Abstract
Many have held that value-added tax (VAT) generates higher revenue which then leads to bigger government spending. They cited the experience of European countries claiming that both the tax rate and the government size have risen over the years. The present study finds that there are exceptions to it based on the state level data of India. The tax was introduced as a replacement for sales tax about a decade ago but an increase in government size was not observed. Instead, higher revenues generated helped in the reduction of budget deficits when introduced along with a number of austerity measures among which was the enactment of the fiscal responsibility law (FRL) placing quantitative restrictions on budget deficits. Some states did witness bigger spending but most of them were observed to be financially weaker states receiving more central transfers. It enabled them to spend more without running into deficits and, hence, meet FRL obligations.
Introduction
One of the characteristics of a good tax is its higher revenue generating capacity and the value-added tax (VAT) has been found to be an ideal one. It can generate large revenue as a result of which it is often called a money machine. But this has been also one of the reasons why the tax is yet to be introduced in the United States, the biggest economy of the world, for the fear that it would lead to bigger government spending. 1 Critics cited the experience of European countries claiming that both the tax rates and the government size have risen over the years. This apprehensive view is justified for those who believe that large government size is inefficient and, hence, should not be fed with a tax like VAT which is known for its revenue-generating capacity. 2
There are also studies which said that the claim of VAT causing larger government size cannot be taken as a general truth. More studies, hence, are necessary before any conclusion can be made. The current article adds to the available literature and is based on data emerging from India which introduced the VAT about a decade ago. It is divided into five sections. The second section narrates the ongoing debate on the relationship between government size and VAT, while the third section examines the existence of the association in India. The fourth section is statistical analysis while the last section concludes the article.
The Debate on the Relationship between VAT and Government Size
Much of the literature on the relationship between VAT and government size seems to be in the context of the USA where economists seem to be embroiled in a discussion for and against the introduction of the tax in the country. 3 The caution against VAT inflating government size was made by McLure (cited in Stockfisch, 1985, p. 547) when he said, ‘[I]f foreign experience is any guide, introduction of a VAT would facilitate growth of the relative size of the federal government, whether VAT was initially introduced to raise additional revenue or only as a substitute for existing taxes.’ Many comments as well as findings have been made since then supporting the positive relationship view. 4 In a paper entitled ‘Beware the Value-Added Tax’, Mitchell (2005b) illustrated how tax burden as a percentage of GDP in the European Union (EU) countries was larger by a small margin as compared to that of the USA in the pre-VAT period. But after the introduction of VAT in the EU countries, tax revenue increased rapidly as well as the size of the government. Holcombe (2010) estimated that among a group of 10 developed countries, Canada is the only country that witnessed a decline in the VAT rate since its introduction, while Denmark witnessed a growth by 177.8 per cent. Echoing the concern of VAT’s capacity to raise lots of revenue, Christian and Robbins (2011) called VAT an ideal choice for those who want government to ‘expand on a grand scale’.
The argument that the VAT leads to larger government spending has not been conclusive. Stockfisch (1985) compared the growth ratio 5 of 12 VAT countries before and after the implementation of the tax with that of 12 non-VAT countries all belonging to the group of Organisation of Economic Cooperation and Development (OECD) countries. It was observed to fall in a majority of both VAT and non-VAT countries in the second period and, hence, concluded that the view of VAT leading to a larger government spending did not hold. Similarly, Keen and Lockwood (2006) using data of a group of OECD countries found a weak statistical link between VAT and government size. On the statement that tax rates increased in EU countries, Bartlett (2011) divided these countries into two, that is, the countries which implemented VAT before 1975 and the other after 1975. He observed that the earlier was a period of high inflation and a number of countries had higher tax rates which were justified to tackle price rise. There was little evidence of the phenomenon in the later period which enjoyed price stability. Barro (2012) said that the tax rates in the EU countries are on the upper side because of the fact that they cannot be less than 15 per cent as per the EU rules. Therefore, the USA did not have to act similarly if it is in place as done by non-EU member countries like Japan, Canada and Australia which maintained VAT rates at reasonable levels.
From two totally different angles, the theory of positive relationship between VAT and government size has been criticized. Mikesell (2014, p. 18) believed that there are misconceptions about VAT and retail sales tax (RST) in the USA though both seek to tax consumption: ‘American and international views of RST and VAT diverge. Americans favor the RST but abhor the VAT and international observers praise the VAT while viewing the RST as archaic. Several misconceptions may shape these attitudes.’ One such misconception is that concerns for increasing VAT rates had been expressed but not for the increase in RST rates. After comparing increase in the average RST rates of 45 states with that of VAT rates in EU and OECD countries, he concluded that the increase in VAT rates was not more than that of RST rates. On the other hand, Gale and Harris (2011) said that there have been findings that show tax deductions followed by larger spending and vice versa. He claimed that rejecting VAT on the argument that it would increase government size is a ‘starve the beast’ concept. It is an idea held by the conservatives which say that government should cut taxes with the aim of reducing government spending. 6 One such person who advocated tax cuts is Milton Friedman (2003) who said: ‘How can we ever cut government down to size? I believe there is one and only one way: the way parents control spendthrift children, cutting their allowance. For government, that means cutting taxes.’ But many have doubted the validity of the theory. Niskanen (2006) argued that there are problems with the theory. First, reducing the price of the commodity will increase demand and cutting tax is like price control which will increase demand for public services. Second point is that federal spending and receipts during 1981 and 2005 moved in the opposite direction which again proves the theory to be untrue in reality. Third, the belief in the theory has reduced the commitment necessary to cut expenditure. Similar view was held by Bartlett (2010) and said: ‘Starve the beast was a theory that seemed plausible when it was first formulated. But more than 30 years later it must be pronounced a total failure … there is growing evidence that its impact has been perverse—raising spending and making deficits worse.’
VAT Revenue, State Level Government Size and Fiscal Responsibility Rule
Sales Tax, VAT and GST in India
Sale tax is a tax levied on the sale of goods. It is simple to administer as it is calculated by multiplying the value of the good with the tax rate. But it also leads to cascading effect by taxing the good again and again resulting in price rise as set off for tax paid at earlier stage is not given. For this reason, sales tax was replaced by state VAT in the year 2005. Soon GST or goods and services tax will be introduced which will replace or subsume CENVAT, state VAT and a number of related taxes levied both by the states and the centre. 7 It is going to be levied in the entire production and distribution chain on the value added at each stage of the process. 8 It is a dual GST consisting of central GST and state GST and they will be sharing proceeds of the tax.
Statements to be Examined
Three statements can be made and examined about its relevance in India after VAT has been introduced replacing state sales tax. The first is that substitution of sales tax by VAT results in higher revenue generation. The second statement is that the VAT rate tends to increase over time and the last one is that higher tax revenue causes larger government spending. Each of them shall be examined one after another.
Increase in VAT Revenue and Tax Rates
India introduced the state level VAT in 2005 replacing state sales tax and has proved to be a good source of revenue (Nepram, 2011; Reserve Bank of India (RBI), 2009). In fact, aggregate VAT revenue of the states rose from 3.09 per cent of the GDP in 2005–2006 to 4 per cent in the year 2013–2014 despite a slump in 2009–2010 which is a remarkable achievement in a short span of time (Table 1). State sales tax was contributing only 49 per cent of the GDP in 2004–2005 but VAT now generates about 59 per cent of own tax receipts of the states.
Composition of Revenue Receipts of the States (% of the GDP)
An important reason why VAT yields more revenue than sales tax is because of its inbuilt mechanism for higher tax compliance. 9 VAT is generally levied using the tax credit method which requires documentation of every transaction, thus, increasing tax compliance. As for the changes in VAT revenue over time, Keen (2013) said that it can be broadly attributed to the changes of three factors, namely the consumption, tax rate and C-efficiency ratio which he called them as drivers of VAT revenue. Of the three factors, he observed that increase in revenue mainly comes from increase in the C-efficiency ratio. 10
It can be seen from Table 2 that consumption as a percentage of GDP in the country witnessed an average annual decline of 0.19 per cent over the period 2005–2013, so this cannot be the driver of tax growth. As far as the tax rate is concerned, many states have increased both the lower and standard rates from the initially planned 4 per cent and 12.5 per cent. This was the issue that was repeatedly raised by opponents of VAT and discussed earlier. For example, Chhattisgarh government raised the reduced VAT rate from 4 per cent to 5 per cent and the standard rate from 12.5 per cent to 14 per cent as early as 2010 and the reason given was to recoup revenue loss due to recession (Business Standard, 2009). Andhra Pradesh government raised the 4 per cent floor rate to 5 per cent in 2011 and said it was done under the advice of the central government to compensate for the loss incurred in reducing the central sales tax from 4 per cent to 2 per cent (Business Standard, 2011). The West Bengal budget for the fiscal year 2013–2014 saw the raising of the two VAT rates from 4 per cent to 5 per cent and from 13.50 per cent to 14.50 per cent for building infrastructure (The Hindu, 2013). On the other hand, the increase in VAT rate from 5 per cent and 14 per cent to 5.5 per cent and 14.5 per cent for Karnataka government in the budget for 2012–2013 for one year only was retained in the subsequent year’s budget to recover money for waiving crops loans to farmers who suffered from drought (Business Standard, 2013).
Tax Performance
*Efficiency ratio is the ratio of VAT collections to GDP and divided by standard tax rate. **Estimated using consumption expenditure in place of GDP. *** Estimated using private consumption expenditure in place of GDP.
The efficiency ratio of VAT has been defined by Ebrill, Keen, Bodin, and Summers (2001) as the ratio of tax collections to GDP and divided by the standard rate. It is given in percentage by multiplying the result by 100. It simply tells what percentage of GDP is collected by each percentage point of the standard tax rate (Sopak, 2012). On the other hand, the C-efficiency ratio is the ratio which uses consumption in place of GDP. Private consumption is also sometimes used in place of consumption in estimating the ratio. The estimation of these ratios is complicated in case of India as there are two VAT rates and there is lack of uniformity in the rates adopted by the states. But a modest attempt is made here using the standard rate of 12.5 per cent till 2009–2010 and 13.5 per cent thereafter as almost all the states in the country have raised their VAT rates.
In the results given in Table 2, the ratios are below 50 per cent which is understandable because of the presence of exempted goods, lower tax rate and possible lack of tax compliance. 11 It is also seen that the two ratios gradually increased despite a slight decline in the year 2009–2010 and 2010–2011 which tells that VAT performance increases with the passage of time. The probable reason why C-efficiency ratio has increased over time in our context is the increase of the literacy rate and the age or the period of VAT in place as administration may become more efficient with experience. 12
Decline in Government Size
In spite of the increase in tax revenue, aggregate expenditure comprising of revenue and capital expenditures declined from a peak of 19.59 per cent of the GDP in 2003–2004 to 16.57 per cent in 2005–2006 and further to 15.98 per cent in 2010–2011 (Figure 1). Since then government expenditure has increased only marginally. Thus, it can be concluded that higher VAT revenue in general has not led to an increase in public expenditure. On the other hand, an increase in tax revenue along with a decline in government size has made revenue and fiscal deficits in a much better position except for the years 2008–2010 and 2013–2014 (Figure 2). Increase in deficits in the first period was unavoidable due to hike in pay expenditure following the Sixth Pay Commission recommendations though with much lower severity than before, while in the second period the economy suffered a slowdown resulting in lower tax revenues. 13
Two possible reasons can be specified on why government spending declined despite of a strong performance from VAT. The first possible reason was a reduction in the revenue from the other components of sales tax. Sales tax is composed of state sales tax/VAT, central sales tax, surcharge on sales tax, receipts of turnover tax, etc. The White Paper on State Level Value Added Tax that came out when state VAT was implemented mentioned that the government will abolish a number of related taxes while the central sales tax would be phased out. As a result, the other components of sales tax which contributed 0.74 per cent of the GDP in 2004–2005 declined to 0.33 per cent in 2013–2014 (Table 1). Though VAT receipts were able to make up for the loss of revenue, the importance of sales tax inclusive of all the related taxes measured by its contribution to GDP increased from 3.75 per cent in 2004–2005 to only 4.33 per cent in 2013–2014.
A major reason for the decline, however, is that VAT implementation took place at a time when both the centre and the states were trying to contain public expenditure. It was in response to sharp deterioration in government finances that emerged after the implementation of the Fifth Pay Commission recommendations in the late 1990s. Revenue deficit increased from 0.77 per cent to 2.94 per cent of the GDP during 1995–1996 and 1999–2000 while fiscal deficit rose from 2.76 per cent to 4.85 per cent (Figure 2) which prompted the Twelfth Finance Commission to say that the six years from 1997–1998 to 2002–2003 ‘have been the worst in the history of state finances’ (Government of India, 2004, p. 37). Fiscal reforms were introduced and the states undertook a number of austerity measures one of which was the enactment of a fiscal responsibility law (FRL) in the year 2005 called the Fiscal Responsibility and Budget Management Act (FRBMA). It aimed at reducing revenue deficit to zero and fiscal deficit to 3 per cent of the respective states incomes. As a result of the measures, the proportion of committed expenditures comprising salary, pension and interest which was 52.81 per cent of the total revenue expenditure in 2000–2001 declined to 46.8 per cent in 2011–2012 leading to an increase in the quality of public spending of the states (Table 3). Expenditure on pension continued to rise though it is expected to decline in future as the states have implemented new pension schemes which are contributory in nature.
State Level Expenditure on Salary, Interest and Pension (₹crores)
There has been a sharp fall in capital expenditure as well since the year 2005–2006 onwards even though there has been a desirable increase in capital outlay. Discharge of internal debt fell from 2.15 per cent of the GDP in 2003–2004 to 0.69 per cent in 2005–2006. The drop in case of repayment of central loans was from 1.20 per cent of the GDP in 2004–2005 to 0.24 per cent in 2005–2006. Overall, the decline was from 5.40 per cent of the GDP in 2003–2004 to 3.65 per cent in 2005–2006. Debt reconsolidation and debt writes off along with recommendations not to make plan transfers in the form of loans enabled improvement of debt position to a large extent. 14
The decline in government size as well as the increase in VAT revenue was not common to all the states. For this, estimation is made which compares the increase in government size with that of the increase in VAT revenue by comparing the averages of the pre-VAT and VAT period figures (Figure 3). It is seen that only three states witnessed a decline in tax revenue while the remaining states gained. On the other hand, 16 of the 26 states under consideration witnessed a decline in government size, while 10 states the majority of which are small states experienced higher expenditure. 15 Thus, it can be concluded that with the exception of some states there has been a reduction in government size after the introduction of VAT.
The states which witnessed higher spending can be grouped into two categories, namely special category states (SCS) and general category states (GCS). The SCS are Arunachal Pradesh, Jammu & Kashmir, Manipur, Meghalaya, Mizoram and Nagaland. Government sector plays an important role in these states due to low presence of private sector or public–private sector partnership in infrastructure projects (RBI, 2010) and as result the increase in government size of them is not unexpected. They are financially weak and the increase in spending for most of them has been largely financed by central transfers and to a small extent by the new tax. A larger central transfer makes states to spend more without incurring into higher deficits and this explains to some extent high government size observed in small states. As for the remaining four GCS also the government size increased (Figure 4) but in the later part of the study period they were, with the exception of Chhattisgarh, within the targets set by the FRBMA which as stated earlier is to reduce revenue account deficit to zero and limit fiscal deficit to 3 per cent of the respective GSDPs (Table 4).
Revenue and Fiscal Deficits of Selected States (% of GSDP)
Empirical Analysis
A panel data regression analysis is used to check whether the introduction of VAT leads to larger government expenditure. The question that emerges is about the methodology to be used. A good choice for examining causality relations is the panel Granger test, though in our case its use is unsuitable as our main intention is to examine whether the replacement of state sales tax with VAT leads to larger government size. Further, our time period is short and two big states, namely Tamil Nadu and UP implemented VAT as late as 2007 and 2008, respectively. A reasonable choice is to adopt a simple panel regression with VAT dummy which will take the value 1 when the new tax is in place and 0 when it is yet to be implemented. In short, it is a time dummy and a negative coefficient would tell a reduction in government size when the new tax is in place. This would help us to validate findings in Figure 3 as well. For this purpose, government size (GOVSIZE) is taken as the dependent variable which is total government expenditure consisting of revenue and capital expenditure divided by the net state domestic product (NSDP). The independent variables chosen are VAT and FRBMA dummies, central transfers, state size, urbanization and pension expenditures. An important observation made is that the years of VAT implementation and the enactment of FRL have been quite similar with a high correlation coefficient. The use of FRBMA dummy, therefore, has been avoided as it would likely give incorrect estimates. Per capita central transfers have been obtained by dividing total central transfers inclusive of tax transfers and grants with population figures. For this purpose, the population of the years 2001 and 2011 are taken to correspond to the fiscal years 2001–2002 and 2011–2012, respectively, and the inter census figures are interpolated. Population can also be a major determinant of government size but its use has been avoided as per capita central transfers have just been estimated by dividing central transfers by its figures. The NSDP is included to measure the size of economic activity. Urbanization is included to check whether more urbanized states have higher government expenditure. Pension expenditure as a percentage of revenue expenditure has also been included to capture the impact of committed expenditure on government size. The equation under consideration is as follows:
The expected signs of the independent variables can also be briefly mentioned here. The VAT dummy variable as stated earlier will have a negative sign if there is a decline in government size in the periods when the new tax is in place. A positive sign is expected as larger tax revenue would increase government spending, but a larger impact of austerity measures like FRL could reduce its size. Dependence on central transfers is likely to have a positive sign as larger government transfer is likely to increase spending without incurring larger deficits. A state with higher economic activity may increase the size of the government, while an increasing urban population may also increase government spending in order to provide the necessary infrastructure. Pension expenditure is likely to increase government size as well.
The dataset consists of 26 states for the period 2001–2002 and 2011–2012 which is 11 years. 16 The data related to state finances and gross domestic product have been obtained from various Reserve Bank of India publications like Handbook of Statistics on State Government Finances, State Finances: A Study of the State Budgets and Handbook of Statistics on Indian Economy. The estimations are made using STATA 11. 17
First, the data are pooled together and estimated results are given in first column of Table 5. However, the model does not take into account the state-wise heterogeneity and, hence, random effect and fixed effect estimates are made, the results of which are given in columns second and third, respectively. The random effects model has been chosen as a better model against the pooled model by the Breusch–Pagan LM test. Similarly, it has been also selected over the fixed effects model as the value of chi2 in the Hausman test has been found to be insignificant. We cannot, however, take it as the result as our panel data have problems of serial correlation, heteroscedasticity and contemporaneous correlation or cross-sectional dependence. In such cases, it is suggested to use the fixed effects method using Driscoll-Kraay standard errors and written as STATA programme by Hoechle (2007). Estimation using the panel corrected standard error method that was suggested by Beck and Katz (1995) is also given in the fifth and sixth columns without and with state effects.
The Impact of VAT on Government Size (GSIZE)
It is observed in majority of the results that the coefficient of VAT dummy variable is not significant even at 10 per cent level of significance which clearly makes us hard to believe that VAT leads to larger government size. As for the other variables, the results are as we expected. Central transfer is found to increase government size, while pension expenditure is observed to have a positive effect as committed expenditures increase along with the revision of every pay commission recommendations. The size of state income is inversely related to the dependent variable. This is possible because NSDP can be used as a fair representative of the population size of states also as larger states have larger NSDPs and vice versa with high correlation. The negative coefficient is because bigger states enjoy economies of scale and, hence, have smaller government size. The coefficient of urbanization has been found to be negative in the fixed effects model.
Conclusion
In this short but eventful period after the introduction of the state level VAT in the country, two important observations have been made. The first one is that the tax indeed has proved to be a money machine and the performance of the tax has improved over time. The second important finding is that it has not led to an increase in government size although there is an increase in tax rates. There is, however, a caveat. The tax was introduced at a difficult time when major fiscal reform measures were being introduced in the country to curtail public expenditure. If the tax was introduced in a normal time, results could have gone in a different way. But one thing is certain that VAT can be used for reducing budget deficit when introduced along with measures restraining public expenditure.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The author received no financial support for the research, authorship and/or publication of this article.
Footnotes
Acknowledgements
I thank an anonymous reviewer for comments and suggestions on two earlier drafts. The remaining errors are mine.