STRUCTURAL DETERMINANTS OF TAX REVENUE IN LATIN AMERICA AND THE CARIBBEAN, 1990-2009

D istr.
L IM IT E D
L C /M E X /L .1 0 8 7
E C L fl C
S U B R E G IO N A L H E A D Q U A R T E R S IN MEXICO
4 D ecem ber, 2012
O R IG IN A L : E N G L IS H
STRUCTURAL DETERMINANTS OF TAX
REVENUE IN LATIN AMERICA AND
THE CARIBBEAN, 1990-2009
Luca Dioda
Comisión Económica para América Latina y el Caribe (CEPAL)
2
This document has been prepared by Luca Dioda, Consultant of the Economic Development Unit, Economic
Commission for Latin America and the Caribbean (ECLAC), Subregional Headquarters in Mexico.
The views expressed in this document, which has been reproduced without formal editing, are the author's sole
responsibility and do not necessarily reflect those o f the Organization.
LC/MEX/L.1087
Copyright © 2012, United Nations. All rights reserved
Press in United Nations • México, D. F. • December 2012 • 2012-041
3
CONTENTS
ABSTRACT
............................................................................................................................................................ 5
IN T R O D U C T IO N ............................................................................................................................................................ 7
I.
R E C E N T S E L E C T E D S T U D IE S O F T H E S T R U C T U R E O F T A X A T IO N O N
D E V E L O P IN G C O U N T R IE S ............................................................................................................................ 9
II.
T H E T A X S T R U C T U R E IN L A T IN A M E R IC A : S T Y L IZ E D F A C T S .............................................12
III.
E C O N O M E T R IC M E T H O D O L O G Y ............................................................................................................ 16
A.
B.
C.
D.
IV .
E c o n o m ic v ariab les in clu d ed in th e em p irical a n a ly s is ...................................................................... 16
P o litical v a r ia b le s ............................................................................................................................................18
S ocio d e m o g rap h ic v a ria b le s ....................................................................................................................... 19
O th e r co n tro l v a r ia b le s .................................................................................................................................. 19
R E G R E S S IO N A N A L Y S IS : F U N C T IO N A L S P E C IF IC A T IO N A N D R E S U L T S ...................... 20
A.
B.
C.
S tructure o f ta x a tio n ...................................................................................................................................... 23
S u b reg io n al d ifferen ces in tax ation: th e p a rticu la rity o f C entral A m e ric a and the
C a rib b e a n ........................................................................................................................................................... 23
T he ta x effo rt i n d e x ...................................................................................................................................... 28
C O N C L U S IO N S .....................................................................................................................................................30
V.
B I B L IO G R A P H Y .............................................................................................................................................................33
IN D E X O F T A B L E S
1
2
3
4
5
6
7
8
IN D E X
1
2
3
4
L A T IN A M E R IC A A N D T H E C A R IB B E A N : T A X R E V E N U E , 1 9 9 0 -2 0 0 9 ....................... 13
L A T IN A M E R IC A : D IR E C T , IN D IR E C T A N D T O T A L T A X R E V E N U E ,
1 9 9 0 -2 0 0 9 ...........................................................................................................................................................14
A D IA G R A M A T IC IL L U S T R A T IO N O F T H E M O D E L S P E C IF IC A T IO N
A P P R O A C H A D O P T E D IN T H E R E G R E S S IO N A N A L Y S IS ....................................................17
S U M M A R Y T A B L E O F R E G R E S S IO N S R E S U L T S ................................................................... 22
T A X R E V E N U E , D IV ID E D B E T W E E N D IR E C T A N D IN D IR E C T T A X E S IN
T H E C A R IB B E A N IN 2 0 0 9 .......................................................................................................................24
S H A R E O F A G R IC U L T U R E IN G D P A N D G D P P E R C A P IT A ............................................. 26
C IV IL L IB E R T IE S A N D P O L IT IC A L R IG H T S IN D E X E S ........................................................27
T A X E F F O R T IN D E X ................................................................................................................................ 29
O F F IG U R E S
TAX REVENUE
TAX REVENUE
TAX REVENUE
TAX REVENUE
AND
AND
AND
AND
GDP
GDP
GDP
GDP
GROW TH
GROW TH
GROW TH
GROW TH
RATE
RATE
RATE
RATE
IN
IN
IN
IN
C E N T R A L A M E R IC A , 1 9 9 0 -2 0 0 8 ... 15
S O U T H A M E R IC A , 1 9 9 0 -2 0 0 8 .............15
M E X IC O , 1 9 9 0 -2 0 0 8................................. 15
T H E C A R IB B E A N , 1 9 9 0 -2 0 0 8 ..............15
A N NEXES:
I
E C O N O M IC V A R IA B L E S ...................................................................................................................... 35
II
E C O N O M IC V A R IA B L E S F IG U R E S ..................................................................................................37
III
R E S U L T S O F T H E R E G R E S S IO N S .................................................................................................... 39
5
ABSTRA CT
In recen t y ears m an y studies have b een conducted w orldw ide in o rd er to investigate the structural o r lo n g ­
term d eterm inants o f ta x revenue (see in ter alia G upta, 2007; M ahdavi, 2007; P ro feta and Scabrosetti, 2010).
T he aim o f th is study is to extend th e em pirical literature on this issue by applying standard m odels to the
case o f L atin A m erica and th e C aribbean. T hrough panel econom etric m ethodologies, the p a p er assesses the
statistical significance o f a n u m b e r o f potential determ inants o f ta x revenue as a share o f G D P , using data
from 32 L atin A m erican countries o v er the perio d 1990-2009. The em pirical analysis pays particular
attention to exam ine th e relevance o f p o litical and historical variables to understand regional differences in
ta x revenue. T he results indicate that, am ong the variables th a t exert a statistically significant influence on
ta x revenue are th e follow ing ones: civil liberties, fem ale lab o r force participation, the age com position o f
th e population, th e degree o f p olitical stability, the level o f education, the population density as w ell as the
size o f th e shadow econom y.
7
INTRODUCTION
T h e le v e l o f ta x a tio n in L a tin A m e ric a , in p a rtic u la r in C e n tra l A m e ric a , is v e ry lo w c o m p a re d w ith
o th e r re g io n s in th e w o rld . T h is is w o rry in g g iv e n th a t an in su ffic ie n t le v el o f fisca l rev e n u e m a k e s it
e x tre m e ly d iffic u lt fo r a n y g o v e rn m e n t to a d e q u a te ly p ro v id e p u b lic serv ices an d to m e e t th e b a sic
n e e d s o f th e p o p u la tio n in te rm s o f h e a lth , e d u c a tio n an d sec u rity , n o t to m e n tio n to p u t in p la ce a
d e v e lo p m e n t a g e n d a to c a rry o u t a stru ctu ral tra n sfo rm a tio n o f th e eco n o m y to b o o s t g ro w th . L ow
fisc a l re v e n u e s are in m a n y c o u n trie s a m a in cau se o f p o o r p u b lic serv ic es an d in fra stru c tu re as w e ll an
im p o rta n t c o n s tra in t o n so c ia l e x p e n d itu re .
T h e aim o f th is p a p e r is to id e n tify th e m a in lo n g -r u n d ete rm in a n ts o f ta x rev e n u e, b o th its
m a jo r item s as w e ll as its a g g re g a te to ta l. T o do so, th e e m p irica l a n a ly sis co n sid e rs th re e g ro u p s o f
e x o g e n o u s v a ria b le s: e c o n o m ic , p o litic a l an d so c io -d e m o g ra p h ic ones. T he e m p iric a l a n a ly sis is b ased
on th e a p p lic a tio n o f p a n e l re g re ssio n s m o d e ls. T he u n d e rly in g a ssu m p tio n is th a t b y k n o w in g th e lo n g ru n d e te rm in a n ts o f ta x re v e n u e s, p o lic y m a k ers in th e reg io n m a y be b e tte r eq u ip p e d to id e n tify som e
o f th e o b sta c le s th a t fiscal refo rm s face and, th e re fo re , to id e n tify p o lic y in stru m e n ts th a t m a y h e lp to
rem o v e th em .
T he p a p e r is o rg an ized as fo llo w s. A fte r th is in tro d u ctio n , c h ap te r I p rese n ts a b rie f review o f
re c e n t studies o n th e stru ctu ral d eterm in an ts o f ta x rev en u es in dev elo p in g countries. C h ap te r II show s the
sty lized facts o f ta x rev en u es in th e sam ples selected, and ch a p ter III in tro d u ces, b o th , th e eco n o m etric
sp ecificatio n th a t is u sed fo r th e em p irical an aly sis and th e list o f exo g en o u s v ariab les ty p ica lly
c o n sid ered in th e re le v a n t literatu re o n th is issue. T he results o f th e eco n o m etric an aly sis are d iscu ssed in
c h a p te r IV . F in ally , c h a p te r V p u ts fo rw ard th e co n clu sio n s an d p o in ts to fu rth er research on this
im p o rta n t to p ic fo r th e region.
9
I. R E C E N T S E L E C T E D S T U D IE S O F T H E S T R U C T U R E O F T A X A T IO N O N
D E V E L O P IN G C O U N T R IE S
In th e last five to te n years there have been a considerable n u m b er o f studies devoted to identify the
h istorical d eterm inants factors b eh in d th e regional and national differences in taxation structure. T heir
conclusions are rath er diverse, in p art due to the heterogeneity in m ethodologies, the set o f countries
covered, th e specific variab les an d th e periods considered fo r the em pirical analysis.
A cco rd in g to a stu d y o f A b h ijit Sen G u p ta (2007), th a t co v ers 105 d ev elo p in g co u n tries o v e r 25
y ears, th e fo llo w in g v ariab les h av e a p o sitive and statistically sig n ifican t asso ciatio n w ith ta x revenue p e r
c ap ita : size o f th e e c o n o m y as reflected b y G D P a t p u rch a sin g p o w e r p arity , trad e o p en n ess, fo reig n aid,
an d a n u m b e r o f in d icato rs o f p o litic a l an d eco n o m ic stability. O n the o th e r h and, th e study co n clu d es th a t
th e share o f ag ricu ltu re in G D P , an d an in d ica to r o f co rru p tio n ex ert a statistically sig n ifican t n egative
in flu en ce o n th e lev el o f ta x rev en u es. D iv id in g his sam ple in th ree g ro u p s acco rd in g to the level o f
in co m e, th e stu d y fin d s th a t fo reig n aid h as a sig n ific an t an d po sitiv e effe ct on ta x rev en u es in low
in co m e co u n tries b u t n o t in m id d le o r h ig h incom e ones. M o reover, he finds a strong, n egative
relatio n sh ip b e tw e e n ta x rev en u e an d c o rru p tio n only fo r the m id d le and lo w incom e countries. P olitical
stab ility is n e g a tiv e ly asso c ia te d w ith ta x revenue only in h ig h incom e co untries; in the others the
a sso ciatio n is p o sitiv e. F in ally , he id en tifies a n eg ativ e re latio n b e tw ee n in d irect ta x e s and revenue
p erfo rm an ce, in th e sense th a t o v erall ta x rev enue as a share o f G D P ten d s to be lo w er in th e p resen ce o f a
relativ ely h ig h level o f ta x e s o n g o o d s an d services.
B ird , M artin ez-V elasq u ez an d T o rg le r (2004) analyze d ata on ta x rev en u es and th e ir determ in an ts
in a sam ple o f 110 d ev elo p in g co u n tries fo r 1990-1999. A m o n g th e ir m ain co n clu sio n s stand o u t the
fo llo w in g o n e s: i) p e r c a p ita G D P is p o sitiv e ly asso ciated w ith ta x revenue, b u t trad e open n ess does n o t
h av e a statistically sig n ific a n t in flu en ce, and ii) dem o g rap h ic g ro w th and th e share o f ag ricu ltu re in G D P
are asso c ia te d w ith lo w e r lev els o f ta x revenue. T h eir results also su g g est th a t the degree o f in eq uality, the
size o f th e sh ad o w eco n o m y an d th e re g u latio n o f en try are n eg a tiv ely asso c iate d w ith ta x revenue. The
in d ices o f civ il lib erties an d p o litic a l rights, political stability, rule o f law an d relative ab sence o f
c o rru p tio n have a statically sig n ific a n t p o sitiv e asso ciatio n w ith ta x revenue. O ne o f th e ir con clu sio n s is
th a t b e tte r an d m o re e ffic ie n t in stitu tio n s lead to a h ig h e r level o f ta x revenue. H o w ev er, one co u ld also
argue th e cau sality in th e o p p o site d irectio n , i.e. h ig h e r levels o f tax a tio n m ake it po ssib le to have
e ffic ie n t in stitu tio n s. F in ally , th e ir stu d y does som e reg io n al co m p ariso n s and finds th a t th e lo w e r level o f
ta x rev en u e in L atin A m e ric a relativ e to o th e r develo p in g co u n tries is m ain ly due to th e lo w e r q u ality o f
its in stitu tio n s, h ig h e r co rru p tio n , la rg e r sh adow e co n o m y an d lo w e r ta x rates.
M ah d av i (2 0 0 7 ), in his an aly sis o f d ata o f 43 dev elo p in g co u n tries o v er th e p erio d 1973-2002,
fin d s a p o sitiv e co rrelatio n b etw een ta x revenue and open n ess o f th e eco n o m y , th e literacy rate and G D P
p e r c a p ita g ro w th rate. O n th e o th e r h and, he co n clu d es th a t an in crease in fo reig n aid, in aging o f
p o p u latio n , in p o p u la tio n d e n sity an d in inflatio n has a neg ativ e relatio n sh ip w ith ta x revenue, w h ereas
v ariab les such as th e share o f ag ricu ltu re on G D P , fem ale la b o r force p articip atio n , eco n o m ic v o latility ,
civ il lib erties an d p o litical rig h ts are statically in significant. H e also finds a po sitiv e co rrelatio n b etw een
ta x rev en u e o b tain ed from in co m e, p ro fits and capital g ain s an d the level o f political rights. H o w e v e r the
in d e x o f civ il lib erties he u ses is n e g a tiv ely asso ciated w ith v alue ad d ed tax , p ro p erty tax e s and social
secu rity revenues.
P ian castrelli (2 0 0 1 ), b ased o n d a ta o f 75 d ev elo p ed and d ev elo p in g co u n tries o v e r the perio d
1985-1995, id en tifies p e r c a p ita G D P , the share o f in d u stry in G D P and tra d e open n ess as th e m o st
10
im p o rta n t d eterm in an ts o f ta x revenue. T he share o f ag riculture on G D P is n eg ativ ely asso ciated w ith ta x
rev en u e, in line w ith o th e r studies. P ro fe ta an d S cabrosetti (2010) an aly ze d determ in an ts o f ta x revenue o f
39 d ev elo p in g co u n tries o v e r th e p e rio d 1 9 9 0 -2 0 0 4 , in clu d in g 11 A sian , 19 L atin A m erican and 9 recen t
m em b ers o f th e E U . T h ey id en tify statistically sig n ifican t d ifferen ces in the reg io n al determ in an ts o f ta x
revenue. F o r in stan ce, G D P p e r c a p ita an d th e d eb t/G D P ratio w ere n o t statically sig n ifican t determ in an ts
o f ta x rev en u es in th e A sia n eco n o m ies in clu d ed in the sam ple, b u t w ere po sitiv e statistically sig n ifican tly
fo r L atin A m erican co u n tries. B u t fo r th e w hole sam ple b o th in d icato rs ap p ea r to have a po sitiv e b u t n o t
alw ay s sig n ific a n t influence. T he share o f ag riculture o v e r G D P in flu en ces ta x revenue n eg a tiv ely in
L atin A m e ric a b u t is n o t sig n ific a n t in A sia, open n ess o f the eco n o m y has a p o sitiv e im p ac t on ta x
rev en u e in A sia an d in E u ro p e, b u t a n eg ative one in L atin A m erica. T he in d e x o f d em o cratizatio n seem s
to be p o sitiv e ly lin k ed to ta x rev en u e an d a h ig h e r lev el o f civil lib erties and p o litica l rights is asso ciated
w ith in creased ta x p erfo rm an ce. F o r L atin A m erican co u n tries ed u catio n , th e share o f p o p u latio n o v e r 65
y ears old, fem ale la b o r force p a rtic ip a tio n and th e size o f th e sh adow ec o n o m y are p o sitiv e and
sig n ifican tly related to ta x rev en u e, w h e re as p o p u la tio n d en sity is not. F o r A sia v ariab les such as
seco n d ary school a tta in m e n t an d u rb a n p o p u la tio n are n o t sig nificant, w h ereas the share o f p o p u latio n
o v e r 65 y ears old is n e g a tiv e ly asso c ia te d w ith ta x revenue.
A s h ig h lig h ted b y th e abo v e an aly ses o f re ce n t eco n o m etric studies, th ere is no cle a r p attern o f
th e sig n ifican ce o f all th e v ario u s p o ten tial d eterm in an ts o f ta x p erfo rm an ce in d ev elo p in g countries.
H o w ev er, alth o u g h som e resu lts do v ary according to th e p e rio d an aly zed and th e sam ple o f countries
ch o sen , in g en eral, in d icato rs like G D P p e r capita, the share o f n o n -ag ricu ltu ral activ ities on G D P
— w h ic h m a y be in terp reted as p ro x ies o f a c o u n try 's stage o f d ev elo p m en t— have a p o sitiv e, sig n ifican t
in flu en ce o n ta x revenue. A h ig h e r d egree o f open n ess and an in creasin g d eb t are u su ally asso cia ted w ith
a h ig h e r lev el o f tax atio n . M o reo v er, d e m o cratizatio n , b e tte r in stitutions, less co rru p tio n , political stability
an d th e rule o f law are u su a lly asso ciated w ith in creased ta x revenue.
O n th e o th e r h an d , th e asso ciatio n b etw e en fo re ig n aid an d th e sh adow eco n o m y on ta x
p erfo rm an ce is n o t c le a r cut. A cco rd in g to M ah d av i fo reig n aid ten d s to be linked w ith lo w er ta x revenue,
w h ereas G u p ta p o in ts to a d irect e ffe c t in th e case o f lo w incom e co u n tries b u t a n o n -sig n ific an t one in
th e case o f ric h e r cou n tries. T he size o f th e shadow eco n o m y is p o sitiv ely asso ciated w ith ta x revenue
acco rd in g to P ro fe ta an d S cab ro setti (2010), w h ereas B ird, M artin ez V asq u ez and T o rg ler (2004) fin d a
n eg ativ e correlatio n . O th e r facto rs th a t, acco rd in g to th e rev iew ed literatu re co u ld have an effect on
ta x a tio n are co n flicts, b o th in tern al an d external. B esley and P ersso n (2007) an aly zed th e effects o f w ars
on th e fiscal cap acity w ith d a ta o f 180 co u n tries fro m 1945 to 1997. T h ey fo u n d th a t a rm ed co nflicts
a ffe c t th e cap acity o f a g o v e rn m e n t to co lle ct ta x revenues: ex tern al w ars m a y b o o st fiscal cap acity as
g o v ern m en ts ca rry o u t in v estm en ts fin an ced b y tax es, w h ereas in ternal w ars decrease ta x revenue as
co n flicts am o n g d iffe re n t g ro u p s w e a k e n state institutions. In a n o th er study, co vering 188 co u n tries o v er
th e p e rio d 1975-2004, C árd en as, E sla v a and R am irez (2010) fin d a neg ativ e correlatio n b etw ee n in ternal
co n flicts an d fiscal cap acity m easu red in tw o d ifferen t w ays: i) as to ta l ta x rev enue as a percen tag e o f
G D P , fo llo w in g B esley an d P ersso n (2007), and ii) as in co m e ta x revenue as share o f G D P . T hey partly
co n firm th e resu lts o f th e p rev io u s study; in fact, w h ereas in tern al co n flicts w ea k en state cap acity , the
au th o rs state th a t ex tern al co n flicts are n o t asso c iated w ith h ig h e r levels o f ta x revenue, in co n trast w ith
B esley an d P ersso n (2007). T h e y p o in t o u t th a t m a jo r in tern atio n al w ars shaped fiscal cap acity an d in p art
co n trib u ted to th e fo rm atio n o f th e m o d ern state. E x tern al co n flicts can have a po sitiv e im p ac t on fiscal
cap acity o n ly i f th e p e rio d a n aly zed is long e n o u g h (fo r in stance B esley and P ersso n a n aly zed five
d ecad es) an d i f m a n y co u n tries are in v o lv e d in it. In addition, th e ir stu d y also in d icates th a t th e in ten sity
o f th e in tern al c o n flic t h as an in flu en ce o n th e S ta te 's cap acity to co llec t tax. T hey fin d th a t in C olom bia,
in 1994-2002, k id n ap p in g s an d fo rced d isp lacem en ts ten d ed to decrease ta x rev enue as a share o f G D P as
th e y w eak e n e d th e c ap acity to c o lle c t taxes. O th er variab les th at, according to th e rev iew ed literature,
11
a ffe c t ta x rev en u e are th e level o f ed u catio n , th e d en sity o f th e po p u latio n , rate o f urb an izatio n , fem ale
la b o r force p a rtic ip a tio n an d th e age co m p o sitio n o f the po p u latio n . T he p re se n t p a p e r has th e aim o f
p ro v id in g re c e n t ad d itio n al eco n o m etric evid en ce on th e determ in an ts o f th e structure o f ta x a tio n in L atin
A m erica, w ith a special focus o n th e C aribbean and C entral A m erica.
12
II. THE TAX STRUCTURE IN LATIN AMERICA: STYLIZED FACTS
The em pirical analysis o f th is p a p e r is b ased on d ata fo r 32 countries in L atin A m erica and the C aribbean 1
o v er th e p erio d 1 9 90-2009. A s m en tio n ed in the introduction, th e level o f ta x revenue in L atin A m erican
countries is low . In 2009, it stood on average at 19% o f G D P, notw ithstanding it h ad increased m ore th an
fo u r percentage points since 1990. In co m parison, in E urope and in the O E C D countries the ta x burden w as
39.7% 2 and 28 .4 % o f G D P. 3
T ax revenue certainly did n o t g ro w at the sam e pace in all countries in the region in 1990-2009, and
n e ith e r th e econom ies expan d ed a t th e sam e pace. O n the one hand, som e C aribbean countries like St K itts
and N ev is, H aiti, T rinidad an d T obago, as w ell as N ic arag u a in C entral A m erica, and C olom bia, Ecuador,
A rgentina, P lurinational State o f B o liv ia and B razil in South A m erica increased th e ir ta x revenues a t an
a n n u a l a v e ra g e ra te o f g ro w th o v e r a n d ab o v e 2% . O n th e o th e r h a n d , in o th e r c o u n trie s it d e c re a se d
— Surinam e, G uyana, B elize— an d in o th er it augm ented at v ery a slow pace. In any case, in general the
o verall ta x bu rd en rem ains v ery sm all especially in C entral A m erica and — to a lesser extent— in South
A m erica. M oreover, it is w orrying th a t in M exico, H aiti, G uatem ala and Panam a, the C entral G overnm ent
collects less th a n 12 p ercen t o f th e G D P from taxes, in contrast to 2 3 .5 % in B razil o r 18.2% in A rg en tin a 4
(see table 1).
A s ta b le 2 show s, in th e re g io n ta x rev en u es dep en d m ore on in d ire ct ta x e s th a n d irec t ones. In
fact, th e rev en u e co llected fro m th e fo rm e r is tw ice as h ig h as th a t com ing fro m th e latter. In addition,
a n o th e r c h aracteristic o f th e stru ctu re o f tax a tio n in th e re g io n is th e h ig h level o f ev asio n and the
p resen ce o f so c alled special reg im es th a t im ply a sig n ifican t loss o f fiscal revenue to g ro u p s o f interest.
In th e n e x t c h a p te r w e p re se n t th e fu n ctio n al specificatio n o f the eco n o m etric m odel — and discuss its
resu lts— u se d h ere to assess th e relev an ce o f a series o f eco n o m ic and political v ariab les as d eterm inants
o f ta x rev en u e in L atin A m e ric a n co u n tries. T he results o f th e em p irical an aly sis m ay serve as an in put
n o t o n ly fo r th e acad em ic d ebate but, h o p efu lly , also fo r p o licy m akers.
A m o n g th e eco n o m ic v ariab les th a t is a prio ri e x p ected to have an im p act on ta x rev enue stand
o u t th e rate o f g ro w th an d th e lev el o f G D P p e r capita. T he g ro w th p erfo rm an ce o f co u n tries in th e region
h as b e e n u n ev en , w ith p erio d s o f h ig h g ro w th fo llo w ed by others o f slow and in som e co u n tries on
o ccasio n e v en n eg ativ e g ro w th . M o reo v er, as is w ell k n o w n in th e la st th ree d ecades L atin A m erica grew ,
o n av erag e, a t a slo w e r p ace th a n E a st A sia n co u n terp arts d espite h av in g a p p lied rad ical m a rk e t reform s
in th e 1980s an d 1990s. F o cu sin g o n th e p erio d 1990-2010 one can d istin g u ish tw o b ro ad g ro u p s o f
co u n tries in C entral A m e ric a an d th e C aribbean. T he first one fo rm e d b y co u n tries th a t ex p erien ced , in
g en eral, h ig h g ro w th rates o f real G D P , and in clu d es P anam a, th e D o m in ican R ep u b lic, T rin id ad and
T o b ag o , C o sta R ic a an d A n tig u a an d B arbuda. T he seco n d co n fo rm ed b y G renada, th e B aham as,
Jam aica, B arb ad o s, St L u c ia an d H aiti h a d w ea k e co n o m ic g row th. Interestin g enough, fo r th is sam ple
an d p erio d , it seem s th a t co u n tries th a t h ave alw ay s had, fo r histo rical reasons, h ig h level o f ta x revenue
— such as th e fo rm e r B ritish c o lo n ie s— , h a d a w eak g ro w th p erfo rm an ce th ese y ea rs co m p ared to
co u n tries w ith relativ e lo w ta x rev en u es level (see ta b le I-1 in an n ex I).
1 Data for Cuba with the desired disaggregation was not available.
2 Source: EUROSTAT, webpage http://epp.eurostat.ec.europa.eu/cache/TTY_OFFPUB/KS-SF-11-026/EN/KS-SF-11-026EN.PDF; data related to tax revenue including social contributions of the general government.
3 Source: OECD, Centre for Tax Policy and Administration, webpage http://stats.oecd.org/Index.aspx.
4 Source: CEPALSTAT, related to fiscal revenue of Central Government excluding grants and donations.
13
TABLE 1
LATIN AMERICA AND THE CARIBBEAN: TAX REVENUE, 1990-2009
(Percentages o f GDP)
Country
Tax revenue as % o f GDP
1990
2000
2005
2009
T otal
13.8
16.0
18.3
19.0
C arib b ean
18.8
20.1
22.6
23.7
S panish colonies
Dominican Republic
Haiti
8.9
10.5
7.3
10.2
12.5
7.9
12.1
14.6
9.7
12.4
13.1
11.7
B ritish an d D utch colonies
Antigua and Barbuda
Bahamas
Barbados
Belize
Dominica
Granada
Guyana
Jamaica
St Kitts
St Vincent and the Grenadines
St Lucia
Suriname
Trinidad y Tobago
21.6
18.4
21.9
15.8
15.6
31.1
22.1
24.2
19.4
16.7
31.4
20.5
28.4
23.0
20.2
23.4
29.0
25.5
23.3
27.8
26.4
25.4
19.3
16.8
32.3
21.6
31.6
22.9
21.6
26.7
26.8
27.1
28.1
31.1
24.7
22.3
22.9
20.7
16.7
24.0
26.0
24.9
22.7
18.3
22.6
21.2
23.7
22.6
C e n tra l A m erica
Costa Rica
El Salvador
Guatemala
Honduras
Nicaragua
Panama
9.8
11.0
9.3
7.7
12.4
8.1
10.3
12.0
12.3
10.2
11.5
13.7
14.5
9.6
12.9
13.6
12.4
11.5
14.5
16.7
8.7
13.3
13.8
12.4
10.7
14.4
17.7
10.9
South A m erica a n d Mexico
Argentina
Bolivia (Plurinational State of)
11.7
10.2
14.5
13.5
12.9
15.3
19.9
17.9
9.4
10.0
9.7
12.0
12.2
16.4
12.9
15.3
15.8
20.5
22.7
18.3
12.4
10.1
8.8
13.0
13.6
18.3
15.3
15.4
18.2
Brazil
Chile
Colombia
Ecuador
Mexico
Paraguay
Peru
Uruguay
Venezuela (Bolivarian Republic of)
Source: ECLAC.
14.7
7.0
7.8
9.8
10.0
10.8
14.1
17.8
23.5
16.1
12.9
13.9
9.6
14.5
13.4
18.0
13.5
14
TA BLE 2
L A T IN A M E R IC A : D IR E C T , IN D IR E C T AND T O T A L TAX REV EN U E, 1990-2009
(Percentages o f GDP)
Tax revenue as share o f GDP
1990
9.8
18.8
11.7
Central America
Caribbean
South America and Mexico
2000
2005
2009
12.0
20.1
13.5
12.9
22.6
15.3
13.3
23.7
15.4
Direct tax revenue as share o f GDP
Central America
Caribbean
2.7
5.7
3.1
6.1
4.0
6.8
4.9
7.8
South America and Mexico
3.8
4.0
5.1
5.3
Indirect tax revenue as share o f GDP
Central America
Caribbean
South America and Mexico
6.5
8.4
8.8
8.2
11.0
6.0
11.6
8.2
14.7
8.6
14.7
8.5
Source: ECLAC, elaborated by the author on the basis of a simple average.
Note: The sum of direct and indirect tax revenue does not coincide with total tax revenue because other taxes
and tax devolutions are not classifiable as direct and indirect taxes.
A n aly zin g th e e co n o m ic p erfo rm an ce o f the S outh A m erican eco n o m ies an d M exico, the p attern
in th is reg ard is so m ew h at sim ila r to th e one m e n tio n ed above o f th e C arib b ean and C entral A m erica.
D u rin g 1990-2009 C hile, A rg en tin a, P eru an d U ru g u ay reg istered an average annual rate o f g ro w th o f real
G D P h ig h e r th a n 2.5% . O n th e o th e r h a n d B o liv arian R ep u b lic o f V en ezu ela, M exico an d P arag u ay had
an e v en less d y n am ic p erfo rm an ce w ith an average annual G D P g ro w th rate aro u n d 1%. In ad d itio n it
sh o u ld be m en tio n ed th a t fo r th e reg io n th e average rate o f eco n o m ic e x p an sio n p o st 1990 has b ee n lo w er
th a n th e one th a t c h aracterized its g ro w th p ath d u ring 1950-1980. In a d d itio n , th e re g io n ’s eco n o m ic
ex p an sio n in re c e n t d ecad es h as b een m ark ed b y episo d es o f h ig h v o la tility an d n o t in freq u en t financial o r
b alan ce o f p ay m en ts crisis.
In g en eral fo r th e w h o le reg io n th e e v o lu tio n o f ta x revenue does n o t ex actly m irro r th a t o f the
rate o f g ro w th o f G D P . T h ere ap p ears to be som e co rrelatio n on th e ir cy clical flu ctu atio n s, b u t the lo n g ­
te rm tre n d o f ta x rev en u e as a share o f G D P in the su b reg io n s fo llo w s d iffe ren t p attern s fro m the
co rresp o n d in g lo n g -term rate o f eco n o m ic e x p an sio n (see fig u res 1 to 4). T he em p irical an alysis, rep o rted
in th e fo llo w in g ch ap ter, ex p lo res fo r L atin A m e ric a the statistical significance o f the am ple set o f
v ariab les — su g g ested b y th e sp ecialized literatu re— th a t m ay have a p e rsiste n t in fluence on ta x revenues
an d cap tu re k e y eco n o m ic, p o litical, h isto rical, dem o g rap h ic and even g eo g rap h ic characteristics o f the
co u n tries in th e region.
FIGURE 1
TAX REVENUE AND GDP GROWTH RATE IN
CENTRAL AMERICA, 1990-2008
i— i 1----1------1---- 1— i-- 1------1-1----- 1— i 1----1------1-1------1— i 1---1990 1992 1994 1996 1998 2000 2002 2004 2006 2008
—
♦
—
Taxrevenue
Years
-4
—
♦
—
GDPgrowthrate
Source: ECLAC.
FIGURE 3
TAX REVENUE AND GDP GROWTH RATE
IN MEXICO, 1990-2008
17
1990
1992
1994
1996
1998
2000
2002
2004
Years
-■
-Tax revenue —
♦
—
GDPgrowthrate
Source: ECLAC.
2006
2008
FIGURE 2
TAX REVENUE AND GDP GROWTH RATE IN
SOUTH AMERICA, 1990-2008
— Tax revenue
Years
—
♦
—
GDPgrowthrate
Source: ECLAC.
FIGURE 4
TAX REVENUE AND GDP GROWTH RATE
IN THE CARIBBEAN, 1990-2008
Years
-Tax revenue —
♦
—
GDPgrowthrate
Source: ECLAC.
16
III. ECONOMETRIC METHODOLOGY
The em pirical analysis is b ased on an application o f panel d ata econom etric m ethods w ith a balanced panel
d ataset an d ran d o m effect regressions utilizing a sam ple o f 32 countries o v er the perio d 1990-2009.
F ollow ing th e standard m ethodology, the individual effects in th e random effects m odel are specified
th ro u g h a v ariable th a t is u ncorrelated w ith the independent variables. In the fixed effects m odels a
correlation b etw een th e individual erro r term and the pred icto r variables is assum ed. A ccording to the
relev an t literature, random effects m o d els are preferred w hen differences betw een countries are assum ed to
influence th e d ep en d en t variable (see in ter alia T orres-R eyna, 2010). R andom effects m odels te n d to be
preferred i f th e n u m b er o f individual o b servations N is large relative to the tim e dim ensions, so th a t the
individual effects can be co n sid ered ran d o m (H siao, 2004). In fact, random effects m odels have been w idely
u sed in th e em pirical studies o n the to p ic o f the presen t p a p er (Profeta, Scabrosetti 20 1 0 ; G upta, 2007;
R anjan, 2011), th o u g h som etim es com b ined w ith o th er m odels. W e ran regressions using fixed effect
specifications b u t th e results o btained ten d ed n o t to be significantly different from those derived through
random effects m odels. 5
In th e ad o p ted m o d el sp ecification, th e d e p e n d en t variab le is ta x revenue as a share o f G D P. It
in clu d es all rev en u es o f th e C en tral G o v ern m en t c o llected th ro u g h ta x es, b u t ex clu d es g ran ts, don atio n s
an d rev en u e co m in g fro m p e tro leu m an d n atu ral resources. 6 D a ta w as d eriv ed from C E P A L S T A T . 7 It is
im p o rta n t to p o in t o u t th a t w e also ran reg ressio n s tak in g th e d ep e n d e n t variab le e ith e r as revenue
c o llected th ro u g h d irect ta x e s o r as revenue co llected th ro u g h in d ire ct ones. T he p ro ced u re fo r the
in clu sio n o f ex o g en o u s v ariab les in th e m o d el sp ecificatio n w as the fo llo w in g one: First, w ere con sid ered
th e k e y v ariab les th a t, acco rd in g to th e em p irical results o f th e literatu re, m ay a prio ri p lay a role in
ex p lain in g ta x revenue. Second, co n tro l v ariab les w ere ad d ed w ith the double p u rp o se to evaluate th e ir
e ffe c t o n ta x reven u e an d to stren g th en th e m odel th ro u g h a ro b u stn ess ch eck (table 3).
A . E C O N O M I C V A R IA B L E S IN C L U D E D IN T H E E M P I R I C A L A N A L Y S IS
G D P p e r capita, at co n stan t U S dollars o f 2000 and expressed in logarithm serves as a proxy fo r the level o f
d ev elo p m en t o f a country. T he m odel tested the relation betw een G D P grow th and ta x revenue in Latin
A m erica b u t did n o t find co n sisten t evidence (see ch apter II). H o w ev er there is a positive correlation
b etw een G D P p e r cap ita (expressed in lo garithm ) and ta x revenue (see figure II-1 in annex II). 8 T he relation
b etw een ta x revenue and th e level o f d ev elopm ent o f a country has been studied in the literature. In ter alia
H inrichs (1966) and T anzi (1992) found a positive correlation betw een them . Indeed, according to W a g n e r’s
law , since the dem an d fo r pub lic services is incom e elastic, econom ic d evelopm ent is associated w ith an
in creased req u est fo r pub lic g oods and services w hich n eed to be financed in ter alia by increasing ta x
revenue (T anzi, 1987). A lso, d ev elo p m en t is associated w ith g rea ter State capacity to levy and collect taxes
(C elliah, 1971).
5 Heteroscedasticity, common in cross sectional data, could not be ruled out and thus error terms do not have constant variance,
the standard errors of the estimates are biased and the estimates may assume “wrong” values. The analysis was done using
STATA and applying standard procedures to correct for this problem.
6 Revenues from Panama Canal transit and from Mexican natural resources are not included. The exception is Trinidad and
Tobago, where total tax revenue of the Central government includes petroleum revenue..
7 Data on tax revenue was obtained from http://www.eclac.cl/estadisticas/. For many Caribbean countries, data was collected
from other official sources by the author.
8 In all graphs in this chapter, the average tax revenue is captured on the y-axis, and the average coefficient of exogenous
variables is captured on the x-axis, for 1990-2009.
17
TA BLE 3
A D IA G R A M A T IC IL L U S T R A T IO N O F TH E M O D E L S P E C IF IC A T IO N A PPR O A C H
A D O PT E D IN T H E R E G R E SSIO N ANALYSIS
D e te rm in a n ts o f ta x r e v e n u e : a g r a p h ic a p p r o a c h
G D P p e r c a p ita g ro w th rate
G D P p e r c a p ita
s h a re o f a g ric o ltu re o n G D P
E c o n o m ic in d ic a to rs
o p en ess o f econom y
d e fic it p r e v o u s y e a r
p e rs o n a l in c o m e ta x
c o r p o r a te ta x
S e c o n d a ry s c h o o l e n ro lm e n t
ta x o n p r o p e rty
s h a re o f o v e r 6 5 s in th e p o p u la tio n
fe m a le la b o u r fo rc e p a r te c ip a tio n rate
u rb a n iz a tio n
Tax re v e n u e
S o c io d e m o g ra p h ic in d ic a to rs
d e n s ity o f p o p u la tio n
p o p u la tio n g ro w th
shadow econom y
C
onG D P
\ -
J
ta x o n in te rn a tio n a l tra d e
In d ire c t t a x
d o m e s tic t a x o n g o o d s a n d s e r v ic e s
C iv il lib e rtie s
p o litic a l rig h ts
r e g im e d u r a b ility
P o litic a l in d ic a to rs
in te rn a l a n d e x te rn a l c o n flic ts
A N G L O dum m y
In d e p e n d e n t v a r ia b le s
D e p e n d e n t v a ria b le
E le m e n t o f d e p e n d e n t v a ria b le
Source: Elaborated by the author.
T he share o f ag ricu ltu re o v e r G D P is a n o th er variab le u se d as a p ro x y fo r d ev elo p m en t. A strong
n eg ativ e re la tio n b e tw e e n a g ric u ltu re ’s share in G D P and ta x revenue can be expected. In th e develo p in g
co u n tries it is n o t e asy to ta x th e ru ral sector, since a large p a rt o f it co nsists o f su b sisten ce and sm all
farm ers, n o to rio u sly d iffic u lt to ta x g iv e n the large n u m b ers th a t sell th e ir p ro d u cts in in fo rm al m arkets
(S to tsk y an d W o ld eM arian , 1997). O n th e o th e r h and, since m a n y p u b lic sec to r activities are u rb a n b ased,
a d eclin in g share o f ag ricu ltu re in G D P ten d s to be lin k ed to an in crease in d em an d fo r p u b lic
ex p en d itu res an d th u s p u t p ressu re to raise ta x revenue (T anzi, 1992). In L atin A m erica and C arib b ean the
level o f ta x rev en u e o n av erag e is low , b u t som e co u n tries like G u yana, D o m in ic a and B elize have a
relativ ely im p o rta n t ag ricu ltu ral secto r an d at the sam e tim e re lativ e ly h ig h ta x revenue. D ue to th e h ig h
co llin earity b e tw e e n th e share o f ag ricu ltu re in G D P and G D P p e r c a p ita (-0 .7 7 ), b o th in d icato rs m ay
serve as p ro x ies fo r d e v elo p m en t, w e to o k only “a g ric u ltu re ” am o n g th e fix ed and p rin cip al variab les fo r
th e em p irical an aly sis, an d ran sep arate reg ressio n s to evalu ate the e ffec t o f G D P p e r c ap ita (ex p ressed in
lo g arith m ) o n ta x revenue.
T h e lite ra tu re id e n tifie s th e o p e n n e ss o f th e e co n o m y , m e a su re d b y th e sum o f ex p o rts an d
im p o rts as a sh are o f G D P (see fig u re II-2 in a n n e x II), as a p o te n tia l d e te rm in a n t o f ta x rev e n u e ev en
th o u g h p re v io u s a n a ly sis are n o t c o n c lu siv e . T ax es on im p o rts and ex p o rts are e a sy to im p o se g iv e n th e
e a sily id e n tifia b le so u rce o f co lle c tio n . F u rth e rm o re , since o p e n e c o n o m ies are m o re e x p o se d to
e x te rn a l risk s, th e ir g o v e rn m e n ts c o u ld b e e x p e c te d to b u ild b e tte r in su ra n c e sy stem s in o rd e r to p ro te c t
th e ir c itiz e n s a g a in s t th e se risk s (R o d rik , 1998). F o r th e se rea so n s a p o sitiv e re la tio n b e tw e e n ta x
18
re v e n u e an d tra d e o p e n n e ss c o u ld b e ex p ecte d . O n th e o th e r h an d , in m a n y d ev elo p in g c o u n trie s, tra d e
lib e ra liz a tio n refo rm s te n d e d to d e c re a se ta x re v en u es as th e y lo w e re d ta riffs (K e en an d S im one, 20 0 4 ).
C a rib b e a n e c o n o m ie s sh o w a le v e l o f tra d e o p e n n ess (1 0 8 % ) h ig h e r th a n C e n tra l A m e ric a (91% ) and
S o u th A m e ric a (4 9 .9 % ). 9
A n o th e r v ariab le co n sid ered in th e em p irical literature is p u b lic d e b t (as a p ercen tag e o f G D P).
U n fo rtu n ately , co m p arab le d a ta w ere n o t av ailable fo r all L atin A m erica n co u n tries fo r th e ch o se n period.
O ften d a ta w ere n o t co m p arab le, o r did n o t co rresp o n d to th e sam e agg reg ate c ateg o ry o f g o v ern m en t, o r
h a d o th e r p ro b lem s th a t im p ed ed its use fo r co m p arativ e p u rp o ses. T hus, and in o rd er to w o rk w ith n o t a
v e ry sm all sam p le, in th e e m p iric a l a n a ly sis w e c o n sid e re d d a ta on th e fiscal d e fic it as a share o f G D P
— o f p re v io u s y e a rs — in ste a d o f d a ta o n debt.
B . P O L I T I C A L V A R IA B L E S
T h ere is no co n sen su s in th e em p irical literature on th e sig nificance o f po litical variab les — such as the
level o f d e m o cracy an d th e d u ratio n o f a p o litic al reg im e— as determ in an ts o f ta x revenue. O n the one
h an d , acco rd in g to som e au th o rs (A cem o g lu an d R o b in so n , 2006; B oix, 2003), d em o cracy is im p o rtan t to
red istrib u te in co m e fro m th e rich to th e p oor, to create an e n larg ed w elfare state, and a stro n g e r an d m ore
e ffic ie n t ta x system , b ased m o re o n d irect tax e s th an on in d ire ct taxes. In ad d itio n , an a prio ri assu m p tio n
is th a t u n d e r a n o n d em o cratic reg im e th e size o f th e p u b lic secto r w o u ld be relativ ely sm all, b ecau se a
large p a rt o f citizen s are ex clu d ed from the d ecisio n m ak in g p ro cess. T hus a tra n sitio n to w a rd s a
dem o cratic g o v e rn m e n t w o u ld co in cid e w ith an in crease in ta x e s and p u b lic sp ending in acco rd an ce w ith
th e th e o ry o f th e m e d ia n v o ter, m o v in g in the d irectio n o f a b e tte r red istrib u tio n o f w ealth. O n th e o th er
h an d , som e au th o rs, such as B arro (1979) an d W ittm an (1989), c o n sid er th a t the m ain drivers o f p u blic
p o licy are n o t p o litic a l factors, b u t e fficien cy con sid eratio n s. M o reo v er, u n lik e B o ix ’s th eo ry , M u lligam
(2004) d id n o t fin d ev id en ce th a t d em o cracies spend m ore in p u b lic services like ed u cation, health,
p en sio n , th a n auto cracies.
In an y case, to te s t th e se h y p o th eses th e em p irical an aly sis relied on tw o v ariab les 10 — civil
lib erties an d p o litic a l rig h ts— as p ro x y o f th e state o f d em ocracy. T he civil lib erties in d ex captures the
deg ree o f freed o m o f e x p ressio n an d b eliefs, o f org an izatio n , and o f assem bly. It can be co n sid ered as a
m easu re b o th o f th e rule o f law an d , p erh ap s to o , o f p e rso n al au to n o m y w ith o u t in terferen ce fro m the
state. T he ran g e o f th e in d ex v aries fro m one, th e h ig h e st lev el o f civil liberties, an d seven, th e low est.
M ah d av i (2008) u se d it as a p ro x y fo r th e lev el o f co rru p tio n con sid erin g th a t an im p ro v e m e n t o f civil
lib erties o f a co u n try sh o u ld be asso c ia ted w ith reduced co rru p tio n , ow ing to m ore tra n sp a re n c y and
acco u n tab ility w ith in th e p u b lic sector. T he p o litic al rights in d ex is a m ix tu re th a t cap tu res th e legal
p re ro g a tiv e s th a t enab le citizen s to p articip ate freely in the po litical p ro cess th ro u g h th e rig h t to vo te and
to be v o te d fo r p u b lic office, th e ex isten ce o f credible op p o sitio n and o f p o litical rights. It go es fro m one,
th e h ig h e st level o f p o litic a l rights, to seven th e low est. T here is h ig h co -lin earity am o n g th e se tw o
in d ices (0.80). A cco rd in g to th e literatu re a p o sitive co rrelatio n b etw e en the level o f d em o cracy and ta x
rev en u e is e x p ected an d a h ig h e r level o f civil lib erties and po litical rights should be asso cia te d w ith
h ig h e r ta x rev en u e (see fig u re II-3, a n n ex II).
The index o f regim e durability 11 reports the num ber o f years since either a regim e took office o r a
transition period has ended as defined b y the absence o f stable political institutions. The index takes the first
9 The degree of openness for the subregion is calculated as the average of the corresponding national figures for 1990-2009.
10 See the website from Freedom House www.freedomhouse.org
11 Derived from the policy IV dataset, see www.systemicpeace.org/polity/polity4.htm
19
y ear in office o f a n ew g overnm ent as the zero baseline, and each additional year b y the sam e governm ent is
counted consecutively until a n ew g overnm ent takes office and the count is restarted from zero. Political
stability m ay be associated w ith a larger capacity o f the state to collect taxes through reliable institutions.
D um m y variables (see sum m ary regressions 12-15 in annex III) w ere introduced to investigate possible
differences o f regim e taxation betw een countries due to other political o r historical variables such as internal
and external conflicts, o r historic heritage like having been a Spanish o r an E nglish colony.
C . S O C IO D E M O G R A P H I C V A R IA B L E S
A m o n g th e im p o rtan t socio dem og rap h ic variables considered in the em pirical literature as a facto r th a t m ay
influence tax atio n is th e average level o f education o f the population, The assum ption is th a t a hig h er level
o f education should enable citizens to b etter understand and com ply w ith ta x codes, to have a b etter access
to form al jo b s and, p erh ap s to o , to have g reater conscience o f the responsibility o r obligation to p ay taxes. In
th e econom etric m o d el it w as captured as the average n u m b er o f years o f secondary school attainm ent o f the
pop u latio n (see figure II-4 in an n ex II). F urtherm ore, according to literature the percentage o f elderly in the
population, m easu red b y th e share o f p eople o v er 65, should be positively associated w ith ta x revenue.
P ensions b ecom e th e m ain, i f an y at all, source o f incom e fo r old people. T hus it could be argued th a t States,
w ith a h ig h o r rapidly expanding p ro p o rtion o f elderly people, face the pressure to create a pension system
and th is can only be done in a sustainable w ay by increasing taxes. A n o th er com m on indicator is fem ale
lab o r force participation. It is ex pected to be positively correlated w ith ta x revenue, as m ore w om en
em p lo y ed in th e form al m ark et enlarge th e ta x base.
U rb a n iz a tio n sh o u ld also b e p o sitiv e ly c o rre la te d w ith ta x rev en u e. O n th e one h an d , it
in c re a se s c itiz e n s ’ d e m a n d fo r p u b lic g o o d s an d serv ices. O n th e o th e r it te n d s to fac ilita te ta x
a d m in istra tio n (T an zi, 1987). T h e d e n sity o f p o p u la tio n sh o u ld b e p o sitiv e ly lin k e d w ith ta x rev en u e as
it te n d s to red u c e th e a d m in istra tiv e co sts o f ta x c o lle c tio n an d o f c o n tro llin g fo r ta x ev asio n
(A n sari, 1982). A n o th e r d e m o g ra p h ic in d ic a to r th a t m ay b e im p o rta n t in ex p la in in g ta x re v e n u e is
p o p u la tio n g ro w th . R a p id ly g ro w in g p o p u la tio n s p u t ad d itio n a l p re ssu re on th e ta x sy stem in o rd e r to
re g iste r an d m o n ito r n e w ta x p a y e rs (B ah l, 20 0 3 ). T h u s a n e g a tiv e c o rre la tio n a m o n g ta x re v en u e and
th e p o p u la tio n g ro w th rate is ex p ected .
D . O T H E R C O N T R O L V A R IA B L E S
T he em pirical analysis here carried o u t considered a n u m b e r o f additional, say control, variables. O ne o f
th em is th e size o f th e shadow econom y o f the inform al sector. To the extent th a t h ig h er taxes m ay induce
inform ality, th ey au g m en t th e u n d erg ro u nd econom y (Schneider, 2005). A ccording to som e authors, the
increase o f ta x rates in th e last decades — m ainly in m an y developed countries— represents an im portant
incentive fo r com panies an d individuals to w ork in th e inform al econom y (T anzi, Schuknecht, 1997). The
enforced legal o b ligation to p ay tax es is correlated w ith the perception o f the citizens regarding the quality
o f pub lic services supplied b y th e state. I f an increase o f ta x rates is associated w ith an im provem ent o f
pub lic goods, a ta x rise p ro b ab ly w ill face less opposition. A n o th er variable considered the G IN I coefficient.
The acute inequality in L atin A m erica is partly due to the fact th a t taxation does n o t play a strong
redistributive role. T here are m an y p ossible explanations: the ta x rates are to o low , the w eig h t o f direct ta x
on in d irect ta x is to o sm all, o r evasion is high. L ow levels o f ta x revenue are ultim ately a consequence o f the
refusal o f th e elites to contribute to finance public services fo r the rest o f the population. M oreover, lack o f
tru st in th e fairness o f th e ta x system an d in the institutions can increase the shadow econom y, ta x evasion
and even en d an g er political stability (A lesina and Perotti, 1996).
20
IV . R E G R E S S I O N A N A L Y S IS : F U N C T IO N A L S P E C IF IC A T IO N A N D R E S U L T S
T he basic regression m o d el o f ta x revenue as a linear function o f selected econom ic and political indicators
perfo rm ed quite w ell in o u r analysis as show n b y the R 2 and the F -test statistics. R elevant results w ere
d erived using th e follow ing ex ogenous variables: the grow th rate o f G D P p e r capita, the share o f agriculture,
G D P p e r cap ita (expressed in logarithm ), openness o f econom y and the fiscal deficit o f the previous y e ar
and including p o litical v ariables as presence o f civil liberties and political rights (regressions 1 and 2 in
a n n ex III). E x ten d ed v ersions o f th e b asic m odel included o ther control variables (see results o f regressions
5 to 11 in an n ex III).
A s th e first co lu m n o f re g re ssio n 1 in an n ex III show s th e share o f agricu ltu re in G D P is statically
sig n ifican t an d in v ersely related to ta x revenue. A one p erc e n t g ro w th in the share o f ag ricu ltu re m ay
d ecrease ta x rev en u e b y 0.18% . T he im p a c t is relativ ely strong an d it is in line w ith p rev io u s fin d in g s (see
c h a p te r III). T rad e o p en n ess is statically sig n ifican t and has a slig h tly p o sitiv e relatio n sh ip w ith ta x
revenue: a one p e rc e n t in crease in th e open n ess o f the eco n o m y , calcu lated as a sum o f im ports and
e x p o rts as share o f G D P , m a y b o o st ta x revenue by 0.04% . F o r L atin A m erica n co u n tries, and, as R o d rik
(1998) arg u es, a g re a te r size o f fo re ig n trad e relative to G D P in duces an in crease in ta x rev en u es as trad e
ta riffs are easily im p o sed an d m o n ito red . P e r c a p ita G D P g ro w th rate and th e fiscal d eficit o f th e prev io u s
y e a r w ere n o t sig n ific a n t in ex p lain in g th e ta x revenue.
R eg ard in g p o litical in d icato rs, a h ig h e r level o f civil lib erties is asso ciated w ith h ig h e r ta x
rev en u e in a statically sig n ifican t w ay . O ne percen tag e p o in t increase in th e in d ex o f C ivil L ib erties is
lin k ed to an in crease o f 1.21 p ercen tag e po in ts in ta x revenue as a share o f G D P; a resu lt su g gesting th a t
less in terferen ce b y th e State in restrictin g civil lib erties m ay stren g th en th e S ta te ’s cap acity to collect
tax es, p erh ap s lin k ed to o to an in creased p erc ep tio n o f tran sp a ren c y and acco u n tab ility in th e use o f fiscal
reso u rces o r ex p en d itu re b y th e p u b lic ad m in istratio n . In tro d u cin g ad d itio n al political v ariab les in the
m o d el led to co n firm a slig h tly stro n g e r in fluence o f agricu ltu re and o p en n ess on ta x revenue. T he in d ex
o f p o litic a l rig h ts, u n lik e th a t o f civ il lib erties, w as n o t statically sig n ifican t in th e re g re ssio n an aly sis o f
ta x revenue.
In re g re ssio n 3 in an n e x III th e relatio n sh ip b etw een ta x rev enue and G D P p e r capita, calcu lated
in lo g arith m a t c o n sta n t U S d o llars o f y e a r 2000, is h ig h ly sig n ifican t and p o sitiv e . A 10% ch ange in G D P
p e r c a p ita is asso ciated w ith a change in ta x rev enue o f 0.6 po in ts (as a percen tag e o f G D P ). T he last
p o litic a l v ariab le in tro d u ced in th e reg ressio n an aly sis w as regim e du rab ility (see reg ressio n 4). It w as
statically sig n ifican t an d p o sitiv e ly related to ta x revenue su ggesting th a t an ex ten sio n o f one y e a r in the
d u ratio n o f a d u ra b ility o f a reg im e is asso ciated w ith an increase o f ta x rev enue b y 0 .1 % as a p ercentage
o f G D P.
W h e n ad d itio n al co n tro l v ariab les w ere in clu d ed , in alm o st all cases th e ir estim ated coefficien ts
sh o w ed th e e x p ected signs, as rep o rted in th e p rev io u s literature (see ch a p ter III and reg ressio n 5-11 in
a n n e x III). T he level o f e d u catio n is p o sitiv e ly and statically significant. 12 T he e ffect o f school
e n ro llm e n t on ta x rev en u e seem ed , h o w ev er, quite w eak. O ne m ore y e a r o f average school a ttain m en t is
lin k ed to a h ig h e r ta x rev en u e ( o f 0 .0 6 % as a share o f G D P ). Fem ale la b o r force p a rtic ip atio n and th e age
o f p o p u la tio n are sig n ifican t an d p o sitiv e, as w ell as th e d en sity o f po p u latio n . The o th er tw o
d em o g rap h ics v ariab les, p o p u la tio n g ro w th and u rb an izatio n , w ere n o t statically significant. T he shadow
12 Source World Bank, World Development Indicators (WDI) (http://data.worldbank.org/data-catalog/world-developmentindicators).
21
eco n o m y w as statically sig n ifican t w ith a neg ativ e sign. F inally w e h av e to no te th a t the G IN I index, an
in d ic a to r o f in eq u ality , is n o t statically significant. A fte r th e in tro d u c tio n o f th e co n tro l in d icato rs the
re su lt o f th e first tw o reg ressio n s are p erh ap s reinforced. The civil liberties variab le rem ain s sig n ifican t
an d p o sitiv e; ag ricu ltu re an d trad e o p en n ess alw ays have th e e x p ected signs and lose sig nificance only
w h en th e sh ad o w e co n o m y an d sch o o ling are in tro d u ced into th e m odel. G D P p e r c a p ita g ro w th rate
rem ain s sig n ifican t o n ly in th ree cases on elev en reg ressio n s, th e sign is po sitiv e as e x p ected and the
e ffe c t o f e co n o m ic g ro w th o n ta x a tio n ap pears m odest. T he fiscal d eficit o f p rev io u s y ears b ecom es
sig n ifican t, b u t o n ly a t 1% an d o n ly a fte r h av in g in tro d u ced in the re g re ssio n th e sh adow eco n o m y and
th e sch o o lin g v ariab les, b u t th e n u m b e r o f o b serv atio n s is lo w e r (see tab le 4).
W e conclude this section underlining the fact th a t the follow ing variables seem to have a significant,
positive im pact o n ta x revenue: civil liberties, durability o f the political regim e, openness, G D P p er capita,
population density, education and fem ale labor force participation. O ther indicators, such as agriculture and the
shadow econom y, have a negative im pact. M oreover, the grow th o f G D P p e r capita is usually n o t significant
and, w hen it is significant, it seem s to have a sm all im pact on tax revenues. T o sum m arize the results w e point
out th a t the share o f agriculture o v er G D P, civil liberties, G D P per capita, fem ale labor force participation and
age o f population m ay have a strong im pact on ta x revenue. T he effect is m ore m odest for: the level o f
openness o f the econom y, secondary school enrolm ent and density o f population.
22
TA BLE 4
SU M M A RY TA BLE O F R E G R E SSIO N RESU LTS
Tax revenue
Gdpvar per capita
Agriculture
Log gap per capita
Openness
Deficit previous year
Lack of civil liberties
Lack of political rights
Durability
ANGLO
Internal conflict 1
Internal conflict 2
External conflict 3
Total
0.0352
4/12 a
-0.2411
8/11
6.1957
1/1
0.0605
11/12
-0.0014
2/12
-1.2792
10/10
-0.2888
1/1
0.1008
1/1
7.5976
1/1
-4.0625
1/1
-5.1053
1/1
-4.3520 *
1/1
Direct
Principal variables
-0.0031
0/5
**
-0.1208
2/4
***
3.1215
1/1
0.0236
5/5
0.0630
3/5
Political variables
***
-0.9490
3/3
-0.4040
1/1
***
0.0528
1/1
Dummies
***
1.9240
1/1
**
0.3392
1/1
***
-0.1898
1/1
*
-1.5568
1/1
Control variables
**
Indirect
**
***
**
**
***
***
*
*
**
0.0663
5/5
-0.1169
4/4
1.7661
1/1
0.0254
5/5
-0.0061
0/5
-0.2242
0/3
0.0720
0/1
0.0669
1/1
6.2182
1/1
-4.9854
1/1
-3.8647
1/1
-2.7287
1/1
***
**
**
**
*
***
***
***
0.0630
1/1
0.1147
**
Female labor force
1/1
0.8025
***
Oldness
1/1
0.2118
Population growth
1/1
0.0264
***
Population density
1/1
0.0232
Urbanization
1/1
-0.1122
**
Shadow economy
1/1
Source: Elaborated by the author.
a
The GDP per capita growth rate is significant in 4 regressions over 12 and the level of significance is almost
always at 10%.
Notes: The variable must be significant in the majority of regressions to become significant in the summary table, the
4 regression with dummy have been counted as one regression.
The stars ***, **, *, indicate, respectively, the statistical significance at the one, five and ten percent level, calculated
as average of significant values more repeated.
The value of BETA coefficient is the highest among the significant cases.
The insignificant cases are reported in grey.
In the cells it has been reported the Beta value, the statistical significance and the number of significant regressions
over the total of regressions run.
Schooling
23
A. STRUCTURE OF TAXATION
T o extend th e analysis to ta x com position, w e look again a t the results o f regressions 1 to 4, and focus on the
second and th ird colu m n s (see an nex III). In th em the dependent variables are respectively direct tax
revenue (including revenue com ing from personal and corporate incom e tax, property ta x and others direct
tax es) and ind irect ta x revenue (including revenue com ing from ta x on sales o r consum ption, taxes on trade
and o th er indirect taxes). In these cases a h ig h er degree o f civil liberties is linked w ith a h ig h er level o f
direct ta x revenue, w hereas in d irect tax es are n o t associated w ith a hig h er level o f civil liberties. M oreover,
th e relation b etw een th e (second) in d ex o f dem ocratization, political rights, and direct ta x is highly
significant an d p ositive, w hereas it is n o t significant in case o f indirect tax. F urtherm ore, the results indicate
th a t m o re d em o cratic countries ten d to have a hig h er level o f direct taxes (% o f G D P), a fact th a t n o t occurs
in th e case o f ind irect taxes. T hus th e view o f A cem oglu (2006) and B o ix (2003) seem s to prevail regarding
direct taxation. First, dem ocratic countries ten d to carry o u t redistributive policies from the rich to the poor,
building up a w elfare state an d a stro n g er and m ore efficient ta x system based to a greater extent on direct
tax es th a n o n indirect taxes. Second, th e tran sitio n to w ard a dem ocratic g o v ernm ent is associated w ith an
increase o f tax es and pub lic spending (m edian v o te r theory), m oving in the direction o f a b etter
redistribution o f w ealth. R egarding o th er variables, G D P p e r capita g row th rate becom es significant if
associated w ith indirect ta x revenue. T he effect is positive b u t quite w eak.
B. S U B R E G IO N A L D IF F E R E N C E S IN T A X A T IO N : T H E P A R T IC U L A R IT Y O F
C E N T R A L A M E R IC A A N D T H E C A R IB B E A N
A m o n g th e d u m m y v ariables w e in troduced is A N G L O to m ark countries th a t have b een in the p ast E nglish
o r D utch colonies. 13 A s seen in table 5, form er E nglish and D utch colonies have a h ig h er level o f ta x
revenue as percen tag e o f G D P, alm o st 7.5 points m ore th an the o ther countries o f L atin A m erica. B arbados,
Jam aica, S urinam e and St K itties and N ev is show a high level o f direct ta x revenue (m ore th a n 10 p ercen t as
share o f G D P ), b u t B aham as and A n tig u a and B arbuda have a level o f d irect ta x revenue, b elo w 4 % o f
G D P. O n th e o th er hand, fo rm er E nglish and D utch colonies do n o t have a tax atio n structure th a t diverges
significantly from th a t o f th e o th er countries o f the region, regarding the prevalence o f indirect ta x over
others. H ow ever, th ey diverge on the level o f ta x revenue, particularly w h en considering the capacity o f the
C aribbean states to co llect a h ig h er am o u n t o f ta x revenue. C olonial heritages ten d to be reflected in the
im p o rtan t role in th e determ ination o f ta x revenue and in the shape o f public institutions. D ifferences
b etw een th e S panish and E nglish co lonialism are reflected in the fiscal policies o f the countries o f the region
(T hirsk, 1997). E nglish colonies inherited institutions able to penetrate m ore the countryside and create a
larg er form al lab o r m arket, to im pose h ig h e r ta x rates. A ll th ese facts in p a rt help to explain th e h ig h er level
o f ta x revenue o f m an y C aribbean countries com pared to C entral A m erican and South A m erican econom ies,
to co llect ta x revenue in a m o re efficient w ay. It w ill be interesting in a future study to investigate the im pact
o f colonial heritage on institutional elem ents and to analyze h o w th ese elem ents interact w ith and perhaps
jo in tly determ ine th e level o f ta x revenue.
13 Anglo includes the following countries: Antigua and Barbuda, Bahamas, Barbados, Belize, Dominica, Granada, Guyana,
Jamaica, Saint Kitts and Nevis, Saint Lucia, Saint Vincent and the Grenadines, Suriname, Trinidad and Tobago.
24
TA BLE 5
TAX R E V E N U E , D IV ID ED B E T W E E N D IR E C T AND IN D IR E C T TAXES IN
T H E CARIBBEAN IN 2009
Tax revenue over GDP - 2009
Country
Total
Direct
Indirect
C arib b ean
23.7
7.8
14.7
S panish colonies
Dominican Republic
Haiti
12.4
13.1
11.7
3.1
4.1
2.0
8.2
9.0
7.5
B ritish a n d D utch colonies
Antigua and Barbuda
Belize
25.4
19.3
21.6
8.6
3.8
7.8
15.7
15.5
13.8
Bahamas
Barbados
Dominica
Grenada
16.8
32.3
31.6
22.9
1.8
15.7
6.4
6.4
11.9
16.6
25.2
16.5
Guyana
Jamaica
St Kitts and Nevis
St Vincent and the Grenadines
21.6
26.7
26.8
27.1
8.4
10.9
10.1
7.2
13.0
15.8
16.7
19.9
St Lucia
Suriname
Trinidad y Tobago *
28.1
31.1
24.7
8.9
16.1
7.9
19.2
14.5
6.2
C e n tra l A m erica
13.3
13.8
12.4
4.9
4.8
4.6
8.2
8.6
7.4
Costa Rica
El Salvador
10.7
3.2
Guatemala
14.4
4.8
Honduras
17.7
6.3
Nicaragua
10.9
6.0
Panama
Source: ECLAC, elaborated by the author on the basis of a simple average.
7.1
9.7
11.4
5.1
* In total tax revenue, the petroleum revenue is included.
W e also a n a ly z e d th e e ffect o f in ternal and external co n flicts on ta x revenue by in tro d u cin g three
d u m m y v ariab les; c o n flic t 1 rep resen ts th e g ro u p o f c o u n tries in v o lv ed in an in ternal co n flict o f a t least
o ne y e a r d u ratio n since 1980, w h ereas co n flic t 2 in d icates co u n tries in v o lv ed in co n flicts o f at le a st tw o
y e a r duration. F in ally , c o n flic t 3 refers to th o se co u n tries in v o lv ed in ex tern al c o n flic t since 1980. 14 B oth
14 The data come from the UCDP/PRIO Armed conflict Dataset built as a result of the collaboration between the Uppsala
Conflict Data Program (UCDP) and the Centre for study of Civil Wars of the International Peace Research Institute of Oslo
(PRIO). Conflict 2 dummy includes the following countries: Colombia, El Salvador Guatemala, Haiti, Mexico, Nicaragua,
Peru, Bolivarian Republic of Venezuela, whereas dummy 1 includes Panama, Paraguay, Suriname, Trinidad and Tobago, in
addition to countries already included in dummy 2. Dummy 3 covers countries involved in external conflict since 1980 such
as Argentina, Ecuador, Grenada, Panama and Peru.
25
in tern al c o n flic t v ariab les (co n flict 1 an d co n flict 2) w ere h ig h ly sig nificance and n egative. C ountries
in v o lv e d in co n flicts o f at le a st one y e a r te n d to h av e a lo w e r level o f ta x revenue, losing m a in ly revenue
c o m in g from in d ire c t tax es; m o re o v e r in case o f co n flicts th a t la st m ore th a n one y ear, the cap acity o f the
state to co lle c t ta x e s d ecreases ev en further, reducing the ta x rev enue b y up to 5% . T hese results support
th e v ie w o f B ailey an d P ersso n (2008). Such co n flicts reduce th e cap acity o f th e state to c o llec t taxes;
c o m p e titio n a m o n g in te rn a l g ro u p s w ith d iv e rg e n t a im s a n d o p p o site in te re s ts w e a k e n s in stitu tio n s
— m ain ly th e S tate— b y u n d erm in in g its fiscal capacity. O n the o th e r h and, co n trary to som e o f th e ir
p red ictio n s, in L atin A m e ric a ex tern al co n flicts do n o t seem n e ith e r to fo ster fiscal c ap acity n o r to induce
th e states to co lle c t m ore reso u rces in o rd e r to face th e ex tern al th reats and the w ar. O n th e contrary,
ex tern al w ars, like in tern al ones, w e a k e n the S ta te ’s cap acity to co llect ta x e s and th u s reduce th e ta x
revenue. T he d u m m y v ariab le c o n flic t 3 w as n o t h ig h ly sig n ifica n t — o n ly at 10% — w ith an inverse
relatio n to d ire c t ta x revenue. T o su m m arize, in ternal co n flicts in L atin A m erican te n d e d to w e ak e n fiscal
p ressu re, an d reduce rev en u es from in d irec t taxes. In case o f ex tern al co n flict th e re su lt is sim ilar, b u t the
im p a c t is fe lt m o re o n d ire c t ones.
T he level o f ta x a tio n as a p ercen tag e o f G D P in C entral A m e ric an co u n tries in c re ase d in the last
tw o d ecad e b y an av erag e 3.5 poin ts. N ev erth eless, it stood a t only 13.3% o f G D P in 2009, m u ch lo w er
th a n in o th ers c o u n tries o f th e reg io n and am ong th e lo w e st in the w orld. In fact, in th a t p erio d , tax
reven u e in creased sig n ifican tly o n ly in N icarag u a, by 10 po in ts o f G D P . It au g m en ted b e tw ee n tw o and
th ree p o in ts o f G D P in E l S alvador, G u atem ala, H o n d u ras and C o sta R ica, an d rem ain ed ra th e r co n stan t
in P anam a. O n th e o th e r h an d , b o th as reflected b y th e ch an g e in th e share o f ag ricu ltu re in G D P an d in
th e level G D P p e r capita, C en tral A m e rica n co u n tries has in th e last tw o d ecad es (199 0 -2 0 0 9 ) gone
th ro u g h an im p o rta n t tra n sfo rm a tio n and d ev elo p m en t (see ta b le 6). E x c ep t in N ica ra g u a, w h ere it
in creased b y 3 p o in ts, in th e o th e r co u n tries o f th e reg io n the share o f ag riculture co n sid erab ly
d im in ish ed . It d e creased m o re th a n 10 p o in ts in H o n d u ras and G u atem ala, b u t less in C o sta R ica, P an am a
an d E l S alvador, w h ere a g ric u ltu re ’s share in G D P w as alrea d y m u ch low er. D uring th e p erio d covered,
G D P p e r c a p ita rose at an an n u al rate o f 2.6% , a lm o st one p o in t above th e reg ional average th u s reducing
th e gap. In fact, P a n a m a ’s, C o sta R ic a ’s and E l S a lv ad o r’s G D P p e r c ap ita is d efin itely h ig h e r th a n th a t in
N icarag u a, H o n d u ra s an d G u atem ala.
H in rich s, 1966; T anzi, 1992 an d 1987, and C elliah, 1971 fo u n d a po sitiv e co rrelatio n b e tw ee n the
level o f d e v e lo p m e n t an d ta x rev en u e (see ch a p te r III). H o w ev er, the ta x b u rd e n in C entral A m erica
in creased o n ly m o d e stly in th e la st tw o d ecades. W h y does ta x revenue rem ain so m o d e st relativ e to o th er
co u n tries in th e reg io n ? First, C en tral A m e ric a ’s co lo n ial heritag e sh ap ed in special w a y s th e in stitu tio n s
th a t h av e ad v ersely in flu en ced , u n d e rm in ed its fiscal capacities. B ritish co lo n ialism le ft b eh in d relativ ely
e ffic ie n t in stitu tio n s able to co lle c t tax es, w ith a form al la b o r m a rk e t and a m o d e rn fiscal system in the
C aribbean. S econdly, e x cep t fo r H o n d u ras and C o sta R ica, th e co u n tries in the re g io n w ere in v o lv ed in
in tern al co n flicts a fte r 1980. T hese co n flicts w e a k en e d th e institu tio n al cap acity and fiscal cap acity o f the
states to co lle c t tax es. T he o n ly c o u n try o f the region in v o lv ed in an ex tern al w a r h a s b e en P anam a.
A cco rd in g to C ard en as, E sla v a an d R am irez (2010), ex tern al co n flicts do n o t in crease th e fiscal cap acity
o f th e states if th e d u ratio n o f th e c o n flic t is sh o rt o r if the co n flict does n o t involve m an y co u n tries, as
occu rred in th e case o f th e U S in v a sio n o f P a n a m a in 1989.
26
TA BLE 6
SH A R E O F A G R IC U L T U R E IN G D P AND G D P P E R C A PITA , 1990-2009
Country
Central America
Caribbean
South America and Mexico
1990
2000
2005
Share of agriculture in GDP
17.6
14.5
12.0
13.7
9.4
8.3
11.7
8.1
8.7
A v e ra g e
14.3
10.7
Costa Rica
El Salvador
Guatemala
Honduras
Nicaragua *
Panama
12.3
17.4
25.9
22.4
na
9.8
A v e ra g e
17.6
Country
1990
Change 1990 to 2009
11.5
6.9
8.6
-6.0
-6.7
-3.1
9.7
9.0
- 5 .3
9.5
10.5
22.8
15.9
20.9
7.2
8.7
10.5
13.4
13.7
19.0
7.0
7.1
12.5
12.4
12.5
18.8
6.0
-5.2
-4.9
-13.5
-10.0
3.3
-3.8
14.5
12.0
2000
2005
GDP per capita
2 306
2 544
5 274
5 769
3 830
4 095
Central America
Caribbean
South America and Mexico
1 800
4 711
3 147
A v e ra g e
3 219
3 803
Costa Rica
El Salvador
Guatemala
Honduras
Nicaragua
Panama
3 111
1 571
1 446
1 049
682
2 940
1 800
A v e ra g e
2009
11.5
2009
- 6 .0
Annual growth rate (%)
2 911
5 828
4 653
2.6
1.1
2.1
4 136
4 464
1.7
4 057
2 209
1 718
1 141
772
3 938
4 501
2 424
1 762
1 294
843
4 440
5 043
2 566
1 857
1 380
875
5 744
2.6
2.6
1.3
1.5
1.3
3.6
2 306
2 544
2 911
2.6
Source: World Bank, World development indicators.
Note: GDP per capita annual growth rate calculated at constant US Dollars off 2000.
F in ally , a th ird reaso n fo r th e low lev el o f ta x atio n in C entral A m e ric a is th e w eak n ess o f
d em o cracy . In fact, in C en tral A m e ric a th e civil liberties and th e political rig h ts in d e x have n o t im proved,
o n av erag e, from 1990 u n til 2 0 0 9 (see ta b le 7). T he first one has w o rsen ed , w h ereas the seco n d one has
rem ain ed u n ch an g ed . In p articu lar, alth o u g h th ere h as b een , du rin g the p erio d an aly zed , a h ig h lev el o f
civ il lib erties in C o sta R ic a an d P an am a, in N icarag u a, H o n d u ras and G u atem a la the in d ic a to r o f civil
lib erties did n o t im p ro v e. A lth o u g h it im p ro v ed in E l S alvador, it d id n o t reach a satisfacto ry level.
R eg ard in g th e in d ic a to r o f po litical rights, C o sta R ica, P an a m a an d E l S alv ad o r show a h ig h e r level, b u t
in N icarag u a, H o n d u ra s an d G u atem ala th e in d e x d eterio rated in the last tw o decades.
27
TA BLE 7
C IV IL L IB E R T IE S AND P O L IT IC A L R IG H T S INDEXES
Index of civil liberties
Central America
Caribbean
1990
2.8
2.3
2000
2.8
1.9
2005
2.7
2.0
2009
3.0
1.9
South America and Mexico
3.0
3.0
2.5
2.6
A verage
2.7
2.6
2.4
2.5
Costa Rica
El Salvador
Guatemala
Honduras
1.0
4.0
4.0
3.0
2.0
3.0
4.0
3.0
1.0
3.0
4.0
3.0
1.0
3.0
4.0
4.0
Nicaragua
Panama
3.0
2.0
3.0
2.0
3.0
2.0
4.0
2.0
2.8
2.8
2.7
3.0
2000
2.2
1.9
2005
2.3
2.0
2009
2.7
1.7
A verage
Index of political rights
Caribbean
1990
2.7
2.3
Central America
South America and Mexico
2.3
2.5
2.4
2.5
A verage
2.4
2.2
2.2
2.3
Costa Rica
El Salvador
Guatemala
Honduras
1.0
3.0
3.0
2.0
1.0
2.0
3.0
3.0
1.0
2.0
4.0
3.0
1.0
2.0
4.0
4.0
Nicaragua
Panama
3.0
4.0
3.0
1.0
3.0
1.0
4.0
1.0
A verage
2.7
2.2
2.3
2.7
Source: Freedom House.
A s seen in c h a p te r III, in co u n tries w h ere th e in d ex su ggests a, say, low level o f dem o cracy , large
p a rt o f th e citizen s m a y be ex clu d ed from th e k ey d ecisio n m ak in g p ro cess. A n d perh ap s th ere are few o r
p ra c tic a lly n o p o litic a l p arties th a t rep resen t th e interests o f th e electo rate, b ein g larg ely in flu en ced b y the
v e ste d in terests o f lo b b ies an d elites (G ro ssm an a n d H elp m an , 1994). A cco rd in g to th is v iew , elites ex ert
p o w e r an d p ressu re o n p o litic a l p arties in o rd e r to d efen d th e ir in terests, and in p a rtic u la r to p re v en t tax es
from rising an d k eep th e ir sp ecial p riv ileg es and e x em p tio n s. T h ere is a p ercep tio n th a t in m an y L atin
A m e ric a n c o u n tries p o w erfu l lan d o w n ers — and o th e r p riv ile g ed classes— has b e en able to b lo c k ta x
refo rm s th a t in ten d to raise d irect ta x e s o n lan d and in co m e (T he W o rld B an k G roup, 2 0 0 8 ).
S u m m arizin g , th e lo w level o f tax e s in C e n tra l A m e ric a is m ain ly due to a p o o rly w o rk in g o r
in effectiv e d e m o cracy lin k ed to a w eak rep resen tativ en ess o f political p arties, to h isto rical heritag e — the
S pan ish co lo n ialism — an d to th e sequels o f in ternal co n flicts th a t h it th e reg io n du rin g th e 1980s and
1990s. O n th e o th e r h an d , th e in g en eral h ig h e r lev el o f th e ta x rev en u es relative to G D P in th e C aribbean,
28
m a in ly in th e fo rm e r B ritish an d D u tch co lo n ies, reflects a m ore effective d em o cracy , a lo w er n u m b er and
less in ten se co n flicts, an d so m ew h at b e tte r institu tio n s due to th e co lo n ial legacy.
C . T H E T A X E F F O R T IN D E X
In line w ith C helliah (1971), B ahl (1971), C helliah, B ass, K elly (1975) and G u p ta (2007) w e estim ated the
ta x effo rt (as p ercentage o f G D P ) as th e ratio o f actual ta x revenue to potential revenue. The potential
revenue w as calculated b y using th e regression 1 (see an nex III), dropping the n o t statically significant
variables. I f th e in d ex o f ta x effo rt is g reater th an one, it m eans th a t the countries are collecting a hig h er
am o u n t o f ta x revenue th a n the am o u n t th a t is predicted, estim ated b y the regression analysis o f the
lo n g -te rm d eterm inants o f ta x revenue considering the co u n tries’ specific econom ic, social and institutional
conditions (Piancastelli, 2001).
W e in serted in th e reg ressio n fo r each country, th e average v alue o f th e th re e in d icato rs fo r the
p erio d s 1990-1994 an d 2 0 0 5 -2 0 0 9 , in o rd e r to calcu late th e p o te n tial ta x burden. T he average v alue o f
five y ears w as u se d in o rd e r to m in im ize the p o ssib le e ffe ct o f ex trao rd in ary events. T able 8 rep o rts the
ta x effo rt in d e x fo r th e co u n tries o f th e region. F o r th e p erio d o f analysis, th e resu lts su g g est th a t countries
th a t e x h ib it a h ig h e r ta x effo rt also h av e a h ig h e r ta x burden. T he ta x effo rt is on averag e h ig h e r in the
C arib b ean th a n in th e re st o f th e reg ion. In the C aribbean, w ith the ex cep tio n o f B ah am as and
D o m in ican R ep u b lic, th e in d e x is o v e r one, th u s in d icatin g th a t th ese co u n tries have b een able to increase
th e ir cap acity to co lle c t ta x rev en u e o v e r and above w h a t it w o u ld have b een p re d icte d b y th e lo n g -term
stru ctu ral d eterm in an ts. C en tral A m erican eco n o m ies have an in d e x o f ta x e ffo rt lo w e r th a n one, w ith
N ic a ra g u a b ein g th e ex cep tio n th a t in th e la st d ecades co n sid erab ly in creased its ta x burden. T hus, this
su b reg io n faces k e y o b stacles th a t im p ed e h e r to ev en achieve th e, actu ally low , lev el o f tax a tio n g iv en by
its reven u e p o ten tial. M ex ico an d S outh A m e ric an co u n tries show on average a lev el o f ta x effo rt close to
o ne b u t w ith large v arian ce. F o r in stan ce, M exico an d E cu ad o r have a lev el o f ta x revenue ra th e r m o d e st
c o m p ared w ith th e ir stru ctu ral p o ten tial, u n like B razil and P lu rin atio n al State o f B olivia, w h ich show a
h ig h value.
C o m p arin g th e p erio d s 1990-1994 and 2 0 0 5-2009, the fiscal e ffo rt in d ex h as in creased in the
th ree su b -reg io n s, w ith som e ex cep tio ns. In C entral A m erica th e in d ex d ecreased in P an am a and
H o n d u ras, in th e C arib b ean it d im in ish ed in B ah am as, G u y an a and S an ta L ucia, and in th e re st o f the
reg io n it d e creased in M ex ico an d C hile. In ad d itio n it seem s th a t th e C arib b ean co u n tries, in p a rtic u la r
th o se th a t h av e b een in th e p a st E n g lish o r D u tch colonies, show a relativ ely strong fiscal cap acity, w ith
m o re e ffic ie n cy to co lle c t ta x revenue. C entral A m erican eco n o m ies have a red u ced cap acity to c o lle ct ta x
revenue. T h u s, th e an aly sis o f th e ta x effo rt co n firm s o r ev en rein fo rces the fin dings o f the prev io u sly
d ev e lo p e d study.
29
TABLE 8
TAX EFFORT INDEX
The effort index calculation
1990-1994
Country
Dominica
Suriname
Barbados
Trinidad y Tobago
Tax revenue
(% of G D P)
Real
Potential
24.1
na
28.2
18.1
15.2
20.2
Index of tax
effort
1.33
na
1.40
2005-2009
Variance
between real
and potential
tax revenue
5.97
na
8.03
Tax revenue
(% of G D P)
Real
Potential
30.0
29.0
32.8
18.5
18.5
21.8
Index of tax
effort
1.62
1.57
1.51
Variance
between real
and potential
tax revenue
11.50
10.5
11.06
na
19.8
na
na
28.3
20.2
1.40
8.04
Jamaica
20.6
18.6
1.11
2.01
24.6
18.0
1.36
6.55
St Vincent and the
Grenadines
St Kitts and Nevis
Granada
22.8
16.4
22.6
18.8
21.0
18.4
1.21
0.78
1.23
4.01
-4.60
4.20
27.0
28.1
23.5
20.3
21.2
19.1
1.33
1.33
1.23
6.68
6.90
4.46
St Lucia
26.7
19.8
1.35
6.97
25.6
21.5
1.19
4.18
na
18.4
na
na
21.7
18.7
1.16
3.0
22.2
17.3
1.28
4.91
20.8
18.7
1.11
2.12
17.6
21.1
0.84
-3.47
20.6
20.6
1.00
0.03
10.5
16.3
0.64
-5.84
14.7
17.7
0.83
-3.03
15.8
19.4
0.81
-3.62
16.7
21.0
0.79
-4.34
Belize
Guyana
Antigua and
Barbuda
Dominican
Republic
Bahamas
Haiti
Average
5.3
na
na
na
10.6
na
na
na
20.7
19.0
1.09
1.69
24.5
19.7
1.25
4.83
Nicaragua
10.9
13.3
0.82
-2.44
17.7
15.5
1.14
2.18
Honduras
12.9
14.6
0.88
-1.72
15.3
17.7
0.87
-2.38
9.8
14.2
0.69
-4.40
12.9
16.1
0.80
-3.18
Guatemala
8.4
10.9
0.77
-2.46
11.6
14.6
0.80
-2.97
Costa Rica
11.6
17.7
0.65
-6.15
14.4
19.9
0.73
-5.45
10.6
21.2
0.50
-10.59
10.2
20.8
0.49
-10.64
10.7
15.8
15.3
14.4
0.70
1.09
-4.63
1.31
13.7
23.0
17.4
15.9
0.79
1.45
-3.74
7.10
E l Salvador
Panama
Average
Plurinational State
of Bolivia
na
14.3
na
na
23.2
16.6
1.40
6.6
Argentina
Brazil
12.1
15.0
0.81
-2.86
17.1
16.7
1.03
0.42
Uruguay
14.4
16.6
0.87
-2.23
18.2
18.1
1.01
0.10
Bolivarian
Republic of
Venezuela
15.4
16.6
0.93
-1.18
14.9
15.6
0.95
-0.75
Chile
16.6
17.1
0.97
-0.51
18.6
19.8
0.94
-1.26
Peru
12.0
13.3
0.90
-1.32
14.6
16.0
0.91
-1.42
Colombia
8.2
13.0
0.63
-4.83
13.1
15.1
0.87
-1.93
Paraguay
10.5
14.3
0.73
-3.87
13.2
15.7
0.84
-2.48
Ecuador
7.3
na
na
na
11.4
17.0
0.67
-5.6
Mexico
10.1
14.8
0.68
-4.67
8.8
17.1
0.52
-8.28
Average
12.8
15.0
0.85
-2.24
16.0
16.7
0.96
-0.68
Source: ECLAC, elaborated by the author.
na: not available.
30
V. CONCLUSIONS
A s is w ell k now n, in p anel d ata regressions the sam ple o f countries chosen and the period covered fo r the
analysis m ay affect th e results on th e significance o f som e o f the, say, exogenous variables considered as is
th e case on such studies on tax atio n . T aking account these considerations, this p ap e r focus on taxation
exclusively in L atin A m erica an d C aribbean countries. D ue to reasons related to d ata availability and
com parability, w e restricted th e analysis to the period 1990-2009. T he m ain purpose o f the study w as to
identify the long-term v ariables — including historical, econom ic, social and political factors— th at
significantly influence tax atio n in th e countries o f the region. A second purpose o f the study w as to have a
b etter understan d in g o f the p otential regional differences in the ta x effort o f each country, m easured b y the
gap b etw een its actual and its p otential ta x revenue.
T he em p irical resu lts o f th e p an el m o d els here b u ilt and eco n o m etrically te ste d in dicate th a t G D P
p e r c a p ita an d o p en n ess o f th e e c o n o m y are p o sitiv e ly related to ta x rev enue in a statically significant
w a y . T he share o f ag ricu ltu re o v e r G D P an d th e size o f th e sh adow eco n o m y are also statically
sig n ifican t, b u t n e g a tiv e ly asso c ia te d w ith ta x rev en u e. O n th e o th e r h and, p e r c a p ita G D P g ro w th rate
an d lag g ed fiscal d eficits o f th e p rev io u s y e a r w ere alm o st alw ays n o t statically sig n ifican t. H o w ev er, the
e stim ated m o d els th a t fo cu sed ex clu siv ely on in direct ta x e s id en tified G D P g ro w th as a sig n ifican t
in flu en ce w ith a p o sitiv e sig n . In line w ith p rev io u s literatu re, o u r w o rk s in d icates th a t in L atin A m erica
th e level o f d ev elo p m en t, as p ro x y b y th e inverse o f the share o f ag ricu ltu re in G D P and b y G D P p e r
capita, h as a strong an d p o sitiv e in flu en ce on th e ta x b u rd e n . M o v in g to socio dem o g rap h ics determ in an ts
th e stu d y fo u n d o u t th a t th e lev el o f ed u cation, fem ale la b o r force p a rticip a tio n and th e p o p u latio n d en sity
h av e a p o sitiv e an d sig n ifican t im p act on ta x revenue, b u t th e level o f u rb a n iza tio n an d th e rate o f
p o p u la tio n g ro w th w ere n o t sig n ifican t. A m o n g o th e r in d icato rs an aly zed , an esp ec ially strong p o sitive
a sso ciatio n w as fo u n d b e tw e e n ta x a tio n and th e share o f w o m en em p lo y ed in the form al m a rk et and by
th e share o f peo p le o v e r 6 5 y e a rs o ld . W ith the ex cep tio n o f u rb an izatio n , th a t is u su a lly sig n ifican t and
p o sitiv e ly asso ciated w ith ta x rev en u e, th e others socio dem o g rap h ics variab les are in line w ith p revious,
recen t stu d ies.
C o n cern in g th e p o litical v ariab les, a h ig h e r degree o f civil lib erties and m ore p o litic al stability, as
m e a su re d b y th e d u ra b ility o f th e p o litical regim e, are asso ciate d w ith h ig h e r ta x re v en u e. T he p o litical
rig h ts in d e x is n o t statically sig n ifican t, u n like P ro feta and S cabrosetti (2010) th a t fin d a p o sitive
c o rrelatio n b e tw e e n th e m . H o w ev er, i f w e d e ep en th e an aly sis on th e structure o f ta x atio n , w e fin d th at
th e lev el o f p o litical rig h ts b ec o m e s h ig h ly sig n ifican t and po sitiv e as a d e term in an t o f d irect ta x
rev en u es, b u t n o t o f in d ire c t tax es. It w as also su g g ested th a t in th e reg io n th e in d icato rs asso ciated w ith a
say m o re d em o c ra tic g o v ern m en ts u su a lly reg ister a h ig h e r lev el o f d irect ta x rev en u es, p erh ap s due to
th e ir c o m m itm e n t to red istrib u tiv e policies. O n th e co n trary, a h ig h e r level o f in d ire ct ta x e s is n o t u su ally
lin k ed w ith an in creased d eg ree o f civ il lib erties and p o litic al rights. T his re su lt on th e po sitiv e co rrelatio n
b e tw e e n level o f d e m o cracy an d stru ctu re o f ta x a tio n d iverges fro m P ro feta an d S cabrosetti (2010), w ho
fin d n o ev id en ce o f th a t. W e m u st u n d erlin e th a t th e ir study covers less co u n tries (19 w ith only the
D o m in ican R ep u b lic am o n g th e C arib b ean) o v e r a sh o rter p erio d o f tim e (1990-2004).
W e fo u n d th a t th e stru ctu re o f tax atio n does n o t diverge sig n ifican tly am ong the th ree regions
a n aly zed (S o u th A m e ric a p lu s M ex ico , C entral A m e ric a an d the C aribbean); all have a p red o m in an ce o f
in d ire c t ta x e s o v e r d ire c t ones. M o reo v er, w e have id en tified h isto rical and p o litica l factors w h ich h elp to
e x p lain reg io n al d ifferen ces on ta x a tio n : th e co lo n ial h eritag e, th e in ternal co n flicts th a t have hit the
reg io n fro m th e b eg in n in g o f th e eig h ties u n til the end o f the last cen tu ry , and a n u m b e r o f ind icato rs o f
th e ex te n t o f d e m o c ra c y . T hese facto rs, in d ifferen t w ays, shape the fiscal cap acity o f th e co u n tries and
31
th e ir d iffe re n t ta x b u rd e n s an d co m p o sition. T he C arib b ean region, in p a rtic u la r th e fo rm er B ritish and
D u tch co lo n ies, sh o w s a h ig h e r level o f ta x revenue, w h ich is clo ser to th e w estern standards. T his is due
to th e ir co lo n ial leg acy , a h ig h e r lev el o f civil lib erties and p o litic al rights and a red u c ed p resen ce o f
co n flicts. O n th e co n trary , C en tral A m e rica n co u n tries have one o f th e lo w est levels o f ta x rev enue in the
w o rld , rep resen tin g o n ly 13.3% o f G D P in 2009. T he ta x effo rt m o d el ca lcu latio n s here carried out,
co n firm o u r findings.
In ad d itio n , d esp ite th e in creased ta x rev en u e, th e im p ro v em e n t in th e level o f d ev elo p m en t and
th e eco n o m ic g ro w th o b serv ed in th e p e rio d an aly zed (1990-2009), th e ta x b u rd en in L atin A m erica and
in th e C arib b ean h as rem ain ed to o m o d e st co m p ared w ith o th e r w o rld regions. C o n seq u en tly , p u blic
ex p en d itu res are v ery lim ited an d th e level o f in eq u ality is acute as th e fiscal system does n o t ex e rt a
re le v a n t red istrib u tiv e im pact. O ne o f th e m a in ch allen g es o f the L atin A m e ric an co u n tries in th e n e a r
future is to stren g th en th e ir ta x a tio n c ap acity in o rd e r to im prove p u b lic services, reduce in eq u alities and
p ro m o te su stain ab le eco n o m ic g ro w th . In fact, th is m ay be th e o n ly w ay to reco n cile eco n o m ic g ro w th
an d eq u ality . F iscal p o lic y is a k e y to o l. A fiscal reform th a t enlarg es the ta x basis, creates an efficien t
fiscal sy stem an d elim in ates special ta x regim es an d ex em p tio n s is a n ec essary step fo r m o st co u n tries in
th e reg io n , an d in p a rtic u la r C en tral A m erica, to have su fficien t fiscal reso u rces th a t m a y help to push
fo rw ard an effectiv e a g e n d a fo r d e v e lo p m e n t th a t does pro m o te g ro w th , ensu res a m a jo r red u ctio n in
p o v e rty an d in eq u ality .
33
B IB L IO G R A P H Y
Acemoglu, D. and J. A R obinson (2006), “Economic origins o f dictatorship and dem ocracy”, Cambridge
University Press.
Alesina, A. and R. Perotti (1996), “Income distribution, political instability and investment”, European Economic
Review, 78, 796-805.
Ansari, M. (1982), “Determinants of tax ratio: A cross country analysis”, Economical and Political Weekly, 19 June,
pp. 1035-1042.
Bahl, R. W. (1971), “A regression approach to tax effort and tax ratio analysis”, IMF Staff Papers, 18 (3), 570-612.
Barro, R. J. (1979) “On the determination o f public debt”, Journal o f Political Economy, 87, 940-71
Beseley, T., and T. Persson (2008). “Wars and State Capacity”. Journal o f the European Economic Association,
6 (2-3), 522-30.
Bird, R. M., J. M artínez-Vásquez and B. Torgler (2004), “ Societal institutions and tax effort in developing
countries”, Andrew Young School o f Policy Studies, Georgia State University, Working Paper 04-06.
Boix, C. (2003), “Democracy and redistribution”, Cambridge University Press.
Cárdenas, M.., M. Eslava and S. Ramírez (2010), “External wars, internal conflicts and State capacity: Panel data
evidence”, Latin America Initiative at Brookings.
Chelliah, R. J. (1971), “Trends in taxation in developing countries”, IMF Staff Papers, 18, 254-331.
Chelliah, R. J., H. J. Bass and K. R. Kelly (1975), “Tax ratio and tax effort in developing countries”, IMF Staff
Papers, 22 (1), 187-205.
Grossman, G. M. and E. Helpman (1994), “Protection for Sale”, American Economic Review, 84, 4, 833-850.
Gupta, A. S. (2007), “Determinants o f tax revenue in developing countries”, IMF Working Paper (WP/07/184).
Hinrichs, H. H. (1966), “A general theory o f tax structure: Change during development”, Cambridge, MA: Harvard
Law School International Program.
Hsiao, C. (2004), “Analysis o f panel data”, Cambridge University Press, Second Edition.
Keen, M. and A. Simone (2004), “Tax policy in developing countries: Some lessons from the 1990s and some
challenges ahead”, in Helping Countries Develop: The Role o f Fiscal Policy, edited by Sanjeev Gupta, Benedict
Clements and Gabriela Inchauste, 302-352. Washington, D. C., International Monetary Fund, 2004.
Mahdavi, S. (2008), “The level and composition o f tax revenue in developing countries: Evidence from unbalanced
panel data”, International Review o f Economics and Finance 17, pp. 607-617.
M ulligan, C. B., R. Gil and X. Sala-i-Martin (2004), “Do democracies have different public policies than non
democracies?”, Journal o f Economic Perspectives, 18(1), pp. 51-74.
Piancastelli, M. (2001), “M easuring the tax effort o f developed and developing countries. Cross country panel data
analysis-1985/1995”, Instituto de Pequisa Económica Aplicada (IPEA).
Profeta, P. and S. Scabrosetti (2010), “The political economy o f taxation: Lessons for developing countries”,
Edward Elgar Publishing Limited.
Ranjan, V. and G. Agrawal (2011), “FDI inflow determinants in BRIC countries: A panel data analysis”,
International Business Research (4), 2011.
Rodrik, D. (1998), “Why do more open economies have bigger governments?”, Journal o f Political Economy 106,
pp. 997-1032.
Schneider, F. (2005), “ Shadow economies around the world: W hat do we really know?" European Journal o f
Political Economy, Elsevier, 21(3), pp. 598-642.
Stotsky, J. G. and A. W oldeMariam (1987), “Tax effort in Sub-Saharian Africa”, IMF Working Paper 97/107,
International M onetary Found, W ashington, D. C.
Tanzi, V. (1987), “Quantitative characteristics o f tax systems o f developing countries”, in D.M.G. Newbery and N.
H. Stern Editions, The Theory o f Taxation for Developing Countries, New York, Oxfort University Press.
(1992), “ Structural factors and tax revenue in developing countries: A decade o f evidence”, I. Goldin and
L.A. W inters Editions, Cambridge University Press.
Tanzi, V. and L. Schuknecht (1997), “Reconsidering the fiscal role o f government: The international perspective”,
American Economic Review 87 (2), pp. 164-168.
Thirsk, W. (1997), “Tax Reform in Developing Countries”, W ashington, D. C., World Bank.
34
Torres-Reyna, O., “Getting started in fixed/random effects models using R ”, Princeton University, Data and
Statistics Services.
Wittman, D. (1989), “Why democracies produce efficient results”, Journal o f Political Economy, 97, pp. 1395-1424.
The W orld Bank Group, in collaboration w ith DFID (2008), The Political Economy o f Taxation in Developing
Countries. Challenges to Practitioners.
35
ANNEXES
A N N E X I: E C O N O M I C V A R IA B L E S
TABLE I-1
E C O N O M IC V A RIABLES
Economic data
country
Tax revenue as %
of G DP
G D P per capita
growth rate
Average
1990-1999
Average
2000-2009
Average
1990-1999
Total
15.4
18.0
1.9
Caribbean
19.5
22.3
8.7
11.7
11.0
13.9
6.5
9.4
B ritish and
Dutch
colonies
Antigua and
Barbuda
21.4
24.0
2.1
17.7
19.2
1.1
2.0
Bahamas
15.8
15.7
0.3
-1.3
Barbados
29.2
32.3
1.8
-0.3
21.9
2.3
Spanish
colonies
Dominican
Republic
Haiti
Belize
Average
2000-2009
G D P per capita U S D
costant 2000
Share of agriculture
over G DP
Average
1990-1999
Average
2000-2009
Average
1990-1999
1.8
3 781
4 556
12.4
2.0
1.5
4 625
5 677
1.0
1.3
1 289
4.2
3.5
-2.3
-0.9
1.5
Average
2000-2009
Trade openness
Average
1990-1999
Average
2000-2009
9.1
84.5
87.2
11.7
8.1
113.0
107.3
1 743
10.6
6.9
58.4
63.4
2 142
3 091
10.6
6.9
436
394
5 139
6 283
11.8
8 205
9 879
16 022
17 845
8 623
0.8
2 840
78.3
71.8
38.4
54.9
8.2
121.4
114.0
4.0
3.6
167.1
129.5
2.8
1.8
101.8
95.1
9 969
6.8
3.7
108.4
119.2
3 596
17.6
14.9
108.1
120.9
108.6
Dominica
23.7
27.3
2.1
1.6
3 398
3 954
21.0
18.1
119.9
Grenada
22.2
23.0
2.7
0.3
3 133
4 430
10.2
6.9
110.2
102.7
Guyana
20.4
19.6
5.5
1.0
793
981
37.5
27.8
217.3
201.0
Jamaica
20.8
23.7
1.0
1.0
3 488
3 705
8.2
6.1
105.0
97.5
St Kitts and
Nevis
19.1
25.5
4.2
0.6
6 336
7 815
5.6
2.9
126.9
113.2
St Vincent
and the
Grenadines
23.1
26.2
3.2
3.6
2 583
3 771
14.3
8.5
125.5
108.2
St Lucia
24.2
23.8
1.9
0.4
4 245
4 648
10.3
5.1
137.1
118.7
28.6
-0.6
3.8
1 964
2 289
12.8
7.0
62.9
67.4
Suriname
Trinidad y
Tobago
19.4
24.7
2.2
5.8
5 175
8 792
2.4
0.8
87.8
100.2
Central
Am erica
11.2
12.9
2.2
2.1
2 038
2 570
17.3
12.3
85.1
97.1
Costa Rica
12.0
13.8
3.0
2.4
3 531
4 534
12.8
8.5
80.8
95.8
69.9
E l Salvador
10.1
11.9
3.7
1.7
1 884
2 421
14.6
10.6
55.7
Guatemala
9.4
11.7
1.8
0.9
1 567
1 786
24.5
14.4
43.4
64.1
Honduras
13.1
14.5
0.5
2.1
1 097
1 277
21.2
13.6
87.2
125.6
Nicaragua
12.2
16.0
1.1
1.4
679
831
22.8
19.3
66.3
86.0
Panama
10.5
9.5
3.2
4.3
3 ,471
4 573
8.0
7.1
177.1
141.0
36
Table I-1 (concluded)
Economic data
country
Tax revenue as %
of G DP
Average
Average
1990-1999
2000-2009
G D P per capita
growth rate
Average
1990-1999
Average
2000-2009
G D P per capita U S D
constant 2000
Share of agriculture
over G DP
Average
1990-1999
Average
2000-2009
Average
1990-1999
Average
2000-2009
Trade openness
Average
1990-1999
Average
2000-2009
South
A m erica
and Mexico
13.0
14.9
1.7
2.2
3 581
4 111
10.3
8.6
45.4
54.5
Argentina
12.3
15.3
3.8
2.8
7 216
8 174
5.9
8.6
18.7
38.3
Plurinational
State of
Bolivia
15.4
18.6
1.6
1.9
942
1 073
16.4
14.4
48.7
61.9
Brazil
18.4
22.2
0.8
2.0
3 503
4 002
6.9
6.1
17.2
25.7
Chile
17.0
18.1
4.9
2.5
4 063
5 543
8.1
4.6
57.4
72.0
15.1
Colombia
8.7
12.0
0.6
2.5
2 488
2 798
Ecuador
7.5
10.9
-0.2
3.5
1 330
1 545
Mexico
9.6
9.4
1.4
0.4
5 214
6 149
Paraguay
11.6
12.6
0.1
0.3
1 432
Peru
12.8
13.6
2.2
2.0
Uruguay
Bolivarian
Republic of
Venezuela
14.8
17.8
3.4
2.4
14.6
13.3
-0.2
1.8
5 087
Source: ECLAC and World Bank.
8.3
35.4
35.8
7.0
56.5
66.2
6.2
3.9
49.0
56.9
1 376
21.9
19.9
95.3
98.1
1 854
2 393
8.8
7.6
30.1
41.9
6 259
7 131
8.1
9.8
38.3
51.9
5 037
5.2
4.2
52.3
50.9
37
ANNEX II
ECONOMIC VARIABLES FIGURES
FIG U R E II-1
C O R R E L A T IO N B E T W E E N TAX REV EN U E AND GDP
P E R C A PITA
GDP per capita in logarithm
•
•
1 95% CI
South America and Mexico
Central America
■
Fitted values
Caribbean
Source: ECLAC.
FIG U R E II-2
C O R R E L A T IO N B E T W E E N TAX R EV EN U E AND TRA D E O PENN ESS
O F AN E C O N O M Y
Lack of civil liberties
•
•
Source: ECLAC.
1 95% CI
South America and Mexico
Central Amercia
■
Fitted values
Caribbean
38
FIG U R E II-3
C O R R E L A T IO N B E T W E E N TAX R EV EN U E AND C IV IL
L IB E R T IE S INDEX
• TTO
•
V C T LCA
ja g r
Bl z a
• BRA
• ATG
• BHS
I VEN
• N IC
•
do
M
HND
cri pry
• COL
G TM?LV
M E ^E c u
• HTI
50
• PAN
100
Openess
150
1 95% CI
200
Fitted values
•
South Am erica and Mexico
•
Central America
•
Caribbean
Source: ECLAC and Freedom House, Inc.
FIG U R E II-4
C O R R E L A T IO N B E T W E E N TAX R EV EN U E AND L E V E L O F ED U C A TIO N
Education
1 95% CI
•
South A m erica and Mexico
•
Central Am erica
Source: ECLAC and World Bank.
•
Fitted va lues
C aribbean
39
A N N EX III
R E S U L T S O F T H E R E G R E S S IO N S
T A B L E III-1
Total
Regression 1
Tax revenue
Const
Gdpvar per capita
Agriculture
Openness
Deficit previous year
L ack of civil liberties
Direct
re
18.9022
2.3880
0.0352
0.0245
-0.1810
0.0752
0.0382
0.0160
-0.0014
0.0608
-1.2190
0.3021
Number of observations
Countries
R2 within
R2 between
R2 overall
re
***
0.0000
0.1500
**
0.0160
**
0.0170
0.9810
***
0.0000
6.7685
1.0157
-0.0031
0.0195
-0.0732
0.0447
0.0209
0.0086
0.0580
0.0244
-0.9490
0.2072
495
31
0.216
0.350
0.346
Gdpvar per capita
Agriculture
Openness
Deficit previous year
L ack of political rights
Number of observations
Countries
R2 within
R2 between
R2 overall
495
31
0.159
0.230
0.222
***
0.0000
0.8750
0.1020
**
0.0150
**
0.0180
***
0.0000
10.7387
1.2471
0.0624
0.0154
-0.0985
0.0385
0.0207
0.0084
-0.0061
0.0345
-0.2242
0.2137
0.1220
***
0.0040
**
0.0200
0.9390
0.2160
5.4098
1.1904
0.0001
0.0190
-0.1109
0.0504
0.0236
0.0113
0.0534
0.0307
-0.4040
0.1255
440
30
0.206
0.143
0.150
0.8590
0.2940
Indirect
re
***
0.0000
***
0.0000
***
0.0000
***
0.0100
**
0.0130
440
30
0.144
0.224
0.212
Direct
re
16.9030
2.4133
0.0392
0.0254
0.2337
0.0806
0.0403
0.0173
0.0044
0.0571
-0.2888
0.2335
re
440
30
0.298
0.121
0.142
Total
Regression 2
Tax revenue
Const
Indirect
re
***
0.0000
0.9950
**
0.0280
**
0.0360
*
0.0810
***
0.0010
10.3487
1.2592
0.0668
0.0156
-0.1151
0.0374
0.0189
0.0092
-0.0007
0.0353
0.0720
0.1491
440
30
0.150
0.095
0.117
***
0.0000
***
0.0000
***
0.0020
**
0.0390
0.9850
0.6290
40
Table III-1 (continued)
Total
Regression 3
Tax revenue
Const
Gdpvar per capita
Openness
Deficit previous year
Lack of civil
liberties
Log gdp p er capita
Number of
observations
Countries
R2 within
R2 between
R2 overall
Direct
re
***
-35.2798
10.3925
0.0296
0.0272
0.0436
0.0141
-0.0290
0.0533
0.0010
0.2760
***
0.0020
0.5870
**
-0.5494
0.2378
6.1957
1.3501
0.0210
***
0.0000
536
32
0.361
0.269
0.290
re
-20.4774
4.7803
-0.0103
0.0181
0.0219
0.0073
0.0306
0.0239
-0.5313
0.1747
3.1215
0.5915
Gdpvar per capita
Agriculture
Openness
Deficit previous year
Lack o f civil liberties
Durability
***
0.0000
0.2000
-5.3654
5.5506
0.0625
0.0154
0.0254
0.0077
-0.0217
0.0356
***
0.0020
***
0.0000
-0.1215
0.1970
1.7661
0.7285
0.5700
***
0.0030
481
31
0.395
0.133
0.180
Total
Regression 4
Tax revenue
Const
Indirect
re
13.7021
2.2214
0.0440
0.0293
-0.0879
0.0598
0.0341
0.0125
0.1402
0.0862
-0.9250
0 3664
0 1008
0 0355
re
***
0.0000
0.1320
0.1420
***
0.0060
0.1040
**
0 0120
***
0 0050
4.8296
1.5219
0.0240
0.0167
-0.0507
0.0452
0.0170
0.0077
0.0653
0.0448
-0.7036
0 2699
0 0528
0 0306
0.5420
0.5370
**
0.0150
481
31
0.158
0.265
0.252
Direct
re
0.3340
***
0.0000
***
0.0010
Indirect
re
***
0.0020
0.1510
0.2630
**
0.0270
0.1450
***
0 0090
*
0 0850
6.6783
1.3780
0.0513
0.0187
-0.0799
0.0329
0.0153
0.0071
-0.0587
0.0373
0.1134
0 1838
0 0457
0 0241
***
0.0000
***
0.0060
**
0.0150
**
0.0310
0.1150
0 5370
*
0 0570
41
T ab le III-1 (co n clu d ed )
Number of observations
Countries
R2 within
R2 between
R2 overall
329
20
0.450
0.121
0.091
326
20
0.386
0.095
0.105
326
20
0.253
0.008
0.003
Source: Elaborated by the author.
Note: The stars ***, **, * indicate respectively the statistical significance at the one, five and ten percent
level. The first value below “re” is the coefficient value that indicates the slope of regression line, the value
below the coefficient is the standard error, whereas the value below the star is the P value.
TABLE III-2
CONTROL VARIABLES REGRESSIONS
6
5
Regressions 5-11
Tax revenue in GDP
Const
Gdpvar per capita
Agriculture
Openness
Deficit previous year
Lack of civil liberties
Schooling
re
***
13.7119
2.8131
0.0744
0.0000
0.0221
0.0010
-0.1224
0.1019
0.0375
0.0271
0.1584
0.0818
-0.9392
0.2989
0 0630
0.0318
***
0.2300
0.1660
*
0.0530
***
0.0020
8.8239
3.0793
0.0523
0.0261
-0.0704
0.0559
0.0605
0.0137
0.0016
0.0793
-0. 8082
0.3010
re
***
0.0040
**
0.0450
0.2080
***
0.0000
0.9840
***
0.0070
10.6967
2.4172
0.0275
0.0255
-0.0975
0.0544
0.0502
0.0134
-0.0097
0.0618
-0.7473
0.3112
***
0.0000
0.2810
*
0.0730
***
0.0000
0.8750
**
0.0160
18.6855
2.2747
0.0369
0.0234
-0.1831
0.0767
0.0394
0.0154
0.0005
0.0596
-1.2792
0.3251
10
9
re
re
***
0.0000
0.1150
**
0.0170
**
0.0110
0.9930
***
0.0000
15.0133
2.3739
0.0364
0.0232
-0.1279
0.0743
0.0379
0.0151
0.0083
0.0577
-1.0602
0.2932
***
0.0000
0.1180
*
0.0850
**
0.0120
0.8860
***
0.0000
17.3763
4.7655
0.0345
0.0239
-0.1693
0.0837
0.0388
0.0160
-0.0026
0.0603
-1.2193
0.3029
***
0.0000
0.1490
**
0.0430
**
0.0150
0.9660
***
0.0000
19.9389
3.2987
0.0472
0.0275
-0.1111
0.0541
0.0503
0.0175
0.1968
0.1167
-0.9900
0.3951
***
0.0000
*
0.0860
**
0.0400
***
0.0040
*
0.0920
**
0.0120
**
0.0310
0.8025
0.2990
***
0.0070
0.2118
0.4362
Population growth
0.6270
0.0264
0.0086
Population density
Urbanization
***
0.0020
0.0232
0.0498
0.6420
-0.1122
0.0529
Shadow economy
R2 within
R2 between
R2 overall
re
0.0480
Oldness
Number of observations
Countries
11
re
**
0.1147
0.0531
Female labor force
8
7
re
205
30
401
27
441
28
495
31
495
31
495
31
201
0.268
0.322
0.316
0.392
0.293
0.335
0.350
0.372
0.399
0.220
0.250
0.418
0.449
0.222
0.298
0.297
0.428
0.079
0.081
0.325
0.324
24
**
0.0340
Table III-2 (continued)
Regressions 12-13
T ax revenue
Const
Gdpvar per capita
Agriculture
Openness
Deficit previous
year
ANGLO
Total
13.8692
1.6744
0.0462
0.0254
-0.2285
0.0676
0.0312
0.0168
0.0164
0.0565
7.5976
1.7280
***
0.0000
*
0.0680
***
0.0010
*
0.0640
0.7720
***
0.0000
12
13
ANGLO
Conflict internal 1
Direct
4.2035
1.1215
0.0071
0.0203
-0.1184
0.0506
0.0196
0.0119
0 0630
0.0308
1.9240
1.1151
***
0.0000
0.7250
**
0.0190
0.1010
**
0.0410
*
0.0840
Indirect
8.0433
0.9145
0.0650
0.0163
-0.1002
0.0363
0.0170
0.0089
-0 0009
0.0361
6.2182
1.4321
Conflict 1
Observations
Countries
R2 within
R2 between
R2 overall
***
0.0000
***
0.0000
***
0.0060
*
0.0540
0.9810
***
Total
18.0537
2.5417
0.0439
0.0255
-0.2383
0.0801
0.0373
0.0182
0 0138
0-0565
440
30
0.159
0.227
0.187
440
30
0.147
0.536
0.517
0.0000
*
0.0840
***
Direct
4.7113
1.3452
0.0066
0.0204
0.0410
-0.1194
0.0519
0.0213
0.0122
0.8070
0 0610
0.0304
0.0030
**
***
0.0000
0.7460
**
0.0210
*
0.0820
**
Indirect
12.3314
1.4256
0.0660
0.0161
-0.1115
0.0379
0.0185
0.0091
0.0450
-0 0002
0.0362
0.7510
-4.9854
1.3626
***
0.0000
***
0.0000
***
0.0030
**
0.0410
0.-9950
0.0000
-4.0625
2.0080
495
31
0.154
0.579
0.522
***
495
31
0.155
0.279
0.348
**
0.0430
0.3392
1.0700
440
30
0.159
0.110
0.119
440
30
0.147
0.370
0.400
***
0.0000
Table III-2 (concluded)
Regressions 14-15
Tax revenue
Const
Gdpvar per capita
Agriculture
Openness
Total
17.8162
2.5229
0.0437
0.0255
-0.2367
0.0795
0.0366
0.0183
Deficit previous
year
Conflict 2
Conflict 3
0.0129
0.0565
-5.1053
1.6450
***
0.0000
*
0.0860
***
14
15
Conflict internal 2
Conflict 3 external
Direct
4.8905
1.3709
0.0066
0.0204
0.0030
**
-0.1191
0.0518
0.0211
0.0460
0.8190
***
0.0020
***
0.0000
0.7450
**
Indirect
11.4856
1.4034
0.0660
0.0161
***
0.0000
***
0.0000
***
Total
17.2653
2.3646
0.0439
0.0256
***
0.0000
*
0.0860
***
Direct
5.1281
1.2730
0.0069
0.0204
0.0210
*
-0.1140
0.0388
0.0179
0.0030
**
-0.2411
0.0804
0.0382
0.0030
**
-0.1208
0.0516
0.0210
0.0123
0.0860
0.0091
0.0500
0.0179
0.0330
0.0612
0.0304
-0.1898
0.8417
**
-0.0002
0.0360
-3.8647
1.3997
0.8220
0.0440
0.8220
0.9950
***
0.0127
0.0563
***
0.0000
0.7350
**
Indirect
11.0021
1.2442
0.0663
0.0163
***
0.0000
***
0.0000
***
0.0190
*
-0.1169
0.0384
0.0186
0.0020
**
0.0120
0.0810
0.0089
0.0360
0.0613
0.0305
**
-0.0004
0.0359
0.9900
-2 7287
2 2443
0 2240
0.0440
0.0060
- 4 3520
2 5534
*
0 0880
-1 5568
0 7392
**
0 0350
495
440
440
495
440
440
31
30
30
31
30
30
0 155
0 159
0 147
0 155
0 159
0 147
R2 between
0 304
0 108
0 219
0 257
0 148
0 156
0 293
0 118
0 235
0 207
0 132
0 194
R2 overall
Source: Elaborated by the author.
Note: The stars ***, **, * indicate, respectively, the statistical significance at the one, five and ten percent level. The first value below “re” is the
coefficient value that indicates the slope o f regression line, the value below the coefficient is the standard error, whereas the value below the star is the P
value.
Observations
Countries
R2 within