Jurnal 3 MSDM Global

The Economic Journal, 132 (January), 361–390 https://doi.org/10.1093/ej/ueab052 C The Author(s) 2021.
Published by Oxford University Press on behalf of
Royal Economic Society. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.or
g/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
Advance Access Publication Date: 28 June 2021
ENFORCEME N T O F L A B O U R RE GUL ATI ON AND T HE

LABOUR MARKET EFFECTS O F TRADE: EVIDENCE FROM
BRAZIL∗
Vladimir Ponczek and Gabriel Ulyssea
Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

How does enforcement of labour regulations shape the labour market effects of trade? We combine local
economic shocks generated by the unilateral trade liberalisation in Brazil and enforcement variation across
regions to show that regions with stricter enforcement observed: (i) lower informality; (ii) larger losses in
overall employment; (iii) greater reductions in the number of formal plants. Regions with weaker enforcement
experienced opposite effects. All these effects are concentrated on low-skill workers. Our results indicate that
greater flexibility introduced by informality allows both formal firms and low-skill workers to cope better
with adverse labour market shocks.
Many developing countries, most notably in Latin America, underwent major trade liberalisation
episodes in the 1980s and early 1990s (Goldberg and Pavcnik, 2007). Despite the many expected
gains from trade, concerns about negative labour market consequences have always been present
in these countries.1 In particular, one major concern is that trade opening could induce a re-
allocation from formal to informal jobs, especially among less skilled workers (Goldberg and
Pavcnik, 2003). Since informal jobs are typically of lower quality and are not covered by labour
regulations nor social security, this informality effect could represent a large welfare loss from
trade opening.
Informality however also introduces greater de facto labour market flexibility, which can
be particularly relevant in the presence of burdensome and strict labour regulations. Higher
flexibility may help firms and workers to cope better with negative economic shocks, which could
reduce employment losses relative to a counterfactual scenario with perfect enforcement and no
informality. This conjecture has important implications for how one interprets the labour market
effects from trade and their potential consequences for welfare. More broadly, it implies that the
rigidity introduced by labour market regulations can lead to worse labour market outcomes and
potentially amplify employment losses from adverse economic shocks. This latter point directly
∗ Corresponding author: Gabriel Ulyssea, Department of Economics, University College London, London, WC1H
0AX, UK. Email: g.ulyssea@ucl.ac.uk
This paper was received on 15 September 2020 and accepted on 19 June 2021. The Editor was Nezih Guner.
The data and codes for this paper are available on the Journal website. They were checked for their ability to reproduce
the results presented in the paper.
This research is supported by the R4D initiative on Employment and labour market outcomes, funded by the Swiss
National Science Foundation and the Swiss Development Cooperation. We are thankful to Penny Goldberg, Rafael
Dix-Carneiro, Brian Kovak, Brian McCaig, Emmanuel Milet, Marcelo Olarreaga, Nicolas Depetris, four anonymous
referees and the Editor, as well as participants at various conferences and seminars for helpful comments and discussions.
Ponczek gratefully acknowledges financial support from CNPq. All remaining errors are ours.
1 A growing literature consistently documents that local economies that become more exposed to foreign competition
observe worse labour market outcomes relative to those that are less exposed (e.g., Autor et al., 2013; Kovak, 2013;
Costa et al., 2016; Dix-Carneiro and Kovak, 2017; 2019).
[ 361 ]
362 the economic journal [january
speaks to the extensive literature that analyses the consequences of labour regulations and labour
market rigidity for labour market performance.2
This paper tackles these issues by exploiting Brazil’s large-scale, unilateral trade liberalisation
episode of the early 1990s. Brazil is an attractive empirical setting for at least three reasons.
First, the unilateral trade liberalisation had substantial and heterogeneous effects across local
labour markets (e.g., Kovak, 2013; Dix-Carneiro and Kovak, 2017; 2019). Second, just before
the beginning of the trade liberalisation process, in 1988, Brazil underwent a major constitutional
reform that substantially increased the restrictiveness and the direct costs associated with labour
regulation (Barros and Corseuil, 2004).3 Third, enforcement of labour regulation varies greatly

across regions in Brazil (Almeida and Carneiro, 2012). We exploit the geographic variation in
the intensity of trade shocks and enforcement of labour regulation to assess if, and to what extent,
the presence of stricter enforcement of a costly regulatory framework shapes the labour market
responses to trade liberalisation. More broadly, we investigate whether greater de facto labour
market flexibility (introduced by informality) leads to lower employment losses in the face of an
adverse economic shock.
To get to these questions, we construct a measure of local, trade-induced shocks based on
changes in tariffs at the industry level combined with the initial sectoral composition of em-
ployment across regions, which remains fixed at the levels observed before the trade opening
process started (e.g., Topalova, 2010; Kovak, 2013). An important identification assumption is
the (conditional) exogeneity of these trade shocks relative to (unobserved) pre-existing trends in
local labour markets. Previous papers that use the same regional trade shock document direct
evidence in support of this assumption (e.g., Kovak, 2013; Dix-Carneiro and Kovak, 2017; 2019).
A second key aspect of our empirical strategy is the measurement of enforcement capacity and
intensity across local economies. In Brazil, enforcement of labour regulation is the sole respon-
sibility of the Ministry of Labour and the technology of enforcement is quite straightforward:
labour inspectors are assigned to labour offices (LOs) located in municipalities across the country
and they travel by car to inspect firms (Almeida and Carneiro, 2012). Hence, greater distances to
LOs imply that firms are less likely to be inspected and enforcement is more likely to be weak
(all things equal). We thus use distance to the nearest labour office as a proxy for enforcement
capacity in a given local market. We collect new data on the date of creation of all labour offices
in Brazil and restrict the analysis to those created before the trade opening process started. Hence,
our measure of enforcement capacity is pre-determined with respect to future trade shocks and
local labour market conditions.
We start by examining the basic effects of regional trade shocks on labour market outcomes. We
use individual-level Census data from 1991 and 2000 to compute local labour market outcomes
net of the influence of socio-demographic variables (e.g., gender, schooling and age). Our results
confirm the findings of previous studies: between 1991 and 2000, regions more exposed to
the trade liberalisation shock experienced a substantial increase in both informality and non-
employment relative to regions less exposed.4 The novelty here comes from the analysis of
these effects across skill levels, which shows striking results. Almost all negative effects on
2 This is an extensive literature, which we discuss below.
3 According to the employment index in Botero et al. (2004), the cost of labour regulation in Brazil is 20% above the
mean and median of 85 countries in the world and more than 2.5 times as large as in the United States.
4 Dix-Carneiro and Kovak (2019) also showed that in the longer run (until 2010) the effects on non-employment
vanish but those on informality persist (and even amplify). We focus on the period up to 2000 because our focus lies on
investigating the extent to which stricter enforcement reduces/amplifies employment effects in the aftermath of a negative
demand shock.

C 2022 Royal Economic Society.
2022] enforcement and labour market effects of trade 363
informality, non-employment and wages come from low-skill workers. High-skill workers show
no informality effects and much smaller and marginally significant effects on non-employment
and wages.
We then move to the focus of the paper, which is the analysis of heterogeneous effects across en-
forcement capacity levels, which is proxied by the maximum driving distance to the nearest labour
office. The results show that regions with higher enforcement capacity observe lower informality
but greater non-employment effects as a response to the trade shock. Symmetrically, regions with
lower enforcement capacity experience greater informality but lower non-employment effects.
Again, all the effects are concentrated on low-skill workers, with no heterogeneous effects among

high-skill workers. The magnitudes of these effects are large. For a strong local trade shock (tariff
reduction of 0.1 log points), a region with very low enforcement capacity (90th percentile of
the distance distribution) would experience an increase of 10 percentage points on informality,
but nearly zero effects on non-employment. In contrast, a region with high enforcement capacity
(10th percentile of the distance distribution) would experience a 3 percentage points increase in
informality but a much stronger increase in non-employment, of 3.9 percentage points. As for
wages, the effects are large in magnitude and go in the expected direction, i.e., when enforcement
is weaker, there are greater wage losses. However, the coefficients are imprecisely estimated and
we find no statistically significant heterogeneous effects on wages for neither skill group in our
benchmark specifications.
To complement these results from the Demographic Census, we use administrative data from
the Ministry of Labour that contain the universe of formal firms and workers. We find that
regions with weaker enforcement experience a stronger reduction in formal employment relative
to regions with greater enforcement capacity. This indicates that, beyond preserving jobs, there
is also a ‘switching effect’ from formal to informal jobs in harder-hit regions with weaker
enforcement. However, we show that the higher flexibility introduced by weaker enforcement
also leads to greater survival of formal establishments, as regions with weaker enforcement
capacity have smaller losses in the number of formal plants. This result is consistent with the fact
that a large fraction of informal employment is located in formal firms, the so-called intensive
margin of informality (Ulyssea, 2018). Moreover, it indicates that the intensive margin plays an
important role in formal firms’ survival in the face of an adverse economic shock.
In sum, our results show that, in the years following the unilateral trade liberalisation, regions
with stricter enforcement (and therefore higher labour market rigidity) experienced the following:
(i) less switching from formal to informal jobs and lower overall informality effects; (ii) larger
employment losses and (iii) greater reductions in the number of formal plants. The opposite is
observed in regions where enforcement capacity is weak and de facto labour market flexibility is
high: there are strong informality effects, but no statistically significant effects on employment
and greater survival of formal firms. These effects are completely driven by low-skill workers. Our
results therefore indicate that greater de facto labour market flexibility introduced by informality
allows both formal firms and low-skill workers to cope better with adverse labour market shocks.
Put differently, the results suggest that informality acts as an employment buffer in the face of
negative economic shocks, but this seems to be the case only for low-skill workers.
Even though our measure of enforcement capacity is pre-determined, one potential concern
about the empirical strategy is that the location of labour offices is obviously not random.
Thus, the enforcement capacity measure could be capturing the effects of other characteristics
that are not accounted for in our specifications. For example, distances to labour offices could
be capturing how remote a given area is, i.e., less connected to important economic centres.

More remote regions may also have a higher proportion of low-productivity firms, which are
more likely to respond to negative shocks by increasing informal employment. We address
this and other potential threats to identification in different ways. We start by noting that all
regressions are estimated in first difference, therefore accounting for time-invariant, unobserved
local economies’ characteristics. Our benchmark specifications also allow for state-specific trends
and for differential trends across micro-regions with different initial demographic conditions
within states.
The threat to identification would hence have to come from some omitted, time-varying,
regional characteristic that drives labour market outcomes within a given state, which is not

captured by local economies’ initial socio-demographic conditions, and which is also correlated
with the location of labour offices. The extensive robustness analysis in Subsection 3.3 shows
that this is unlikely to be the case, as our results are robust to a variety of additional potential
confounders. We start by investigating whether reversion to the mean across regions with lower
and higher initial levels of informality and non-employment could be driving our results. This
would be the case if our enforcement capacity measure was capturing heterogeneity in initial
informality levels, rather than enforcement capacity per se. This is not the case, as the results
are robust to the inclusion of 1980’s informality and non-employment rates. We also investigate
whether our results are capturing the ‘remoteness effect’ mentioned above. We do so by including
the median driving distance to the state’s capital, thus allowing for differential trends by proximity
to the capital, and the results remain unchanged (if anything, they become stronger). Finally, we
control for local government per capita spending and the Gini coefficient, both measured in
1991. The former aims to control for the possibility that distance to the labour office could be a
proxy for availability of local public goods and infrastructure in general, rather than enforcement
capacity per se. The latter controls for the initial level of inequality in the region, which is also a
relevant indicator of local economic conditions. Our results remain unchanged in both cases.
In the final part of the paper, we turn our focus to a more direct measure of enforcement
intensity, which is the total number of inspections per formal firms in a given local market.
This measure is potentially subject to measurement error and it is likely to be an endogenous
regressor. We thus use distance to the nearest LO as an instrument for enforcement in a limited
information maximum likelihood estimator. Again, considering a high intensity local trade shock
(tariff reduction of 0.1 log points), low-skill workers in a region with weak enforcement—0.9
inspections per 100 firms (the 10th percentile)—would experience an increase of 12.1 percentage
points in informality but no disemployment effects. In contrast, low-skill workers in a region
with strict enforcement—17.2 inspections per 100 firms (the 90th percentile)—would experience
no informality effects but an increase of 10.3 percentage points in non-employment rates. It is
worth noting that the regional trade shock is associated with an increase of 5.2 and 2.7 percentage
points in informality and non-employment among low-skill workers, respectively. Hence, the
strength of enforcement (or lack thereof) can lead to labour market responses in both informality
or non-employment that are substantially larger than the average effect from the trade shock.
Our paper contributes to three literature streams. First, the literature on trade and local labour
markets, which includes (but is not restricted to) Topalova (2010); Autor et al. (2013); Kovak
(2013); Hakobyan and McLaren (2016); and Carneiro and Kovak (2017; 2019). Second, the
literature that analyzes the relationship between trade opening and informality (Goldberg and
Pavcnik, 2003; Menezes-Filho and Muendler, 2011; Bosch et al., 2012; Dix-Carneiro and Kovak,
2017; 2019). In contrast with both literature streams, we focus on a new dimension: the interaction
between the enforcement of labour regulation and trade policies, and how these interactions shape

the labour market adjustment to trade shocks. Also, importantly, we document that the effects of
trade opening on informality and non-employment are mostly concentrated on low-skill workers,
both on average and across different enforcement levels. Finally, our paper also dialogues with the
literature that argues that informality introduces de facto flexibility to otherwise very rigid formal
labour markets that are subject to burdensome and costly regulatory frameworks (e.g., Meghir
et al., 2015; Ulyssea, 2018).5 More broadly, our results speak to the extensive literature about the
consequences of labour regulations and labour market rigidity (e.g., Besley and Burgess, 2004;
Botero et al., 2004; Nickell et al., 2004; Tella and MacCulloch, 2005; Freeman, 2010; Adhvaryu
et al., 2013; Almeida and Poole, 2017).

The remainder of the paper is structured as follows. In Section 1 we briefly describe the trade
liberalisation process and the measure of local trade shock used, the institutional background and
data. In Section 2 we describe the empirical strategy, while in Section 3 we present the empirical
results. We conclude in Section 4.
1. Background and Data

1.1. Trade Liberalisation and Local Trade Shocks in Brazil
Until 1990, Brazil was characterised by a complex system of protection against foreign competi-
tion that included both tariff and non-tariff barriers (Kume et al., 2003). In 1988–1989, there was
a first move towards reforming the structure of protection, which reduced tariff redundancy and
special regimes, among other measures. In March 1990, the newly elected president unexpectedly
eliminated non-tariff barriers, typically replacing them with higher import tariffs in a process
known as ‘tariffication’. This implied that, from 1990 onwards, tariffs started to accurately reflect
the actual level of protection faced by Brazilian industries.6
From 1990 until 1995, Brazil implemented a major unilateral reduction in trade tariffs. During
this period, the average tariff fell from 30.5% to 12.8% and the SD fell from 14.9% to 7.4%.7
Hence, not only did the overall level of protection decrease, but the variation across industries
was also substantially reduced. Figure 1 shows the percentage change in tariffs across the main
industries, which is one of the sources of variation we exploit in our identification strategy, as
discussed below. As the figure shows, there was substantial variation across sectors in tariff
reductions. Moreover, tariff cuts were strongly and negatively correlated with pre-liberalisation
tariff levels: industries with initially higher levels of protection (i.e., tariffs) experienced larger
tariff reductions (Kovak, 2013).
The measure of local trade shocks exploits the fact that regions with larger employment shares
in industries that experienced greater tariff reductions were more likely to be affected by the
trade opening process. Put differently, the unilateral trade liberalisation episode is more likely
to represent a substantial negative labour demand shock in regions with a larger fraction of its
labour force employed in industries that faced larger tariff cuts (relative to regions with a larger
fraction of employment in industries less affected). We use tariff data from Kume et al. (2003)
5 In an earlier paper, Boeri and Garibaldi (2005) argued that despite advances in monitoring capacity by the government,
informality is ‘tolerated’ because it attenuates unemployment. Similarly, Ulyssea (2010) quantitatively showed in a two-
sector matching model that higher enforcement leads to higher unemployment and lower welfare. Even though our
empirical results refer to a very different setting, they are consistent with the mechanisms highlighted in both papers.
6 A more detailed description of the trade liberalisation in Brazil can be found in Kovak (2013) and Dix-Carneiro and
Kovak (2017).
7 There were minor changes in tariffs after 1995, which are not relevant compared to the changes that occurred in the
1990–1995 period.

Change in log(1+tariff), 1990–1995
0.00
–0.05
–0.10
–0.15
–0.20

–0.25
Agriculture
Metals
Apparel
Food processing
Wood, furniture, peat
Textiles
Paper, publishing, printing

Mineral mining
Chemicals
Electric, electronic equip.

Auto, transport, vehicles
Machinery, equipment
Plastics
Other manuf.
Pharma., perfumes, detergents
Petroleum refining
Petroleum, gas, coal

Footwear, leather
Rubber
Non-metallic mineral manuf
Fig. 1. Changes in log(1 + tariff), 1990–1995.

Source. Dix-Carneiro and Kovak (2017). [Copyright American Economic Association; reproduced with
permission of the American Economic Review].
to construct the ‘regional tariff change’ (RTC) as proposed by Kovak (2013):

RTCr = βri d ln(1 + τi ).
i
Here
λri /θi
βri = ,
i λri /θi
with λri = L ri /L r the fraction of labour allocated to industry i in region r and θi equal to
one minus the wage bill share of industry i. It is worth emphasising that we use changes in
output tariffs to construct RTCr . Alternatively, one could use effective rates of protection, which
incorporate both input and output tariffs. However, at the level of industry classification used
here—which is standard in the literature—changes in input and output tariffs are highly correlated
and the regional tariff changes computed using either measure (output tariffs or effective rates of
protection) are almost perfectly correlated (Dix-Carneiro et al., 2018).8
Since we investigate the effects of trade liberalisation on skilled and unskilled workers sepa-
rately, we also compute different regional tariff shock measures for skilled and unskilled workers,
which we denote RTCr,k , where k denotes the skill group. For that, we follow a similar approach
8 In the Online Appendix, we examine the robustness of our results to different measures of local trade shocks, such
as the ratio of imports to production and the import penetration coefficient. All of our results remain largely unchanged.

to that of Autor et al. (2018) and compute weights that are specific to the skill groups, which
are given by λrik = L rik /L r k .9 As Figure A.1 in the Online Appendix shows, the measure of
regional trade shock for low-skill workers is highly correlated with the overall measure (RTCr ).
This is expected, as low-skill workers correspond to the vast majority of the labour force. The
RTC measure for high-skill workers is also strongly correlated with the overall measure, but less
than that for low-skill workers.
1.2. Labour Regulations and Enforcement

In Brazil, the permissible types of labour contract, their conditions and terms for termination
are completely regulated by a labour code based on the civil law system, the Consolidação das
Leis Trabalhistas, which dates back to 1943. As part of the labour regulations in Brazil, formal
workers are required to hold a booklet issued by the Ministry of Labour that must be signed by
the employer and that contains workers’ entire formal employment histories. Having the labour
contract registered in this booklet entitles workers to a series of rights and benefits, such as
unemployment insurance, severance payment, a one-month paid vacation and a 50% premium
for overtime hours.
The labour regulation introduced in 1943 was already quite detailed, extensive and rigid. The
new Federal Constitution enacted in 1988 substantially extended the range of labour regulations
and workers’ benefits, further increasing the regulatory hiring and firing costs.10 Labour taxes are
also quite high in Brazil, with the main components being social security contributions (20%),
direct payroll taxes (9%) and contributions to a severance fund (8.5%). Computing the overall
labour tax rate as the share of commercial profits (which provides a cross-country comparison),
it amounts to 42.1% in Brazil, 12.9% in Canada and 10% in the United States. Not only is the
tax rate is high, but there are also substantial compliance costs involved. The time required to
pay labour taxes in Brazil is nearly 5 times higher than in the United States, 491 and 100 hours,
respectively (The World Bank, 2007).
Given how cumbersome and costly the labour regulation is, both firms and workers have
incentives to either partially comply or avoid it entirely via informal labour contracts. In such an
environment where incentives to formalisation are arguably weak, enforcement plays a substantial
role in determining not only informality levels, but labour market outcomes more broadly.
The Ministry of Labour is directly responsible for enforcing labour regulations, but it only
inspects registered firms and therefore it does not tackle informal labour in informal firms. En-
forcement is implemented in a very decentralised way, both at the state level (with a labour office
called delegacia do trabalho) and within states through local labour offices called subdelegacias.
The state-level office (delegacia) is always located in the state’s capital and the local offices
(subdelegacias) are spread out across municipalities. The number of local offices is a function of
the state’s size and economic relevance (Almeida and Carneiro, 2012).
Inspectors are allocated to a specific subdelegacia and they travel by car to inspect firms. Most
inspections are triggered by anonymous reports and inspectors must assess compliance with all
the relevant dimensions of the labour code and not only if workers are formally registered or
not. There is substantial regional, within-state variation in enforcement intensity (Almeida and
Carneiro, 2012). In particular, one of the factors that determines regional variation in enforcement
9 See Dix-Carneiro and Kovak (2015) for a more general discussion within a specific-factors model of regional
economies with two types of worker.
10 See Barros and Corseuil (2004) for a complete description of the changes introduced by the 1988 Constitution.

is the relative density of local labour offices, which by its turn determines the travel distances that
inspectors face in order to carry out inspections. Thus, these travel distances are a key determinant
of the capacity of enforcement across local labour markets.
1.3. Data
Throughout the paper, our main unit of analysis is the micro-region, which is a collection of
contiguous municipalities that are economically integrated. The micro-regions are defined by the
National Bureau of Statistics (IBGE) and closely reproduce the idea of local economies (similar

to commuting zones in the United States, as in Autor et al., 2013), which have been extensively
used in the recent literature (e.g., Costa et al., 2016; Dix-Carneiro and Kovak, 2017).11 We use a
mapping between municipalities and micro-regions that results in 411 consistent micro-regions
between 1980 and 2000. Our analysis focuses on the changes between 1990 and 2000, but we
use the 1980 census to control for baseline characteristics.
We use four datasets in our empirical exercise. The first is the Decennial Population Census,
which contains information on individuals’ socio-economic characteristics, as well as labour
market outcomes. Particularly important for our exercise, the census provides information on
workers’ informality status. We define as informal workers those employees who do not hold a
formal contract, which in Brazil is characterised by a ‘signed work booklet’ (as discussed in the
previous section). Workers who report being employees are directly asked whether they have a
formal contract, which is the information we use to define if an employee is formal or not. Our
measure of informality therefore excludes the self-employed. We do so because the mechanisms
on which we focus here, and in particular our measures of enforcement, refer to employees only.
Nevertheless, we also separately analyse the effects on the self-employed, as individuals can
respond to worsening labour market conditions by resorting to self-employment.
The second data set contains administrative data from the Ministry of Labour related to
enforcement activity. This dataset contains yearly information on the number of firms inspected
by municipality from 1995 to 2013, the number of inspectors responsible for the auditing process
in each state of the country and the locations of all labour offices. We add to this administrative
data set a crucial piece of information, namely the date of creation of each labour office (i.e.,
subdelegacia). It was necessary to directly call each of the 121 labour offices in Brazil to collect
this information. Among those, ninety-two offices were created prior to 1990 (the start year of the
trade opening process), nineteen offices were created between 1990 and 2000 and the remaining
were created after 2000. The third data set contains the driving distance to the nearest labour
office in each municipality, the distance to the state’s capital and the number of inspectors at
the state level compiled by Almeida and Carneiro (2012). Finally, we use the Relação Anual
de Informações Sociais (RAIS), which is an administrative data set collected by the Ministry of
Labour that contains the universe of formal firms and workers. We use the RAIS to compute the
number of formal establishments in each micro-region.
To obtain a measure of enforcement capacity at the micro-region level, we compute the
maximum distance to the nearest labour office within each micro-region.12 As discussed in the
previous section, a greater distance is associated with weaker enforcement capacity in the micro-
region. As for the measure of actual enforcement intensity, we compute the ratio between the
11 Indeed, the data show that only 3.2% of workers lived and worked in different micro-regions in 2000 (Dix-Carneiro
et al., 2018).
12 We also use the average distance in our robustness analysis and our results remain largely unchanged.

Table 1. Basic Descriptive Statistics at the Micro-Region Level.
Sample: All workers Low skill High skill
Mean SD Mean SD Mean SD
Non-employment in 1991 0.390 0.045 0.411 0.043 0.221 0.039
Informality in 1991 0.443 0.202 0.464 0.205 0.290 0.170
Self-employment in 1991 0.182 0.065 0.190 0.068 0.104 0.025
Log wages in 1991 0.795 0.420 0.650 0.376 1.605 0.402
RTCr 0.045 0.040 – – – –
RTCr unskilled – – 0.043 0.039 – –
RTCr skilled – – – – 0.091 0.035
Distance to the LO (per 100 km) 2.666 1.855 – – – –

Distance to the capital 4.096 2.484 – – – –
Number of inspections per 100 firms 7.430 8.140 – – – –
Share female in 1991 0.500 0.018 – – – –
Share high skill in 1991 0.116 0.060 – – – –
Share urban areas in 1991 0.616 0.197 – – – –
Number of micro-regions 411
Notes: We use individual-level data and sampling weights from the Demographic Census to compute simple non-
employment and informality rates, average wages, and shares of female, high-skill and urban populations at the micro-
region level. Distance to the LO, distance to the state’s capital, the number of inspections and the regional trade shocks
are calculated as described in the text. Means and SDs refer to the distribution of these means at the micro-region level.
total number of inspections carried out in 1995–1999 and the total number of formal firms in a
given micro-region.
Figure 2 displays the kind of variation we will be exploring throughout our analysis. Panel
(a) shows how the regional trade shock, RTCr , varies across micro-regions in Brazil. Panel (b)
displays the regional variation in enforcement intensity (i.e., inspections per hundred firms), as
well as the location of all ninety-two local labour offices (subdelegacias) created prior to 1990
in Brazil.13 As the figure shows, there is substantial variation in the trade shock, the intensity of
enforcement and in the density of local labour offices across micro-regions.
Table 1 provides the descriptive statistics of the variables used in our analysis, which are all
constructed at the micro-region level. We define low-skill workers as those who did not complete
high school, and high-skill workers as those with at least a high school diploma. Non-employment
is defined as the sum of individuals actively looking for jobs (the unemployed) and those out of
the labour force. We use this measure (instead of unemployment) to reduce measurement error,
as these two states tend to be less distinct in less-developed economies (Donovan et al., 2021).
The table shows that micro-regions are quite heterogeneous in terms of initial conditions, as most
variables have a high level of dispersion. The same is true for the trade shock, which is high on
average but also has a very high SD.
2. Empirical Strategy
In this section we describe our empirical strategy. However, before proceeding to the discussion
of the empirical specifications, it is useful to describe the economic mechanisms that guide our
empirical exercise. We do so by using the framework developed in Ulyssea (2018). The interested
13 One could also argue that forward-looking policy makers would choose the location of the pre-1990 labour offices
based on the expected impact of the tariff reduction policy. However, it is important to note that a new administration
took office in 1990. Hence, all labour offices used in our exercise were created by previous administrations, which could
not anticipate the sudden and unexpected change in trade policy.

Fig. 2. Regional Tariff Changes and Enforcement of Labour Regulation across Regions.
Notes: Map of RCTr from Dix-Carneiro et al. (2018). Regional distribution of inspections and labour
offices obtained using administrative data from the Ministry of Labour (see the text).

reader is referred to that paper for a complete discussion of the model, while the reader only
interested in the empirical exercise can skip directly to Subsection 2.2.
2.1. Theoretical Framework

In the model developed by Ulyssea (2018), firms can exploit two margins of informality. The
first is the extensive margin, which refers to firms’ decision to pay entry fees and register their
business or not. The second is the intensive margin, which refers to the decision of a formal
(registered) firm to hire workers without a formal contract. Firms sort between sectors upon entry

based on their expected productivity and the (in)formal sector is comprised of (un)registered
firms.
If a firm enters the informal sector, it avoids registration costs and taxes altogether, but faces
an expected cost associated with informality that increases with the firm’s size. This can be
rationalised by the fact that larger firms are more visible to the government and are detected with
greater probability.14 If a firm decides to enter the formal sector, it faces fixed registration costs
and must pay revenue and labour taxes. The latter can be avoided by hiring informal workers.
However, formal firms that hire informal workers also face a probability of detection, which
increases with in the number of informal workers. Thus, smaller formal firms will hire a larger
fraction of their labour force informally and this share decreases with firm size. Since productivity
and size are one-to-one in the model, more productive firms (in expectation) self-select into the
formal sector and less productive firms enter the informal sector. Similarly, conditional on being
formal, less productive firms hire a larger fraction of informal workers.
Firms hire both low- and high-skill workers that are aggregated into a composite labour input
through a CES production function. Skill shares may differ across formal and informal firms,
and the author’s estimates imply that the formal sector is more intensive in high-skill workers.
This is consistent with the fact that informality is more prevalent among low-skill workers.
The estimation results also indicate that the expected cost of hiring informal workers is lower
for low-skill than high-skill individuals, which suggests that higher enforcement would have
stronger effects on low-skill workers. The government has two possible enforcement levers:
to increase enforcement on the intensive margin of informality, by increasing inspections on
formal firms; and to increase enforcement on the extensive margin, by increasing the intensity
of inspections on informal firms. In Brazil, these two types of enforcement activity are carried
out by different government branches. The enforcement capacity that we examine in this paper
refers to enforcement on the intensive margin, which is conducted by the Ministry of Labour.
Even though Ulyssea’s framework does not include international trade, the unilateral trade
opening episode in Brazil can be interpreted through the lenses of the model as a competition
shock, which drives down prices for domestic firms (both formal and informal). In order to
illustrate the mechanism that we have in mind, we take the estimated model in Ulyssea (2018)
and simulate the value functions of being formal before and after the trade opening shock, which
is parameterised as a permanent decline in the equilibrium price. Even though in Ulyssea’s model
prices (i.e., wages) fully adjust in equilibrium, Figure 3 shows the results of a partial equilibrium
simulation, where we show firms’ payoffs after a one-time price reduction (equivalent to an
increase in real wages). The horizontal axis represents firms’ productivity, denoted by θ , and
the vertical axis represents firms’ payoffs under different scenarios. For the sake of expositional
14 This is a common formulation in the literature; see, for example, de Paula and Scheinkman (2010) and Leal Ordonez
(2014), among others.


f
ɾ1 ɾ2 ɾ3 ɾ4
Fig. 3. Trade Opening under Low and High Enforcement.
simplicity, we assume that the negative price shock only affects formal firms, but all that we need
to assume is that formal firms are at least as adversely affected as informal ones (but possibly
more).
We consider two scenarios for informal firms: low and high enforcement on the extensive
margin of informality. We do so because this is the margin that generates greater effects on firms
and therefore are easier to visualise in the graph. The exercise would be analogous if we were to
consider enforcement on the intensive margin, as they go in the same direction. Figure 3 shows
the four corresponding curves.
Consider first the situation prior to the trade shock with the two markets, low and high levels
of enforcement (dashed red line, solid black lines and red lines). In the market with a high level
of enforcement, all firms with productivity θ < θ1 will optimally choose to be informal as their
payoff is higher than that associated with formality (red dashed line above the black solid line).
In the market with a low level of enforcement, firms with productivity θ < θ2 will choose to be
informal, which shows that, for any given distribution of firm productivity, the market with lower
levels of enforcement will have a larger share of informal firms (and workers), as expected.
When the trade shock hits, the high and low enforcement markets observe an increase in
the informality thresholds from θ1 to θ3 and θ2 to θ4 , respectively. However, the impact on

informality will be larger in the market with a lower level of enforcement, as more firms can
resort to informality to cope with the shock instead of simply exiting. The intuition here is that
firms adversely affected can downsize and resort to informality if enforcement is weak (the costs
of informality increase with the firm’s size), but this option is not available if enforcement is high.
Thus, with weaker enforcement, there is a larger increase in informality, but a smaller reduction
in the mass of active firms. With stronger enforcement, there are lower informality effects, but
greater reductions in the mass of active firms. The model does not have unemployment, but the
effects on the mass of firms indicate what the effects on unemployment could be, as greater
displacement of firms is likely to be associated with higher unemployment. Finally, note that

informal firms are more intensive in low-skill workers and formal firms hire a larger fraction of
low-skill workers without a formal contract (in contrast to high-skill workers). Hence, the model
would predict that these movements towards informality would be stronger for low-skill than
high-skill workers.
In sum, this model implies that, in the aftermath of a unilateral trade shock, markets that have
lower levels of enforcement would experience a stronger increase in informality, but a lower
reduction in the number of plants, including formal ones. Conversely, markets that have higher
levels of enforcement would observe a smaller increase in informality, but greater firm exit and
potentially larger disemployment effects. The effects on employment and informality would be
largely concentrated among low-skill workers.
2.2. Empirical Specification

Our empirical strategy consists of two steps. In the first step, we capture the changes in the
outcome of interest at the micro-region level, netting out the influence of individuals’ socio-
demographic characteristics. More concretely, we run the following regressions at the individual
level:

Yit = γr t Dr + xi,t βt + i,t . (1)
r
Here i indexes individuals, t = 1991, 2000 denotes the year, Dr denotes the set of micro-region
dummies and xi,t is a vector of individual characteristics that includes age, age squared, schooling,
gender and race.
The outcomes considered are wages, a dummy for whether the individual is an informal
employee, dummy for self-employed and a dummy for non-employment, which includes both
unemployment and out-of-the-labour-force statuses. As discussed in Section 1, the informality
dummy considers only those individuals who work as employees in the private sector, and we
define as informal workers those without a formal contract (no signed work booklet). Since we are
analysing enforcement of labour regulation, which only applies to employees, this informality
definition is the most consistent with the goals of our empirical exercise. Nevertheless, we
also discuss the effects on self-employment, formal employment and the number of formal
establishments in Section 3, as these can be important margins of adjustment in the aftermath
of the trade shock (Dix-Carneiro and Kovak, 2019) and therefore can shed light on the different
forces at play.
The first step thus provides us with a measure of average wages, informality and non-
employment rates at the micro-region level. In order to assess the heterogeneous effects across

skill groups, we also estimate regression (1) separately for low- and high-skill workers and obtain
separate estimates of γ̂ by skill level.
In the second step, we run regressions in first difference at the micro-region level. The first set
of regressions we estimate re-visit the overall labour market impacts of the local trade shock and
provide new evidence on the heterogeneity across skill levels. The basic specification is
ŷr = ζ0 + ζ1 RTCr + α4 Zr + δs + u r , (2)
where r indexes regions, ŷr ≡ γ̂r,2000 − γ̂r,1991 , Zr denotes the set of controls used in all
regressions and δs denotes the state dummies, which absorb differential state-level trends. The

vector Zr includes the controls that are used in all specifications in the paper. It includes the main
baseline demographic characteristics of the micro-regions, which could influence the labour
market outcomes that we analyse, in particular, the share of women, high-skill individuals, urban
population and total population (in logarithms), all measured in 1991.
Our main goal, however, is to investigate how enforcement of labour regulations interacts
with the trade shock in shaping local labour market responses. For that, we directly assess how
distances to the labour offices might shape the labour market effects from trade. More specifically,
we estimate the regression
ŷr = α0 + α1 RTCr + α2 RTCr × Distr + α3 Distr + α4 Zr
+ α5 Distr × Inspectorss + δs + εr , (3)
where again r denotes the micro-region, δs the state dummies, Zr is the same vector of controls
described above (used in all regressions) and Distr denotes the maximum distance to the nearest
labour office of micro-region r , which is our enforcement capacity variable.15 We also follow
Almeida and Carneiro (2012) and include the interaction between the number of inspectors at the
state level (Inspectorss ) and the distance to the labour office, Distr . The motivation is that, for any
given distance, the number of inspectors available at the state level (lowest level of disaggregation
available) is an important determinant of enforcement, as it provides a measure of the resources
available at the state level.
Since we are using the first-stage estimates as dependent variables, we follow the previous
literature and weight the second-stage regressions by the inverse of the SEs of the first stage (e.g.,
Kovak, 2013; Dix-Carneiro and Kovak, 2017; Dix-Carneiro et al., 2018).16 When estimating
heterogeneous effects across skill levels, we use the same specification but with the appropriate
γ̂ estimated separately by skill level and the appropriate definition of RTC discussed in the
previous section.
2.2.1. Identification: discussion

A possible critique to our empirical strategy is the absence of random variation in enforcement
capacity levels, as the distance to the nearest labour office is certainly not randomly distributed
across micro-regions. Therefore, one could argue that this variable might be capturing the effect
15 We first calculate the distance from the centroid of each municipality to the nearest labour office. The maximum of
the distances to municipalities that belong to the micro-region r is defined as the enforcement capacity of micro-region r .
16 In Online Appendix C, we show that our results are robust to not using any weights as well. It is also worth noting
that part of the literature uses clustered SEs at the level of aggregation immediately above micro-regions, which in the
Brazilian case would correspond to the meso-region. However, the intra-cluster correlation of our variable of interest,
RTCr × Distr , is very close to zero, which indicates that this clustering is not adequate in our context. Nevertheless,
Online Appendix C shows that our results are robust to using clustered SEs at the meso-region level.

of other characteristics of local economies that are not accounted for in our specifications. We
start by noting that, as discussed in Section 1, we restrict ourselves to the labour offices created
up until 1990 when constructing the variable Distr . Thus, we are only using the pre-determined
enforcement capacity, which is not responding to the (future) local trade shock and local labour
market conditions.
As we estimate the regressions in first difference, our specification accounts for micro-region
fixed effects and state-specific trends,17 as well as differential trends across micro-regions with
different initial conditions in terms of demographics and size (we control for the share of women,
high-skill individuals, urban population and total population in 1991).

That said, a potential threat to identification could be the existence of reversion to the mean
across regions with lower and higher levels of informality and non-employment. If our variable of
interest (RTCr × Distr ) is simply capturing heterogeneity in initial informality levels, our results
would not be identifying the heterogeneous effects of trade across enforcement capacity levels.
Instead, they would simply reflect differential trends across high- and low-informality regions. In
Subsection 3.3 we discuss a series of robustness tests, which shows that our results are robust to
the inclusion of 1980’s informality and non-employment rates in the first-differenced regression
(expression (3)). We also examine how our results are affected by the inclusion of other controls
that account for potential confounding effects, including: (i) differential trends across more and
less remote regions (further away from large urban centres); (ii) local supply of public goods,
proxied by local government spending and (iii) initial level of inequality in the micro-region. As
we discuss in Subsection 3.3, the results remain unchanged and, if anything, become stronger.
Even if one argues that some important determinant of local economic development remains
unaccounted for—and that it is being captured by the enforcement capacity measure—one
would expect this omitted variable to produce heterogeneous effects on informality and non-
employment that go in the same direction. Put differently, it is hard to think of an omitted
variable that could lead simultaneously to greater informality and lower disemployment effects
as a response to trade liberalisation, within the same state and after controlling for the different
variables discussed above. Even more so because this pattern is only observed among low-skill
workers, while high-skill workers do not seem to be affected by the level of enforcement in their
region.
3. Enforcement of Labour Regulations and the Labour Market Effects of Trade

We start our analysis by re-visiting the basic average results on informality, non-employment and
wages found in the literature. Table 2 shows the estimates of regression (2) using as the dependent
variable the decadal changes in informality (columns (1)–(3)), non-employment (columns (4)–
(6)) and wages (columns (7)–(9)). For all three outcomes, we assess the effects on all workers,
low-skill and high-skill workers. Importantly, since the variable RTCr refers essentially to tariff
cuts, it is negative for almost all regions. We therefore work with the negative of RTCr , so that
the results are more easily interpretable.
Columns (1), (4) and (7) show the same patterns found in previous studies (e.g., Dix-Carneiro
and Kovak, 2017; Dix-Carneiro et al., 2018): between 1991 and 2000, regions that were hit
harder by the trade opening process experienced an increase in informality and non-employment
17 The inclusion of state dummies is important because many relevant policies and resources are defined at the state
level (e.g., police force and a substantial fraction of health and education expenditures).

376
Table 2. Basic Effects on Informality, Non-Employment and Wages.

Informality Non-employment Wages
Sample (by workers’ skill level): All Low High All Low High All Low High
(1) (2) (3) (4) (5) (6) (7) (8) (9)
RTCr 0.451∗∗∗ 0.206∗∗ −1.062∗∗∗
(0.130) (0.082) (0.221)
RTCr unskilled 0.520∗∗∗ 0.267∗∗∗ −0.930∗∗∗
(0.131) (0.091) (0.231)
RTCr skilled −0.093 0.119∗ −0.408
(0.191) (0.066) (0.295)
Observations 411 411 411 411 411 411 411 411 411
the economic journal
R2 0.375 0.409 0.315 0.395 0.402 0.319 0.608 0.558 0.588
Notes: Robust SEs reported. ∗∗∗ , ∗∗ , ∗ denote significance at the 1%, 5% and 10% levels, respectively. All regressions include state fixed effects and the following demographic
controls: share of women, high-skill individuals, urban population and log population in 1991.
C

[january
2022 Royal Economic Society.

Table 3. Effects on Informality and Non-Employment by Enforcement Capacity Level.
Informality Non-employment
Sample (by workers’ skill level): All Low High All Low High
(1) (2) (3) (4) (5) (6)
RTCr 0.211 0.348∗∗∗
(0.155) (0.106)
RTCr × Dist. LOr 0.208∗∗ −0.083
(0.097) (0.066)
RTCr unskilled 0.162 0.453∗∗∗
(0.158) (0.113)
RTCr unskilled × Dist. LOr 0.326∗∗∗ −0.148∗∗

(0.106) (0.073)
RTCr skilled −0.205 0.121
(0.323) (0.127)
RTCr skilled × Dist. LOr 0.148 0.004
(0.199) (0.089)
Observations 411 411 411 411 411 411
R2 0.394 0.434 0.327 0.418 0.428 0.328
Notes: Robust SEs reported. ∗∗∗ , ∗∗ , ∗ denote significance at the 1%, 5% and 10% levels, respectively. All regressions
follow the specification in (3), which also includes the interaction between the number of inspectors at the state level
(Inspectorss ) and the distance to the labour office, Distr , the distance in levels (Distr ) and the state fixed effects. The
demographic controls are the following: share of women, high-skill individuals, urban population and log population in
1991.
and a decrease in wages relative to regions less affected. The effects are statistically significant
and economically meaningful. To illustrate the magnitude of these effects, consider moving a
region from the 10th to the 90th percentile of the distribution of RTCr , which implies a change
of a 0.1 log point in the trade shock. The results from Table 2 imply that this micro-region would
experience an increase of 4.5 (0.1 × 0.451 × 100) and 2.1 (0.1 × 0.21 × 100) percentage points
in informality and non-employment rates, respectively. To put these numbers in perspective,
they correspond to around 54% and 58% of an SD in decadal changes in informality and non-
employment rates, respectively.
The novel results in Table 2 come from the analysis of heterogeneous effects across skill levels.
As the table shows, almost all negative effects on informality, non-employment and wages come
from low-skill workers. High-skill workers show no statistically significant effects on informal-
ity and wages, and a much smaller, marginally significant effect on non-employment. Using the
same reasoning as above, moving a region from the 10th to the 90th percentile of the distribu-
tion of RTCr would imply an increase of 5.2 (0.1 × 0.520 × 100) and 2.7 (0.1 × 0.267 × 100)
percentage points in informality and non-employment rates among low-skill workers, respec-
tively. These effects account for 60% and 67% of an SD in decadal changes in informality and
non-employment rates for low-skill workers, respectively.
3.1. Effects Across Enforcement Capacity Levels

We now turn to our main analysis—the heterogeneous effects of trade liberalisation across
enforcement capacity levels (i.e., distance to the labour office), as described by regression (3).
Table 3 shows the results for employment-related outcomes: columns (1)–(3) show the results
for informality and columns (4)–(6) for non-employment. We estimate regression (3) using all
individuals and separately by skill level. The results for all workers indicate that regions with lower
enforcement capacity (greater distances to the LO) experience stronger effects on informality.

The same is not true for non-employment, as the interaction term is small in magnitude and not
significant.
Once we disaggregate the results by skill level, however, the same striking pattern observed in
the basic regression emerges. We find sizable and statistically significant heterogeneous effects for
both informality and non-employment among low-skill workers, but none for high-skill workers.
For a strong trade opening shock (RTCr = 0.1, as in the previous section), a region with very low
enforcement capacity (90th percentile of the distance distribution) would experience an increase
of 10 percentage points on informality, but nearly zero effects on non-employment. In contrast, a
region with high enforcement capacity (10th percentile of the distance distribution) experiences

a lower informality effect—a 3 percentage points increase—but much stronger negative effects
on employment, with an increase in non-employment rates of 3.9 percentage points.
The intuition for these results is as follows. In regions with stricter enforcement, firms cannot
resort to informality and therefore there is more de facto labour market rigidity, which leads
to greater employment losses. The opposite happens in regions where enforcement capacity is
weaker and de facto labour market flexibility is higher: there are strong informality effects, but
no statistically significant effects on employment. These results therefore indicate that, when
enforcement is low, informality acts as an employment buffer in regions more adversely affected
by the trade opening episode, but only for low-skill workers.
These stronger informality effects could, in principle, be a combination of two main forces.
First, the buffer effect mentioned above, in which formal jobs that would be lost are converted
to informal contracts. Second, it could also be the case that formal jobs that would be otherwise
retained in a high-enforcement region are converted to informal jobs in low-enforcement regions
(i.e., a ‘switching effect’). While the former could be seen as a potential advantage of informality,
as it helps preserving employment, the latter could be seen as detrimental, as it leads to greater
losses of formal jobs. To examine this issue, we turn to the RAIS data set, which contains the
universe of formal establishments and workers. One main advantage is that the RAIS is available
since 1988, which allows us to examine pre-trends as well. Given the nature of the data set, we
cannot run the first stage regressions at the individual level, so we run the following regressions
for each year t:
log yr,t − log yr,1991 = β0,t + β1,t RTCr + β2,t RTCr × Distr + β3,t Distr + β4,t X r
+ β5,t Distr × Inspectorsr + δs,t + εr,t . (4)
Here yr,t represents total formal employment or total number of formal establishments in region
r at time t = 1992, . . . , 2000; we use the same set of controls as in (3). For t = 1988, . . . , 1991,
we define the dependent variable as log yr,1991 − log yr,t .
For the sake of simplicity, we only plot the coefficients β̂2,t in Figure 4, while the complete
results from regression (4) are shown in Online Appendix B. The first thing to note is that we find
no evidence of pre-trends on either formal employment or the number of formal establishments
before 1991. The heterogeneous effects only become strong and statistically significant from 1994
onwards, after the unilateral trade opening process was concluded. Consistent with the conjecture
of the ‘switching effect’, Figure 4(a) shows that regions with weaker enforcement experience a
stronger reduction in formal employment relative to regions with greater enforcement capacity.
However, panel (b) shows that the greater flexibility introduced by weaker enforcement also
leads to greater survival of formal establishments: regions with weaker enforcement capacity
(greater distances to labour offices) observe smaller losses in the number of formal plants. This


(a)
(b)
Fig. 4. Effects on the Number of Formal Establishments and Workers.

Notes: Both panels show estimates of β̂2,t from (4).

Table 4. Effects on Wages by Enforcement Capacity Level.
Sample (by workers’ skill level): All Low High
(1) (2) (3)
RTCr −0.846∗∗∗
(0.281)
RTCr × Dist. LOr −0.078
(0.184)
RTCr unskilled −0.710∗∗
(0.297)
RTCr unskilled × Dist. LOr −0.169
(0.202)
−0.066

RTCr skilled
(0.453)
RTCr skilled × Dist. LOr −0.139
(0.299)
Observations 411 411 411
R2 0.612 0.562 0.596
F-statistic (joint significance) 6.271 4.876 0.373
p-value 0.00209 0.00812 0.689
Notes: Robust SEs reported. ∗∗∗ , ∗∗ , ∗ denote significance at the 1%, 5% and 10% levels,
respectively. All regressions follow the specification in (3), which also includes the inter-
action between the number of inspectors at the state level (Inspectorss ) and the distance
to the labour office, Distr , the distance in levels (Distr ) and the state fixed effects. The
demographic controls are: share of women, high-skill individuals, urban population and
log population in 1991.
result is consistent with the fact that a large fraction of informal employment is located in formal
firms, the so-called intensive margin of informality (Ulyssea, 2018). Moreover, it shows that
the intensive margin plays an important role in formal firms’ survival in the face of an adverse
economic shock.
Finally, we examine the effects on wages. As Table 4 shows, we find no statistically significant
heterogeneous effects across different enforcement capacity levels. The point estimates for both
skill levels are nevertheless large in magnitude and go in the expected direction, i.e., when
enforcement is weaker, there are greater wage losses. These point estimates are consistent with
the idea that the greater de facto flexibility provided by informality leads to stronger adjustments
in prices (wages) but lower effects on quantities (employment). However, they are not precisely
estimated and we cannot reject the hypothesis that the effects are the same across regions with
different enforcement capacity levels.18
3.2. Self-Employment
The focus of our analysis lies on informality defined as the share of informal employees. We
do so because this definition is the most consistent with our measures of enforcement capacity
and intensity. These refer to the enforcement activities conducted by the Ministry of Labour in
Brazil, which are targeted at formal firms and their employees.
Another important dimension of informality refers to self-employment. Even though the self-
employed are not directly affected by the inspections conducted by the Ministry of Labour, they
18 However, as we discuss in the robustness analysis (Subsection 3.3), in some specifications we do find significant
heterogeneous effects on low-skill workers’ wages.

Table 5. Effects on Self-Employment.
Sample (by workers’ skill level): All Low High All Low High
(1) (2) (3) (4) (5) (6)
RTCr 0.277∗∗ −0.070
(0.121) (0.137)
RTCr × Dist LOr 0.390∗∗∗
(0.080)
RTCr unskilled 0.332∗∗ −0.029
(0.133) (0.151)
RTCr skilled × Dist. LOr 0.425∗∗∗
(0.096)
0.103∗

RTCr skilled 0.143
(0.055) (0.092)
RTCr skilled × Dist. LOr −0.032
(0.052)
Observations 411 411 411 411 411 411
R2 0.675 0.726 0.185 0.698 0.744 0.187
Notes: Robust SEs reported. ∗∗∗ , ∗∗ , ∗ denote significance at the 1%, 5% and 10% levels, respectively. Regressions
follow the specification in (3), which also includes the interaction between the number of inspectors at the state level
(Inspectorss ) and the distance to the labour office, Distr , the distance in levels (Distr ) and the state fixed effects. The
demographic controls are the following: share of women, high-skill individuals, urban population and log population in
1991.
can be indirectly affected via general equilibrium effects. For example, displaced formal em-
ployees might transit to self-employment rather than unemployment. Additionally, our measure
of enforcement might be correlated with other dimensions of enforcement, such as inspections
on informal firms. Thus, in this section we investigate whether the results observed for labour
informality are also observed for self-employment.
Table 5 shows the results using the specification described in (3) and using the share of self-
employed among occupied individuals as the outcome of interest. The results show the same
patterns observed so far: low-enforcement regions show stronger increases in self-employment
relative to regions with higher enforcement capacity (small distances to the LO), and these
heterogeneous effects are entirely concentrated among low-skill workers. Considering a strong
trade opening shock (RTCr = 0.1), a region with low enforcement capacity (75th percentile of
the distance distribution) would experience an increase of 6.5 percentage points on the share of
self-employment. This corresponds to 75.7% of an SD in decadal changes in self-employment
shares. In contrast, a region with high enforcement capacity (25th percentile of the distance distri-
bution) would experience an increase of 2.6 percentage points on the share of self-employment,
or 30% of an SD in decadal changes.19 These results are consistent with the conjecture that
the different margins of enforcement move together and that self-employment is also an im-
portant adjustment margin for low-skill formal workers affected by negative labour market
shocks.20
19 Self-employment is strongly counter-cyclical in Latin American countries (e.g., Bosch and Esteban-Pretel, 2012),
which is consistent with the substantial magnitudes of our estimated effects in the face of a strong negative shock, such
as RTCr = 0.1.
20 To directly compare the results obtained for informal employees and self-employed, in Online Appendix Section
C we estimate regression (3) using as outcomes the shares of informal employees and self-employed over the working
age population. Hence, this exercise uses the same denominator for both employment categories, which allows a direct
comparison. The results remain qualitatively unchanged.

3.3. Robustness
In this section, we discuss several different exercises that assess how robust our results are.
We start by examining robustness to changes in inference. Table C.1 in Online Appendix C
shows our main results from Tables 3 and 4 with the p-values from our benchmark specification
(simple robust SEs), with clustered SEs at the meso-region level, and bootstrapped SEs. Part of
the previous literature uses clustered SEs at the level of aggregation immediately above micro-
regions, which in the Brazilian case would correspond to the meso-region (e.g., Dix-Carneiro
and Kovak, 2015; Dix-Carneiro et al., 2018). As briefly discussed in Section 2, the intra-cluster
correlation of our variable of interest, RTCr × Distr , is very close to zero, which indicates that

this clustering is not adequate in our context. We nevertheless report results with clustered SEs
for completeness and comparability with the previous literature. As Table C.1 shows, our results
remain as significant as in the benchmark specification. We also bootstrap our entire estimation
procedure, which includes the first-stage regressions at the individual level. For that, we use a
10% sample of the demographic census and five hundred replications. If anything, our results
become more significant. Interestingly, the coefficient of the interaction between RTCr and the
distance to the labour office in the wage regression for low-skill workers becomes marginally
significant (p-value of 0.108).
In Tables C.2–C.10 of Online Appendix C, we re-visit the results from Tables 3 and 4, but
including one control at the time and also expanding the set of controls used. These additional
controls are the following: informality and unemployment levels in 1980; the logarithm of local
government spending (per capita) and the Gini coefficient, both measured in 1991; and median
driving distance to the state’s capital. As discussed in Section 2, the existence of reversion to the
mean across regions with lower and higher levels of informality and non-employment could be a
potential threat to identification. The inclusion of baseline informality and unemployment rates
addresses these concerns.
We obtain the annual local government spending at the municipality level from the Ministry
of Finance (Ministério da Fazenda—Secretaria do Tesouro Nacional) and then aggregate it at
the micro-region level. The inclusion of this variable addresses the possibility that the distance
to the labour office can be a proxy for local public goods and infrastructure in general, rather
than enforcement capacity per se. The initial level of inequality might also be an important
determinant of labour market trends and local economic conditions, so we also control for that.
Finally, the driving distance to the state’s capital tackles the concern of a ‘remoteness effect’, i.e.,
the distance to the labour office could be capturing differential trends across more and less remote
regions (further away from large urban centres). These regions are likely to have different levels
of development, and can react differently to the trade shocks. In particular, more remote regions
might have a higher proportion of low-productivity firms, which are more likely to respond to
the trade shocks by increasing informal employment. If this is the case, we would confound this
‘remoteness effect’ with enforcement capacity per se.21
As the results in Tables C.2–C.7 show, the estimates of the interaction term—trade shock ×
enforcement capacity levels—generally remain stable or increase in magnitude after we include
the basic demographic controls (share of females, high-skill individuals, urban population and log
population in 1991). The one exception is the result for non-employment among low-skill workers,
21 It is worth noting that the utilisation of this variable implies the loss of sixteen micro-regions, for which it is not
available. We run the same regressions with the same sample, but without controlling for distance to the capital to verify
that our results are not driven by this sample variation.

which becomes slightly weaker and only marginally significant once we include all the additional
controls.22 Nevertheless, RTCr and the interaction term remain jointly very significant, with a
p-value of 0.005. As for wages, Tables C.8–C.10 show that including the additional regressors
increases precision and the point estimates of the interaction term. The regressions with all
additional regressors indicate that regions with weaker enforcement capacity had stronger wage
losses, and these effects were entirely concentrated on low-skill workers. This is consistent with
the stronger informality effects and lower employment losses also being concentrated among
low-skill workers. Finally, Tables C.11–C.13 in the Online Appendix show that the results for
self-employment are robust as well.

To further assess the robustness of our conclusions, we estimate a different specification, which
is to use the industry-by-micro-region cell as the unit of analysis. In this specification, we move
away from our measure of regional tariff change and directly explore the tariff cut variation
across industries and the variation in enforcement capacity across regions. The main limitation
of this specification is that one cannot analyse the effects on non-employment. More concretely,
we estimate the regression
yr,k = β0 + β1 Tariffsk × Distr + β2 X r,k + γr + νk + εr,k ,
where yr,k = yr,k,2000 − yr,k,1991 and yr,k,t is the outcome of interest in region r , industry k and
year t; X r,k is the same set of basic demographic controls used in the previous regressions, but
computed at the industry-by-micro-region level; and γr and νk denote region and industry fixed
effects, respectively. As Table C.15 in the Online Appendix shows, we confirm our previous
results on informality—strong heterogeneous effects among low-skill workers—but we also find
heterogeneous effects for high-skill workers, albeit weaker. We also find significant heterogeneous
effects on wages, especially for low-skill workers.
As an additional specification test, in Table C.16 of the Online Appendix we assess whether
the choice of enforcement capacity measure affects our results. Instead of using the maximum
distance, we use the average distance to the labour offices as the measure of enforcement. The
results remain unchanged. Since our main specifications focus on the interaction term between
the distance to the labour office (Distr ) and RTCr , we also examine the distribution of the
shock across different distances. We do so in Figure C.1 of the Online Appendix. As the figure
shows, the support of the shock varies across deciles of Distr . To examine whether this has any
implications to our results, we estimate our main specification (expression (3)), including the
triple interaction RTCr × Distr × Commom supr (and all the appropriate double interactions),
where Commom supr denotes a dummy for regions whose RTCr lies within the common support
interval defined across the different deciles of Distr (we provide more details in Online Appendix
C). As Table C.17 in the Online Appendix shows, we do not find differential effects for regions
within or outside the common support of RTCr . If anything, the point estimates indicate that the
effects become stronger in regions within the common support of the trade shock.
Finally, in Tables C.18 and C.19 of the Online Appendix we use two alternative measures of
local trade shock: (i) the ratio of imports to production and (ii) the import penetration coeffi-
cient, respectively. As the tables show, our results on informality remain strong, significant and
concentrated among low-skill workers. The effects on unemployment remain strong, but are less
precisely estimated and become marginally significant for low-skill workers (p-values of 0.101
22 Part of this loss in precision could be due to the loss of sixteen observations for which there is no information about
the driving distance to the capital.

and 0.106). The main disadvantage of these measures is that they are not skill specific, so they
introduce some measurement error.
3.4. Effects Across Enforcement Levels: Instrumental Variable Results

To obtain a more direct interpretation in terms of actual enforcement, we estimate an instrumental
variable model that has as an endogenous regressor the number of inspections per hundred firms
in the micro-region. This is the measure of enforcement intensity discussed in Section 1. The
second stage regression is analogous to the benchmark specification (expression (3)) and is given

by
ŷr = α0 + α1 RTCr + α2 RTCr × Enforcer + α3 Enforcer + α4 Zr + δs + νr , (5)
where all variables remain the same as in (3) and Enforcer denotes the number of inspections per
hundred firms.
The measure of enforcement is clearly endogenous. For example, the government might
respond to changes in local labour market conditions by increasing the resources available to
enforce the labour regulation in a given market. Conversely, the government might choose to relax
enforcement of labour regulations in the face of a negative labour demand shock. Moreover, this
variable can be subject to substantial measurement error, which also highlights the need for an
instrument. We use Distr and RTCr × Distr as instruments for Enforcer and RTCr × Enforcer .
We use as an additional instrument the distance (Distr ) interacted with the number of inspectors
at the state level. This term captures the fact that, for a given distance, states with more inspectors
will have more effective enforcement (conditional on state-specific trends captured by state
dummies). We estimate (5) using the limited information maximum likelihood estimator, as it
is known to have better small sample properties than two-stage least squares in the presence of
weak instruments.23
Table 6 shows the first set of IV estimates pooling all workers together, while Table 7 reports
the results for low- and high-skill workers separately (all first-stage results can be found in
Online Appendix Table D.1). We focus on informality and non-employment only, as the results
for wages confirm those discussed in the previous section (i.e., we find no significant effects).
Table 6 shows the expected patterns: regions with more enforcement (i.e., more inspections)
show lower informality effects and higher unemployment effects, although the former are not
statistically significant (column (2)).
Once again, the results by skill levels show a striking pattern, as all effects are concentrated
on low-skill workers (see Table 7). In regions with more enforcement there are lower informality
effects and much higher disemployment effects, but only among low-skill workers. Interestingly,
contrasting the ordinary least-squares (OLS) and IV estimates indicates that there is an attenuation
bias (both in Tables 6 and 7). That is, the OLS estimates tend to underestimate the negative effect
on informality and the positive effect on unemployment of higher levels of enforcement. This is
consistent with the presence of measurement error in the number of inspections, but also with
the government responding to this economic shock by weakening enforcement in regions that
are hit harder by the trade opening.
We report the relevant diagnostic statistics in the bottom of both tables. We report the Kleibergen
and Paap (2006) test statistics for underidentification and weak instruments. The Kleibergen-Paap
23 We also compare the LIML estimates with those obtained with a just-identified two-stage least squares and find
very similar and statistically identical results. These are available upon request.

Table 6. Effects on Informality and Non-Employment: IV Estimation.
Informality Non-employment
OLS IV OLS IV
(1) (2) (3) (4)
RTCr 0.259 0.784∗∗∗ 0.238∗∗ −0.275
(0.157) (0.254) (0.095) (0.188)
RTCr × inspections 0.014∗ −0.036 0.007 0.068
(0.008) (0.031) (0.005) (0.024)
Observations 411 411 411 411
R2 0.381 0.261 0.413 0.185

Kleibergen-Paap rk LM statistic – 12.44 – 14.02
p-value – 0.002 – 0.001
Kleibergen-Paap rk Wald F-statistic – 6.692 – 6.818
Anderson-Rubin ξ 2 test (robust inference) – 14.23 – 15.62
p-value – 0.003 – 0.001
Notes: ∗∗∗ , ∗∗ , ∗ denote significance at the 1%, 5% and 10% levels, respectively. Estimates obtained using the limited
information maximum likelihood estimator. All regressions control for state fixed effects and the following demographic
controls: share of women, high-skill individuals, urban population and log population in 1991.
LM statistic indicates that we can strongly reject the null that the model is underidentified in all
of our results. Similarly, the Kleibergen and Paap (2006) heteroskedasticity-robust Wald rk F-
statistic shows that we can reject the null that the instruments are weak for both outcomes for the
pooled and low-skill workers sample, but not for high-skill workers.24 Nevertheless, we use the
Anderson and Rubin (1949) weak-instrument robust Wald test for assessing the joint significance
of our endogenous regressors (Enforcementr and Enforcementr × RTCr ). The results indicate
that the endogenous regressors are strongly significant both for non-employment and informality,
with the exception of the results for non-employment of high-skill workers.
We provide a graphic representation of these results by plotting the effects of the regional trade
shock on informality and non-employment as a function of different enforcement levels. More
concretely, we plot the parameter
ζ̂ (Enforcement) = α̂1 + α̂2 × Enforcement,
which gives the effect of the regional trade shock for a given enforcement level.
To make the results easily interpretable, we plot ζ̂ (Enforcement) for each decile of the en-
forcement distribution, i.e., each decile of the distribution of inspections per hundred firms across
micro-regions. Since the previous results show that all the effects come from low-skill workers,
we only show the plots for this group.
Figure 5 shows a very clear pattern: in regions with low levels of enforcement, the local trade
shock produces strong informality effects but no disemployment effects; in regions with high
levels of enforcement, the trade shock does not lead to any increase in informality but has very
strong disemployment effects. In order to assess the magnitude of these heterogeneous effects,
we again consider a high intensity local trade shock (RTCr = −0.1). A region in the first decile
of enforcement—with 0.9 inspections per hundred firms—would experience an increase of 12.1
percentage points in informality but no disemployment effects. A region in 90th percentile of
enforcement—with 17.2 inspections per hundred firms—would experience no informality effects
but an increase of 10.3 percentage points in non-employment rates. Considering the results from
24 The critical values for the Kleibergen-Paap test have not yet been tabulated in the literature, but the common practice
is to use the critical values of Stock and Yogo (2005).

386
Table 7. Effects on Informality and Non-Employment by Workers’ Skill Level: IV Estimation.

Low skill High skill
Informality Non-employment Informality Non-employment
OLS IV OLS IV OLS IV OLS IV
(1) (2) (3) (4) (5) (6) (7) (8)
RTCr unskilled 0.293∗ 1.292∗∗∗ 0.324∗∗∗ −0.307
(0.161) (0.314) (0.107) (0.210)
RTCr unskilled × inspections 0.010 −0.087∗∗ 0.006 0.078∗∗∗
(0.008) (0.038) (0.005) (0.026)
RTCr skilled −0.103 −0.829 0.064 −0.332
(0.249) (0.799) (0.083) (0.304)
RTCr skilled × inspections 0.011 0.080 0.008 0.065
(0.022) (0.133) (0.009) (0.050)
Observations 411 411 411 411 411 411 411 411
R2 0.418 0.164 0.417 0.152 0.327 0.175 0.322 0.175
Kleibergen-Paap rk LM – 11.89 – 12.78 – 6.71 – 5.56
the economic journal
statistic
p-value – 0.003 – 0.002 – 0.035 – 0.062
Kleibergen-Paap rk Wald – 6.945 – 6.488 – 2.584 – 2.070
F-statistic
Anderson-Rubin ξ 2 test – 20.08 – 16.42 – 10.62 – 5.633
C

(robust inference)
p-value – 0.000 – 0.001 – 0.014 – 0.131
Notes: ∗∗∗ , ∗∗ , ∗
denote significance at the 1%, 5% and 10% levels, respectively. Estimates obtained using the limited information maximum likelihood estimator. All regressions
control for state fixed effects and the following demographic controls: share of women, high-skill individuals, urban population and log population in 1991.
[january
2022 Royal Economic Society.


d
(a)
(b)
Fig. 5. Informality and Non-Employment Effects across Enforcement Deciles—Low-Skill Workers.

Table 2, on average, the regional trade shock is associated with an increase of 5.2 and 2.7
percentage points in informality and non-employment among low-skill workers, respectively.
Therefore, the lack or strength of enforcement can lead to labour market responses in either
informality or non-employment that are substantially larger than the effects observed on average
(without accounting for heterogeneity).
These results are very much in line with the framework discussed in Section 2. In regions where
there is greater enforcement, firms are more likely to be detected by the government and therefore
the expected cost of informality is higher. Hence, firms are less likely to respond to a negative
shock by hiring a greater share of their workers informally. This inability to resort to informality

as a coping strategy implies that more firms either fire workers or exit the industry altogether,
which implies greater employment losses. The opposite would be observed in a market with low
levels of enforcement.
It is important to emphasise that our measure of enforcement is mostly relevant for the intensive
margin of informality, as the labour offices only audit formal firms. Nevertheless, the results from
Subsection 3.2 suggest that our measures of enforcement capacity and intensity are likely to be
positively correlated with other margins of enforcement (such as enforcement of tax regulation),
which would also discourage the extensive margin of informality. If this is the case then our
empirical estimates are capturing the importance of both margins of informality for firms and
workers to cope with adverse economic shocks.
4. Final Remarks
In this paper we investigate how, and to what extent, enforcement of labour regulations shapes the
labour market effects of trade. We do so in the context of Brazil, a country that underwent a major
unilateral trade liberalisation episode in early 1990s, and which is characterised by burdensome
labour regulations that are imperfectly enforced. We exploit variation across regions in both the
intensity of the trade shock and enforcement of labour regulations to assess whether different
levels of enforcement lead to heterogeneous effects on employment, informality, wages and the
number of surviving formal plants across local labour markets.
The main results of the paper show that, in the ten years after the trade opening, regions
with stricter enforcement observed: (i) substantially lower informality effects; (ii) much larger
disemployment effects; (iii) lower reductions on formal employment and (iv) greater reductions
in the number of formal plants. Regions with weaker enforcement observed symmetric effects.
All the effects are concentrated on low-skill workers, while the effects on high-skill workers
do not seem to vary across high- and low-enforcement regions. Our results thus indicate that
greater de facto labour market flexibility introduced by informality allows both formal firms and
low-skill workers to cope better with adverse labour market shocks.
We believe that these results highlight the importance of labour market rigidities introduced
by burdensome and costly regulations, and how they can amplify employment losses in the
face of negative shocks. Even though our empirical setting is a mid-income country with high
informality, our results can be relevant for developed countries as well. This is particularly true
given the rise of the ‘gig economy’ and the increasingly common co-existence between more
stable, permanent labour contracts and very flexible, largely unregulated employment relations
in more-developed economies. Even though the coexistence of these different types of contract
generates non-trivial distributional issues, our results indicate that having the option to rely on

more flexible contracts may help prevent greater job losses and firm exit in the face of a negative
economic shock.
The welfare implications of our results, however, remain as an open question. On the one
hand, our findings indicate that the de facto flexibility introduced by informality can prevent
greater employment losses and formal firm closures in the face of a negative demand shock.
On the other hand, the literature has extensively shown that, on average, informal workers
have lower earnings and worse working conditions than their (observationally equivalent) formal
counterparts. Moreover, results on wages suggest that workers in regions with weaker enforcement
might experience wage losses, which potentially introduce additional distributional concerns

(although these results are not precisely estimated). Finally, workers might attribute some intrinsic
value to formal jobs and their benefits. Hence, there are important trade-offs at the individual
level, and the net welfare effects are unclear a priori.
Finally, even if informality acts as an employment buffer in the mid-run—in which case one
could argue that it is a second-best relative to a counterfactual scenario with perfect enforcement
and no informality—it remains unclear what the long-run consequences are. Hysteresis can lead
to persistent informality effects even after the dissipation of the economic shock. If informal
jobs are indeed of lower quality, and given that informality can lead to substantial misallocation
of resources and worse aggregate outcomes (see Ulyssea, 2020 for a review), workers in low-
enforcement regions can be worse off in the long run.
São Paulo School of Economics-FGV, Brazil

University College London, UK
Additional Supporting Information may be found in the online version of this article:
Online Appendix
Replication Package
References
Adhvaryu, A., Chari, A.V. and Sharma, S. (2013). ‘Firing costs and flexibility: evidence from firms’ employment
responses to shocks in India’, Review of Economics and Statistics, vol. 95(3), pp. 725–40.
Almeida, R. and Carneiro, P. (2012). ‘Enforcement of labor regulation and informality’, American Economic Journal:
Applied Economics, vol. 4(3), pp. 64–89.
Almeida, R.K. and Poole, J.P. (2017). ‘Trade and labor reallocation with heterogeneous enforcement of labor regulations’,
Journal of Development Economics, vol. 126(May), pp. 154–66.
Anderson, T.W. and Rubin, H. (1949). ‘Estimation of the parameters of a single equation in a complete system of
stochastic equations’, The Annals of Mathematical Statistics, vol. 20(1), pp. 46–63.
Autor, D., Dorn, D. and Hanson, G. (2013). ‘The China syndrome: Local labor market effects of import competition in
the United States’, American Economic Review, vol. 103(6), pp. 2121–68.
Autor, D., Dorn, D. and Hanson, G. (2018). ‘When work disappears: Manufacturing decline and the falling marriage-
market value of young men’, American Economic Review: Insights, vol. 1(2), pp. 161–78.
Barros, R.P.D. and Corseuil, C.H. (2004). ‘The impact of regulations on Brazilian labor market performance’, in (J.
Heckman and C. Pages, eds.), Law and Employment: Lessons from Latin America and the Caribbean, pp. 273–350,
Chicago: University of Chicago Press.
Besley, T. and Burgess, R. (2004). ‘Can labor regulation hinder economic performance? Evidence from India’, The
Quarterly Journal of Economics, vol. 119(1), pp. 91–134.
Boeri, T. and Garibaldi, P. (2005). ‘Shadow sorting’, in (C. Pissarides and J. Frenkel, eds.), NBER Macroeconomics
Annual, pp. 125–63, Cambridge, MA: MIT Press.
Bosch, M. and Esteban-Pretel, J. (2012). ‘Job creation and job destruction in the presence of informal markets’, Journal
of Development Economics, vol. 98(2), pp. 270–86.

Bosch, M., Goñi-Pacchioni, E. and Maloney, W. (2012). ‘Trade liberalization, labor reforms and formal–informal
employment dynamics’, Labour Economics, vol. 19(5), pp. 653–67.
Botero, J.C., Djankov, S., Porta, R.L., Lopez-de Silanes, F. and Shleifer, A. (2004). ‘The regulation of labor’, The
Quarterly Journal of Economics, vol. 119(4), pp. 1339–82.
Costa, F., Garred, J. and Pessoa, J.P. (2016). ‘Winners and losers from a commodities-for-manufactures trade boom’,
Journal of International Economics, vol. 102(September), pp. 50–69.
de Paula, A. and Scheinkman, J.A. (2010). ‘Value-added taxes, chain effects, and informality’, American Economic
Journal: Macroeconomics, vol. 2(4), pp. 195–221.
Dix-Carneiro, R. and Kovak, B. (2015). ‘Trade liberalization and the skill premium: A local labor markets approach’,
American Economic Review, vol. 105(5), pp. 551–7.
Dix-Carneiro, R. and Kovak, B. (2017). ‘Trade liberalization and regional dynamics’, American Economic Review, vol.
107(10), pp. 2908–46.

Dix-Carneiro, R. and Kovak, B. (2019). ‘Margins of labor market adjustment to trade’, Journal of International Economics,
vol. 117(March), pp. 125–42.
Dix-Carneiro, R., Soares, R. and Ulyssea, G. (2018). ‘Economic shocks and crime: Evidence from the Brazilian trade
liberalization’, American Economic Journal: Applied Economics, vol. 10(4), pp. 158–95.
Donovan, K., Lu, W. and Schoellman, T. (2021). ‘Labor market dynamics and development’, Mimeo, Manuscript.
Freeman, R.B. (2010). ‘Labor regulations, unions, and social protection in developing countries: Market distortions or
efficient institutions?’, in (D.R.M. Rosenzweig, ed.), Handbook of Development Economics, vol. 5, pp. 4657–702,
Amsterdam: North-Holland.
Goldberg, P.K. and Pavcnik, N. (2003). ‘The response of the informal sector to trade liberalization’, Journal of Develop-
ment Economics, vol. 72(2), pp. 463–96.
Goldberg, P.K. and Pavcnik, N. (2007). ‘Distributional effects of globalization in developing countries’, Journal of
Economic Literature, vol. 45(1), pp. 39–82.
Hakobyan, S. and McLaren, J. (2016). ‘Looking for local labor market effects of NAFTA’, Review of Economics and
Statistics, vol. 98(4), pp. 728–41.
Kleibergen, F. and Paap, R. (2006). ‘Generalized reduced rank tests using the singular value decomposition’, Journal of
Econometrics, vol. 133(1), pp. 97–126.
Kovak, B. (2013). ‘Regional effects of trade reform: What is the correct measure of liberalization?’, The American
Economic Review, vol. 103(5), pp. 1960–76.
Kume, H., Piani, G. and Souza, C. (2003). ‘A politica brasileira de importação no periodo 1987–1998: Descrição e
avaliação’, in (C.H. Corseuil and H. Kume, eds.), A Abertura Comercial Brasileira nos Anos 1990: Impacto Sobre
Emprego e Salário, pp. 9–37, Brasilia: IPEA.
Leal Ordonez, J.C. (2014). ‘Tax collection, the informal sector, and productivity’, Review of Economic Dynamics, vol.
17(2), pp. 262–86.
Meghir, C., Narita, R. and Robin, J.M. (2015). ‘Wages and informality in developing countries’, American Economic
Review, vol. 105(4), pp. 1509–46.
Menezes-Filho, N.A. and Muendler, M.A. (2011). ‘Labor reallocation in response to trade reform’, Working Paper
17372, National Bureau of Economic Research.
Nickell, S. and Layard, R. (1999). ‘Labor market institutions and economic performance’, in (O.C. Ashenfelter and D.
Card, eds.), Handbook of Labor Economics, pp. 3029–84, Amsterdam: North-Holland.
Nickell, S., Nunziata, L. and Ochel, W. (2004). ‘Unemployment in the OECD since the 1960s. What do we know?’,
ECONOMIC JOURNAL, vol. 115(500), pp. 1–27.
Stock, J.H. and Yogo, M. (2005). ‘Testing for weak instruments in linear IV regression’, in (D.W.K. Andrews and J.H.
Stock, eds.), Identication and Inference for Econometric Models: Essays in Honor of Thomas Rothenberg, pp.
80–108, Cambridge: Cambridge University Press.
Tella, R.D. and MacCulloch, R. (2005). ‘The consequences of labor market flexibility: Panel evidence based on survey
data’, European Economic Review, vol. 49(5), pp. 1225–59.
The World Bank. (2007). Doing Business 2007: How to Reform, Washington, DC: The World Bank.
Topalova, P. (2010). ‘Factor immobility and regional impacts of trade liberalization: Evidence on poverty from India’,
American Economic Journal: Applied Economics, vol. 2(4), pp. 1–41.
Ulyssea, G. (2010). ‘Regulation of entry, labor market institutions and the informal sector’, Journal of Development
Economics, vol. 91(January), pp. 87–99.
Ulyssea, G. (2018). ‘Firms, informality, and development: Theory and evidence from Brazil’, American Economic
Review, vol. 108(8), pp. 2015–47.
Ulyssea, G. (2020). ‘Informality: Causes and consequences for development’, Annual Review of Economics, vol. 12, pp.
525–46.


Jurnal 3 MSDM Global

Uploaded by

Copyright:

Available Formats

Jurnal 3 MSDM Global

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Jurnal 3 MSDM Global

Uploaded by

Copyright:

Available Formats

The Economic Journal, 132 (January), 361–390 https://doi.org/10.1093/ej/ueab052 C The Author(s) 2021.

Published by Oxford University Press on behalf of

ENFORCEME N T O F L A B O U R RE GUL ATI ON AND T HE

Vladimir Ponczek and Gabriel Ulyssea

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

1. Background and Data

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Paper, publishing, printing

Electric, electronic equip.

Petroleum, gas, coal

Fig. 1. Changes in log(1 + tariff), 1990–1995.

to construct the ‘regional tariff change’ (RTC) as proposed by Kovak (2013):

1.2. Labour Regulations and Enforcement

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

2.1. Theoretical Framework

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Fig. 3. Trade Opening under Low and High Enforcement.

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

2.2. Empirical Specification

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

2.2.1. Identification: discussion

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

3. Enforcement of Labour Regulations and the Labour Market Effects of Trade

Table 2. Basic Effects on Informality, Non-Employment and Wages.

R2 0.375 0.409 0.315 0.395 0.402 0.319 0.608 0.558 0.588

2022 Royal Economic Society.

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

3.1. Effects Across Enforcement Capacity Levels

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Fig. 4. Effects on the Number of Formal Establishments and Workers.

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

yr,k = β0 + β1 Tariffsk × Distr + β2 X r,k + γr + νk + εr,k ,

3.4. Effects Across Enforcement Levels: Instrumental Variable Results

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Table 7. Effects on Informality and Non-Employment by Workers’ Skill Level: IV Estimation.

2022 Royal Economic Society.

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Fig. 5. Informality and Non-Employment Effects across Enforcement Deciles—Low-Skill Workers.

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

São Paulo School of Economics-FGV, Brazil

Downloaded from https://academic.oup.com/ej/article/132/641/361/6310581 by guest on 18 March 2024

You might also like

yr,k = β0 + β1 Tariffsk × Distr + β2 X r,k + γr + νk + εr,k ,