Indirect taxes and price formation - a model for the Polish economy

Transcription

1 Jakub Boratyńsk Insttute of Econometrcs and Statstcs Unversty of Lodz ul. Rewoluc 1905 r. 41 Łódź, Poland e-mal: Indrect taxes and prce formaton - a model for the Polsh economy 1. Introducton Ths study ams at extendng the elementary nput-output prce model to represent the role of ndrect taxes (formulaton of the nput-output prce model can be found e.g. n R. E. Mller & P. D. Blar [1985]). The proposed extenson s manly based on the solutons presented by R. Bardazz & M. Grassn [1991] and M. Grassn [1997]. The fnal formulaton meets the problem of unavalablty of detaled data on tax components n ndvdual cells of nput-output tables, whch s lkely not only the Polsh-specfc dffculty. Buldng such a model was dctated manly by some casual needs,.e. examnng effects of possble mport tax ntroducton and effects of adaptng VAT rates to the EU standards. However, t can also functon as a block of the IMPEC model of the Polsh economy. 2. Indrect taxes and prce formaton - extenson of the elementary nput-output prce model An mportant feature of the presented extenson s that t s based on the structure of the actually avalable nput-output data, whch - n the case of Poland - conform wth prncples of the System of Natonal Accounts (SNA). Thus, for transparency of further consderatons t s necessary at ths place to focus on selected data ssues and prce concepts.

2 In the SNA, the two man categores of prces are used,.e. basc prces and fnal prces (see The System of Natonal Accounts vol. II [1997]). Basc prces are defned as amounts of money receved by producers for ther products, not countng any taxes and subsdes, as well as trade and transport margns. Fnal prces are amounts as seen from consumers' pont of vew,.e. they are nclusve of all taxes and margns. The relaton between basc and fnal prces can be presented as follows (see Zenkowsk [2002]): basc prce + customs dutes and other mport charges + excse and taxes on selected servces (gamblng, lottery etc.) subsdes + VAT + trade margn + transport margn = fnal prce (purchaser s prce, market prce). Accordng to prncples of the SNA, ntermedate and fnal consumpton are evaluated n fnal prces, whereas global output, mports and exports - n basc prces. Referrng to the quoted defntons, global output of a gven branch n basc prces s obtaned by addng up materal costs - evaluated n fnal prces - and value added of that branch. As a consequence, basc prce of a gven product s a functon of, among other thngs, all ndrect taxes pad on products and servces used n the producton process. Among ndrect taxes, the role of VAT n prce formaton requres a closer consderaton. The

3 producer s tax lablty s gven by the dfference between the tax charged on hs sales and that pad on hs purchases of ntermedate goods and servces (see R. Bardazz and M. Grassn [1991]). In such case, VAT on ntermedate products and servces makes n fact no cost to the producer, whle t can be fully deducted from hs tax lablty. In terms of prces, VAT does not affect basc prce but rather the fnal prce and, therefore, can be treated as a tax on fnal consumpton. However, often there are departures from such a pure VAT system,.e. there are rules lmtng deducton of tax pad on producer s nputs. R. Bardazz and M. Grassn [1991] menton three typcal reasons for whch deducton of VAT s restraned n most of the EU countres. Frstly, sectors sellng products or servces whch are exempted from VAT (n Poland: educaton, health care, publc admnstraton servces, fnancal servces etc.) have no rght to deduct the tax. Also, small frms, whch do not exceed a certan lmt of turnover, can beneft from VAT exempton, regardless of product or servce they offer. Secondly, the general rule s that VAT can be deducted only for those products and servces whch are strctly connected wth the output of a gven sector. Snce such classfcaton of ntermedate goods s often dffcult, especally for unncorporated famly frms, specal rules are usually appled, for example the rules lmtng deducton of VAT pad on partcular goods (e.g. fuels). Fnally, for small busness there usually exst smplfed and standardsed methods of tax settlement. In such cases the deductble VAT s establshed basng on some fxed economc parameters rather than the actual value of ntermedate goods and servces used n producton (e.g. for farmers). In all of those cases, nondeductble VAT takes part n the formaton of basc prces and, as well as other ndrect taxes, should be present n the extended prce equaton. Apart from tax elements, the extended model must also take nto account the exstence of mports n ntermedate use of products and servces. Obvously, changes n costs of domestc producton should not affect mport prces n the model (unless exchange rates are consdered), otherwse prce effects of the cost-push nflatonary spral could be overestmated n smulatons. Therefore, n the proposed model, use of mported products and servces s dstngushed from that of domestc ones. The above remarks lead to formulatng the nput-output prce model extended wth ndrect taxes and margns. Assume that an economy s output can be dvded nto n groups of homogeneous products and

4 servces. Denote by x (, = 1,..., n) the value of outlays on ntermedate products and servces of type used n producton of goods of type ( x are, thus, elements of the frst part of a product-to-product nput output table). Denote further by X ( =...,n) 1, the global producton of goods or servces of type. Global output of a gven type s equal to materal costs augmented by costs of prmary producton factors (value added),.e.: X = n = 1 x + V (1) where V stands for value added assocated wth goods and servces of type. Accordng to the SNA rules of evaluatng transactons, as well as some smplfyng assumptons concernng taxes and margns, global output and ntermedate use are gven as follows: X = Q ~ p (2) and: ( q p ( 1+ s + d ) + q ~ p ( 1+ s + d ))( + h t ) x = ~ 1 (3) where Q stands for global output n quantty terms,, q and q represent quanttes of goods of type used as ntermedate nputs n producton of branch, domestc and mported, respectvely, p~ - basc prces of domestc goods, p~ - basc prces of mported goods. Symbol s stands for average rate of all ndrect taxes except VAT (net - tax mnus subsdy rates) pad on domestc goods of type (.e. excse tax and specal taxes on selected servces), s - average rate of all ndrect taxes except VAT (net) pad on mported goods of type (.e. dutes and other mport charges as well as excse tax, mnus subsdes), d = b + c, where b s an average rate of trade margn, c - average rate of transport margn. Symbol t stands for average nomnal VAT rate for products or servces of type (the same for domestc and mported goods), h - a coeffcent showng what part of an ndvdual ntermedate purchase (n terms of value ncludng all ndrect taxes except VAT) s subect to nondeductble VAT. In other words, h t show rates whch can be named as effectve VAT rates for

5 ndvdual nputs of ntermedate products and servces. As can be seen from equaton (3), all ndrect taxes are assumed to be ad valorem taxes wth fxed rates. Smlarly, trade and transport margns are consdered fxed n proporton to value of products and servces (n basc prces) belongng to a gven group. Such treatment of taxes s perhaps not fully adequate, as t does not represent the actually exstent non-lneartes n calculaton of taxes. Ths restrcton can be, though, ustfed by the fact that most of the consdered ndrect taxes are ad valorem taxes. Rates are named average, snce ndvdual groups of products or servces n many cases are not homogeneous and, thus, wthn a gven group there may coexst products or servces taxed at dfferent nomnal rates. Actually, the smplfyng assumptons - explct n the formulaton of the model - are strctly connected wth the structure of the usually avalable nput-output data. Regardng (2) and (3), cost equaton (1) can be wrtten as: n = 1 ( q ~ p ( 1+ s + d ) + q ~ p ( 1+ s + d )( 1+ ht ) Q ~ p = ) + V (4) Dvdng by Q yelds: ~ n ( ) ( ) q q V p = ~ p 1+ s + d + ~ p 1+ s + d ( 1+ ht ) + = 1 Q Q Q (5) Denotng: q a ~ = (6) Q q a ~ = (7) Q

6 and: V v ~ = (8) Q equaton (5) can be then wrtten as: n = 1 ( a~ ~ p ( 1+ s + d )( 1+ ht ) + a~ ~ p ( 1+ s + d )( 1+ ht )) ~ p = + v~ (9) where a ~ and a ~ are techncal coeffcents showng - n physcal terms - the amount of products or servces of type - domestc and mported, respectvely - necessary to supply a unt of products or servces of type. Fnally, f stands for element-by -element multplcaton, equaton (9) can be wrtten n the matrx form: ~ ( A ( J H T ))( I S D) p ~ ~ ( A ( J H T ))( I S D) p ~ d + + d + d + m + + m + + v p ~ ~ (10) d = m whch after solvng yelds: ~ ~ ( I ( A d ( J+ H T) )( I+ Sd + D) ) ( A m ( J+ H T) )( I+ Sm + ) ( D p ~ + v) 1 ~ d = m p ~ (11) where p ~ d = [ ~ p ], ~ m = [ ~ ~ p p ], d = [ a ~ ~ A ], m = [ a ~ ] s, S m - dagonal matrx of elements of elements s A, = [ h ] H, S d s a dagonal matrx of elements, T - dagonal matrx of elements t, D - dagonal matrx d, I s a untary matrx, J - matrx wth all elements equal 1 (, =...,n) 1,. It s obvous, however, that nput-output tables n physcal terms are practcally unavalable and, thus, techncal coeffcents can not be obtaned. Instead, value-based coeffcents can be used n prce equaton, whch s n fact equvalent to assumng that ntally all prces equal 1. Under such assumpton, solvng prce formula leads to obtanng prce ndces p rather than levels p~ (see R. E. Mller and P. D. Blar [1985]). Defne :

7 z a = (12) X z a = (13) X and: V v = (14) X where: z = q ~ p (15) and: z = q ~ p (16) where z and z represent ntermedate consumpton evaluated n basc prces (whch can be consdered observable at a certan stage of emprcal analyss - for detals see the next secton), a and a represent value-based nput-output coeffcents, v - value added per unt of output evaluated n basc prces. The fnal, applcable verson of the extended prce model can be, thus, wrtten as: 1 ( I ( A ( J + H T) )( I+ S + D) ) (( A ( J + H T) )( I+ S + D) p v) p + d = d d m d m (17) where p d = [ p ], p m = [ p ], A d = [ a ], m = [ a ] A, v = [ v ] (, =...,n) contan prce ndces for domestc and mported goods, respectvely. 1,. Vectors p d and p m Model represented by equaton (17) provdes a wde range of possble applcatons. Apart from analyses of prce reacton to changes n untary value added, t enables examnng the effects of changes n tax rates ( S d, S and T ), as well as effects of fluctuatons of mport prces ( p m m ) and

8 changes of trade and transport margn rates ( B, C ). However, a problem arses of how to obtan the requred parameters of model (17) usng the avalable nput-output data. The next secton s dedcated to a soluton of ths problem, whch nvolves procedure of nput-output table decomposton. 3. Decomposton of the nput-output table The maor problem n practcal applcaton of the model gven by equaton (17) s that the necessary parameter matrces ( A A, S, S, D, T, H ) are not mmedately derved from nput-output tables. d, m d m Generally, calculaton of these parameters requres nformaton on tax components n ndvdual transactons of fnal and ntermedate use. Such detaled data are usually unavalable. Therefore, n the Polsh case, as perhaps for most European countres, applcaton of the proposed prce model reles on decomposton of nput-output table, based on ncomplete data and smplfyng assumptons. Perhaps, the ssue of the greatest concern s to determne h coeffcents, showng contents of nondeductble VAT n ndvdual ntermedate consumpton transactons. Soluton of ths problems s farly ambguous and, consequently, any specfc assumptons are not emboded n the formulaton of model (17). Thus, the model n that form s ndependent of the actually chosen decomposton method. It must be emphassed that the method of decomposton ntroduced below s not the only possble one. It should rather be consdered as presentaton of a certan general approach, whch can be a subect of further dscusson and development. Decomposton procedure strctly reles on the structure of the nput-output table, whch - for the Polsh case - s presented n table 1 (all calculatons presented n ths paper are based upon the product-toproduct nput-output table for the year 2000 was elaborated by the author basng on supply and use tables, provded by the Central Statstcal Offce, unpublshed). For convenence of the analyss, whch s anyway focused on the tax effect on prces, the value added s aggregated nto one row,.e. ts components are not dstngushed n the presented table.

9 Table 1. Structure of the nput-output table 2000 for Poland. [1] [2] [3] Row totals [1] [ x ] n n [ y ] [ ] k n l E [ G ] n 1 n 1 [2] [ V ] m n Subtotals ([1], [2]) [ X ] 1 n [3] [ M ] 1 n [4] [ ] n S 1 [5] [ ] n S 1 [6] [ T ] 1 n [7] [ B ] 1 n [8] [ C ] 1 n Column totals [ G ] 1 n where y k represent fnal demand, l beng the number of the dstngushed fnal demand categores (not countng exports, whch s treated separately), G - global supply of products or servces of type, - exports of products or servces of type, M - mports of products or servces of type, amount of ndrect taxes, other than VAT, pad on domestc products or servces of type, amount of ndrect taxes, other than VAT, pad on mported products or servces of type, E S - total S - total T - total amount of non-deductble VAT pad on both domestc and mported products and servces of type, B - total amount of trade margn pad on products and servces of type, transport margn pad on products and servces of type. C - total amount of In the nput-output table'2000, transport, storage and communcaton servces account were dvded nto two parts one concernng transport servces connected wth trade margns (name t transport-asmargn ), the other ncludng transport servces treated as ntermedate use of other sectors, as well as storage and communcaton servces. As a result, global output of the transport-as-margn servces equals total of the transport margns pad on all products n the economy. Smlarly, trade and repar were dvded nto trade and repar, however n the case of trade, ts output equals the total of trade margns across all transactons by defnton. Such separaton of trade and transport s necessary for the

10 model as well as the decomposton procedure to work properly. It s worth notcng that snce margns are recorded n expendtures on other products and servces, total supply of trade and transport-asmargn, equals zero, as do all values n the correspondng rows of the nput-output table. Extractng VAT from the nput-output table At the frst stage, VAT s extracted nput-output table. At the same tme, VAT rates t and coeffcents h - showng contents of non-deductble VAT n ntermedate flows - are determned. The procedure s based on the followng relaton (see also J. C. Collado & F. Sancho [2002] for method of recoverng hdden tax rates n nput-output tables n whch all transactons contan full amount of deductble): T = t h t +t n l x + = 1 1+th k= 1 1 y k (18) In (18) we have n equatons wth n ( n +1 ) varables ( t and h ). Thus, to obtan those parameters, a pror knowledge and/or smplfyng assumptons must be appled. For example, consder the sectors whose products and servces are fully exempted from VAT. In the case of Poland these are: fshery, fnancal servces, publc admnstraton and defence, educaton, health care and socal securty servces. As they do not have rght to VAT deducton, t s defnte that all ntermedate nputs n these sectors nclude full amounts of VAT. Thus, coeffcents n the correspondng columns of the H matrx are all equal 1. Equaton (18) assumes that all fnal purchases are nclusve of full VAT amount. However, ths need not be true for nvestment. Smplfyng, t can be assumed that enterprses fully exempted from VAT pay the tax on nvestment goods, other do not (t s deducted). Among nsttutonal sectors of the Polsh economy, the non-proft nsttutons, government and fnancal enterprses can be treated as the performers of most actvtes exempted from VAT. Consequently, t was assumed that nvestment outlays of the mentoned nsttutonal sectors nclude full VAT amounts, whle non-fnancal enterprses and households (household frms) deduct the whole tax. The only excepton s that for prvate expendtures on buldngs, whch are also recorded as household nvestment, but VAT cannot

11 be deducted. Thus, for clarty of the soluton presented below, nvestment n buldngs were moved to household consumpton. Followng the above remarks, equaton (18) should be substtuted wth: T = t h 1 x + Ω + th 1 Ω2 t 1+t x + k Γ t 1+ t y k (19) ( 1 ) ( 1 ) where Ω = {,,, ω } ( 1 ) 1... s a class of subsequent numbers of groups of commodtes whch are not 1 2 ( 2 ) ( 2 ) exempted from VAT, ω beng the total number of such products and servces, Ω = {,,, n } ( ω ) - a class of subsequent numbers of groups of commodtes whch are fully exempted from VAT, { k,k2,,kγ Γ = 1... } - a class of subsequent numbers of fnal expendture categores fully nclusve of VAT, γ beng the total number of such categores. Stll, however t s mpossble to solve (19) for t and h wthout further smplfcaton. 1 2 In order to llustrate the approach better, assume for a whle that VAT s present only n fnal demand (except nvestment of households and non-fnancal enterprses) and n ntermedate expendtures of branches fully exempted from VAT. In terms of h coeffcents t means that h = 0 for Ω1. Thus, we get: T = Ω2 t 1+t x + k Γ t 1+t y k (20) Solvng (20) for t yelds: 1 t = T x + yk T (21) Ω2 k Γ Result of ths experment for the nput-output table 2000 are presented n table 2.

12 Table 2. VAT rates (n %) and h coeffcents. Products and servces Results of (21) Reference rates Destnaton rates 1 Agrculture and forestry Fshery Mnng Food Tobacco Fabrcs Textle Leather Wood Paper Publshng and prntng Petrol Chemcals Rubber and plastc Other non-metallc Metal Metal products Machnes Offce machnes and computers Electrc machnes Rado and TV devces Medcal and optcal devces Motor vehcles Other transport equpment Furnture and other goods Recyclng Electrcty, gas, water Constructon Hotels and restaurants Fnancal servces Busness and real estate servces Publc admnstraton and defence Educaton Health care and socal securty Other servces Repar Transport, storage, communcaton Trade Transport-as-margn The frst column contans rates calculated accordng to equaton (21). Many of those rates prove to be hgher than the hghest nomnal VAT rate actually used n Poland (22%). The only explanaton of such results (apart from naccuracy of data) s that the denomnator n formula (21) s to small for certan types of goods and, consequently, t should be augmented by a part of those ntermedate costs, whch were assumed VAT-free. Thus, t gves an ndrect evdence for what s not explct n the nput-output table, that s for the exstence of non-deductble not only n costs of the sectors fully exempted. h The above observaton s the bass for the further procedure. The dea amounts to searchng for such

13 h values (for Ω1 ) whch would brng down the resultng t rates at least to the level of 22%, (the hghest nomnal rate) thus allowng to fnd VAT amounts whch are defntely present n ntermedate flows. In fact, nstead of usng the termnal level of 22%, one can calculate reference rates, basng on vald tax regulatons, whch assgn nomnal VAT rates to dfferent knds of products and servces. Ths sort of calculaton s farly approxmate. If wthn a group of products there are goods charged at dfferent VAT rates, they should be weghted by shares n global supply of those goods. However, reference rates should perhaps be treated as upper bounds for actual effectve rates rather than as actual effectve rates themselves. Ths restrcton results from the fact that a gven good can be usually suppled by ether a VAT- payng producer or by a VAT-exempted one. In such a case, the effectve VAT rate becomes lower than the nomnal (reference) one. Accordng to the proposed approach, the so called destnaton rates are determned. Destnaton rates are set to the reference rates for those products and servces, for whch rates resultng from equaton (21) prove hgher than the reference ones (bold font). For the remanng commodtes, destnaton rates are set at the level determned by applyng formula (21). Destnaton rates actually compose the T matrx (see model (17)). Gven VAT destnaton rates, t s possble to calculate h coeffcents, provded that the number of unknown varables s reduced to n. Ths can be obtaned by assumng that those coeffcents are unform across rows of the nput-output table. It means that n all branches whch are not fully exempted from VAT, the same fracton of value of a partcular materal s charged wth the nondeductble tax. Regardng ths assumpton, equaton (19) can be rewrtten as: T = t h x + 1 Ω +th 1 Ω2 t 1+t x + k Γ t 1+t y k (22)

14 That smplfcaton enables solvng (22) for each h separately. The soluton yelds: ( 2 ) 2 h = T +t T x yk t +t x T +t yk + x (23) 2 1 Ω k Γ Ω k Γ Ω2 1 where t are VAT destnaton rates, as shown n table 2. Coeffcents h are set to zero whenever dvson by zero arses n (23). Fnally elements of T and H can be set as: t t = 0 for for = (24) h for Ω1 h = (25) 1 for Ω2 It can be seen from table 2 that the hghest contents of non deductble VAT are found n ntermedate purchases of food, leather, products of publshng, fuels and fabrcs. The last step n whch VAT s nvolved s purfyng nput-output table of ths tax, n order to enable calculaton of the remanng parameters. Denotng by w ntermedate flows purfed of VAT and by f k - fnal demand purfed of VAT, one can formulate the followng purfcaton rules: w 1 = 1+ h t x (26) h 1 = 1+t yk, y k, for for k Γ k Γ (27) It must be emphassed that the proposed method of determnng T and H parameter matrces s based on rather strong assumptons, whch can lead to some undesrable effects n the model. Frstly, unaquantance wth the actual effectve rates and usng reference rates nstead can lead to

15 underestmatng contents of non-deductble VAT n ntermedate flows. Thus, prce responses to VAT rate changes n the model can be nterpreted as the mnmum of whch one can be certan,.e. actual responses are lkely to be greater. The only reason for such lmtaton n the functonalty of the model s the lack of data on non-deductble VAT. Secondly, a researcher must stll deal wth the noton of effectve rather than pure nomnal rates whch often causes dffcultes n adequate formulaton of smulaton scenaros. Thrdly, the rules n the model are rather compatble wth rules lmtng deducton of VAT on partcular goods, whle perhaps n the Polsh condtons a greater role should be attrbuted to VAT exempton for small enterprses. Modellng of the latter case, however, complcates the procedure of determnng non-deductble VAT n nput-output table. As a result, the proposed method leaves a substantal margn for further consderaton and, possbly, mprovement. Purfyng the nput-output table of ndrect taxes other than VAT Next stages of the decomposton procedure are perhaps more transparent than the prevous one. To determne dagonal elements of S d, S m, B and C matrces, the followng formulas are used: S s = (28) X S s = (29) M B b = (30) X + M C c = (31) X + M The resultng rates are presented n table 3. Negatve values of tax rates mean that subsdes exceed taxes.

16 Table 3. Rates of ndrect taxes excludng VAT (n %), margn rates. Products and servces s 100 s 100 b 100 c Agrculture and forestry Fshery Mnng Food Tobacco Fabrcs Textle Leather Wood Paper Publshng and prntng Petrol Chemcals Rubber and plastc Other non-metallc Metal Metal products Machnes Offce machnes and computers Electrc machnes Rado and TV devces Medcal and optcal devces Motor vehcles Other transport equpment Furnture and other goods Recyclng Electrcty, gas, water Constructon Hotels and restaurants Fnancal servces Busness and real estate servces Publc admnstraton and defence Educaton Health care and socal securty Other servces Repar Transport, storage, communcaton Trade Transport-as-margn Dstngushng between domestc and mported producton reles on the assumpton, accordng to whch, for partcular commodty there s a fxed share of mports, dentcal n all ndvdual purchases of that commodty,.e.: z = μ z (32) where z represent ntermedate flows evaluated at basc prces (see (15) and (16)): z = z + z (33)

17 Coeffcents μ can be calculated accordng to the followng rule: M μ = (34) X + M The above assumpton makes t possble to purfy transactons of both ndrect taxes on domestc goods and those on mported ones, as well as trade and transport margn, at once. The approprate rule s gven by: z w = 1+b + c + μ s + ( 1 μ ) s (35) Havng z and μ, t s possble to determne elements of the last two parameter matrces, A m : a z = ( 1 μ ) (36) X A d and z a = μ (37) X Fnally, all of the acqured parameter matrces should be dmnshed by removng rows and columns correspondng wth trade and transport-as-margn. Otherwse the model could not be solved as all elements of A d and A m matrces n rows correspondng wth trade and transport-as-margn equal zero (margns are present explctly n the prce equaton, so the soluton remans correct). The test for correctness of the decomposton procedure s done by calculatng prces accordng to formula (17), usng the acqured parameter matrces all of the resultng prce ndces should equal one.

18 4. Fnal prces and aggregate prce ndces At the last stage of the analyss, fnal prces, as well as aggregate prce ndces are determned. Fnal prce for a partcular group of products or servces can be calculated as: ( +t )( + s + d ) p r = 1 1 (38) Values of r show the relaton of the fnal to the basc prce. Changes of fnal prces of domestc goods are represented by the followng ndces: π = r / `r (39) where r stands for ntal values of ( ) ` d r, that s the values calculated basng on orgnal parameters of the model (not changed due to scenaro assumptons). Analogously, fnal prces as well fnal prce ndces for partcular mported commodtes can be calculated,.e.: ( +t )( + s + d ) p r = 1 1 (40) and: π r = `r (41) For nterpretaton purposes, t s usually convenent to use a weghted prce ndex for subsequent groups of products and servces, ncludng both domestc and mported ones: π = ( 1 μ ) r + μr ( 1 μ ) r + μ r ` ` (42) Aggregate prce ndex for category k of fnal expendtures (not exempted from VAT - e.g. fnal

19 consumpton) can be calculated as follows: 1 n n λ k = π yk yk (43) = 1 = 1 where k Γ. Values of y k are taken drectly form the nput-output table (see table 1). Aggregate prce ndex of global domestc output s gven by: 1 n n (X) λ = p X X (44) = 1 = 1 Defne also: δ = ( 1+ s + d ) 1+ `s p + `d (45) δ = ( 1+ s + d ) 1+ `s p + `d (46) and δ = ( 1 μ )( 1+ s + d ) p + μ ( 1+ s + d ) p ( 1 μ )( 1+ `s + `d ) + μ ( 1+ `s + `d ) (47) where ` s, ` s and d` are ntal values of ndrect tax rates (except VAT) and margns - as determned n the decomposton procedure. Values of δ can be treated as fnal prce ndces for commodtes for whch full deducton of VAT s allowed. Thus, t leads to formulatng aggregate prce ndces for those fnal demand categores whch are free of VAT burden (e.g. exports and nvestment of non-fnancal enterprses), as well as to the mxed aggregated prce ndex for a wder range of fnal goods, respectvely (e.g. GDP):

20 1 n n (F) λ k = δ yk yk (48) = 1 = 1 where k { 1,2,...,l} and: n n n n (A) λ k = π yk + δyk yk + yk (49) k U 1 = 1 k U 2 = 1 k U 1 = 1 k U 2 = 1 1 where U Γ, U { 1,2,...,l}. 1 2 Accordng to analytcal needs, other prce ndces can be defned. 5. Smulaton results Smulaton assumptons were nspred manly by the two facts. Frstly, t was the dscusson among the Polsh poltcs and economsts on the possblty of ntroducng a 5% mport tax. Secondly, t s the EU accesson, whch entals adaptaton of VAT rates on certan products and servces n the next years. Therefore, the emprcal part of the study s dvded nto two parts, adequate for the analysed ssues. Smulaton 1 - mport tax Applyng scenaro for the frst smulaton s farly smple. It amounts to addng 0.05 to each rate of ndrect taxes on mported goods s. The results are presented n tables 4 and 5. The relatvely hghest growths of fnal prces can be observed for offce machnes, motor vehcles and rado and TV devces, manly as a result of hgh mport shares for these products. However, fnal prces are also drven by sgnfcant ncrease n materal costs, whch s vsble n basc prces (for ndustry rates of growth of basc prces oscllate around 1% - 2%). It s the offce-machne ndustry who bears the hghest ndrect costs of ntroducng mport tax. Among commodtes showng the weakest reacton, as far as prces are concerned, there are essentally those whch absorb lttle materal costs. One should menton here manly educaton, publc admnstraton and defence and health care servces.

21 Table 4. Results of smulaton 1. Rates of growth of prces, share of mports n supply (n %). Products and servces Basc prces (domestc) Fnal prces (domestc + mported) Share of mports n total supply 1 Agrculture and forestry Fshery Mnng Food Tobacco Fabrcs Textle Leather Wood Paper Publshng and prntng Petrol Chemcals Rubber and plastc Other non-metallc Metal Metal products Machnes Offce machnes and computers Electrc machnes Rado and TV devces Medcal and optcal devces Motor vehcles Other transport equpment Furnture and other goods Recyclng Electrcty, gas, water Constructon Hotels and restaurants Fnancal servces Busness and real estate servces Publc admnstraton and defence Educaton Health care and socal securty Other servces Repar Transport, storage, communcaton Table 5. Results of smulaton 1. Rates of growth of aggregate prces (n %). Categores of fnal expendtures + Prce change global output Prvate consumpton 1.3 Government consumpton 0.3 Investment 2.0 Export 1.9 Global domestc output (basc prces) 0.8

22 Table 5 shows changes of prces for maor categores of fnal demand, as well as global output. Dfferences among the rates of growth can be explaned, frstly, by dfferences n expendture structures across categores. The second reason s that to evaluate global output, basc prces are used, whle for the demand fnal prces. The hghest prce ncrease s that of nvestment, the lowest - that of government consumpton. In general, ntroducng mport tax would rather have unfavourable mpact on nvestment condtons, snce relaton of prces of nvestment goods to prces of consumpton would worsen. The ncrease of prces s also comparatvely hgh for exports. On the other hand, however, mport tax seems an attractve source of budget revenues. Government consumpton becomes relatvely cheap, meanng that besdes drect outcome n a form of tax revenue ncrease, addtonal redstrbuton effect appears. Smulaton 2 - VAT adaptaton to the EU standards In the second experment, potental effects of the forthcomng VAT rate changes are analysed. The pont of changes s lmtng the range of reduced and zero rates (e.g. n constructon and agrculture). The maor changes n VAT rates are ponted out n table 6. Table 6. Maor forthcomng changes n VAT rates n Poland (rates n %). Products and servces Current rate New rate Agrculture 3 7 Means of agrculture producton 3 7 Constructon 7 22 Materals for constructon 7 22 To set proper values of t for smulaton, new reference rates were calculated (see the prevous secton) takng nto account the planned changes. Calculatng dfferences between these rates and the old reference rates (see table 2), leads to fndng the necessary adustments of t n the smulaton. The adustments are presented n table 7. Tables 8 & 9 show results of smulaton 2.

23 Table 7. Changes of VAT rates n smulaton 2 (n percentage ponts). Products and servces Changes of VAT rates 1 Agrculture and forestry Wood Publshng and prntng Chemcals Other non-metallc Metal Metal products Machnes Furnture and other goods Constructon 15.0 In the case of VAT ncrease, prce changes are predomnated by drect effects, vsble n fnal prces. However, a slght reacton of basc prces can also be observed - generally beng at the level of 0.1% - 0.2%. For products and servces fully exempted from VAT ths reacton s reasonably stronger. For example, growths of basc prces of fshery products exceeds 1%, whle for publc admnstraton and defence, educaton, as well as publshng and prntng, these rates are close to 0.5%. As far as aggregate categores are concerned, t s consumpton that goes up the hghest, for other categores, prce reacton s very slght. Agan, as n the case of mport tax, the state budget addtonally benefts from relatve prce changes, but the cost ths tme s not transferred nether to nvestment nor exports. It can be sad that n the case of ncrease of the consdered VAT rates (agrculture products, constructon servces, materals for agrculture, materals for constructon, publshng and prntng products), the role of ndrect cost-push effects n prce formaton proved rather nsgnfcant. It means that VAT on these partcular products have lttle share n the non-deductble VAT n the economy. can be mostly deducted by producers, beng rather lucky concdence as far as nflaton n Poland s concerned. One should have n mnd, however, that the currently proposed procedure of purfyng nput-output table of VAT s mperfect and may lead to underestmatng the role of non-deductble VAT n formaton of basc prces. Thus, ths s the man ndcaton the man pont n further development of the prce model.

24 Table 8. Results of smulaton 2. Rates of growth of prces (n %). Products and servces Basc prces (domestc) Fnal prces (domestc + mported) 1 Agrculture and forestry Fshery Mnng Food Tobacco Fabrcs Textle Leather Wood Paper Publshng and prntng Petrol Chemcals Rubber and plastc Other non-metallc Metal Metal products Machnes Offce machnes and computers Electrc machnes Rado and TV devces Medcal and optcal devces Motor vehcles Other transport equpment Furnture and other goods Recyclng Electrcty, gas, water Constructon Hotels and restaurants Fnancal servces Busness and real estate servces Publc admnstraton and defence Educaton Health care and socal securty Other servces Repar Transport, storage, communcaton Table 9. Results of smulaton 2. Rates of growth of aggregate prces (n %). Categores of fnal expendtures Prce change + global output Prvate consumpton 1.7 Government consumpton 0.3 Investment 0.1 Export 0.1 Global domestc output 0.1

25 References Bardazz R., Grassn M. (1991), Value-added Taxes and Other Indrect Taxes n an EEC Country Model: the Italan Case, Economc Systems Research, Vol. 3, No. 1. Collado J. C., Sancho F. (2002), Recoverng Hdden Indrect Tax Rates for Improved Calbraton n Multsectoral Modellng, Economc Systems Research, Vol. 14, No. 1. Grassn M. (1997), The Structure of the Modern Multsectoral Model. The Input-Output Model of INFORUM System, [n:] Proceedngs of the 3rd World INFORUM Conference, Absolwent, Lodz. Mller R. E., Blar P. D. (1985), Input-Output Analyss: Foundatons and Extensons, Prentce-Hall, Inc., Englewood Clffs, New Jersey. System Rachunków Narodowych t. II (The System of Natonal Accounts vol. II), (1997), GUS (Central Statstcal Offce), Warszawa. Tomaszewcz Ł., Boratyńsk J (2003), Analza cenowych efektów zman podatków pośrednch na przykładze podatku mportowego (Prce Effects of Indrect Tax Changes. The Case of Import Tax), [n:] Wzrost gospodarczy, restrukturyzaca I rynek pracy w Polsce. Uęce teoretyczne empryczne, Katedra Ekonom UŁ, Łódź. Zenkowsk L. (2002), Co to est PKB? (What s the GDP?), Warszawa.