THE DEGREE OF COMPLIANCE BASED ON EXCISE DUTIES IN ROMANIA BETWEEN 2002 AND 2015

Transcription

1 DOI: /sues THE DEGREE OF COMPLIANCE BASED ON EXCISE DUTIES IN ROMANIA BETWEEN 00 AND 015 Slva Paula Todor Ph.D. student Faculty of Fnance The Bucharest Unversty of Economc Studes, Romana e-mal: Rodca Ghur PhD student Faculty of Fnance The Bucharest Unversty of Economc Studes, Romana e-mal: Petre Brezeanu Ph.D. Professor Faculty of Fnance The Bucharest Unversty of Economc Studes, Romana e-mal: Florn Dumter Ph.D. Professor Faculty of Economcs, Engneerng and Informatcs, Vasle Goldș Western Unversty of Arad, Romana e-mal: Marus Boță Ph.D. Assocate Professor Faculty of Economcs, Engneerng and Informatcs, Vasle Goldș Western Unversty of Arad, Romana e-mal: (Receved Aprl 017; Accepted July 017) Abstract: The fscal polcy s an nstrument that collects resources for the state budget necessary to perform state functons; stablze the economy; regulaton of the economy and recovery growth. The most mportant nstruments are taxes and taxpayers. Snce 1989, wth the transton to a market economy, a specal tax polcy has been known by excse taxes. Analyzng the case of Romana and the evoluton of excse dutes (herenafter ED) conformty we created some regressons that llustrate the ED correlaton n Romana between 00 and 015. The methodology used: three unfactoral regresson models showng how ED mpacts economc ndcators such as GDP, power purchase expressed through the net annual average salary, and household fnal consumpton expendture. Moreover, each model has been tested and verfed usng statstc tests to gve relable results. In a frst stage, we analyzed the correlaton between GDP and consumpton as endogen varable and ED, then we created another model that we kept ED as an ndependent varable, but we changed the dependent varable usng the purchasng power as a dependent varable. Because accordng to the used tests we demonstrated that correlaton coeffcents are sgnfcant, we proceeded to explan them startng from fscal polcy and economc realty, own of these analyzed 14 years. In concluson we hghlghted below as 1

2 the ED depends on GDP and consumpton and the purchasng power can nfluence the ED. For future dscusson and studes we ntend to compare the results wth other countres n dfferent geographc areas n Europe Keywords: Tax conformty, regresson, evoluton, excse dutes, fscal polcy. JEL: G 40, H Introducton Excse dutes are the consumer charges set by the legslaton n force by whch any consumer (a natural or legal person) owes the state budget for consumpton of products subject to ths tax, whether t s domestc producton or mport. The regulatons n force very clearly defne excse dutes and classfy them, of course also detalng the stuatons n whch they are exempt from payment. Takng nto account the compettve market mechansms and the tendency of modern states to ntervene n the economy wth tax, we can easly notce what a sgnfcant role the taxes play, whether we refer to drect or ndrect ones, beng used as a lever of economc polcy through whch we can Accentuate or brake certan areas of actvty, and ncrease or curb consumpton n a partcular category of goods or servces. Due to taxes and taxes, fnancal ncomes are accumulated at the state's dsposal, whch s why the fscal polcy of any country s very mportant and clear and unnterruptble regulaton s needed. The state through ths fscal system accumulates resources to the state budget, whch wll then fnance ts expenses n order to carry out ts dutes. In practce, there are a number of taxes and fees that can be categorzed by content and form The excse duty regme appled to other products shall be determned by the competent authortes of the Communty countres, provded that ther taxaton does not requre a border control. Those authortes are enttled to charge taxes (other than value added tax) and servce charges. Specal consumpton taxes or excse dutes are applcable to products that are consumed n hgh quanttes and can not be replaced by other products of the same knd. Thus, consumpton s approxmately lnear wthout large oscllatons, so f the level of excses remans roughly the same and the fscal yeld remans constant. We can easly as a result of the tests that excse dutes have an mpact on gross domestc product and mplctly on fnal consumpton but not on a very large weght because they are by nature not applcable to a large category of products and mplctly to consumers such as VAT.. Research methodology Accordng to the abstract, ths artcle ams to hghlght the mpact of voluntary complance from the excse tax perspectve. In practce, we want to outlne how the amount of revenue from the collecton of these taxes affects, on the one hand, gross

3 domestc product (GDP) and, on the other hand, fnal consumpton. Another aspect that we propose through the artcle s to dentfy how purchasng power explans the varaton n excse dutes. In ths respect, we wll proceed to the constructon of 3 sngle-factor models, n of them the excses (meanng the revenues collected n the budget from ther collecton) consttuted the ndependent varable, and n the thrd varable dependent on purchasng power Ths thrd model s a reversed form precsely because, startng from the fact that n most cases we are dscussng an nelastc demand for excsable products n relaton to the prce, we wanted to see how purchasng power over a perod of 14 Do not nfluence the revenue from excse collecton, just as the dfferent purchasng power nfuses the consumpton of excsable products and mplctly the revenues collected from the budget from ths consumpton tax. In order to acheve these regressons, we used the Evews software platform, the 9th verson. Regardng the veracty of the data, we menton that these have been taken from the European Commsson webste as follows: Table no.1 ÎA* GDP * C* PP* *ÎA the excse tax *GDP Gross domestc power C consumpton and PP power purchase. All values n the table are amounts expressed n mllons of Euros. Because all three models that we wll buld next are sngle regresson lnear regresson models, we wll present some general aspects of lnear regresson. A frst aspect looks at ts form:, = 1,,..., n. A second aspect s related to the data by whch the statstcs generally operate, so only estmates of maxmum veracty are used n statstcs, whch can only be obtaned f certan condtons are 3

4 cumulatvely met. These condtons are n fact those assumptons that are assocated wth each regresson model n order to obtan certan propertes for model parameter estmators. Therefore, as we are n the context of a classcal lnear regresson model, we wll present the sx standard assumptons: The fgure s lnear and has the followng form: y = a + x +, =1,,,n. 1. Random errors are of average zero (because random error s regarded as the sum of random factors wth dfferent sgns, t s admtted that these unregstered factors wll not produce an average effect on the dependent varable): E( ) = 0, =1,,,n. Homoscedastctatea of random errors (accordng to whch the lnk between the dependent and ndependent varable s relatvely stable):var ( ) = X, n. =, = 1,. Random errors are not correlated (accordng to ths hypothess by randomly choosng two errors these are not correlated): cov( ) = 0, pentru j. Regsters and random errors are not correlated (random errors do not depend on explanatory varables): cov ( ) = 0, pentru ( ) ș j. Random errors are normally dstrbuted ( ) lnear functon of normally dstrbuted dstrbutons s normally dstrbuted; Central Lmt Theorem): N (0, ) (Damalan, 016). So we wll stll predct model parameters usng the smallest squares method (MCMMP). We wll test the models so obtaned usng the Student test (Penu, 016), we wll also see to what extent the resultng coeffcents are sgnfcant and what s the lkelhood of acceptng the alternatve hypothess. Fnally, we wll try to analyze the tme seres from three perspectves: the dstrbuton asymmetry over ts meda, the heght of seres dstrbuton and statonarty. 3. Results and dscusson 3.1. The mpact of excse revenue on gross domestc product Although ntutvely we can assume that the average of the dependent varable depends on a lnear relaton to the average of the ndependent varable (Anghelache, Paglacc and Prodan, 013), however, from dlgence reasonng, n order to dentfy the typology of the regresson functon, we wll graphcally represent the pars of ponts The amounts representng the amounts collected from the excse budget and the GDP. 4

5 ÎA Todor S.P., Ghur R., Brezeanu P., Dumter F., Boță M. (017) ,000 60,000 80, , ,000 Dagram no. 1 So, by analyzng the pont cloud or the rght scatterng dagram, we can deduce that the phenomenon can be descrbed by a lnear regresson lnear model (Anghelache, Anghelache and all 01). Usng the Evews platform we proceeded to estmate the parameters as can be seen from the table below: Table no. Varable Coeffcent Std. Error Statstc t Prob. IA C R-squared Mean dependent var Adjusted R-squared S.D. dependent var S.E. of regresson Akake nfo crteron Sum squared resd 1.08E+10 Schwarz crteron Log lkelhood Hannan-Qunn crtter F-statstc Durbn-Watson stat Prob(F-statstc) PIB Replacng the values thus obtaned n the form of the lnear functon, as t was presented above, we obtan: y = x + PIB = *ÎA

6 Thus, a frst observaton s that there s a drect lnk between the two varables, whch mples that the ncrease of the excse tax revenues automatcally nvolves GDP growth, whch s qute logcal, we can not say that ths lnk would be very strong Analyzng ts values R ș R. Accordng to these ndcators, the change n GDP s explaned n a small percentage by the change n excse recepts. Bascally, accordng to R, the lnk between the two varables s 40.85%, and s approprate R The GDP varaton s explaned n a proporton of 35.9% of the proceeds of ths tax. Moreover, assumng ad absurdum that excse goods dsappeared on our market and mplctly the tax collected from them, GDP would be * = The fact that GDP varaton s explaned n such a small percentage by the varaton n excse recepts can also be explaned by the role of product excse, one of whch s even dscouragng consumpton. But when you dscourage consumpton of excsable products (by analyzng the stuaton n the state's perspectve), t s napproprate to want an ncrease n excse tax revenues and a strong lnk between them and GDP, because t would go aganst logc, so the model we bult Reflects an exstng economc realty (so the objectve that the state promotes a polcy of hnderng consumpton reduces the achevement of another objectve of the statute - ensurng budget revenues) and a sustanable logc n ths respect. As shown above, we wanted to use the Student test for testng. Accordng to the table above t ÎA = ș c onstants = , s table = t 0,05;14; = 1,761, Where we deduce that trcycles s lower than both tc and tc, therefore the coeffcents calculated from the unfactoral model are sgnfcant. The above probabltes show to what extent the latter are sgnfcant. Gven that n both cases we dsagree wth values below 0.05, we nfer that we can accept the alternatve hypothess for both coeffcents wth a probablty of 95%, statng that the two coeffcents are sgnfcant. 3.. The mpact of revenue from excse dutes on household fnal consumpton expendture Next, we wll try to outlne how the tax revenues nfluence another macroeconomc ndcator, namely household fnal consumpton expendture (herenafter and fnal consumpton or CFP). Lke the prevous subchapter we wll start from the pont cloud chart analyss that represents these two seres to ensure that we have a sngle factor lnear model. Therefore, accordng to the scatter plot no. we can say that we are n the presence of a lnear regresson, aspect that allows us to pass on parameter estmaton wth the software and testng t, as can be seen from table no. 3. 6

7 ÎA Todor S.P., Ghur R., Brezeanu P., Dumter F., Boță M. (017) ,000 50,000 70,000 90,000 Dagram no. Table no. 3 Varable Coeffcent Std. Error Statstc T Prob. IA C R-squared Mean dependent var Adjusted R-squared S.D. dependent var S.E. of regresson Akake nfo crteron Sum squared resd 3.80E+09 Schwarz crteron Log lkelhood Hannan-Qunn crter F-statstc Durbn-Watson stat Prob(F-statstc) CFP We wll replace and n ths case the values thus obtaned n the form of the lnear regresson, as t was presented above, resultng: y = x + CFP = *ÎA Thus, a frst observaton and n ths case s that between the two varables s a drect lnk, whch mples that the ncrease of the excse tax revenues automatcally nvolves the ncrease of the consumpton, whch n the frst phase may seem napproprate n relaton to the role of the excse dutes, gven that these (excse dutes) are meant to dscourage consumpton, but an another 7

8 partcularty of excsable products s that they "enjoy" of the elastcty greater than 1 of ther demand n relaton to the populaton's ncome, whch may explan why the ncrease of excse duty revenue and mplctly excse duty has not resulted n a decrease n consumpton. However, lke the prevous model, we can not say that ths lnk would be a very strong one by analyzng the values of R and R. Accordng to these ndcators the varaton n consumpton s explaned n a small percentage by the varaton n revenues from excses. Practcal, accordng to R, the lnk between the two varables s about 38.86%, and accordng to R, the consumpton varaton s explaned n the proporton of 33.77% of the recepts of ths tax. Gven that excse goods represent only a lmted range of products by reference to the whole volume of products n the cvl crcut, t s obvous that, as shown by the followng equaton: * = , even though excsable products would no be marketed and the state would not charge ths tax, the value of the consumpton would be postve. And for ths model, we wll use the Student test to test the model. Accordng to the above table t ÎA = and c onstants = , t table = t 0,05;14; = 1,761, we deduce that the crtcal value s smaller than t ÎA and c onstants, therefore the coeffcents calculated from the unfactoral model are sgnfcant. The above probabltes show the percentage n whch the coeffcents are sgnfcant. Consderng that n both cases we have values below 0.05, we deduce that we can accept the alternatve hypothess for both coeffcents wth a probablty of 95%, statng that the two coeffcents are sgnfcant Impact of purchasng power on state budget revenues from excse dutes In the begnnng of ths thrd chapter, we have specfed that ths unfactoral model has another form, an nverted one, not by the regresson model, but by the fact that n ths scenaro we ntended to explan the excse recepts through the purchasng power (herenafter and PC) and not explan a macroeconomc ndcator lke PC by excses, practcally n ths thrd unfactoral model the dependent varable model wll be represented by the excse recepts. Before we proceed to model estmaton, we note that purchasng power has not been found by us on natonal or nternatonal portals whch collect and process data n ts own form. The purchasng power s the generc name that we (the authors of ths artcle) attrbuted to a seres of data that defnes the annual average of net earnngs over the last 14 years. So, ths model wll show how the net average earnngs nfluence the consumpton of excsable products and mplctly excse tax revenues to the state budget. 8

9 PC Todor S.P., Ghur R., Brezeanu P., Dumter F., Boță M. (017) Dagram no. 3 So, and n ths thrd case, we can say that we are n the presence of a lnear regresson, whch allows us to pass on parameter estmaton wth software and testng t, as can be seen from the followng table: Table no. 4 Varable Coeffcent Std. Error Statstc t Prob. PC C R-squared Mean dependent var Adjusted R-squared S.D. dependent var S.E. of regresson Akake nfo crteron Sum squared resd Schwarz crteron Log lkelhood Hannan-Qunn crter F-statstc Durbn-Watson stat Prob(F-statstc) ÎA Substtutng and n ths case the resulted values n the form of the lnear regresson, as presented above, we obtan: y = x + ÎA = 1.307*PC Thus, a frst observaton and n ths case s between these two varables 9

10 s a drect lnk, whch mples that the ncrease n average earnngs automatcally mples an ncrease n the consumpton of excsable products and mplctly an ncrease of revenues to the state budget from excses, an ntutve aspect f we remember that for these products, demand s nelastc, applyng a fortor argument, all the more so revenue ncreases, the populaton wll be tempted to acqure such products (the more obvous aspect n the case of luxury products). However, lke prevous models, we can not say that ths lnk would be a very strong one by analyzng the values of R and R. Accordng to these ndcators, the varaton n excse recepts s explaned n a small percentage by the average earnngs varaton, whch s explaned by the nelastcty of demand. So, accordng to R, the lnk between the two varables s approxmately 43.80%, and accordng to R the varance of ÎA (recepts form excse dutes) s explaned n the rato of 39.1% of purchasng power. And for ths model we wll use the Student test to test the model. Accordng to the table above t PC = and c onstants = , t table = t 0.05; 14; = 1.761, so we deduce that crtcal value s only smaller than t PC, ths s the only sgnfcant coeffcent, aspect also ndcated by the probablty value much below 0.05, whch shows that we can accept the alternatve hypothess only for the purchasng power coeffcent wth a probablty of 95%. As we menton above, a part of the analyss s about the way n whch the analyzed seres are presented n tme. The frst two perspectves relate to the seres dstrbuton asymmetry over ts own average and the heght of the dstrbuton. In ths sense we wll buld the hstograms for all four tme seres, and accordng to the Skewness and Kurtoss tests, we wll make a comparatve analyss of these. The followng 4 hstograms are bult usng the same software platform: Seres: PIB Sample Observatons 14 Graphc no. 1 The hstogram of GDP (herenafter and PIB) Mean Medan Maxmum Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probablty Seres: CFP Sample Observatons 14 Mean Web: publcat.uvvg.ro/ndex.php/studaeconoma.pages Medan Maxmum Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probablty

11 Graphc no. The hstogram of household fnal consumpton Seres: PC Sample Observatons 14 Mean Medan Maxmum Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probablty Graphc no. 3 The hstogram of purchasng power Seres: IA Sample Observatons 14 Mean Medan Maxmum Mnmum Std. Dev Skewness Kurtoss Jarque-Bera Probablty Graphc no. 4 The hstogram of recepts from excses Accordng to the Skewness test, a test that ndcates the devaton of emprcal dstrbuton by reference to the symmetrc dstrbuton around the mean, n the case of GDP, consumpton and purchasng power, we are n a stuaton where, gven that the values are less than zero, the dstrbuton s tlted to the rght, wth more extreme values to the left. As respects excse dutes, the value of ndcates that the dstrbuton s tlted to the left, wth many extreme values to the rght. Accordng to the Kurtoss test, a test ndcatng the degree of flattenng or sharpenng of a dstrbuton, gven that for all four seres the test ndcates values less than 3, we are n the stuaton of platykurtc dstrbuton, beng more flat than 11

12 the normal dstrbuton wth dspersed values on a larger nterval around the average. Therefore, for all four tme seres, the probablty of occurrence of extreme values s lower than f they had a normal dstrbuton. The last proposed perspectve of analyss for these artcle remans testng the statonary of the seres. We wll test the statonary wth the Dckey Fuller Test, as shown n the followng 4 tables. Table no. 5. Statstc t Prob.* Augmented Dckey-Fuller test statstc Test crtcal values: 1% level % level % level *MacKnnon (1996) one-sded p-values. Table no. 6. Statstc t Prob.* Augmented Dckey-Fuller test statstc Test crtcal values: 1% level % level % level *MacKnnon (1996) one-sded p-values. Table no. 7. Statstc t Prob.* Augmented Dckey-Fuller test statstc Test crtcal values: 1% level % level % level *MacKnnon (1996) one-sded p-values. Table no. 8 Statstc t Prob.* Augmented Dckey-Fuller test statstc Test crtcal values: 1% level % level % level *MacKnnon (1996) one-sded p-values. Because the values of statstcal t are much lower than t tabel (crtcal value - 1,761), we are n the stuaton of rejectng the null hypothess for all four tme seres, 1

13 resultng that each of them are statonary seres over tme. Therefore, the statonary of the tme seres shows that f there was a shock on these seres, t would absorb over tme and would not be one wth permanent effects. 4. Conclusons and dscusson By ths artcle has acheved the man objectve of hghlghtng and mpactng voluntary complance from excse dutes perspectve. In the future we are gong to compare the result wth other countres wth a developed economy and wth countres wth emergent economy. Thus, we have succeeded to demonstrate that revenues collected from excse dutes to the state budget are nfluenced by fnal consumpton and purchasng power. Of course, excse taxes have a small contrbuton to a naton's budget revenue and the overall mpact s not very hgh, but we have managed to establsh a connecton between the varables. The ncrease n average earnngs automatcally mples an ncrease n the consumpton of excsable products and budget revenues, ntutve aspect f we start at the premse that the demand for these products s nelastc, especally f revenue ncreases, the populaton t wll be temptng to buy such products (the more obvous n the case of luxury products). However, as a general concluson we can not say that ths lnk would be a very strong one. 5. Selectve bblography 1. Anghelache, C., Paglacc, M, Prodan, L., Model de analză macroeconomcă bazat pe funcţa de regrese, Romanan Journal of Statstcs no. 1/013.. Anghelache, G.-C., Anghelache C., Prodan, L., Dumtrescu, D., Soare, D., V., Elemente teoretce prvnd utlzarea modelulu econometrc de regrese multfactorală, Romanan Statstcal Revew Trm III / 01- Supplement. 3. Dervs, K., De Melo, J. and Robnson, S. (198). General Equlbrum Models for Development Polcy New York, Cambrdge Unversty Press Devarajan, S., Lews, J.D. and Robnson, S. (1994). Gettng the Model Rght: The General Equlbrum Approach to Adjustment Polcy Mmeo 4. OECD Åsa Johansson, Chrstopher Heady, Jens Arnold, Bert Brys and Laura Varta 008,,Tax and economc growth Economcs Department, Workng Papers nr Parry, C.D.H. (1997). Alcohol msuse and publc health: A 10-pont acton plan. Polcy Bref No1, February South Afrcan Medcal Research Councl. Avalable [onlne] 6. Penu, D. Indrect Taxes n Romana an Econometrc Analyss, Academc Journal of Economc Studes, Vol., No.1, March 016, pp , ISSN , ISSN On-lne Sngle, E., Robson, L, Xe, X. and Rehm, J. (1998). The Economc Costs of Alcohol, Tobacco and Illct Drugs n Canada, 199. Addton 93:

14 8. Sorensen, P.B (1998), Recent Innovatons n Nordc Tax Polcy: from the Global Income Tax to the Dual Income Tax, n: Tax Polcy n the Nordc countres, Macmllan Press. 9. Van As, S The taxng ssue of alcohol abuse. Establshng an alcohol njury fund Scence n Afrca, January 004. Avalable [onlne]: 14