ISSN (Print) - 0012-9976 | ISSN (Online) - 2349-8846

A+| A| A-

An Assessment of the Revenue Impact of State-Level VAT in India

Revenue and GSDP data for 29 states for 1993-94 to 2008-09 are used to study the revenue performance of the state value added tax in India. The direct revenue impact was assessed by testing if VAT introduction increased VAT or state's own revenue buoyancies or the VAT or SOR to GSDP ratios. The indirect impact of VAT introduction on the VAT base (proxied by GSDP) and base growth were also examined. No indirect impacts of the VAT on its base was found. The direct revenue impact of the VAT was found to be positive in two-thirds of sample jurisdictions. A positive impact on SOR was however found only in Orissa and Haryana among 11 major states and 50% of other jurisdictions. Limited VAT revenue performance can partly be traced to large-scale evasion given weaknesses in VAT administration identified in a 2009 performance audit by the Comptroller and Auditor General. The implications of this study for the planned move to a goods and services VAT (from the current goods only VAT) are drawn and a suggestion is made for a non-VAT goods and services tax which should be less vulnerable to tax evasion.

SPECIAL ARTICLE

An Assessment of the Revenue Impact of State-Level VAT in India

Arindam Das-Gupta

Revenue and GSDP data for 29 states for 1993-94 to 2008-09 are used to study the revenue performance of the state value added tax in India. The direct revenue impact was assessed by testing if VAT introduction increased VAT or state’s own revenue buoyancies or the VAT or SOR to GSDP ratios. The indirect impact of VAT introduction on the VAT base (proxied by GSDP) and base growth were also examined. No indirect impacts of the VAT on its base was found. The direct revenue impact of the VAT was found to be positive in two-thirds of sample jurisdictions. A positive impact on SOR was however found only in Orissa and Haryana among 11 major states and 50% of other jurisdictions. Limited VAT revenue performance can partly be traced to large-scale evasion given weaknesses in VAT administration identified in a 2009 performance audit by the Comptroller and Auditor General. The implications of this study for the planned move to a goods and services VAT (from the current goods only VAT) are drawn and a suggestion is made for a non-VAT goods and services tax which should be less vulnerable to tax evasion.

This is an extensively revised and extended version of “A Preliminary Evaluation of the Revenue Impact of the State Level VAT in India” co-authored with Joy Chowdhury presented at the National Institute of Public Finance and Policy, New Delhi in March 2011. For this paper data have been corrected and updated, and the analysis has been altered drawing largely on comments received from seminar participants. Their suggestions are gratefully acknowledged individually where appropriate. Thanks are due to Fernanda Andrade for excellent research assistance. The usual disclaimers apply.

Arindam Das-Gupta (ronnie@gim.ac.in) is at the Centre for Economic Research, Goa Institute of Management, Goa.

Introduction and Motivation

T
he revenue performance of the state-level value added tax (VAT) in India relative to the turnover-type sales taxes it replaced is assessed here. Besides being the fi rst econometric assessment of sub-national VAT revenue performance, this assessment may serve as a benchmark for the proposed national and state Goods and Services Tax (GST). Barring further consensus building or implementation problems, the GST is to replace several central and state levies, including the central and state VATs over the next few years.

In developing countries the VAT is the consumption tax of choice of most applied public economists.1 However, Stiglitz and Dasgupta (1971) identified conditions under which VATlike exemption of productive intermediate inputs would not ensure economic efficiency. Some recent theoretical papers on the VAT also found it wanting when imperfect markets or i nformal sectors exist in the economy.2 On the other hand by granting input tax credits (ITC), the base of the VAT is narrower than a consumption tax without ITC, thus violating a widely accepted rule of thumb for practical design of general taxes, broad bases permitting low tax rates.3 One justification for this violation is that opposed interests of input suppliers (who benefit from evasion of VAT on their output) and buyers (who would like to claim ITC) make the VAT partly “self-enforcing”.4

Whatever its merits or drawbacks, the VAT is now implemented in at least 138 nations.5 In at least three of them ( Brazil, India and Quebec province of Canada) a sub-national VAT is also in place. The empirical assessments of the revenue i mpact of the VAT in Ebrill et al (2001) and also in Keen and

Table 1: Revenue Gain from VAT Adoption Region (% of GDP)

Region Ebrill et al (2001) Keen and Lockwood (2007)3, 4 Average Gain Number of: Average Gain (%) over Predecessor Countries Gaining Sales Tax (% of GDP)2, 4 Revenue from the VAT (Countries Losing Revenue)

Sub-Saharan Africa 1.10 11 (14) -0.81

Asia and Pacific 0.70 19 (3) 2.10

Americas 1.42 14 (9) 0.51

Central Europe and BRO1 -1.88 Not studied Not studied

EU (plus Norway and Switzerland) 1.05 17(0) 4.15

North Africa and west Asia 0.10 3 (2) 0.45

Small Islands 1.96 8 (0) 4.03

  • (1) BRO: Baltic states, Russia and other states of the former Soviet Union.
  • (2) Figures based on IMF staff calculations.
  • (3) Illustrative calculations by the authors based on their equation 4 (of 7 equations) estimated with panel data for 143 countries having at least 10 years of data between 1975 and 2000.The authors also estimate predicted revenue gain from VAT adoption for countries not having a VAT.
  • (4) Revenue variables: Ebrill et al (2001): VAT to GDP ratio over predecessor sales tax to GDP ratio. Keen and Lockwood (2007): tax-GDP ratio pre- and post-VAT.
  • Economic & Political Weekly

    EPW
    march 10, 2012 vol xlvii no 10

    Lockwood (2006, 2007) were limited to national level VATs. So an assessment of the VAT in India, which has had VATs on goods at both national and sub-national levels for around six years, is of interest. The revenue performance of the VAT reported in Ebrill et al and in Keen and Lockwood (2007) are reproduced in Table 1 (p 55).6 Clearly revenue gain from VAT adoption, while fairly widespread, were not universal.

    As Ebrill et al (2001) point out, these results need not refl ect poor VAT revenue potential. First, many countries intended a revenue neutral replacement of their earlier consumption taxes by the VAT, as in the Indian states. Second, design differences in VATs in different countries cause them to depart in various ways from a textbook VAT, again as in the states. Third, if the VAT causes less economic distortion than the tax it r eplaces, this may lead to VAT base (proxied by GDP) gains, i ncreasing the denominator in Table 1’s revenue ratios. The VAT fails on revenue grounds only if both the direct revenue impact plus its indirect impact on the VAT base are negative.

    Note that revenue is not the only performance criterion. A dministrative and cost efficiency, predictability, simplicity, impact on economic effi ciency, evasion proneness, equity and economic welfare as a whole are other important evaluation criteria. Of these, only administrative efficiency and evasion proneness are partly addressed below.

    After describing Indian state VAT design features, modelling and data issues are discussed, followed by the presentation of empirical results. Two robustness tests of the main e mpirical fi ndings and a review of a recent performance audit by the Comptroller and Auditor General (CAG) in 2010 are then presented. Policy suggestions based on the analysis conclude the paper.

    The VAT in Indian States

    Starting with Haryana and ending with Uttar Pradesh, between 2003-04 and 2007-08 VATs on goods were implemented in all Indian states and several union territories.7 Implementation dates for the 29 states are in Table 2.

    Table 2: Dates of VAT Implementation by States in India

    Haryana 1st Apr 2003
    Andhra Pradesh, Bihar, Haryana, Karnataka, Kerala, Maharashtra,
    Orissa, Punjab, West Bengal, Arunachal Pradesh, Assam,
    Himachal Pradesh, Goa, Jammu and Kashmir, Manipur,
    Meghalaya, Mizoram, Nagaland, NCT New Delhi, Sikkim, Tripura 1st Apr 2005
    Uttarakhand 1st Oct 2005
    Chhattisgarh, Madhya Pradesh, Gujarat, Rajasthan, Jharkhand 1st Apr 2006
    Tamil Nadu 1st Jan 2007

    Uttar Pradesh 1st Jan 2008

    Source: Halakhandi (2007) except Tamil Nadu: Government of Tamil Nadu (no date), and Uttar Pradesh: CA.inINDIA.Org (2011).

    Though VAT designs differ across states, among major widely prevailing design differences compared to a destination based consumption-type VAT are:8

  • The continuing origin-based central sales tax (CST) on interstate sales.
  • No VAT on imports from abroad.
  • Thresholds (differing across states) for registration of VAT dealers. Also in some states a simplified tax regime without input crediting for dealers below the VAT threshold but above a fl oor turnover.
  • Exclusion of certain goods including basic necessities, petroleum, oil and lubricants from the VAT.
  • Limits on VAT crediting for inputs and capital goods, and disallowance or carry forward of refunds in excess of tax paid on sales except for exports.
  • Consequently commodity taxes in the states continue to be partly origin based, tax intermediate inputs, and result in differential cascading across both goods and services. Even so, there are fewer design differences across the states than, for example, in the cross-country studies cited above. Furthermore, in India it is likely that the VAT was introduced to, inter alia, improve revenue but indirectly by reducing economic distortions and increasing the tax base.9

    Data and Modelling Issues

    To assess the revenue impact of the state VAT, the (a) gross state domestic product (GSDP) buoyancy of sales taxes (ST), and (b) the revenue to GSDP ratio, before and after VAT are examined. GDP (here GSDP) is the standard proxy for the base of general consumption taxes in most revenue performance studies. Two issues are examined. First, has the VAT done better than the sales tax it replaced? Second, has the VAT contributed to an improved own revenue performance? The latter is not assured if VAT gains are eroded by losses from other revenue sources, unintended or intended.10

    For the first question two equations, ST revenue pre- and post VAT implementation were compared:

    LNST = B0 + B1LNG + B2(VAT.LNG) ...(1)

    tttt

    (ST/G) = B0 + B1VAT ...(2)

    tt

    In (1) LN prefixed to a variable name denotes its natural logarithm, GSDP is abbreviated to G and the t is an annual time

    tt

    period subscript ranging from 1993-94 to 2008-09. VAT is a

    t

    dummy variable taking the value 1 for years in which the VAT prevailed and zero otherwise. Thus VATt.LNG is a slope dummy

    t

    variable. An increased coefficient of the VAT dummy in the buoyancy equation (1) is consistent with higher secular revenue productivity of the VAT compared to the earlier sales tax. An increase only in the VAT/GSDP ratio may reflect a one time increase in revenues due to the VAT, with no trend impact.

    For the second question, the same two equations but with state’s own revenue receipts (SORR) replacing ST are estimated:

    LNSORR = B0 + B1LNG + B2(VAT.LNG), ...(3)

    tttt

    (SORR/G) = B0 + B1VAT ...(4)

    tt

    An alternative to equations (1) and (3) with lagged Gt-1 replacing current G, (equations 1a and 3a) is also reported.11

    t

    A fifth equation was estimated to check if, even if there was no revenue increase, the VAT at least contributed a larger share of state revenue:12

    (ST/SORR) = B0 + B1VAT ...(5)

    tt

    These models do not include other possible determinants of revenue performance. Keen and Lockwood (2007), for example, estimate pooled regressions and so include additional “tax effort” determinants including a per capita income variable, a trade openness variable and the share of agriculture in GDP. These variables, which will vary little over the sample period in Indian states, are unlikely to contribute to the explanatory

    march 10, 2012 vol xlvii no 10

    power of the time series models analysed here. Furthermore trade openness data are not available for Indian states.13 However, as in other Indian studies, states are classified as major states and non-major states, the latter including the 10 special category states. Special category states are officially held to suffer from poor infrastructure, difficult terrain and in most cases large tribal populations.14

    The equations above neglect the indirect impact, if any, of VAT introduction on the VAT base. To assess this, two more equations were estimated using pooled data for the jurisdictions studied. The reason for data pooling was to take into a ccount possible cross-state economic spillovers on the VAT base.15 Using the subscript j for the jth state, the estimated equations were:

    LNGSDPjt = B + B1VATjt + B2Time + B3Statej, ...(6)

    o t

    ǻLNGSDPjt = B + B1VATjt + B2Time + B3Statej. ...(7)

    o t

    There is little alternative to the admittedly weak methodology of using a VAT dummy variable to assess the impact of the VAT. This methodology, with all its problems, is also used in earlier VAT impact studies including Ebrill et al (2001) and Keen and Lockwood (2006, 2007).16 However, this implies that differences between VAT and pre-VAT periods rather than the impact of the VAT are being studied. The technique cannot distinguish between the VAT’s impact and the impact of other tax and fi scal reforms during the period. For this detailed, state by state, for BJ and UPU in all equations. VAT1equals 1 for years in

    t

    which only one sibling state had the VAT and zero otherwise.

  • (b) Data for two states, Jammu and Kashmir and Karnataka were only available to 2007-08.
  • (c) Tamil Nadu and Uttarakhand (then Uttaranchal) implemented the VAT mid-year rather than on 1 April. A dummy variable for mid-year implementation was tried but, being insignificant, was dropped from the regressions reported here.
  • (d) GSDP data were from three different series: 1993-94, 19992000 and 2004-05. A chained GSDP series was, estimated by projecting the ratio of overlapping years of these series backward using a linear projection equation fitted by ordinary least squares. The resulting chained series thus has GSDP even for years before 2004-05 to the base year 2004-05. Equations (1) to
  • (4) were estimated with both chained and unchained GSDP s eries. With unchained GSDP data, VAT revenue performance turns out to be worse than with chained GSDP. So only chained series results are reported in the main text. Differences with unchained GSDP series are footnoted.
  • Empirical Results

    In Table 3, VAT dummy coefficients and their signifi cances are summarised from the detailed Appendix Tables A1 to A7 (pp 61-64). Table 4 (p 58) reports the mean values of the GSDP and SORR

    inquiries on the quality of VAT implementation Table 3: VAT Dummy Variable Signs and Significances for Equations (1) to (5)

    and also other reforms are needed. The quality of State ST GSDP ST Lagged ST/GSDP SORR GSDP SORR SORR/GSDP ST/SORR Buoyancy GSDP Buoyancy (Eq 2) Buoyancy Lagged GSDP (Eq 4) (Eq 5)

    VAT implementation is partly examined below by

    (Eq 1) (Eq 1a) (Eq 3) Buoyancy drawing on a VAT performance audit.17 Two (Eq 3a)

    other statistical exercises to check the robust-

    Major states Andhra Pradesh (AP) 0.000 0.003 0.004* 0.004 0.005 0.008 -0.003

    ness of the basic results were carried out.

    Gujarat (Guj) 0.011* 0.011* -0.002 0.003 0.004 -0.015* 0.236*

    Current rupee data on GSDP, ST and SORR

    Haryana (Har) 0.004 0.007 0.010* 0.019* 0.027* -0.023 0.142*

    are used for all 29 Indian states (clubbed into

    Karnataka (Kar) 0.008* 0.012* 0.001 0.008 0.013 0.010 -0.032 26 jurisdictions as explained below) for 2003-Kerala (Ker) -0.005* 0.000 0.003 0.002 0.007 0.003 0.01

    04 to 2008-09. ST and SORR data were from Maharashtra (Mah) -0.004 0.002 0.001 0.001 0.006 0.003 -0.006

    the website of the Reserve Bank of India (RBI) Orissa (Ori) 0.008 0.010 0.008* 0.011 0.014 0.017* 0.000 Punjab (Pun) 0.001 0.005 0.005* -0.005 0.004 -0.002 0.056*

    and GSDP data were from the website of the

    Rajasthan (Raj) 0.004 0.009 0.008* 0.010 0.015* 0.005 0.067*

    Ministry of Statistics and Programme Imple-

    West Bengal (WB) 0.007* 0.011* 0.001 0.007 0.01 0.004 -0.025

    mentation (MOSPI).18 Four data problems and

    Tamil Nadu (TN) 0.000 0.002 -0.008* 0.005 0.007 -0.008 -0.039*

    the manner in which they were dealt with are

    Non-major states now described. Arunachal Pradesh (ArP) 0.007 -0.032 0.013* 0.077* 0.074* 0.074* 0.082*

    (a) Chhattisgarh, Jharkhand, and Uttarakhand Assam (Asm) 0.007 0.005 0.015* 0.013* 0.011* 0.025* 0.047 Himachal Pradesh (HP) 0.022* 0.024* 0.013* 0.024* 0.027 0.035* 0.034

    were carved respectively out of Madhya Pradesh,

    Goa 0.018* 0.019* -0.014 -0.007 -0.003 -0.073* 0.065

    Bihar and Uttar Pradesh in 2000. So com-

    Jammu and Kashmir (JK) 0.023* 0.019* 0.022* 0.014* 0.012 0.029* 0.166*

    bined data for Bihar-Jharkhand (BJ), Madhya

    Manipur (Man) 0.029 0.018 0.007* 0.05* 0.053* 0.012* 0.215*

    Pradesh-Chhattisgarh (MPC), and Uttar Pradesh-

    Meghalaya Meg) 0.019* 0.023* 0.01* 0.011 0.015 0.008* 0.138* Uttarakhand (UPU) were used. This reduced Mizoram (Miz) 0.026 0.015 0.011* 0.05* 0.051* 0.008 0.168*

    the number of jurisdictions to 26 instead of 29. Nagaland (Nag) 0.020* 0.018 0.005* 0.024* 0.028 0.005 0.098*

    Since differences could arise after the split, an Sikkim (Sik) -0.020* -0.022 0.013* -0.041 0.004 -0.301 0.021 Tripura (Tri) 0.010 0.007 0.008* -0.008 -0.011 0.004 0.177

    additional dummy variable term, B3Split, was

    tNCT Delhi (ND) -0.010 0.007 0.006 -0.011 0.007 0.014 -0.038*

    added to equations (1) to (5) for these states,

    of which combined states

    with Split equalling one from the year of the

    tBihar+Jharkhand (BJ) 0.001 0.001 -0.002 0.007 0.005 -0.001 -0.021 split (2000 in all three cases) and zero before Madhya Pradesh+ that. Furthermore, in BJ and UPU, Bihar and Chhattisgarh (MPC) 0.003 0.007 0.006* 0.001 0.004 0.007 0.039* Uttar Pradesh+

    Uttarakhand implemented the VAT before

    Uttarakhand (UPU) 0.008* 0.013* 0.007* 0.017* 0.020* 0.012 0.01

    their sibling states (Table 2). So additional

    (1) *: Significant at 95% or better. P-values are reported in the Appendix. dummy variable terms, B4VAT1, were added (2) Of the combined states, Bihar, Madhya Pradesh and Uttar Pradesh are major states.

    t

    Economic & Political Weekly

    EPW
    march 10, 2012 vol xlvii no 10 57

    Table 4: Buoyancies and Mean Values of Ratios The last column of Table 3 shows that the
    State ST GSDP Buoyancy (Eq 1) ST Lagged GSDP Buoyancy (Eq 1a) ST/GSDP (Eq 2) SORR GSDP Buoyancy (Eq 3) SORR Lagged GSDP SORR/GSDP (Eq 4) ST/SORR (Eq 5) share of ST in SORR increased in only 11 states (excluding Arunachal), including four major
    Major states Andhra Pradesh 1.085 1.057 0.047 1.013 Buoyancy (Eq 3a) 1.029 0.091 0.520 states, after VAT introduction and reduced in one major and one non-major state. So reliance on sales taxes did not increase in the majority
    Gujarat 0.797 0.797 0.047 0.799 0.791 0.094 0.510 of states after the VAT reform.
    Haryana 1.147 1.128 0.039 0.548 0.421 0.127 0.388 On the indirect impact of VAT introduction,
    Karnataka 0.894 0.858 0.050 0.959 0.917 0.104 0.472 coeffi cients of VAT dummies in pooled regres-
    Kerala 1.154 1.073 0.051 1.005 0.931 0.080 0.643 sions with LNGSDP and the GSDP growth rate
    Maharashtra Orissa 1.095 1.117 1.034 1.119 0.040 0.027 1.015 1.061 0.958 1.062 0.082 0.062 0.484 0.440 in Table 6 (p 59) are uniformly insignifi cant. The conclusion is that VAT introduction did
    PunjabRajasthan West Bengal Tamil Nadu 1.166 1.171 0.894 0.875 1.156 1.141 0.835 0.835 0.033 0.032 0.026 0.062 1.079 0.893 0.959 0.865 0.935 0.836 0.907 0.826 0.108 0.082 0.047 0.105 0.320 0.401 0.543 0.580 not lead to any base expansion.22 So the direct revenue impact of the VAT is also the total r evenue impact.23
    Non-major states
    Arunachal Pradesh 4.510 5.476 0.003 0.856 0.984 0.050 0.080 Robustness Checks
    Assam 1.467 1.53 0.025 1.214 1.294 0.054 0.475 Given the questionable data, especially for
    Himachal Pradesh Goa Jammu and Kashmir Manipur Meghalaya Mizoram 1.288 0.643 1.700 1.353 1.310 1.992 1.267 0.607 1.75 1.519 1.258 2.155 0.017 0.064 0.017 0.009 0.015 0.005 1.103 0.756 1.297 0.634 1.017 0.581 1.052 0.673 1.299 0.561 0.958 0.573 0.064 0.203 0.055 0.033 0.047 0.049 0.284 0.335 0.329 0.304 0.347 0.151 GSDP, and methodological weakness, two r obustness checks are now presented. Further, in the next section findings of the performance audit (CAG 2010), which also tend to suggest negative or weak VAT revenue per-
    Nagaland 1.261 1.266 0.009 0.789 0.671 0.031 0.311 formance are presented.
    Sikkim 1.811 1.854 0.020 1.271 0.501 0.869 0.042
    Tripura 1.372 1.349 0.014 1.223 1.212 0.037 0.412 States Gaming the Centre: The centre agreed
    New Delhi 1.294 0.974 0.047 1.367 1.029 0.071 0.655 to compensate states implementing the VAT in
    of which combined states Bihar+Jharkhand Madhya Pradesh+ Chhattisgarh 0.887 1.232 1.015 1.315 0.028 0.026 0.821 1.128 0.975 1.198 0.059 0.079 0.478 0.347 2005 for any revenue loss in the initial years relative to sales tax revenue in 2004-05. The compensation would equal 100%, 75% and
    Uttar Pradesh+Uttarakhand 1.077 1.018 0.029 0.812 0.792 0.064 0.458 50% of the revenue loss in the fi rst, second
    Averages major states 1.036 1.003 0.041 0.927 0.874 0.089 0.482 and third years of the VAT, r espectively. Could
    All states 1.330 1.361 0.030 0.964 0.899 0.106 0.396 this have led to higher than normal state tax

    All buoyancies are significant at 99%: See Tables A1, A2, A4 and A5.

    ratios and also the buoyancies to help interpretation of Table 3.19 State by state narrative assessments are in Table 5 (p 59).

    Results for three states are difficult to interpret. In Andhra Pradesh the three ratios in the table appear mutually contradictory. They are, in any case, small. In Arunachal Pradesh, sales tax/VAT revenue grew 15,000% (in nominal terms) over the sample period while SORR grew by 1,000% or over twice as much as GSDP. Clearly, these gains cannot be attributed to VAT introduction alone. In Sikkim ratios and buoyancies appear to be mutually contradictory. In any case VAT appears to have no impact on revenue performance in Sikkim, the negative VAT buoyancy and ST/SORR may contradict this, implying questionable results. Leaving aside these states, Table 5 suggests that VAT revenue performance was positive in 15 of the remaining 23 jurisdictions including in six of 10 major states (excluding AP). Of these, in Karnataka, Kerala and UPU revenue gains were small.20

    Own revenue performance after VAT introduction improved in only two major states (Haryana and Orissa) and seven nonmajor states. Overall, even if VAT performance was positive in two-thirds of the states, improved own revenue performance after VAT introduction occurred in less than 40% of jurisdictions including only two major states.21

    effort in 2004-05 followed possibly by lower than normal tax effort particularly in 2005-06?24 If so VAT dummy coefficient estimates reported above would be biased downward and could turn insignifi cant.25 To examine this augmented versions of equations (1) and (2), equations (1b) and (2b) were estimated for the 21 states which implemented the VAT in 2005. The additional variables included were dummy variables for 2004-05 (PreVAT) and 2005-06 (PostVAT). These were slope dummies in (1b) and intercept dummies in (2b). The hypothesis is confirmed if PreVAT is positive and significant and, perhaps, PostVAT is negative and significant. Results are summarised in T able 7 (p 59). Further detail is in Tables A8 and A9 (p 64).

    For both equations the first of the four columns for each equation reports VAT dummy signs and significances from Table 3. In equations (1b) and (2b) PreVAT is positive and signifi cant in two and five states, respectively. In no case is PostVAT signifi cant. However, PreVAT/PostVAT have the expected positive/negative sign pattern in eight cases in (1b) and 10 cases in (2b). Thus the hypothesis of gaming has weak support. What of VAT dummy coefficients? In fact addition of PreVAT and PostVAT robs some VAT dummy coefficients (including two negative coeffi cients) of their significance. In no case does addition of these dummies cause an insignifi cant VAT dummy to become signifi cant

    march 10, 2012 vol xlvii no 10

    though they appear to cause downward bias in some cases. Thus the hypothesis of VAT dummy coefficients being insignifi cant due to states gaming the centre can be safely rejected. Downward bias of VAT dummy variables is, however, possible.

    Can Winners Compensate Losers? Instead of counting states with revenue improvements postVAT, an alternative is to see if states gaining revenue from the VAT could compensate states that

    Table 5: Impact of VAT Introduction on Sales Tax and State's Own Revenues: State by State Assessment

    State Assessment

    Major states Andhra Pradesh See discussion in the text

    Gujarat VAT not a success but other revenue sources performed even worse.

    Haryana Improved revenue performance including of the VAT.

    Karnataka VAT had no impact on revenue performance in Karnataka. ST buoyancy improved but by under 1%.

    Kerala VAT had no impact on revenue performance in Kerala. ST buoyancy worsened but by under 1%.

    Maharashtra VAT appears to have had no impact on revenue performance in Maharashtra.

    Orissa ST/GSDP and SORR/GSDP dummy coefficients are both

    large relative to the mean. Suggests improved revenue

    performance including of the VAT.

    Punjab From ST/SORR and ST/GSDP, VAT was successful. Other revenue sources eroded VAT gains.

    Rajasthan From ST/SORR and ST/GSDP, VAT was successful. Other revenue sources eroded VAT gains.

    West Bengal VAT had no impact on revenue performance. ST buoyancy improved but by under 1% leaving it below unity.

    Tamil Nadu VAT performance was worse than the sales tax it replaced, but overall revenue performance is unchanged.

    Non-major states Arunachal Pradesh See discussion in the text

    Assam Improved revenue performance including of the VAT.

    Himachal Pradesh Improved revenue performance including of the VAT.

    Goa VAT appears to have had no impact on revenue performance in Goa. ST buoyancy may have improved by around 3% still leaving it well below unity.

    Jammu and Kashmir Improved revenue performance including of the VAT.

    Manipur Improved revenue performance including of the VAT.

    Meghalaya Improved revenue performance including of the VAT.

    Mizoram Improved revenue performance including of the VAT.

    Nagaland VAT performance is positive but overall SORR performance has not improved.

    Sikkim See discussion in the text.

    Tripura ST/GSDP increased from a low level of 1%. No impact on overall own revenue performance.

    New Delhi VAT performance worse than the sales tax it replaced. Overall revenue performance is unchanged.

    of which combined states Bihar+Jharkhand VAT had no impact on revenue performance.

    Madhya Pradesh+ VAT performance is positive but had no impact on Chhattisgarh revenue performance.

    Uttar Pradesh+ Improved revenue performance including of the VAT, Uttarakhand though magnitude is small.

    lost revenue so that the country as a whole gained.26 To test this, data were aggregated across all 29 states in the sample and the following aggregate versions of equations (1) to (4) were estimated:

    LNST = B0+B1LNG+B2[VAT2003LNG] +B3[VAT2005LNG]+B4[VAT2006LNG]+B5[VAT2007LNG] (1c)

    ST/G = B0+ B1+B2+B4

    VAT2003VAT2005VAT2006 +B5 (2c)

    VAT2007

    LNSORR = B0+B1LNG+B2[VAT2003LNG] +B3[VAT2005LNG]+B4[VAT2006LNG]+B5[VAT2007LNG] (3c)

    SORR/G = B0+ B1+B2

    VAT2003VAT2005 +B4+B5 (4c)

    VAT2006VAT2007

    Four VAT dummy variables were needed given that states implemented the VAT in different years. For example, VAT2003 takes on the value 1 from 2003-04 onward to capture the VAT effect of states implementing the VAT in 2003 (from Table 2 this was only Haryana). Results, including F-tests for the joint significance of the four VAT dummies are in Table 8 (p 60).

    In Table 8, only the VAT dummies in equation (2c) are signifi cant. However, looking at the individual dummies in the equation only VAT2003, when Haryana alone introduced the VAT, is significant. Furthermore states’ own revenues in equations (3c) and (4c) were not significantly affected by the VAT. So it may be concluded that revenue gainers from the VAT could not compensate the losers.

    Did Tax Evasion Reduce VAT Performance?

    To what extent was VAT performance eroded by poor administration permitting leakage through tax non-compliance? For this the findings of the performance audit in CAG (2010) are revealing. The audit conducted during April-November 2009 covered 23 states27 and the post-VAT period 2005-06 to 2008-09, which is precisely the VAT years included in the sample in this paper (barring Haryana’s early VAT years). Using the 2005 white paper of the Empowered Committee of State Finance Ministers (ECSFM) which set out desirable basic VAT design and tax administration (TA) features as a benchmark, CAG (2010) assessed VAT performance.28

    The main findings of importance for this paper were:

  • Defi ciencies in VAT acts and rules existed in many states.
  • The large backlog of pending assessments under the predecessor taxes burdened TAs.
  • Incomplete automation, limited electronic return fi ling, and differences in VAT returns and documents across states seriously handicapped cross-verification of information in VAT r eturns across VAT dealers within and across states.
  • • Inability or unwillingness to

    Table 6: Impact of VAT Introduction on GSDP (pooled regressions for all states on state dummy variables, a time trend, and VAT

    cross-check information with

    period dummy variable)

    LNGSDP (Eqn 6) ǻLNGSDP (Eqn 7)
    Regression VAT Dummy Variable Regression VAT Dummy Variable
    Significance Coefficient Significance Significance Coefficient Significance
    Regression without combined states 0.000 0.024 0.484 0.000 -0.037 0.129
    Regression with combined states 0.000 0.029 0.344 0.000 -0.030 0.178

    Additional dummy variables for combined states are (a) from the year of states splitting, and (b) years during which only one of the combined states implemented the VAT.

    Economic & Political Weekly

    EPW
    march 10, 2012 vol xlvii no 10

    that available in other tax departments like the central excise and customs departments.

    • Ineffective procedures for verifying ITC claims and detecting fake ITC claims.

    Table 7: Signs and Significances of VAT, PreVAT and PostVAT Dummy Variables These official performance audit findings, based on extensive

    ST Buoyancy (eq 1b) ST/GSDP (eq 2b)

    test checks, provide independent verification of the relatively

    VAT - (eq 1) VAT PreVAT PostVAT VAT - (eq2) VAT PreVAT PostVAT

    poor revenue performance of the VAT found in this paper. The

    Andhra Pradesh 0.000 0.000 -0.020 -0.031 0.004* 0.005* 0.002 -0.003

    audit traces this to incomplete reforms and ineffective TAs.29 It

    Karnataka 0.008* 0.013* 0.173* -0.056 0.001 0.002 0.003 -0.001

    would be of interest to see if TA weakness can statistically ex-

    Kerala -0.005 -0.004 0.008 -0.036 0.003 0.004 0.005 -0.004 Maharashtra -0.004 -0.002 0.114 0.033 0.001 0.001 0.006* 0.001 plain poor revenue performance if state by state information Orissa 0.008 0.009 0.126 0.077 0.008* 0.008* 0.005 0.001 for the CAG report were made available. Note, however, that ad-Punjab 0.001 -0.001 0.127 0.207 0.005* 0.004 0.007 0.006 ministration of the predecessor sales taxes was also ineffective West Bengal 0.007* 0.012 0.163* -0.050 0.001 0.001 0.002 0.000 as documented by several studies and o ffi cial reports.30 The

    Arunachal incapacity of TAs to successfully cope with administering a Pradesh 0.007 0.005 0.311 0.448 0.013* 0.015* 0.006 -0.004

    new, sophisticated, tax like the VAT is strongly suggested by

    Assam 0.007 0.010 0.263 0.164 0.015* 0.015* 0.015* 0.004

    the CAG performance audit.

    Himachal Pradesh 0.022* 0.028* 0.164 -0.102 0.013* 0.015* 0.006* -0.005

    Implications for Near Term Reform

    Goa 0.018* 0.020* 0.041 -0.050 -0.014 -0.017 -0.021 0.002 Jammu Given the poor ability of states to cope with tax reforms docuand Kashmir 0.023* 0.027* 0.243 0.039 0.022* 0.024* 0.013* -0.002

    mented by the CAG and the possible negative impact of this on

    Manipur 0.029 0.038 0.120 -0.269 0.007* 0.009* 0.002 -0.005

    revenue is several states, further large-scale tax reform at this

    Meghalaya 0.019* 0.021* 0.136 0.039 0.010* 0.011* 0.005 -0.001

    stage appears premature, despite the three years of planning. TAs

    Mizoram 0.026 0.031 0.299 0.099 0.011* 0.012* 0.006* -0.002

    will have to cope with a greatly expanded number of dealers

    Nagaland 0.020* 0.021 0.024 0.008 0.005* 0.005* 0.002 0.000

    under the GST. Furthermore state TAs have no experience deal-

    Sikkim -0.020* -0.027 -0.208 0.022 0.013* 0.015* 0.009 -0.005

    ing with dealers providing services as there have been no gen-

    Tripura 0.010 0.011 0.035 -0.016 0.008* 0.009* 0.005 -0.002 New Delhi -0.010 -0.016 -0.114 0.179 0.006 0.006 0.006 0.004 eral state taxes on services. So while base broadening by

    (1) *: Significant at 95% or better. P-values are reported in the Appendix. i ncluding services is desirable in due course, this should not be

    attempted unless TAs expertise in taxing service providers.

  • Most states were without tax administration procedure Instead, performance benchmarks for TAs should be laid manuals. down with respect to current TA weaknesses and procedures
  • Problems with VAT dealer registration procedures allowing non-in implementing the VAT on goods. Moving to a GST should registration of some dealers and multiple registration of others. only be suggested if states can achieve the performance
  • Penalties for VAT non-compliance were at the discretion of benchmark as verified, for example, by another CAG perform-TAs and often not levied. ance audit.
  • On account of these TA deficiencies audit test checks of For states which had a positive VAT revenue performance around 1,00,000 dealers found widespread tax evasion and but poor own revenue performance, attention should possiavoidance through a variety of channels including (1) Under-bly be diverted to other revenue sources. Such states include declaration of sales and incorrect or false ITC claims by Chhattisgarh, Karnataka, Kerala, Madhya Pradesh, Naga50% of VAT dealers; (2) granting of incorrect VAT exemptions; land, Punjab and Rajasthan. For Goa and Gujarat causes of and (3) collection of VAT from customers which was not re-apparently declining tax effort should be identifi ed and mitted to state treasuries by some exempt dealers who con-c orrected. For Arunachal, Sikkim and Maharashtra further tinued to receive transitional benefits from earlier tax assessment to identify causes of apparently contradictory or i ncentive schemes. insignificant revenue performance indicators is needed.

    Are any base broadening (and conse-

    Table 8: Aggregate Regression Results for Equations (1c) to (4c)

    Variable/Statistic ST Buoyancy (Eqn 1c) ST/GSDP (Eqn 2c) SORR Buoyanc (Eqn 3c) SORR/GSDP (Eqn 4c) quent tax rate lowering) options avail-Coeff P-Value Coeff P-Value Coeff P-Value Coeff P-Value able for the existing VATs on goods? LNGSDP (Buoyancy) 1.050* 0.000 0.943* 0.000 One option is a move from 100% ITC to

    VAT*LNGSDP 0.003 0.227 0.003 0.197

    VAT*LNGSDP 0.002 0.550 0.002 0.501

    VAT*LNGSDP 0.000 0.934 0.002 0.580

    VAT*LNGSDP -0.002 0.569 0.000 0.968

    VAT0.004 0.028 0.002 0.467

    VAT0.002 0.402 0.003 0.598

    VAT0.001 0.809 0.003 0.638

    VAT-0.001 0.614 -0.001 0.876

    R-Squared 0.996 0.712 0.995 0.393

    F-Significance 0.000 0.009 0.000 0.244

    partial ITC at, say, 20% of input taxes paid by suppliers. As noted in the introduction, there is no theoretical justifi cation of any effi ciency benefit in countries like India from a 100% ITC. Evidence in Table 6 also suggests the absence of effi ciency benefi ts, though data and methodological weaknesses are present. Instead revenue loss due to

    F-Test: Joint significance of VAT evasion and TA inability to administer dummy variables 1.125 6.181* 1.800 2.037

    the ITC documented by the CAG will be

    F-test degrees of freedom (4.9) (4.10) (4.9) (4.10)

    limited as will loss from a narrow base

    (1) Sample period was 1993-94 to 2007-08 due to missing 2008-09 data for two states.

    (2) *: Significant at 99%. with a partial ITC. Furthermore, “self

    march 10, 2012 vol xlvii no 10

    enforcement” benefits from an ITC, if Appendix

    Table A1: LNST = B0+B1 LNGSDP+B2 [VAT.LNGSDP] (Equation 1)

    present, will continue with a 20% ITC. State R-square F Buoyancy Slope Dummy Dummy: Only Bihar/ Dummy: Years with
    Significance Variable (VAT) Uttarakhand VAT Bifurcated States
    Conclusions Coeff P-Value Coeff P-Value Coeff P-Value Coeff P-Value

    The state VAT was implemented in 2005 after a decade of preparation and at an unknown but large cost. From the statistical results and the CAG audit it can be

    Andhra Pradesh 0.993 0.000 1.085* 0.000 0.000 0.905
    Gujarat 0.992 0.000 0.797* 0.000 0.011* 0.001
    Haryana 0.992 0.000 1.147* 0.000 0.004 0.336
    Karnataka 0.991 0.000 0.894* 0.000 0.008* 0.023
    Kerala 0.998 0.000 1.154* 0.000 -0.005* 0.011

    inferred that the economic return in Maharashtra 0.988 0.000 1.095* 0.000 -0.004 0.262

    terms of revenue and efficiency gains to this expenditure of public funds is at best zero for the country as a whole. However, in Haryana, Orissa and the six identifi ed special category states in Table 5, the re-

    Orissa 0.985 0.000 1.117* 0.000 0.008 0.173
    Punjab 0.964 0.000 1.166* 0.000 0.001 0.902
    Rajasthan 0.979 0.000 1.171* 0.000 0.004 0.437
    West Bengal 0.989 0.000 0.894* 0.000 0.007* 0.016
    Tamil Nadu 0.990 0.000 0.875* 0.000 0.000 0.951
    Arunachal Pradesh 0.816 0.000 4.510* 0.000 0.007 0.924

    turn may have been large enough to jus-Assam 0.969 0.000 1.467* 0.000 0.007 0.398

    tify the cost of reform planning and implementation. Given the apparent lack of readiness of states, implementing the GST in 2012-13 is a high risk step whose returns may not repay the cost of planning

    Himachal Pradesh 0.992 0.000 1.288* 0.000 0.022* 0.000
    Goa 0.990 0.000 0.643* 0.000 0.018* 0.000
    Jammu and Kashmir 0.985 0.000 1.700* 0.000 0.023* 0.007
    Manipur 0.726 0.000 1.353* 0.004 0.029 0.441
    Meghalaya 0.992 0.000 1.31* 0.000 0.019* 0.002
    Mizoram 0.972 0.000 1.992* 0.000 0.026 0.160

    and implementing the GST. Nagaland 0.976 0.000 1.261* 0.000 0.020* 0.019

    Further state by state investigation is needed, particularly of tax administration and tax compliance, to throw more light on the costs and benefits of the 2005 VAT reform and devise a more extensive

    Sikkim 0.982 0.000 1.811* 0.000 -0.020* 0.077

    Tripura 0.991 0.000 1.372* 0.000 0.010 0.08
    New Delhi 0.932 0.000 1.294* 0.000 -0.010 0.433
    Bihar+ Jharkhand 0.970 0.000 0.887* 0.000 0.001 0.849 -0.011 0.933 0.120 0.299
    Madhya Pradesh+ Chhattisgarh 0.993 0.000 1.232* 0.000 0.003 0.399 0.227 0.003
    benchmarks for the proposed GST reform. Uttar Pradesh+
    Uttarakhand 0.999 0.000 1.077* 0.000 0.008* 0.000 0.147 0.000 0.188 0.000
    Notes Table A2: LNST = B0+B1 LNGSDP–1+B2 [VAT–1 LNGSDP–1] (Equation 1a)
    1 See Bird and Gendron (2007). State R-square F Buoyancy Slope Dummy Dummy: Only Bihar/ Dummy: Years with
    2 See Das-Gupta (2005), Emran and Stiglitz Significance (lagged GSDP) Variable (VAT) Uttarakhand VAT Bifurcated States
    (2005) and Keen and Ligthart (2005). These Coeff P-Value Coeff P-Value Coeff P-Value Coeff P-Value
    papers look variously at welfare, revenue and Andhra Pradesh 0.992 0.000 1.057 0.000 0.003 0.424
    efficiency of the VAT in relation to other taxes. 3 See, for example, Bird (2011) who terms this the BBLR approach. Gujarat Haryana 0.990 0.992 0.000 0.000 0.797 0.000 1.128 0.000 0.011 0.004 0.007 0.142
    4 See, for example, Chelliah and Kavita Rao Karnataka 0.979 0.000 0.858 0.000 0.012 0.024
    (1999). The other claimed advantage, removal Kerala 0.994 0.000 1.073 0.000 0.000 0.944
    of cascading or tax-on-tax, merely makes estimation of effective tax rates diffi cult without Maharashtra 0.980 0.000 1.034 0.000 0.002 0.684
    necessarily having any effi ciency impact. Orissa 0.983 0.000 1.119 0.000 0.010 0.092
    5 Jafari and Salehani (2010). Punjab 0.968 0.000 1.156 0.000 0.005 0.430
    6 The important study by Ebrill et al (2001) also Rajasthan 0.978 0.000 1.141 0.000 0.009 0.117
    assesses efficiency, equity and administrative aspects of the VAT for 100+ countries. 7 Maharashtra introduced a subtraction type West Bengal Tamil Nadu 0.984 0.978 0.000 0.000 0.835 0.000 0.835 0.000 0.011 0.005 0.002 0.588
    VAT in 1995. Due to design deficiencies, it re- Arunachal Pradesh 0.855 0.000 5.476 0.000 -0.032 0.649
    verted to a single point sales tax in 1999, Assam 0.969 0.000 1.530 0.000 0.005 0.564
    though partial setting off of input costs from sales revenue continued till 2005 (Government of Maharashtra 2000 and Chelliah and Kavita Himachal Pradesh Goa 0.995 0.989 0.000 0.000 1.267 0.000 0.607 0.000 0.024 0.000 0.019 0.001
    Rao 1999). This should be kept in mind when Jammu and Kashmir 0.994 0.000 1.750 0.000 0.019 0.001
    interpreting empirical results. 8 Updated VAT Acts, Rules, Circulars and Notifi cations of different states and union territories Manipur Meghalaya 0.732 0.991 0.000 0.000 1.519 0.005 1.258 0.000 0.018 0.647 0.023 0.001
    are available at www.stvat.com. Mizoram 0.984 0.000 2.155 0.000 0.015 0.298
    9 See, among others, Chelliah and Kavita Rao Nagaland 0.963 0.000 1.266 0.000 0.018 0.081
    (1999). Sikkim 0.978 0.000 1.854 0.000 -0.022 0.081
    10 If the cost of raising revenue through a VAT is less than that of the tax it replaces, then a state not wanting more revenue, could lower its Tripura New Delhi 0.994 0.987 0.000 0.000 1.349 0.000 0.974 0.000 0.007 0.118 0.007 0.120
    overall cost of funds by reducing dependence Bihar+Jharkhand 0.970 0.000 1.015 0.000 0.001 0.920 -0.105 0.452 0.027 0.822
    on other revenue sources. Theoretical arguments for this and related propositions are in Keen and Lockwood (2006). 11 This suggestion by M Govinda Rao is gratefully acknowledged. Regressions with slope and Madhya Pradesh+ Chhattisgarh Uttar Pradesh+ Uttarakhand 0.988 0.997 0.000 0.000 1.315 0.000 1.018 0.000 0.007 0.219 0.013 0.001 0.223 0.000 0.144 0.118 0.169 0.002
    Economic & Political Weekly march 10, 2012 v ol xlvii no 10 61
    EPW
    intercept dummies variables and both GSDPt-1 Table A3: (ST/GSDP) = B1 + B2 VAT (Equation 2)
    and GSDPt were unreliable with high multicol-State R-Square F VAT Dummy Variable Dummy: Only Bihar/ Dummy: Years with
    linearity. These are not reported here. Significance (VAT) Uttarakhand VAT Bifurcated States
    12 (5) also serves as a partial data consistency Coeff P-Value Coeff Coeff P-Value Coeff
    check by comparing its VATt coeffi cient sign and significance with that of VAT coeffi cients Andhra Pradesh 0.261 0.043 0.004* 0.043
    tin (2) and (4). Gujarat .0160 0.645 -0.002 0.645
    13 Keen and Lockwood (2006, 2007) use log (reve-Haryana 0.652 0.000 0.010* 0.000
    nue/GDP) as their dependent variable. This is Karnataka 0.032 0.526 0.001 0.526
    14 equivalent to a restricted regression with GDP buoyancy constrained to have the value 1. Special category states include Arunachal Kerala Maharashtra 0.147 0.010 0.143 0.719 0.003 0.001 0.143 0.719
    Pradesh, Assam, Himachal Pradesh, Jammu Orissa 0.607 0.000 0.008* 0.000
    15 and Kashmir, Manipur, Meghalaya, Mizoram, Nagaland, Tripura and Sikkim (Saxena 2009). Given the poor estimation results from this model, a simultaneous model with VATt and GSDPt as dependent variables and additional GSDP determinants was not specifi ed. Punjab Rajasthan West Bengal Tamil NaduArunachal Pradesh 0.247 0.418 0.079 0.299 0.786 0.050 0.007 0.291 0.028 0.000 0.005* 0.008* 0.001 -0.008* 0.013* 0.050 0.007 0.291 0.028 0.000
    16 As Keen (2009) puts it “Such a dummy varia- Assam 0.588 0.001 0.015* 0.001
    ble, though, is a very noisy indicator…. VATs differ enormously amongst themselves: in the extent of exemptions, threshold, number of Himachal Pradesh Goa 0.845 0.146 0.000 0.144 0.013* -0.014 0.000 0.144
    rates, ease of obtaining refunds, treatment of Jammu and Kashmir 0.749 0.000 0.022* 0.000
    17 18 services …” (p 162). Government of India, Comptroller and Auditor General (2010) abbreviated CAG. An Excel file with data used is available at Manipur Meghalaya Mizoram 0.592 0.817 0.762 0.000 0.000 0.000 0.007* 0.010* 0.011* 0.000 0.000 0.000
    http://www.gim.ac.in/data/32 states chained Nagaland 0.752 0.000 0.005* 0.000
    GSDP and RBI rev data 93-94 to 08-09.xls. Sikkim 0.391 0.010 0.013* 0.010

    19 The very high VAT buoyancy in Arunachal and its low VAT/GSDP ratio, discussed later, should be noted in interpreting results.

    20 That Maharashtra already had partial ITC prior to 2005 could be the cause of the insignificant dummies found here. This requires further study.

    21 With unchained GSDP, major states with significant, positive VAT dummies decreased from four to two while states with signifi cant, negative VAT buoyancy dummies increased from one to two. Significances of either buoyancy or ratio dummies changed for Andhra Pradesh, Gujarat, Karnataka, Rajasthan and West Bengal. VAT buoyancy and GSDP shares were both significantly positive for Rajasthan and both significantly negative for Kerala. The buoyancy dummy for Maharashtra was signifi cantly negative. For SORR, no major state had a positive significant VAT buoyancy dummy. The SORR/ GSDP VAT dummy for Andhra Pradesh was insignificant. Thus VAT performance is worse with unchained GSDP. For non-major states, dummy signs and significances were identical to those with chained GSDP except that the SORR/GSDP ratio VAT dummy for Mizoram was insignifi cant.

    22 The average annual GSDP growth rate of sample states fell from 12.3% in pre-VAT years to 11.7% post -VAT implementation.

    23 Significances of coefficients with unchained GSDP were identical though estimated coeffi cients were somewhat larger.

    24 The Economic Times (2004): Describes the compensation scheme. The possibility of states gaming the centre is reported, 23 September, for example, in Gupta (2005), 16 September.

    25 Thanks are due to Kavita Rao who fl agged this possibility, which led to this robustness check.

    26 Grateful thanks are due to Pulin Nayak for sug-

    Tripura 0.647 0.000 0.008* 0.000

    New Delhi 0.107 0.216 0.006 0.216

    Jharkhand+Bihar 0.065 0.839 -0.002 0.477 -0.002 0.657 0.001 0.437

    Madhya Pradesh+ Chhattisgarh 0.884 0.000 0.006* 0.004 0.010 0.000

    Uttar Pradesh+ Uttarakhand 0.982 0.000 0.007* 0.000 0.006 0.000 0.007 0.000

    Table A4: LNSORR = B0+B1 LNGSDP+B2 [VAT.LNGSDP] (Equation 3)

    State R-square F Buoyancy Slope Dummy Dummy for Only Bihar/ Dummy: Years Significance Variable (VAT) Uttarakhand VAT with Bifurcated Implementation States

    Coeff P-Value Coeff P-Value Coeff P-Value Coeff P-Value

    Andhra Pradesh 0.987 0.000 1.013 0.000 0.004 0.315

    Gujarat 0.986 0.000 0.799 0.000 0.003 0.373

    Haryana 0.923 0.000 0.548 0.000 0.019 0.046

    Karnataka 0.978 0.000 0.959 0.000 0.008 0.140

    Kerala 0.994 0.000 1.005 0.000 0.002 0.394

    Maharashtra 0.988 0.000 1.015 0.000 0.001 0.766

    Orissa 0.979 0.000 1.061 0.000 0.011 0.097

    Punjab 0.937 0.000 1.079 0.000 -0.005 0.533

    Rajasthan 0.968 0.000 0.893 0.000 0.010 0.094

    West Bengal 0.968 0.000 0.959 0.000 0.007 0.178

    Tamil Nadu 0.995 0.000 0.865 0.000 0.005 0.063

    Arunachal Pradesh 0.879 0.000 0.856 0.007 0.077 0.003

    Assam 0.982 0.000 1.214 0.000 0.013 0.024

    Himachal Pradesh 0.943 0.000 1.103 0.000 0.024 0.044

    Goa 0.952 0.000 0.756 0.000 -0.007 0.428

    Jammu and Kashmir 0.986 0.000 1.297 0.000 0.014 0.023

    Manipur 0.901 0.000 0.634 0.001 0.050 0.003

    Meghalaya 0.971 0.000 1.017 0.000 0.011 0.146

    Mizoram 0.965 0.000 0.581 0.000 0.050 0.000

    Nagaland 0.914 0.000 0.789 0.000 0.024 0.040

    gesting this check.

    Sikkim 0.462 0.018 1.271 0.016 -0.041 0.418

    27 That is all states studied here excluding

    Tripura 0.986 0.000 1.223 0.000 -0.008 0.156

    Haryana, Uttar Pradesh, Uttarakhand, Punjab,

    New Delhi 0.937 0.000 1.367 0.000 -0.011 0.383

    Arunachal Pradesh and Tamil Nadu.

    28 The ECSFM was set up by the centre in 1999 to Bihar+Jharkhand 0.932 0.000 0.821 0.001 0.007 0.537 0.095 0.628 0.057 0.728 coordinate VAT designs across states and arrive

    Madhya Pradesh+

    at a consensus design. The consensus design

    Chhattisgarh 0.983 0.000 1.128 0.000 0.001 0.875 0.085 0.297

    was described in the white paper (CAG 2010).

    Uttar Pradesh+

    The ECSFM currently plays the same role across states with respect to the planned GST. Uttarakhand 0.987 0.000 0.812 0.000 0.017 0.013 0.259 0.008 0.214 0.019

    62 march 10, 2012 vol xlvii no 10

    29 It should be noted that administration of the Table A5: LNSORR = B0+B1 LNGSDP-1+B2 [VAT-1 LNGSDP-1] (Equation 3a) predecessor sales taxes was also ineffective as

    State R-square F Buoyancy Slope Dummy Dummy for Only Bihar/ Dummy: Years

    was documented by several studies and offi cial

    Signi ficance (lagged GSDP) Variable Uttarakhand VAT with Bifurcated

    reports.

    (lagged VAT) Implementation States

    30 An example is Chapter 3 in World Bank (2005).

    Coeff P-Value Coeff P-Value Coeff P-Value Coeff P-Value

    References

    Bird, Richard (2011): “The BBLR Approach to Tax Reform in Emerging Countries” in M G Rao and M Rakshit (ed.), Public Economics: Theory and Policy (New Delhi: Sage Publishers).

    Andhra Pradesh 0.990 0.000 1.029 0.000 0.005 0.168

    Gujarat 0.990 0.000 0.791 0.000 0.004 0.214

    Haryana 0.908 0.000 0.421 0.006 0.027 0.013

    Karnataka 0.960 0.000 0.917 0.000 0.013 0.086

    Kerala 0.987 0.000 0.931 0.000 0.007 0.066

    Maharashtra 0.988 0.000 0.958 0.000 0.006 0.075

    Bird, Richard M and Pierre-Pascal Gendron (2007): The VAT in Developing and Transitional Countries (Cambridge and New York: Cambridge University Press).

    CA.inINDIA.Org (2011): e-bible for chartered accountants, available at http://www.cainindia. org/news/6_2008/vat_value_added_tax_2008_ india_news_.html, accessed 8 March.

    Chelliah, R J and Kavita Rao (1999): A Primer on the Value Added Tax, National Institute of Public Finance and Policy, New Delhi.

    Das-Gupta, Arindam (2005): “With Non-competitive Firms, a Turnover Tax Can Dominate the VAT”, Economics Bulletin, Vol 8, No 9, pp 1-6, available at http://www.economicsbulletin.com/ 2005/volume8/EB−05H20003A.pdf, last accessed March 2011.

    Diamond, P and J Mirrlees (1971): “Optimal Taxation and Public Production I: Production Effi -

    Orissa 0.965 0.000 1.062 0.000 0.014 0.111

    Punjab 0.939 0.000 0.935 0.000 0.004 0.610

    Rajasthan 0.957 0.000 0.836 0.000 0.015 0.030

    West Bengal 0.959 0.000 0.907 0.000 0.010 0.074

    Tamil Nadu 0.987 0.000 0.826 0.000 0.007 0.063

    Arunachal Pradesh 0.873 0.000 0.984 0.011 0.074 0.009

    Assam 0.986 0.000 1.294 0.000 0.011 0.043

    Himachal Pradesh 0.930 0.000 1.052 0.000 0.027 0.041

    Goa 0.948 0.000 0.673 0.000 -0.003 0.679

    Jammu and Kashmir 0.982 0.000 1.299 0.000 0.012 0.068

    Manipur 0.890 0.000 0.561 0.005 0.053 0.004

    Meghalaya 0.962 0.000 0.958 0.000 0.015 0.080

    Mizoram 0.958 0.000 0.573 0.000 0.051 0.000

    Nagaland 0.918 0.000 0.671 0.000 0.028 0.015

    Sikkim 0.399 0.047 0.501 0.109 0.004 0.902

    ciency”, American Economic Review, 61, 8-27.

    Ebrill, L, M Keen, J Bodin and V Summers (2001): The Modern VAT (Washington DC: International Monetary Fund).

    Emran, M Shahe and Joseph E Stiglitz (2005): “On Selective Indirect Tax Reform in Developing Countries”, Journal of Public Economics, Elsevier, Vol 89(4), pp 599-623, April.

    Government of India, Comptroller and Auditor General (2010): “Implementation of Value Added Tax in India – Lessons for Transition to Goods and Services Tax – A Study Report”, Comptroller and Auditor General, New Delhi, available at http://cag.gov.in/SRA-value-added-tax.pdf, accessed on 10 October 2011.

    Government of India, Ministry of Statistics and Programme Implementation (MOSPI) (2007): “Statement: Gross State Domestic Product at Current Prices” available at http://mospi.nic.in/ statewise_sdp1999_2000_8feb10.pdf on 11 Nov-

    Tripura 0.984 0.000 1.212 0.000 -0.011 0.075

    Delhi 0.992 0.000 1.029 0.000 0.007 0.071

    Bihar+Jharkhand 0.932 0.000 0.975 0.002 0.005 0.665 -0.005 0.980 -0.046 0.794

    Madhya Pradesh+ Chhattisgarh 0.976 0.000 1.198 0.000 0.004 0.520 0.010 0.920

    Uttar Pradesh+ Uttarakhand 0.982 0.000 0.792 0.000 0.020 0.013 0.307 0.007 0.191 0.077

    Table A6: (SORR/GSDP) = B1+B2 VAT (Equation 4)

    State R-Square F VAT Dummy Variable Dummy: for Only Bihar/ Dummy: Years with Significance

    Uttarakhand VAT Bifurcated States

    Coeff P-Value Coeff P-Value Coeff P-Value

    Andhra Pradesh 0.228 0.061 0.008 0.061

    Gujarat 0.265 0.042 -0.015 0.042

    Haryana 0.131 0.168 -0.023 0.168

    Karnataka 0.173 0.123 0.010 0.123

    Kerala 0.150 0.138 0.003 0.138

    ember 2009.

    Maharashtra 0.055 0.382 0.003 0.382

    Government of Maharashtra (2000): “Report of the

    Orissa 0.579 0.001 0.017 0.001

    Expert Group to Review Value Added Tax in

    Punjab 0.003 0.832 -0.002 0.832

    Maharashtra” (Valluri Narayan Committee), Government of Maharashtra, Mumbai. Rajasthan 0.051 0.401 0.005 0.401

    Government of Tamil Nadu, Commercial Taxes West Bengal 0.152 0.135 0.004 0.135 D epartment (no date): “Tamil Nadu Value

    Tamil Nadu 0.124 0.180 -0.008 0.180

    A dded Tax”, http://www.tnvat.gov.in, ac-

    Arunachal Pradesh 0.634 0.000 0.074 0.000

    cessed on 8 March 2011.

    Gupta, Monica (2005): “States Underplay VAT Assam 0.749 0.000 0.025 0.000 Gains to Get Aid”, 16 September, Business Himachal Pradesh 0.522 0.002 0.035 0.002 Standard online edition, available at http://

    Goa 0.397 0.009 -0.073 0.009

    www.business-standard.com/india/news/

    states-underplay-vat-gains-to-get-aid/220653/,

    Jammu and Kashmir 0.669 0.000 0.029 0.000 accessed on 20 October 2011. Manipur 0.280 0.035 0.012 0.035

    Halakhandi, Sudhir (2007): “CA Club India – Inter-Meghalaya 0.355 0.015 0.008 0.015 active Platform for Finance Professionals and

    Mizoram 0.083 0.280 0.008 0.280

    Tax Payers”, http://www.caclubindia.com/articles/value-added-tax-for-students-1508.asp,

    Nagaland 0.099 0.253 0.005 0.253 accessed on 8 March 2011. Sikkim 0.126 0.178 -0.301 0.178

    Jafari, Samimi, Ahmad and Fereshte Talesh Salehani (2010): “VAT and Governance: Evidence from Countries around the World”, Australian Journal of Basic and Applied Sciences, 4(10): 4852-56, available at http://www.insipub.com/ajbas/ 2010/4852-4856.pdf, last accessed March 2011.

    Keen, Michael (2009): “What Do (and Don’t) We Know about the Value Added Tax?”, A Review of Richard M Bird and Pierre-Pascal Gendron’s

    Economic & Political Weekly march 10, 2012

    EPW
    Tripura New Delhi 0.085 0.192 0.273 0.089 0.004 0.014 0.273 0.089
    Jharkhand+Bihar 0.051 0.856 -0.001 0.895 0.002 0.868 -0.004 0.497
    Madhya Pradesh+ Chhattisgarh Uttar Pradesh+ Uttarakhand 0.584 0.709 0.003 0.002 0.007 0.012 0.213 0.027 0.014 0.014 0.013 0.007 0.008 0.079
    vol xlvii no 10 63

    “The VAT in Developing and Transitional Countries”, Journal of Economic Literature 2009, 47(1), 159-170, available at http:www.aeaweb.org/articles.php?doi=10.1257/jel.47.1.159, accessed on 15 April 2011.

    Keen, Michael and Ben Lockwood (2006): “Is the VAT a Money Machine?”, National Tax Journal, 59(4), 905-928, available at http://www2.warwick.ac.uk/fac/soc/economics/staff/academic/ lockwood/mm.pdf, accessed on 20 October 2011.

    – (2007): “The Value Added Tax: Its Causes and Consequences”, http://www2.warwick.ac.uk/ fac/soc/economics/research/papers/twerp_ 801.pdf, accessed on 20 October 2011.

    Keen, Michael and Jenny E Ligthart (2005): “Coordinating Tariff Reduction and Domestic Tax Reform under Imperfect Competition”, Review of International Economics, Blackwell Publishing, Vol 13(2), pp 385-90.

    MO SPI (2011): “Statement: Gross State Domestic Product at Current Prices”, available at http:// mospi.gov.in/State-wise_SDP_1999-2000_ 20nov09.pdf, accessed on 11 November 2009.

    – (2009): “Statement: Gross State Domestic Product at Current Prices”, available at http:// mospi.gov.in/State-wise_SDP_1999-2000_ 20nov09.pdf, accessed on 11 November.

    Nellor, David (1987): “The Effect of the Value-Added Tax on the Tax Ratio”, IMF Working Paper, pp 1-28, 9 July, available at http://ssrn.com/ abstract=884798, accessed on 20 October 2011.

    Table A7: ST/SORR = B1+ B2 VAT (Equation 5)

    State R-Square F VAT Dummy Variable Dummy: for Only Bihar/ Dummy: Years with
    Significance Uttarakhand VAT Bifurcated States
    Coeff P-Value Coeff P-Value Coeff P-Value
    Andhra Pradesh 0.002 0.865 -0.003 0.865
    Gujarat 0.411 0.007 0.236 0.007
    Haryana 0.431 0.006 0.142 0.006
    Karnataka 0.187 0.108 -0.032 0.108
    Kerala 0.014 0.658 0.010 0.658
    Maharashtra 0.008 0.740 -0.006 0.74
    Orissa 0.000 0.996 0.000 0.996
    Punjab 0.375 0.012 0.056 0.012

    Newbery, D (1986): “On the Desirability of Input Taxes”, Economics Letters, 20, 267-70.

    Piffano, Horacio L P (2007): “Argentina and Brazil: Fiscal Harmonisation and Subnational Sales Taxation – State/Provincial VAT versus State/ Provincial Retail”, Working Paper, Departmento de Economia, Universidad Nacional de la Plata.

    Reserve Bank of India (2010): “Handbook of Statistics on State Government Finances – 2010”, available at: http://www.rbi.org.in/scripts/OccasionalPublications.aspx?head=Handbook% 20of%20Statistics%20on%20State%20Gov-

    Rajasthan West Bengal Tamil Nadu 0.249 0.089 0.543 0.049 0.262 0.001 0.067 -0.025 -0.039 0.049 0.262 0.001 ernment%20Finances%20-%202010, accessed on 3 January 2010. – (2011): “State Finances: A Study of Budgets”, Revenue Receipts of States and Union Territo-
    Arunachal Pradesh Assam Himachal Pradesh Goa 0.263 0.137 0.079 0.234 0.042 0.158 0.292 0.058 0.082 0.047 0.034 0.065 0.042 0.158 0.292 0.058 ries with Legislature, 31 March, Appendix I, available at http://www.rbi.org.in/scripts/ AnnualPublications.aspx?head=State+Finances +%3a+A+Study+of+Budgets, accessed on 3 July 2011.
    Jammu and Kashmir Manipur Meghalaya Mizoram Nagaland 0.596 0.312 0.661 0.493 0.325 0.001 0.025 0.000 0.002 0.027 0.166 0.215 0.138 0.168 0.098 0.001 0.025 0.000 0.002 0.027 Saxena, N C (2009): “Medium-term Fiscal Reforms Strategy for States”, Government of India, Planning Commission available at http://planningcommission.nic.in/reports/articles/ncsxna/index.php?repts=fi scal.htm#V.Special, accessed on 22 October 2011.

    Sikkim 0.089 0.262 0.021 0.262 Stiglitz J and P Dasgupta (1971): “Differential Taxation, Public Goods and Economic Effi ciency”,

    Tripura 0.824 0.000 0.177 0.000

    Review of Economic Studies, 38, 151-74.

    New Delhi 0.248 0.050 -0.038 0.050

    The Economic Times (2004): “VAT: States to Get Bihar+Jharkhand 0.283 0.247 -0.021 0.473 -0.042 0.353 0.049 0.055 Full Compensation, 23 September, available Madhya Pradesh+ at http://economictimes.indiatimes.com/vatstates-to-get-full-compensation/articleshow/

    Chhattisgarh 0.872 0.000 0.039 0.008 0.067 0.000

    860867.cms, accessed 20 October 2011.

    Uttar Pradesh+

    World Bank (2005): State Fiscal Reforms in India

    Uttarakhand 0.443 0.063 0.010 0.751 -0.003 0.915 0.055 0.024

    (New Delhi: Macmillan).

    Table A8: LNST = B0+B1 LNGSDP+B2 [VAT.LNGSDP]+B3 PreVAT+B4 PostVAT Table A9: LNSORR = B0+B1 LNGSDP+B2 [VAT.LNGSDP]+B3 PreVAT+B4 PostVAT

    State R-Square F VAT Slope Dummy PreVAT PostVAT State R-Square F VAT Slope PreVAT PostVAT
    Significance Variable Significance Dummy
    Coeff P-Value Coeff P-Value Coeff P-Value Coeff P-Value Coeff P-Value Coeff P-Value
    Andhra Pradesh 0.993 0.000 0.000 0.963 -0.020 0.801 -0.031 0.696 Andhra Pradesh 0.318 0.189 0.005 0.038 0.002 0.552 -0.003 0.441
    Karnataka 0.997 0.000 0.013 0.000 0.173 0.003 -0.056 0.255 Karnataka 0.085 0.796 0.002 0.480 0.003 0.460 -0.001 0.817
    Kerala 0.998 0.000 -0.004 0.082 0.008 0.825 -0.036 0.362 Kerala 0.353 0.143 0.004 0.052 0.005 0.129 -0.004 0.303
    Maharashtra 0.991 0.000 -0.002 0.609 0.114 0.161 0.033 0.673 Maharashtra 0.327 0.176 0.001 0.558 0.006 0.035 0.001 0.859
    Orissa 0.987 0.000 0.009 0.202 0.126 0.286 0.077 0.501 Orissa 0.707 0.002 0.008 0.001 0.005 0.072 0.001 0.659
    Punjab 0.972 0.000 -0.001 0.892 0.127 0.374 0.207 0.170 Punjab 0.466 0.050 0.004 0.109 0.007 0.102 0.006 0.206
    West Bengal 0.995 0.000 0.012 0.000 0.163 0.003 -0.050 0.286 West Bengal 0.160 0.536 0.001 0.273 0.002 0.307 0.000 0.888
    Arunachal Pradesh 0.819 0.000 0.005 0.960 0.311 0.827 0.448 0.748 Arunachal Pradesh 0.853 0.000 0.015 0.000 0.006 0.077 -0.004 0.202
    Assam 0.977 0.000 0.010 0.294 0.263 0.126 0.164 0.317 Assam 0.774 0.000 0.015 0.000 0.015 0.010 0.004 0.527
    Himachal Pradesh 0.995 0.000 0.028 0.000 0.164 0.066 -0.102 0.234 Himachal Pradesh 0.924 0.000 0.015 0.000 0.006 0.013 -0.005 0.062
    Goa 0.991 0.000 0.020 0.001 0.041 0.629 -0.050 0.555 Goa 0.242 0.326 -0.017 0.143 -0.021 0.243 0.002 0.898
    Jammu and Kashmir 0.988 0.000 0.027 0.009 0.243 0.122 0.039 0.801 Jammu and Kashmir 0.855 0.000 0.024 0.000 0.013 0.017 -0.002 0.695
    Manipur 0.730 0.004 0.038 0.438 0.120 0.876 -0.269 0.723 Manipur 0.672 0.003 0.009 0.000 0.002 0.486 -0.005 0.146
    Meghalaya 0.993 0.000 0.021 0.004 0.136 0.171 0.039 0.683 Meghalaya 0.883 0.000 0.011 0.000 0.005 0.026 -0.001 0.617
    Mizoram 0.975 0.000 0.031 0.174 0.299 0.377 0.099 0.763 Mizoram 0.847 0.000 0.012 0.000 0.006 0.031 -0.002 0.422
    Nagaland 0.977 0.000 0.021 0.068 0.024 0.877 0.008 0.957 Nagaland 0.784 0.001 0.005 0.000 0.002 0.236 0.000 0.824
    Sikkim 0.984 0.000 -0.027 0.061 -0.208 0.266 0.022 0.900 Sikkim 0.464 0.051 0.015 0.010 0.009 0.289 -0.005 0.537
    Tripura 0.991 0.000 0.011 0.126 0.035 0.766 -0.016 0.891 Tripura 0.718 0.001 0.009 0.000 0.005 0.130 -0.002 0.560
    New Delhi 0.935 0.000 -0.016 0.328 -0.114 0.695 0.179 0.539 New Delhi 0.142 0.591 0.006 0.342 0.006 0.573 0.004 0.702

    march 10, 2012 vol xlvii no 10

    Dear reader,

    To continue reading, become a subscriber.

    Explore our attractive subscription offers.

    Click here

    Comments

    (-) Hide

    EPW looks forward to your comments. Please note that comments are moderated as per our comments policy. They may take some time to appear. A comment, if suitable, may be selected for publication in the Letters pages of EPW.

    Back to Top