A MULTI-DIMENSIONAL APPROACH TO THE DETERMINANTS OF TAX REVENUE: THE CASE OF THE STATE OF JAMMU AND KASHMIR (INDIA)

Transcription

1 A MULTI-DIMENSIONAL APPROACH TO THE DETERMINANTS OF TAX REVENUE: THE CASE OF THE STATE OF JAMMU AND KASHMIR (INDIA) Samir- ul-hassa., 1 Research Scholar, Departmet of Ecoomics, North Easter Hill Uiversity Shillog, Meghalaya, Idia,793022, Samirul.hassa@yahoo.com Prof. Biswambhara Mishra., 2 Professor: Departmet of Ecoomics, North Easter Hill Uiversity Shillog, Meghalaya, Idia,793022, bmishra_ehu@hotmail.com P. Sriivasa Suresh 3 Associate Prof: Departmet of Ecoomics, North Easter Hill Uiversity, Shillog, Meghalaya, Idia, Sureshp03@yahoo.co.i ABSTRACT: The state of Jammu & Kashmir is oe of the special category states of Idia, that faces a severe resource cruch o the oe had ad a explosive public expediture tred o the other had. The iability of the state govermet to raise adequate resources of its ow cast s serious doubt about the tax efforts carried out by the govermet from time to time. Agaist this backgroud, this paper tries to aalyze the major log ad short ru determiats of tax reveue i the state of Jammu ad Kashmir by applyig recet ecoometric methods such as Autoregressive Distributed Lag (ARDL) ad by takig a broader set of variables which comprises ecoomic, political ad demographic dimesios. The result shows that all the ecoomic variables, except for the share of agriculture ad the uemploymet rate, have positive ifluece o the tax reveue. Regardig political stability variables, some like political crises ad law ad order are sigificat, while others like the electio cycle were foud to be isigificat. Iterestigly, both the variables of the demographic dimesio, viz., the seasoal break i populatio desity ad urba populatio, were foud to be isigificat betwee to ad sigificat from to to the chages i the tax reveue of the state. Key words: Tax reveue, Ecoomic, ARDL, Political stability, Populatio, itegratio JEL Classificatio: H2, H7, H3, H71, H26, H23, E62 1. Itroductio The state of Jammu ad Kashmir is oe of the special category states of Idia, which is typically characterized by a greater depedece o agriculture. Aroud 70 percet of its populatio depeds o agriculture as a mai source of livelihood. The regio is also uique with its great potetial i tourism. Sigificat

2 developmet has bee witessed i differet spheres of ecoomic life i recet years. Yet access to opportuities for a reasoable miimum stadard of livig i the state is comparatively lower to that of other special category states of the coutry. The fiscal health of the state is by o meas ecouragig at all, where the states ow tax reveue cotributes hardly 19.7 percet of the total reveue receipts of the state. I the state where ow tax reveue cotributes o more tha 13 percet of the state icome, the aggregate govermet expediture costitutes as high as 48.9 percet. As a result, the state has developed a depedecy sydrome ad that is evidet from a explosive cycle of public expediture growth. Coupled with this, there is a icreasig demad for grats-i-aids ad other Cetral assistace to help bridge the gap of large budgetary deficits. This reflects a iadequacy o the part of the state govermet to geerate eough resources to meet the chagig volatile fiscal situatio. There are umber of reasos that ca be attributed for this poor state of fiscal health of the state govermet. We believe that the major factors that have bee resposible are (i) static tax base due to low level of ecoomic activities which might have bee due to level of ifrastructural developmet, (ii) emergece of a parallel ecoomy due to various tax prefereces that the govermet accord from time to time ad (iii) political ad ecoomic itolerace to the expaded ecoomic activities, ad the social urest that the state ecoomy experieces from time to time Therefore, the repercussio from all these forces at work might have resulted i various leakages ot oly i tax geeratig capacity but also i arrowig dow the tax base of various taxes i the state. If we are to assig a cause-effect relatioship to this type of vexed problem the it ca be argued that the failure o the part of the state govermet o the resource mobilizatio frot, which has bee maily resposible for their low level of ecoomic activities, low level of ecoomic base ad their fial culmiatio i the form of social urest. I a moder welfare state, fulfillmet of social desire to have a better quality of life is depedet ot oly o the capacity of the govermet to mobilize adequate resources but also o the degree of mometum of the ecoomic activities that a state i questio attais. Ay jolts to this by the erratic behavior i the social, ecoomic ad political istitutios of the society at large proves to be a hidrace ot oly to the expaded ecoomic activities but also arrows dow the tax base of the ecoomy i questio. The iterplay of these two forces ca be take as a startig poit for ay systematic attempt to explai the social, ecoomic ad political implicatios of the tax effort of the state to attai a reasoable degree of sustaiable ecoomic growth with a scietific ad reliable ecoometric model. With this itesio, Autoregressive distributed Lag (ARDL) ad multiple regressio models has bee used for time series data for period of 30 years ( ) to explai the

3 social, ecoomic ad political implicatios of the tax effort i the state of Jammu ad Kashmir. The result shows that all the ecoomic variables, except for the share of agriculture ad the uemploymet rate, have positive ifluece o the tax reveue. Regardig political stability variables, some like political crises ad law ad order are sigificat, while others like the electio cycle were foud to be isigificat. Iterestigly, both the variables of the demographic dimesio, viz., the seasoal break i populatio desity ad urba populatio, were foud to be isigificat betwee to ad sigificat from to to the chages i the tax reveue of the state. The results idetify that Chages i political ad ecoomic variables have a larger impact o the level of Tax reveue due to the matter that most of the ecoomic activates i the state are subjected to the peace coditio ad level of ormalcy. The slow growth of ecoomic activities ad large exemptio of taxes has also made these variables ielastic. O other had demographic determiats are positively correlated with the growth of Tax reveue. The socio-ecoomic ad political characteristic prevailig i the state of Jammu ad Kashmir is more or less same to most developig coutries. The huge gap betwee reveue ad expediture, poor ifrastructure, mass social ad ecoomic iequalities, uemploymet, lack of techology, burde of debt ad political istability are the commo features of Jammu ad Kashmir ecoomy ad so as of differet developig ad uder developed coutries. It is with all these forces, that the efficiecy to geerate reveue from ow sources has reduced cosiderably over the years. Therefore, by aalyzig the determiates of tax reveue i the state of Jammu ad Kashmir with these broader dimesio, we ca geeralize a idea how far the socioecoomic ad political setup of a regio or a uderdeveloped coutry ca affect the growth i tax reveue. 2. Research Objectives With the above backgroud, the preset study iteds to make a i depth aalysis of: 1). Aalyze the ecoomic determiats of tax reveue of the state of Jammu ad Kashmir. 2). To idetify the major political ad demographic determiats of tax reveue i the state of Jammu ad Kashmir 3). To aalyze the tedecy of the variables to brig the log ru equilibrium i tax reveue. 3. Fiscal sceario of Jammu ad Kashmir As poited out i the precedig paragraphs, the ecoomy of the state depeds mostly o traditioal forms of occupatio ad agriculture still remais the pivotal of all other ecoomic activities i the absece of desired level of idustrializatio. The idigeous traditioal occupatio of farmig, aimal husbadry, tourism ad horticulture forms the backboe of the ecoomy. Agriculture is the mai source of

4 livelihood i the state where 70 % of populatio ekes out their livig from agriculture, ad 49 % of total workig force directly depeds o this sector for their livelihood. The slow growth i agriculture ad allied sectors is a major cause of cocer. It is true that ecoomic developmet i the moder times has come to be associated with idustrializatio, but Jammu & Kashmir has ot bee able to attract ivestmets i this sector ad remaied a idustrially backward state due to its uique ecoomic disadvatages arisig out of remoteess ad poor coectivity, hilly ad ofte ihospitable terrai, weak resource base, poor ifrastructure, sparse populatio desity, shallow markets ad most importatly the political ucertaity. Cotemporary political situatio i Jammu ad Kashmir is well uderstood by the electoral politics of the state sice the assembly electios of Over the last oe decade, the average aual rate of growth of state domestic product has remaied at 4.51 % i , as compared to 5.19 % durig the decade of A disaggregate picture about patter of growth i the state domestic product i the state shows that durig the last decade, the state agriculture grew at a average growth rate of 3.21 % aually, while the average aual growth rate for the idustrial sector stood at 2.10 % durig , as compared to 3.69 %, ad 2.55 % respectively durig the decade Over the years, there has bee a tremedous expasio of the service sector i the state. The service sector has registered a average aual rate of growth of 9.38 % i as compared to 9.03 % i The per capita icome of Jammu ad Kashmir at costat prices i which was Rs i , rose to Rs i as compared to Rs 7164 i The per capita icome of the state has grow at a average aual rate of growth of 4.78 % durig the period of Accordig to the latest comparable data, Jammu ad Kashmir is raked at the 21st positio i terms of per capita icome amog all the Idia states (DES, 2014). The state has highest uemploymet rate of 5.3 % (5.4 % for males ad 3.5 % for females) as compared to its sister states uder special category. At all Idia level the figures of uemploymet for the states is 2.6 % (3.1 % for males ad 3.0 % for females) ad uemploymet is more prevalet i urba areas tha i rural areas of the state, which is uique (MOSPI). With the expasio of the govermet activities, the magitude of pla expediture of the state govermet has icreased tremedously, which i tur has give rise to the eed for a rapid icrease i reveue. It is expected that the sources of reveue should grow automatically at the required rate. But the experiece of the state of Jammu ad Kashmir egates the above propositio. As a result of which, this has created a wideig gap betwee the state s expediture resposibilities o the oe had, ad available resources o the other, thereby givig rise to the problem of attaiig a appropriate degree of fiacial self-reliace o the part of the state

5 govermet. The performace of the state o the resource mobilizatio frot provides rather a poor ad dismal picture. It is worth metioig here that states ow tax reveue ad cetral share of taxes ad duties are two mai sources of total tax reveue of the state. The tred ad growth of tax reveue i the state of Jammu ad Kashmir ca be predicted from figure 1.1 ad 1.2 below. The figures show the tred i growth of state tax reveue over last thirty years from to with both curret ad costat prices (usig GDP deflator). The figure shows that over the years the tax reveue of the state has show icreasig tred i both curret ad costat prices but with a cosiderable fluctuatio. It ca be see from the figure that the tax reveue of the state at curret prices was growig very less i 80 s especially till It might be due to the low collectio of taxes, mass tax evasio ad tax exemptio i this period due to slow ecoomic activities i the state, slow growth of trade ad busiesses etc which was because of prevailig political turmoil i the state durig this period which affect tax base ad tax rates. (Refer Fig.1.1) While as, a brisk tred i growth i tax reveue starts from 2000 owards whe the ecoomic activities i the state start growig slowly ad the political urest has slow dow as well. If we take look at curret positio of tax reveue of the state, it shows a upward tred, were tax reveue is icreasig upwards; it might be due to the impositio of taxes which were exempted i 90 s period, which icrease the tax base of the state ad thus icrease the tax reveue. Similarly at costat prices a similar upward ad dowward tred ca be predicted i the growth of tax reveue i the state. The slow ad decliig growth over certai years might be due to the social tesio i the state, were the militacy has ruied each ad every ecoomic as well as social sector of the state. Similarly, the aual growth of tax reveue over the years is also showig a fluctuatio tred. The total tax reveue of the state was growig at 19.1 % per aum betwee the period to , the growth i this period might be due to the ability of the state to mobilize its resources by differet ecoomic activities like tourism, idustries, horticulture, trade etc. But durig the period to the aual growth of tax reveue has decreased to 10.4 % per aum which might be due to severe coditios i the state, which disturb the whole ecoomic setup of the state ad most of the ecoomic activities have come to a stadstill. I the last 10 years from to , the tax reveue of the state, has icreased at 18.9 % per aum. Ad it might be due to the improvig coditios i the state ad growth of the ecoomy through icreasig idustrial activities ad trade i Jammu, ad tourism ad horticulture i Kashmir. The states ow tax reveue has costatly show a growig tred over the period of time. The aual growth

6 rate of states ow tax reveue was % per aum i the first te years of study period i.e to , which icreased to % per year i ext te years ad further to % per aum, over the period to It shows that the states ow tax reveue is growig at the rate of 15.9 % per aum, over the period from to , with icreasig tred i growth The mai sources of states ow tax reveue like, VAT, Services Tax, GST, Passeger tax, Registratio fee, stamp duty, Toll ad Excise duty, Vehicle tax ad Electricity duty tax have fluctuated a lot over the last thirty years, due to chagig political ad ecoomic status i the state. While as the cetral share of taxes ad duties was growig i the first phase of the period at 26.65% per aum ad it has reduced betwee to to 4.95 % per aum ad icrease to % durig to I last thirty years the cetral share of taxes ad duties is growig at % per aum while as the ow tax reveue is growig at 15.9 % per aum (ASFRs). Thus the overall tax structure of the state has goe through a difficult period which ot oly reduced the efficiecy of state to collect reveue through taxes but also hampered the potetial, by destructio of major sources of taxes reveue. It is oly sice last years, that the state has etered ito a phase of trasformatio ad ecoomic growth which opeed ew ways ad base for growig reveue through taxes. But still due to earlier destructio of sources of reveue, the growth of reveue through taxes is very low. Therefore, all these observed treds oted above provide a solid groud for the ecessity ad the desirability of udertakig a aalysis of the determiats of taxatio of the state of Jammu ad Kashmir, to have a proper uderstadig of the factors which have bee resposible for pushig up ad dow the tax reveue or keepig the level of taxatio rather at a miimal level o the other had. May istitutioal, ecoomic, demographic ad political variables affect fiscal outcomes. Further eve with the emergece ad growth of public choice as a ew perspective from which to examie the operatios of govermets, the cosesus view asserted by Dye i 1984 remaied at least a implicit assumptio of efforts to idetify determiats of taxatio. It is evidet from the above discussio that over the last thirty years, the basic macro-ecoomic idicators of ecoomic developmet has remaied at a pathetically lower level. This provides eough evidece that the ecoomic activities vis- a- vis the tax base of various taxes staggered at a low level of vicious circle. As a result, the state has ot bee able to geerate sufficiet reveue from its ow resources ad has bee facig serious fiacial problems [41].The problem became all the more serious due to the prevailig circumstaces i the state affectig both reveue ad expediture. The state suffered from political dispute for a log period, sice 1989 owards, resultig i the erosio of the tax base, icrease i expediture due to destructio

7 of ifrastructure ad various other factors related with disturbed law ad order. Thus, havig all those costraits i the ecoomy ad i the regio, the importace of mobilizig the iteral reveue for overall developmetal process i the state has become a promiet issue of the state. Taxatio is a importat mechaism to geerate ad mobilize iteral reveue ad stregthe the fiacial system ad attai fiacial self sufficiecy. Therefore, the paper is a attempt to look i to the itricate relatioship betwee a set of complex socioecoomic ad political variable for determiig the major determiats of tax reveue of the state to ascertai whether these variables have played ay role i resource mobilizatio process of the state or they have bee proved detrimetal i the way of tax geeratig capacity of the state. Keepig cosistecy with the above metioed objectives, the study iteds to test the followig hypotheses. 4. Hypotheses 1. Chages i political ad ecoomic variables may have a larger impact o the level of Tax reveue. 2. Demographic determiats are positively correlated with the growth of Tax reveue. 5. Review of literature Over the years ecoomists ad researchers have foud differet factors that affect the growth of tax reveue. Amog them the most importat are factors from ecoomic, social, demographic ad political spheres. [54] i their study of determiats of taxatio used pael data from 30 coutries over the period ad foud that, the share of agriculture ad miig i GDP has a egative impact o tax reveue. However, export share i GDP ad per capita GDP are positively ad sigificatly associated with tax reveue performace. [47] Foud that per capita icome ad the ratio of trade to GDP are positively strog determiats of tax reveue, whereas, share of agriculture i GDP is egatively associated with tax reveue. [12] foud that a tax rate is positively related to the populatio size of the commuities eve whe cotrollig for desity. [30] foud that tax reveues i Turkey are sigificatly affected by agricultural, idustrial sector share i GDP, foreig debt stock, moetizatio rate of the ecoomy ad urbaizatio rate, while the agriculture share i GDP foud egatively associated with the tax reveue. The results also suggest that opeess to foreig trade has o sigificat impact o tax reveues i Turkey. [55] foud that tax evasio, agriculture ratio ad populatio desity determie the tax reveue i Ugada. He revealed that tax evasio is the most importat factor which reduces the tax reveue i the coutry. [58] has empirically ivestigated the determiats of value added tax i Keya. His study showed that GDP, chage i level of tax, istitutio ad demographic variables determie the VAT reveue i Keya. [35] made a attempt to

8 idetify the obstacles of tax reveue geeratio i developig coutries. His study showed that the structure of ecoomies, tax systems, patters of political system, ad low icome of these coutries are resposible for their low tax reveue geeratio. [36] explai i a empirical aalysis of determiats of tax reveue i Nigeria that tax reveue teds to be sigificatly resposive to chages i icome level, exchage rate ad iflatio rate. He cocludes that macroecoomic istability ad level of ecoomic activities are the mai drivers of tax buoyacy ad tax effort i Nigeria. [10] foud that the quality of istitutios ad resource reveues are strog determiats of tax ratio i GDP. His study fids that Per-Capita GDP ad trade opeess improves the tax ratio i GDP. He also idetifies that the structure of value-added, agriculture, service ad idustry shares are strog detrimets of the tax ratio of GDP. [18] foud that the tax collectio rate (especially direct taxes) i Armeia did ot icrease with the same pace as GDP. They also foud that istitutioal quality, urbaizatio ad shadow ecoomic activity are the mai factors behid low tax-to- GDP ratio i Armeia. [27] aalyzed the determiats of tax reveue i developig coutries where, he foud that the structural factors such as per capita GDP, agriculture share i GDP, trade opeess ad foreig aid sigificatly affected tax reveue performace of a ecoomy. He also showed that corruptio, political stability ad share of direct ad idirect taxes also determies tax reveue i developig coutries. [25] exteral coflicts do ot icrease the fiscal capacity of the states, if the duratio of the coflict is short or if the coflict does ot ivolve may coutries, as occurred i the case of the US ivasio of Paama i [3] Fids that viable state ad sustaied peace is essetial for costructio of the Tax Reveue Base. [38] made a attempt to study the tax performace ad its determiats of some Idia states by takig data for the period of They employed a multiple regressio equatio to measure the impact. They ivestigated the relatio with the explaatory variables like per capita icome, degree of urbaisatio as measured by the ratio of urba populatio to total populatio, share of o-agricultural icome i total state icome ad per capita developmetal expediture. [7] foud that the agriculture, export ratio ad miig share i GDP as importat variables of tax reveue determiatio. [1] ivestigate the determiats of tax reveue, were he has used the direct ad idirect taxes as a explaatory variables. His study compares the determiats of tax reveue i Idia ad Pakista o these two variables. His results show that Pakista is geeratig more tax reveue through idirect taxes whereas Idia from direct taxes. [51], [42] aalyzed the tax efforts i poor states of Idia. They show that factors such as per capita SDP, proportio of urba populatio ad degree of literacy have sigificat impact o the tax efforts or tax reveue i the poor states of Idia. [20] foud

9 that the govermet expediture has a sigificat impact o the govermet tax reveue i Idia. [8] idetify the mai determiats of tax reveue with referece to twety two states of Idia, by employig multiple regressio models. Their study showed that per capita deficit, urba populatio, per capita expediture ad per capita icome of the states has sigificat impact o tax reveue while as primary sector icome, literacy rate, desity of populatio, schedule cast populatio ad political variables are ot sigificat. [37]measure the horizotal imbalace betwee reveue ad expediture i Idia. Their study shows that Variatios i tax base, tax effort, ifrastructural facilities - both physical ad social - ad political ucertaity are importat determiats of horizotal imbalaces betwee reveue ad expediture i Idia. [53] made a attempt to idetify the determiats of idividual taxes ad their aggregate i the state of Gujarat for period They worked o the some ecoomic ad demographic variables. [49] foud the effect of various ecoomic ad political variables o the tax reveue i four Idia states amely Karataka, Orissa, Kerala ad West Begal. His study shows that per capita icome, share of Agriculture i SDP, cosumer price idex ad Deficit has sigificat relatio with tax reveue i these states, while Political variables like political ideology has o sigificat relatio with tax reveue i these states. [21] i his study shows that high ecoomic subsidies reduce the o tax reveue i Gujarat. [19] Study the tax efforts of the state of Pujab for period of He cosiders four major ecoomic variables to examie the determiats of tax reveue i Pujab. [43] aalyzed ad showed that icrease i Icome ad a chage i prices have sigificat impact o the growth of Tax reveue i Nagalad. 6. Data Sources ad Methodology The study tries to aalyze the impact of differet ecoomic, political ad demographic compoets o the growth of tax reveue i the state of Jammu ad Kashmir. I the study we ited to use the data set for the period, to , (which is cosidered as a importat period for chagig the ecoomy as well as political setup of state) for the variables like tax reveue i NSDP, per capita icome, Idirect taxes [58], [1], total outstadig [57], [44], Share of Agriculture to NSDP [49], [27], Share of Idustries to NSDP, share of Services to NSDP, [54], [30], share of Exports to NSDP [7], rate of uemploymet, Populatio desity, [12], [55], Urba populatio [8], [34], Political crisis, [29], [22], Law ad order, [3], [27], [9], ad electio cycle,[35] ad [37],. The variables chose for the study represet ecoomic, political ad demographic status of the state which by our uderstadig directly or idirectly affect the tax reveue or are importat sources of tax reveue i the state. The state brought its reveue by lavig taxes ad duties o agriculture, maufacture ad services sector i which services sector is highest

10 cotributor to the state ecoomy ad to taxes as well. Thus these three variables have impact o tax reveue. The state has high export of taxable primary products which geerate bulk of reveue, so its share i taxes as well. Per capita icome i the state has bee icreasig, so the moey i the hads of the people icreases, so higher opportuity of tax emerges. The ecoomic activities have ormally started to grow i the state, which results growth i icome, ad thus ope sources to impose differet idirect taxes. The rate of uemploymet either reduces tax reveue or icreases deped upo tred, the state is cout i the highest uemploymet regios thus its effect o tax reveue i state ca be positive. Higher populatio desity ad urbaizatio meas high icome groups came ito existece ad thus affect tax reveue. The political stability i the state has always a importat issue for ruig the public activities smoothly [9]; the state has goe through log period of political ad law ad order crisis which reduces the growig stregth of ecoomy ad sources of tax reveue as well. Further, the year of electio, brig more focus of favour groups to give may tax relaxatios to gai their help i comig electio. The study uses time series data collected from RBI ad other state govermet authorities. The variables has bee coverted ito real prices usig GDP deflator ad also ito atural log equatios for time series so that the coefficiets represet the Elasticity [26]. As our prime aim is to uderstad the ecoomic, political ad demographic determiats of tax reveue thus three regressio models have bee used separately for each determiat i order to avoid multi-colliarity issue. We employ the autoregressive distributed lag (ARDL) approach [46] ad [45], to test for existece of a relatioship betwee ecoomic variables ad tax reveue ad to obtai robust results [33], [1] ad [5]. While as multiple regressios were used for political ad demographic determiats like [55], [54] ad [49]. Estimates provided by ARDL model avoid problems such as autocorrelatio ad edogeetiy, they are ubiased ad efficiet. Autoregressive distributed lag (ARDL) is the combiatio of both autoregressive models ad distributed lag models. So, a time series is ot oly a fuctio of its lagged values but also the fuctio of curret ad lagged values of oe or more regressors. ARDL techique has several advatages ad it has superiority over other ecoometric techiques which are used for logru relatioship (Ahmed, 2016). I this paper ecoomic determiates has bee divided i to two ARDL equatios i order to avoid multicolliarity issue which ca affect the sigificace if the variable. The defiitio of Variables ad the basic form of variables i two ecoomic determiats equatio ad of political ad demographic models is as uder: Basic from of Ecoomic determiats model TAX REVENUE = f(idirect taxes, icome from Agriculture sector, icome from services

11 sector ad value of exports) (1) TAX REVENUE = f(total outstadig of govermet, icome from idustry sector, Per capita icome ad rate of uemploymet) (2) Where Tax Reveue (Tr): The reveue collected by the state govermet through taxes, it is the total collectio of direct ad idirect taxes. Idirect taxes (idtax): Reveue collected from Taxes levied o goods ad services rather tha o icome or profit. Icome from agriculture (sagi): Net icome geerated by the state through Agriculture ad allied sectors. Icome from services sector (sserv): Net icome geerated by the state through services sector Value of Exports (sexpo): Moetary value of exports the state has geerated through export of goods ad services i a fiacial year. Rate of Uemploymet (uemp): Rate of uemploymet is the situatio of uemploymet i the state. It is the rate at which uemploymet icreases. Basic form of Political determiats model TAX REVENUE = f(political crises, Law ad order ad electio cycle) (3) Where Political crises (pcrises): The chage i rulig from elected govermet to goveror s rule Law ad order (Law): Situatio of strikes, protests ad civilia killigs i a fiacial year. Electio cycle (elecy): The year i which electio was held i the state. Basic form of Demographic determiats model TAX REVENUE = f(populatio desity ad rate of Urbaizatio) (4) Where Populatio desity (pod): Number of people per sq km i a fiacial year Rate of urbaizatio (urb): Populatio livig i urba ceters likes tows ad cities. (Refer Table No. 1.1) 6.1 Estimatio procedure Lag Legth Criteria The ARDL model allows each variable to have its ow lag optimal lag legth structure. I estimatig the ARDL model used for ecoomic determiats i this paper, we applied the Akaike Iformatio Criterio (AIC) to arrive at the optimal lag structures for each of the variables i Equatio (1) ad (2) used i our aalysis Statioary test/uit root test Statioarity test of a time series is a importat procedure to avoid spurious regressio results. The statioary test is carried out to measure the reliability of the time series variables. The time series statioarity is a statistical characteristic of a series like its mea ad variace [26], So if i a series, both mea ad variace are costat over time the the series has o uit root or is statioary, otherwise if ot costat over time,

12 the the series has a uit root or is o statioary, ad thus we eed to chage the series i to respective differeces. The differecig procedure is set o observatio as first differece ad secod differece o itercept, tred ad itercept or without tred. I this paper Augmeted Dickey Fuller (1979) ad Phillips ad Perro (1988) tests have bee used for statioary test. These test aalyze the equatios like X level x 1 x 1st-diifereced value x x t t 1 x 2d-diifereced value x x t t 2 The hypothesis tested for each variable for statioarity ad o-statioarity are: The ull hypothesis will be H 0 : (α0,) = (α0, 0, 1) (No Statioarity) The alterative hypothesis H 1 : (α0, ) (α0, 0, 1) (Statioarity) After aalysis if a series is statioary without differece or i other words is statioary at level it will be as I(0) form or itegrated as order 0. O other had if a series is statioary at 1 st differece it will be desiged as I(1) form or itegrated as order 1. Similarly if series is statioary at 2 d differece it will be cosidered i I(2) or itegrated as order Estimated models Ecoomic determiats model As discussed, the study has bee divided i to three ecoometric models to idetify the sigificat variables from ecoomic, political ad demographic dimesio which affect the tax reveue collectio. Autoregressive Distributed Lagged (ARDL) model has bee coducted to kow the ecoomic determiats of tax reveue as per the statiority results, while multiple regressio method has bee used for political ad demographic determiats. Autoregressive Distributed Lagged (ADRL) is a modelig techique which allows each variable to have its ow lag optimal lag legth ad adds error correctio features to a multi-factor model to uderstad the log ru as well as short ru relatioship amog the variables after kowig that the variables are havig itegratio order of either I(0) or/ad I(1) ad are havig log ru coitegratio [56], [11], [40] ad [32]. The study follows the approach adopted by [27], [33], ad[6] to develop our model for the study. We have divided the ecoomic variables further ito two ARDL model equatios i order to avoid the problem of multi-collariity [30]. The subsequet ARDL models for equatio (1) ad (2) of ecoomic determiats are show below:

13 Dltr t = α 1 + δ 1 (ltr t i ) + δ 2 + (lidtax t i ) + δ 3 (lsagr t i ) + δ 4 (lserv t i ) + δ 5 (lexpo t i ) + β 1 Dltr t i + β 2 Didtax t i + β 3 DSagr t i + β 4 DSserv t i + β 5 Dsxpo t i + ε 1t (5) Dltr t = α 2 + θ 1 (ltr t i ) + θ 2 (outstad t i ) + θ 3 (lpci t i ) + θ 4 (lsid t i ) deotes icome from services sector ad expo deotes value of exports i equatio (5). While as outstad deotes total outstadig of debt, Pci deotes per capita icome, sid represets icome from idustry sector ad uemp deotes rate of uemploymet i equatio (6). α 1 ad α 2 are the itercept coefficiets of the two equatios. δ 1, δ 2, δ 3, δ 4 ad δ 5 are the correspodig log ru multipliers whereas β 1, β 2, β 3, β 4, ad β 5 are the short ru dyamic coefficiets of the respective ARDL model equatio (5). Similarly θ 1, θ 2, θ 3, θ 4 ad θ 5 are the correspodig log ru multipliers whereas γ 1, γ 2, γ 3, γ 4 ad γ 5 are the short ru dyamic coefficiets of the respective ARDL model Equatio (6). ε 1t, ad ε 2t are the white oise error terms of the two ARDL models. The hypothesis of both the equatios is tested o probability value of t- statistics at 5% ad 10 % level of sigificace Boud testig for co-itegratio of + θ 5 (luemp t i ) + γ 1 Dltr t i ecoomic determiats + γ 2 Doutsatd t i + γ 3 DPci t i ecoomic determiates ad tax reveue is examied usig the ARDL bouds testig + γ 4 DSid t i + γ 5 Dump t i The log-ru relatioship betwee variables from procedure. The boud test has bee employed to aalyze the presece of coitegratio amog the + ε 2t (6) variables [46] ad [45]. Boud testig ca Where D is the differece level of the variable idetify the log ru relatioship with a ad l is the atural log. Tr represets tax depedet variable followed by its forcig reveue, idtax represets idirect taxes, sagr variables. The F-test statistic of bouds test for deotes icome from agricultural sector, serv Equatio (5) ad (6) will be examied o the

14 basis of critical value at 5% level of sigificace i order to establish log ru relatioship betwee the variables i these two equatios. The ull hypothesis of o coitegratio through ARDL boud testig i ARDL model Equatio (5) ad (6) is δ 1 = δ 2 = δ 3 = δ 4 = δ 5 = 0 ad δ 1 = δ 2 = δ 3 = δ 4 = δ 5 =0. The hypotheses are tested by computig the geeral F-statistics ad comparig them with critical values [46] ad [45]. After the ARDL boud testig for log ru co-itegratio of ARDL model (5) ad (6), If log ru relatioship exists betwee the ecoomic variables i both the models, the log ru parameters ca be estimated by usig the followig models for both equatio (5) ad (6): ltr t = α 1 β 1 ltr t i + β 2 idtax t i + β 3 Sagr t i ltr t = α 2 γ 1 ltr t i + γ 2 outsatd t i + γ 3 Pci t i + γ 4 Sid t i + γ 5 ump t i + ε 2t (8) Where l is the atural log of the variables, α 1 ad α 2 are the itercept coefficiets. β 1, β 2, β 3, β 4, ad β 5 ad γ 1, γ 2, γ 3, γ 4 ad γ 5 are the log ru multiplier coefficiets of the respective variables i equatio (7) ad (8). ε 1t, ad ε 2t are the white oise error terms of the two ARDL models Similarly after boud testig of ARDL model (5) ad (6), the short-ru dyamics ca be foud by estimatig the followig equatios for ecoomic determiats: + β 4 Sserv t i + β 5 sxpo t i + ε 1t (7)

15 Dltr t = α 1 + φ 1 Dltr t i Dltr t + φ 2 Didtax t i + φ 3 DSagr t i + φ 4 DSserv t i + φ 5 Dsxpo t i + ECT t 1 + ε 1t (9) = α Dltr t i + 2 Doutsatd t i + 3 DPci t i + DSid t i + 5 Dump t i + ECT t 1 + ε 2t (10) Where D is the differece level of the variable; l is the atural log form of respective variable ad, α 1 ad α 2 are the itercept coefficiets. Parameters φ 1, φ 2, φ 3, φ 4 a d φ 5 ad 1, 2, 3, 4 ad 5 are the short ru coefficiets of equatio (9) ad (02). The coefficiet of ECM speed of adjustmet towards the log-ru equilibrium. Coefficiet of adjustmet should be egative ad statistically sigificat for covergece Political determiats model The study uses OLS multivariate regressio model, [49] ad [1] to test the political determiats of tax reveue. The dummy variables have bee chose as explaatory political variables like Political crisis [16] ad [22] were 0 is for the years, whe there was political parties rulig, ad 1 whe there was Presidets rule i the state. Law ad order, [3], were 0 whe there were less tha 500 civilia deaths ad 1 whe there were more tha 500 civilia deaths i a year i the state, [29]. Fially electio cycles were 0 for ormal year ad 1 for electio year. The regressio equatio tested for Political determiats of tax reveue is show below: Dltr t = α 1 + δ 1 Pcrisis t + δ 2 law t + δ 3 elecy t + ε t (11) Where D is differece level of the variable, l is the atural log ad α 1 is the itercept of the model. δ 1, δ 2, ad δ 3 are the coefficiets of Political crisis, law ad order ad electio cycle. ε t is the Error term of the model. The coefficiets ad the hypothesis of the model will be tested o probability value of t-statistic at 5 ad 10% level of sigificace. i both equatios represets ECT shows the

16 6.3.4 Demographic determiats The determiats of demographic variables have structural breaks as the demographic variables have isigificat relatioship with tax reveue up to certai period ad sigificat relatio i other period. Before goig to aalyze the determiats of the demographic variable, we will try to obtai the structural break poit ad the divide the period of study accordigly. Chow Breakpoit test has bee used for the structural break Chow Breakpoit test The chow test (1968) is used to test whether a sigle regressio is more efficiet tha two separate regressios ivolvig splittig the data ito two sub-samples (Lee, 2008). The test is used to realize the structural break i a time series data [23]. The chow test is carried out first sigle regressio equatio o the full data. The equatio tested for chow test will be Dltr t = α 0 + β 1 Dpod t + β 2 Durb t + ε t (1) After checkig the structural break the above equatio will be split ito two data set equatio o the bases of structural break poit. The model will the be of two separate equatios as show below Dltr t = α 1 + γ 1 Dpod t + γ 2 Durb t + ε t (2) Dltr t = α 2 + δ 1 Dpod t + δ 2 Durb t + ε t (3) Where α 0, α 1 ad α 2 are the itercept of the Equatios ad β 1, β 2, γ 1, γ 2, δ 1, ad δ 2, are the Coefficiets of the variables i differet equatios. The chow test is estimated o the basis of ull hypothesis which states that α 1 = α 2, γ 1 = γ 2, ad δ 1 = δ 2. The chow test is thus estimated by obtaiig residual sum of squared (RSS) of all the data set before ad after structural break. Let RSS 0 is Residual sum of square of combied data set, RSS 1 is residual sum of square of first data group ad RSS 2 is the Residual su of square of secod data group. N 1 ad N 2 are the umber of observatios i each group ad K is the total umber of parameter estimated(here we estimate 3 parameters). The the Chow test statistics will be Chow Test statistic = (RSS 0 (RSS 1 +RSS 2 )) (K) (RSS 1 +RSS 2 ) /(N 1 +N 2 2K) The test statistic is thus estimated with the F statistic o (N 1 + N 2 2K) degrees of freedom ad o Log likelihood ratio ad compared with the probability value at 5% or 10 % level of sigificace. 6.4 Diagostic tests I order to check the stregth of our models estimated, differet diagostic tests have bee carried out. Breusch-Godfrey Serial Correlatio or LM Test was doe for serial correlatio of the model, ARCH Test (autoregressive coditioal

17 heteroskedasticity) has bee carried for Heteroskedasticity. Similarly, the test for parameter stability of the model has bee performed by the CUSUM statistics ad the Normality test has bee doe through Jarque- Bera test. All the diagostic tests are estimated through ull hypothesis which are tested through the test statistic value of each test ad the probability value at 5% level of sigificace. 7. Results ad Discussio 7.1 Uit root test The Augmeted Dickey-Fuller test was coducted to pretest the variables for uit roots to verify that the variables are ot itegrated of a order higher tha oe. The purpose is to geerate the results free of spurious regressio. Before goig for ADF test the Akaike Iformatio Criterio were used to determie the optimal umber of lags for each variable icluded i the test. Table 1.2 preset the results of the uit root tests both at levels ad 1st differeces. (Refer Table No.1.2) The test results show that the ADF statistics or T- statistic for all the variables at the levels do ot exceed the critical values at 5% level of sigificace which implies that all the variables are o statioary at levels. All the variables have to be checked at first differeces. The ADF test carried out at first differece shows that T- statistic of ADF test is higher tha their respective critical values at 5% level of sigificace, which implies that all the variables are statioary after first differeces. Thus we coclude that all the variables, i.e, tax reveue, share of idirect taxes to total tax reveue, total outstadig, per capita icome, share of agriculture to NSDP, Share of Idustries to NSDP, Share of Services sector to NSDP, value of Exports, uemploymet rate, populatio desity ad urba populatio are havig a itegrated order o I(1), meas that all the variables are statioary at 1 st differece accordig to ADF test. Though our itegrated order of the variables is I(1) we ca use Johase (1988) co itegratio test for estimatig log ru relatioship. But i order to obtai robust results for log as well as short ru, we ca use ARDL method which apply boud test despite the order of itegratio is I(1) ot I(0), [45]. The ARDL approach ca be applied to time series variables irrespective of whether they are I(0), I(1), or mutually co-itegrated [52]. Thus we have applied ARDL model to test the log ad short ru relatioship of the variables uder study. 7.2 Boud Testig The ARDL boud test has bee applied to estimate weather there exist ay log ru relatioship betwee the variables i ARDL model (5) ad (6). Table 1.3 shows the results of ARDL boud test of two ARDL model. (Refer Table No.1.3) The table idicates that there is uique co itegratig relatioships betwee the ecoomic variables i the two ARDL models (5) ad (6).

18 As the ull hypothesis of the two tests is o co itegratio ad it ca be rejected oly if calculated F statistic is higher tha upper critical boud value. Calculated F-statistic of ARDL boud test for equatio (5) is which is greater tha critical value of upper boud at 1%, 5% ad 10%, respectively. It implies that the idepedet variables, like idirect taxes, icome from agriculture sector, icome from services sector ad value of exports i ARDL model equatio (5) have log ru relatioship. So, the ull hypothesis was rejected ad alterative hypothesis was accepted. Similarly calculated F- statistic for ARDL boud test for equatio (6) is which is also greater that critical value of upper boud at 1%, 5% ad 10% level of sigificace which implies that variables of ARDL model equatio (6), like outstadig, per capita icome from idustry sector ad rate of uemploymet have log ru associatio. These results idicate that i all relatioships, betwee the variables i two ARDL models are the forcig variables that move first whe a commo stochastic shock hits the system. Therefore, our two ARDL models for ecoomic determiates of tax reveue, have log-ru relatioship, so we ca ow estimate the log rui ad short ru estimates of the variables to obtai robust results. Also Johase Co itegratio test has bee carried out to kow the log ru relatioship betwee the variables. 8. Results ad discussio of the models 8.1 Ecoomic determiats As we discuss i the methodology sectio that i order to remove the problem of multicolliearity we will split the ecoomic variables ito two ARDL model equatios, to kow the sigificat variable which affects the tax reveue i the state of Jammu ad Kashmir. Havig foud log ru relatioships (i.e. coitegratio) amog tax reveue ad various other ecoomic variables, i the ext step the log ru ad short ru relatioship are estimated usig the selected ARDL model equatio of (7) ad (8) for log ru estimates ad (9) ad (10) for short ru estimates. The estimates log ru ad short ru results of ARDL model (5) are preseted i table 1.4 i pael A ad B. The lag legths of (1,2,2,1,1) for idepedet variables are determied by Akaike Iformatio Criterio(AIC) followig the suggestio of [46]. Tests for models of Tax reveue as depedet variable ad idirect taxes, icome for agriculture sector, icome from services sector ad value of exports as idepedet variable, miimum of 1 lag for depedet variable has fixed to esure lagged explaatory variables are preset i the error correctio model (ECM). (Refer Table No. 1.4) The log ru estimates of variables like idirect taxes, icome from agriculture sector, icome from services sector ad value of exports of equatio (5) obtaied from equatio (7) i pael A, reveals that idirect taxes, icome from services sector, icome from agriculture sector

19 ad value of exports are the key determiats of tax reveue. The log ru impact of idirect taxes has positive ad sigificat impact o tax reveue as expected. 1% icrease i idirect taxes will lead to 0.86% icrease i tax reveue. Idirect taxes like sales tax, excise duty, stamp ad registratio duty etc are the taxes easily collected by the govermet over the years thus with icrease idirect taxes the tax reveue icreases. The result is i tue with the fidigs of [58] ad is statistically sigificat at 1% level of sigificace. Agricultural icome to NSDP is egatively related to tax reveue collectio. 1% percet growth i agriculture icome to NSDP will reduce tax reveue by %. It is statistically sigificat at 1 percet level ad idicates that more share of agriculture sector reduces the tax reveue. Agriculture has almost 29 percet cotributio i GDP of Jammu ad Kashmir but its cotributio i tax reveue is almost 1 percet because of low tax o the icome from agriculture sector. [56] ad [9] support this egative relatioship of icome from agriculture sector to tax reveue. The sig of icome from services sector is positive ad is statistically sigificat at 1% level of sigificace. It implies that i log ru 1% icrease i icome from services sector icrease the tax reveue by 0.30%. The results are i lie with [30]. Similarly the value of exports also shows positive ad sigificat relatioship with tax reveue. It implies that 1% icrease i value of exports i the state will icrease the tax reveue by 0.343%. It reveals that with the icrease of export value of goods i the state the tax reveue will also icrease. These results are also supported by [49] ad [27]. Next step is to estimates of short ru dyamic coefficiets of equatio (5) obtaied from equatio (8). The short ru dyamic results are provided Peal B i table 1.4. I terms of sigs ad sigificaces, the results are geerally cosistet with the log ru fidigs. The table reveals that all the variables are statistically sigificat i short ru to produce chage i tax reveue but the tame lag impact differs i each variable. The table shows that Idirect taxes at lag 1 (Accordig AIC criteria) are sigificat determiats i the short ru. The short ru error coefficiets show that previous year idirect taxes has positive ad sigificat impact o the curret year s tax reveue at 1% level of sigificace. It shows that 1% icrease i idirect taxes at lag 1 will icrease the tax reveue at 0.96%. The share of agriculture icome shows egative but has a sigificat impact o curret year s tax reveue at lag 1 at 5% level of sigificace but positive ad isigificat at lag 2 at 5% level of sigificace. The short ru results of error coefficiet model shows that, at lag 1 of SAGR, 1% icrease i SAGR i previous year will reduce the tax reveue of curret year at -0.71%, ad at lag 2, 1% icrease i SARG will icrease tax reveue by 0.07% but is isigificat at 5% level of sigificace. Icome from services sector ad value of exports also shows positive ad

20 sigificat impact o tax reveue i short ru. The coefficiet of share of services sector to NSDP shows that it has positive ad sigificat impact o tax reveue at both the time lags at 5% level of sigificace. It implies that, 1% icrease i services sector icome at lag 1 will icrease the tax reveue by 0.129% ad by 1% icrease i services sector icome at lag 2 will icrease tax reveue by 0.14% as the variable is sigificat at 5 % level of sigificace. The value of exports i NSDP also shows that it has a positive ad sigificat impact o tax reveue i short ru. The results obtaied for ARDL model (5) with ARDL model equatio (7) ad (8), are satisfactory i terms of Jammu Ad Kashmir State is cocered. As idirect taxes are major sources of tax reveue, so the effect of Idirect taxes will be more o tax reveue also the less tax base ad exemptio of various direct taxes over log period of time i the state, like commercial taxes, wealth taxes, property taxes etc have icrease the importace of idirect taxes i the state. Also, the agriculture sector of the state is ot taxed much, so icreases i share will reduce tax reveue. As far as services sector of the state is cocered, it is the oly growig sector of the ecoomy but due to lot of costraits like ifrastructure of the state ad law ad order problems, the sector also shows less coefficiet to tax reveue, but as the SSERV variable has positive impact o tax reveue it is due to the tourism sector ad telecom sector. The state is kow for its hadicraft ad hadloom works which geerates goods of export quality thus as the share of exports to NSDP has icreased over the years the tax reveue has also icreased. The error coefficiet of the Error Correctio Term (ECM) which is deoted by ecm(-1)) is egative( ) ad statistically sigificat at 5% level of sigificace. It reveals the evidece of fast pace of respose to brig equilibrium i tax reveue whe there are shocks i short ru. The egative coefficiet of error correctio model determies the speed of adjustmet to log-ru equilibrium by the idepedet variables. The egative coefficiet is a idicatio that ay shock that takes place i the short-ru by the idepedet variables metioed i above model would be corrected i the log-ru. It shows that ay fluctuatio caused i previous years, or i the short ru will brig equilibrium i log ru at 71% or i other words it meas that it will take at least two years to restore ay disequilibrium i tax reveue. The rule of thumb is that, the larger the error correctio coefficiet (i absolute term), the faster the variables equilibrate i the log-ru whe shocked [2]. The R 2 of equatio (.9878) suggests that 98% variatio i the tax reveue is explaied by the variables used i the model Diagostic Tests Various diagostic tests have bee carried to test the goodess of fit of the ARDL model equatio (5). Breusch-Godfrey (LM Test) was carried out to kow whether the model has the problem of