Do Neighbours Influence Value‐Added‐Tax Introduction? A Spatial Duration Analysis
| Author | Pavel Čížek,Jenny E. Ligthart,Jinghua Lei |
| Date | 01 February 2017 |
| DOI | http://doi.org/10.1111/obes.12136 |
| Published date | 01 February 2017 |
25
©2016 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
Do Neighbours Influence Value-Added-Tax
Introduction? A Spatial Duration Analysis*
Pavel ˇ
C´iˇzek†, Jinghua Lei‡ and Jenny E. Ligthart†,§
†CentER, Department of Econometrics and OR, Tilburg University, Tilburg,
The Netherlands (e-mail: p.cizek@tilburguniversity.edu)
‡School of Finance, China Financial Policy Research Center, Renmin University of China,
Beijing, China (e-mail: j.lei@ruc.edu.cn)
Abstract
The spatial survival models typically impose frailties, which characterize unobserved
heterogeneity,to be spatially correlated. However,the spatial effect may not only exist in the
unobserved errors, but it can also be present in the baseline hazards and the dependent vari-
ables.A new spatial survival model with these three possible spatial correlation structures is
explored and used to investigate the implementation of value-added tax (VAT) in 99 coun-
tries over the period 1970–2009. Estimation is performed by a Bayesian approach through
the Markov chain Monte Carlo method. The estimation results suggest the presence of a
significant spatial correlation among the VAT introductions of neighbouring countries.
I. Introduction
The value-added tax (VAT), first introduced more than 50 years ago, remained confined to
a few countries until the late 1960s. However, after another 30 years, roughly 150 countries
have implemented a VAT, which on average raises about 25% of their tax revenue (Ebrill,
Keen and Bodin, 2001). The VAT is a tax on value added, which can be defined as the
value that a producer adds to his raw materials or purchases before selling the improved
product or service. Its invoice-credit mechanism – which seeks to tax the value added at
each stage of the production-distribution chain – causes it to fundamentally differ from a
retail sales tax or a turnover tax. Despite negative results concerning taxation of inputs,
conjectured efficiency benefits and possible administrative advantages1of the VAT have
led to its remarkable spread around the world, by being introduced or replacing other
JEL Classification numbers: C11, C23, C41, H20, H70
*The authors thank the IMF’sTax Policy Division for making availablethe data on revenues and VAT rates, the
conference participants at the VI World Conference of the Spatial Econometrics Association (Salvador, Brazil, July
2012), seminar participants at TilburgUniversity and Renmin University of China, and three anonymous referees for
their helpful comments. This research was financed bythe g rant 404-10-131 of NWO,the Netherlands Organization
for Scientific Research.
1Das-Gupta and Gang (2003) give a theoretical examination of the advantages of transactions cross-matching,
which is claimed to be a possibly important administrative advantageof the VAT. Limited empirical evidences are in
Ebrill et al. (2001) and Keen and Lockwood (2010).
§Jenny E. Ligthart died on 21 November 2012 after a short illness at the age of 45.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 79, 1 (2017) 0305–9049
doi: 10.1111/obes.12136
26 Bulletin
taxes on production and sales, often the turnover tax. As the rise of the VAT has been the
most significant development, by any standards, in tax policy and administration of the
recent decades (Keen and Lockwood, 2010), our aim is to study the factors influencing the
introduction of aVAT in a country, and in particular, the dynamic effects of the neighbouring
countries’ (VAT-)decisions on the VAT enactment. For this purpose, we propose a new
spatial duration model, discuss its estimation, and apply it to data onVAT adoption covering
the last 40 years.
Despite the fact that the VAT introduction can have a substantial effect on govern-
ment revenues and their stability or imports and exports (e.g. see the discussion in Keen
and Lockwood, 2010), the VAT enactment has received remarkably little attention in the
academic literature, especially on the empirical side.2Ebrill et al. (2001) provide some
informal guidance on selecting potential determinants of VAT adoption.3They informally
argue that countries are more likely to adopt the VAT if they have a higher GDP per capita,
are less open, have a higher literacy rate, and feature a larger population. Recently, Keen
and Lockwood (2010) are the first ones to formally explore the causes and consequences
of VAT adoption by using a dynamic probit model for a sample of 143 countries during
the 1975–2000 period. Their analysis makes a first step in capturing possible neighbour-
hood effects of VAT adoption: countries are more inclined to implement a VAT when other
countries in the same region have done so, the so-called copycat effect. However, Keen
and Lockwood (2010) neither employ a formal spatial econometric framework nor make
use of survival analysis to measure possible neighbourhood effects.4
To examine the spatial correlation among neighbouring countries’ decisions, it is not
sufficient to incorporate a variable that indicates the proportion of countries with a VAT
in the same region. The correlation among neighbours can be caused by various factors
such as the copycat effect founded in the ‘yardstick competition’(Besley and Case, 1995),
trade competition, or common global influences of economic or institutional nature (e.g.
the International Monetary Fund, IMF). Consequently, these neighbourhood effects do not
only exist between the observations, they might also occur in the unobserved factors and
thus in error term. With regard to the spatial dependence among the observations and in
the unobserved components, the traditional (non-spatial) estimation procedures may not be
consistent to draw appropriate inferences as their assumptions have been violated; hence,
appropriate inference is not feasible. On the other hand, standard spatial survival models
always assume that the spatial correlation structure only exists in the unobserved errors,
which is not realistic and does not facilitate examining the spatial correlation explicitly
(e.g. Li and Ryan, 2002; Bastos and Gamerman, 2006).
2Variousstudies focus on the economic effects of VAT enactment. Nellor (1987) considers empirically the revenue
effect of VAT adoption by analyzing a sample of 11 European countries in the 1960s and 1970s and provides
evidence that VAT introduction raises the tax revenue-to-GDP ratio. Desai and Hines (2005) examine the effect of
VAT implementation on international trade and find that reliance on VAT is associated with less exports and imports.
Furthermore, this negative effect on exports is stronger among low-income countries than it is among high-income
countries.
3Wefocus on the date of VAT implementation. However,we will refer to adoption, enactment, and implementation
interchangeably.
4Brockmeyer (2010) uses the Coxpropor tional hazard model to estimate the impact of lending bythe Inter national
Monetary Fund (IMF) on VAT adoption in a panel of 125 countries during the period 1975–2000, but she does not
focus on the spatial dimension.
©2016 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial