage points and unemployment fell by 7.5 points. This decline in

labor informality is particularly puzzling because, during the same period, the minimum wage
                                                                                                      18
increased by 61 percent in real terms and changes in other labor regulations were negligible.

    Haanwinckel and Soares (2020) develop a search model of informal labor markets with

worker and rm heterogeneity, intra-rm bargaining with imperfect substitutability across

types of workers, and a comprehensive set of labor regulations.            They estimate the model

using Brazilian data and use it to understand the main determinants of the reduction in

labor informality. The model accounts for 57 percent of the decline in informality and 58

  18 See Haanwinckel and Soares (2020) . Barbosa Filho and Veloso (2016) provide a detailed discussion of
the decline in labor informality in Brazil during the 2000s.

                                                   17
percent of the reduction in unemployment observed during the 2003-2012 period.

   Subsequently, they assess the contribution of several factors in generating the observed

decline in informality. Overall, the composition of the labor force, productivity and demand

parameters are, together with the minimum wage, the main determinants of the changes

observed in the Brazilian labor market during this period. The change in the composition of

the labor force appears as the main driving force behind the reduction in informality. The

model predicts that without an increase in the share of skilled workers, informality would

have gone up instead of declining.

   Moreover, positive productivity and demand shocks contributed to reducing informality.

In their model these shocks beneted mainly the formal rms and induced their expansion.

This in turn increased the formalization of labor through the intensive margin. In particular,

the sizable improvement in terms of trade associated with the commodity boom may have

contributed to labor formalization. In Section 5 we will discuss some evidence that the credit

expansion of the 2000s may have also played an important role.

   In the model, an increase in the relative supply of skilled workers makes their labor

market less tight. Since formal rms are more intensive in skill workers, the improvement in

the skill composition of the labor force induces an expansion of formal rms. Moreover, due

to the complementarity between skilled and unskilled workers, the productivity of unskilled

workers increases. Both forces end up reducing informality through both the intensive and

extensive margins.       There is a reallocation of labor to formal rms and marginal informal

rms choose to become formal. show that these model predictions are consistent with the

Brazilian evidence.



4.2.2 Minimum wage
Between the mid-1990s and the mid-2010s, Brazil has seen a remarkable decline in income
              19
inequality.        Over the same period, Brazil's real minimum wage more than doubled. This

fact has motivated several studies about the eects of the minimum wage and other variables

on earnings inequality. However, very few papers addressed the eects of the minimum wage

increase on the eciency of labor allocation among rms.

   Engbom and Moser (2021) combine rich administrative and survey data with an equi-

librium model of the Brazilian labor market and quantify the eects of the increase in the

minimum wage in Brazil from 1996 to 2012 on inequality and employment. They show that

the rise in the minimum wage accounts for a large part of the decline in earnings inequality

in Brazil during this period.         The paper focus on the role played by heterogeneous rms

 19 Firpo   and Portella (2019) analyze the decline in wage inequality in Brazil since the mid-1990s.


                                                      18
in mediating the eects of the minimum wage policy.       In their model the increase in the

minimum wage aects mostly low-productivity rms, for which the minimum wage is more

binding. The results show that these rms cut employment creation as the increase in the

minimum wage squeezes their prot margins. The easier recruiting environment in turn in-

duces higher-productivity rms to increase hiring. As a result, in their model the minimum

wage primarily reallocates employment from lower- to higher-productivity rms rather than

to unemployment. The magnitudes of their estimated eects of the minimum wage on in-

equality are driven by how binding the minimum wage is together with the extent of rm

productivity dispersion in Brazil. The authors argue that they nd muted negative eects

of the minimum wage on employment and aggregate output due to the heterogeneous eects

of the minimum wage across the rm productivity distribution. However, Haanwinckel and

Soares (2020) obtain dierent results. Specically, they nd that the decline in unemploy-

ment between 2003 and 2012 would have been greater in the absence of a minimum wage

rise in the magnitude observed in the data. They also show that the decline in informality

in Brazil would have been considerably larger if the minimum wage had not increased. To

the extent that higher informality is associated with lower productivity, their results suggest

that the increase in the minimum wage might have been detrimental to productivity growth.

It thus remains an open issue how the rise in the minimum wage aected informality and

productivity in Brazil during the 2000s. In any case, it should be noted that in Brazil, an in-

crease in the minimum wage has signicant scal eects since many social benets, including

pension payments and unemployment insurance, are indexed to the minimum wage. Hence,

possible positive eects of an increase in the minimum wage have to be compared to the

associated scal costs and their implications for investment and innovation decisions.



4.2.3 Misallocation of labor between the public and private sectors
There is a large body of evidence showing that for many countries the structure of wages

and pensions and the labor law legislation are dierent for public and private employees.

Such dierences aect the occupational choice of agents and might generate some type of

misallocation. This is particularly relevant for Brazil, a country with a high public sector
            20
premium.

   Cavalcanti and Santos (2021) develop a life-cycle model with endogenous occupational

choice and heterogeneous agents to study the implications of an overpaid public sector in

Brazil. The paper investigates whether dierences in labor compensation and labor legislation

between private and public workers aect individuals' occupational choice, investment and

 20 See   Bank (2017c).


                                              19
aggregate productivity.

   The authors analyze the eects of both static and dynamic types of misallocation. First,

the existence of a public earnings premium inuences the occupational decision of agents.

The main idea is that such premium might attract high-productivity entrepreneurs who

would not apply for a job in the public sector in the absence of such a public wage premium.

Moreover, there may be negative long-run consequences, since investment decisions depend

on the agents´ occupational choices.           The results show that an overpaid public sector has

sizable productivity losses. They show that a decrease in the public wage premium from 19

percent to 15 percent can produce a signicant positive eect on long-run aggregate output

(7 percent increase) without any substantial decrease in public infrastructure.                      They also

show that the reallocation of factors of production accounts for about 19 percent of the

aggregate change in output. The rest is due to changes in factor accumulation. The authors

also argue that pension reforms can have similar aggregate eects, but such reforms change

more the incentive of agents to invest in nancial assets and most of the impact on output

is due to changes in factor accumulation. Changes in the legislation law which decrease job

stability in the public sector also produce important quantitative eects on aggregate output

and productivity. They nd that when the job destruction rate in the public sector increases

from 0 percent to 1 percent the number of applicants to public sector jobs decreases by nearly

15 percent and output increases by more than 3 percent.



4.3 Subsidies to national innovation and low management quality
In addition to a more ecient allocation of resources across rms, one important source of

productivity gains is innovation. In the last decades Brazil has implemented several innova-

tion policies, most of them involving scal incentives. For instance, the Lei de Informática

(Informatics Law), instituted in 1991 and renewed in 2001, 2004 and most recently in 2014,

promotes increased local content of ICT hardware and related electronics assembly plus in-

vestments in local R&D operations. Similarly, the Lei do Bem (Fiscal Incentives Law), which

was instituted in 2007 and replaced a prior 2005 law, expanded incentives for investments in

R&D, authorizing companies that invest in R&D and meet certain requirements to claim tax

incentives automatically for certain types of spending.

   Several papers show that the incentives for innovation and R&D provided by the Lei
                                                21
de Informática have not been eective.               In particular, Kannebley Júnior and Porto (2012)

show that it has been ineective in stimulating productivity-enhancing R&D. Although the

incentives have contributed to leading global ICT hardware rms to produce locally, the ben-

 21 See   Bank (2017b) for a review of the eect of innovation policies on productivity in Brazil.


                                                       20
eciaries have not been able to produce internationally competitive ICT products. Regarding

the Lei do Bem, the evidence suggests that the program had a positive but modest impact

on innovation. For instance, Devereux and Guceri (2015) nd that the average realized R&D

intensity is relatively low. They argue that this is because the legislation design favors in-

cumbent, older and larger rms and it does not reach most small or new companies. This

is due to the fact that Lei do Bem excludes rms that declare income tax returns based on
                                                                       22
their presumed prot, which includes most of the young rms.

   To the extent that Lei do Bem favors incumbent rms, it may have contributed to slow

down the reallocation process from low-growth incumbents to high-growth young rms. A

recent paper by De Souza (2021) studies a Brazilian innovation program implemented in

2001 that taxed the leasing of international technology to subsidize national innovation.

The Brazilian Technology Substitution Program (TSP) created a 10 percent marginal tax

rate on the payments of any international intellectual property. The revenue raised by the

tax was used to subsidize innovation projects by rms in targeted sectors.                 The author

uses a novel rm-level dataset with information on international technology leasing, patent

applications, and employment of Brazilian rms and calibrates a model to analyze the eects

of the innovation policy. He shows that the program led rms to replace technology licensed

from developed countries with in-house innovations, which led to a decline in employment

and an increase in the share of low-skilled workers. He nds that rms more exposed to the

TSP program had an increase of 4.47 percentage points in the probability of having at least

one patent, in comparison to the control rms, while drastically reducing the probability of

leasing technology from developed countries.

   Another important source of eciency gains at the level of the rm is through an increase
                         23
in management quality.        Bank (2017b) provides evidence on the average management quality

of rms in Brazil relative to some peers (Mexico, Chile, Turkey and Argentina) and to higher-
                              24
income OECD countries.             It shows that average management quality of Brazilian rms is

lower than most comparator countries. Moreover, management quality dispersion of rms in

Brazil is higher than in other countries, with a much larger share (18 percent) of rms being

poorly-managed. This contrasts with 2 percent of rms in the U.S., 6 percent in China, and

11 percent in Mexico.

   This evidence is consistent with the stylized facts previously discussed according to which

Brazil has a larger productivity dispersion and a higher proportion of low-productivity rms

than other emerging economies. The existence of a large segment of poorly-run rms suggests

 22 SeeKlenow and Li (2020) for evidence on the importance of young rms for innovation.
 23 SeeBloom et al. (2013).
 24 There is available data for many countries, including Brazil, in the World Management Survey.




                                                  21
                                                                             25
insucient management training and lack of market competition.



4.4 Rent-seeking
Another potentially important source of misallocation in Brazil is rent-seeking. A few recent

papers have explored rich data sets to examine the eects of patronage and corruption on

public employment selection and rm performance in Brazil. Colonnelli et al. (2020) use a

detailed matched employer-employee data on the universe of public employees in Brazil over

19972014 and a regression discontinuity design in close electoral races to examine the role of

political connections for public employment selection in Brazil. In particular, they empirically

investigate whether discretion in public employment decisions is used as a patronage tool,

and the consequences on the selection process in the context of the Brazilian public sector.

Patronage represents an obvious friction in the selection of a high-quality public workforce,

since political support can act as a substitute for individual competence in the process of

government hiring.      They establish three main ndings.          First, political connections are a

key and quantitatively large determinant of employment in public organizations in Brazil,

for both bureaucrats and frontline providers. Second, patronage is an important mechanism

behind this result.     Third, political considerations lead to the selection of less competent

individuals.   In addition to patronage, there is recent evidence on the eect of corruption

on rm performance in Brazil.         Colonnelli et al. (2020) estimate the economic eects of a

randomized anticorruption crackdown on local economic activity.                   On average, they nd a

0.9 percent increase in the number of rms and business establishments operating in treated

municipalities in the three years after the audit.          They also nd that audits increase real

economic activity, with an increase in sales by local rms by about 6 percent.                    There is

also evidence of large spillover eects, as a nearby audit has an impact on non-audited

municipalities that is similar to their main eects. They also show that their magnitudes can

be rationalized by an average corruption tax on rms in the range of 5-23 percent. Finally,

they provide evidence that is informative of the role of corruption as a source of misallocation

across rms.      Consistent with the presence of a corruption tax, they nd that incumbent

rms in government-dependent sectors grow the most after an audit.                   On the other hand,

politically connected rms shrink considerably, suggesting that a potential channel through

which corruption aects local economic activity is by allowing connected but inecient rms

to operate and stie competition. They conclude the paper by highlighting the central role

  25 Ranasinghe  (2020) nds substantial dierences in the extent of misallocation across male and female
establishments spanning over 75 low and middle-income countries. In South American and South Asian
countries female establishments face higher distortions on production, whereas in Eastern European countries
male establishments face higher distortions.


                                                    22
played by public procurement in the empirical ndings they document. They suggest that an

eective anti-corruption policy would be to improve governance and best practices in public
                 26
procurement.




5       Reforms and productivity growth
This section will discuss the literature that analyzes the eect on productivity of several

reforms implemented since the 1990s. Several of them tried to reduce distortions that were

discussed in the previous section and thus improve the eciency of factor allocation.



5.1 Trade liberalization
In the late 1980s Brazil started a process of trade liberalization that lasted until the mid-

1990s. This episode consisted of large, unilateral, import tari reductions. Average taris

fell from 31 to 13 percent, and there was ample variation in the magnitude of tari cuts

across sectors. After 1995, taris remained relatively constant. Using a panel of manufac-

turing rms, Ferreira and Rossi (2003) measure the impact of trade reform on productivity

growth. The results show that the tari reduction had a substantial positive impact on the

growth rates of TFP and output per worker.                Moreover, this eect was widespread in the

manufacturing sector, with most activities experiencing large productivity improvements.

Lisboa et al. (2010) show that one important mechanism through which the tari reduction

increased productivity was that it allowed national producers to have access to more tech-

nologically advanced equipment and components from abroad.                   Despite the evidence that

the trade liberalization reform increased productivity at the rm level, there were not siz-

able aggregate eects in the 1990s.         Menezes Filho and Muendler (2011) provide evidence

that is informative about this issue.        They nd that even though output shifted to high-

productivity export-oriented rms, there was a labor ow away from these rms because

their labor productivity increased faster than their production.              The labor that was shed

from these productive rms moved to low-productivity service activities, unemployment and

out of the labor force.     As a result, despite sizable gains at the rm level, the labor real-

location induced by trade liberalization seems to have reduced its aggregate eect.                   Using

    26 Zaourak
             (2019) argues that rent-seeking activities, in the form of bribes, have aggregate eects through
two channels. First, they generate misallocation of resources across rms, since they prevent resources from
owing to the most productive rms. Second, rent-seeking activities aect the allocation of resources within
rms, since they drive resources away from innovation activities. He then quantitatively shows that these
two channels can help explain why Brazil has both more misallocation across rms and less investment in
research and development than developed economies.


                                                     23
data from Brazil, Dix-Carneiro (2014) shows that the slow reallocation of workers and capital

toward export-oriented industries leads to substantially lower gains from trade compared to

traditional frictionless models. He estimates 11 to 26 percent lower gains from trade com-

pared to a situation where reallocation occurs immediately. In addition, he estimates that

adjustment costs are heterogeneous across the population.      In particular, older, less edu-

cated, and female workers face substantially higher barriers to mobility across sectors. These

workers in import-competing sectors experience substantial losses following liberalization, so

governments willing to compensate the losers from trade should pay particular attention to

workers with these characteristics. Dix-Carneiro and Kovak (2017) investigate how Brazilian

region-specic labor market outcomes have responded to local shocks induced by the tari

cuts, tracking the evolution of these eects over time. They show that regions facing larger

negative shocks induced by liberalization experienced declines in formal sector employment

relative to regions facing smaller shocks. These eects gradually increased following the be-

ginning of liberalization and only stabilized fteen years afterward. These results indicate

that formal sector employment adjustment in response to trade liberalization was large but

slow. They also document that wages in harder-hit locations steadily declined for years and

never recovered. The slow reallocation of capital led to a steady amplication of the initial

local labor demand shock, making workers in harder-hit regions even less productive over

time compared to those in more favorably aected regions. Agglomeration economies also

amplify the labor market eects of trade liberalization. As rms in harder-hit regions leave

the market, the productivity of remaining local rms gradually declines, further reducing

local wage and employment growth. This evidence focused on how aggregate formal sector

and regional-level labor market outcomes responded to liberalization. In another study, Dix-

Carneiro and Kovak (2019) analyzed how individual labor market trajectories responded to

the trade-induced labor demand shocks.     Following individual workers over time using the

Relação Anual de Informações Sociais (RAIS) data, they show that tradable sector workers

initially employed in harder-hit locations were more likely to switch to nontradable sectors

in response to liberalization. However, this response did not oset the large losses in employ-

ment in tradable sectors. Finally, they document that the formal employment trajectories of

workers initially employed in nontradable sectors were aected almost as much as those of

tradable sector workers, indicating the presence of important spillovers from tradable to non-

tradable sectors locally. Dix-Carneiro and Kovak (2019) also investigate how the structure

of local labor markets responded to the trade-induced local shocks over longer time periods.

In the medium run (1991 to 2000), nonemployment and informal employment increased in

harder-hit locations relative to the national average.   Yet in the long run (1991 to 2010),

nonemployment did not respond to local trade shocks, but informal employment strongly



                                              24
increased in harder-hit regions. This indicates that trade-displaced workers spend some time

unemployed or out of the labor force, but eventually nd reemployment in the informal sec-

tor. The evidence thus suggests that informal employment constitute an important margin

of labor market adjustment in response to trade shocks. In order to shed light on this issue,

Dix-Carneiro et al. (2021) build an equilibrium model of a small open economy with labor

market frictions and imperfectly enforced regulations. They estimate the model using data

from Brazil and use counterfactual simulations to understand how trade aects economic out-

comes in the presence of informality. They nd that trade openness leads to large declines

in informality in the tradable sector. On the other hand, the eects of trade openness on

informality in the non-tradable sector are more context-dependent. As a result, the overall

eect of trade on informality is ambiguous, and generally small. They also nd that trade

openness is associated with substantive increases in productivity and that the productivity

gains from trade are understated in studies focusing exclusively on the formal sector of the

economy.



5.2 Credit reforms
5.2.1 Repossession of collateral
There is a large literature documenting the relationship between collateral and better credit
                                                                          27
conditions, in particular larger loan volumes and lower interest rates.        To remove inecien-

cies in the auto loan and other credit markets, the Brazilian government enacted in 2004 the

Law 10.931, the Lei de Alienação Fiduciária (Fiduciary law), which aected the auto loan,

mortgage, and capital markets. Among other changes, the credit reform simplied the sale of

repossessed cars used as collateral for auto loans. Before the reform, even though banks were

allowed to repossess the autos of borrowers who failed to repay their loans, they could not

resell these repossessed cars without court approval. As a result, the time from repossession

of a car to its resale by the bank averaged more than two years. After the reform the process

of reselling a repossessed car became simpler and faster. Borrowers and creditors engaged

in direct relationships and courts played a signicant role only when borrowers explicitly

requested it. The law thus avoided unnecessary trials, reduced the reliance on courts and

increased the enforceability of auto loan contracts.

   Assunçao et al. (2014) use micro-level data from one of the largest banks in Brazil to

provide direct evidence on the consequences of the reform.       They show that the reform

resulted in larger loans with lower spreads and longer maturities.    It also expanded credit

 27 See   Djankov et al. (2007).


                                              25
to riskier, low-income borrowers for newer, more expensive cars. They thus concluded that

collateral and repossession play a crucial role in the liberalization and democratization of

credit in Brazil.



5.2.2 Payroll lending reform
In December 2003, the Brazilian Congress passed a law (Law 10.820) that allowed banks to

oer loans with repayment through automatic payroll deduction (Lei do Crédito Consignado),

which turned future income into collateral. The law regulated the procedures through which

commercial banks underwrite payroll loans to private sector employees and to those receiving

social security benets from the Instituto Nacional do Seguro Social (INSS), the federal pay-

as-go pension system. Payroll lending has existed in Brazil since the establishment of Law

8.112, which was enacted in December 1990 to regulate the provision of such loans to public

sector retirees and public servants.    However, private sector retirees and employees were

not included in the scope of the law.      The new legislation introduced in 2003 extended

the mechanism of payroll lending to private sector workers and social security beneciaries.

Coelho et al. (2012) estimate the impact of the new law. They nd that payroll lending caused

a reduction in interest rates and a large surge in personal loans in Brazil. Their results suggest

that policies that strengthen collateral have a major impact on lenders' ability to underwrite,

and they thus improve borrowers' access to nance.



5.2.3 Bankruptcy reform
In early 2005, the Brazilian Congress approved a new bankruptcy law (Law 11.101, Lei

de Falências). One of the most signicant changes was to enhance the protection given to

creditors, which was implemented through two channels. First, the new law increased their

priority order to receive proceeds.    Second, it allowed them to actively participate in the

reorganization procedure. The law also aimed at increasing the eciency of the bankruptcy

system by reducing the cost and time of both reorganization and liquidation, leading to an

increase in the value of distressed rms. It improved on existing rules by providing in- and

out-of-court options to reorganize, and by allowing the conversion of reorganization proceed-

ings into liquidation. Using data from Brazilian and non-Brazilian rms, Araujo et al. (2012)

estimated the eect of the bankruptcy reform on debt variables. In particular, they compared

Brazilian rms to non-Brazilian rms from Argentina, Chile and Mexico with respect to the

behavior of debt related variables.    They found a reduction of approximately 8 percent in

the cost of debt and increases of 10 percent and 23 percent in the amount of both total debt

and long-term debt in Brazil, respectively. Since secured creditors have beneted more from



                                               26
the new law than unsecured ones, the eect is more pronounced on long-term debt, which

is known to be more correlated with secured debt.        Brazilian judicial districts are highly

heterogeneous in terms of eciency. In some districts, cases are closed within time frames

comparable to those observed in the United States. In others, the functioning of courts is

undermined by the large number of pending cases. Therefore, when the new bankruptcy law

entered into force, the eciency of local courts became a key determinant of the ability of

both creditors and rms to reap the benets of the reform. Ponticelli and Alencar (2016) ex-

ploit variation in the congestion of civil courts across Brazilian municipalities to estimate the

eect of enforcement of the new bankruptcy law on rm access to nance, investment, and

size. The paper assessed the impact of a potential misallocation frictionthe ineciency of

local judicial institutions -on credit markets as well as rm investment and growth. They nd

that rms operating in municipalities with less congested courts experienced a larger increase

in the use of secured loans, as well as a larger increase in investment and value of output

in the years after the reform.    To establish the direction of causality, they use an instru-

mental variable strategy that exploits dierences in court congestion across otherwise similar

neighboring municipalities located across judicial district borders within the same state. The

evidence shows that dierences in court enforcement aect the impact of bankruptcy reform

on rm access to nance, investment, and size. The authors argue that their results make

the case that an ecient judiciary is a necessary precondition for rms to benet from these

reforms. In this sense, to increase the impact of reforms on productivity it is important to

nd the right balance between promoting necessary changes in legal rules and investing to

increase the eciency of the judicial institutions in charge of enforcing. Judicial decisions in

bankruptcy are often inuenced by the goal to preserve employment in nancially distressed

rms. Araujo et al. (2021) construct a new court-level measure of pro-labor bias based on

the text of judicial decisions and exploit the random assignment of cases to courts within

judicial districts in the state of São Paulo in Brazil to study the eect of pro-labor bias on

labor market outcomes. They motivate their focus on Brazil for a number of reasons. First,

Brazil has relatively strict labor-protection laws and a judicial system often described by local

observers as biased in favor of debtors and workers. They cite survey evidence that shows

that a large fraction of the Brazilian Judiciary perceives itself as having a redistributive role.

Second, according to the data collected for the paper, there is a large degree of variation

in the degree of pro-labor bias across Brazilian courts dealing with bankruptcy cases. They

nd that workers of rms assigned to high-pro-labor courts experience 4.4 percent lower

post-bankruptcy earnings.     They also try to identify what explains the negative eect of

pro-labor bias on workers' earnings. They show that courts with higher pro-labor bias tend

to facilitate the continuation of insolvent rms, by either rejecting liquidation requests at a



                                               27
higher rate, or converting reorganization cases into liquidations at a lower rate. Then they

show that employees of insolvent rms whose cases were assigned to a high-pro-labor court

are more likely to stay with the same employer in the post-bankruptcy period. Finally, the

authors try to understand why do employees remain with the same employer when they

could potentially earn more if searching for a new job. They nd that the relative decline

in earnings is stronger in areas with limited internet coverage. Their interpretation of this

result is that when information costs are high, individuals limit their search and are more

likely to stay with their current employer for longer, even if this implies a wage reduction.

Overall, their empirical ndings suggest that judicial bias matters for employees' earnings

and employment trajectories in Brazil and have important policy implication for the role of

bankruptcy institutions. This is consistent with the results of Ponticelli and Alencar (2016)

and their implication that an ecient judiciary is an important precondition for rms to

benet from these reforms.



5.2.4 Financial deepening
Catão et al. (2009) examine empirically a link between credit markets and formalization.

Since access to bank credit typically requires compliance with tax and employment legisla-

tion, rms are more likely to incur such formalization costs once bank credit is more widely

available at lower cost.   Since the cost of remaining an unregistered rm is that of hav-

ing either limited or no access to formal credit markets, nancial deepening would tend

to shrink the relative size of the informal sector.   They exploit the fact that in the 2000s

Brazil experienced both a sizable credit expansion and a large decline in labor informality.

The share of workers with a formal labor contract (carteira assinada) rose from 38 percent

in 2003 to nearly 45 percent of the urban labor force by early 2008.     They note that this

upward trend in formality took place with a simultaneous credit expansion.       The ratio of

credit to rms rose from 15 percent of GDP in 2003 to 22 percent in 2008. They investi-

gate this credit channel for Brazil using the Rajan-Zingales measure of nancial dependence

and a dierence-in-dierences approach applied to household survey data.      They nd that

formalization rates increased with nancial deepening, especially in sectors where rms are

typically more dependent on external nance.      They also highlight that the main channel

through which formalization took place was through shifts in the formality rate within each

rm size category. To a lesser extent, they also nd some evidence that nancial deepening

shifted the composition away from self-employment and towards larger rms.         Regarding

the interpretation of their results, they note that one important implication is that nancial

deepening may lead to better resource allocation and hence to higher productivity through




                                             28
                               28
a reduction of informality.         Another implication is that public programs of credit subsidies

to small enterprises should be designed with the requirement that such rms comply with a

host of regulations and nancial obligations associated with formal employment.



5.3 Formalization policies
5.3.1 Special tax regimes for micro and small rms
Most of the formalization policies analyzed in the literature have been in the direction of

reducing the costs of formality. In particular, great emphasis has been placed on reducing

registration costs, which are often regarded as a major constraint to rm creation and for-
               29
malization.         However, the available evidence suggests that reducing the costs of entering
                                                                        30
the formal sector has a very limited eect on formalization.                 This suggests that the per-

ceived benets of formalization are very low for most small-scale entrepreneurs. In the last

decades several formalization policies were implemented in Brazil. In general, they reduced

the costs of entering and operating in the formal sector. Most prominent among them were

special tax regimes for micro and small rms.             I will discuss next some of these programs

and their evaluations.       In 1996 a federal reform was enacted to reduce red tape costs and

simplify the tax system for micro and small enterprises (MSEs).                 The SIMPLES program

combined six dierent federal taxes and social contributions into one single monthly-based

rate.   The reform reduced the tax burden considerably and red tape costs as well.                    In the
                                                                        31
beginning, only a few sectors were covered by the new law.                   With the Complement Law

No.123 of December 14, 2006, several additional activities were included. Moreover, under

the new system (SIMPLES Nacional), state and municipal taxes were combined with the

federal taxes and contributions into one single monthly-based rate. Fajnzylber et al. (2011)

used the change in legislation at the end of 1996 as a natural experiment to estimate the

causal eects of the program on formalization rates and performance outcomes. They found

that the program led to an increase in the formalization rates of about 11 percentage points

(or 50 percent). They also found a positive impact on revenues, prots and employment. A

study published a year later obtained dierent results. Monteiro and Assunção (2012) found

positive eects of SIMPLES on formalization rates only for rms in the retail sector. For the

whole sample of rms, their estimates suggest that SIMPLES did not aect formalization

  28 Barbosa  Filho and Veloso (2016) provide some evidence that the reallocation of labor from the informal
to the formal sector contributed to increase aggregate productivity in the 2000s.
  29 Djankov et al. (2002).
  30 Andrade et al. (2013).
  31 Retail, manufacturing, transportation, construction, and other services that do not require professionals
with regulated occupations, such as auditors and engineers, among others.


                                                     29
         32
rates.

    Piza (2018) evaluates the SIMPLES program with the same data sets used in both studies

and shows that the large eects of the program on formalization rates found previously were

likely confounded by measurement error in the assignment variable and seasonal shocks that

aected more intensely the sectors that the reform initially targeted. He then suggests two

alternative empirical strategies to deal with the identication problems present in the two

studies. The new estimates suggest that SIMPLES did not increase formalization rates of

micro- and small rms.         The author then argues that given the expansion of the program

over the years, it would be important to evaluate SIMPLES in its current format so that

the potential eects of the program on tax collection through higher formalization rates of
                                                                                           33
micro- and small businesses could be confronted with its large scal costs.

    Rocha et al. (2018) estimate the impact of another large-scale formalization program

implemented in Brazil, the Individual Micro-Entrepreneur Program (Programa do Microem-

preendedor Individual, MEI). Introduced in 2009, the program targeted entrepreneurs with

at most one employee, and was designed to reduce both registration and ongoing costs of

remaining formal, by reducing monthly taxes and red tape. In its rst phase, the program

eliminated entry costs for eligible entrepreneurs, and in the second phase it substantially

reduced the tax burden.         The authors exploit this staggered implementation to assess the

eects of reducing the tax burden faced by small formal rms. They nd that the rst phase

had no eect on formalization, while the second led to an increase of around 11 percent. This

result is entirely driven by the formalization of existing informal rms and not by the creation

of new formal businesses, nor by greater survival among formal rms. However, the authors

argue that these interventions were not cost-eective, since the program led to net losses

in tax revenues. Taken together, these results indicate that reducing the costs of entering

the formal sector had very limited or no formalization eects in Brazil, while reducing the

ongoing costs of formality was more eective, but the eects were not large and the policies
                                         34
do not seem to be cost eective.              This raises the question of why reducing the costs of

formality have such limited eects.

    As discussed previously, Ulyssea (2018) nds that only 9.3 percent of informal rms in

Brazil are potentially productive rms hindered by high regulatory costs.                        The informal

  32 Monteiro  (2016) shows that SIMPLES is not cost-eective. Even in the retail sector the increase in
revenues that results from formalization does not compensate for the revenue loss of already formal micro
and small rms that have a reduction in their tax burden.
  33 Corseuil and Moura (2016) evaluated the eect of SIMPLES on several rm performance indicators in
the manufacturing sector for several years, including the SIMPLES Nacional in 2007. They found no eect
of SIMPLES on rm performance.
  34 See Ulyssea (2020b) for a discussion of the evidence on formalization policies in Brazil and other countries.




                                                       30
rms that could formalize once formal sector's entry costs are removed, but choose to remain

informal to enjoy the cost advantages of informality, correspond to 41.9 percent of all informal

rms. Finally, the remaining 48.8 percent correspond to rms that are too unproductive to

ever become formal.    Hence a very small fraction of informal rms in Brazil seems to be

constrained by high formalization costs, and therefore reducing them will have a limited

eect on formalization. Since these policies involve scal costs, they are in general not cost-

eective even when they have positive formalization eects.



5.3.2 Other formalization policies (reduction of entry costs, increase in enforce-
      ment)
Even though reforms targeted at the reduction of formal sector's entry costs have not been

eective in inducing rms to formalize, they might still produce important aggregate eects.

Indeed, some papers show that reducing entry costs into the formal sector can produce posi-

tive and sizable aggregate eects in general and specically in Brazil. Ulyssea (2010) develops

a two-sector matching model that incorporates the main features of Latin American labor

markets.   The model is numerically solved using Brazilian data and several policy simula-

tions are performed. The simulation results show that lowering the costs of entry into the

formal sector substantially improves employment composition. Making formal sector's entry

costs similar to informal sector's would increase formal employment by nearly 31 percent

and reduce unemployment by 36 percent. As a consequence of the changes on employment

composition and the reduction of unemployment rate, average productivity and specially

welfare show a substantial increase. Hence, high entry costs not only play an important role

in determining employment composition, but also generate a considerable deadweight loss.

This latter eect comes from the fact that high creation costs reduce the number of formal

vacancies created and, as the informal sector is not able to completely absorb all workers,

unemployment increases. In a recent article, Ulyssea (2018) shows that reducing entry costs

into the formal sector eliminates deadweight losses from wasteful barriers to entry, which

leads to more competition and an increase in aggregate production in the formal sector.

It also increases high-skill wages, since the formal sector is more intensive in high-skill la-

bor. Since the formalization process is concentrated among low-productivity rms, there is

a negative composition eect that leads to a decrease in aggregate TFP. Nevertheless, total

output increases because there is a substantial increase in the mass of active rms in the

economy. In contrast, increasing enforcement can have a substantial impact on informality

and productivity, since a large fraction of rms could be formal but optimally choose to re-

main informal. However, this policy can have potentially large negative eects, since nearly




                                              31
half of all informal rms in Brazil are not productive enough to ever survive in the formal

sector. De Andrade et al. (2016) randomly assign municipal inspectors to rms in order to

assess whether higher enforcement can induce rms to formalize in the state of Minas Gerais,

Brazil. Their results show that being assigned to the enforcement treatment increases regis-

tration rates with the municipality by 24 percentage points. The authors nd no evidence

of spillovers on neighboring rms, which they attribute to the relatively low increase in in-

spections and to the fact that many rms indicate that they do not communicate with their

neighbors.   Meghir et al. (2015a) nd that reducing informality by increasing enforcement

does not increase unemployment and increases welfare by enabling the reallocation of work-

ers to higher productivity jobs.   As a consequence, overall wages go up and inequality is

decreased. These results keep labor market regulation xed and greater welfare gains may be

obtainable from deregulation. In any case, they argue that an intermediate world where the

informal sector is tolerated at the current levels of enforcement is not a welfare-enhancing

policy. Ulyssea (2018) also analyzes the eects of enforcement on informality and produc-

tivity. He uses the estimated model to conduct counterfactual analyses of two enforcement

policies. One is to increase the cost of the extensive margin of informality through greater

enforcement on informal rms.      The other is to increase the costs of the intensive margin

through tighter enforcement on formal rms that hire informal workers. The results show

that increasing enforcement on the extensive margin nearly eradicates informal rms, which

generates a large positive eect on aggregate TFP due to composition eects, as this policy

eliminates many small and unproductive informal rms.       This positive eect on aggregate

TFP more than compensates the reduction in the mass of active rms, and total output

increases by more than 3 percent.     On the other hand, an increase in enforcement on the

intensive margin may induce small formal rms to move to the informal sector. As a result,

aggregate informality may increase when there is more enforcement on the intensive margin.

Despite reducing informality and increasing productivity, a policy of increased enforcement

may generate an overall welfare loss in the economy due to the elimination of informal rms

and possible unemployment consequences. If the adjustment of labor from the informal to

the formal sector takes a long time and diers across skill levels, the overall welfare cost

might be substantially higher. In addition to adjustment costs, an increase in enforcement

probably involves signicant implementation costs, given that informal rms are numerous,

small and often geographically spread.




                                              32
5.3.3 Payroll tax reform
The evidence of reducing payroll taxes on labor market outcomes is quite mixed.               A few

studies nd that the incidence of such costs is fully on wages with no signicant eects on

employment for the U.S. and Latin American countries. Other papers show that only part

of the increase in the payroll tax burden was passed onto wages. More recent work nds a
                                                                      35
sizable impact on employment in Sweden and Colombia.

   A few studies reviewed in Bank (2017b) nd that a large reduction in payroll taxes for

selected sectors in Brazil had very limited eects on formal employment creation. However,

a recent paper found dierent results. Baumgartner et al. (2021) study the eects of Plano

Brasil Maior (PBM). Such tax reform altered the tax base upon which payroll taxes are

calculated for selected sectors in Brazil.          Instead of paying a payroll tax of 20 percent on

wages, rms were required to contribute to social security with tax rates between 1 percent

and 2 percent on gross revenue.            This amounted to a de facto average reduction in the

total tax burden associated with payroll of more than 50 percent.               The authors exploit

the gradual implementation of the PBM payroll tax reform across 65 sectors from 2011 to

2014.     Their empirical ndings show a sizable positive eect on total employment due to

both rm entry and rm growth in treated sectors, but no impact on wages.                 Moreover,

the eects are driven by labor markets with relatively low levels of concentration, which is

consistent with predictions of an oligopsony model. They also perform a cost-benet analysis

of the PBM payroll tax reform. They nd that each created job was around 2.1 times more

costly than estimates from a major federal program that transfers funds to municipalities to

nance local public spending in Brazil. Hence, despite having achieved a signicant impact on

employment in treated sectors, the costs of the program in terms of foregone tax income were

substantial. The authors also argue that this cost-benet analysis does not take into account

other consequences of this kind of reform, such as the likely welfare consequences related to

lower social security tax revenues and the deterioration of government´s scal position. They

conclude that a more complete and quantitative general equilibrium would be needed to assess

such tradeos. Other researchers have used structural models in order to assess the general

equilibrium impact of payroll tax reductions. In a recent review of the literature, Ulyssea

(2020b)argues that the counterfactual results from both macro and structural models indicate

that reductions in payroll tax seem to generate some formalization but with a low elasticity. If

one dierentiates the eects on the intensive and extensive margins of informality, the eects

are stronger on labor informality (intensive margin) and weaker on rm informality (extensive

margin). Consistent with the limited eects on informality, the results from dierent papers

 35 See   Baumgartner et al. (2021) for a review of the literature.


                                                       33
indicate that reducing payroll taxes has positive by small eects on productivity. Haanwinckel

and Soares (2020) estimate a model for Brazil to analyze a reduction in payroll taxes for low-

wage workers while also keeping track of the scal burden imposed on the government. Lower

payroll tax rates can lead to a decline in informality, but they can also substantially reduce

government revenues. However, only a fraction of government revenues come from payroll

taxes on unskilled workers, since their wages are lower and they account for a smaller fraction

of formal employment.    Thus, a policy in which governments subsidize the employment of

low-wage formal workers through a progressive payroll tax may have a larger impact on

informality while simultaneously minimizing the implications for government revenue. They

compare the introduction of a progressive payroll tax with a reduction in the overall payroll

tax. Their results show that progressive payroll taxes achieve unequivocally better results for

unemployment and formalization while minimizing impacts on government revenue. Lower

taxes among unskilled workers induce marginal rms to comply, enlarging the tax base.

Taxes raised from rms that formalize help oset part of the foregone revenue. Moreover, the

policy is clearly benecial for unskilled workers, lowering their unemployment and increasing

average wages. Thus, they conclude that progressive payroll taxes can help alleviate poverty

and income inequality without placing an excessive burden on the government scal position.




6    Potential reform agenda
Demographic dividends are nearly exhausted, with Brazil being one of the fastest aging

societies in the world. This implies that in the next decades improvements in living standards

and reductions in poverty and inequality will crucially depend on productivity growth. This

paper discussed the literature on reforms and growth in Brazil in order to understand why

productivity growth has been so low despite several reforms in the last decades. The evidence

indicates that reforms such as trade liberalization and improvements in credit collateral

had a positive impact on productivity.    However, these eects were not strong enough to

place Brazil in a sustained growth trajectory.     Alongside reforms that aimed to increase

competition, several policy-induced distortions were introduced in the last decades, such

as credit subsidies, tax exemptions and local content requirements. The evidence reviewed

in this paper suggests that these distortions may have contributed to oset the positive

eect of reforms on productivity. Moreover, the analysis showed that the eects of reforms

may have been dampened by shortcomings related to other determinants of productivity,

such as the low average schooling of the labor force and poor management quality. At the

heart of Brazil's low and stagnant productivity is a business environment that discourages

competition and induces misallocation of resources. The lack of competition in both domestic


                                              34
and external markets in turn reduces eciency and undermines productivity growth. High-

productivity rms face several barriers to grow and low-productivity rms stay in the market.

To a large extent, the development model of state intervention and import-substitution that

prevailed from the 1930s until the 1980s is still present.     In order for Brazil to increase

productivity growth in the next decades it will be necessary to a competitive market-based

economy. It is thus necessary to remove distortions that protect incumbents, that is reduce

entry barriers, and prevent the reallocation of factors of production from low-productivity

to high-productivity rms. In order to facilitate factor reallocation, it will be important to

complement market reforms with policies that enable rms and workers to take advantage

of the new opportunities that will be created, especially by enhancing the human capital of

workers and entrepreneurs. Between 2003 and 2005, several credit reforms aimed at improving

collateral were implemented. These included the simplication of selling repossessed cars used

as collateral for auto loans, allowing banks to oer loans with automatic payroll deduction,

and bankruptcy reform.    These reforms increased loan volumes, extended maturities, and

lowered interest rates. They also expanded credit to riskier, low-income borrowers, promoting

nancial inclusion. Additionally, the credit expansion in the 2000s contributed to the decline

in informality, potentially boosting aggregate productivity.

   However, after the 2008 nancial crisis, there was a signicant expansion of credit by

public banks, especially BNDES. The evidence suggests that this had a small impact on pro-

ductivity, possibly increasing credit misallocation in the manufacturing sector. These policies

were likely not cost-eective due to heavy subsidies. Since 2015, government intervention in

the credit market has been gradually reversed. The creation of TLP in 2017 reduced subsidies

in credit provided by BNDES, but housing and rural credit still receive substantial subsidies.

Extending the TLP to these sectors could increase competition.

   Another issue is the high level and dispersion of private credit spreads. Despite reforms,

there is still room for improvement. Reforms are underway to increase the dissemination of

debtor information, enhance competition, and improve collateral through electronic guaran-

tees. Regulatory changes are facilitating the entry of new nancial institutions, especially

ntechs. These reforms have the potential to increase competition, expand credit, lower in-

terest rates, and reduce misallocation by providing credit to young innovative rms lacking

collateral.In addition to removing domestic barriers, it is important to increase the exposure

of Brazilian rms to international competition. There is evidence that the trade liberaliza-

tion reform of the 1990s improved productivity through several channels, such as competitive

selection and imports of better quality capital and intermediate goods. However, since the

mid-1990s there has been little progress. In fact, several distortions were created in the last

decade, especially an increase in non-tari barriers such as local content requirements. There



                                              35
are some promising liberalization possibilities, in particular the Mercosur-European Union

trade agreement. Another important initiative is the process that has already been initiated

of application to OECD membership. This could trigger a process of regulatory convergence

to developed countries best regulatory practices, which in turn would contribute to a more

competitive domestic business environment.           Finally, there is much room for tari reduc-

tion of capital and intermediate goods. That could be an important channel of technology
              36
adoption.

   The evidence reviewed in this paper shows the crucial role of the informal sector in the

productivity agenda in Brazil, both in its extensive (rm registration) and intensive margins

(labor registration). Since formal rms are much more productive than informal enterprises,

a reallocation of labor from the informal to the formal sector has the potential to increase

productivity signicantly. However, formalization policies in Brazil had a very limited eect.

Special tax regimes for micro and small rms, such as SIMPLES and MEI, had little impact

on formalization rates and rm performance variables. In addition, they are not cost-eective.

One additional negative eect of size-dependent tax policies such as SIMPLES and MEI is the

potential increase in misallocation that they entail by favouring small rms. One implication

of this analysis is that one important reform would be to simplify the tax system for all rms

and reduce the scope of special tax regimes for micro and small rms. This would increase

competition and reduce the large scal costs associated with ineective programs. There are
                                                                                        37
already some tax reforms in Congress that make proposals along these lines.

   The small impact of formalization policies in Brazil is consistent with a study reviewed in

this paper which shows that almost half of informal rms in Brazil are too unproductive to
                          38
ever become formal.            Moreover, there is some evidence that average management quality is

low and that the proportion of rms that are poorly managed in Brazil is higher than in other
              39
countries.         This suggests that policies targeted at micro and small business should focus on

human capital and management improvements instead of tax reduction. The credit reforms

that are underway will also help providing more credit at lower rates for talented but credit-

restricted entrepreneurs. This points to an important theme that is present in the analysis

of several studies reviewed in this paper. Since labor reallocation to high-productivity rms

is a crucial mechanism for productivity growth in Brazil, the smooth functioning of the labor

market is crucial to maximize the eects of reforms and reduce possible negative welfare

consequences associated with labor transitions.

  36 Lisboaet al. (2010).
  37 PEC 45/2019 and PEC 110/2019 combine federal, state and municipal taxes into one single VAT tax.
Another legislative project (PL 3.887/2020) proposes the unication of two federal taxes into one VAT tax.
 38 Ulyssea (2018).
 39 Bank (2017b).



                                                    36
   In 2017, a labor market reform was enacted with the goal of increasing exibility in hours

worked and by allowing for new types of contracts (Law 13.467). It also increased the scope of

labor benets that may be freely negotiated between employers and employees. It is dicult

to evaluate the eects of this reform since in the recent period the labor market in Brazil has

been much aected by the great recession of 2014-2016 and the pandemic in 2020. It will

be important to assess whether the labor reform will generate more exibility in the coming

years and if new reforms will be necessary.

   Another crucial area for productivity gains is education. As discussed in the text, there

is evidence that the improvement in skill composition was the most important determinant
                                             40
of the informality decline in the 2000s.          Better schooling is also necessary to facilitate labor
                                                                          41
market transitions associated with trade liberalization reforms.               Moreover, there is some

evidence that the human capital of the entrepreneur is one the main determinants of the
                                                                                 42
decision to formalize a rm and for the upgrading of rm capabilities.                In addition to bet-

ter schooling, an improved infrastructure may facilitate the process of factor reallocation to
                                        43
more productive sectors and rms.             A recent paper nds that an increase in electricity

access contributed to the process of structural transformation in Brazil, beneting mostly
                             44
the manufacturing sector.         One study suggests that diculties in setting up national dis-

tribution systems because of the low quantity and quality of rail and road networks might
                                                                45
have prevented productivity gains in the retail sector.              There is also some evidence that

better internet connections may improve access to information about job opportunities and
                                                                     46
thus facilitate labor reallocation to more productive rms.

   Finally, the studies reviewed in this paper point to the crucial role of a public sector reform

in order to reduce rent-seeking and increase the eectiveness of business support policies. In

particular, there is evidence that patronage and corruption have signicant negative impacts

on productivity in Brazil, and that there is misallocation on the allocation of labor between

the public and the private sector. In addition, there is evidence that business support policies
                                                                          47
in Brazil are in general not eective and have high scal costs.               One important reason is

that they are very poorly coordinated and there is almost no evaluation of their eects. The

  40 Haanwinckel   and Soares (2020).
  41 Dix-Carneiro  (2014).
  42 See La Porta and Shleifer (2014) for evidence on the importance of entrepreneur´s schooling for for-
malization. Cusolito and Maloney (2018) highlight the importance of entrepreneur´s human capital for the
identication of business opportunities and the upgrading of rm capabilities.
  43 See Frischtak (2013)for a review of the literature about infrastructure and development in Brazil.
  44 Perez Sebastian et al. (2020).
  45 De Vries (2008).
  46 Araujo et al. (2021).
  47 Bank (2017b).




                                                      37
agenda of public sector reform should involve several components, such as more transparency

and accountability, as well as a reduction of the public wage premium. Moreover, there should

be a thorough revision of existent business support policies and the establishment of regular

evaluation procedures. It is necessary to establish clear goals, monitor policy implementation

and conduct ex-ante and ex-post evaluation. It would also be important to introduce sunset

clauses so that policies that do not show positive eects could be discontinued.                  In order

to increase policy coordination, one interesting possibility would be to create a Productivity

Commission, following the model rst introduced in Australia and later reproduced in several
                                                           48
countries, such as New Zeland, Mexico and Chile.

   This a challenging but politically feasible productivity agenda that could potentially re-

duce Brazil´s income gap with developed countries in the next decades. The recent pandemic

caused by Covid-19, as well as global megatrends, including demographic ageing, globaliza-

tion, technological and climate change, represent an additional challenge, but they could also

serve to disrupt the existing equilibrium and be harnessed for change.

 48 See   Azevedo (2017) for a discussion of Productivity Commissions and their possible adoption in Brazil.




                                                     38
References
Akcigit, U., Alp, H., and Peters, M. (2021). Lack of selection and limits to delegation: Firm

  dynamics in developing countries. American Economic Review, 111(1):231275.


Andrade, G. H., Bruhn, M., and McKenzie, D. (2013). A helping hand or the long arm of

  the law? experimental evidence on what governments can do to formalize rms. World

  Bank Policy Research Working Paper, (6435).


Araujo, A., Ferreira, R., and Funchal, B. (2012). The brazilian bankruptcy law experience.

  Journal of Corporate Finance, 18:9941004.


Araujo, A., Ferreira, R., Lagaras, S., Moraes, F., Ponticelli, J., and Tsoutsoura, M. (2021).

  The labor eects of Pro-Labor bias in bankruptcy. NBER Working Paper, (28640).


Assunçao, J. J., Benmelech, E., and Silva, F. S. (2014). Repossession and the democratization

  of credit. The Review of Financial Studies, 27(9):26612689.


Azevedo, L. (2017).   Agências de promoção da produtividade: Experiência internacional e

  lições para o brasil. In Anatomia da Produtividade no Brasil, pages 429446. Elsevier.


Banerjee, A. V. and Duo, E. (2005).      Growth Theory through the Lens of Development

  Economics, volume 1 of Handbook of Economic Growth, chapter 7. Elsevier.


Bank, I.-A. D. (2017a). Assessing Firm-Support Programs in Brazil.


Bank, W. (2017b). Business support policies: Large spending, little impact. Technical report,

  World Bank.


Bank, W. (2017c). Um ajuste justo: Análise da eciência e qualidade do gasto público no

  brasil. Technical report, World Bank.


Barbosa Filho, F. and Correa, P. (2017). Distribuição de produtividade do trabalho entre as

  empresas e produtividade do trabalho agregada no brasil. In Anatomia da Produtividade

  no Brasil, pages 109142. Elsevier.


Barbosa Filho, F. and Veloso, F. (2016). A contribuição da formalização para a elevação da

  produtividade do trabalho no brasil nos anos 2000: Uma análise exploratória. In Causas

  e Consequências da Informalidade no Brasil, pages 303325. Editora Elsevier.


Baumgartner, E., Corbi, R., and Narita, R. (2021).      Payroll tax, employment and labor

  market concentration. Technical report, Universidade de São Paulo.


                                             39
Bloom, N., Eifert, B., Mahajan, A., McKenzie, D., and Roberts, J. (2013). Does management

  matter? evidence from india. The Quarterly journal of economics, 128(1):151.


Bonelli, R., Veloso, F., and Pinheiro, A. (2017).     Anatomia da Produtividade no Brasil.

  Elsevier, Rio de Janeiro.


Bonomo, M. A. C., Brito, R. D., and Martins, B. (2015).        The after crisis Government-

  Driven credit expansion in brazil: a rm level analysis. Journal of International Money

  and Finance, 55:111134.


Bustos, P., Garber, G., and Ponticelli, J. (2020). Capital accumulation and structural trans-

  formation. Quarterly Journal of Economics, 135(2):10371094.


Bustos, P., Vincenzi, J., Monras, J., and Ponticelli, J. (2019).   Structural transformation,

  industrial specialization, and endogenous growth. NBER Working Paper No. 25871.


Calice, P., Ribeiro, E., and Byskov, S. (2018). Ecient nancial allocation and productivity

  growth in brazil. Technical report, World Bank.


Catão, L., Pagés, C., and Rosales, M. (2009).     Financial dependence, formal credit and

  informal jobs: New evidence from brazilian household data. IDB Working Paper.


Cavalcanti, T., Kaboski, J., Martins, B., and Santos, C. (2021). Dispersion in nancing costs

  and development. NBER Working Paper, 28635.


Cavalcanti, T. and Santos, M. (2021). (MIS)Allocation eects of an overpaid public sector.

  Journal of the European Economic Association, 19(2):953999.


Cavalcanti, T. and Vaz, P. (2017). Access to Long-Term credit and productivity of small and

  medium rms: a causal evidence. Economics Letters, 150:2125.


Coelho, C., de Mello, J., and Funchal, B. (2012). The brazilian payroll lending experiment.

  Review of Economics and Statistics, 94(4):925934.


Colonnelli, E., Prem, M., and Teso, E. (2020).      Patronage and selection in public sector

  organizations. American Economic Review, 110(10):30713099.


Corseuil, C. H. and Moura, R. L. (2016).    O impacto do simples no nível de emprego da

  indústria brasileira. In Barbosa Filho, F., Ulyssea, G., and Veloso, F., editors, Causas e

  Consequências da Informalidade no Brasil, pages 187207. Editora Elsevier.




                                             40
Cusolito, A. and Maloney, W. (2018). Productivity Revisited: Shifting Paradigms in Analysis

  and Policy. World Bank, Washington, DC.


De Andrade, G. H., Bruhn, M., and McKenzie, D. (2016). A helping hand or the long arm

  of the law? experimental evidence on what governments can do to formalize rms. The

  World Bank Economic Review, 30(1):2454.


De Souza, G. (2021). The labor market consequences of appropriate technology.


De Vries, G. J. (2008). Did liberalization start a retail revolution in brazil? Technical report,

  GGDC Working Paper.


De Vries, G. J. (2014). Productivity in a distorted market: The case of brazil's retail sector.

  Review of Income and Wealth, 60(3):499524.


Devereux, M. and Guceri, I. (2015). An impact study of isenção scal em pesquisa de desen-

  volvimento: 'the good law'. Technical report, Centre for Business Taxation, University of

  Oxford.


Dix-Carneiro, R. (2014).    Trade liberalization and labor market dynamics.       Econometrica,

  82(3):825885.


Dix-Carneiro, R., Goldberg, P., Meghir, C., and Ulyssea, G. (2021). Trade and informality

  in the presence of labor market frictions and regulations.


Dix-Carneiro, R. and Kovak, B. K. (2017).        Trade liberalization and regional dynamics.

  American Economic Review, 107(12):29082946.


Dix-Carneiro, R. and Kovak, B. K. (2019). Margins of labor market adjustment to trade.

  Journal of International Economics, 117:125142.


Djankov, S., McLiesh, C., and Shleifer, A. (2007). Private credit in 129 countries. Journal

  of Financial Economics, 84:299329.


Djankov, S., Porta, R., Lopez-de Silanes, F., and Shleifer, A. (2002). The regulation of entry.

  Quarterly Journal of Economics, 117:137.


Dutz, M. (2018). Jobs and Growth: Brazil's Productivity Agenda. World Bank.


Engbom, N. and Moser, C. (2021). Earnings inequality and the minimum wage: Evidence

  from brazil. NBER Working Paper, 28831.




                                               41
Fajnzylber, P., Maloney, W., and Montes-Rojas, G. (2011). Does formality improve Micro-

  Firm performance? evidence from the brazilian SIMPLES program. Journal of Develop-

  ment Economics, 94(2):262276.


Ferreira, P. C. and Rossi, J. L. (2003). New evidence from brazil on trade liberalization and

  productivity growth. International Economic Review, 44(4):13831405.


Firpo, S. and Portella, A. (2019). Decline in wage inequality in brazil: A survey. World Bank

  Policy Research Working Paper, (9096).


Frischtak, C. (2013).   Infraestrutura e desenvolvimento no brasil.   In Veloso, F., Ferreira,

  P., Giambiagi, F., and Pessôa, S., editors, Desenvolvimento Econômico: Uma Perspectiva

  Brasileira, pages 322346. Elsevier.


Haanwinckel, D. and Soares, R. (2020).      Workforce composition, productivity, and labor

  regulations in a compensating dierentials theory of informality. BFI Working Paper.


Hsieh, C.-T. and Klenow, P. J. (2009). Misallocation and manufacturing tfp in china and

  india. The Quarterly Journal of Economics, 124(4):14031448.


Hsieh, C.-T. and Klenow, P. J. (2014). The life cycle of plants in india and mexico*. The

  Quarterly Journal of Economics.


Inklaar, R. and Timmer, M. (2014). The relative price of services. Review of Income and

  Wealth, 60(4):727746.


Joaquim, G. and Van Doornik, B. (2019). Bank competition, cost of credit and economic

  activity: Evidence from brazil. Working Paper.


Kannebley Júnior, S. and Porto, G. S. (2012). Incentivos scais à pesquisa, desenvolvimento

  e inovação no brasil: Uma avaliação das políticas recentes. Working Paper.


Klenow, P. and Li, H. (2020).    Innovative growth accounting.   In NBER Macroeconomics

  Annual 2020, volume 35. University of Chicago Press, Chicago, IL.


La Porta, R. and Shleifer, A. (2014).    Informality and development.   Journal of economic

  perspectives, 28(3):109126.


Lisboa, M., Menezes Filho, N., and Schor, A. (2010). The eects of trade liberalization on

  productivity growth in brazil: Competition or technology? Revista Brasileira de Economia,

  64(3):277289.



                                             42
Meghir, C., Narita, R., and Robin, J.-M. (2015a).      Wages and informality in developing

  countries. American Economic Review, 105(4):15091546.


Meghir, C., Narita, R., and Robin, J. M. (2015b).      Wages and informality in developing

  countries. American Economic Review, 105(4):15091546.


Menezes Filho, N. and Muendler, M.-A. (2011).       Labor reallocation in response to trade

  reform. NBER Working Paper, (17372).


Monteiro, J. (2016). Saindo da informalidade: O impacto do simples na formalização de em-

  presas. In Barbosa Filho, F., Ulyssea, G., and Veloso, F., editors, Causas e Consequências

  da Informalidade no Brasil, pages 169186. Editora Elsevier, Rio de Janeiro.


Perez Sebastian, F., Steinbucks, J., Féres, J., and Trotter, I. (2020). Electricity access and

  structural transformation: Evidence from brazil's electrication. Technical report, World

  Bank.


Piza, C. (2018). Out of the shadows? revisiting the impact of the brazilian SIMPLES program

  on rms' formalization rates. Journal of Development Economics, 134:125132.


Ponticelli, J. and Alencar, L. (2016).    Court enforcement, bank loans, and rm invest-

  ment:   Evidence from a bankruptcy reform in brazil.     Quarterly Journal of Economics,

  131(3):13651413.


Qian, R., Araújo, J., and Nucifora, A. (2018). Brazil's productivity dynamics. Background

  paper for Dutz, M. (2018). Jobs and Growth: Brazil's Productivity Agenda.


Ranasinghe, A. (2020).   Misallocation across establishment gender.     University of Alberta

  Working Paper.


Restuccia, D. and Rogerson, R. (2008). Policy distortions and aggregate productivity with

  heterogeneous establishments. Review of Economic Dynamics, 11(4):707720.


Restuccia, D. and Rogerson, R. (2017). The causes and costs of misallocation. Journal of

  Economic Perspectives, 31(3):151174.


Rocha, R., Ulyssea, G., and Rachter, L. (2018). Do lower taxes reduce informality? evidence

  from brazil. Journal of Development Economics, 134:2849.


Ulyssea, G. (2010). Regulation of entry, labor market institutions and the informal sector.

  Journal of Development Economics, 91(1):8799.




                                             43
Ulyssea, G. (2018). Firms, informality, and development: Theory and evidence from brazil.

  American Economic Review, 108(7):20152047.


Ulyssea, G. (2020a). Formal and informal rm dynamics. University College London Working

  Paper.


Ulyssea, G. (2020b). Informality: Causes and consequences for development. Annual Review

  of Economics, 12:525546.


Vasconcelos, R. (2017). Misallocation in the brazilian manufacturing sector. Brazilian Review

  of Econometrics, 37(2):191232.


Veloso, F., Matos, S., Ferreira, P., and Coelho, B. (2017). O brasil em comparações interna-

  cionais de produtividade: Uma análise setorial. In Bonelli, R., Veloso, F., and Pinheiro,

  A., editors, Anatomia da Produtividade no Brasil, pages 63107. Elsevier.


Zaourak, G. (2019). Rent-seeking activities, misallocation and innovation in brazil. Working

  Paper.




                                             44