Publications

Featured Opinions

Budget Math: How Accurate Have the Centre’s Past Revenue Estimates Been?

The Wire

  • 31 January, 2019
  • Suraj Jaiswal


The Union Budget, in essence, gives details of how much the government plans to allocate for schemes and projects, and how much revenue it expects to collect in the ensuing financial year.

The total amount that government plans to spend is determined by the revenue that it estimates to raise.

Many a times, the actual revenue raised is different from the budget estimates, as a majority of the revenue comes from taxes and tax collections, which are affected by factors such as income growth, economic activities, international trade, etc – all of which have a degree of uncertainty and are difficult to predict with absolute accuracy.

Notwithstanding the difficulty, it is imperative that budget estimates are accurate to the highest degree possible.

This piece examines the accuracy of budget estimates for tax collection over the period of 2004-05 to 2016-17 and tries to answer how accurate these estimates have been. For the period when the actuals have turned out to be much different, it also tries to analyse the potential causes behind that. For this purpose, budget estimates (BE) is compared with revised estimates (RE) and actuals (AE) – the three figures the budget document gives in a given financial year.

The differences between three numbers are as their names suggest – BE refers to the original estimates, RE figures suggest the actual revenue or expenditure for the initial six months and the projected revenue and expenditure for the rest six months. Actuals are the real numbers for revenue and expenditure in terms of how much was collected and how much was spent.

By comparing AE and RE with BE, the accuracy or inaccuracy of the estimates given in the budget documents can be established.

The graph below presents the comparison of RE and AE with BE for the overall tax revenue. In the graph, BE has been normalised, i.e. – it is taken as the base and is plotted as the X-axis.

The lines depicting RE and AE show how different the RE and AE turned out in percentage terms of BE. The corresponding sign reflects if it was overestimation or underestimation. For example – in the year 2004-05, AE was 4% less than the BE, while for the year 2007-08, AE turned out to be 8.2% more than the BE.

Graph 1: Overall Tax Revenue

Graph 1: Overall Tax Revenue

In the period of analysis – 2004-05 to 2016-17 – AE turned out to be less than BE in nine out of 13 years, and it exceeded the BE only four times. The biggest deviation occurred in 2008-09 when AE turned out to be 12% less than BE. It can be argued that this large deviation was due to the global financial crisis where the tax revenue did not grow much. It was compounded by the fact that the growth rate of the last two years had led to BE expecting a much higher growth of tax collection.

India’s total tax revenue is made of various individual components, such as corporate income tax (CIT), personal income tax (PIT), excise duty (ED), customs duty (CD) and service tax (ST). Since all individual components are estimated separately, and each of them may be affected by different factors, it is important to view each one individually.

Corporate Income Tax – The actual numbers for CIT were less than BE 11 out of 13 times, the exceptions being 2006-07 and 2007-08. It seems that the high economic growth rate of those years have resulted in the AE turning out to much higher than BE. However, it is clear that the budget has been overestimating the corporate income tax collection, though in the last four years, the estimates seem to be getting more accurate.

Graph 2: Corporate Income Tax

Graph 2: Corporate Income Tax

Personal Income Tax – In case of personal income tax, the actuals have exceeded BE on five occasion and were below BE for eight years. The overall trend is highly random to be able to draw a definite conclusion, though few pointers can be highlighted.

Graph 3: Personal Income Tax.

Graph 3: Personal Income Tax.

The sharp deviation (-23%) in case of 2008-09 can be attributed to the global financial crisis. In 2008-09, the growth rate assumed (BE to BE) was also very high at 40%, and that also contributed to AE turning out less than BE. After the experience of 2008-09, the growth rate assumed (BE to BE) in 2009-10 estimation was actually negative– which is why AE was 17% more than BE.

Another pointer refers to the financial year 2016-17 when AE was 3.2% more than the BE, which can be explained by the one-time collection of around Rs 10,000 crore arising from the income disclosure schemes, where government provided amnesty to anyone disclosing previously undeclared income.

Also read: Interim Health Budget May Rise 5-11%, Primary Care to Get Boost

Excise Duty – For the period of study, AE for excise duty was greater than BE on three occasions, most notably during 2015-16 and 2016-17, when it exceeded BE by close to 20% in two consecutive years.

Graph 5: Excise Duty

Graph 4: Excise Duty

The reason for this is the increase in the excise duty in petrol and diesel when the global prices for crude oil fell to a record low, and that resulted in windfall gain for the government, and helped actual revenue from excise exceeding the BE. The overestimation for 2008-09 can be attributed to the global financial crisis.

Custom Duty – In case of custom duty, AE has been higher than BE in five out of the 13 years. Some of the outcomes are on expected lines, for example, AE exceeding BE in the high growth years from 2004-05 to 2007-08, then AE falling behind the BE in the immediate years after the global financial crisis.

Graph 5: Customs Duty

Graph 5: Customs Duty

One notable feature of custom revenue collection is that in the last four years, the growth of revenue estimated (BE to BE) has been on the relatively lower side (in single digits), and still the AE has fallen behind the BE.

Service Tax – The trends of AE vs BE is most random in the case of service tax.

Graph 6: Service Tax

Graph 6: Service Tax

While it also has some features common with other components, such as AE exceeding the BE in high growth years and falling behind after global financial crisis, the trends in remaining periods swing wildly on either side. So, AE exceeds BE by over 10% in 2011-12, and then it lags by more than 20% in 2014-15.

While budgets are traditionally analysed based on their allocations for schemes, new programmes, tax regulations and so on, mismatch of BE with RE and AE has received rather less attention.

However, since the actual revenue collection – most of which comes from taxes – can, in turn, affect the actual fund release for the programmes, it is imperative that more efforts be made to get the budget estimates as accurate as possible.