Financial VAT May Improve Trade Openness

Abstract

This paper theoretically and empirically analyzes the influence on the rate of trade openness of the taxation of financial services under VAT. The empirical analysis is carried out using data from the OECD and 36 European Union countries for the period 1961-2012. Dynamic panel data techniques are used, concretely the GMM System, and an unbalanced panel is handled. The results corroborate that financial VAT, and in particular the “option-to-tax” method applied by some countries in the European Union, have a positive impact on a country’s rate of trade openness.

Full text

In
International Center for Public Policy
Working Paper 20-08
June 2020
Financial VAT May Improve Trade Openness
Julio López-Laborda
Guillermo Peña
INTERNATIONAL
CENTER FOR
PUBLIC POLICY

International Center for Public Policy
Working Paper 20-08
Financial VAT May Improve Trade Openness
Julio López-Laborda
Guillermo Peña
June
2020
International Center for Public Policy
Andrew Young School of Policy Studies
Georgia State University
Atlanta, Georgia 30303
United States of America
Phone: (404) 413-0235
Fax: (404) 651-4449
Email: paulbenson@gsu.edu
Website: http://icepp.gsu.edu/
Copyright 2020, the Andrew Young School of Policy Studies, Georgia State University.
No part of the material protected by this copyright notice may be reproduced or utilized
in any form or by any means without prior written permission from the copyright owner.

International Center for Public Policy
Andrew Young School of Policy Studies
The Andrew Young School of Policy Studies was established at Georgia State University with
the objective of promoting excellence in the design, implementation, and evaluation of public
policy. In addition to four academic departments, including economics and public
administration, the Andrew Young School houses eight leading research centers and policy
programs, including the International Center for Public Policy.
The mission of the International Center for Public Policy (ICePP) at the Andrew Young School
of Policy Studies is to provide academic and professional training, applied research, and
technical assistance in support of sound public policy and sustainable economic growth in
developing and transitional economies.
ICePP is recognized worldwide for its efforts in support of economic and public policy reforms
through technical assistance and training around the world. This reputation has been built
serving a diverse client base, including the World Bank, the U.S. Agency for International
Development (USAID), the United Nations Development Programme (UNDP), finance
ministries, government organizations, legislative bodies, and private sector institutions.
The success of ICePP reflects the breadth and depth of its in-house technical expertise. The
Andrew Young School's faculty are leading experts in economics and public policy and have
authored books, published in major academic and technical journals, and have extensive
experience in designing and implementing technical assistance and training programs.
Andrew Young School faculty have been active in policy reform in over 40 countries around
the world. Our technical assistance strategy is not merely to provide technical prescriptions
for policy reform, but to engage in a collaborative effort with host governments and donor
agencies to identify and analyze the issues at hand, arrive at policy solutions, and implement
reforms.
ICePP specializes in four broad policy areas:
 Fiscal policy (e.g., tax reforms, public expenditure reviews)
 Fiscal decentralization (e.g., reform, intergovernmental transfer systems, urban finance)
 Budgeting and fiscal management (e.g., local, performance-based, capital, and multi-
year budgeting)
 Economic analysis and revenue forecasting (e.g., micro-simulation, time series
forecasting)
For more information about our technical assistance activities and training programs, please
visit our website at icepp.gsu.edu or contact us at paulbenson@gsu.edu.

Financial VAT May Improve Trade Openness
Julio López-Laborda and Guillermo Peña1
June 2020
Abstract
This paper theoretically and empirically analyzes the influence on the rate of
trade openness of the taxation of financial services under VAT. The empirical
analysis is carried out using data from the OECD and 36 European Union
countries for the period 1961-2012. Dynamic panel data techniques are used,
concretely the GMM System, and an unbalanced panel is handled. The results
corroborate that financial VAT, and in particular the “option-to-tax” method
applied by some countries in the European Union, have a positive impact on a
country’s rate of trade openness.
Key words: financial VAT, trade openness, panel data
JEL classification: H25, H21, E62, F14
Keywords: municipal finance, revenue capacity, expenditure needs, fiscal gap
1 Department of Public Economics, University of Zaragoza, Spain
Acknowledgments: The authors are grateful for the funding received from the Regional Government of Aragon
European Social Fund and the European Regional Development Fund (Public Economics Research Group).

1
1. Introduction
Although in some countries financial services are levied with indirect taxes, in most countries
these services are exempt from VAT. This exemption has several consequences for the economy.
Those most studied relate to efficiency, as the exemption causes several distortions, mainly due
to irrecoverable VAT for businesses.1 Furthermore, there are other distortions. As far as equity is
concerned, and to the extent that financial services are consumed to a greater proportion by
wealthier individuals, the exemption increases inequality in income distribution (Huizinga, 2002;
López-Laborda and Peña, 2017a). Tax revenue is also affected by the exemption, with no
consensus among scholars concerning the impact. The results of a hypothetical VAT collection
on financial services range from 6 to 15 billion euros for Europe (Huizinga, 2002; European
Commission, 2011; Lockwood, 2011).
One way the exemption affects efficiency is its impact on trade openness, because financial
services are under-taxed services for households and, as a kind of non-traded good, this under-
taxation discourages the consumption of correctly-taxed goods, as is the case of all traded goods,
which are taxed by the general VAT rate. Therefore, removing the VAT exemption of financial
services and taxing them at a positive tax rate could reduce disincentives for traded services, and
therefore, trade openness would increase. The aim of this paper is to empirically test whether
applying indirect taxes to financial services, and applying VAT in particular, positively affects
the rate of trade openness.
1 For a detailed review of the economic distortions of the exemption of financial services on VAT, see López-Laborda
and Peña (2018).

2
The literature has proposed several methods for taxing financial services, some of which are
currently applied in international practice. Table 1 shows the main methods applied around the
world in different countries.
The “zero rate” method consists of establishing a VAT rate of 0 percent on financial services,
allowing financial institutions to claim input VAT. The “exemption with partial input credits”
method, also known as “partial income recovery”, is the intermediate method between exemption
and zero-rating, where a percentage of the input VAT is allowed for crediting. In the method of
“taxation of fees and commissions”, there is a mandatory taxation of all the explicit fees and
charges of financial services and a recoverable input VAT. The “option-to-tax” method allows
financial entities the possibility of charging VAT on financial services or not, applying the tax on
the interest margin, and fees and commissions, or only on the latter. “Net operating income” and
“gross interest” methods take net operating income and gross lending interest, respectively, as
the tax basis for VAT. In the “addition” method, the tax is calculated by considering the sum of
wages, rents, interests, and net profits as the tax base. In the “subtraction” method, the tax base is
the difference between revenues and purchases, being both financial and non-financial. The
“separate taxes” method consists of a new type of tax on financial services distinct from general
VAT; a specific example is the Financial Activities Tax (FAT), which also includes aspects of
the addition method. These last five methods do not allow financial entities to credit their input
VAT.2
2 For a more in-depth description and analysis of the various methods, see López-Laborda and Peña (2017b, 2018).

3
Table 1. Methods of Taxing Financial Services Applied Around the World
Method
Countries where applied
Method
Countries where applied
Zero-rating Quebec (up to 2013), New Zealand (since 2005;
Merrill 2011),
Net
operating
income
Mexico (since 1992; Schatan 2003)
Exemption
with partial
input credits
Australia (since 2000; De la Feria and Walpole
2009), Singapore (since 1994; Jenkins and
Khadka 1998), Malaysia (since 2015; IMF
2015)
Subtraction
method
Italy (since 1998; Keen et al. 2010), proposed in
Japan to be established in 1950, but rejected (De la
Feria and Krever 2012), also proposed in Canada on
1987(Schenk 2009), and in the Philippines (Xu and
Krever 2016) proposed on 2000, but abandoned before
implementing
Taxing fee-
based
services
Australia, Singapore, South Africa (since 1996;
Merrill 2011), Malaysia, the Philippines (since
1988), India (since 1994; Deloitte 2013), China
(since 1994; Owens 2014), Korea (since 1982;
MSF 2012), Belgium (1971–1977; Ernst and
Young 2009), Slovenia (since March 2013, PKF
2014), Andorra (since 2013), Gahana (since
2015; PWC 2015), Mexico (since 1980; Schatan
2003), Thailand (since 1992; BOI 2016),
Taiwan
Separate
taxes
Quebec, Israel (since 1981; Gillis 1987), France (since
1968; Pons 2006), Denmark, Italy, Andorra (from
June 2002 to 2013, as a sales equalization tax;
ABA 2010), China (from 1994 (Owens 2014) up to
1 May 2016 (KPMG 2016)), India (since 1994
(Deloitte 2013), proposed under GST in 2016, but
postponed until 2017), the Philippines (since 1946;
except for the year 2003 when it was taxed under
VAT, ZGLO 2006), Taiwan (since 1 April 1986;
ROC 2016), Thailand, Iceland and Korea
Option to
tax
Option to tax only fees (partial taxation):
Belgium (since 1978), Lithuania (since 1 May
2004), France (since 1979)
Option to tax fees and margin (full taxation):
Austria (since 1997 with retroactive effect),
Estoniai (since 2002), Germany (since 1968)
Source: Ernst and Young (2009)
Addition
method
Quebec, Michigan (since 1953; De la Feria and
Krever 2012), France (since 1979; Pons 2006), Israel
(since 1976; Gillis 1987), Denmark (since 1988;
Møller and Hjerrild 2013)
Taxation of
gross
interest
Argentina (since 1992; Zee 2004). Proxy taxes
(Burns 2007): China (since 1994; Owens 2014),
on VAT since 1 May 2016; KPMG 2016), the
Philippines (since 1946; ZGLO 2006), Taiwan
(since April 1986, ROC 2016), Thailand and
Korea
Financial
Activities
Tax
Iceland (since 2012; Keen et al. 2016)
Source: López-Laborda and Peña (2017b).
As Guttmann and Richards (2006) state, the literature on the determinants of trade openness is
scarce, in spite of such seminal papers as Alesina and Wacziarg (1998). These authors include
geographical variables, the tax import ratio, the terms of trade, and public expenditure to explain
trade openness. Subsequently, there have been new contributions to the topic. Concretely, there
are advances in the study of geographical and commercial variables, such as Hau (1999), Alcalá
and Ciccone (2004), Guttmann and Richards (2006), Ram (2009) and Marjit et al. (2014). Other

4
authors incorporate financial depth as a determinant, such as Svaleryda and Vlachos (2002) or
Aizenman and Noy (2009). Finally, some authors study the influence of consumption or the
public sector size on the trade openness rate (Garen and Trask, 2005; Benarroch and Pandey,
2008, 2012; Benarroch and Pandey, 2012, and Jetter and Parmeter, 2015). Other papers also
study the cointegration of the dependent variable with other variables, such as energy
consumption (Nasreen and Anwar, 2014), or look for determinants of other variables distinct
from but relating to trade openness, such as international competition (Chang et al., 2009).
The literature on the determinants of the rate of trade openness has not yet analyzed the impact
of financial VAT on trade openness. 3 In order to analyze this, we manage an unbalanced data
panel of 36 countries for the period 1961-2012. The selected countries are developed and
developing countries of the EU (27) and the OECD, with the exception of Switzerland, Cyprus,
Romania and Malta. Due to the temporal dependence of the data on the dependent variable (trade
openness rate), a dynamic panel data is estimated, following the Generalized Method of
Moments (GMM) in two steps.
The paper is divided into five sections. Section 2 develops the framework that theoretically
establishes the influence of financial VAT on the rate of trade openness. Section 3 proposes the
specification of the econometric model and the variables to be incorporated. Section 4 estimates
the econometric model that analyzes the impact of financial VAT on trade openness and
discusses the results. Our estimates suggest, first, that financial VAT, and in particular the
“option-to-tax” method applied by some countries in the European Union, has a positive impact
3 At present, as far as we know, the impact of financial VAT on any variable has not yet been studied with real data
and econometric techniques, except for the impact of this tax on the size of the financial sector (López-Laborda and
Peña, 2017c) and on income distribution (López-Laborda and Peña, 2017a).

5
on a country’s rate of trade openness, and second, that the “separate taxes” do not appear to have
a significant effect on trade openness.
2. Conceptual Framework
In this section, we propose a theoretical framework for analyzing the effects of the financial
VAT on trade openness, based on Feldstein and Krugman (1990), and assuming full pass-
through of VAT to prices (Benedek et al, 2019). We consider a country that produces and
consumes an exported good X, an imported good M, and a non-traded good N. The country is
assumed to be small on world goods markets, so that it can trade X for M at a fixed relative price.
The rate of trade openness of the country is defined as the sum of exports and imports over total
GDP.
The country applies a typical VAT, with tax refund on exports and taxation on imports, so that
imports and exports are both reduced in the same proportion by the application of the VAT,
which allows us to aggregate X and M into a composite traded good T. Non-traded good is
exempted from VAT. As we will see below, the exemption, which does not allow input VAT to
be credited, results in an under-taxation of the non-traded sector compared to the traded sector,
which encourages an increase in non-tradable consumption and production while reducing the
size of the trading sector, and therefore a decrease in the rate of trade openness would be
expected.
We can consider financial services to be more similar to a non-traded than to a traded service, as
those services are currently more often provided in physical branch offices than on the Internet.
As shown by Freund and Weinhold (2002), while the Internet has improved trading with many
services, this result is stronger when excluding some services as financial intermediation.
Nonetheless, this effect could currently be lessening, since the development of regulation in the

6
main trading conduits is reducing the trading costs of financial services (Miroudot, Sauvage, and
Shepherd, 2013). While some authors, as Krugman (1991, p. 65) states, consider that “[s]ome
services, however, especially in the financial sector, can be traded”, the literature has
traditionally considered them non-traded services (Benigno and Fornaro, 2014). Indeed, the
consumption of financial services reached 76.7 percent of the final demand for these services in
2015 in Spain, while exports only reached 23.3 percent.4 The export share of financial services is
significantly lower than that of traditionally traded products: textile products reach a share of
46.7 percent and motor vehicles, 65.3 percent.
Next, we analyze the expected differences in trade between the following five scenarios
concerning financial VAT, representative of the various methods applied in international practice
as shown in Table 1: exemption, zero-rate, separate taxes, option to tax, and the taxation of
financial services under VAT with a positive tax rate.
First, the inefficiency derived from the exemption is analyzed. Considering are the prices of
the traded goods in the country,are the prices of the non-traded goods (financial services) in
this country,
G
t
is the general VAT rate,
f
t
is the tax rate or the VAT applied to financial
services, and is the percentage of the traded goods that are used as input in non-traded
goods, and assuming the tax collection is higher than the irrecoverable VAT; we can represent
the relative price of traded to non-traded goods as follows:
(1)
4 Data from the Input-Output Table for basic prices from the Spanish National Institute for Statistics (INE, 2015).
Accessed on 23 April 2020.

7
This expression shows that the exemption (identified by sub-index e) reduces the size of the
tradable sector,, by increasing its price relative to non-traded goods, with respect to the
general method of taxation in VAT, . The next expression compares the exemption and “zero-
rate” cases:
(2)
If “zero-rate” is applied to financial services, non-traded goods are not taxed, but the input VAT
is refunded. Therefore, the tax levy and the price of these services is lower than in a case where
the non-traded sector cannot deduct input VAT, which is the case for the exemption. Hence, the
“zero-rate” method would further increase the price of traded goods relative to non-traded ones,
discouraging traded goods, , more than the exemption method.
We now compare the exemption method and “separate taxation”. The latter applies a positive tax
rate on financial services but, as it is a different tax from VAT, the VAT chain is also broken as
in the exemption method. The relative price of traded to non-traded goods is then as follows:
(3)
So, the tradable sector is encouraged with “separate taxes”, , compared to the exemption or
“zero-rate” methods.
The fourth method we will discuss is financial VAT with positive tax rate:
(4)

8
Assuming that the irrecoverable input VAT is lower than the collected financial VAT, we can
see that the relative price is lower with the full financial VAT method than with the exemption
method and therefore the tradable sector will be incentivized in the first case, , compared to
the exemption method.
It is worth noting that, if
Gf
tt=
, then . In this case, the
following will be fulfilled:
(5)
Consequently, the method of separate taxation could encourage an inefficiently high tradable
sector.
And finally, the “option-to-tax” method is considered. This method allows financial entities to
opt between the exemption and the taxation of financial services in VAT with a positive tax rate
and the deduction of input VAT:
(6)
Where is the proportion of businesses from country A that opt to tax. The aggregate
results of this method are between those of full taxation (for , expression 6 equates
expression 4) and those of the exemption method (for , expression 6 equates expression 3).
In short, if financial services are considered as non-traded services, the results summarized in (5)
suggest that financial VAT can enhance the size of the tradable sector, at the expense of the non-
tradable sector, and hence increase the level of trade openness. According to our theoretical
analysis, the most suitable methods seem to be full taxation and “option-to-tax”.

9
3. Specifications
According to the theoretical results obtained in the previous section, our objective is to
empirically test whether VAT on financial services positively affects the degree of trade
openness of countries.
Like Chang et al. (2009) and Marjit et al. (2014), we will estimate a model that applies the
System GMM method for dynamic panel data (Arellano and Bover, 1995, and Blundell and
Bond, 1998). The specification is as follows:
(7)
1
*
it it T it t it
openness openness T a c
γ ββ ε
−
= + + ++ +x
Where
it
openness
is a variable that reflects, in percentage, the trade openness rate of the country
i in year t. This variable is the sum of exports and imports of goods, divided by the value of
GDP, in US dollars and current prices. The variable
1
it
openness
−
is the first lag of the
endogenous variable and
γ
is its coefficient. T is our vector of interest variables (if financial
services are subject to VAT, and, if so, the method and the tax rate applied),
it
x
are the control
variables,
β
are the coefficients,
a
is the constant,
t
c
is the trend, and
it
ε
is the disturbance
term.
Two complementary specifications are formulated, which differ in the variables of interest they
incorporate. The first specification uses fvat *fr and separate*fr as interest variables. The first is
the interaction of fvat, a binary variable taking the value 1 if financial services are subject to
VAT according to Table 1 (excluding FAT and separate taxes), and 0 otherwise; with fr, the
financial services tax rate applied as a percentage. The second is the interaction of fr with
separate, a binary variable taking the value 1 if financial services are subject to a separate tax,
and 0 otherwise. As seen in Section 2, a country with financial VAT would have a higher rate of

10
trade openness than with the exemption, because it would avoid discouraging traded goods
compared to non-traded goods such as financial services. In addition, the expected effect will be
greater as the financial VAT rate approaches the general VAT rate.
In the second specification, we focus on determining the effect on the ratio of trade openness of
the financial VAT method most used by the countries in the sample, which is the “option-to-tax”
method established by the European Union (EU). Article 137(1)(a) of the VAT Directive
currently in force allows EU Member States to introduce an option-to-tax financial services. In
the EU, the exemption is generally applied, but since 1978 several countries, such as Austria,
Belgium, Estonia, France, Germany and Lithuania, have successively introduced the option-to-
tax system. The “option-to-tax” method allows financial entities to opt to levy VAT on financial
services. If an entity decides not to levy VAT, the exemption is applied. If it opts to tax, then
financial services are subject to VAT using the VAT method chosen by the country where the
financial services provider is established. In this way, each financial entity chooses the most
profitable option depending on the volume of input VAT that the company incurs. This method
is designed for financial entities that provide services to businesses and apply for a large amount
of deductible input VAT (López-Laborda and Peña, 2017b).
This specification uses O2T*fr, alter*fr and separate*fr as variables of interest. The first is the
interaction of the financial tax rate, fr, and O2T, a dummy variable that reflects whether a
country applies (value 1) or not (value 0) the “option-to-tax” method. The second variable of
interest is the interaction of fr with alter, a dummy that reflects whether a country applies (value
1) or not (value 0) financial VAT with a method other than the “option-to-tax”.5
5 If any data is zero in both fVAT and separate variables, it means that either the country exempted financial services
from VAT that year, or the country was not taxing VAT at that time (e.g. US). In the case of France, where financial

11
The following variables are used as controls in both specifications, according to the literature on
trade openness (see Table 2). The variables related with the demand side are gdppc, cpc and
electricity. The gdppc variable is the logarithm of GDP per capita, lagged one period to avoid
simultaneity and endogeneity problems. Per capita capital, incorporated through cpc variable, is
measured by gross investment divided by the country’s population, expressed in thousands of
millions of dollars and considering investment as the purchase of fixed assets plus net changes in
stock. Electricity production, net from energy losses occurred during transformation, distribution
and consumption, lagged one year, is measured in kW hour per capita by the electricity variable.
The following variables reflect public affairs. The size of the public sector is incorporated by the
psize variable, measured by public expenditure over GDP, considering public expenditure such
as government payments for operational activities for the provision of goods and services,
including workers’ remuneration (as wages and salaries), interests and subsidies, donations,
social benefits and other costs like income and dividends, according to the World Bank. A
country’s public surplus is controlled by the surplus variable, which is the percentage of surplus
over the total GDP. Finally, the experience variable reflects the total years since the entry into
force of VAT.
Institutional variables are also included in the specifications. Political stability is considered with
stability, a variable that captures the probability expectations of a destabilization of the
government. The second variable measures the gross secondary school enrolment rate, which is
the total number of secondary school students divided by the total number of persons of
secondary school age. Language is an indicator of institutional development, and measures the
VAT and a separate tax were in force simultaneously for a while, we have considered it as financial VAT during that
period.

12
presence of at least a significant minority of the population whose mother tongue is one of
Europe’s five main languages (English, French, German, Spanish and Russian). Infrastructures
are considered through the mobiles variable, which measures mobile phone lines per 100 people.
The models also include some geographical variables. The local variable takes the value 0 if the
country is an island, and otherwise, is equal to the result of dividing one by the number of
countries that have a common border.6 The population is incorporated by the population variable,
which is the de facto population estimated at the middle of the year. The density variable reflects
the de facto population divided by the surface area of the country. We also control by the area
variable, which is the size of a country measured by its area.
Finally, financial and trading variables are also incorporated. The size of the financial sector is
included through the fsize variable, which is the percentage of national private credit provided by
the financial sector over total GDP, lagged one year. Inflation is the rate of growth of the price
index. Financial openness is measured by fopenness, which is the sum of the capital and current
accounts of the balance of payments, with a lag of one year. The terms of trade adjustment, TOT,
is the level of import minus export of goods and services.
All variables have been obtained from the World Bank database (World Bank, 2018), with the
exceptions of language, local, experience, fr, fvat, O2T, alter and separate, all created by the
authors. The expected signs for the coefficients of each variable are shown in Table 2. We use
panel data, with information from the years 1961 to 2012 from 36 countries: all European Union
6 Other location variables have been used, such as the mean distance to France, USA and Japan, but we obtained worse
results in the estimates of the models.

13
countries (27) and the OECD, with the exceptions of Switzerland, Cyprus, Romania and Malta.
Table 3 contains the main descriptive statistics of the variables used in the estimates.
Table 2. Expected Signs
Variable
Exp.
sign
Literature Variable
Exp.
sign
Literature
fvat*fr (+)
Feldstein and Krugman
(1990)
second (+) Chang et al. (2009).
O2T*fr (+) By the authors language (+) Alcalá and Ciccone (2004)
alter*fr (+) By the authors mobiles (+) Chang et al. (2009)
separate*fr (+) By the authors local (+/-)
(+): Chang et al. (2009), (-):
Guttmann and Richards (2006)
gdppc (-) Guttman and Richards
(2004) pop (+/-)
(+): Ram (2009), (-): Alesina and
Wacziarg (1998), Alcalá and Ciccone
(2004), Guttman and Richards
(2004).
cpc (+) Marjit et al. (2014) density (+/-)
Ram (2009): theoretically (-),
empirically (+)
electricity (+) Nasreen and Anwar (2014) area (-)
Alesina and Wacziarg (1998), and
Guttman and Richards (2004).
psize (+)
Alesina and Wacziarg
(1998), Rodrik (1998),
Garen and Trask (2005) and
Ram (2009)
fsize (+) Chang et al. (2009)
surplus (+) Aizenman and Noy (2009) inflation (0/-)
Chang et al. (2009). Aizenman and
Noy (2009)
experience (+) Alesina and Wacziarg (1998) fopen (+) Aizenman and Noy (2009)
stability (+/-)
(+): Hau (1999), (-):
Aizenman and Noy (2009),
Marjit et al. (2014).
TOT (-) Camagni (2002)

14
Table 3. Descriptive Statistics
Variable
No
observations
Mean
Standard
Deviation
Minimum Maximum Skewness Kurtosis
openness (%)
1583
54.593
32.235
6.816
184.901
1.274
4.616
gdppc (ln)
1611
8.909
1.267
4.516
11.627
-0.395
2.561
psize (%)
1133
31.556
11.563
1.148
88.608
0.143
4.546
cpc ($)
1527
0.000
0.000
0.000
0.000
4.726
27.989
fopen ($)
252
-1.32E+10
1.15E+11
-8.09E+11
2.97E+11
-3.859
26.173
density(km2
pc)
1794
105.496
104.458
1.354
500.545
1.484
4.762
local (0-1)
1689
0.370
0.325
0
1
1.088
2.875
surplus (%)
575
-1.546
4.351
-29.420
20.010
0.397
9.273
stability
504
0.704
0.641
-1.623
1.668
-1.429
5.030
mobiles
1848
25.750
43.299
0
172.322
1.450
3.643
inflation (%)
1518
11.162
39.335
-4.480
1058.374
16.569
374.716
TOT ($)
1527
7.04E+11
6.09E+12
-3.58E+13
6.97E+13
7.307
74.094
fvat (0-1)
1872
0.115
0.319
0
1
2.415
6.838
O2T(0-1)
1872
0.080
0.271
0
1
3.098
10.604
alter (0-1)
1872
0.035
0.183
0
1
5.061
26.646
separate (0-
1)
1872
0.056
0.229
0
1
3.881
16.059
fr (%)
1829
2.6
6.3
0
25
2.237
6.409
4. Estimates and Results
To avoid problems of multicollinearity, the correlation matrix is analyzed and the VIF test is
applied and, as a consequence, the variables fsize, electricity, area, population, language,
experience and second are initially eliminated. Next, due to the high temporal period, unit root
tests are applied to the dependent variable, in particular Im–Pesaran–Shin and Phillips–Perron.
These tests indicate a problem of unit root. Therefore, a time trend has been incorporated into the
model (Phillips and Perron, 1988). The GMM System is applied to the two specifications

15
formulated in Section 3, and the methodology is the same for estimating both models. First, each
model is estimated taking all non-correlated variables into account. Once this is done, the Sargan
test (over-identification of the instruments) and Arellano and Bond test (non-autocorrelation of
residues) are applied. The Sargan test assumes the validity of the applied instruments as null
hypothesis. In this first step, no good econometric properties are obtained, so better models are
needed.
Second, we sequentially eliminate the non-significant variables from previous models until we
obtain estimates in which the validity of the instruments and non-autocorrelation of the residues
are corroborated. In these resulting models, the residuals are obtained by a WC-robust estimator
derived by Windmeijer (2005), which is a robust and bias-corrected estimator for two-step VCEs
(variance-covariance matrix estimators). This gives final Models I and II shown in Table 4.
A positive and statistically significant impact in trade openness is obtained for the coefficients of
financial VAT (fvat*fr, Model I) and of taxation through the “option-to-tax” method (O2T*fr,
Model II), as predicted by our theoretical framework. A 1-percent increase in the financial rate of
VAT in a country increases the short-term rate of trade openness by 2.57 percent, while a 1-
percent increase in the financial rate through the application of the “option-to-tax” method raises
the trade openness rate by 2.74 percent in the short term. However, the taxation of the financial
services by means of an out-of-VAT tax, which does not allow the full credit of input VAT, as
well as financial VAT types other than the “option-to-tax” method, do not seem to have any
significant influence on the rate of trade openness, as shown by the low significance of the
coefficient associated with these variables.

16
Table 4. Estimates Results
Dependent variable:
openness
Model I Model II
Explanatory variables
Coeff.
Std. e.
p-value
Coeff.
Std. e.
p-value
openness t-1
0.726
0.088
0
0.683
0.112
0
trend
1.854
0.57
0.001
1.801
0.549
0.001
gdppc
-15.420
5.24
0.003
-16.438
4.516
0
surplus
0.784
0. 342
0.022
0.684
0.397
0.085
stability
2.072
3.206
0.518
0.654
2.985
0.827
inflation
0.305
0.373
0.413
TOT -2.88E-13 1.32E-13 0.03
-3.67E-
13
1.10E-13
0.001
fvat*fr
2.572
1.273
0.043
O2T*fr
2.736
1.557
0.079
alter*fr
-0.704
2.649
0.79
separate*fr
-0.596
1.694
0.725
-0.75
1.779
0.673
constant
83.827
30.961
0.007
99.000
25.573
0
Sargan (p-value)
0.257
0.2
Arellano-Bond (p-
value 1st, 2nd Order)
0.000 0.608 0.000 0.728
No Observations
443
435
No Instruments
32
33
* 10%, **5% and *** 1% signification level. Std. e.: standard errors
As for the control variables, some of them have significant coefficients and with the expected
sign. The coefficient of the lagged dependent variable is positive and significant, consistent with
Marjit et al. (2014), and the model has good econometric properties, which confirms the
hypothesis developed at the beginning of section 4, so the dynamic character of the model is
corroborated. The coefficient of time trend is also significant, avoiding potential unit root

17
problems. The coefficient of the logarithm of the GDP per capita has a negative sign, as in
Guttman and Richards (2004), in contrast with other authors such as Chang et al. (2009) and
Ram (2009), who obtain positive coefficients. Guttman and Richards (2004) suggest that if trade
variables are incorporated, as is our case with the variable TOT, the trade openness relationship
with income is negative. They explain that, according to the literature, non-traded prices are
lower in developing countries, so, based on the assumption that all countries produce the same
proportion of traded and non-traded goods, the value of the non-traded goods would be lower in
developing than developed countries. Hence, incorporating trading variables, the GDP would be
negatively related with trade openness. Another explanation they provide is that the non-traded
sector is higher in developed countries. The variable relating to the public surplus, surplus, is a
variable of macroeconomic stability, and a positive correlation is obtained with the trade
openness rate, as in Aizenman and Noy (2009). Finally, the coefficient of the variable TOT has a
negative sign because, as Camagni (2002) states, the terms of trade have a negative correlation
with competitiveness. Nonetheless, the coefficient of the terms of trade variable does not appear
to be economically significant, due to its low magnitude. The coefficients of all these variables
have a significance greater than 5 percent, with the exception of the public surplus variable of
Model II, with a significance greater than 10 percent.
In our model, no significance is obtained for the coefficients of the geographical, financial and
institutional variables and the variables related with factor endowment or human capital.
The estimated coefficients inform us of the short-term effects of exogenous variables on the
endogenous variable. Long-term effects are calculated by dividing the former variables by one
minus the coefficient of the lag of the endogenous variable. The elast icities of the short- and
long-term effects of the significant variables of Models I and II are summarized in Table 5. This

18
table shows that the long-term effects are higher than the short-term effects in absolute terms for
all variables.
Table 5. Short- and Long-Term Effects
Model
I
II
Variable
Short run effect
Long run effect
Short run effect
Long run effect
gdppc
-15.420
-56.325
-16.438
-51.853
surplus
0.784
2.863
0.684
2.156
TOT
-2.88E-13
-1.05196E-12
-3.67E-13
-1.15767E-12
fVAT
2.572
9.393
O2T
2.736
8.631
For our interest variables, the long-term effect of financial VAT is 9.39 percent and that of the
impact of the O2T variable is 8.63 percent. We can therefore assert that financial VAT, and
specifically the “option-to-tax” method, seems to contribute to increasing the efficiency of the
economy, improving the degree of competitiveness of a country through its rate of trade
openness, in the short and especially in the long term.
5. Concluding Remarks
We have theoretically and empirically analyzed the effects on trade openness of levying VAT on
financial services. Theoretically, we expect financial VAT to reduce the price of traded goods
relative to the price of non-traded goods, allowing an increase in the tradable sector. The results
obtained in our empirical exercises suggest, first, that financial VAT, and in particular the
“option-to-tax” method applied by some countries in the European Union, has a positive impact
on a country’s rate of trade openness; and second, that the “separate taxes” do not appear to have
a significant effect on trade openness.

19
Therefore, eliminating the exemption and establishing financial VAT would benefit the
economy. The problem is how to apply the levy method. Many methods have been designed, but
they are either too simple and do not allow full taxation of the financial services, such as zero-
rate, or they produce distortions, such as the addition method, or they are theoretically accurate
but difficult to apply, such as the cash flow method with TCA (“tax calculation account”). For a
discussion of the methods, see López-Laborda and Peña (2017b).
At the mid-point of this trade-off between simplicity and accuracy, in López-Laborda and Peña
(2018) we developed a sufficiently precise but feasible method for taxing financial services
under VAT. This is the “mobile-ratio” method, which taxes the financial margin of each
company using a mobile-ratio approach. The tax base is constructed by applying the same ratio
to each interest transaction carried out by the company in a given period: e.g., each loan or
deposit interest. The ratio consists of the margin generated by financial services provided by the
company (i.e., the difference between interest receipts and interest payments) during the period
closest to the current one for which the information is available, divided by the total value of the
interests of the company (i.e., interest receipts plus interest payments) in that same period. The
VAT rate is then applied to the tax base. Under this method, VAT rate is also directly applied to
net explicit fees and commissions. Thus, all the financial value added provided by a company is
taxed. Furthermore, the mobile-ratio method is applied to financial services provided by financial
and non-financial entities in order to achieve neutrality.

20
References
AIZENMAN, J., and NOY, I. (2009): “Endogenous financial and trade openness”, Review of
Development Economics, 13(2): 175-189.
ALCALÁ, F. and CICCONE, A. (2004): “Trade and productivity”, Quarterly Journal of
Economics, 2: 613–646.
ALESINA, A. and WACZIARG, R. (1998): “Openness, country size and government”, Journal
of Public Economics, 69: 305–321.
ARELLANO, M. and BOND, S. (1991): “Some tests of specification for panel data: Monte
Carlo evidence and an application to employment equations”, Review of Economic
Studies 58: 277–297.
ARELLANO, M. and BOVER, O. (1995): “Another look at the instrumental variable estimation
of error-components models”, Journal of Econometrics 68: 29–51.
BENARROCH, M. and PANDEY, M. (2008): “Trade openness and government size”,
Economics Letters, 101: 157–159.
BENARROCH, M. and PANDEY, M. (2012): “The relationship between trade openness and
government size: does disaggregating government expenditure matter?”, Journal of
Macroeconomics, 34 (1): 239–252.
BENEDEK, D., DE MOOIJ, R.A., KEEN, M. and P. WINGENDER (2019): “Varieties of VAT
pass through”, International Tax and Public Finance, 1-14.
BENIGNO, G. and FORNARO, L. (2014): “The financial resource curse”, The Scandinavian
Journal of Economics, 116(1): 58-86.
BLUNDELL, R., and BOND, S. (1998): “Initial conditions and moment restrictions in dynamic
panel data models”, Journal of Econometrics, 87: 115–143.
BOI (2016): “A Business Guide to Thailand 2016”, Thailand Board of Investment
(www.boi.go.th/upload/content/BOI-BusinessGuide2016-20160318_87145.pdf).
BURNS, L. (2007): “The Taxation of Financial Services under a VAT in Developing Countries”,
p.13 (paper presented at the 4th annual Asian Tax Forum, Hanoi, Vietnam, 19-20 April
2007).
CAMAGNI, R. (2002): “On the Concept of Territorial Competitiveness: Sound or Misleading?”,
Urban Studies, 39(13): 2395–2411.
CHANG, R., KALTANI, L. and LOAYZA, N. V. (2009): “Openness can be good for growth:
The role of policy complementarities”, Journal of Development Economics, 90: 33–49.
DE LA FERIA, R. and KREVER, R. (2012): “Ending VAT Exemptions: Towards A Post-
Modern VAT”, Oxford University Centre for Business Taxation Working Papers,
WP12/28. (eureka.bodleian.ox.ac.uk/4387/1/WP1228.pdf).

21
DE LA FERIA, R. and WALPOLE, M. (2009): “Options for taxing financial supplies in Value
Added Tax: EU vat and Australian GST models compared”, International and
Comparative Law Quarterly, 58: 897–932.
(www.sbs.ox.ac.uk/sites/default/files/Business_Taxation/Docs/Publications/Working_Papers/Ser
ies_08/WP0818.pdf).
DELOITTE (2013): Service Tax in India. A Handbook.
(www.cefi.in/wp-content/uploads/2014/09/Deloitte_Service-Tax-in-India.pdf).
ERNST and YOUNG (2009): “Design and Impact of the “Option to Tax” System for application
of VAT to financial services”, Report prepared for the European Banking Federation,
October 28, 2009.
EUROPEAN COMMISSION (2011): “Impact Assessment Accompanying the Document
Proposal for a Council Directive on a Common System of Financial Transaction Tax and
Amending Directive 2008/7/EC”, SEC (2011) 1102 final, final Vol. 6 (Annex 5).
Brussels, Belgium.
FELDSTEIN, M. and KRUGMAN, P. R. (1990): “International Trade Effects of Value-Added
Taxation”, in E. Assaf Razin and J. Slemrod, eds., Taxation in the Global Economy (pp.
263-282), University of Chicago Press, Chicago.
FREUND, C., and WEINHOLD, D. (2002): “The Internet and international trade in services”,
American Economic Review, 92(2): 236-240.
GAREN, J. and TRASK, K. (2005): “Do more open economies have bigger governments?
Another look”, Journal of Development Economics, 77: 533–551.
GILLIS, M. (1987): “The VAT and Financial Services”, World Bank Departmental Working
Papers, DRD220.
(documents.worldbank.org/curated/en/514711468740445677/The-VAT-and-financial-services).
GUTTMANN, S., and RICHARDS, A. (2006): “Trade openness: an Australian perspective”,
Australian Economic Papers, 45(3): 188-203.
HAU, H. (2002): “Real Exchange Rate Volatility and Economic Openness: Theory and
Evidence”, Journal of Money, Credit and Banking, 34(3): 611-630.
HUIZINGA, H. (2002): “A European VAT on Financial Services?”, Economic Policy, 17(35):
497-534.
JENKINS, G. and KHADKA, R. (1998): “Tax Reform in Singapore”, International Tax
Program, (Vol. 644), Harvard Law School, Harvard Institute for International
Development, John F. Kennedy School of Government, Harvard University.
JETTER, M. and PARMETER, C. F. (2015): “Trade openness and bigger governments: The role
of country size revisited”, European Journal of Political Economy, 37: 49–63.
KEEN, M., KRELOVE, R. and NORREGARD, J. (2010): “The Financial Activities Tax”, in S.
Claessens, M. Keen and C. Pazarbasioglu, eds., Financial Sector Taxation: The IMF´s

22
Report to the G-20 and Background Material, (pp. 118-143) International Monetary
Fund. Washington D.C., United States.
KEEN, M., KRELOVE, R. and NORREGARD, J. (2016): “The Financial Activities Tax”,
Canadian Tax Journal, 64(2): 389- 400
KPMG (2016): “China’s new VAT rates & rules - high level policies and general impacts across
all industries”, China Tax Alert, 9.
(www.kpmg.com/CN/en/IssuesAndInsights/ArticlesPublications/Newsletters/ChinaAlerts/Docu
ments/china-tax-alert-09-vat-implementation-rules.pdf).
KRUGMAN, P. R. (1991): Geography and Trade, Cambridge, MA: MIT Press.
LOCKWOOD, B. (2011): “Estimates from national accounts data of the revenue effect of
imposing VAT on currently exempt sales of financial services companies in the EU,” in ,
PWC, eds., How the EU VAT exemptions impact the banking sector, Study to assess
whether banks enjoy a tax advantage as a result of the EU VAT exemption system
(www.pwc.com/gx/en/financial-services/pdf/2011-10-18_vat_study_final_report.pdf)
LÓPEZ-ALASCIO, J.M. (2013): “La opción por la tributación de las operaciones financieras en
el IVA”, Cuadernos de Formación, vol. 16/2013, col. 7/13, 125-136, Instituto de
Estudios Fiscales.
LÓPEZ-LABORDA, J., and PEÑA, G. (2017a): “Efectos sobre la redistribución de la renta del
gravamen de los servicios financieros en el IVA”, Working Paper 11, Grupo de
Investigación de Economía Pública, Universidad de Zaragoza.
(www.economiapublica.net/wp-content/uploads/2017/06/Redistribución.pdf)
LÓPEZ-LABORDA, J., and PEÑA, G. (2017b): “International Practices of Financial VAT”,
International VAT Monitor, 28 (6): 457-465.
LÓPEZ-LABORDA, J., and PEÑA, G. (2017c): “Does financial VAT affect the size of the
financial sector”, Economics: The Open-Access, Open-Assessment E-Journal, 11(2017-
7), 1-28.
LÓPEZ-LABORDA, J., and PEÑA, G. (2018): “A New Method for Applying VAT to Financial
Services”, National Tax Journal, 71 (1): 155-182.
MARJIT, S., MANDAL, B. and ROY, S. (2014): “Trade Openness, Corruption and Factor
Abundance: Evidence from a Dynamic Panel,” Review of Development Economics,
18(1): 45–58.
MIROUDOT, S., SAUVAGE, J. and SHEPHERD, B. (2013): “Measuring the cost of
international trade in services”, World Trade Review, 12(4): 719-735.
MERRILL, P.R. (2011): “VAT Treatment of the Financial Sector”, Tax Analysts, 163-185
(www.taxwire.org/www/freefiles.nsf/Files/MERRILL-13.pdf/$file/MERRILL-13.pdf).

23
MØLLER, H. and HJERRILD B. (2013): “Taxation of the Financial Sector in Denmark”,
Nordisk Skattevidenskabeligt Forskningsråd, spring, p.93-132 (www.djoef-
forlag.dk/sites/ntj/files/2013/2013_10.pdf).
MSF (2012): “Korean Taxation”, Ministry of Strategy and Finance, Korea.
(www.nts.go.kr/eng/data/KOREANTAXATION2012.pdf).
NASREEN, S. and ANWAR, S. (2014): “Causal relationship between trade openness, economic
growth and energy consumption: A panel data analysis of Asian countries”, Energy
Policy, 69: 82–91.
OWENS, J. (2014): “Aligning China’s Economic and Tax Systems–The Role of China’s1994
Tax Reform”, Manuscript, WU Global Tax Policy Center at the Institute for Austrian and
International Tax Law (Vienna),
(www.wu.ac.at/fileadmin/wu/d/i/taxlaw/institute/WU_Global_Tax_Policy_Center/46_Re
view_China_1994_Tax_Reform-20141026-NL.pdf).
PHILLIPS, P. C. and PERRON, P. (1988): “Testing for a unit root in time series regression”,
Biometrika, 75(2): 335-346.
PONS, T. (2006): “Option for Taxation of Financial Services in France”, International VAT
Monitor, International Bureau of Fiscal Documentation, May/June 2006, 182-185
(http://www.empcom.gov.in/WriteReadData/UserFiles/file/2006-33.pdf).
PWC (2015): “VAT on Financial Services,” On Point, Tax Services, Report of
PricewaterhouseCoopers LLP, Ghana, (http://www.pwc.co.za/en/assets/pdf/on-point-vat-
on-financial-services-3rd-edition.pdf).
RAM, R. (2009): “Openness, country size, and government size: additional evidence from a
large cross-country panel,” Journal of Public Economics, 93: 213–218.
ROC (2016): “Guide to ROC Taxes 2016”, Ministry of Finance Republic of China (Taiwan).
RODRIK, D. (1998): “Why do more open economies have bigger governments?” Journal of
Political Economy, 106(5): 997–1032.
SCHATAN, R. (2003). VAT on banking services: Mexico’s experience. International VAT
Monitor, 14(4): 287-294.
(http://www.empcom.gov.in/WriteReadData/UserFiles/file/2003-17.pdf).
SCHENK, A. (2009): “Taxation of Financial Services (Including Insurance) under a US Value
Added Tax.” Tax Law Review, 63: 409–442.
SVALERYDA, H. and VLACHOS, J. (2002): “Markets for risk and openness to trade: how are
they related?”, Journal of International Economics, 57: 369–395.
XU, Y., and KREVER, R. (2016): “Applying VAT to financial services in China: Opportunities
for China and lessons for the world?”, Australian Tax Review, 45(1): 38-49.

24
WINDMEIJER, F. (2005). “A finite sample correction for the variance of linear efficient two-
step GMM estimators”, Journal of Econometrics 126(1): 25–51,
(www.sciencedirect.com/science/article/pii/S0304407604000387).
WORLD BANK (2018): World Bank Open Data, Database
(data.worldbank.org).
ZEE, H. H. (2004): Taxing the Financial Sector: Concepts, Issues, and Practice, Washington,
DC, United States: International Monetary Fund.
ZGLO (2006): “Technical Report Policy Review of Financial Sector”, Zambrano & Gruba Law
Office Report, Prepared for the Taxation Capital Market Development Council Republic
of the Philippines, 30 June 2006. (pdf.usaid.gov/pdf_docs/Pnadh620.pdf).