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 states 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.

Introduction and Motivation

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 first
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
informal 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
impact of the VAT in Ebrill et al (2001) and also in Keen and
Table 1: Revenue Gain from VAT Adoption Region (% of GDP)
Region

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.
Economic & Political Weekly

EPW

march 10, 2012

vol xlvii no 10

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
Asia and Pacific
Americas
Central Europe and BRO1
EU (plus Norway and Switzerland)
North Africa and west Asia
Small Islands

1.10
0.70
1.42
-1.88
1.05
0.10
1.96

11 (14)
19 (3)
14 (9)
Not studied
17(0)
3 (2)
8 (0)

-0.81
2.10
0.51
Not studied
4.15
0.45
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.

55

SPECIAL ARTICLE

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 reflect
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
replaces, this may lead to VAT base (proxied by GDP) gains,
increasing the denominator in Table 1s 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.
Administrative and cost efficiency, predictability, simplicity,
impact on economic efficiency, 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 empirical findings 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
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
Uttarakhand
Chhattisgarh, Madhya Pradesh, Gujarat, Rajasthan, Jharkhand
Tamil Nadu
Uttar Pradesh

1st Apr 2003

1st Apr 2005

1st Oct 2005
1st Apr 2006
1st Jan 2007
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 floor turnover.
56

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:
LNSTt = B0 + B1LNGt + B2(VATt.LNGt)
...(1)
(ST/G)t = B0 + B1VATt
...(2)
In (1) LN prefixed to a variable name denotes its natural logarithm, GSDPt is abbreviated to Gt and the t is an annual time
period subscript ranging from 1993-94 to 2008-09. VATt is a
dummy variable taking the value 1 for years in which the VAT
prevailed and zero otherwise. Thus VATt.LNGt is a slope dummy
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
states own revenue receipts (SORR) replacing ST are estimated:
LNSORRt = B0 + B1LNGt + B2(VATt.LNGt),
...(3)
(SORR/G)t = B0 + B1VATt
...(4)
An alternative to equations (1) and (3) with lagged Gt-1 replacing current Gt, (equations 1a and 3a) is also reported.11
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)t = B0 + B1VATt
...(5)
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

EPW

Economic & Political Weekly

SPECIAL ARTICLE

power of the time series models analysed here. Furthermore for BJ and UPU in all equations. VAT1t equals 1 for years in
trade openness data are not available for Indian states.13 How- which only one sibling state had the VAT and zero otherwise.
ever, as in other Indian studies, states are classified as major (b) Data for two states, Jammu and Kashmir and Karnataka
states and non-major states, the latter including the 10 special were only available to 2007-08.
category states. Special category states are officially held to (c) Tamil Nadu and Uttarakhand (then Uttaranchal) implesuffer from poor infrastructure, difficult terrain and in most mented the VAT mid-year rather than on 1 April. A dummy
cases large tribal populations.14
variable for mid-year implementation was tried but, being inThe equations above neglect the indirect impact, if any, of significant, was dropped from the regressions reported here.
VAT introduction on the VAT base. To assess this, two more equa- (d) GSDP data were from three different series: 1993-94, 1999tions were estimated using pooled data for the jurisdictions 2000 and 2004-05. A chained GSDP series was, estimated by
studied. The reason for data pooling was to take into account projecting the ratio of overlapping years of these series backpossible cross-state economic spillovers on the VAT base.15 Using ward using a linear projection equation fitted by ordinary least
the subscript j for the jth state, the estimated equations were:
squares. The resulting chained series thus has GSDP even for
LNGSDPjt = Bo + B1VATjt + B2Timet + B3Statej,
...(6) years before 2004-05 to the base year 2004-05. Equations (1) to
LNGSDPjt = Bo + B1VATjt + B2Timet + B3Statej.
...(7) (4) were estimated with both chained and unchained GSDP
There is little alternative to the admittedly weak methodo- series. With unchained GSDP data, VAT revenue performance
logy of using a VAT dummy variable to assess the impact of the turns out to be worse than with chained GSDP. So only chained
VAT. This methodology, with all its problems, is also used in ear- series results are reported in the main text. Differences with
lier VAT impact studies including Ebrill et al (2001) and Keen and unchained GSDP series are footnoted.
Lockwood (2006, 2007).16 However, this implies that differences between VAT and pre-VAT periods rather than the impact Empirical Results
of the VAT are being studied. The technique cannot distinguish In Table 3, VAT dummy coefficients and their significances are
between the VATs impact and the impact of other tax and fiscal summarised from the detailed Appendix Tables A1 to A7 (pp 61-64).
reforms during the period. For this detailed, state by state, 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)
ST GSDP
ST Lagged
ST/GSDP SORR GSDP
SORR
SORR/GSDP ST/SORR
and also other reforms are needed. The quality of State
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
17
(Eq 3a)
drawing on a VAT performance audit. Two
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
-0.004
0.002
0.001
0.001
0.006
0.003
-0.006
04 to 2008-09. ST and SORR data were from Maharashtra (Mah)
0.008
0.010
0.008* 0.011
0.014
0.017*
0.000
the website of the Reserve Bank of India (RBI) Orissa (Ori)
0.001
0.005
0.005* -0.005
0.004 -0.002
0.056*
and GSDP data were from the website of the Punjab (Pun)
Rajasthan (Raj)
0.004
0.009
0.008* 0.010
0.015* 0.005
0.067*
Ministry of Statistics and Programme ImpleWest 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
Arunachal Pradesh (ArP)
0.007
-0.032
0.013* 0.077* 0.074* 0.074*
0.082*
now described.
0.007
0.005
0.015* 0.013* 0.011*
0.025*
0.047
(a) Chhattisgarh, Jharkhand, and Uttarakhand Assam (Asm)
0.022*
0.024*
0.013* 0.024* 0.027
0.035*
0.034
were carved respectively out of Madhya Pradesh, Himachal Pradesh (HP)
Goa
0.018*
0.019* -0.014 -0.007 -0.003 -0.073*
0.065
Bihar and Uttar Pradesh in 2000. So comJammu 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*
0.020*
0.018
0.005* 0.024* 0.028
0.005
0.098*
the number of jurisdictions to 26 instead of 29. Nagaland (Nag)
-0.020*
-0.022
0.013* -0.041
0.004 -0.301
0.021
Since differences could arise after the split, an Sikkim (Sik)
Tripura
(Tri)
0.010
0.007
0.008*
-0.008
-0.011
0.004
0.177
additional dummy variable term, B3Splitt, was
NCT 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 Splitt equalling one from the year of the Bihar+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+
0.003
0.007
0.006* 0.001
0.004
0.007
0.039*
that. Furthermore, in BJ and UPU, Bihar and Chhattisgarh (MPC)
Uttarakhand implemented the VAT before Uttar Pradesh+
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, B4VAT1t, were added (2) Of the combined states, Bihar, Madhya Pradesh and Uttar Pradesh are major states.
Economic & Political Weekly

EPW

march 10, 2012

vol xlvii no 10

57

SPECIAL ARTICLE
Table 4: Buoyancies and Mean Values of Ratios
State

Major states
Andhra Pradesh
Gujarat
Haryana
Karnataka
Kerala
Maharashtra
Orissa
Punjab
Rajasthan
West Bengal
Tamil Nadu
Non-major states
Arunachal Pradesh
Assam
Himachal Pradesh
Goa
Jammu and Kashmir
Manipur
Meghalaya
Mizoram
Nagaland
Sikkim
Tripura
New Delhi
of which combined states
Bihar+Jharkhand
Madhya Pradesh+
Chhattisgarh
Uttar Pradesh+Uttarakhand
Averages major states
All states

ST GSDP
ST Lagged
ST/GSDP
Buoyancy GSDP Buoyancy (Eq 2)
(Eq 1)
(Eq 1a)

SORR GSDP
Buoyancy
(Eq 3)

SORR
SORR/GSDP
Lagged
(Eq 4)
GSDP
Buoyancy
(Eq 3a)

1.085
0.797
1.147
0.894
1.154
1.095
1.117
1.166
1.171
0.894
0.875

1.057
0.797
1.128
0.858
1.073
1.034
1.119
1.156
1.141
0.835
0.835

0.047
0.047
0.039
0.050
0.051
0.040
0.027
0.033
0.032
0.026
0.062

1.013
0.799
0.548
0.959
1.005
1.015
1.061
1.079
0.893
0.959
0.865

1.029
0.791
0.421
0.917
0.931
0.958
1.062
0.935
0.836
0.907
0.826

0.091
0.094
0.127
0.104
0.080
0.082
0.062
0.108
0.082
0.047
0.105

0.520
0.510
0.388
0.472
0.643
0.484
0.440
0.320
0.401
0.543
0.580

4.510
1.467
1.288
0.643
1.700
1.353
1.310
1.992
1.261
1.811
1.372
1.294

5.476
1.53
1.267
0.607
1.75
1.519
1.258
2.155
1.266
1.854
1.349
0.974

0.003
0.025
0.017
0.064
0.017
0.009
0.015
0.005
0.009
0.020
0.014
0.047

0.856
1.214
1.103
0.756
1.297
0.634
1.017
0.581
0.789
1.271
1.223
1.367

0.984
1.294
1.052
0.673
1.299
0.561
0.958
0.573
0.671
0.501
1.212
1.029

0.050
0.054
0.064
0.203
0.055
0.033
0.047
0.049
0.031
0.869
0.037
0.071

0.080
0.475
0.284
0.335
0.329
0.304
0.347
0.151
0.311
0.042
0.412
0.655

0.887

1.015

0.028

0.821

0.975

1.232
1.077
1.036
1.330

1.315
1.018
1.003
1.361

0.026
0.029
0.041
0.030

1.128
0.812
0.927
0.964

1.198
0.792
0.874
0.899

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
58

ST/SORR
(Eq 5)

The last column of Table 3 shows that the

share of ST in SORR increased in only 11 states
(excluding Arunachal), including four major
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
of states after the VAT reform.
On the indirect impact of VAT introduction,
coefficients of VAT dummies in pooled regressions with LNGSDP and the GSDP growth rate
in Table 6 (p 59) are uniformly insignificant.
The conclusion is that VAT introduction did
not lead to any base expansion.22 So the direct
revenue impact of the VAT is also the total
revenue impact.23
Robustness Checks

Given the questionable data, especially for

GSDP, and methodological weakness, two
robustness 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 performance are presented.

States Gaming the Centre: The centre agreed

to compensate states implementing the VAT in
2005 for any revenue loss in the initial years
0.059
0.478
relative to sales tax revenue in 2004-05. The
compensation would equal 100%, 75% and
0.079
0.347
0.064 0.458
50% of the revenue loss in the first, second
0.089
0.482
and third years of the VAT, respectively. Could
0.106
0.396
this have led to higher than normal state tax
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 insignificant.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
Table 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 significant in
two and five states, respectively. In no case is PostVAT significant.
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 coefficients)
of their significance. In no case does addition of these dummies cause an insignificant VAT dummy to become significant
march 10, 2012

vol xlvii no 10

EPW

Economic & Political Weekly

SPECIAL ARTICLE

though they appear to cause downward bias in some cases.

Thus the hypothesis of VAT dummy coefficients being insignificant 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

Major states
Andhra Pradesh
Gujarat
Haryana
Karnataka
Kerala
Maharashtra
Orissa

Punjab
Rajasthan
West Bengal
Tamil Nadu

Assessment

See discussion in the text

VAT not a success but other revenue sources performed
even worse.
Improved revenue performance including of the VAT.
VAT had no impact on revenue performance in Karnataka.
ST buoyancy improved but by under 1%.
VAT had no impact on revenue performance in Kerala. ST
buoyancy worsened but by under 1%.
VAT appears to have had no impact on revenue
performance in Maharashtra.
ST/GSDP and SORR/GSDP dummy coefficients are both
large relative to the mean. Suggests improved revenue
performance including of the VAT.
From ST/SORR and ST/GSDP, VAT was successful.
Other revenue sources eroded VAT gains.
From ST/SORR and ST/GSDP, VAT was successful.
Other revenue sources eroded VAT gains.
VAT had no impact on revenue performance. ST buoyancy
improved but by under 1% leaving it below unity.
VAT performance was worse than the sales tax it replaced,
but overall revenue performance is unchanged.

Non-major states
Arunachal Pradesh
Assam
Himachal Pradesh
Goa

See discussion in the text

Improved revenue performance including of the VAT.
Improved revenue performance including of the VAT.
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+ B1VAT2003+B2VAT2005+B4VAT2006

+B5VAT2007

(2c)

LNSORR = B0+B1LNG+B2[VAT2003LNG]
+B3[VAT2005LNG]+B4[VAT2006LNG]+B5[VAT2007LNG] (3c)
SORR/G = B0+ B1VAT2003+B2VAT2005
+B4VAT2006+B5VAT2007

(4c)

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 significant. 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 Haryanas 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:
Deficiencies 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 filing, and
differences in VAT returns and documents across states seriously handicapped cross-verification of information in VAT
returns 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)
that available in other tax deLNGSDP (Eqn 6)
LNGSDP (Eqn 7)
partments like the central exRegression
VAT Dummy Variable
Regression
VAT Dummy Variable
Significance
Coefficient
Significance
Significance
Coefficient
Significance
cise and customs departments.
Regression without combined states
0.000
0.024
0.484
0.000
-0.037
0.129
Ineffective procedures for veriRegression with combined states
0.000
0.029
0.344
0.000
-0.030
0.178
fying
ITC claims and detecting
Additional dummy variables for combined states are (a) from the year of states splitting, and (b) years during which only one of the
fake ITC claims.
combined states implemented the VAT.
Economic & Political Weekly

EPW

march 10, 2012

vol xlvii no 10

59

SPECIAL ARTICLE
Table 7: Signs and Significances of VAT, PreVAT and PostVAT Dummy Variables
ST Buoyancy (eq 1b)
ST/GSDP (eq 2b)
VAT - (eq 1) VAT
PreVAT PostVAT VAT - (eq2) VAT
PreVAT

Andhra Pradesh 0.000 0.000

Karnataka
0.008* 0.013*
Kerala
-0.005 -0.004
Maharashtra -0.004 -0.002
Orissa
0.008 0.009
Punjab
0.001 -0.001
West Bengal
0.007* 0.012
Arunachal
Pradesh
0.007 0.005
Assam
0.007 0.010
Himachal
Pradesh
0.022* 0.028*
Goa
0.018* 0.020*
Jammu
and Kashmir
0.023* 0.027*
Manipur
0.029 0.038
Meghalaya
0.019* 0.021*
Mizoram
0.026 0.031
Nagaland
0.020* 0.021
Sikkim
-0.020* -0.027
Tripura
0.010 0.011
New Delhi
-0.010 -0.016

-0.020
0.173*
0.008
0.114
0.126
0.127
0.163*

-0.031
-0.056
-0.036
0.033
0.077
0.207
-0.050

0.004*
0.001
0.003
0.001
0.008*
0.005*
0.001

0.005*
0.002
0.004
0.001
0.008*
0.004
0.001

0.002
0.003
0.005
0.006*
0.005
0.007
0.002

PostVAT

-0.003
-0.001
-0.004
0.001
0.001
0.006
0.000

0.311 0.448 0.013* 0.015* 0.006 -0.004

0.263 0.164 0.015* 0.015* 0.015* 0.004
0.164 -0.102 0.013* 0.015* 0.006* -0.005
0.041 -0.050 -0.014 -0.017 -0.021 0.002

These official performance audit findings, based on extensive

test checks, provide independent verification of the relatively
poor revenue performance of the VAT found in this paper. The
audit traces this to incomplete reforms and ineffective TAs.29 It
would be of interest to see if TA weakness can statistically explain poor revenue performance if state by state information
for the CAG report were made available. Note, however, that administration of the predecessor sales taxes was also ineffective
as documented by several studies and official reports.30 The
incapacity of TAs to successfully cope with administering a
new, sophisticated, tax like the VAT is strongly suggested by
the CAG performance audit.
Implications for Near Term Reform

Given the poor ability of states to cope with tax reforms documented by the CAG and the possible negative impact of this on
revenue is several states, further large-scale tax reform at this
stage appears premature, despite the three years of planning. TAs
will have to cope with a greatly expanded number of dealers
under the GST. Furthermore state TAs have no experience dealing with dealers providing services as there have been no general state taxes on services. So while base broadening by
(1) *: Significant at 95% or better. P-values are reported in the Appendix.
including services is desirable in due course, this should not be
attempted unless TAs expertise in taxing service providers.
Instead, performance benchmarks for TA s should be laid
Most states were without tax administration procedure
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 performance audit.
TAs and often not levied.
For states which had a positive VAT revenue performance
On account of these TA deficiencies audit test checks of
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 identified and
mitted to state treasuries by some exempt dealers who con- corrected. For Arunachal, Sikkim and Maharashtra further
tinued to receive transitional benefits from earlier tax assessment to identify causes of apparently contradictory or
insignificant revenue performance indicators is needed.
incentive schemes.
Are any base broadening (and conseTable 8: Aggregate Regression Results for Equations (1c) to (4c)
quent tax rate lowering) options availVariable/Statistic
ST Buoyancy (Eqn 1c)
ST/GSDP (Eqn 2c)
SORR Buoyanc (Eqn 3c)
SORR/GSDP (Eqn 4c)
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
0.003
0.227
0.003
0.197
VAT2003*LNGSDP
partial ITC at, say, 20% of input taxes
VAT2005*LNGSDP
0.002
0.550
0.002
0.501
paid by suppliers. As noted in the intro0.000
0.934
0.002
0.580
VAT2006*LNGSDP
duction, there is no theoretical justifi-0.002 0.569
0.000
0.968
VAT2007*LNGSDP
cation of any efficiency benefit in counVAT2003
0.004
0.028
0.002 0.467
tries like India from a 100% ITC. Evi0.002
0.402
0.003 0.598
VAT2005
dence in Table 6 also suggests the ab0.001
0.809
0.003 0.638
VAT2006
sence of efficiency benefits, though
-0.001 0.614
-0.001 0.876
VAT2007
data and methodological weaknesses
R-Squared
0.996
0.712
0.995
0.393
are present. Instead revenue loss due to
F-Significance
0.000
0.009
0.000
0.244
evasion and TA inability to administer
F-Test: Joint significance of VAT
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
60

0.243
0.120
0.136
0.299
0.024
-0.208
0.035
-0.114

0.039
-0.269
0.039
0.099
0.008
0.022
-0.016
0.179

0.022*
0.007*
0.010*
0.011*
0.005*
0.013*
0.008*
0.006

0.024*
0.009*
0.011*
0.012*
0.005*
0.015*
0.009*
0.006

0.013*
0.002
0.005
0.006*
0.002
0.009
0.005
0.006

-0.002
-0.005
-0.001
-0.002
0.000
-0.005
-0.002
0.004

march 10, 2012

vol xlvii no 10

EPW

Economic & Political Weekly

SPECIAL ARTICLE

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
Significance

Coeff

P-Value

0.993
0.992
0.992
0.991
0.998
0.988
0.985
0.964
0.979
0.989
0.990
0.816
0.969
0.992
0.990
0.985
0.726
0.992
0.972
0.976
0.982
0.991
0.932
0.970

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000

1.085*
0.797*
1.147*
0.894*
1.154*
1.095*
1.117*
1.166*
1.171*
0.894*
0.875*
4.510*
1.467*
1.288*
0.643*
1.700*
1.353*
1.31*
1.992*
1.261*
1.811*
1.372*
1.294*
0.887*

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.004
0.000
0.000
0.000
0.000
0.000
0.000
0.000

0.000
0.011*
0.004
0.008*
-0.005*
-0.004
0.008
0.001
0.004
0.007*
0.000
0.007
0.007
0.022*
0.018*
0.023*
0.029
0.019*
0.026
0.020*
-0.020*
0.010
-0.010
0.001

0.993

0.000

1.232*

0.000

0.003

0.999

0.000

1.077*

0.000

0.008* 0.000

Conclusions

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
inferred that the economic return in
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 identified
special category states in Table 5, the return may have been large enough to justify 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
and implementing the GST.
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
benchmarks for the proposed GST reform.

Andhra Pradesh
Gujarat
Haryana
Karnataka
Kerala
Maharashtra
Orissa
Punjab
Rajasthan
West Bengal
Tamil Nadu
Arunachal Pradesh
Assam
Himachal Pradesh
Goa
Jammu and Kashmir
Manipur
Meghalaya
Mizoram
Nagaland
Sikkim
Tripura
New Delhi
Bihar+ Jharkhand
Madhya Pradesh+
Chhattisgarh
Uttar Pradesh+
Uttarakhand

Buoyancy

Slope Dummy
Dummy: Only Bihar/ Dummy: Years with
Variable (VAT)
Uttarakhand VAT
Bifurcated States
Coeff
P-Value Coeff P-Value Coeff P-Value

0.905
0.001
0.336
0.023
0.011
0.262
0.173
0.902
0.437
0.016
0.951
0.924
0.398
0.000
0.000
0.007
0.441
0.002
0.160
0.019
0.077
0.08
0.433
0.849 -0.011 0.933

0.120 0.299

0.399

0.227 0.003

Notes

Table A2: LNST = B0+B1 LNGSDP1+B2 [VAT1 LNGSDP1] (Equation 1a)

1 See Bird and Gendron (2007).

2 See Das-Gupta (2005), Emran and Stiglitz
(2005) and Keen and Ligthart (2005). These
papers look variously at welfare, revenue and
efficiency of the VAT in relation to other taxes.
3 See, for example, Bird (2011) who terms this
the BBLR approach.
4 See, for example, Chelliah and Kavita Rao
(1999). The other claimed advantage, removal
of cascading or tax-on-tax, merely makes estimation of effective tax rates difficult without
necessarily having any efficiency impact.
5 Jafari and Salehani (2010).
6 The important study by Ebrill et al (2001) also
assesses efficiency, equity and administrative
aspects of the VAT for 100+ countries.
7 Maharashtra introduced a subtraction type
VAT in 1995. Due to design deficiencies, it reverted to a single point sales tax in 1999,
though partial setting off of input costs from
sales revenue continued till 2005 (Government
of Maharashtra 2000 and Chelliah and Kavita
Rao 1999). This should be kept in mind when
interpreting empirical results.
8 Updated VAT Acts, Rules, Circulars and Notifications of different states and union territories
are available at www.stvat.com.
9 See, among others, Chelliah and Kavita Rao
(1999).
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
overall cost of funds by reducing dependence
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

State

Economic & Political Weekly

EPW

march 10, 2012

Andhra Pradesh
Gujarat
Haryana
Karnataka
Kerala
Maharashtra
Orissa
Punjab
Rajasthan
West Bengal
Tamil Nadu
Arunachal Pradesh
Assam
Himachal Pradesh
Goa
Jammu and Kashmir
Manipur
Meghalaya
Mizoram
Nagaland
Sikkim
Tripura
New Delhi
Bihar+Jharkhand
Madhya Pradesh+
Chhattisgarh
Uttar Pradesh+
Uttarakhand
vol xlvii no 10

R-square

F
Significance

Buoyancy
(lagged GSDP)
Coeff
P-Value

0.992
0.990
0.992
0.979
0.994
0.980
0.983
0.968
0.978
0.984
0.978
0.855
0.969
0.995
0.989
0.994
0.732
0.991
0.984
0.963
0.978
0.994
0.987
0.970

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000

1.057
0.797
1.128
0.858
1.073
1.034
1.119
1.156
1.141
0.835
0.835
5.476
1.530
1.267
0.607
1.750
1.519
1.258
2.155
1.266
1.854
1.349
0.974
1.015

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.005
0.000
0.000
0.000
0.000
0.000
0.000
0.000

0.003
0.011
0.007
0.012
0.000
0.002
0.010
0.005
0.009
0.011
0.002
-0.032
0.005
0.024
0.019
0.019
0.018
0.023
0.015
0.018
-0.022
0.007
0.007
0.001

0.988

0.000

1.315

0.000

0.997

0.000

1.018

0.000

0.147 0.000

0.188 0.000

Slope Dummy
Dummy: Only Bihar/ Dummy: Years with
Variable (VAT)
Uttarakhand VAT
Bifurcated States
Coeff
P-Value Coeff P-Value Coeff P-Value

0.424
0.004
0.142
0.024
0.944
0.684
0.092
0.430
0.117
0.005
0.588
0.649
0.564
0.000
0.001
0.001
0.647
0.001
0.298
0.081
0.081
0.118
0.120
0.920 -0.105 0.452

0.027 0.822

0.007

0.219

0.144

0.013

0.001

0.223 0.000

0.118

0.169 0.002

61

SPECIAL ARTICLE

12

13

14

15

16

17
18

19

20

21

22

23

24

25
26
27

28

62

intercept dummies variables and both GSDPt-1

and GSDPt were unreliable with high multicollinearity. These are not reported here.
(5) also serves as a partial data consistency
check by comparing its VATt coefficient sign
and significance with that of VATt coefficients
in (2) and (4).
Keen and Lockwood (2006, 2007) use log (revenue/GDP) as their dependent variable. This is
equivalent to a restricted regression with GDP
buoyancy constrained to have the value 1.
Special category states include Arunachal
Pradesh, Assam, Himachal Pradesh, Jammu
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 specified.
As Keen (2009) puts it Such a dummy variable, though, is a very noisy indicator. VATs
differ enormously amongst themselves: in the
extent of exemptions, threshold, number of
rates, ease of obtaining refunds, treatment of
services (p 162).
Government of India, Comptroller and Auditor
General (2010) abbreviated CAG.
An Excel file with data used is available at
http://www.gim.ac.in/data/32 states chained
GSDP and RBI rev data 93-94 to 08-09.xls.
The very high VAT buoyancy in Arunachal and
its low VAT/GSDP ratio, discussed later, should
be noted in interpreting results.
That Maharashtra already had partial ITC
prior to 2005 could be the cause of the insignificant dummies found here. This requires further study.
With unchained GSDP, major states with significant, positive VAT dummies decreased from
four to two while states with significant, 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 significantly 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 insignificant.
The average annual GSDP growth rate of sample states fell from 12.3% in pre-VAT years to
11.7% post -VAT implementation.
Significances of coefficients with unchained
GSDP were identical though estimated coefficients were somewhat larger.
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.
Thanks are due to Kavita Rao who flagged this
possibility, which led to this robustness check.
Grateful thanks are due to Pulin Nayak for suggesting this check.
That is all states studied here excluding
Haryana, Uttar Pradesh, Uttarakhand, Punjab,
Arunachal Pradesh and Tamil Nadu.
The ECSFM was set up by the centre in 1999 to
coordinate VAT designs across states and arrive
at a consensus design. The consensus design
was described in the white paper (CAG 2010).
The ECSFM currently plays the same role
across states with respect to the planned GST.

Table A3: (ST/GSDP) = B1 + B2 VAT (Equation 2)

State

R-Square

F
Significance

VAT Dummy Variable

(VAT)
Coeff
P-Value

Andhra Pradesh
Gujarat
Haryana
Karnataka
Kerala
Maharashtra
Orissa
Punjab
Rajasthan
West Bengal
Tamil Nadu
Arunachal Pradesh
Assam
Himachal Pradesh
Goa
Jammu and Kashmir
Manipur
Meghalaya
Mizoram
Nagaland
Sikkim
Tripura
New Delhi
Jharkhand+Bihar
Madhya Pradesh+
Chhattisgarh
Uttar Pradesh+
Uttarakhand

0.261
.0160
0.652
0.032
0.147
0.010
0.607
0.247
0.418
0.079
0.299
0.786
0.588
0.845
0.146
0.749
0.592
0.817
0.762
0.752
0.391
0.647
0.107
0.065

0.043
0.645
0.000
0.526
0.143
0.719
0.000
0.050
0.007
0.291
0.028
0.000
0.001
0.000
0.144
0.000
0.000
0.000
0.000
0.000
0.010
0.000
0.216
0.839

0.004*
-0.002
0.010*
0.001
0.003
0.001
0.008*
0.005*
0.008*
0.001
-0.008*
0.013*
0.015*
0.013*
-0.014
0.022*
0.007*
0.010*
0.011*
0.005*
0.013*
0.008*
0.006
-0.002

0.043
0.645
0.000
0.526
0.143
0.719
0.000
0.050
0.007
0.291
0.028
0.000
0.001
0.000
0.144
0.000
0.000
0.000
0.000
0.000
0.010
0.000
0.216
0.477

0.884

0.000

0.006*

0.004

0.982

0.000

0.007*

0.000

Dummy: Only Bihar/

Uttarakhand VAT
Coeff
Coeff

-0.002

0.006

0.657

0.000

Dummy: Years with

Bifurcated States
P-Value
Coeff

0.001

0.437

0.010

0.000

0.007

0.000

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

State

Andhra Pradesh
Gujarat
Haryana
Karnataka
Kerala
Maharashtra
Orissa
Punjab
Rajasthan
West Bengal
Tamil Nadu
Arunachal Pradesh
Assam
Himachal Pradesh
Goa
Jammu and Kashmir
Manipur
Meghalaya
Mizoram
Nagaland
Sikkim
Tripura
New Delhi
Bihar+Jharkhand
Madhya Pradesh+
Chhattisgarh
Uttar Pradesh+
Uttarakhand

R-square

F
Significance

Buoyancy

Coeff

P-Value

Coeff

0.987
0.986
0.923
0.978
0.994
0.988
0.979
0.937
0.968
0.968
0.995
0.879
0.982
0.943
0.952
0.986
0.901
0.971
0.965
0.914
0.462
0.986
0.937
0.932

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.018
0.000
0.000
0.000

1.013
0.799
0.548
0.959
1.005
1.015
1.061
1.079
0.893
0.959
0.865
0.856
1.214
1.103
0.756
1.297
0.634
1.017
0.581
0.789
1.271
1.223
1.367
0.821

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.007
0.000
0.000
0.000
0.000
0.001
0.000
0.000
0.000
0.016
0.000
0.000
0.001

0.004
0.003
0.019
0.008
0.002
0.001
0.011
-0.005
0.010
0.007
0.005
0.077
0.013
0.024
-0.007
0.014
0.050
0.011
0.050
0.024
-0.041
-0.008
-0.011
0.007

0.315
0.373
0.046
0.140
0.394
0.766
0.097
0.533
0.094
0.178
0.063
0.003
0.024
0.044
0.428
0.023
0.003
0.146
0.000
0.040
0.418
0.156
0.383
0.537

0.983

0.000

1.128

0.000

0.001

0.875

0.987

0.000

0.812

0.000

0.017

0.013

march 10, 2012

Slope Dummy
Variable (VAT)

Dummy for Only Bihar/ Dummy: Years

Uttarakhand VAT
with Bifurcated
Implementation
States
P-Value Coeff P-Value Coeff P-Value

vol xlvii no 10

0.095 0.628

0.057 0.728
0.085 0.297

0.259 0.008
EPW

0.214 0.019

Economic & Political Weekly

SPECIAL ARTICLE
29 It should be noted that administration of the
predecessor sales taxes was also ineffective as
was documented by several studies and official
reports.
30 An example is Chapter 3 in World Bank (2005).

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).
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/EB05H20003A.pdf, last accessed March 2011.
Diamond, P and J Mirrlees (1971): Optimal Taxation and Public Production I: Production Efficiency, 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 November 2009.
Government of Maharashtra (2000): Report of the
Expert Group to Review Value Added Tax in
Maharashtra (Valluri Narayan Committee),
Government of Maharashtra, Mumbai.
Government of Tamil Nadu, Commercial Taxes
Department (no date): Tamil Nadu Value
Added Tax, http://www.tnvat.gov.in, accessed on 8 March 2011.
Gupta, Monica (2005): States Underplay VAT
Gains to Get Aid, 16 September, Business
Standard online edition, available at http://
www.business-standard.com/india/news/
states-underplay-vat-gains-to-get-aid/220653/,
accessed on 20 October 2011.
Halakhandi, Sudhir (2007): CA Club India Interactive Platform for Finance Professionals and
Tax Payers, http://www.caclubindia.com/articles/value-added-tax-for-students-1508.asp,
accessed on 8 March 2011.
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 Dont) We
Know about the Value Added Tax?, A Review
of Richard M Bird and Pierre-Pascal Gendrons
Economic & Political Weekly

EPW

march 10, 2012

Table A5: LNSORR = B0+B1 LNGSDP-1+B2 [VAT-1 LNGSDP-1] (Equation 3a)

State

Andhra Pradesh
Gujarat
Haryana
Karnataka
Kerala
Maharashtra
Orissa
Punjab
Rajasthan
West Bengal
Tamil Nadu
Arunachal Pradesh
Assam
Himachal Pradesh
Goa
Jammu and Kashmir
Manipur
Meghalaya
Mizoram
Nagaland
Sikkim
Tripura
Delhi
Bihar+Jharkhand
Madhya Pradesh+
Chhattisgarh
Uttar Pradesh+
Uttarakhand

R-square

F
Signi ficance

Buoyancy
(lagged GSDP)

Slope Dummy Dummy for Only Bihar/ Dummy: Years

Variable
Uttarakhand VAT
with Bifurcated
(lagged VAT)
Implementation
States
Coeff
P-Value Coeff P-Value Coeff P-Value

Coeff

P-Value

0.990
0.990
0.908
0.960
0.987
0.988
0.965
0.939
0.957
0.959
0.987
0.873
0.986
0.930
0.948
0.982
0.890
0.962
0.958
0.918
0.399
0.984
0.992
0.932

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.047
0.000
0.000
0.000

1.029
0.791
0.421
0.917
0.931
0.958
1.062
0.935
0.836
0.907
0.826
0.984
1.294
1.052
0.673
1.299
0.561
0.958
0.573
0.671
0.501
1.212
1.029
0.975

0.000
0.000
0.006
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.011
0.000
0.000
0.000
0.000
0.005
0.000
0.000
0.000
0.109
0.000
0.000
0.002

0.005
0.004
0.027
0.013
0.007
0.006
0.014
0.004
0.015
0.010
0.007
0.074
0.011
0.027
-0.003
0.012
0.053
0.015
0.051
0.028
0.004
-0.011
0.007
0.005

0.976

0.000

1.198

0.000

0.004

0.520

0.982

0.000

0.792

0.000

0.020

0.013

0.168
0.214
0.013
0.086
0.066
0.075
0.111
0.610
0.030
0.074
0.063
0.009
0.043
0.041
0.679
0.068
0.004
0.080
0.000
0.015
0.902
0.075
0.071
0.665 -0.005 0.980 -0.046 0.794
0.010 0.920
0.307 0.007

0.191 0.077

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

State

R-Square

F
Significance

Coeff

P-Value

Dummy: for Only Bihar/

Uttarakhand VAT
Coeff
P-Value

Andhra Pradesh
Gujarat
Haryana
Karnataka
Kerala
Maharashtra
Orissa
Punjab
Rajasthan
West Bengal
Tamil Nadu
Arunachal Pradesh
Assam
Himachal Pradesh
Goa
Jammu and Kashmir
Manipur
Meghalaya
Mizoram
Nagaland
Sikkim
Tripura
New Delhi
Jharkhand+Bihar
Madhya Pradesh+
Chhattisgarh
Uttar Pradesh+
Uttarakhand

0.228
0.265
0.131
0.173
0.150
0.055
0.579
0.003
0.051
0.152
0.124
0.634
0.749
0.522
0.397
0.669
0.280
0.355
0.083
0.099
0.126
0.085
0.192
0.051

0.061
0.042
0.168
0.123
0.138
0.382
0.001
0.832
0.401
0.135
0.180
0.000
0.000
0.002
0.009
0.000
0.035
0.015
0.280
0.253
0.178
0.273
0.089
0.856

0.008
-0.015
-0.023
0.010
0.003
0.003
0.017
-0.002
0.005
0.004
-0.008
0.074
0.025
0.035
-0.073
0.029
0.012
0.008
0.008
0.005
-0.301
0.004
0.014
-0.001

0.061
0.042
0.168
0.123
0.138
0.382
0.001
0.832
0.401
0.135
0.180
0.000
0.000
0.002
0.009
0.000
0.035
0.015
0.280
0.253
0.178
0.273
0.089
0.895

0.002

0.584

0.003

0.007

0.213

0.709

0.002

0.012

0.027

vol xlvii no 10

VAT Dummy Variable

0.014

0.868

0.014

Dummy: Years with

Bifurcated States
Coeff
P-Value

-0.004

0.497

0.013

0.008

0.007

0.079

63

SPECIAL ARTICLE
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.

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.
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%20Government%20Finances%20-%202010, accessed
on 3 January 2010.
(2011): State Finances: A Study of Budgets,
Revenue Receipts of States and Union Territories 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.
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=fiscal.htm#V.Special, accessed on 22 October 2011.
Stiglitz J and P Dasgupta (1971): Differential Taxation, Public Goods and Economic Efficiency,
Review of Economic Studies, 38, 151-74.
The Economic Times (2004): VAT: States to Get
Full Compensation, 23 September, available
at http://economictimes.indiatimes.com/vatstates-to-get-full-compensation/articleshow/
860867.cms, accessed 20 October 2011.
World Bank (2005): State Fiscal Reforms in India
(New Delhi: Macmillan).

(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.
MOSPI (2011): Statement: Gross State Domestic

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

State

R-Square

F
Significance

VAT Dummy Variable

P-Value

Dummy: for Only Bihar/

Uttarakhand VAT
Coeff
P-Value

Dummy: Years with

Bifurcated States
Coeff
P-Value

Coeff

Andhra Pradesh
Gujarat
Haryana
Karnataka
Kerala
Maharashtra
Orissa
Punjab
Rajasthan
West Bengal
Tamil Nadu
Arunachal Pradesh
Assam
Himachal Pradesh
Goa
Jammu and Kashmir
Manipur
Meghalaya
Mizoram
Nagaland
Sikkim
Tripura
New Delhi
Bihar+Jharkhand
Madhya Pradesh+
Chhattisgarh
Uttar Pradesh+
Uttarakhand

0.002
0.411
0.431
0.187
0.014
0.008
0.000
0.375
0.249
0.089
0.543
0.263
0.137
0.079
0.234
0.596
0.312
0.661
0.493
0.325
0.089
0.824
0.248
0.283

0.865
0.007
0.006
0.108
0.658
0.740
0.996
0.012
0.049
0.262
0.001
0.042
0.158
0.292
0.058
0.001
0.025
0.000
0.002
0.027
0.262
0.000
0.050
0.247

-0.003
0.236
0.142
-0.032
0.010
-0.006
0.000
0.056
0.067
-0.025
-0.039
0.082
0.047
0.034
0.065
0.166
0.215
0.138
0.168
0.098
0.021
0.177
-0.038
-0.021

0.865
0.007
0.006
0.108
0.658
0.74
0.996
0.012
0.049
0.262
0.001
0.042
0.158
0.292
0.058
0.001
0.025
0.000
0.002
0.027
0.262
0.000
0.050
0.473

-0.042

0.049

0.055

0.872

0.000

0.039

0.008

0.067

0.000

0.443

0.063

0.010

0.751

0.055

0.024

0.353

-0.003

0.915

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

State

Andhra Pradesh
Karnataka
Kerala
Maharashtra
Orissa
Punjab
West Bengal
Arunachal Pradesh
Assam
Himachal Pradesh
Goa
Jammu and Kashmir
Manipur
Meghalaya
Mizoram
Nagaland
Sikkim
Tripura
New Delhi

64

R-Square

0.993
0.997
0.998
0.991
0.987
0.972
0.995
0.819
0.977
0.995
0.991
0.988
0.730
0.993
0.975
0.977
0.984
0.991
0.935

F
VAT Slope Dummy
PreVAT
PostVAT
Significance
Variable
Coeff
P-Value Coeff P-Value Coeff P-Value

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.004
0.000
0.000
0.000
0.000
0.000
0.000

0.000
0.013
-0.004
-0.002
0.009
-0.001
0.012
0.005
0.010
0.028
0.020
0.027
0.038
0.021
0.031
0.021
-0.027
0.011
-0.016

0.963
0.000
0.082
0.609
0.202
0.892
0.000
0.960
0.294
0.000
0.001
0.009
0.438
0.004
0.174
0.068
0.061
0.126
0.328

-0.020
0.173
0.008
0.114
0.126
0.127
0.163
0.311
0.263
0.164
0.041
0.243
0.120
0.136
0.299
0.024
-0.208
0.035
-0.114

0.801
0.003
0.825
0.161
0.286
0.374
0.003
0.827
0.126
0.066
0.629
0.122
0.876
0.171
0.377
0.877
0.266
0.766
0.695

-0.031
-0.056
-0.036
0.033
0.077
0.207
-0.050
0.448
0.164
-0.102
-0.050
0.039
-0.269
0.039
0.099
0.008
0.022
-0.016
0.179

0.696
0.255
0.362
0.673
0.501
0.170
0.286
0.748
0.317
0.234
0.555
0.801
0.723
0.683
0.763
0.957
0.900
0.891
0.539

R-Square

F
Significance

Andhra Pradesh
0.318
Karnataka
0.085
Kerala
0.353
Maharashtra
0.327
Orissa
0.707
Punjab
0.466
West Bengal
0.160
Arunachal Pradesh 0.853
Assam
0.774
Himachal Pradesh 0.924
Goa
0.242
Jammu and Kashmir 0.855
Manipur
0.672
Meghalaya
0.883
Mizoram
0.847
Nagaland
0.784
Sikkim
0.464
Tripura
0.718
New Delhi
0.142
march 10, 2012

0.189
0.796
0.143
0.176
0.002
0.050
0.536
0.000
0.000
0.000
0.326
0.000
0.003
0.000
0.000
0.001
0.051
0.001
0.591

VAT Slope
Dummy
Coeff P-Value

0.005
0.002
0.004
0.001
0.008
0.004
0.001
0.015
0.015
0.015
-0.017
0.024
0.009
0.011
0.012
0.005
0.015
0.009
0.006

0.038
0.480
0.052
0.558
0.001
0.109
0.273
0.000
0.000
0.000
0.143
0.000
0.000
0.000
0.000
0.000
0.010
0.000
0.342

vol xlvii no 10

PreVAT
Coeff P-Value

0.002
0.003
0.005
0.006
0.005
0.007
0.002
0.006
0.015
0.006
-0.021
0.013
0.002
0.005
0.006
0.002
0.009
0.005
0.006

EPW

0.552
0.460
0.129
0.035
0.072
0.102
0.307
0.077
0.010
0.013
0.243
0.017
0.486
0.026
0.031
0.236
0.289
0.130
0.573

PostVAT
Coeff

P-Value

-0.003
-0.001
-0.004
0.001
0.001
0.006
0.000
-0.004
0.004
-0.005
0.002
-0.002
-0.005
-0.001
-0.002
0.000
-0.005
-0.002
0.004

0.441
0.817
0.303
0.859
0.659
0.206
0.888
0.202
0.527
0.062
0.898
0.695
0.146
0.617
0.422
0.824
0.537
0.560
0.702

Economic & Political Weekly