Understanding GST revenue performance

Contrary to belief, GST underperformed the old tax regime in its initial years but has now begun to exceed expectations, six years after implementation

The goods and services tax (GST) has been one of the major policy successes of recent years. It has stimulated the growth of a truly national market, replacing a welter of distorting taxes with a single rational system. It has also provided a stable source of revenue to both the Centre and states, notwithstanding some teething issues and discord between the two levels of government. But how much has the government actually received from the GST? Is it really an improvement on the pre-GST regime? The answers are surprising. 

Nearly all of the analysis of GST, whether in reporting by journalists or research by economists, is based on the data from the GSTN portal. In both the portal and the analysis, including some excellent recent academic studies, the GST figure reported for 2022-23 is Rs 18.1 trillion, equivalent to 6.6 per cent of gross domestic product or GDP.
 

But this figure actually measures GST collections, and collections are not the same as government revenue. That’s because some of the collections are refunded to taxpayers. We do not have the refunds data but we can calculate the difference between collections as reported by GSTN and net revenues after refunds accruing to the government. Figure 1 suggests that in 2022-23, the last full year of data, they amounted to no less than Rs 1.8 trillion. As a result, government GST revenues (both Centre and state, including the cess) amounted to Rs 16.1 trillion, or 5.9 per cent of GDP, about 0.7 per cent of GDP less than what the headline number suggests. It also shows that last year’s difference was not a fluke. Ever since the GST was implemented, this gap has hovered around 0.6-0.7 per cent of GDP.

 

What are these refunds? GST refunds are different from those on individual and corporate income taxes. With income taxes, there are advance tax payments, which require ex post reconciliation and hence refunds. But in the case of GST, taxes are paid on actual revenues, net of the taxes actually paid on inputs. So, in most cases there is no need for refunds.  
 

There is one major exception. In GST, exports are zero-rated, which means that exporters don’t pay taxes on their output but are entitled to refunds on the taxes they paid on their inputs. This constitutes the major chunk of refunds under the GST. (Recall that zero-rated items are different from exempt items, which are not taxed but which are also not entitled to input tax credits). 
 

At the same time, we can infer that refunds are related to the Integrated GST (IGST) because we can compare IGST collections and IGST settlements (the amounts actually distributed as revenue to the Centre and the states). The red line in Figure 2 shows that the discrepancy between IGST collections and IGST settlements is remarkably close to the gap between overall collections and revenues shown in Figure 1. 
 

Putting the pieces together, it seems that the gap between collections and revenues reflects refunds paid to exporters to reimburse them for the IGST they paid. We can hypothesise, but cannot be sure, that this IGST relates mainly to imports used in production, rather than domestic goods purchased from other states. (We assume this because other data suggests that exports often rely heavily on imported parts, notably in the case of iPhones.) In other words, the larger share of refunds is to compensate exporters for the IGST paid on their imported inputs.
 

The large amount of refunds has important implications for how one views GST performance. Going back to Figure 1, gross collection data gives the impression that the GST regime immediately overtook the pre-GST average (the dotted line as calculated by Professor S Mukherjee in a recent NIPFP paper), then dipped during the pandemic period, and once again surpassed the pre-GST regime in 2021-22. But budgetary revenue figures tell a very different story. They reveal that GST revenue overtook the pre-GST regime only in the current fiscal year, six years after the tax was first implemented.  
 

If refunds make such a material difference to the GST story, why have they been ignored? Perhaps because the GSTN does not publish the refunds data. As a result, we can only infer the numbers by comparing the gross collections reported by the GSTN with the revenues in the Central and state government Budgets and by comparing IGST collections with settlements. 

Is the fact that GST revenues have only recently surpassed pre-GST levels an indictment of the GST? Not at all. Revenues have declined for two reasons. The first reason is that export refunds have become much smoother, quicker, and fuller with the GST than they were under the previous regime. This is surely a good thing. Less welcome, however, there was a rate-cutting spree in the years leading up to the pandemic, which reduced the weighted average collection rate from 14.4 per cent in May 2017 to 11.6 per cent in September 2019, according to the Reserve Bank of India. 
 

As a result, we now need to focus on further boosting GST collections. At this point, there is little room to strengthen administration without risking over-zealousness and unnerving taxpayers; the real need is to address the remaining major design flaws. Specifically, the rate cuts of 2018 and 2019 need to be reversed, even if not fully, as part of a rationalisation of the overall rate structure. At the same time, the rate structure needs to be simplified, as the current complexity, especially for the cesses, is probably dragging down revenue collections and is certainly complicating enforcement. 
 

The most natural change would be to move to a three-rate structure (recommended by the Revenue Neutral Rate (RNR) Committee in 2015) with a standard rate of 18 per cent, a lower rate of say 10 per cent, and a demerit rate of 40 per cent. And if GST compensation is a thing of the past, the now-redundant cesses should be incorporated into the normal rate structure at the top rate of 40 per cent. Not only would such a move simplify the system, it would also eliminate a major distortion whereby a significant source of revenues has been walled off from the standard divisible pool of taxes that provides resources for both the Centre and the states.
 

Three conclusions follow: The GST did less well in the initial years than we have thought, but is now doing much better than we could have hoped for. It could do even better if it were redesigned in the direction of simplicity. And our appreciation and understanding of this historic reform would improve if the refunds data (and their composition) were to be published.  

The writers are, respectively, senior fellow, Peterson Institute for International Economics;  principal, JH Consulting; and program associates at the Center for Effective Governance of Indian States (CEGIS).  Ankit Chatri and Devashish Deshpande of CEGIS also contributed to this piece