Misreporting trade: Tariff evasion, corruption, and auditing standards

• Author: Derek Kellenberg,Arik Levinson
• Date: 01 February 2019
• Published date: 01 February 2019
• DOI: http://doi.org/10.1111/roie.12363

106

|

© 2018 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/roie Rev Int Econ . 2019;27:106–129.

1 | INTRODUCTION

News stories about misreported trade periodically surface in the business press.1 A recent UN report

claimed that a large share—67 percent in some cases—of exports goes unreported (UNCTAD, 2016).

Common themes in these stories include tariff and tax evasion, corruption, regulatory enforcement,

and firms and organizations spanning countries at different levels of economic development. In this

paper, we show that this type of trade misreporting is widespread and varies systematically not only

with tariffs, but also with country characteristics such as economic development, domestic taxes, cor-

ruption, and participation in regional trade agreements (RTAs).

The importance of tariff evasion has not gone unnoticed by economists. Bhagwati (1964) first

wrote about the phenomenon more than 50 years ago, showing that in Turkish trade data the gap

between reports by importers and exporters was correlated with tariffs and import controls. More re-

cently, Fisman and Wei (2004) found that the gap between China’s reported imports from Hong Kong

and Hong Kong’s reported exports to China was larger for industries facing higher Chinese tariffs.

Subsequent papers have examined exports from Germany to 10 transition economies, imports to India,

direct exports from China to the United States, trade between the United States and Canada, imports to

Received: 27 February 2018

|

Revised: 22 May 2018

|

Accepted: 27 May 2018

DOI: 10.1111/roie.12363

ORIGINAL ARTICLE

Misreporting trade: Tariff evasion, corruption, and

auditing standards

Derek Kellenberg1

|

Arik Levinson2,3

1Department of Economics, University of

Montana, Missoula, Montana, USA

2Department of Economics, Georgetown

University, Washington, DC, USA

3NBER, Cambridge, Massachusetts, USA

Correspondence

Derek Kellenberg, University of Montana, 32

Campus Dr., Missoula, MT 59812, USA.

Email: Derek.Kellenberg@mso.umt.edu

Abstract

Official international trade statistics report commerce be-

tween every pair of countries twice: once for the importing

country and once for the exporter. In principle, the two

values differ only by transport costs, but as has long been

recognized, they also differ systematically with product‐

level tariffs. We aggregate across products to construct a

dataset of annual aggregate bilateral trade, separately for

the importer and exporter reports. With these data, we

show that the reporting differences also vary systemati-

cally with country characteristics aside from tariffs: in-

comes, auditing standards, corruption, and trade

agreements.

|

107

KELLENBERG aNd LEVINSON

Kenya, Mauritius, and Nigeria, and most recently, Tanzanian imports from three developing country

trading partners.2 Although global assessments are scarce, one recent study estimated that in 2012,

more than U.S.$729 billion flowed out of developing countries as the result of trade mis‐invoicing and

tariff evasion (Kar & Spanjers, 2014).

Such reports mostly confirm the Fisman and Wei result that higher tariff rates lead to more tariff

evasion for the samples of countries studied. Like Fisman and Wei, they all identify evasion by com-

paring product‐level tariffs with product‐level trade data reported by exporters and importers between

select pairs of countries. As a result, even though the combined body of evidence suggests that higher

tariffs lead to more misreporting, the results are not generalizable beyond the specific sets of countries

examined. We do not really know the extent to which trade misreporting occurs across a broader set of

countries or goods, or whether it is more prevalent among rich or poor countries.

More importantly, the focus on tariff evasion disregards other country characteristics that may reward

or penalize firms for misreporting trade. Many of these, like income taxes, corruption, accounting stan-

dards, and capital controls, do not vary across industries within a country but do vary across countries

and over time. So their effects on misreported trade cannot be identified using isolated pairs of countries,

and disaggregated product level data provides little in return as national level variables do not vary at the

product level. A large panel of trade data and country characteristics, like the one we have assembled

for this project, is necessary to identify the effects of these country characteristics on misreported trade.

We begin by describing a model in which firms or countries choose how much to misreport their

imports or exports. Those misreports are functions of characteristics such as tariffs, corruption, taxes,

and the strength of auditing and accounting standards. We then use that model, along with a simple

accounting identity, to estimate the effects of those country characteristics on the gap between total

reported exports and imports among pairs of countries, using data on annual trade among 126 coun-

tries from 2002 to 2012.

Our approach using aggregate trade, as opposed to the industry‐by‐industry measures used by

Fisman and Wei (2004) and others, does come with a few trade‐offs. The aggregate approach cannot

identify trade misreporting that results from misclassifying products, say from high‐tariff to low‐tariff

categories, because those misreports are netted out in the aggregate.3 What we observe is pure mis-

reporting: cases where firms have reported to trade authorities a value different from the true value

transacted. In this sense, our analysis should be viewed as a conservative estimate of overall trade

misreporting.

This country‐level approach also measures tariffs using an average rate applied to all products by

each country, possibly introducing a degree of aggregation bias into the estimates of tariff evasion,

but our results are similar whether we use trade‐weighted average tariffs or simple averages of tariff

rates. More importantly, economists have been estimating tariff evasion using product‐level data for

many years. Our focus is elsewhere, on country characteristics that do not vary across products, have

not been studied before, and cannot be identified except at the country level.

In return, our aggregated approach has several advantages. First, by examining aggregate trade

among many countries, we are able to estimate the effects of policy‐relevant country characteris-

tics other than tariffs. These include auditing and accounting standards,4 corruption, participation in

RTAs, and other domestic tax rates. Second, by netting out misclassification through aggregation and

focusing on pure misreporting effects, we avoid difficult empirical challenges associated with disen-

tangling the two in industry‐level analyses. Third, when we do examine tariffs, our empirical strategy

allows us to estimate their average effect on misreporting across a wide variety of importing and

exporting countries. Finally, by aggregating across industries and using country variables as proxies

for misreports, we account for the potential endogeneity of true unobserved trade volumes and those

misreports.