Trade Openness and Tax Revenue Performance in East African Countries

Transcription

1 Modern Economy, 2017, 8, hp:// ISSN Online: ISSN Prin: Trade Openness and Tax Revenue Performance in Eas African Counries Micah Samuel Gaalya 1, Bbaale Edward 2, Hisali Eria 2 1 Commissioner General s Office, Uganda Revenue Auhory, Kampala, Uganda 2 College of Business and Managemen Science, Makerere Universy, Kampala, Uganda How o ce his paper: Gaalya, M.S., Edward, B. and Eria, H. (2017) Trade Openness and Tax Revenue Performance in Eas African Counries. Modern Economy, 8, hps://doi.org/ /me Received: March 12, 2017 Acceped: May 13, 2017 Published: May 16, 2017 Copyrigh 2017 by auhors and Scienific Research Publishing Inc. This work is licensed under he Creaive Commons Aribuion Inernaional License (CC BY 4.0). hp://creaivecommons.org/licenses/by/4.0/ Open Access Absrac Despe he well-known gains from rade, he effecs of rade openness are a priori ambiguous. For his reason s imporan o esablish he effecs of rade openness on differen sources of governmen revenue for any counry opening s borders o rade. This sudy sough o esablish he effecs of rade openness on differen caegories of axes. A panel daa coinegraion echnique ha uses he Fully Modified Ordinally Leas Squares and Dynamic Ordinally Leas Squares were employed. The daa are annual cross counry panels of Eas Africa counries covering he period The daa were obained from he IMF s Inernaional Finance Saisics, he African Developmen Bank s African Economic Oulook and he World Bank s World Developmen Indicaors. We found ha he average ariff rae used as a measure for rade openness posively influences oal ax, indirec ax and rade ax while he average ariff rae squared is negaive, illusraing a Laffer effec for he hree ax caegories. The relaionship beween rade openness and direc axes is found o be insignifican. The policy implicaion is ha governmens of EAC counries should asymmerically implemen rade openness policies, paricularly lowering he ariff rae o help in improving ax performance. Keywords Trade Openness, Tax Revenue Performance, EAC Counries 1. Inroducion The laer par of he wenieh cenury has been associaed wh subsanial expansion in rade flows, capal movemens as well as mobily of labour across borders. During he period world rade in goods and services has grown dramaically from abou US $6.199 rillion in 1994 o approximaely US $26.02 ril- DOI: /me May 16, 2017

2 lion in This reflecs a growh rae of 76.1 percen [1]. For he period, world rade grew on average nearly wice as fas as world producion, suggesing ha counries are increasingly rading wh each oher. In he case of EAC counries rade in goods and services increased from abou US $4.4 billion in 1994 o approximaely US $36.78 billion in This shows a growh rae of 88.1 percen [1]. An analysis of he growh rae suggess ha EAC counries have a higher rade growh rae in conras o he world rade. The above saisics indicaes ha for he period, EAC counries rade grew much faser han world rade. Accordingly, he [2] idenifies hree major reasons for he growh in world rade; and hese reasons are likely o coninue increasing growh in rade in he medium and long erm. The firs reason is ha improvemens in he echnology of ransporaion and communicaion have grealy reduced he coss of ransporing goods, services as well as facors of producion. Addionally, he increasing convergence of ases and preference of individuals and socieies has increased demand for goods and services across counries. Lasly, he global economic cooperaion has led o rade openness or reducion and removal of barriers o free rade. These hree reasons have influenced he growh in world rade [2]. In his paper we discuss he influence of rade openness or reducion and removal of barriers o free rade on EAC counries. The increasing prominence of rade openness is moivaed by four major gains; however, here are considerable overlaps among hem [2]. These gains come from unilaeral rade openness policies as well as from rade openness policies ha ake place hrough regional and mulilaeral negoiaions. The five major gains from rade openness are presened below; o begin wh, rade openness allows counries o expor hose goods and services ha hey make efficienly and o impor hose goods and services ha hey make inefficienly. Nex, rade openness resuls in lower prices, enabling an increase in real income which increases consumer and producer welfare. In he same way, rade openness leads o gains in oal facor producivy i.e., freer rade exposes counries o new producion echnologies ha foser higher producivy a boh firm and indusry level. Lasly, rade openness enables low income counries o raise heir income levels owards high income counries [3] and [4]. These four facors have influenced he increase in rade openness across counries. Theoreically, he influence of rade openness on impors and revenue performance is considered o be an indirec oucome. This indirec oucome is derived from he response of consumpion and producion decisions o price changes, of which he price changes are riggered by rade reforms [5] [6] [7]. For example, a reducion in impor ariffs is likely o influence impors and revenue performance depending on he elasicy of impor demand and price elasicy of supply for impor subsues i.e., if he demand for impors is inelasic s likely ha impor volumes and revenue performance will remain unchanged irrespecive of he changes in impor ariffs and prices. On he oher hand, if he demand for impors is elasic s possible ha impor volumes and revenue performance will increase owing o changes in impor ariffs and prices. 691

3 This heoreical relaionship has been examined by [4] [7]-[12] among ohers. This leraure can be summarized ino wo groups, he firs group shows ha removal of barriers o free rade increases revenue performance while he second group shows ha removal of barriers o free rade leads o a decline in revenue performance. An example of sudies in he firs group are presened below; Firsly, [3] using panel daa on developing counries concludes ha srucural facors such as GDP per capa, share of agriculure o GDP, rade openness and foreign aid significanly increase revenue performance in developing counries. The sudy shows ha rade openness has a srong posive relaionship wh oal revenue performance. However he inadequacy of his sudy is ha focusses on oal revenue performance and ignores he influence of rade openness on disaggregaed caegories of axes such as direc and indirec axes. Secondly, a sudy by [13] using a se of facors ha influence revenue performance sugges ha revenue performance is higher in more open and less agriculural dependen economies ha are less populous and peaceful. Overall his sudy shows ha rade openness has a posive relaionship wh rade axes and GDP per capa. On he oher hand sudies in he second group which shows ha removal of barriers o free rade increases impors bu leads o a decline in revenue performance. These include; In he firs place, [14] examining he relaionship beween ariff reform and rade axes in Uganda shows ha exchange rae depreciaion has had a pass hrough effec o he domesic marke price of impors and his increases rade axes in he shor run bu reduces rade axes in he long run. This sudy predics ha rade openness will o some exen decreases revenue performance. The shorcoming of his sudy is ha does no consider oher key caegories of axes ha influence overall revenue performance such as direc, indirec and oal revenues. Nex, [10] examining he fiscal effecs of ariff reducion for Caribbean communy concludes ha Caribbean counries have experienced revenue shorfalls as a consequence of rade openness. Similarly [15] examining he fiscal effecs of ariff reforms on Uganda s rade wh EAC counries concludes ha Uganda has experienced revenue shorfalls as a resul of reducing s impor ariff raes. However he shorcoming of his sudy is ha only considers effecs of rade openness on impor ariff and does no consider oher key caegories of axes ha influence overall revenue performance such as direc and indirec ax revenues. Also, [19] using a panel of 117 counries over a 32-year period, examine [16] heory of ax srucure developmen which saes ha as counries develop hey progressively change heir ax srucure o adop o he changes in economic srucure. More specifically he counries end o replace rade revenue wh domesically based revenues. In relaion o his heory [17] show ha low income counries have no recovered from domesic axes he revenues hey have los from rade openness. Noneheless he sudy shows ha he replacemen of rade axes wh domesic axes has become higher han previous sudies 692

4 have suggesed. In addion, [11] examining he argumen ha rade openness depresses he revenue performance in developing counries show ha low income counries and upper middle income counries have experienced declining revenues on accoun of rade openness. They conclude ha rade openness leads o falling direc and rade revenues and ha he srucural characerisic of low income and developing counries have been significan in explaining he decline in revenue performance. Lasly, [18] examining he relaionship beween rade openness and revenue performance in sub-saharan Africa, conclude ha rade openness raises overall axes in French colonies of Africa (CFA franc counries), hough he disaggregaed ax oucome sugges ha rade openness raises rade axes bu lowers indirec axes. To sum all up, from he above leraure, s clear ha rade openness affecs revenue performance, he effecs are eher negaive or posive depending on he economic srucures of an economy or region. In he case of EAC counries despe implemening rade openness reforms, hhero here are no sudies ha invesigaed he effecs of rade openness on differen caegories of axes. Sudies by [13] and [14] have only invesigaed effecs of rade openness on rade axes and do no consider effecs of rade openness on oher key caegories of axes. Furhermore, according o [8] he average ariff rae is aken o indicae he level of rade openness for an economy, as such a decline in he average ariff rae is aken o indicae greaer rade openness while an increase in he average ariff rae is aken o indicae lower rade openness. From he Figure 1 we conclude ha EAC counries have experienced a reducion in ariff raes which reflecs an increase in he level of rade openness. Figure 1. EAC counries average ariff rae. Source: World developmen indicaors, world bank, April

5 Example of rade openness reforms implemened in EAC counries include; implemenaion of ariff reforms under he World Trade Organizaion and General Agreemen on Tariffs and Trade (GATT), ariff reforms under he IMF and World Bank Srucural Adjusmen Program and ariff reforms under he EAC cusoms union proocol [2]. These reforms have involved he reducion of ariff and non-ariff barriers across EAC counries. Correspondingly, according o Figure 2 he revenue performance measured by he ax o GDP raio shows ha for he period 1994 o 2012, revenue increased however, despe he increase in revenue performance, he EAC counry s average oal ax o GDP raio ha sands a percen is less han he sub-saharan Africa average oal ax o GDP raio of 16.1 percen [15]. The lower EAC counry s average oal ax o GDP raio appears o imply ha EAC counries are experiencing low revenue performance as compared o oher sub- Saharan African counries. From economic heory and leraure, appears ha he lower ax performance and higher levels of impor demand could be driven by increased levels of rade openness. Therefore his sudy aemps o answer he quesion how does rade openness affec revenue performance for he EAC counries? Saemen of he Problem I s argued ha he reducion or removal of barrier o free rade, such as impor ariffs lowers impor prices bu also reduces an array of axes charged a imporaion. The gains from removal of barriers o free rade are expeced o increase domesic oupu hrough he use of beer impored skills and echnology o foser high producivy a boh firm and indusrial level; hereby lowering impor for cerain caegories of impors bu also increasing domesic producions and revenue performance. Figure 2. EAC Revenue o GDP raio. Source: World Developmen Indicaors, World Bank, April

6 In he case of EAC counries, despe implemening rade openness reforms he revenue performance measured by he ax o GDP raio shows ha EAC counries are experiencing lower revenue performance as compared o he average sub Saharan Africa revenue performance. This is expeced o cause limed funding for governmen, if revenue performance are no improved. If he low ax performance is no improved, EAC counries will have o review heir rade openness policies in order o avoid problems associaed wh limed funding for governmen. Thus, s imperaive ha evidence regarding he influence of rade openness on ax performance is provided given he sraegic imporance of axes o developmen of EAC counries. 2. Leraure Review 2.1. Leraure on Direc Taxes The leraure on ax performance indicaes ha direc axes are composed of personal income axes, corporae income axes, renal income axes, whholding axes and capal gains axes. The revenue from hese ax handles are defined as income axes or direc axes [14]. The heoreical linkage beween rade openness and direc axes is likely o be hrough price change or economic growh. This relaionship has been examined by a number of sudies such as [4] [8] [16] among ohers. Amongs hese sudies [8] poins ou ha rade openness is relaed o high levels of economic growh and ha counries ha have opened up o rade have increased heir levels of growh and income axes. Anoher sudy by [4] invesigaes he effecs of rade openness on income axes by applying he gravy model. They find ha rade openness subsanially raises income. They predic ha since a one percen increase in he shares of impors and expors in GDP leads o a one-half o wo percen increase in income per person. This implies ha rade openness which induces higher volumes of rade is associaed wh higher incomes. A sudy by [15] uses a gravy model and shows ha rade openness increases rade flows, which furher suppors he hypohesis ha rade openness leads o an increase in rade volumes and economic growh. Addionally he growh in he economy is likely o have a direc impac on income axes, his happens when higher per capa income leads o widening of he ax base for income axes. For example, [16] provides a solid heory supporing a causal relaionship beween per capa income and ax level in his ax base and ax handle heory. He saes ha an increase in per capa income raises he size of public secor which in urn increases a counry s ax base and axable capacy. Reference [17] and [18] demonsraes he associaion beween economic growh and ax by using regression analysis and found ha overall ax is posively influenced by he level of per capa income. Alhough he correlaion is no srong, he finds ha per capa income has a posive impac on personal income ax. 695

7 In anoher sudy by [7] he poins ou ha income axes may be difficul o mobilize in low income counries because of he high adminisraive coss and oher srucural consrains. In low income counries, income axes are relaively easy o evade because mos developing counries generally have weak penaly on ax evasion. There also oher srucural challenges ha increase he ease of ax evasion. Therefore, in developing and less developed counries, income ax is usually levied on wages of public secor employees, foreign corporaions and a few small and medium enerprise [7]. These facors depress he income ax base, resuling in a narrower base in developing counries han in developed counries, and hence lower ax from his ax caegory irrespecive of rade openness effors Leraure on Indirec Taxes The leraure on indirec axes shows ha indirec axes are composed of Value Added Tax (VAT) or Commercial Transacion Levy and Sales ax as well as Excise ax [14]. These axes are considered o be a good soluion for offseing he decrease in ax arising from changes in he inernaional rade ax regime because hey have a broader ax base as compared o income axes [15]. According o [15] he relaionship beween rade openness and consumpion ax is more complicaed o assess as compared o he relaionship beween rade openness and income axes. This is because he effecs of rade openness on indirec axes depends on many facors such as he price elasicy of demand for impors and he price elasicy of supply of impor subsues [15]. For example, when impor ariffs are reduced, he relaive price of impors o impor subsues may also decrease, which may make domesic consumers swch o consumpion of more impors. Consequenly his may lead o a decrease in axes colleced from domesically produced impor subsues and an increase in axes from impors. The effec of rade openness on indirec ax can also be viewed hrough s impac on economic growh. For example, he assumpion ha ax bases grow as economic growh proceeds also rue for he consumpion ax. The growh in he economy is also relaed o he growh in he consumpion ax base [15]. Therefore consumers should have more income in heir hand as he economy grows, which means ha here is higher purchasing power and higher demand for domesic consumpion. However, surprisingly, he resuls shown in [7] sae ha here is no correlaion beween consumpion axes and income per capa. Despe he resul in his sudy, [10] shows ha a counries size plays an imporan role in deermining he domesic consumpion ax. Furhermore, [15] show ha consumpion axes are more significan ax source in he bigger counries han in he smaller counries. In he same way he amoun of consumpion axes colleced depends direcly on he domesic consumpion, ha is, larger counries end o have a high populaion and a large domesic marke whereas smaller counries seem o have 696

8 a smaller populaion and heir size of domesic marke is smaller. As a resul, swching sources of ax from rade ax o a broad-based consumpion ax, alhough applicable for developed counries, may cause fiscal problems for developing and less developed counries which have smaller marke sizes. There for he impac of rade openness ough o be evaluaed Leraure on Trade Taxes According o leraure rade axes are composed of impor and expor axes as well as a broad array of axes charged on goods and services a he poin of impor or expor [14]. Trade ax is hough o decline afer counries open up heir rade ax regime. Economic heory poins ou ha he reducion in impor ariff ofen leads o a decline in rade ax especially for small open economies [3]. A number of empirical sudies have been underaken o assess hese effecs, mos of he sudies have used ax share in GDP as he dependen variable wh differen combinaions of explanaory variables. For example, [10] invesigaes he effecs of rade openness on rade ax in he Caribbean communy. The sudy finds evidence ha rade openness in he Caribbean counries lead o a reducion in rade ax. Reference [3] sudies he principal deerminans of ax performance across developing counries by using a broad daase of 105 counries over 25 years. In his sudy he noed ha per capa GDP, agriculure share in GDP, rade openness, foreign aid, corrupion, polical sabily, he share of direc and indirec axes are significan in deermining ax performance. The paper concluded ha ax collecions are low in counries, which heavily depend upon axing goods and services, while counries ha depend on income axes have high ouurn of ax. However for counries ha have opened up here rade regimes, rade axes appear o have declined. Similarly [8] proposes ha if rade openness is accompanied wh a reducion in ariff dispersion, hen ax revenue may increase. The raionale is ha a reducion in he dispersion of ariff is ofen done by lowering he higher ariff and increasing lower ariff in order o obain average values. The sudy furher proposes ha if he inial ariff raes are high, hen ariff reducion may lead o an increase in ax revenue since price elasicies of demand and supply are no consan over he enire range of prices. The above effec can be illusraed by a Laffer curve, which shows he relaionship beween ax revenue and ariff resricions (Figure 3). The illusraions shows ha when he inial ariff rae is prohibively high, he rade volumes are likely o be severely boled-up and ax will be low. As such reducing ariffs will lead o a subsanial increase in rade volumes and a decrease in he incenive o evade axes. However, if here is a furher ariff reducion afer rade is fairly liberalized a he ax maximizing rae he increase in rade volume will no be large enough o offse he lower ariffs and now he direc effec of ariff reducion will resul in he loss of axes. As a resul, overall ariff revenue will decrease [8] [19] [20] [21]. 697

9 Figure 3. Laffer curve. Source; Laffer (2004). Economic heory proposes ha he higher he ariffs, he higher he ax evasion since evasion leads o high marginal benef for ax payers. Reference [22] shows ha a one percen increase in ariff raes is associaed wh a hree percen increase in ax evasion. The assumpion is ha ariff reducion raises he coss for ax evaders and hence lowers he level of ax evasion. Therefore reducion in ariff brings an increase in ax revenue [23]. Among he sudies conduced on rade ax performance in he EAC is a sudy by [24]. The sudy examines he relaionship beween ariff reforms and cusoms revenue by explicly capuring he insuional feaures of decision making in Uganda. The sudy uses he Johansen mulivariae approach and Error Correcion Model o esablish he long-run and shor run relaionship beween rade openness and rade ax. The resuls show ha exchange rae depreciaion has pass hrough effecs o he domesic marke price of impors which reduces rade ax o GDP raio in he long-run, hough increases rade ax in he shor erm Synhesis of he Leraure The secion provides a synhesis of leraure on sudies invesigaing effecs of rade openness on ax revenue. This leraure can be summarized ino wo groups, he firs group shows ha removal of barriers o free rade increases revenue performance [3] and [13], while he second group shows ha removal of barriers o free rade leads o a decline in revenue performance [10] [15] [17] [24] among ohers. The sudies have used he following daa esimaion echniques o esimae ax revenue funcions i.e., Error Correcion Model, Vecor Error Correcion Models, Auo Regressive Disribued Lag models, Fixed and Random Effecs models as well as he General Mehod of Momens. The leraure only examines he influence of rade openness on rade axes bu does no examine he influence of rade openness on oher ax caegories such as direc and indirec axes for EAC counries. As such, despe implemening rade openness reforms in he EAC counries, hhero here are no sudies ha invesigaed he effecs of rade openness on direc and indirec 698

10 axes for EAC counries. This sudy conribues o rade leraure by underaking a sudy ha esimaes he effecs of rade openness on differen ax caegories in he EAC counries. 3. Mehodology for Esimaing Tax Revenue Funcions 3.1. Model Specificaions Analysis of ax leraure by [3] [11] [25] [26] [27] [28] reveals ha sudies looking a ax performance have used a variey of mehods. The mos commonly used approach is he behavioral approach. The approach was proposed by [27] and [29] o measure ax performance. The approach regresses he ax o GDP raio on a se of variables ha serve as proxies for a counry s ax handles. In a funcional form, he ax funcion is presened below; T Y ( ) = f V (3.1) where, T = ax performance. Y = real GDP. T Y = ax o GDP raio. V = vecor of ax handles. = is a ime subscrip. We rewre Equaion (3.1) ino an economeric form i.e. T Y = β + β V + ε (3.2) 1 2 where; β ( 1 and 2) = are coefficiens for he ax performance variables. ε = is he error erm. Tax performance is considered as an endogenous variable while he vecor of ax handles are considered as exogenous variables. Equaion (3.2) is ransformed ino Equaion (3.3) where he endogenous and exogenous variables are inroduced. Hence, oal ax, indirec, direc and rade ax funcions can be expressed as follows: T Y = β0 + β1urban + β2gdp + β3aid + β4reer (3.3) + β Inflaion + β Agric ++ β Deb + ε where; T Y = Toal ax, direc, indirec and rade ax o GDP raio. Urban = Share of urban populaion o he oal populaion. GDP = GDP per capa. Aid = Aid per capa. REER = Real effecive exchange rae. Inflaion = Consumer Price Index. Agric = Share of agriculural secor o GDP. Deb = Deb sock as a percenage of Gross Naional Income. = is a ime subscrip. ε = is he error erm. 699

11 To esablish he effecs of rade openness on differen caegories of axes for EAC counries, we conver model o a panel esimaion and also inroduce he average ariff rae alongside oher conrol variables. These variables are drawn from leraure as proposed by [3] [11] [25] [26] [27] [28] among ohers. Based on he leraure he average ariff rae, average ariff rae squared, governmen effeciveness, rule of law, regulaory qualy and conrol of corrupion are inroduced ino he ax funcion in Equaion (3.3). Therefore Equaion (3.3) becomes Equaion (3.4) which is he ax funcion ha esablishes he effecs of rade openness for EAC counries. The equaion is expressed as follows: where, T T Y = β + βurban + β GDP + β Aid + β REER + β Inflaion β Agric + β Deb + β ATR + β ATR + β GE β LAW + β RQ + β CC + ε Y = Toal ax, direc, indirec and rade ax o GDP raio. Urban = Share of urban populaion o oal populaion. GDP = GDP per capa. Aid = Aid per capa. Exch = Real effecive exchange rae. Inflaion = Consumer Price Index. Agric = Share of agriculural secor o GDP. Deb = Deb sock as a percenage of Gross Naional Income. GE = Governmen effeciveness. LAW = Rule of law. RQ = Regulaory qualy. CC = Conrol of corrupion. (3.4) ε = is he error erm. = is a ime subscrip. i = 1, 2, 3, 4, Daa Type and Sources The sudy employs a cross counry panel which includes he following counries; Burundi, Kenya, Rwanda, Tanzania and Uganda. Daa on he oal ax o GDP, indirec ax o GDP, direc ax o GDP and rade ax o GDP are obained from he Africa Economic Oulook, Saisical Annex, however he daa range is limed only o period since he year 1994 which became he base year for his research. Daa on share of urban populaion o he oal populaion, average ariff rae, deb sock, share of agriculure secor o GDP, inflaion, real effecive exchange rae and foreign aid are obained from he World Bank s World Developmen Indicaors (WDI). Daa on polical sabily, governmen effeciveness, regulaory qualy and rule of law are obained from he World Bank s Worldwide Governance Indicaors (Table 1). The use of panel daa offers several advanages in economeric analysis, firs, panel daa conains more degrees of freedom and more sample variabily, hence 700

12 Table 1. Descripion of he daa se used in ax models. Variable Ne oal ax GDP raio Ne indirec ax GDP raio Ne direc ax GDP raio Ne rade ax GDP raio GDP per capa Deb sock as a percenage of Gross Naional Income Share of urban populaion o he oal populaion Share of agriculural secor o GDP Source Annual daa from Africa Economic Oulook, Saisical Annex, April Annual daa from Africa Economic Oulook, Saisical Annex, April Annual daa from Africa Economic Oulook, Saisical Annex, April Annual Daa from Africa Economic Oulook, Saisical Annex, April Annual daa from World Bank, World Developmen Indicaors (WDI), July Annual daa from World Bank, World Developmen Indicaors (WDI), July Annual daa from World Bank, World Developmen Indicaors (WDI), July Annual daa from World Bank, World Developmen Indicaors (WDI), July Trade raio o GDP World Bank, World Developmen Indicaors (WDI), July Aid per capa Consumer Price Index Average ariff rae Average ariff rae squared Real effecive exchange rae Voice and Accounabily Polical Sabily & Absence of Violence/Terrorism Governmen Effeciveness Regulaory Qualy Rule of Law Annual daa from World Bank, World Developmen Indicaors (WDI), July Annual and quarerly daa from IMF s Inernaional Finance Saisics (IFS), July Annual daa from World Bank, World Developmen Indicaors (WDI), July We ake he square roo of he average ariff rae series. Annual and quarerly daa from IMF s Inernaional Finance Saisics (IFS), July Worldwide Governance Indicaors, Worldwide Governance Indicaors, Worldwide Governance Indicaors, Worldwide Governance Indicaors, From Worldwide Governance Indicaors, improving he efficiency of economeric esimaes. Second, panel daa has a greaer capacy for capuring he complexy of impor demand behavior han a single ime series daa. I is frequenly argued ha he reason ha a researcher finds or does no find cerain causal effecs in economeric analysis is due o omission of cerain variables in one s model specificaion which are correlaed wh he included explanaory variables. However, since panel daa conain informaion on boh he iner-emporal dynamics and he individualy of he en- 701

13 ies, is capable of conrolling for he effecs of missing or unobserved variables. Panel daa generaes more accurae predicions for individual oucomes by pooling he daa raher han generaing predicions of individual oucomes. If individual behaviors are similar on cerain variables, panel daa provides he possibily of learning an individual s behavior by observing he behavior of ohers. Thus, is possible o obain a more accurae descripion of an individual s behavior by supplemening observaions of he individual in quesion wh daa on oher individuals. 4. Empirical Resuls 4.1. Panel Un Roo Tess The firs sep in he analysis is o esablish he order of inegraion of he variables. We use he [29] panel un roo es and he [30] panel un roo es. The resuls are repored under Table 2. The resuls show ha he T-bar and he w (T-bar) es saisic of [30] panel un roo es fail o rejec he null hypohesis of presence of un roos a level for he following variables; oal ax o GDP raio, indirec axes o GDP raio, direc axes o GDP raio, rade axes o GDP raio, average ariff rae, share of agriculure o GDP, share of urban populaion o he oal populaion and rade raio o GDP. This suggess ha he variables are non-saionary a level. On he oher hand he null hypohesis of presence of un roos a level for he following variable i.e., inflaion, real effecive exchange rae, regulaory qualy, polical sabily, aid per capa, conrol of corrupion and rule of law is rejeced which suggess ha he variables are saionary a level. When he non-saionary variables are ransformed ino firs difference he [30] es rejec he null hypohesis of presence of un roos which implies ha he variables are saionary a firs difference. The resuls of he [30] panel un roo es show ha he Modified Chi Squared Pm panel un roo es fail o rejec he null hypohesis of presence of un roos a level for he following variables; oal ax o GDP raio, indirec axes o GDP raio, direc axes o GDP raio, rade axes o GDP raio, average ariff rae, share of agriculure o GDP, share of urban populaion o he oal populaion and rade raio o GDP which suggess ha he variables are non-saionary a level. When he variables are ransformed o firs difference he [30] panel un roo es rejec he null hypohesis of un roos. This implies ha he variables are saionary a firs difference and inegraed of I(1). The oher variables i.e., inflaion, real effecive exchange rae, regulaory qualy, polical sabily, aid per capa, conrol of corrupion and rule of law, implying hey are inegraed of I(1) Panel Coinegraion Tes We use he Kao panel coinegraion ess o examine for presence of coinegraion relaionship in he variables in he ax funcions. The ADF es saisic rejecs 702

14 Table 2. Panel un roo es resuls. Variables IPS (2003) panel un roo es Maddala and Wu (1999) Order of inegraion T-bar W (T-bar) Toal ax o GDP Toal ax o GDP 4.86* 4.72* 17.2* Indirec axes o GDP Indirec axes o GDP 4.12* 4.32* 10.5* Direc axes o GDP Direc axes o GDP 5.40* 5.2* 22.68* Trade axes o GDP Trade axes o GDP 5.93* 5.55* 28.44* Average Tariff Rae Average Tariff Rae 6.15* 5.28* 13.74* Agriculural share o GDP Agriculural share o GDP 4.44* 4.63* 11.71* I(1) I(1) I(1) I(1) I(1) I(1) GDP per capa 1.39* 0.27* 20.31* I(0) Aid per capa 5.31* 4.64* 27.24* I(0) Trade raio Trade raio 3.35* 3.24* 13.04* I(1) Share of urban populaion o he oal populaion Share of urban populaion o he oal populaion * 0.368* 6.64* I(1) Real effecive exchange rae * I(1) Real effecive exchange rae 0.462* 2.669* 1.94* Inflaion 4.585* 3.030* 6.42* I(0) Regulaory qualy 5.37* 5.12* 8.54* I(0) Polical sabily 6.40* 4.45* 31.06* I(0) Conrol of corrupion 4.07* 4.12* 34.43* I(0) Accounabily 5.66* 4.98* 0.59* I(0) Rule of Law 3.33* 2.54* 1.19* I(0) Noe: In he specificaion of he ess above, we use consan and rend as he deerminisic erms. We use one lag for oal ax o GDP raio, indirec axes o GDP raio, direc axes o GDP raio, rade axes o GDP raio, average ariff rae, share of agriculure o GDP, share of urban populaion o he oal populaion and rade raio o GDP for he IPS and Maddala and Wu. (*) denoes rejecion of he null hypohesis a 1%. he null hypohesis of no coinegraion a 1 percen level of significance for he variables in he four ax funcions. This implies ha here exiss a long-run relaionship in he variables in he ax funcions Panel Coinegraion Regressions The resuls from he panel un roos es and he panel coinegraion ess show ha he variables across he four ax models are inegraed of I(0) and I(1) bu 703

15 also coinegraed. According o [31] and [32] he asympoics of large T, large N panels are differen from he asympoics of small T and small N panels. Therefore esimaions for small T and small N panels rely on Fixed or Random effecs esimaor if he variables are inegraed of he same order and also coinegraed. On he oher hand if he variables are no inegraed of he same order and or coinegraed, he small T and small N panels rely on he Fully-Modified Ordinally Leas Squares (FMOLS) and Dynamic Ordinally Leas Squares (DOLS). Esimaions for small T and large N panels rely on a combinaion of fixed effecs esimaors and insrumenal variable esimaors, such as he [1]. Reference [33] generalized mehod-of-momens esimaor. The large N and large T panel esimaions, rely on he mean-group (MG) and pooled mean-group (PMG) esimaors [31]. In our specific case i.e., T = 19 and N = 5. We have a relaively small T and small N panel ha is inegraed of I(0) and I(1) bu coinegraed. From he leraure we adop he FMOLS and DOLS esimaors. Reference [31] shows ha he FMOLS and DOLS esimaors performs well in small samples. The FMOL and DOLS model is considered superior o oher esimaion echniques because inherenly correc for endogeney, serial correlaion and asympoic bias Discussion of Empirical Resuls Table 3 shows he empirical resuls of he FMOLS and DOLS ax models. The models presened include; oal ax, indirec axes, direc axes and rade axes. The resuls from he regression esimaion reveal he following findings. The coefficien for he average ariff rae ha is used as a measure for rade openness is posive and saisically significan wh respec o oal ax, indirec and rade ax bu insignifican for direc axes. The posive resuls implies ha when ariff are increased, oal axes, indirec axes and rade axes also increases. Nex we examine he sign of he coefficien of average ariff rae squared o es if he Laffer curve effec exiss for oal axes, indirec axes and rade axes. The resuls show ha he sign of he coefficien of average ariff rae squared is negaive and saisically significan for oal ax, indirec and rade ax bu insignifican for direc axes. This illusrae he expeced Laffer effec i.e., he ax-maximizing ariff rae. Therefore, for counries in his panel he ax maximizing ariff rae is esimaed a approximaely 9.2 percen for oal ax, 12.2 percen for indirec ax and 14 percen for rade axes for he period 1994 o Thus, for counries in his group, an increase in ariff raes beyond hese rae would resul ino a decrease in revenue for he hree ax caegories. The resuls are expeced and are similar o findings from previous sudies such as [20] who show ha he increase in ariff rae can only increase ax revenue up o a cerain rae hereafer ax revenue decreases. In Table 4 we conrol for he resuls of he average ariff rae variable by inroducing anoher measure for rade openness i.e., he rade raio o GDP. The resuls from his measure show ha wh respec o oal ax, indirec, direc and rade ax models, he coefficien for rade raio o GDP is posive and saisically 704

16 Table 3. FMOL and DOLS ax models using average ariff rae. (1) (2) (3) (4) (5) (6) (7) (8) VARIABLES Toal ax/ GDP Toal ax/ GDP Direc ax/gdp Direc ax/gdp Indirec ax/gdp Indirec ax/gdp Trade axes/gdp Trade axes/gdp FMOL DOLS FMOL DOLS FMOL DOLS FMOL DOLS Urban Populaion 0.307* 0.296* 0.297* 0.342* * 0.187* 0.156** (0.070) (0.09) (0.064) (0.067) (0.055) (0.062) (0.073) (0.076) GDP per capa 2.51* 1.854* *** 0.633** 0.642** 0.784** (0.46) (0.519) (0.387) (0.376) (0.330) (0.343) (0.441) (0.448) Deb/GNI 0.510** 0.140** * 0.219* 0.540* 0.138** (0.06) (0.074) (0.055) (0.053) (0.047) (0.049) (0.059) (0.052) Foreign aid 0.15** (0.207) (0.264) (0.172) (0.191) (0.147) (0.175) (0.197) (0.211) Exchange rae 1.388* 1.433* 0.981* 0.638** 1.157* 1.069* 1.073* 0.895** (0.412) (0.475) (0.343) (0.344) (0.292) (0.314) (0.393) (0.395) ATR 0.180*** * 0.104** 0.350** 0.343** (0.093) (0.083) (0.078) (0.060) (0.066) (0.055) (0.178) (0.181) ATR squared 0.97*** ** 0.425** 0.012** (0.510) (0.471) (0.425) (0.341) (0.362) (0.311) (0.070) (0.007) Inflaion 0.028* 0.023* 0.127* 0.014** 0.013* (0.008) (0.008) (0.006) (0.006) (0.001) (0.006) (0.011) (0.048) Agriculure 0.005** 0.051** 0.023** Governmen effeciveness (0.017) (0.019) (0.014) (0.014) (0.012) (0.013) (0.017) (0.018) 2.365* 2.002* (0.613) (0.735) (0.510) (0.533) (0.434) (0.486) (0.578) (0.644) Rule of law * * 1.581* 1.276** (0.558) (0.714) (0.464) (0.518) (0.395) (0.472) (0.523) (0.032) Polical Sabily ** 0.420** (0.340) (0.317) (0.283) (0.230) (0.241) (0.210) (0.323) (0.270) Regulaory Qualy ** * 1.942** (0.634) (0.798) (0.528) (0.579) (0.449) (0.527) (0.567) (0.629) Conrol of corrupion 1.555* 1.221* * 1.258** ** (0.410) (0.470) (0.341) (0.341) (0.290) (0.311) (0.350) (0.368) R-squared Noe: Sandard errors in parenheses *p < 0.01, **p < 0.05, ***p < 0.1. The dependen variable for Models (1) and (2) is oal ax of GDP raio, (3) and (4) direc ax o GDP raio (5) and (6) indirec ax o GDP raio while (7) and (8) is rade ax o GDP raio. significan. This implies ha he rade raio o GDP posively influences he four caegories of ax. The resul imply ha rade openness leads o a growh in axes for he four ax caegories. The posive relaionship is in line wh findings from previous sudies such as [8] and [21] hese sudies show ha openness has a posive relaionship wh ax. From he resul, we can conclude ha an increase rade openness increases ax for counries in his group. 705

17 Table 4. FMOL and DOLS ax models using rade raio o GDP. (1) (2) (3) (4) (5) (6) (7) (8) VARIABLES Toal ax/ GDP Toal ax/ GDP Direc ax/gdp Direc ax/gdp Indirec ax/gdp Indirec ax/gdp Trade axes/gdp Trade axes/gdp FMOL DOLS FMOL DOLS FMOL DOLS FMOL DOLS Urban Populaion (0.0385) (0.0647) (0.0408) (0.0702) (0.0227) (0.0411) (0.0333) (0.0700) GDP per capa 2.290* 2.183* 1.393* 1.805* 0.769* 0.765** 0.576* 1.113** (0.233) (0.505) (0.247) (0.549) (0.138) (0.321) (0.202) (0.547) Deb/GNI ** * (0.0647) (0.181) (0.0685) (0.197) (0.0382) (0.115) (0.0561) (0.196) Foreign aid *** (0.232) (0.426) (0.246) (0.463) (0.137) (0.271) (0.201) (0.461) Exchange rae 0.22*** 0.511*** 0.385* *** * 0.871* (0.126) (0.277) (0.133) (0.300) (0.0744) (0.176) (0.109) (0.299) Trade raio o GDP * *** 0.028* 0.010* * ** 0.016* 0.015* (0.0109) (0.0274) (0.008) (0.006) ( ) (0.0174) (0.007) (0.006) Inflaion (0.0127) (0.0445) (0.0135) (0.0483) ( ) (0.0283) (0.011) (0.048) Agriculure 0.095* * 0.105*** * (0.0205) (0.0528) (0.0217) (0.0574) (0.0121) (0.0336) (0.0177) (0.0572) Governmen effeciveness 3.606* 4.571** (0.803) (1.832) (0.851) (1.990) (0.474) (1.163) (0.696) (1.983) Rule of law (0.801) (1.768) (0.849) (1.921) (0.473) (1.123) (0.694) (1.913) Polical Sabily (0.355) (0.837) (0.376) (0.909) (0.210) (0.532) (0.308) (0.906) Regulaory Qualy 1.585* 2.365** * 1.944* (0.482) (1.056) (0.511) (1.148) (0.285) (0.671) (0.418) (1.143) Conrol of corrupion 1.419* 1.709** * Inflaion (0.352) (0.723) (0.372) (0.786) (0.208) (0.459) (0.305) (0.783) R-squared Noe: Sandard errors in parenheses *p < 0.01, **p < 0.05, ***p < 0.1. The dependen variable for Models (1) and (2) is oal ax of GDP raio, (3) and (4) direc ax o GDP raio (5) and (6) indirec ax o GDP raio while (7) and (8) is rade ax o GDP raio. The resuls from oher variables used in he ax funcions show ha he coefficien for he raio of urban populaion o oal populaion is posive and saisically significan wh respec o oal axes, direc ax and rade axes bu insignifican for indirec axes. The resul implies ha an increase in urban populaion increases oal ax, direc axes and rade axes. Therefore an increase in he urban populaion is expeced o increase oal axes, direc ax and rade axes for his group of counries. The resul is expeced and s in line wh sudy findings 706

18 by [11] who show ha urbanizaion increase oal axes, direc ax and rade axes for developing counries. The coefficien for GDP per capa is posive and saisically significan across he four ax caegories. This suggess ha an increase in GDP per capa increases oal ax, indirec ax, direc ax and rade ax. The posive relaionship beween GDP per capa and ax suggess ha a high level of developmen are associaed wh a higher abily for ax payers o pay axes. This is accouned for by he fac ha because income levels are so low in his group of counries, higher income facilaes increased rade and hus higher rade ax. This resul is expeced and is in line wh a sudy by [28] who ha found ha he capacy o collec and pay axes increases wh he level of developmen The coefficien for deb sock as a percenage of gross naional income is negaive and saisically significan wh respec o oal ax, direc ax and rade ax. The resuls shows ha he deb sock as a percenage of gross naional income is an imporan deerminan of ax. The resul suggess ha an increase in deb decreases oal ax, direc ax, indirec ax and rade ax for his group of counries. This resul is expeced and shows ha an increase in deb decreases he ax base for he hree ax caegories. The resuls are similar o findings by [3] and [34] who show ha governmen deb decreases ax performance in developing counries. The coefficien for foreign aid is posive and saisically significan wh respec o oal ax bu insignifican for direc, indirec and rade axes. This implies ha foreign aid increases oal ax. The resul is suppored by [3] who poined ou ha if foreign aid comes primarily in he form of loans, hen he burden of fuure loan repaymens may induce policymakers o mobilize higher axes. For counries in his group a large par of foreign aid is in form of concessional and non-concessional loan ha arac ineres. This resul is expeced and s in line wh previous sudies such as [34] and [35] who find ha foreign aid leads o an increase in ax performance for low income counries. The coefficien for inflaion and real effecive exchange rae are negaive and saisically significan wh respec o oal ax, direc ax and indirec ax. The resuls implies ha an increase in inflaion and real effecive exchange rae decreases ax performance. This resul is expeced and s similar o findings from previous sudies such as [36]. The real effecive exchange rae and inflaion, sugges ha real exchange rae appreciaion and higher inflaion depress revenues, which is consisen wh Tanzi s hypoheses. The hypohesis observes ha here is ofen an inverse relaionship beween a counry s ax revenue and he real level of s official exchange rae. The resuls on coefficien for he share of agriculure secor o GDP is negaive and saisically significan wh respec o oal ax and direc axes. For example, a one percen increase in he share of agriculure secor could reduce revenue performance by as much as 0.4 percen. The resul implies ha he agriculure secor decreases ax from boh ax caegories. The possible explanaion for his resul is ha he agriculure secor has small farmers ha are nooriously difficul o ax and he secor does no generae large axable surplus. In addion 707

19 many low income counries are unwilling o ax he agriculure secor because he agriculure secor provides food and livelihood for a larger par of he populaion. The negaive relaionship is expeced and is consisen wh sudies by [3] [26] [37] who show ha in developing counries he agriculure secor negaively influences ax. The resuls on he governance indicaors are mixed, he coefficien for rule of law has a negaive and saisically significan relaionship wh respec o rade axes. This implies ha rule of law negaively influences rade ax in his group of counries. The variable rule of law reflecs percepions of he exen o which agens have confidence in and abide by he rules of sociey, and in paricular he qualy of conrac enforcemen, propery righs, he police, and he cours, as well as he likelihood of crime and violence. The resul sugges he group of counries need o improve heir rule of law as a means of growing rade axes. The coefficien of regulaory qualy is negaive and saisically significan wh respec o indirec axes and direc axes. The resul implies ha he regulaory qualy in EAC counries decreases indirec and direc axes. The variable regulaory qualy reflecs percepions of he abily of he governmen o formulae and implemen sound policies and regulaions ha perm and promoe privae secor developmen. The negaive relaionship beween regulaory qualy and he wo ax caegories suggess ha EAC counries need o underake regulaory reforms o improve indirec axes and direc axes. The reforms could relae o he ease of doing business as well as harmonizaion of service delivery by governmen insuions. The coefficien for polical sabily shows a negaive and saisically significan relaionship wh respec o indirec ax. The variable polical sabily reflecs percepions of he likelihood ha he governmen will be desabilized or overhrown by unconsuional or violen means, including polically-moivaed violence and errorism. This resul sugges ha indirec axes in he EAC counries are negaively affeced by polical sabily. This resul is suppored by [3] who shows ha a sable economy facilaes growh in revenue for low income counries. The coefficien of conrol of corrupion is posive and saisically significan wh respec o rade axes and indirec axes. A reducion in corrupion (implying an increase in he corrupion index) would subsanially increase rade axes and indirec axes. The resul implies ha he conrol of corrupion increases rade axes and indirec axes. The variable conrol of corrupion reflecs percepions of he exen o which public power is exercised for privae gain, including boh pey and grand forms of corrupion, as well as capure of he sae by eles and privae ineress. The posive resul suggess ha conrol of corrupion in EAC counries could improve indirec and rade ax performance. 5. Conclusions and Policy Recommendaions 5.1. Conclusions Despe he well-known gains from rade, he effecs of rade openness are a pri- 708

20 ori ambiguous. For his reason, s imporan o esablish effecs of rade openness on differen sources of governmen revenue for any counry opening s borders o rade. This sudy esablishes he effecs of rade openness on direc, indirec, rade as well as oal ax revenue. A panel daa coinegraion echnique ha uses he Fully Modified Ordinally Leas Squares and Dynamic Ordinally Leas Squares was employed. The daa are annual cross counry panels of Eas Africa counries covering he period The daa were obained from he IMF s Inernaional Finance Saisics, he African Developmen Bank s African Economic Oulook and he World Bank s World Developmen Indicaors and he World Governance Indicaors, which is an imporan conribuion of he paper o rade openness leraure. Afer esing for robusness of he resuls, our main findings show ha he ariff rae posively influences oal ax, indirec and rade axes bu he relaionship beween he ariff rae and direc axes is found o be insignifican. The sign of he coefficien of average ariff rae squared is negaive and saisically significan for oal ax, indirec and rade ax. The resuls illusrae he expeced Laffer effec i.e., he ax-maximizing ariff rae. The ax maximizing ariff rae is esimaed a approximaely 9.2 percen for oal ax, 12.2 percen for indirec ax and 14 percen for rade axes for he counries in his panel. Thus, for counries in his group, an increase in ariff raes beyond hese raes would resul ino a decrease in revenue for he hree ax caegories. In addion o he rade openness variable, he following variable are found o posively influence axes in EAC counries, including average ariff rae, rade raio o GDP, GDP per capa, he raio of urban populaion o oal populaion, foreign aid and conrol of corrupion. While he following variables are found o negaively influence ax performance i.e., deb sock as a percenage of gross naional income, inflaion, real effecive exchange rae, share of agriculure secor o GDP, regulaory qualy, polical sabily and rule of law Policy Recommendaions The policy implicaion is ha governmens of EAC counries should implemen rade openness policies, paricularly reducing of he ariff rae o help in improving ax revenue performance. Acknowledgemens I acknowledge work by Micah Samuel Gaalya (2015) who previously focused on Trade liberalizaion and ax revenue performance in Uganda. This sudy exends he sudy by using a more robus esimaion echnique he Fully Modified Ordinally Leas Squares and Dynamic Ordinally Leas Squares for seleced Eas African Counries. References [1] World Bank (2012) Revenue o GDP Raio, Alas Mehod [Daa File]. hp://daa.worldbank.org/indicaor/gc.tax.totl.gd.zs?view=char 709