WPS8363
Policy Research Working Paper 8363
Size-Dependent Tax Enforcement and Compliance
Global Evidence and Aggregate Implications
Pierre Bachas
Roberto N. Fattal Jaef
Anders Jensen
Development Research Group
Macroeconomics and Growth Team
March 2018
Policy Research Working Paper 8363
Abstract
This paper studies the prevalence and consequences of and compliance increase with size. Size-dependence is more
size-dependent tax enforcement and compliance. The iden- prevalent in low-income countries, and concentrated at the
tification strategy uses the ranking of industries’ average firm top of the size distribution. When quantified in a general
size in the United States as an instrument for the size rank- equilibrium model, removing size dependent enforcement
ing of the same industries in developing countries. Data on leads to gains in Total Factor Productivity of up to 0.8 percent.
125,000 firms in 140 countries show that tax enforcement
This paper is a product of the Macroeconomics and Growth Team, Development Research Group. It is part of a larger
effort by the World Bank to provide open access to its research and make a contribution to development policy discussions
around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors
may be contacted at pbachas@worldbank.org and rfattaljaef@worldbank.org
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.
Produced by the Research Support Team
Size-Dependent Tax Enforcement and Compliance:
Global Evidence and Aggregate Implications
Pierre Bachas, Roberto N. Fattal Jaef, and Anders Jensen∗
JEL codes: H25, H26, O23, O43, D61
∗ Pierre Bachas: World Bank Research, pbachas@worldbank.org - Roberto Fattal Jaef: World
Bank Research, rfattaljaef@worldbank.org - Anders Jensen: Harvard Kennedy School and NBER, an-
ders_jensen@hks.harvard.edu We are grateful to Alan Auerbach, Tim Besley, Michael Best, Natalie Cox,
Tom Cunningham, Simon Galle, Roger Gordon, Michael Keen, Henrik Kleven, Camille Landais, Florian
Misch, Ted Miguel, Torsten Persson, Andres Rodriguez-Clare, Emmanuel Saez, Johannes Spinnewijn, Mu-
nir Squires, Owen Zidar and seminar participants at the LSE, the 69th IIPF Congress and the 2014 ZEW
Public Finance conference for valuable comments. Bachas gratefully acknowledges ﬁnancial support from
the Center for Equitable Growth and the Julis-Rabinowitz Center for Public Policy and Finance. Jensen
gratefully acknowledges ﬁnancial support from the Peter G. Foundation Grant No.16017 and the ESRC.
1 Introduction
An inﬂuential literature explains cross-country differences in income and productivity
through the misallocation of resources across ﬁrms (Restuccia and Rogerson 2008, Hsieh
and Klenow 2009). One salient property of measured misallocation is its productivity
dependence: the most productive ﬁrms are smaller and the least productive ﬁrms are
larger than the output-maximizing allocation. This property of misallocation can emerge
through government policies that target ﬁrm size; for example size-dependent labor reg-
ulations, tax rates or accounting requirements can discourage ﬁrm growth and impact the
ﬁrm size distribution (Gollin 1995, Guner et al. 2008, Bento and Restuccia 2017). In this
paper, we perform a quantitative evaluation of a speciﬁc feature of taxation which exerts a
heterogeneous distortionary effect: size-dependent tax enforcement and compliance. We
document the pervasiveness of this phenomenon around the world and provide estimates
of the size-gradients across development levels and across the ﬁrm size distribution. We
then characterize the implications of these size gradients for Total Factor Productivity in
the context of a general equilibrium model of ﬁrm heterogeneity.
Exploiting arguably exogenous variation in an industry’s optimal labor scale, we ﬁnd
a robust positive slope of industry average ﬁrm-size, as measured by the number of em-
ployees, on tax inspection probability and compliance. This slope captures the average
effect and masks non-linearities: inspection and compliance increase at the top of the size
distribution and appear similarly lax among small and medium industries. Moreover,
the magnitude of the gradient increases with development, from zero for rich countries
to its highest value in the poorest economies. When feeding our estimated size-gradients
into a general equilibrium model of ﬁrm dynamics, we ﬁnd improvements in TFP of up
to 0.8% when removing the size dependence component of taxation. These gains accrue
both from the reversal of the misallocation as well as the subsequent increase in the ﬁrms’
expenses on innovation and growth.
Tax enforcement is relevant for the study of policy-driven distortions for at least three
1
reasons. First, enforcement reduces the scope for tax evasion and therefore directly im-
pacts ﬁrms‘ effective tax rates. Second, many ﬁrms report facing large costs of dealing
with the tax administration (World Bank 2017).1 Third, reliance on size-dependent poli-
cies for tax enforcement has increased over time, as international institutions encouraged
tax administrations to segment taxpayers (Kanbur and Keen 2014). To illustrate this trend,
Figure 1 shows that over the past 20 years, more than 70 countries adopted special en-
forcement units for large taxpayers. While large taxpayer units hint at stringent enforce-
ment at the top of the ﬁrm-size distribution, countries have also adopted enforcement
policies targeted at small and medium ﬁrms.2 Therefore how tax inspection and compli-
ance vary with ﬁrm size and in countries around the world, is an empirical question.
The empirical analysis uses the comprehensive World Bank Enterprise Surveys (WBES),
which contain ﬁrm-level data on self-reported tax inspection and tax compliance3 for a
sample of 125,000 ﬁrms in 140 countries. Identiﬁcation is based on the idea that ﬁrm size
is partially determined by technological factors (Lucas 1969; Kremer 1993). These tech-
nological factors pin down ﬁrms’ optimal scale of operation (Bain 1954, Burnside 1996,
Kumar et al. 1999, Basu and Fernald 2016). If ﬁrms in an industry share a common opti-
mal scale across countries, then the relative ranking of two industries’ scale in a plausibly
undistorted market such as the US can serve as an instrument for the relative ranking of
the same two industries’ optimal scale in a developing country’s distorted market.4 We
measure scale as the number of employees and take the average at three digit ISIC sectors.
Identiﬁcation relies on predicting the size ranking of industries at the 3-digit ISIC level in
1 The World Bank 2017 Doing Business publication reports that "on average taxpayers spend 25 hours com-
plying with the requirements of an auditor, and go through several rounds of interactions during 10.6 weeks."
2 Many countries implemented small and medium taxpayer ofﬁces alongside the large taxpayer ofﬁce,
which tailored audit algorithms to the evasion risks speciﬁc to smaller ﬁrms.
3 The question on tax compliance is asked at the industry level. Since all our speciﬁcations are at the
industry-country level, we do not require that the reported answers represent the ﬁrm’s own tax compli-
ance. Instead we only require that the answers accurately represent the industry’s average compliance.
4 This idea follows Rajan and Zingales 1998, who studied if industries reliant on external ﬁnance grew
faster in countries with more developed ﬁnancial institutions. They instrument an industry’s reliance on
external ﬁnance with the external ﬁnance usage of the same sector in the US.
2
a WBES country from the size ranking in the US of the same ISIC3 industries.5
The ﬁrst stage relation, that is the conditional expectation of a WBES country’s indus-
try size rank across ranks of US industry size, is positive and linear. When repeated across
subsamples of countries at different income levels, the slope remains constant: the US dis-
tribution has the same power to predict industry size ranking in Ethiopia, Indonesia or
Brazil.6 This suggests that the identiﬁcation strategy captures technological differences in
labor demand, which vary across industries but not across countries.
The US Census industry size distribution is a valid source of exogenous variation un-
der the assumption that US ﬁrms determine their size orthogonally to tax enforcement.
This assumption might be reasonable in the US, where the Internal Revenue Service has
access to comprehensive sources of third-party information and might not need to rely
on imprecise size-proxies for tax enforcement. In contrast, tax authorities in developing
countries with low ﬁscal capacity may be constrained to use size-proxies. We take the fol-
lowing steps to alleviate concerns with the identifying assumption. First, the estimates are
robust when using the industry ranking based on European ﬁrms from Amadeus data,7
instead of the US Census. These ﬁrms are subject to comprehensive ﬁnancial reporting,
and hence size-proxies might no longer be determinants of tax inspection. Second, since
it has been documented that tax evasion is important among the self-employed in the
U.S. (Blumenthal et al. 2001), we only consider ﬁrms with more than ﬁve employees in
the US Census.8 Third, for the few countries in the WBES with similar GDP as the US and
thus similar tax enforcement capacity, we ﬁnd no size gradient in tax inspection.9 Finally
we support the exclusion restriction by showing that the predicted WBES size-ranking is
5 Since our identiﬁcation relies on industry comparisons of ﬁrm size, we do not use the within-industry
size variation. In effect, this reduces our sample to 12,152 ISIC3-country-year observations.
6 This supports the following type of statement: “If the average car manufacturer requires more work-
ers than the average retail ﬁrm in the US, then this ranking of industries by size also holds in Ethiopia,
Indonesia, and Brazil.”
7 We use Amadeus data for German and British ﬁrms. Germany data derive from ﬁnancial statements
ﬁled with the business registry. British data come from audited annual reports presented to shareholders.
8 The selection of ﬁrms with ﬁve or more employees also matches the sampling strategy of the WBES.
9 We cannot test this directly on the US since it does not have a World Bank Enterprise Survey.
3
not just proxying for competing channels that might drive tax inspection policies, namely
capital intensity and reliance on external ﬁnance (Gordon and Li 2009).
The IV estimated size-gradients imply that a 10 percentile increase in the WBES size-
rank increases a ﬁrm’s probability of tax inspection by 2.3% (a 3.8% increase relative to a
mean of 61%) and its tax compliance rate by 2.2% (a 2.7% increase relative to a mean of
81%). We estimate two main control models which allow non-parametrically for the tax
inspection function to vary over ﬁrm characteristics or over two digit ISIC industries in
every country-year. The latter speciﬁcation exploits size variation between narrow ISIC3
industries such as “Manufacture of rubber products” (category 251) compared to “Man-
ufacture of plastics products” (category 252), within the ISIC2 category 25. In addition,
we ﬁnd comparable coefﬁcients in panel models which exploit within a country variation
in the ranking of 3-digit ISIC industry over time. The size gradients mask non-linearities
over the size distribution. Inspection and compliance appear concentrated among large
ﬁrms, but no different for small and medium ﬁrms. The symmetry of results on inspec-
tion and compliance suggests that size-based tax inspection may explain part of the com-
pliance behavior. Finally, we study heterogeneity across countries’ income levels: the
size-gradient appears to fall with development and the size-gradient for countries with
similar income level as the United States is zero and statistically different from that of
low-income countries. A decreasing reliance on size-dependent policies with develop-
ment is consistent with evidence showing that countries with weak ﬁscal capacity rely on
production-inefﬁcient tax instruments (Best et al. 2015, Bachas and Soto 2016).
In order to quantitatively evaluate the macroeconomic implications of the estimated
size-gradients, we appeal to a standard general equilibrium model of ﬁrm dynamics. Our
closest reference in the literature is a closed economy version of Atkeson and Burstein
[2010]. It features three channels through which size dependent effective taxation can
affect TFP: resource misallocation among incumbents, entry and exit of ﬁrms, and in-
4
centives to invest in innovation.10 Our strategy to calibrate a productivity-dependent
enforcement proﬁle in the model is consistent with the identiﬁcation strategy in the em-
pirical analysis. Taking the estimates for the average size-gradient from the IV regressions
for each income group, we use the ﬁrm size distribution in the US, and the models’ im-
plication that size maps one to one with productivity in the undistorted equilibrium, to
back-out a gradient between the probability of tax enforcement and the underlying pro-
ductivity of the ﬁrms. We then evaluate the TFP gains from reversing the size-dependence
in taxation. Our baseline exercise ﬁxes the probability of compliance at the level corre-
sponding to the median size and applies it to all ﬁrms in the economy while keeping the
statutory revenue and proﬁt tax rates unchanged. As an alternative, we consider a case
where, in addition to ﬁxing the probability of taxation across ﬁrms, we readjust the tax
rates so as to preserve the overall share of tax revenue to GDP, and the share of proﬁt-tax
in total revenue.
Our baseline counterfactual yields a TFP gain of 0.8% for the least developed group
of countries, where the size-gradient is the highest, and is neutral for the richest group,
where the compliance proﬁle is ﬂat. At the micro-level, the model yields predictions
that are consistent with the evidence on cross-country differences in the average size and
life-cycle growth of ﬁrms (Bento and Restuccia 2017, Hsieh and Klenow 2014). Average
ﬁrm size increases by up to 30%, and the aggregate innovation intensity in the economy
expands by more than 10% . The magnitude of the aggregate gains are weakened in the
counterfactual with constant tax collection, reaching 0.3% for the lowest income group.
The reason for the decline is that revenue and proﬁt tax rates are increased in order to
raise tax revenue to maintain its share in GDP. Since proﬁt taxes discourage entry and
innovation, even in the absence of size-dependence, the magnitude of the TFP gain is
mitigated.
10 By investment in innovation we do not constraint to thinking just about R&D that leads to frontier
changing innovations or new patents, but rather we think of a broader concept of intangible capital accu-
mulation that may indeed constitute path-breaking innovations but could also refer to adoption of frontier
technologies or implementation of better management practices.
5
Quantitatively, we ﬁnd lower distortions than those generated by ﬁnancial frictions
but comparable in size to those found when evaluating labor market policies. In the
context of the latter, for instance, Hopenhayn and Rogerson [1993] ﬁnd TFP losses of 1
to 2% from taxes to ﬁring and hiring workers. Gourio and Roys [2014] and Garicano
et al. [2016] ﬁnd almost zero gains from reversing a legislation that reduces the taxation
of labor for small ﬁrms. Financial frictions are the most costly distortion, with potential
gains from improving credit markets ranging from 5 to 40% depending on the margins
of adjustment allowed for in the models (Midrigan and Xu 2014, Buera et al. 2011, Moll
2014).
1.1 Related Literature
Our paper participates to two distinct literatures. First, an inﬂuential literature analyzes
cross-country income differences through the misallocation of factor inputs across ﬁrms
and sectors. Since the seminal work of Hsieh and Klenow [2009] and Bartelsman et al.
[2013], many studies applied the same methodologies to characterize the full extent of
misallocation around the world. Furthermore, Restuccia and Rogerson [2008] and Guner
et al. [2008] helped gain awareness of the importance of the size-dependent component
of the idiosyncratic distortions, showing that this is an important feature of the under-
lying policies generating the misallocation that magniﬁes the TFP losses associated with
it. The pervasiveness of misallocation motivated the emergence of a number of stud-
ies investigating the allocative properties of speciﬁc policies or distortions. Among the
most notable ones in this group are Hopenhayn and Rogerson [1993], Gourio and Roys
[2014], and Garicano et al. [2016], evaluating policies related to labor-market regulation,
and Buera et al. [2011], Midrigan and Xu [2014] and Moll [2014] focusing on credit market
distortions. Our work contributes to the misallocation literature from the same angle as
the latter set of studies. We provide identiﬁed estimates of a particular distortionary pol-
icy, size-dependence in tax enforcement and compliance, and quantify the implications of
6
this policy for aggregate TFP and ﬁrm behavior.
Second, our exercise is related to the literature on tax enforcement and third-party in-
formation (Kopczuk and Slemrod 2006, Gordon and Li 2009, de Paula and Scheinkman
2010, Pomeranz 2015, Naritomi 2016). Almunia and Lopez-Rodriguez [2017] show that
Spanish ﬁrm bunch at the size-threshold of the large taxpayer unit, while Zareh and Peichl
[2016] ﬁnd that Armenian ﬁrms bunch at the full account reporting threshold. Our results
provide empirical support to theories where ﬁrm size is correlated with tax compliance
(Kleven et al. 2016, Bigio and Zilberman 2011). In Kleven et al. [2016], tax avoidance
strategies such as double book-keeping and collusion with employees to hide operations
are impossible to sustain for large ﬁrms, since a single whistle blower can reveal the en-
tire operation. Therefore, large ﬁrms disclose third-party information and comply with
their tax obligations. In this model the government enforcement strategy is ﬁxed, and
increased tax compliance with development is driven by ﬁrms’ size growth. However,
our results are also consistent with models where increased tax inspection with size is an
optimal government policy, given ﬁscal capacity constraints (Bigio and Zilberman 2011,
Ito and Sallee 2016).
The paper is structured as follows. Section 2 discusses the data. Section 3 presents
the identiﬁcation strategy and empirical speciﬁcations. Section 4 shows the empirical
results and their robustness. Section 5 presents the general equilibrium model, which
is calibrated in Section 6 to quantify distortions from size-dependent tax enforcement.
Section 7 concludes.
2 Data
To provide global evidence on the relation between industry average ﬁrm size, tax inspec-
tion and tax compliance, we use the comprehensive ﬁrm-level data from the World Bank
Enterprise Surveys (WBES). The surveys cover 125,000 ﬁrms in 140 countries between
7
2003 and 2015. A subset of countries have multiple surveys over time, with an aver-
age of 1.9 surveys per country. The World Bank outsources data collection to third-party
agencies in order to remove the ofﬁcial afﬁliation of the surveyors and not contaminate
responses. The survey agencies draw upon the list of registered establishments provided
by the national statistics ofﬁce.11 The random stratiﬁed sampling is done at the industry-
level, corresponding to the 2-digit ISIC level, and over-samples from large ﬁrms to cap-
ture a large share of economic activity. Given the industry stratiﬁcation, over-sampling
of large ﬁrms does not impact the relative size of ISIC-2 industries. However it could im-
pact the relative size of ISIC3 industries, which we use in some speciﬁcations, and where
we have to assume that within ISIC3, ﬁrm sizes follow similar distribution shapes such
that over-sampling at the top does not impact relative rankings. Firms with fewer than
ﬁve employees, government-owned establishments and co-operatives are dropped from
the sampling frame. The surveyors contact ﬁrms from this stratiﬁed-random sample and
conduct the surveys with the person who most often deals with banks or government
agencies.
We measure size as the average of the log of number of employees per ﬁrm in a three
digit ISIC sector,12 excluding part-time and temporary workers. We calculate size for all
ISIC3-country-year cells in the WBES. The size distribution in the US is drawn from the
2002 Census of Employment and Wages. To be consistent with sampling in the WBES,
we exclude ﬁrms with ﬁve employees or less from the Census.13 One caveat to is that
we solely use industry level variation in average ﬁrm size. In the data, the intraclass
correlation between log employee size and industries is 15%, and hence a large share of
ﬁrm size variation arises within industries.
We construct extensive and intensive margin measures of tax inspection within an
11 Sometimes supplemented with the list of ﬁrms registered with the chamber of commerce.
12 Sectors follow version 3.1 of the ISIC international industry classiﬁcation.
13 In robustness tests we use an alternative size measure which places additional weight on the ﬁrms
which have a higher proportion of the total sectoral employment as suggested by Davis and Henrekson
[1999] and Kumar et al. [1999].
8
ISIC3-country-year cell as, respectively, the share of ﬁrms which report an inspection by
tax ofﬁcials in the past 12 months, and the number of inspections over that period. To
study tax compliance, we use the answer to the question:
“Recognizing the difﬁculties many enterprises face in fully complying with taxes
and regulations, what percentage of sales would you estimate the typical establishment
in your area of activity reports for tax purposes?”.
Full compliance is deﬁned as the share of ﬁrms that report all of their sales. The reference
to a “typical establishment in your area of activity” is meant to encourage ﬁrms to truthfully
report – either a reference group’s behavior or the ﬁrm’s own behavior. While we cannot
infer whose behavior the ﬁrm is precisely referring to, we only require that the ﬁrm’s
reported compliance rate corresponds to its own ISIC 3 industry, a plausible assumption,
and weaker than assuming that the answer corresponds to the ﬁrm’s own compliance
rate. We also construct the effective tax rate, deﬁned as the product of the compliance
rate and the statutory tax rate, where the tax rate is the sum of the corporate income tax
rate and the general sales tax rate (or VAT).14
In robustness tests we use informal payments, deﬁned with the question: “It is said that
establishments like this one are sometimes required to make gifts or informal payments to public
ofﬁcials to “get things done” with regard to customs, taxes, licenses, regulations. On average,
what percent of total annual sales do establishments like this one pay in informal payments or gifts
to public ofﬁcials for this purpose?”. This question provides a direct measure of the informal
tax rate, since the informal payments are expressed as a percentage of sales.
Since we study outcomes at the industry level, we report summary statistics at the
country-year-industry level, where the industry corresponds to the 3-digit ISIC. The 272
country-year surveys cover 140 countries, with on average 50 ISIC3 industries repre-
sented and a median at 52. The average industry surveys 10 ﬁrms. Table 1 displays the
14 We collected the statutory sales and corporate tax rates in the relevant year of the WBES country-year
cell using the KPMG worldwide tax summaries.
9
number of observations, mean, standard deviation, and quartiles for each of the variables
described above. The average number of tax inspections is just under 2 and 62% of ﬁrms
receive at least one tax inspection visit in the year. The average compliance rate with
taxes is 81%, and 57% of ﬁrms report full compliance. The average informal payment
corresponds to 1.6% of ﬁrms’ sales and 26% of ﬁrms make such informal payments. Since
the survey over-samples from larger ﬁrms it is not surprising to see that 22% of ﬁrms are
exporters. Since the sampling frame is deﬁned at the two digit ISIC level, manufacturing
ﬁrms, which occupy a disproportionate share of the ISIC 2 categories, represent 58% of
the sample of 3-digit industries. Finally, the average ISIC3 industry has 69 workers in the
WBES, while the average industry has 53 workers in the 2001 US Census.
Two points worth highlighting concern sample size and survey weights. First, while
the core tax inspection variable is always available, data on tax compliance are only avail-
able for earlier surveys.15 Second, when possible we apply survey weights. However
weights are missing from some early surveys. The core tax inspection results are drawn
from the sample with survey weights, but to preserve sample size, we report results on
tax compliance without dropping observations with missing weights. We show in the
appendix that results remain unchanged in the smaller sample with full survey weights.
3 Identiﬁcation and Econometric Speciﬁcations
3.1 Identiﬁcation and First-Stage
Our objective is to estimate how ﬁrm size impacts tax inspection:
Tax inspectionict = α + β · Sizeict + γct + εict (1)
15 The
question on formal tax compliance was dropped from the harmonized survey after 2007. In sur-
veys after 2007 administered in Angola, Botswana, Congo (Dem. Rep), Ethiopia, Iraq, Mali and Rwanda,
we extracted the tax compliance question from the non-harmonized raw data.
10
The OLS regression of ﬁrm size on tax inspection is likely to suffer from reverse causality
and omitted variable bias. In particular, ﬁrms might reduce their size to prevent facing
more stringent tax inspection. To address this issue we turn to an instrumental variable
strategy. A valid instrument predicts ﬁrm size and only impacts tax inspection through
its effect on ﬁrm size. A vast literature ﬁnds that industries vary in their optimal scales
and that there exists a structural technological demand for labor at the industry level.
Our identiﬁcation strategy follows the intuition of Rajan and Zingales [1998]: if we con-
sider the US as an undistorted market, then US ﬁrms achieve their optimal unconstrained
size, which depends on the structural scale parameter of their industry and idiosyncratic
shocks. This suggests using the average size of ﬁrms in industries in the US as an instru-
ment for the average size of ﬁrms in the same industry in lower-income countries.
The ﬁrst stage is estimated with the following regression:
Rank sizeict = α0 + α1 · Rank sizei,U S 01 + γct + εict (2)
Where Rank sizeict is the average ﬁrm-size rank of ISIC3 industry i, in country c, at time
t, and Rank sizei,U S ,01 is the rank by ﬁrm-size, of industry i, in the US census in 2001.
γct are country-year ﬁxed effects. The US industry ranking is drawn relative to the set
of industries present in a given country-year, and we weight the regression results by
the number of observations in a given ISIC3-country-year cell.16 The slope-coefﬁcient
α1 measures the increase in the size-ranking of an industry in its country, when moving
along the ranking of industry size in the US. Table A1 shows that the results are robust
to using average number of workers per industry rather than industry ranking based on
number of workers. We implement the ﬁrst stage with a rank-rank speciﬁcation for two
reasons. First, by using an industry‘s ranking in terms of average workers, the coefﬁcient
β only depends on the joint distribution of average size in the WBES country and aver-
16 This allows for ISIC3-country-year cells measured with greater precision to carry more weight in the
estimates. Omitting weights does not change qualitatively the results, which remain signiﬁcant.
11
age size in the US. Unlike the log-log speciﬁcation, it does not depend on the marginal
distributions of WBES and US industry-size. In other words, in a rank-rank speciﬁcation
β is not impacted by the ratio of US to WBES industry-size variances. The rank-rank
speciﬁcation is thus more stable across subsamples of different development levels with
widely varying WBES size-variances.17 Second, the rank-rank speciﬁcation appears more
able to untangle technological size-differences from non-technological differences. Non-
technological drivers such as availability of labor-saving instruments, labor regulations,
and legal quality may differ substantially across countries and impact relative ﬁrm size,
thus a wedge between WBES level-differences in size and US level-differences in size.
Such non-technological drivers of size will not impact β so long as they do not overturn
the ranking of industries in WBES relative to the US.
Figure 2 displays non-parametrically the ﬁrst stage relation between industries’ size
ranks in the WBES countries and the US Census. ISIC3 industries are ranked relative
to other industries in the same country-year survey. The WBES and US Census ranks are
then grouped into 50 equal sized (two percentile) bins. Figure 2 plots the mean WBES size
rank percentile within each 50-quantile US size rank percentile and the best-ﬁt line. We
ﬁnd a steep positive slope and a linear relation between industries’ size-ranks in the US
Census and their size-ranks in WBES countries. In Figure 3 we show that the rank-rank
coefﬁcients are constant by countries’ income levels: the US industry-size distribution has
the same power to predict average ﬁrm size over the full size-distribution, in for exam-
ple, Ethiopia, Indonesia and Mexico. In robustness analysis in Section 4.4, we show that
predictive power does not hinge on using the 2001 Census, and remains almost identical
when using either an earlier wave of the Census (1991) or a later wave (2015). Finally, the
ﬁrst stage remains strong and positive when we restrict our analysis to the manufacturing
sector, a common restriction for studies using cross-industry variation.
The WBES country size-ranking predicted from the US ranking of industries is an
17 This is also the main reason why a rank-rank speciﬁcation is chosen over the log-log speciﬁcation in
recent studies of income mobility (Chetty et al. 2014).
12
exogenous source of size-variation under the assumption that the size of US ﬁrms is or-
thogonal to tax enforcement. Arguably, for large ﬁrms the IRS bases its decision directly
on third-party reports and risk scores from economic activity and not indirectly based on
size. In countries with high ﬁscal capacity, it is well-documented that non-compliance is
concentrated among the self-employed and family ﬁrms with few employees (Blumen-
thal et al. 2001, Kleven et al. 2011). To alleviate concerns, we remove all ﬁrms with fewer
than 5 employees in the US Census, which also matches the WBES sampling design. In
Section 4.4, we show that the results hold when constructing the exogenous size-ranking
with British and German ﬁrms in the Amadeus dataset, which are subject to stringent
information reporting requirements.18 Finally, in section 4.3 we show that for the richest
countries in the WBES (countries with per capita GDP above $21,000)19 the estimated co-
efﬁcients of size on tax inspection is zero. This evidence supports the ﬁrst stage identify-
ing assumption that in high ﬁscal capacity countries, ﬁrm size is not driven by size-based
inspection.
The IV strategy provides a causal estimate of the size-gradient under the ﬁrst stage
validity, discussed above, and the exclusion restriction. The exclusion restriction assumes
that the US size-ranking of WBES industries only impacts tax inspection in the WBES
country through the WBES size-ranking. If tax inspection depends on employee-size
only indirectly (e.g. by proxying for capital) the exclusion restriction holds as long as
technological demand for labor does not impact capital other than through labor input.
However, this assumption fails if the ﬁrst-stage coefﬁcient does not capture technological
differences in employee-size, but is instead an imperfect proxy for capital input demand.
To try to address this issue we construct two measures of demand for capital input in the
US industry distribution: the demand for external reliance (Rajan and Zingales 1998) and
18 Amadeus data are collected from annual ﬁnancial statements ﬁled with the business registry and au-
dited annual reports presented to shareholders. These ﬁrms are subject to a broader and deeper set of re-
porting requirements and hence in this sample tax inspection is unlikely to be driven by crude size-proxies.
19 The $21,000 cutoff was chosen to correspond to the 90th percentile of income in the WBES such that
these countries are comparable to the US in terms of ﬁscal capacity.
13
the capital to labor ratio (Gordon and Li 2009). We show in Section 4.4 that when control-
ling for capital intensity, the coefﬁcient on size remains unchanged, while the coefﬁcient
on capital intensity is insigniﬁcant. This suggests that technological demand for labor
does not impact tax inspection indirectly through its interaction with capital.
3.2 Empirical Speciﬁcations
The reduced-form size gradient is estimated by directly regressing the WBES industries’
tax outcomes on the US industries size rank:
Tax outcomeict = δ0 + δ1 · Rank sizei,U S ,01 + γct + εict (3)
Where Tax outcomeict is a tax outcome of industry i in country j at time t (e.g. likelihood
of tax inspection over the past 12 months), Rank sizei,U S ,01 is the size-ranking of industry
i in the US, and γct are country-year ﬁxed effects. The coefﬁcient δ1 identiﬁes the reduced-
form size gradient.
The IV speciﬁcation is:
Tax outcomeict = β0 + β1 · Rank sizeict + γct + εict (4)
Where Rank sizeict of industry i in country c at time t is instrumented with Rank sizei,U S ,01
of industry i in the US.
In practice, we estimate three different empirical models. The ﬁrst model adds a set
of controls to the above equations. It allows for tax outcomes to differ non-parametrically
and interactively in every country and year across a set of industry characteristics. We
code all industries as belonging to above or below their country-year median age, share
of exporters and share of foreign ﬁrms. We then create the matrix (Characteristics)ict con-
taining the full set of interactions across characteristics. The model allows for all factors
14
to impact tax outcomes in an interactive way, resulting in 1,937 ﬁxed effects:
Tax outcomeict = β0 + β1 · Rank sizeict + (Characteristics)ict × (Year)t × (Country)i + εict
(5)
The second model allows for the tax outcome to differ in every country, year and ISIC2
industry. This implies that variation relies on size differences of 3-digit ISIC industries
within a 2-digit ISIC industry. For example, within ISIC category 25 "Manufacture of
rubber and plastics products" it exploits variation in ﬁrm size between "Manufacture of
rubber products" (Category 251) and "Manufacture of plastics products" (Category 252).
This model estimates 6,130 ﬁxed effects:
Tax outcomeict = β0 + β1 · Rank sizeict + (ISIC2)ict × (Year)t × (Country)c + εict (6)
Note that in practice, some ISIC2 sectors deﬁne the ISIC3 sectors, which leads to a drop
in sample size. Further, the drop in size is larger for less developed countries where a
smaller degree of specialization implies that some ISIC3 sectors are not represented for a
given ISIC2. For transparency, we present results from each model side by side.
The third speciﬁcation exploits the panel structure of the data, which is available for
a subset of countries (the average number of surveys per country is 1.9).20 In the panel
model, we add ﬁxed-effects at the 3-digit ISIC level to the two previous models. Identiﬁ-
cation comes from variation in industries’ relative size ranks within a country and across
time. For example the panel model mirroring equation 6 is deﬁned as:
Tax outcomeict = β0 + β1 · Rank sizeict + (ISIC2)ict × (Year)t × (Country)c + ISIC3i + εict
(7)
We report the coefﬁcients β1 of size on tax outcomes from all three sets of speciﬁcations
20 Note that in the panel regressions we do not use the instrumental variable strategy.
15
for tax inspection and informal payments. Since the question on tax compliance was
discontinued after 2007, we have very few repeated country surveys. Therefore we only
report the coefﬁcients β1 of size on tax compliance for the ﬁrst two models.
4 Results
4.1 Reduced Form and IV Estimates of Size Gradients
Tax Inspection
In this section we implement the econometric speciﬁcation described in section 3.2. Fig-
ure 4 plots for each of the six largest countries in our sample,21 an industry’s size ranking
on its average probability of tax inspection. Each dot represents an ISIC 3 industry and
the size of the dot is proportional to its share of total employment within the country
(based on the WBES). We plot the linear ﬁt of size rank on tax inspection which slopes up
in all six countries. We also note that on average industries with higher total employment
have larger average ﬁrm sizes, however there is signiﬁcant variation and some industries
with a high share of total employment rank in the bottom half of average ﬁrm size.
Table 2 reports the size gradient in tax enforcement along the extensive margin (any tax
inspection over the past 12 months, Panel A) and the intensive margin (number of tax
inspections over the past 12 months, Panel B). Panel A shows that industry-size is associ-
ated with a higher likelihood of tax inspection. Columns 3 and 4 show that the reduced
form coefﬁcients are signiﬁcant in both types of (cross-sectional) ﬁxed effect models. In
columns 5 and 6, we estimate the corresponding IV-coefﬁcients for each model. The ﬁrst
stage coefﬁcients are strong (F-statistic of 260 and 105, respectively), and the size gradi-
ents are precisely estimated. The coefﬁcient from our preferred speciﬁcation in column 5
implies that a 10 percentile increase in exogenous WBES size-rank is associated with a 2.3
21 Bangladesh, Brazil, China, India, Indonesia and Mexico.
16
percentage point increase in the likelihood of tax inspection, a 3.8% increase relative to a
mean tax inspection probability of 61.9%. Columns 7 and 8 exploit the panel dimension
of the data, for the subset of WBES countries with multiple surveys. In these speciﬁca-
tions the size gradients are estimated from changes in inspection from a switch in the
relative ranking of ISIC3 industries across time. The coefﬁcients in the panel regression
are positive and signiﬁcant and of similar magnitude as the IV estimates. Panel B, re-
peats the above regressions using as an outcome the number of tax inspections in the past
12 months. The coefﬁcients remain signiﬁcant and positive across speciﬁcations. The IV
coefﬁcient in column 5 suggests that a 10 percentile points increase in exogenous WBES
size-rank is associated with a 0.14 increase in the number of tax inspections, relative to
a mean of 1.98 inspections. The panel coefﬁcients (columns 7-8) are very similar to the
IV estimates. To gauge the extent to which the size gradient endogenously determines
ﬁrm-size, we compare the OLS to the IV coefﬁcient (respectively, comparing columns 1
and 5, and columns 2 and 6). The IV coefﬁcient is larger in three out of four speciﬁcations,
suggesting that ﬁrms might reduce their size to avoid increased tax inspection.
Tax Compliance
We now study the size gradient in tax compliance, which could, in part, be explained by
increasing tax inspection with size. While the identiﬁcation strategy is the same as above,
the question on tax compliance only appears in the ﬁrst waves of the WBES (2003 to 2007),
which implies a smaller sample size and insufﬁcient repeated country surveys to run
panel regressions. Table 3 shows the size gradient on the extensive and intensive margins
of tax compliance, where the extensive margin is the probability of full compliance (Panel
A), and the intensive margin is the share of sales reported for tax purpose (Panel B).
The IV coefﬁcient in column 5 points to sizable effects of ﬁrm size on tax compliance:
a 10 percentile increase in the WBES size-rank is associated with a 5.2 percentage point
increase in the likelihood of full compliance, a 10.6% increase relative to a mean of 61.8%
17
and a 2.2 percentage point increase in the share of sales reported for tax purpose (relative
to a mean of 80.9%). The wedge between the OLS and the IV coefﬁcients suggests that
ﬁrms depress their reported size in order to reduce tax compliance.
4.2 Non-linearities of Tax Inspection and Compliance with Size
In Section 4.1 we show that tax inspection and tax compliance increase with industry size.
Here we study whether this relation is linear or concentrated among speciﬁc segments of
the size distribution. We present non-parametrically the reduced form relation between
tax outcomes and the industry size ranking in the US. We ﬁrst residualize the industry
size-ranking in the US Census, NU S , with respect to controls and country and year ﬁxed
effects as in equation 5. Similarly, we residualize the tax outcome of interest (e.g. tax
inspection likelihood) with respect to the controls and ﬁxed effects. We then split the
residualized industry size-ranking NU S into the size deciles, and normalize to zero the
median industry’s size.
Figure 5 shows for each decile of industry size how tax inspection and compliance com-
pare to the level of the 5th decile. Panel A shows that tax inspection is concentrated at the
top of the industry size distribution and is ﬂat in the bottom and middle of the distribu-
tion. Industries at the top of the size distribution appear 3% more likely to be inspected
compared to the median industry. Panel B shows that the tax compliance rate mirrors
inspection: an industry at the top of the size distribution reports 2% more sales compared
to an industry in the middle. The mirroring pattern of tax inspection and tax compliance
suggests that tax compliance is partly driven by size-dependent tax inspection.
4.3 Heterogeneity across Development Levels
Reliance on size-dependent policies for tax enforcement could originate in state capacity:
at lower levels of development the tax authority is constrained by a lack of information
18
on ﬁrms and has to resort to imperfect proxies such as size-dependent polices (Best et
al. 2015, Bachas and Soto 2016). As a country’s ﬁscal capacity grows (Kleven et al. 2016,
Jensen 2016), the tax authority’s reliance on size-dependent policies can be weakened. We
therefore hypothesize that the size gradient in tax inspection decreases with a country’s
income level. To test this hypothesis, we estimate the tax inspection size gradient across
subsamples of countries at ﬁve different levels of development.22
Figure 6 shows the reduced form and IV coefﬁcients for the likelihood of tax inspection at
each income level.23 The magnitude of the tax inspection size-gradient is decreasing over
levels of development, both in the reduced form and IV speciﬁcations.24 While we cannot
reject that lower-middle, middle and higher-middle income groups have different size-
gradients,25 we can reject that their size-gradient is equal to that of high-income countries.
In high-income countries both the reduced form and IV coefﬁcients are centered at zero.
The absence of size-dependent policies at high income levels is consistent with the theory
that countries with strong ﬁscal capacity decrease their reliance on production-inefﬁcient
tax instruments. It also provides some support to the identifying assumption that tax
inspection is orthogonal to size in a high tax capacity environment.
22 The groups correspond to the World Bank income classiﬁcation, except for the top income group
(above USD 21,000) which corresponds to the 90th percentile of income in the WBES and was chosen such
that countries are comparable to the US in terms of ﬁscal capacity.
23 The ﬁrst stage coefﬁcients are constant across income levels and the 1st stage F-statistic is well above
10 at each income level.
24 The coefﬁcients for the lowest income countries (GDP per capita below $1,100) are the largest but are
only signiﬁcantly different from zero at the 10% level. There are two explanations for this. First, this group
contains fewer country-year observations than lower-middle, middle and higher-middle income groups.
Second we weight ISIC-3 industries by their number of observations. Surveys in the poorest countries tend
to be smaller in terms of number of ﬁrms and have fewer ISIC-3 industries represented.
25 These levels of development include the countries often studied in the misallocation literature such as
China, India and Mexico.
19
4.4 Robustness
Using different instruments
The core results use the industry-size rank in the 2001 US Census. This section shows that
results are robust to the choice of the exogenous size-distribution used as an instrument.
Table A1 presents the ﬁrst stage and IV coefﬁcients on tax inspection, for six different
deﬁnitions of the instrumental variable. Columns 1-3 report results using each of the last
three waves of the US ﬁrm Census (1991, 2001, 2015). We obtain the same ranking of
WBES size and similar IV coefﬁcients.26 Since the validity of the instrument rests on the
assumption that the ﬁrm-size distribution in the US is undistorted by tax enforcement,
using size rankings from other high ﬁscal capacity OECD countries should yield similar
results. We test this by constructing the IV with British and German ﬁrms in the Amadeus
database. In Great Britain the data are derived from audited annual reports presented to
shareholders. In Germany, the data come from annual ﬁnancial statements ﬁled with
the business registry. While these ﬁrms represent a selected sample compared to the US
Census, they are subject to stringent reporting requirements and hence it is less likely that
size proxies are used by tax authorities. Column 4 in Table A1 shows that the Amadeus
size-ranking predicts a similar ﬁrst-stage, and a large and signiﬁcant IV coefﬁcient.
Restricting the sample to manufacturing
To exclude the possibility that results are driven by a small set of peculiar ISIC3 industries,
we limit the sample to manufacturing, which is over-sampled in the WBES (60% of WBES
ﬁrms) and where most industries are well represented across countries. Table A2 shows
the reduced form and IV results on the intensive and extensive margins of tax inspection.
The 1st stage coefﬁcient is the same as the coefﬁcient in the full sample. The reduced form
and IV coefﬁcients are larger than in the full sample.
26 The ﬁrst stage is constant across development-levels for each census wave (Figures A1 & A2).
20
Is worker-size proxying for capital intensity?
A potential limitation to our identiﬁcation strategy is that ﬁrm size, as measured by num-
ber of workers, could proxy for capital intensity. If this was the case, we are not capturing
the direct impact of industry-size but rather the impact of more capital on tax inspection.
To mitigate this concern, we construct an industry capital intensity ranking, using (1) the
measure of reliance on external funds as in Rajan and Zingales [1998] and (2) the capital
to labor ratio as in Gordon and Li [2009]. We then regress jointly industry-size ranking
and capital ranking measures on tax inspection (Table A3). The inclusion of controls for
capital intensity does not impact the results, and the coefﬁcients on capital rankings are
very small in magnitude and insigniﬁcant.27
Substitution between formal and informal taxes along the size distribution
Table A4 shows the size gradient on the likelihood of making an informal payment (Panel
A) and the intensive margin as informal tax payments as a share of sales (Panel B). We
ﬁnd no effect on the extensive margin but do ﬁnd an effect on the intensive margin. The
IV coefﬁcient in column (5) suggests that a 10 percentile increase in exogenous WBES
size-rank is associated with a 0.11 percentage point decrease in the informal tax rate, a 7%
increase relative to a base of 1.58% in sales for informal payments.28
The coefﬁcients on tax compliance and informal payments have opposite signs, which
suggests substitution between formal and informal taxation. Full substitution could mit-
igate the importance of size-dependent tax enforcement. To test for substitution, we need
both the informal and formal effective tax rates. Since ﬁrms report informal payments
as a share of sales, we already have the effective informal tax rate. we construct the ef-
27 As an additional robustness test, we ﬁnd that in the reduced form the US-size ranking is associated
with outcomes which are predicted by theory to vary with labor demand (Lucas 1969; Kremer 1993): we
ﬁnd an increase in sales growth, labor cost per permanent employee, perceived constraints due to labor
regulation, likelihood of having quality certiﬁcation and likelihood of being part of a larger ﬁrm.
28 The negative slope shows that informal payments are regressive across ﬁrms. This ﬁnding comple-
ments Olken and Singhal [2011], who ﬁnd that informal tax payments are regressive across households.
21
fective formal tax rate as the product of an industry’s compliance rate with its country’s
statutory tax rate (deﬁned as the sum of the corporate and sales tax rates).29 Figure A4
plots the residualized formal and informal effective tax rates, for each vingtile of the ﬁrm
size distribution. While the informal tax rate rises (falls) over the portions of the size-
distribution where the formal tax rate falls (rises), Panel A shows that the effective formal
tax payments are much larger than informal payments. As a result, substitution between
formal and informal taxes only explains a small share of the size-gradient.
Size gradient interaction with statutory tax rates
When the statutory tax rate is high, ﬁrms have incentives to evade more taxes. For the
tax authority this implies that the returns to size dependent taxation are larger, and hence
the gradient in tax inspection steeper. To test this hypothesis we regress tax outcomes on
the interaction between rank-size and a high tax dummy variable. The high tax dummy
equals one if a country’s statutory tax rates is above the sample’s median and we deﬁne
the statutory tax rate as the sum of the corporate and sales tax rates. Table A5 shows that
the interaction term between the industry’s size rank and the high tax dummy is positive
for tax inspection and compliance. This suggests that in countries with high tax rates the
tax inspection and compliance gradients are steeper.
5 Aggregate Implications of Size-Dependent Enforcement
This section proposes a model of ﬁrm dynamics to evaluate the aggregate implications of
size-dependent tax enforcement and compliance. Our closest reference in the literature
is a closed economy version the model in Atkeson and Burstein [2010]. It features three
channels through which size dependent effective corporate proﬁt and revenue taxes can
29 We recognize that this methodology faces several issues. (1) Statutory tax rates could depend on ﬁrm
size. (2) We assume that the tax compliance question applies to the sales and corporate income tax, while
the question does not refer to a speciﬁc tax. (3) The tax base could vary across taxes.
22
affect aggregate productivity: resource misallocation among incumbents, entry and exit
of ﬁrms, and incentives to invest in innovation.30 The former connects our policy counter-
factuals to the traditional literature on idiosyncratic distortions and resource misalloca-
tion (Hsieh and Klenow 2009, Restuccia and Rogerson 2008). Entry and exit, and endoge-
nous ﬁrm growth represent potential propagating channels for which size-dependent
enforcement affects the macroeconomy and hence should be taken into account for the
quantiﬁcation of its aggregate effects.
5.1 Technologies
There is a competitive representative producer of a ﬁnal good that is produced combin-
ing a continuum of differentiated varieties of intermediate inputs into a CES production
function with elasticity ρ
ρ
ρ −1 ρ−1
Y = q (ω ) ρ dM (ω )
where q (ω ) is the quantity demanded of variety ω , and M (ω ) denotes the mass of pro-
ducers of variety ω . Proﬁt maximization under perfect competition yields the familiar
demand functions
−ρ
d p (ω )
qk (ω , λ) = Y (8)
P
where P is the price index for the ﬁnal good, and p (ω ) is the price of the variety. The
price index for the ﬁnal good is deﬁned by
1
1−ρ
1−ρ
P = p (ω ) dM (ω )
Producers of intermediate varieties, in turn, operate under monopolistic competition
30 By investment in innovation we are not constraining the analysis to thinking just about R&D that
leads to frontier changing innovations or new patents, but rather we think of a broader concept of intangible
capital accumulation that may indeed constitute path-breaking innovations but could also refer to adoption
of frontier technologies or implementation of better management practices.
23
and produce their variety according to the following technology:
1
y (ω ) = (eω ) ρ−1 l (ω ) (9)
1
The idiosyncratic productivity, denoted with (eω ) ρ−1 , evolves endogenously as a re-
sult of the ﬁrms’ investments in innovation subject to risky outcomes. More concretely,
productivity follows a binomial process in which technology experiences upward or down-
ward jumps of exogenously determined size ∆, according to a probability q (ω ) which is
under control of the ﬁrm through expenses in innovation.31 Formally, consider a ﬁrm
with current productivity eω . In the next period, productivity transitions to
eω + ∆
with probability q (ω )
eω =
eω − ∆
with probability 1 − q (ω )
In order to achieve a probability of upgrading equal to q (ω ) , the ﬁrms must incur in a
labor-denominated cost given by
χt (q , ω ) = eω × µ eφq (ω ) − 1
Notice that the innovation cost is scaled by the current productivity of the ﬁrm. This is an
important assumption that allows the model to be consistent with innovation patterns of
large ﬁrms in the U.S. (Gibralt’s Law) , which is our target economy for the calibration of
parameters that are kept constant across economies.
Size-Dependent Taxation The operations of intermediate good producers are affected
by a corporate proﬁt tax τ π and a sales tax τ s . The key property of taxation systems in
31 The appeal of this shock process, as opposed to any other mean reverting one, is that it constitutes
a discrete approximation to a geometric brownian motion, a process under which the properties of the
equilibrium size distribution of ﬁrms can be characterized in closed form (Luttmer 2007) and hence from
which the calibration of parameters can be easily mapped into moments of the size distribution in the data.
24
developing countries, that we incorporate in the model, is that each producer confronts
an idiosyncratic probability of being inspected by the tax authorities and being enforced
to comply with the payment of the taxes. This probability is positively related with the
size of the ﬁrm. We defer for later the discussion of the exact parameterization of the
implementation of the size dependent inspection probability, for now lets denote such
probability with ε (ω ).
The expected proﬁts of an intermediate good producer, then, are given by
e
ε (ω ) (1 − τ π ) [(1 − τ s ) p (ω ) y (ω ) − wl (ω )]
π (ω ) = (10)
+ [1 − ε (ω )] [p (ω ) y (ω ) − wl (ω )] − wχt (q , ω )
The ﬁrst term represents the after-tax stream of proﬁts in the event of enforcement, while
the second terms are the proﬁts in the event of no compliance. Notice that, when enforced,
proﬁts are taxed gross of innovation expenses.
Working out the algebra it follows that a more convenient presentation of the expected
proﬁts is the following32
π (ω ) = [1 − τ π (ω )] {[1 − τ s (ω )] p (ω ) y (ω ) − wl (ω )} − wχt (q , ω ) (11)
where we are deﬁning τ π (ω ) and τ s (ω ) as the “effective” tax rates confronted by the
ﬁrms which we formally deﬁne as follows
τ π (ω ) = ε (ω ) τ π
τ s (ω ) = ε (ω ) τ s
Notice, then, that we have recast the ﬁrms’ problem into the standard representation in
the misallocation literature where ﬁrm face idiosyncratic taxes. Our contribution is to
32 Strictly speaking, there is a difference between the expected proﬁts in 10 and the one in 11, which is
equal to ε (ω ) τ π τ s (1 − ε (ω )). This difference, however, is quantitatively negligible.
25
offer a calibration of such proﬁle of idiosyncratic taxes based on our estimate of size-
dependent tax compliance.
5.2 Static and Dynamic Choices
We can now turn to describing the ﬁrms’ optimal decisions regarding labor demand, in-
vestments in innovation, entry, and exit. Firms choose the labor demand and the price for
their varieties solving the following static proﬁt maximization problem
maxl(ω ),p(ω ) {[1 − τ π (ω )] {[1 − τ s (ω )] p (ω ) y (ω ) − wl (ω )} − wχt (q , ω )}
subject to the demand function for varieties given in 8 and subject to the production tech-
nology deﬁned in 9. Solving it gives
ρ
ρ−1 Yt ω s ρ
l (ω ) = ρ e [1 − τ (ω )]
ρ wt
The expression shows that only the sales tax carry a distortionary effect on labor de-
mand, which is independent from corporate proﬁt taxation. Thus, size-dependent en-
forcement will have a detrimental effect on aggregate productivity through the misallo-
cation of labor, only to the extent determined by the interaction of the size gradient in tax
inspection efforts and the sales tax rate.
Corporate taxation, on the other hand, does have a distortionary effect on the dynamic
decisions of the ﬁrms. To see this, consider the value of an operating ﬁrm with current
productivity eω
[1 − τ π (ω )] π v (ω ) − weω ×µ eφq (ω ) − 1 − wfc +
v o (ω ) = maxq (ω )
β (1 − δ ) q (ω ) v (ω + ∆) + β (1 − δ ) [1 − q (ω )] v (ω − ∆)
where π v (ω ) = [1 − τ s (ω )] p∗ (ω ) y ∗ (ω ) denotes the indirect variable proﬁt function that
26
embodies the optimal choices of labor and prices described above, and where wfc is the
labor-denominated ﬁxed costs of production, which is a driving force of the exit decisions
of ﬁrms. Taking ﬁrst order conditions with respect to innovation choices we get
wµφeφq (ω ) = β (1 − δ ) [v (ω + ∆) − v (ω − ∆)]
The expression shows that corporate taxation affects innovation through its impact on the
rate of return to ﬁrm growth, which is embodied in the value-differential that ﬁrms enjoy
in the event of a technological upgrade. To the extent that corporate taxation is increasing
in idiosyncratic productivity, the differential gain in valuation that an innovation would
allow for is reduced by the proﬁle of corporate taxation, so we should expect to see less
innovation, and hence a lower aggregate productivity, through this channel.
Having characterized the decisions of an operating establishment, we turn now to
describing the process of entry and exit. The latter follows in part from the exogenous
probability δ which affects all ﬁrms uniformly,33 but also reacts to the ﬁrms’ endogenous
decisions about the proﬁtability of continuing in operation. Let ι (ω ) = 1 if the ﬁrm stays
in operation, and equal to zero otherwise, then
v (ω ) = maxι(ω ) [v o (ω ) , 0]
which is the standard deﬁnition of the value of a ﬁrm as the maximum between staying
in operation and the exit value of zero.
In terms of entry, we assume that there is an inﬁnite pool of potential entrants that
must incur a labor-denominated sunk cost of entry equal to fe , in order to get a random
realization of its initial idiosyncratic productivity from a known distribution Γ (ω ). As-
33 Thisexit shock helps rationalize the model with empirical exit by large ﬁrms, which the model cannot
account for endogenously. Thus, it will be calibrated by the exit rate of large ﬁrms in the US economy.
27
suming positive entry in equilibrium, the following free-entry condition must hold:
wfe = β (1 − δ ) v (ω ) dΓ (ω )
where notice that, as is standard in the literature, we are assuming a one-period lag be-
tween the payment of the entry cost and the realization of the productivity draw.
5.3 Household’s Problem
There is a representative household with preferences of the form ∑∞ t
t=0 β [log (Ct )], where
Ct is the single ﬁnal good produced in the economy. Lifetime utility maximization is done
subject to a standard inter-temporal budget constraint of the form:
Σ∞
t=0 Qt [Ct − Wt L − Tt ] ≤ M0
where Qt denotes inter-temporal prices, M0 is the initial endowment of wealth (claims to
the proﬁts and losses from the initial distribution of ﬁrms), and Tt represents the lump-
sum tax/transfer that balances the deﬁcit or surplus from the collection of idiosyncratic
taxes and subsidies. Note that by rebating revenue back to (or taking it away from) the
household, we are ensuring that all the aggregate implications of misallocation frictions
manifest solely through their effect on aggregate productivity, rather than from waste-
ful consumption of goods from the government. Lastly, notice that the inter-temporal
Qt + 1
accumulation of wealth pins down the economy’s interest rate, given by Rt = Qt .
5.4 Deﬁnition of Equilibrium
The remaining step in the characterization of the equilibrium is to aggregate the outcomes
of the ﬁrm up to the level of the macroeconomy. Key for the aggregation is the distribution
of ﬁrms across ﬁrm-wide productivity, which we denote with Mt (ω ). This distribution
28
evolves according to the following law of motion:
Mt+1 (ω ) = (1 − δ )q (ω − ∆) Mt (ω − ∆) (12)
+ (1 − δ ) [1 − q (ω + ∆)] Mt (ω + ∆) + (1 − δ ) Me,t Γ (ω )
The expression establishes that a fraction (1 − δ ) q (ω − ∆) of ﬁrms with productivity less
than or equal to (ω − ∆) survives the exogenous exit shock and transitions to a produc-
tivity level that is less than or equal to ω . A fraction (1 − δ ) (1 − q (ω + ∆)) of the mass of
ﬁrms with productivity between ω and (ω + ∆) survives the exit shock and jumps down-
ward to have productivity less than or equal to ω . There is also an inﬂow of new ﬁrms
into this group which is given by the mass of entrants, (1 − δ ) Me,t Γ (ω ) . Endogenous
exit will be driven by the mass of ﬁrms that transition downwards from the productivity
cutoff, (1 − δ ) [1 − q (ω + ∆)] Mt (ω + ∆).
A stationary competitive equilibrium in this economy is given by: 1) consumption deci-
sions from the household C 2) sequences of prices, labor demands,innovation decisions,
value functions, and exit decisions for the producers of varieties, {p(ω ), l (ω ) , q (ω ) , v (ω ), ι (ω )};
3) a sequence of ﬁnal good quantities and demand functions for intermediate variety
Y , y d (ω ) ; 4) a stationary distribution of ﬁrmsM (ω ) and its law of motion (equation
12), 5) a measure of entrants Me 6) a vector of prices and transfers {w, R, P , Q, T } ; 7) a
proﬁle of size-dependent inspection probabilities ε (ω ) and the proﬁt and sales tax rates
τ π and τ s , 8) a distribution of productivity at entry Γ(ω ), and 9) an initial wealth of the
household M0 such that: a) given 6, 4, and 9, 1 solves household’s optimization problem,
b) given 6, 3 and 8, 2 solves the incumbents’ dynamic optimization problem, c) given
p (ω ), 3 solves the ﬁnal good sector’s proﬁt maximization problem, d) Me is such that the
free entry condition is satisﬁed in every period, and e) markets clear in every period:
L = Lp + LI + Lf c + fe Me
29
C=Y
where Lp,t , Lf c,t , and Lf p,t are the aggregate demands for labor in production, labor in
innovation, and ﬁxed costs of operation; deﬁned by:
ρ−1 Y ρ
Lp = eω [1 − τ s (ω )] dM (ω ) (13)
ρ wρ
Lf c = fc dM (ω )
LI = eω × µ eφq (ω ) − 1 dM (ω )
Deﬁnition of GDP and TFP
A well known property of CES production functions is that aggregate output inherits the
functional form from individual technologies, with the aggregate T F P given by a geo-
metric weighted average of individual productivity. In the context of the current model
economy with distortions, we can write aggregate output as follows:
Y = ALp (14)
ρ
(1 − τ s (ω ))ρ−1 dM
ρ−1
1
eω (ω )
A = M θ −1 θ
(15)
eω (1 − τ s (ω )) dM (ω )
where M (ω ) is the probability distribution function associated with the distribution of
M (ω )
the mass of ﬁrms, M (ω ) = dM (ω )
.
Equation 15 reveals the channels through which size-dependent tax enforcement man-
ifests in aggregate productivity. For a given number of ﬁrms and a given distribution of
ﬁrms across productivity levels M (ω ), size dependent taxes reduce aggregate productiv-
ity by inefﬁciently allocating labor across production units. As mentioned earlier, only the
sales tax exerts a distortion on allocative efﬁciency, with no effect stemming from proﬁt
taxation. The aggregate effects are also shaped by the response in innovation decisions
30
of ﬁrms, which reduce aggregate productivity by discouraging investments in technolog-
ical upgrading. In this case, both the proﬁt and the sales tax contribute to shaping the
response in innovation expenses. Lastly, size-dependent taxation affects T F P through
entry and exit, which manifest in the total number of varieties in the economy, M .
6 Quantitative Analysis
We turn now to quantifying the macroeconomic implications, and the contributions of
the micro-channels underlying it, of our estimates of size-dependent tax compliance.
We calibrate parameters of the shock process, ﬁxed costs, entry costs, innovation costs,
and the distribution of productivity at entry to replicate properties of the ﬁrm size distri-
bution and the life-cycle of ﬁrms in the US. Then, we implement a strategy to map our
estimates of the size-gradient to turn it into a productivity-dependent proﬁle of proﬁt and
sales taxes.
We consider two counterfactuals for the evaluation of the gains from reversing size
dependent taxation. Our baseline exercise consists of computing the TFP gains34 accru-
ing from ﬁxing the probability of compliance of all ﬁrms at the level corresponding to
the median decile in the size distribution, while keeping statutory tax rates unchanged.
Since this policy does not ensure that aggregate tax collection is unchanged, we balance
the government’s budget through lump sum transfers to or from the household. As a
complementary exercise, we consider a case where in addition to ﬂattening the proﬁle
of enforcement, we adjust tax rates so as to keep the share of tax collection in GDP and
the relative contribution of proﬁt taxation constant at their levels prior to the reform. We
provide more details about the construction of the counterfactuals below.
34 Notice
that the undistorted version of our model, which is the basis for the calibration, is one in which
the equilibrium is Pareto Optimal. This means that any form of taxation will be detrimental to welfare
and, hence, we could interpret our aggregate results as proxies for the welfare losses from size dependent
taxation. Since we are abstracting from other distortions that this pattern of taxation could be helping
minimize, we prefer to refrain from making inferences about welfare and characterize our results solely in
terms of the implications for T F P .
31
6.1 Calibration of Parameter Values
Table 5 reproduces our calibration values for parameters that remain ﬁxed across all ex-
periments, which are determined in order to match properties of the ﬁrm size distribution
and the life-cycle of ﬁrms in the U.S. manufacturing sector. Speciﬁcally, the parameters in
the innovation cost function, µ and φ, are calibrated to jointly match the share of employ-
ment accounted for by the top 10% of ﬁrms in the U.S. manufacturing sector in 2007, and
the ratio of employment of ﬁrms aged 21-25 relative to ﬁrms of age 1 in the same year.
The exogenous component of the exit rate of ﬁrms in the model, given by δ , is set so as
to replicate the exit rate of large ﬁrms in the U.S. manufacturing sector, which is equal
to 2.5%. The variance of the shock process, in turn, which is given by ∆, is chosen so as
to match the standard deviation of employment growth rates in the cross section of U.S.
ﬁrms, as reported by Atkeson and Burstein [2010]. Lastly, we are assuming a Pareto dis-
tribution of entrants’ productivity, the tail parameter of which we discipline by requiring
that the models’ relative size of the average entrant to the average incumbent equals 20%,
which is the value observed in the U.S. in 2007.
6.2 Calibration of the Enforcement Size-Gradient
The identiﬁcation strategy in the empirical analysis adopts the U.S. industry-size distribu-
tion as an instrument for exogenously assessing the size variation of tax enforcement in all
countries in the sample. We follow a strategy for calibrating the productivity-dependent
size enforcement in the model that is consistent with this empirical strategy. Taking the
estimates for the average size-gradient from the IV regressions, we use the U.S. ﬁrm size
distribution, and the model’s implication that size maps one to one with productivity in
the undistorted equilibrium, to back-out a gradient between the probability of tax en-
forcement and the underlying productivity of the ﬁrms.
There is, however, a remaining difference between the identiﬁcation strategy in the
32
empirical analysis and the calibration of the model. This difference lies in that, while the
identiﬁcation of the average size-gradients stems from the rank of industries across the
industry-level size distribution, these size-gradients are then fed into the ﬁrm size distri-
bution in the model. The implicit assumption which rationalizes this choice is that the
industry-level gradient of taxation applies also across ﬁrms of different sizes within each
industry. Based on theory, we cannot unambiguously conclude whether this assumption
magniﬁes the quantitative results we ﬁnd below or weakens them. A supporting piece
of evidence in favor of the latter is that the average size-gradient masks non-linearities
across the industry size distribution (Section 4.2). In particular, larger industries carry a
disproportionate burden of taxation than smaller industries. Given the higher variance in
the within industry ﬁrm-size distribution, this non-linearity in taxation would translate
in higher average gradients if these were estimated at the ﬁrm level. However, we cannot
exclude the possibility that there is less variation in tax-enforcement within industries,
and that most distortions occur at the industry level, in which case the macroeconomic
effects would be weaker.
To formalize the parameterization of the size-dependent enforcement proﬁle in the
model, we proceed as follows. First, we identify 10 deciles of the size distribution in the
undistorted economy. Then, using a given estimate of the gradient from the IV rank regres-
sions, we compute the probability of enforcement at a given decile of the size-distribution
q (decilei>1 ) = q (decile1 ) + gradient ∗ decilei>1
where decile1 is the smallest decile. This representation maps directly into the interpreta-
tion of the IV coefﬁcients in tables 2 and 3. For instance, taking the average gradient of
0.234 from column 5 of table 2 implies that going from a given decile of the size distribu-
tion to the next one increases the probability of inspection by 2.3%.
As mentioned above, deciles in the space of employment map directly into the space of
physical productivity in the case of the undistorted economy, so the equation above also
33
provides us with a probability of inspection at ten points of the productivity distribution.
We generate probabilities of tax inspection for the entire spectrum of productivity levels
taking a linear interpolation for the points in between the deciles.
As an example, consider ﬁgure 7 which illustrates the enforcement probability and
the effective tax proﬁles under the assumption of a probability of inspection at the lowest
decile of 40%, a tax rate of 30%, and a size gradient of 0.234 (Equal to the average size-
gradient in our preferred speciﬁcation). Notice that in the rightmost ﬁgure, where we
measure the effective sales tax rate on the vertical axis, we get two different slopes. That
is in reﬂection of the skewness of the ﬁrm size distribution in the US, whereby the top
10% largest ﬁrms account for the bulk of employment. So, the ﬁrms at the top decile are
much larger than in the bottom 9 deciles. When we take the linear interpolation, the ﬁrst
9 interpolation nodes are close to each other, whereas the node corresponding to the 10th
decile is much further apart. Even though, in the space of probability-rank, each rank is
equidistant from each other, that is not true in the space of the sizes associated with each
rank, thus the ﬂatter portion of the tax proﬁle when going from the 9th to the 10th decile.
In the quantitative exploration that follows, we split countries into 5 income groups,
from poorest to richest, and use the income-speciﬁc estimates of enforcement gradients
to evaluate the macro and micro implications of a hypothetical reform that eliminates all
taxes for countries at various stages of development. The income-speciﬁc gradients, as
well as the income-speciﬁc statutory corporate and sales tax rates that we feed into our
model are presented in table 4.
6.3 Results
Our baseline exercise for the quantiﬁcation of the gains from reversing size-dependent
taxation consists of ﬁxing the probability of taxation at a common value across ﬁrms,
while keeping the tax rates constant at their statutory levels. Since taxation is disruptive
34
in our model, even absent size-dependence,35 the level at which we ﬁx the probability of
compliance matters for the aggregate implications. Thus, as our benchmark, we adopt
the case where the size-dependent proﬁle of probability of compliance rotates at the level
corresponding to the median of the size distribution, and becomes ﬂat across all ﬁrms. We
also conduct a complementary exercise in which we adjust the statutory rates of revenue
and proﬁt taxation to ensure that the aggregate collection of tax revenue as a share of
GDP, and the breakdown between revenue and proﬁt collection, remain the same as in
the allocation with size-dependent taxation. Notice that in this case the level at which we
ﬁx the probability of compliance does not matter, given that the tax rates will ultimately
be adjusted to achieve a common target.
We report in ﬁgure 8 the counter-factual gains in T F P , as well as the underlying
changes in the average ﬁrm size, the rate of aggregate expenditure on innovation, and
the number of entering ﬁrms. All variables are measured as ratios between their values
in the counterfactual and the initial allocation. Income groups are ordered from poor to
rich on the horizontal axis.
Consider ﬁrst the results from the benchmark counterfactual, labeled "Median Prob.
of Enforcement" in ﬁgure 8. The top-left panel shows that size-dependent tax compliance
is preventing the least developed countries from appropriating a 0.8% gain in TFP relative
to a regime that taxes all ﬁrms with the median probability. Underlying these productivity
gains are an increase in average ﬁrm size of more than 30%, an increase of up to 10% in
aggregate innovation intensity, and a 25% potential drop in the number of entrants. The
aggregate and ﬁrm-level response featured by development group 5 is virtually none.
This is clearly due to the fact that the size gradient in tax compliance was almost equal to
zero among these countries.
The ﬁrm-level responses behind our aggregate results are consistent with stylized facts
35 An established property of our underlying model is that while a common revenue tax rate across all
ﬁrms is neutral for TFP, a common proﬁt tax rate is not. Proﬁt taxes reduce TFP through lower entry and
reduced innovation.
35
about differences in the size distribution and the life-cycle growth of ﬁrms across coun-
tries. For instance, Bento and Restuccia [2017] document the strong positive relationship
between average ﬁrm size and economic development. Our results suggest that part of
these differences can be attributed to cross-country differences in size-dependent tax com-
pliance. Furthermore, the decline in innovation intensity in the model, which translates
into slower growth of ﬁrms over the life cycle, is consistent with the evidence presented
in Hsieh and Olken [2014], which establishes the ﬂatter path of ﬁrm growth in India and
Mexico relative to the U.S. Since the aggregate effects in our counterfactuals are tightly
linked to the ﬁrm-level adjustments in the economy, it is reassuring to ﬁnd that the prop-
erties of these adjustments in the model are consistent with the data.
The increasing nature of the probability of tax enforcement is the key for understand-
ing the decline in entry and the rise in average ﬁrm size that takes place when the size-
dependence is reversed. To see this, consider ﬁrst the allocation under size-dependent
taxation assuming that innovation decisions do not change. Given the pattern of life-
cycle growth endowed to the ﬁrms through the calibration of the shock process, a proﬁle
of taxation probabilities that is more binding as ﬁrms get large is more detrimental to the
average incumbents than it is to to the average entrant. The reason for this is that en-
trants beneﬁt from the lack of taxation when young and suffer the higher taxation when
old. With a positive interest rate, however, this higher probability of taxation in the fu-
ture is discounted at the equilibrium interest rate ex-ante, which reduces the detrimental
effect of taxation on the entrant’s expected proﬁtability. However, from the point of view
of the cross section of ﬁrms, such weakening of effect does not happen. Thus, the average
proﬁts of incumbents fall relative to entrants, requiring an increase in entry to restore the
equilibrium.
The endogenous response in innovation provides another channel through which av-
erage ﬁrm size falls, and justiﬁes the decline in innovation intensities illustrated in the
bottom-left plot. In this case, the outcome is more intuitive: ﬁrms anticipate a higher
36
effective tax rate if they become more productive, thus respond by cutting down on inno-
vation expenditure.
We turn now to discussing the counterfactual in which we adjust tax rates at the same
time we remove size-dependence in compliance. We ﬁnd that the gains from this exercise
(illustrated in ﬁgure8 under the label "Fixed Tax Revenue") are weakened relative to the
baseline, reaching a maximum of 0.3%. The mechanisms at work are the same, as reﬂected
by the identical qualitative response of the economy. However, the increase in average
proﬁt tax rates that is required to attain the targeted revenue mitigates the efﬁciency gains
from reversing the size-dependent component of taxation. The baseline counterfactual
delivered shares of revenue to GDP that fell short of what was being collected by the
government under size-dependent taxation. Except for the highest income group, for
which the size gradient was almost zero and hence there is no difference between the
initial allocation and the counterfactual, the share of tax revenue to GDP was 80% of the
share under size-dependence. In order to attain the exact same share, without appealing
only to revenue taxes (which are neutral to TFP) we had to increase the proﬁt tax rate,
which, as mentioned earlier, hinders TFP by holding down the rise in innovation intensity
(bottom left panel) and strengthening the decline in entry (bottom right panel).
We close the numerical analysis with a discussion of the quantitive relevance of our re-
sults. At ﬁrst sight, keeping in mind the magnitude of the productivity gaps across coun-
tries, our ﬁndings do not seem to be ﬁrst order in the quest to catch up with the rest of the
world. Even if we benchmark our gains relative to the productivity improvements that
arise from removing all forms of idiosyncratic distortions, which average at about 60%,
the magnitude of the gains in our exercise are still small. However, when benchmarking
against other studies that have carefully evaluated a single policy, our results start to fall
in the ballpark of the literature. Evaluations of size dependent labor market regulations,
such as Gourio and Roys [2014] and Garicano et al. [2016] ﬁnd gains that are even lower
than ours, in the order of 0.03%. Also in the context of labor regulation, but considering a
37
potentially more damaging policy, such as taxes to hire and ﬁre workers, Hopenhayn and
Rogerson [1993] ﬁnd TFP gains of up to 2% from removing plausibly calibrated values of
these taxes. The highest gains are obtained in the ﬁnancial frictions literature, where TFP
rises between 7% and 40% when improving access to credit, depending on the features
of the model (Buera et al. 2011, Moll 2014, Midrigan and Xu 2014). We interpret these
ﬁndings as suggesting that observed misallocation is the outcome of a collection of poli-
cies that combine mild distortions with more damaging ones, and that size-dependent tax
enforcement and compliance appears to be part of the former group.
7 Conclusion
This paper studies the prevalence and consequences of size-dependent tax enforcement
and compliance in 140 countries. The identiﬁcation strategy uses the relative ranking of
industries’ average industry-size in the US as an instrument for the relative ranking of
the same industries’ size in developing countries. We ﬁnd a positive size gradient in tax
inspection and tax compliance. This result is robust across speciﬁcations with ﬂexible
controls, variation in size within narrowly deﬁned industries and variation in industries’
size over time within a country. Moreover, we ﬁnd that size-based enforcement is concen-
trated at the top of the industry size distribution and prevalent in lower-income countries,
while it does not appear to be used in rich countries. When quantiﬁed in an general equi-
librium model of ﬁrm dynamics, size-dependence in tax enforcement generates a drop
in TFP of up to 0.8% in lowest-income countries, which is driven by a decline in average
ﬁrm size and a reduction in innovation intensity.
There are several limitations to our study. First, given that the WBES sample only covers
registered ﬁrms, it is unclear to what extent our estimates of size-dependent tax policy
would generalize to non-registered ﬁrms. The second limitation is that we proxy for tax
enforcement using only in-person physical inspections by tax inspectors. Tax authorities
38
have other tools at their disposal to collect information and enforce taxes, including third
party data-sources which allow tax inspectors to ﬂag irregular reporting behavior and
deter evasion through letters and on-line communication with taxpayers. These indirect
means of enforcement are more prevalent in rich countries and it is possible that they sub-
stitute for physical audits, however recent research shows that even in the US, in-person
visits by tax inspectors remain prevalent and have positive impacts on compliance (Bon-
ing et al. 2018).
Finally, we recognize that size-dependent tax inspection may arise endogenously as an
optimal policy under information and resource constraints (Kleven et al. 2016, Bigio and
Zilberman 2011). Therefore while we quantify the TFP distortions generated by size-
dependence, we do not measure the revenue gains from this policy, and do not make a
normative statement. However, we hope that our description and quantiﬁcation of distor-
tions induced by size-dependence helps in understanding tax policy in low ﬁscal capacity
countries, where both positive and normative issues remain active areas of research.
39
References
Almunia, Miguel and David Lopez-Rodriguez, “Heterogeneous Responses to Effective
Tax Enforcement: Evidence from Spanish Firms,” American Economic Journal: Economic
Policy, 2017.
Atkeson, Andrew and Ariel Tomás Burstein, “Innovation, Firm Dynamics, and Interna-
tional Trade,” Journal of Political Economy, 2010, 118 (3), 433–484.
Bachas, Pierre and Mauricio Soto, “Not(ch) Your Average Tax System: Corporate Taxa-
tion Under Weak Enforcement,” 2016.
Bain, Joe S., “Economies of Scale, Concentration, and the Condition of Entry in Twenty
Manufacturing Industries,” American Economic Review, 1954, 44 (1), 15–39.
Bartelsman, By Eric, John Haltiwanger, and Stefano Scarpetta, “Cross-Country Differ-
ences in Productivity : The Role of Allocation and Selection,” American Economic Review,
2013, 103 (1), 305–334.
Basu, Susanto and John G Fernald, “Returns to Scale in U.S. Production: Estimates and
Implications,” Journal of Political Economy, 2016, 105 (2), 249–283.
Bento, Pedro and Diego Restuccia, “Misallocation, Establishment Size, and Productiv-
ity,” American Economic Journal: Macroeconomics, 2017.
Best, Michael, Anne Brockmeyer, Henrik Kleven, Johannes Spinnewijn, and Mazhar
Waseem, “Production vs Revenue Efﬁciency With Limited Tax Capacity: Theory and
Evidence From Pakistan,” Journal of Political Economy, 2015, 123(6) (9717), 1311–1355.
Bigio, Saki and Eduardo Zilberman, “Optimal Business Income Tax Enforcement,” Jour-
nal of Public Economics, 2011, (June), 1021–1035.
Blumenthal, Marsha, Charles Christian, and Joel Slemrod, “Do Normative Appeals Af-
fect Tax Compliance? Evidence from a Controlled Experiment in Minnesota,” National
Tax Journal, 2001, 54, 125–38.
Boning, Will, John Guyton, Ronald Hodge, Joel Slemrod, and Ugo Troiano, “Heard
It Through the Grapevine: A Randomized Experiment Assessing Direct and Network
Effect of Tax Enforcement Strategies,” NBER Working Paper Series, No. 24305, 2018.
Buera, Francisco J., Joseph P. Kaboski, and Yongseok Shin, “Finance and Development:
A Tale of Two Sectors,” American Economic Review, August 2011, 101 (5), 1964–2002.
Burnside, Craig, “Production function regressions, returns to scale and externalities,”
Journal of Monetary Economics, 1996, 37, 177–201.
Chetty, Ray, Nathaniel Hendren, Patrick Kline, and Emmanuel Saez, “Where is the
Land of Opportunity? the Geography of Intergnerational Mobility in the United
States,” Quarterly Journal of Economics, 2014.
Davis, Steven J and Magnus Henrekson, “Explaining National Differences in the Size
and Industry Distribution of Employment,” Small Business Economics, 1999, 12 (1), 59–
83.
de Paula, Aureo and Jose A Scheinkman, “Value Added Taxes, Chain Effects and Infor-
mality,” American Economic Journal: Macroeconomics, 2010, 2 (October), 195–221.
Garicano, Luis, Claire LeLarge, and John Van Reenen, “Firm Size Distortions and the
Productivity Distribution: Evidence from France,” American Economic Review, 2016, 106
(18841), 3439–3479.
Gollin, Douglas, “Do Taxes on Large Firms Impede Growth? Evidence from Ghana,”
40
Bulletins 7488, University of Minnesota, Economic Development Center., 1995.
Gordon, Roger and Wei Li, “Tax structures in developing countries: Many puzzles and a
possible explanation,” Journal of Public Economics, 2009, 93 (7-8), 855–866.
Gourio, François and Nicolas Roys, “Size-dependent regulations, ﬁrm size distribution
and reallocation,” Quantitative Economics, 2014, 5, 377–416.
Guner, Nezih, Gustavo Ventura, and Yi Xu, “Macroeconomic implications of size-
dependent policies,” Review of Economic Dynamics, 2008, 11 (4), 721–744.
Hopenhayn, Hugo and Richard Rogerson, “Job Turnover and Policy Evaluation: A Gen-
eral Equilibrium Analysis,” Journal of Political Economy, 1993, 101 (5), 915–938.
Hsieh, Chang-Tai and Benjamin a. Olken, “The Missing “Missing Middle”,” Journal of
Economic Perspectives, 2014, 28 (3), 89–108.
and PJ Klenow, “Misallocation and Manufacturing TFP in China and India,” Quarterly
Journal of Economics, 2009, CXXIV (November).
Hsieh, CT and PJ Klenow, “The Life Cycle of Plants in India and Mexico,” The Quarterly
Journal of Economics, 2014, 129 (August), 1035–1084.
Ito, Koichiro and James M Sallee, “The Economics of Attribute-Based Regulation: The-
ory and Evidence from Fuel-Economy Standards,” 2016.
Jensen, Anders, “Employment Structure and the Rise of the Modern Tax System,” 2016.
Kanbur, Ravi and Michael Keen, “Thresholds, informality, and partitions of compli-
ance,” International Tax and Public Finance, 2014, 21 (4), 536–559.
Kleven, Henrik Jacobsen, Claus Thustrup Kreiner, and Emmanuel Saez, “Why can
modern governments tax so much? an agency model of ﬁrms as ﬁscal intermediaries,”
Economica, 2016, (15218), 51.
, Martin B Knudsen, Claus Thustrup Kreiner, S\oren S\oren Pedersen, and Em-
manuel Saez, “Unwilling or Unable to Cheat? Evidence From a Tax Audit Experiment
in Denmark,” Econometrica, 2011, 79 (3), 651–692.
Kopczuk, Wojciech and Joel Slemrod, “Putting ﬁrms into optimal tax theory,” American
Economic Review Papers and Proceedings, 2006, 96 (2), 130–134.
Kremer, M., “The O-Ring Theory of Economic Development,” The Quarterly Journal of
Economics, 1993, 108 (3), 551–575.
krishna Kumar, Raghuram Rajan, and Luigi Zingales, “What Determines Firm Size?,”
NBER WORKING PAPER SERIES, 1999, Working Paper 7208.
Lucas, Robert, “Labor-capital substitution in US manufacturing.,” in “In Harberger, A.
C. and Bailey, M. J., editors, The taxation of income from capital. Washington, D. C.:
Brookings Institution.” 1969.
Luttmer, Erzo G. J., “Selection, Growth, and the Size Distribution of Firms,” The Quarterly
Journal of Economics, 2007, 122 (3), 1103–1144.
Midrigan, Virgiliu and Daniel Yi Xu, “Finance and misallocation: Evidence from plant-
level data,” American Economic Review, 2014, 104 (2), 422–458.
Moll, Benjamin, “Productivity Losses from Financial Frictions: Can Self-Financing Undo
Capital Misallocation?,” American Economic Review, October 2014, 104 (10), 3186–3221.
Naritomi, Joana, “Consumers as Tax Auditors,” 2016.
Olken, Benjamin A and Monica Singhal, “Informal Taxation,” American Economic Jour-
nal: Applied Economics, 2011, 3 (October), 1–28.
Pomeranz, Dina, “No Taxation without Information: Deterrence and Self-Enforcement in
41
the Value Added Tax,” American Economic Review, 2015, (8), 2539–2569.
Rajan, Raghuram G and Luigi Zingales, “American Economic Association,” The Ameri-
can Economic Review, 1998, 88 (3), 559–586.
Restuccia, Diego and Richard Rogerson, “Policy distortions and aggregate productivity
with heterogeneous establishments,” Review of Economic Dynamics, 2008, 11 (4), 707–720.
World Bank, “Doing Business 2017: Equal Opportunity for All.,” Washington, DC: World
Bank. DOI: 10.1596/978-1-4648-0948-4. License: Creative Commons Attribution CC BY 3.0
IGO, 2017.
Zareh, Asatryan and Andreas Peichl, “Responses of Firms to Tax, Administrative and
Accounting Rules: Evidence from Armenia,” 2016.
42
F IGURE 1: N UMBER OF C OUNTRIES WITH A L ARGE TAXPAYER U NIT O VER T IME
100
Countries with a Large Taxpayer Unit
80
60
40
20
0
1960 1970 1980 1990 2000 2010
Year
Source: Data collected by the authors. All 113 countries, with more than one million inhabitants.
In the past twenty years more than a 100 countries adopted a Large Taxpayer Unit (LTU). This development
is part of an increased trend of taxpayer segmentation, recommended by international institutions.
F IGURE 2: US I NDUSTRY S IZE ON WBES I NDUSTRY S IZE [F IRST S TAGE ]
40
35
Mean WBES Size Rank
20 25 15 30
Rank−Rank Slope=.439
(.008)
10
0 10 20 30 40 50
US Census Size Rank
Source: US Census 2001 and World Bank Enterprise Surveys 2003-2015.
Figure 2 presents non-parametric binned scatterplot of the relationship between WBES countries and US
ﬁrm size ranks, as discussed in Section 3.1. WBES size rank is the rank of an ISIC3 by number of permanent
employees per ﬁrm in the WBES country-year. US Census size rank is the ISIC3 rank of number of employ-
ees per ﬁrm in 2001, excluding ﬁrms with less than 5 employees. In each country-year, ISIC3 industries are
ranked relative to other industries in the same country-year survey. The WBES and US Census ranks are
then grouped into 50 equal sized (two percentile) bins.
43
F IGURE 3: I NDUSTRY S IZE IN THE US ON I NDUSTRY S IZE IN THE WBES BY I NCOME
L EVEL
Low income countries Lower−middle income countries
20
20
Mean WBES Size Rank
Mean WBES Size Rank
15
15
10
10
Rank−Rank Slope=.471 Rank−Rank Slope=.479
(.040) (.017)
5
5
0 5 10 15 20 25 0 5 10 15 20 25
US Census Size Rank US Census Size Rank
Upper−middle income countries High−income countries
20
20
Mean WBES Size Rank
Mean WBES Size Rank
15
15
10
10
Rank−Rank Slope=.424 Rank−Rank Slope=.411
(.013) (.014)
5
5
0 5 10 15 20 25 0 5 10 15 20 25
US Census Size Rank US Census Size Rank
Source: US Census 2001 and World Bank Enterprise Surveys 2003-2015.
Figure 3 presents non-parametric binned scatterplot of the relationship between WBES countries and US
ﬁrm size ranks, in the four per capita income groups based on the World Bank income classiﬁcation. These
results are discussed in Section 3.1 The cut-off per-capita income values are: 1100 for low income countries;
between 1100 and 3900 for lower-middle income countries; between 3900 and 12730 for upper-middle in-
come countries; above 12730 for high income countries. WBES size rank is the ISIC3 rank of number of
permanent employees per ﬁrm in the WBES country-year. US Census size rank is the ISIC3 rank of number
of employees per establishment in 2001, where we exclude establishments with less than 5 employees. In
each country-year, ISIC3 industries are ranked relative to other industries in the same country-year sur-
vey. Within each per capita income group, the WBES and US Census ranks are then grouped into 25 equal
sized bins. For each per capita income group, this graph plots the mean WBES size rank percentile within
each 25-quantile US size rank percentile, together with a best-ﬁt line. The rank-rank slope coefﬁcient and
standard error are estimated in each per capita income group.
44
F IGURE 4: I NDUSTRY S IZE AND TAX I NSPECTION IN M AJOR C OUNTRIES
Bangladesh Brazil China
100
100
100
80
80
80
Likelihood of tax inspection
Likelihood of tax inspection
Likelihood of tax inspection
60
60
60
40
40
40
20
20
20
0
0
0
0 20 40 60 80 0 20 40 60 80 0 20 40 60 80
Size Rank Size Rank Size Rank
India Indonesia Mexico
100
100
100
80
80
80
Likelihood of tax inspection
Likelihood of tax inspection
Likelihood of tax inspection
60
60
60
40
40
40
20
20
20
0
0
0
0 20 40 60 80 0 20 40 60 80 0 20 40 60 80
Size Rank Size Rank Size Rank
ISIC3 Industry Average Linear fit
Source: World Bank Enterprise Surveys 2003-2015.
Figure 4 plots each ISIC 3 industry by its ﬁrm size rank on the probability of tax inspection in the six most
populous countries, as discussed in Section 4.1. When multiple surveys exist for a country we use the latest
survey. The size of the dots is the share of total employment (in our sample) within the country. Therefore
dot size are not comparable across countries but only show relative size of industries within a country. The
red line plots the linear ﬁt of size rank on tax inspection.
45
F IGURE 5: N ON -L INEAR I MPACT OF F IRM S IZE ON TAX I NSPECTION & C OMPLIANCE
( A ) PANEL A: TAX I NSPECTION
Probability of tax inspection, relative to median industry size]
−1 −.5 0 .5 1 1.5 2 2.5 3 3.5
1 2 3 4 5 6 7 8 9 10
Deciles of industry average size
( B ) PANEL B: TAX C OMPLIANCE
Probability of tax inspection, relative to median industry size]
−.5 0 .5 1 1.5 2 2.5 3
1 2 3 4 5 6 7 8 9 10
Deciles of industry average size
Source: US Census 2001 and World Bank Enterprise Surveys 2003-2015.
Figure 5 shows for each decile of industry-size the likelihood of tax inspection and tax compliance com-
pared to the median industry-size, as discussed in section 4.2. To construct these ﬁgures, we ﬁrst residualize
the tax outcomes and industry-size variables with respect to the country-year-control interactive ﬁxed ef-
fects. We then group the residualized industry-size into deciles and plot the average of the residualized tax
outcome within each bin. We winsorize at the 5% level within industry size-deciles. The speciﬁcation cor-
responds to the regressions in Column 3 of Table 2 (Panel A) and Column 3 of Table 3 (Panel B) respectively,
and use the same sample restrictions and variable deﬁnitions.
46
F IGURE 6: S IZE G RADIENTS BY I NCOME L EVELS
.04
Size−Gradient in Tax Enforcement
.03
.02
.01
0
[540,1100] (1100,3900] (3900,12730] (12730,21000] (21000,47100]
Income Group (GDP per capita in PPP)
Reduced form 5pct signif Reduced form 5pct non−signif
IV 5pct signif IV 5pct non−signif
Source: US Census 2001 and World Bank Enterprise Surveys 2003-2015.
Figure 6 show the reduced form (diamonds) and IV coefﬁcients (triangles), replicating the regressions in
Columns 3 and 5 of Table 2 for each development level, as discussed in Section 4.3. The groups correspond
to the World Bank income classiﬁcation, except for the top income group (above USD 21,000) which corre-
sponds to the 90th percentile of income in the WBES and was chosen such that countries are comparable to
the US in terms of economic development.
F IGURE 7: TAX P ROFILES
Probability of Inspection vs Decile Effective Tax vs Log Empl
0.24
0.60
0.22
0.55
0.20
effective tax
proba[, 1]
0.50
0.18
0.45
0.16
0.40
0.14
2 4 6 8 10 0 10 20 30
decile log Empl
The left panel of Figure 7 shows the enforcement probability against deciles of the employment size dis-
tribution, as discussed in Section 6.2. It assumes an average size-gradient of 0.023, and a probability of
inspection at the lowest decile of 40%. The right panel shows the corresponding effective sales tax rates,
assuming a statutory sales tax rate of 30%
47
F IGURE 8: A GGREGATE AND M ICRO I MPLICATIONS OF S IZE -D EPENDENT I NSPEC -
TION
TFP Av.Firm Size
1.010
1.5
1.008
Median Prob. of Enforcement
1.4
Fixed Tax Revenue
1.006
1.3
1.004
1.2
1.002
1.1
1.000
1.0
1 2 3 4 5 1 2 3 4 5
Development Group (1 = poorest) Development Group (1 = poorest)
Average Innovation Intensity Entry of Firms
1.15
1.00
0.95
1.10
0.90
0.85
1.05
0.80
1.00
0.75
1 2 3 4 5 1 2 3 4 5
Development Group (1 = poorest) Development Group (1 = poorest)
Figure 8 shows results from baseline counter-factual that equalizes probability of compliance at the level
of the median-size ﬁrm, keeping ﬁxed the revenue and proﬁt tax rates; and the counterfactual that also
adjusts tax rates to achieve the same share of tax revenue to GDP, and the same share of proﬁt revenue
over total revenue, as in the initial allocations. The former is labeled "Median Prob. of Enforcement", while
the latter is labeled "Fixed Tax Revenue". All variables are illustrated as ratios between their values in
the counterfactual and their values in the equilibrium with size-dependent taxation. Income groups in the
horizontal axes are ranked from low to high.
48
TABLE 1: S UMMARY S TATISTICS
(1) (2) (3) (4) (5) (6)
Variable N Mean s.d. p25 p50 p75
Panel A: Industry Level Outcomes (ISIC3)
Any Tax Inspection [0,100%] 12,152 61.88 35.65 37.50 66.67 100
Number of Tax Inspections 12,152 1.98 2.13 0.67 1.48 2.72
Full Compliance [0,100%] 3,343 57 38.1 25 60 100
Compliance Rate 3,343 80.93 23.6 70 89.6 100
Any Informal Payments [0,100%] 10,013 26.00 36.41 0 0 50
Informal Payments, Share of Sales 10,013 1.58 4.22 0 0 1.50
Formal Tax Payments, Share of Sales 3,048 33.28 11.9 25.8 35 42
Formal and Informal Tax Payments, Share of Sales 2,864 34.97 12.04 27.63 36 42.46
Panel B: Inudustry Level Covariates (ISIC3)
Age 12,152 18.64 13.16 10.75 15.57 22.67
Share of Establishment Owned by Foreign Companies 12,149 0.137 0.266 0 0 0.15
Share of Annual Sales Destined for Direct Export 12,093 0.228 0.326 0 0 0.36
Manufacturing Sector {0,1} 12,152 0.58 0.49 0 1 1
Statutory Corporate Tax Rate + Sales Tax Rate 11,524 0.39 0.09 0.35 0.40 0.45
GDP Per Capita, Constant USD 2011 12,117 10,673 8,554 3,736 8,837 16,459
Panel C: Industry Size Characteristics (ISIC3)
WBES Size 12,146 69.42 289.5 12.94 24.66 52.84
US Census 1991 Size 11,330 48.2 39.09 27.33 38.28 66.18
US Census 2001 Size 11,962 52.95 43.73 28.20 40.48 67.95
US Census 2015 Size 11,958 51.65 40.89 27.79 37.40 62.47
EU Amadeus 2007 Size 12,032 393.7 891.7 116.76 184.05 346.84
WBES RKZ-Weighted Size 12,146 262.3 840.3 21.06 67.40 200.57
US Census 2001 RKZ-Weighted Size 12,001 302.2 306.60 125.58 218.41 384.67
Source: World Bank Enterprise Surveys 2003-2015.
Table 1 displays the summary statistics at the industry-country-year level, as described in section 2. In-
dustries are deﬁned at the 3-digit ISIC level, using the UN classiﬁcation version 3.1. Column 1 reports the
number of industry level observations, column 2 the mean, column 3 the standard deviation and columns
4-6 the quartiles (25% percentile, median and 75% percentile). All statistics are reported for the full sample
which includes some observations for which a sampling weights are missing.
49
TABLE 2: I MPACT OF F IRM S IZE ON TAX I NSPECTION
OLS Reduced Form IV Panel: OLS
(1) (2) (3) (4) (5) (6) (7) (8)
Panel A
Outcome: Tax Inspection {0,1}
Industry Size Rank (World Bank) 0.197 0.232 0.234 0.189 0.303 0.276
(0.054) (0.043) (0.055) (0.109) (0.065) (0.067)
Industry Size Rank (US Census) 0.073 0.050
(0.017) (0.030)
FE country year controls
FE country year ISIC2
FE ISIC3
F-Stat 259.68 105.49
R-squared 0.650 0.772 0.641 0.766 0.650 0.772 0.890 0.906
Panel B
Outcome: # of Tax Inspections
Industry Size Rank (World Bank) 0.0112 0.0161 0.0142 0.0181 0.0170 0.0155
(0.0030) (0.0025) (0.0034) (0.0056) (0.0046) (0.0038)
Industry Size Rank (US Census) 0.0045 0.0048
(0.0010) (0.0017)
(0.017) (0.030)
FE country year controls
FE country year ISIC2
FE ISIC3
F-stat 259.68 105.49
R-squared 0.600 0.734 0.591 0.725 0.599 0.734 0.886 0.896
Observations 9,772 7,864 9,772 7,864 9,772 7,864 5,048 4,176
Number of clusters 131 131 131 131 131 131 75 77
Source: US Census 2001 and World Bank Enterprise Surveys 2003-2015.
Robust standard errors in parentheses, clustered at the country level.
Table 2 shows the coefﬁcients of regressions of industry ﬁrm-size ranking on tax inspection, discussed in
section 4.1. Panel A measures tax inspection as a dummy which takes the value one if the ﬁrm was vis-
ited during the year. Panel B measures tax inspection as the number of times the ﬁrm was visited by tax
inspectors. Odd number columns include interactive ﬁxed effects between country, year and 3 dummies
indicating whether the ISIC3 industry was above median in the country-year distributions of age, export-
share, and foreign-industry share. Even number columns include interactive ﬁxed effects between country,
year and 2-digit ISIC codes. In this case, size-rank coefﬁcients are estimated using variation between 3-digit
ISIC codes, controlling non-parametrically for country-year-2digit ISIC effects. For a given speciﬁcation,
the sample size in the even columns is smaller than in the odd columns. This is because the ISIC2 interac-
tive ﬁxed effect speciﬁcation drops all cells where there does not exist more than one ISIC3-country-year
observation within a ISIC2-country-year observation. The sample drop occurs in industries that are less
specialized, which are more prevalent in less developed countries. In columns (5) & (6), we instrument for
the World Bank size rank using the US Census size-rank. The F-statistic comes from the ﬁrst stage rank-
rank regression of World Bank size rank on US Census size rank. In columns (7) & (8), we add ISIC 3-digit
ﬁxed effects, so that the size-rank coefﬁcients are estimated using variation within ISIC3 over time in its
country-year industry rank. The ISIC3 ﬁxed effect models result in a drop in sample size because the ISIC3
panel structure does not exist for some of the early WBES country surveys.
50
TABLE 3: I MPACT OF F IRM S IZE ON TAX C OMPLIANCE
OLS Reduced Form IV
(1) (2) (3) (4) (5) (6)
Panel A
Outcome: Full Tax Compliance {0,1}
Industry Size Rank (World Bank) 0.244 0.256 0.519 0.401
(0.045) (0.072) (0.138) (0.169)
Industry Size Rank (US Census) 0.175 0.132
(0.045) (0.062)
FE country year controls
FE country year ISIC2
F-stat 35.12 38.80
R-squared 0.626 0.736 0.624 0.733 0.617 0.735
Panel B
Outcome: % Sales Reported
Industry Size Rank (World Bank) 0.152 0.150 0.221 0.183
(0.034) (0.047) (0.088) (0.114)
Industry Size Rank (US Census) 0.075 0.060
(0.035) (0.041)
FE country year controls
FE country year ISIC2
F-stat 35.12 38.80
R-squared 0.719 0.795 0.715 0.793 0.718 0.795
Observations 3,187 2,348 3,187 2,348 3,187 2,348
Number of clusters 83 83 83 83 83 83
Source: 2001 US Census and World Bank Enterprise Surveys 2003-2007.
Robust standard errors in parentheses, clustered at the country level.
Table 3 shows the coefﬁcients of regressions of industry ﬁrm-size ranking on tax compliance, discussed in
section 4.1. Tax compliance is measured as the response to the survey question: "Recognizing the difﬁculties
many enterprises face in fully complying with taxes and regulations, what percentage of sales would you
estimate the typical establishment in your area of activity reports for tax purposes?". Panel A deﬁnes full tax
compliance as a dummy equal to one if a ﬁrm reports reporting 100% of its sales for tax purposes. Panel B
uses the percentage of sales reported for tax purpose. Odd number columns include interactive ﬁxed effects
between country, year and 3 dummies indicating whether the ISIC3 industry was above median in the
country-year distributions of age, export-share, and foreign-industry share. Even number columns include
interactive ﬁxed effects between country, year and 2-digit ISIC codes. In this case, size-rank coefﬁcients
are estimated using variation between 3-digit ISIC codes, controlling non-parametrically for country-year-
2digit ISIC effects. For a given speciﬁcation, the sample size in the even columns is smaller than in the odd
columns. This is because the ISIC2 interactive ﬁxed effect speciﬁcation drops all cells where there does not
exist more than one ISIC3-country-year observation within a ISIC2-country-year observation. The sample
drop occurs in industries that are less specialized, which are more prevalent in less developed countries.
In columns (5) & (6), we instrument for the World Bank size rank using the US Census size-rank. The F-
statistic comes from the ﬁrst stage rank-rank regression of World Bank size rank on US Census size rank.The
tax compliance question was only administered in the earlier waves of the WBES surveys which were not
structured as panels. Hence, we cannot estimate panel models for the tax compliance outcomes.
51
TABLE 4: G RADIENTS AND TAX R ATES BY I NCOME G ROUP
Income Level Gradient Sales Tax Corporate Tax
1 (poorest) 3.98% 9.06% 25.66%
2 2.44% 12.94% 21.35%
3 3.27% 11.98% 19.47%
4 2.04% 16.14% 18.6%
5 0 16.13 23.59
Note: Estimates of gradients by income group are drawn from Figure 7, panel A. Statutory tax rates are
averages among countries in each income group
TABLE 5: PARAMETER VALUES AND C ALIBRATION TARGETS
Parameter Value Target
ρ 3 Hsieh and Klenow (2009), Broda and Weinstein (2006)
1
β 1.05
Interest Rate of 5%
δ 0.025 Employment-Based Exit Rate of Large Firms of 2.5%
∆ 0.25 Std Dev. Employment Growth rate
φ 15 Top 10% Employment Share
µ 0.00041 Employment Age 21-25 relative to Age 1
G(ω ) P areto,η = 4.46 Empl. Ratio Entrants to Incumbents
fc
fe
0.1 Exit Rate of 5%
Note: The top 10% employment share, the average employment ratio between 21-25 and 1 year old ﬁrms,
and the average employment ratio between entrants and incumbents were computed from Business Dy-
namics Statistics database for the year 2007. Numbers are for the manufacturing sector. Standard deviation
of employment growth rates for large ﬁrms are reported in Atkeson and Burstein (2010).
52
A Appendix Figures & Tables (Not for Publication)
F IGURE A1: I NDUSTRY S IZE IN US C ENSUS 91 ON I NDUSTRY S IZE IN WBES BY I N -
COME
Low−income countries Lower−middle income countries
20
20
Mean WBES Size Rank
Mean WBES Size Rank
15
15
10
10
Rank−Rank Slope=.438 Rank−Rank Slope=.485
(.041) (.017)
5
5
0 5 10 15 20 25 0 5 10 15 20 25
US 1991 Census Size Rank US 1991 Census Size Rank
Upper−middle income countries High−income countries
20
20
Mean WBES Size Rank
Mean WBES Size Rank
15
15
10
10
Rank−Rank Slope=.416 Rank−Rank Slope=.406
(.013) (.014)
5
5
0 5 10 15 20 25 0 5 10 15 20 25
US 1991 Census Size Rank US 1991 Census Size Rank
Source: 1991 US Census and World Bank Enterprise Surveys 2003-2015.
Figure A1 presents non-parametric binned scatterplot of the relationship between WBES countries’ and US’
1991 percentile ﬁrm size ranks in the four per capita income groups based on the World Bank classiﬁcation.
The USD cut-off values used for the classiﬁcation are: 1100 for low income countries; between 1100 and 3900
for lower-middle income countries; between 3900 and 12730 for upper-middle income countries; above
12730 for high income countries. This ﬁgure is based on the full WBES sample and the US Census 1991.
WBES size rank is the ISIC3 rank of number of permanent employees per ﬁrm in the WBES country-year.
US Census size rank is the ISIC3 rank of number of employees per establishment in 199, where we exclude
establishments with less than 5 employees. Within each per capita income group, the WBES and US Census
ranks are then grouped into 25 equal sized bins. For each per capita income group, this graph plots the
mean WBES size rank percentile within each 25-quantile US size rank percentile, together with a best-ﬁt
line. The rank-rank slope coefﬁcient and standard error are estimated in each per capita income group
using the underlying ISIC3-year-country data.
53
F IGURE A2: I NDUSTRY S IZE IN US C ENSUS 15 ON I NDUSTRY S IZE IN WBES BY I N -
COME
Low income countries Lower−middle income countries
20
20
Mean WBES Size Rank
Mean WBES Size Rank
15
15
10
10
Rank−Rank Slope=.411 Rank−Rank Slope=.451
(.046) (.018)
5
5
0 5 10 15 20 25 0 5 10 15 20 25
US 2015 Census Size Rank US 2015 Census Size Rank
Upper−middle income countries High income countries
20
20
Mean WBES Size Rank
Mean WBES Size Rank
15
15
10
10
Rank−Rank Slope=.399 Rank−Rank Slope=.402
(.014) (.016)
5
5
0 5 10 15 20 25 0 5 10 15 20 25
US 2015 Census Size Rank US 2015 Census Size Rank
Source: 2015 US Census and World Bank Enterprise Surveys 2003-2015.
Figure A2 presents non-parametric binned scatterplot of the relationship between WBES countries’ and US
2015 percentile ﬁrm size ranks in the four per capita income groups based on the World Bank classiﬁcation.
The USD cut-off values used for the classiﬁcation are: 1100 for low income countries; between 1100 and 3900
for lower-middle income countries; between 3900 and 12730 for upper-middle income countries; above
12730 for high income countries. This ﬁgure is based on the full WBES sample and the US Census 2015.
WBES size rank is the ISIC3 rank of number of permanent employees per ﬁrm in the WBES country-year.
US Census size rank is the ISIC3 rank of number of employees per establishment in 2015, where we exclude
establishments with less than 5 employees. Within each per capita income group, the WBES and US Census
ranks are then grouped into 25 equal sized bins. For each per capita income group, this graph plots the
mean WBES size rank percentile within each 25-quantile US size rank percentile, together with a best-ﬁt
line. The rank-rank slope coefﬁcient and standard error are estimated in each per capita income group
using the underlying ISIC3-year-country data
54
F IGURE A3: I NDUSTRY S IZE IN EU A MADEUS 07 ON I NDUSTRY S IZE IN WBES BY
I NCOME
Low income countries Lower−middle income countries
20
20
Mean WBES Size Rank
Mean WBES Size Rank
15
15
10
10
Rank−Rank Slope=.403 Rank−Rank Slope=.372
(.041) (.017)
5
5
0 5 10 15 20 25 0 5 10 15 20 25
EU Amadeus Size Rank EU Amadeus Size Rank
Higher−middle income countries High−income countries
20
20
Mean WBES Size Rank
Mean WBES Size Rank
15
15
10
10
Rank−Rank Slope=.306 Rank−Rank Slope=.269
(.014) (.015)
5
5
0 5 10 15 20 25 0 5 10 15 20 25
EU Amadeus Size Rank EU Amadeus Size Rank
Source: 2007 EU Amadeus and World Bank Enterprise Surveys 2003-2015.
Figure A3 presents non-parametric binned scatterplot of the relationship between WBES countries’ and EU’
percentile ﬁrm size ranks in the four per capita income groups based on the World Bank classiﬁcation. The
USD cut-off values used for the classiﬁcation are: 1100 for low income countries; between 1100 and 3900 for
lower-middle income countries; between 3900 and 12730 for upper-middle income countries; above 12730
for high income countries. This ﬁgure is based on the full WBES sample and the EU Amadeus sample of
UK and German ﬁrms in 2007. WBES size rank is the ISIC3 rank of number of permanent employees per
ﬁrm in the WBES country-year. EU Amadeus Census size rank is the ISIC3 rank of number of employees
per establishment in 2007 in Germany and UK. Within each per capita income group, the WBES and EU
Amadeus ranks are then grouped into 25 equal sized bins. For each per capita income group, this graph
plots the mean WBES size rank percentile within each 25-quantile EU Amadeus size rank percentile, to-
gether with a best-ﬁt line. The rank-rank slope coefﬁcient and standard error are estimated in each per
capita income group using the underlying ISIC3-year-country data.
55
F IGURE A4: N ON -L INEAR I MPACT OF F IRM S IZE ON F ORMAL & I NFORMAL TAX
R ATES
( A ) PANEL A
2
Residualized coefficent [percentage points]
−.5 0 .5 1 1.5
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Vingtiles of residualized firm size
Formal effective tax rate Informal effective tax rate
( B ) PANEL B: AS P ERCENTAGE OF M EAN OF O UTCOMES
.1
Residualized coefficient [% of mean]
0
−.1
−.2
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Vingtiles of residualized firm size
Formal effective tax rate Informal effective tax rate
Source: US Census 2001 and World Bank Enterprise Surveys 2003-2015.
Figure A4 shows binned scatter plots of the effective formal and informal tax rates on ﬁrm size, discussed
in section ??. These plots correspond to the regressions in Column 3 of Table 3 Panel B and Table 4 Panel
A respectively, and use the same sample restrictions and variable deﬁnitions. To construct these panels,
we ﬁrst residualize the y-axis and x-axis variables with respect to the country-year-control interactive ﬁxed
effects (deﬁned in Section X equation Y). We then group the residualized ﬁrm-size into twenty equal-sized
bins (vingtiles) and plot the median of the y-variable residual within each bin. The median is chosen to
reduce the impact of residual outliers.
56
TABLE A1: F IRST-S TAGE AND IV FOR D IFFERENT I NSTRUMENTAL VARIABLES
Panel A: First Stage Industry Size Rank (WBES)
(1) (2) (3) (4) (5) (6)
Industry Size Rank (US Census 01) 0.308
(0.019)
Industry Size Rank (US Census 91) 0.337
(0.022)
Industry Size Rank (US Census 15) 0.253
(0.020)
Industry Size Rank (EU Amadeus 07) 0.101
(0.017)
Industry Size Rank (KRZ weighting) 0.082
(0.031)
Industry Size Mean (US Census 01) 0.342
(0.048)
R-squared 0.552 0.562 0.524 0.501 0.624 0.140
Panel B: IV Results Tax Inspection {0,1}
(1) (2) (3) (4) (5) (6)
Industry Size Rank (WBES) 0.242
(0.053)
Industry Size Rank (WBES) 0.188
(0.048)
Industry Size Rank (WBES) 0.167
(0.050)
Industry Size Rank (WBES) 0.765
(0.217)
Industry Size Rank (WBES) 0.492
(0.152)
Industry Size Mean (WBES) 0.091
(0.020)
1st stage rank-rank based on sample USCen01 USCen91 USCen15 EUAma07 USCen01RKZ USCen01
1st stage F-statistic 276.83 227.40 159.62 35.52 6.94 50.87
R-squared 0.662 0.662 0.659 0.584 0.628 0.554
FE country year controls
Observations 11,675 11,671 11,031 11,743 11,710 11,675
Number of clusters 140 140 140 140 140 140
Sources: World Bank Enterprise Surveys 2003-2015 and US ﬁrm census 2001.
Robust standard errors in parentheses, clustered at the country level.
Table A1 shows ﬁrst stage regressions and IV regressions on tax inspections across different samples of
ﬁrst stage industry ﬁrm-size. All regressions include interactive ﬁxed effects between country, year and
3 dummies indicating whether the ISIC3 industry was above median in the country-year distributions of
age, export-share, and foreign-industry share. In Panel A row 1-3, US industry size is constructed using
the 2001, 1991, and 2015 US Census of ﬁrm employment. In Panel A row 4, the industry size is constructed
using the Amadeus sample in the UK and Germany between 1998 and 2007. In Panel A row 5, the industry
average ﬁrm-size is constructed using the Kumar et al. [1999] ’employee-weighted’ methodology (KRZ).
This methodology effectively places more weight on ﬁrms that concentrate large shares of total employment
when calculating the industry-wide average ﬁrm size. In Panel A row 6, the industry size mean (as opposed
to rank) in the US Census 2001 is used as the regressor. In Panel A Cols.1-4, the outcome variable is the
WBES size rank as deﬁned in the main tables. In Panel A Column 5, the outcome variable is the average ﬁrm
size in all WBES countries calculated using the employee-weighted KRZ methodology. In Panel A Column
6, the outcome variable is the industry mean ﬁrm size (as opposed to rank). In Panel B, the outcome variable
is a dummy taking a value of one if the ﬁrm has had any tax inspections over the past year: please see Table
1 for more details. In Panel B, Columns 1-5, the WBES industry size rank is instrumented for using a size
rank constructed from different samples, respectively: US Census 2001, US Census 1991, US Census 2015,
EU Amadeus, US Census 2001 RKZ reweigthed. In Panel B Column 6, the WBES industry size mean is
instrumented for using the US Census 2001 industry size mean.
57
TABLE A2: I MPACT OF F IRM S IZE ON TAX I NSPECTION , R ESTRICTED TO M ANUFAC -
TURING
Outcome: Industry Size Rank (WBES) Tax Inspection {0,1} # Inspections
(1) (2) (3) (4) (5)
1st Stage Red. Form IV Red. Form IV
Industry Size Rank (US Census) 0.310 0.135 0.009
(0.029) (0.034) (0.002)
Industry Size Rank (WBES) 0.434 0.0289
(0.089) (0.005)
1st stage F-statistic 114.36 114.36
FE country year controls
R-squared 0.667 0.685 0.688 0.614 0.617
Observations 6,726 6,726 6,726 6,726 6,726
Number of clusters 138 138 138 138 138
Sources: World Bank Enterprise Surveys 2003-2015 and US ﬁrm Census 2001.
Robust standard errors in parentheses, clustered at the country level.
Table A2 shows the coefﬁcients of regressions of industry ﬁrm-size ranking on various outcomes, limiting
the sample to manufacturing industries. All regressions include interactive ﬁxed effects between country,
year and 3 dummies indicating whether the ISIC3 industry was above median in the country-year dis-
tributions of age, export-share, and foreign-industry share. Manufacturing is deﬁned as the set of ISIC2
industries which lie between code-number 15 and 37, using the ISIC Classiﬁcation version 3.1. Size rank
is constructed similarly to the main tables: please see Table 1 for further details. In Column 1, the out-
come is WBES industry size rank. In Columns 2-3, the outcome variable is constructed similarly to Table
1. In Columns 4-5, the outcome variable is constructed similarly to Table 2. In Columns 3 and 5, the WBES
industry size rank is instrumented for using the US industry size rank.
58
TABLE A3: TAX INSPECTION ON F IRM S IZE AND C APITAL N EEDS
Outcome: Tax Inspection {0,1} # Inspections
(1) (2) (3) (4)
Industry Size Rank (US Census) 0.063 0.090 0.003 0.006
(0.020) (0.032) (0.001) (0.002)
Industry External Reliance Rank (US Compustat) -0.001 -0.001
(0.023) (0.001)
Industry Capital-Labor Ratio Rank (US Compustat) -0.001 0.002
(0.027) (0.002)
FE countryyearcontrols
R-squared 0.580 0.624 0.534 0.562
Observations 9,715 5,320 9,715 5,320
Number of clusters 131 106 131 106
Sources: World Bank Enterprise Surveys 2003-2015 and US ﬁrm census 2001.
Robust standard errors in parentheses, clustered at the country level.
Table A3 shows the coefﬁcients of regressions of industry size ranking and industry capital ranking on
various outcomes. All regressions include interactive ﬁxed effects between country, year and three dum-
mies indicating whether the ISIC3 industry was above median in the country-year distributions of age,
export-share, and foreign-industry share. Columns 1 and 3 include the industry external reliance rank as
an additional explanatory variable. External reliance is deﬁned as the share of capital expenditure not ﬁ-
nanced by cash ﬂow of operations. This measure can directly be constructed for a subsample of ﬁrms in
WBES. In the US, we rely on Compustat ﬁrm-level data between 1997 and 2007. We follow the methodol-
ogy of Rajan and Zingales [1998] to construct the ISIC3 measure of external reliance. Having constructed
the external reliance at the ﬁrm level in both the US and WBES, we construct ISIC3-country-year rankings
similar to the industry employee-size rankings. Columns 2 and 4 include the industry capital-labor ratio
rank as an additional explanatory variable. The capital-labor ratio is deﬁned as the ratio of assets to number
of permanent employees. This measure can be constructed for a subsample of ﬁrms in WBES. In the US,
we rely on Compustat ﬁrm-level data between 1997 and 2007. The drop in sample size in columns 2 and 4
is due to the missing observations on assets for ﬁrms in the WBES surveys.
59
TABLE A4: I MPACT OF F IRM S IZE ON I NFORMAL TAX PAYMENTS
OLS Reduced Form IV Panel: OLS
(1) (2) (3) (4) (5) (6) (7) (8)
Panel A
Outcome: Any Informal Payment? {0,1}
Industry Size Rank (World Bank) 0.011 -0.024 -0.048 -0.019 -0.041 -0.046
(0.015) (0.017) (0.032) (0.073) (0.041) (0.058)
Industry Size Rank (US Census) -0.015 -0.008
(0.010) (0.017)
FE country year controls
FE country year ISIC2
FE ISIC3
F-stat 184.39 116.58
R-squared 0.745 0.825 0.745 0.825 0.744 0.896 0.912 0.924
Panel B
Outcome: % Sales Paid Informally
Industry Size Rank (World Bank) -0.0036 -0.0072 -0.0114 -0.0135 -0.0198 -0.0221
(0.0024) (0.0025) (0.0047) (0.0089) (0.0059) (0.0075)
Industry Size Rank (US Census) -0.0036 -0.0036
(0.0015) (0.0024)
FE country year controls
FE country year ISIC2
FE ISIC3
F-stat 184.39 116.58
R-squared 0.466 0.604 0.466 0.604 0.464 0.604 0.778 0.777
Observations 9,538 7,261 9,538 7,261 9,538 7,261 4,381 3,390
Number of clusters 137 137 137 137 137 137 74 78
Source: US Census 2001 and World Bank Enterprise Surveys 2003-2015.
Robust standard errors in parentheses, clustered at the country level.
Table A4 shows the coefﬁcients of regressions of industry ﬁrm-size ranking on informal tax payments, dis-
cussed in section ??. Informal tax payments are measured as the answer to the question: "We’ve heard
that establishments are sometimes required to make gifts or informal payments to public ofﬁcials to ‘get
things done’ with regards to customs, taxes, licenses, regulations, services,etc. On average, what percent of
total annual sales do establishments like this one pay in informal payments or gifts to public ofﬁcials for
tihs purpose?". Panel A deﬁnes informal tax payment as a dummy equal to one if a ﬁrm reports paying
any strictly positive amount of informal payments. Panel B uses the percentage of sales paid in informal
payments. World Bank size rank is the ISIC 3-digit ranking of number of permanent employees per ﬁrm in
the WBES country-year. US Census size rank is the ISIC 3-digit rank of number of employees per establish-
ment in 2001, where we exclude establishments with less than 5 employees. Odd number columns include
interactive ﬁxed effects between country, year and three dummies indicating whether the ISIC3 industry
was above median in the country-year distributions of age, export-share, and foreign-industry share. Even
number columns include interactive ﬁxed effects between country, year and 2-digit ISIC codes. For a given
speciﬁcation, the sample size in the even columns is smaller than in the odd columns. This is because the
ISIC2 interactive ﬁxed effect speciﬁcation drops all cells where there does not exist more than one ISIC3-
country-year observation within a ISIC2-country-year observation. The sample drop occurs in industries
that are less specialized, which are more prevalent in less developed countries. In columns (5) & (6), we
instrument for the World Bank size rank using the US Census size-rank. The F-statistic comes from the ﬁrst
stage rank-rank regression of World Bank size rank on US Census size rank. In columns (7) & (8), we add
ISIC 3-digit ﬁxed effects, so that the size-rank coefﬁcients are estimated using variation within ISIC3 over
time in its country-year industry rank. The ISIC3 ﬁxed effect models result in a drop in sample size because
some countries do not have repeated surveys.
60
TABLE A5: S IZE - GRADIENTS INTERACTION WITH STATUTORY TAX RATES
(1) (2) (3) (4)
Outcome: Tax Inspection {0,1} # Inspections Tax Compliance Informal Tax
Industry Size Rank (US Census) 0.041 0.003 -0.003 -0.003
(0.024) (0.002) (0.053) (0.002)
Industry Size Rank1(High Tax Rate) 0.061 0.003 0.129 -0.002
(0.030) (0.002) (0.062) (0.003)
FE countryyearcontrols
F-test joint signiﬁcance [Rank + Rank1(High Rate)] 32.66 57.08 4.43 14.93
P-value 0.0000 0.0000 0.0373 0.0002
R-squared 0.649 0.579 0.712 0.468
Observations 11,103 11,103 2,920 9,033
Number of clusters 127 127 125 74
Sources: World Bank Enterprise Surveys 2003-2015 and US ﬁrm Census 2001.
Table A5 shows the coefﬁcients of regressions of industry ﬁrm-size ranking, and its interaction with statu-
tory tax rate, on tax inspection, formal compliance, and informal tax payments. US Census size rank is
the ISIC 3-digit rank of number of employees per establishment in 2001, where we exclude establishments
with less than 5 employees. All regressions include interactive ﬁxed effects between country, year and 3
dummies indicating whether the ISIC3 industry was above median in the country-year distributions of
age, export-share, and foreign-industry share. In addition, all regressions include an interaction between
the Census size rank and a country-year varying dummy which takes a value of 1 if the WBES country-year
statutory rate was above the median in the WBES sample distribution. The statutory rate is constructed as
the sum of the corporate tax rate and the indirect tax rate. The F-test value and p-value report the results
from an F-test on the joint signiﬁcance of the industry size rank coefﬁcient and the interaction coefﬁcient
between size rank and the high tax rate dummy.
61
TABLE A6: C ORRELATION OF SIZE WITH PERFORMANCE
(1) (2) (3) (4) (5) (6) (7)
Outcome: Log Sales Cost per Employee Labor regulations Intl. Certiﬁcate Part of Larger Firm Formal Training Finance Constraints
Industry Size Rank (US Census) 0.011 0.011 0.002 0.003 0.001 0.001 -0.001
(0.001) (0.002) (0.000) (0.000) (0.000) (0.000) (0.000)
FE countryyearcontrols
R-squared 0.892 0.896 0.516 0.456 0.580 0.624 0.213
Observations 9,415 9,175 8,160 9,544 9,553 9,691 9,705
Number of clusters 131 130 129 131 131 130 131
Sources: World Bank Enterprise Surveys 2003-2015 and US ﬁrm census 2001.
Robust standard errors in parentheses, clustered at the country-year level. p<0.01, p<0.05, p<0.1
Table A6 shows the coefﬁcients of regressions of industry size ranking on various outcomes. All regressions
include interactive ﬁxed effects between country, year and 3 dummies indicating whether the ISIC3 indus-
try was above median in the country-year distributions of age, export-share, and foreign-industry share.
Industry size rank is constructed similarly to the main tables: please see Table 1 for further details. In Col-
umn 1, the outcome is the natural logarithm of annual ﬁrm sales. In Column 2, the outcome is the ratio of
annual full-time permanent employee labor costs to the total number of full-time permanent employees. In
Column 3, the outcome variable is the response to the answer "are labor regulations no/minor/major/very
severe obstacle to the current operations of this establishment?" (higher value meaning more of an obsta-
cle). In Column 4, the outcome variable is a dummy taking value one if the establishment reports having
an internationally-recognized quality certiﬁcation. In Column 5, the outcome variable is a dummy taking
value one if the establishment is part of a larger ﬁrm. In Column 6, the outcome variable is a dummy
taking value one if the establishment reports having formal training courses for its permanent employ-
ees. In Column 7, the outcome variable is the response to the answer "is access to ﬁnance (availability and
cost) no/minor/major/very severe obstacle to the current operations of this establishment?" (higher value
meaning more of an obstacle).
62
TABLE A7: I MPACT OF F IRM S IZE ON F ORMAL AND T OTAL E FFECTIVE TAX R ATES
OLS Reduced Form IV
(1) (2) (3) (4) (5) (6)
Panel A
Outcome: Formal Effective Tax Rate
= Tax Rate Share of Sales Reported
Industry Size Rank (World Bank) 0.065 0.069 0.112 0.100
(0.015) (0.019) (0.035) (0.045)
Industry Size Rank (US Census) 0.038 0.034
(0.015) (0.017)
FE country year controls
FE country year ISIC2
F-Stat 33.99 38.73
R-squared 0.719 0.795 0.715 0.793 0.718 0.795
Observations 2,920 2,149 2,920 2,149 2,920 2,149
Number of clusters 74 74 74 74 74 74
Panel B
Outcome: Total Effective Tax Rate
= Sum of Formal and Informal
Industry Size Rank (World Bank) 0.064 0.069 0.033 0.024
(0.023) (0.022) (0.059) (0.075)
Industry Size Rank (US Census) 0.011 0.008
(0.021) (0.026)
FE countryyearcontrols
FE countryyearISIC2
F-stat 33.38 38.73
R-squared 0.753 0.811 0.750 0.809 0.752 0.810
Observations 2,731 1,970 2,731 1,970 2,731 1,970
Number of clusters 74 74 74 74 74 74
Source: 2001 US Census and World Bank Enterprise Surveys 2003-2007 and tax rates collected by the au-
thors based on the KPMG’s corporate and indirect tax rates tables.
Robust standard errors in parentheses, clustered at the country level.
Table A7 shows the coefﬁcients of regressions of industry ﬁrm-size ranking on the formal effective tax rate
(Panel A) and the total effective tax rate (Panel B), discussed in Section ??. In Panel A, the effective formal
tax rate is calculated as the product of the (ﬁrm-speciﬁc) share of sales reported for tax purpose (outcome
in Table 3) and the (country-yea speciﬁc) sum of the indirect and corporate tax rates. In Panel B, the total ef-
fective tax rate is calculated, at the ﬁrm level, as the sum of formal tax payments and informal tax payments
(outcome in Table A4) as a share of a ﬁrms’ sales. Odd number columns include interactive ﬁxed effects
between country, year and 3 dummies indicating whether the ISIC3 industry was above median in the
country-year distributions of age, export-share, and foreign-industry share. Even number columns include
interactive ﬁxed effects between country, year and 2-digit ISIC codes. In this case, size-rank coefﬁcients
are estimated using variation between 3-digit ISIC codes, controlling non-parametrically for country-year-
2digit ISIC effects. For a given speciﬁcation, the sample size in the even columns is smaller than in the odd
columns. This is because the ISIC2 ﬁxed effect speciﬁcation drops all cells where there does not exist more
than one ISIC3-country-year observation within a ISIC2-country-year observation. In columns (5) & (6), we
instrument for the World Bank size rank using the US Census size-rank. The F-statistic comes from the ﬁrst
stage rank-rank regression of World Bank size rank on US Census size rank. The question on formal tax
payments was only administered in the earlier waves of the WBES surveys which were not structured as
panels. Hence, we can not estimate panel models for the formal and total effective tax rate.
63