Using Tax Data To Fine-Tune India's Economic Policy For More Equitable Growth

A sustained high growth, other than productivity and efficiency improvements, given emerging employment, income, and wealth patterns, will necessarily require policies to redistribute incomes, to provide the consumption levels that will be critical for sustained growth. This is the crux of the current debate.

Much of this debate uses the multiple data on consumption that are presently available, although the privately generated data is priced and not available for public validation and testing.

One supplemental, albeit highly aggregated, series is income tax, which provides actual, measured, income data, which can then be used as an anchor to validate various other survey data, many of which have been shown to have various limitations, and especially, to be non-comparable due to their own specific sampling and other methodologies.

A problem with income tax data is the large variation in the formats of the presentation of data. Data releases on the official Income Tax site provide data in multiple formats, with differing dates (some (Assessment Year 2019 (AY19), which corresponds to the Financial Year 2018 (FY18), some AY21, some AY23, etc.), categories (income tax filers, taxpayers, individuals vs business, etc.), and many more. There is little standardisation in reporting. The numbers of returns, taxpayers, gross incomes, etc., differ across the sources of data, yet are broadly congruent, which, we feel, makes trends and ratios from the separate data series broadly comparable.

One specific caveat. We use a distribution of personal income tax filers between salaried and business professionals for FY2020-21, located from a private source. Unfortunately, this was the desperate lock-downed COVID year, where businesses, especially small, were subject to extreme stress, and which is likely to have severely disrupted operations and hence earnings for small business categories. Hence, FY21 returns are likely to be large outliers. Given this caveat, treat the imbalances in the following narrative with some caution, that the inferences are likely to be exaggerated.

Given these reservations, the following observations relate to the particular debate on income distribution (which is loosely termed as inequality) and hence to policies that, if necessary, might help to ameliorate this.

Taxpayer Data Sets

Let’s start with the taxpayer data. Individual (including HUF or Hindu Undivided Family submissions) tax returns filed have grown at a compound annual growth rate of 9.5 per cent over the 9-year period FY14 to FY23, which is more than the 5.9 per cent growth for corporate tax filers. It is difficult to comment on the differential rates of growth, although it is understandable that the number of businesses (however micro and small) will find it difficult to keep up with their employees, vendors and shareholder
beneficiaries. It is true, however, that there is an indication (as we will see later) that small businesses are not generating significant revenues and profits.

The next question is the ratio of tax return filers to actual taxpayers, as a proxy for collection efficiency. A priori, one would have expected given the enormous data analytics which the income tax department has invested, that tax evasion will have been significantly reduced. Prima facie, the data do point to this. The ratio of tax returns filers to actual taxpayers has fallen from 1.65 times in FY15 to 1.27 in FY23. However, this shows that there still exist 30 per cent more filers than payers.

The number of taxpayers includes collections from Tax Deduction at Source (TDS), which indicates incomes earned, but which have not been reported. How much of this discrepancy is due to income reporters who fall below the payable income tax slabs versus those who fail to report incomes is not clear. Salaried income taxpayers are unlikely to be in this category since their deductions are (likely to be mostly) correct.

Given this ambiguity, we turn to other measures of validation of income tax collection. One, there is dated data available till AY2019 (FY2013 to FY2018, which is quite dated) on taxpayer filings as well as gross incomes, arranged by income slabs. This data is not (publicly) available for later years (to the best of our knowledge). Note that there have been large changes in the way incomes are tracked and MIS (Management Information System) on these generated, and hence, the reported incomes are likely to have become significantly more accurate.

Tax Filing Growth Versus Income Growth

Interpreting this comparison of the tax filers and the incomes throws up some fascinating and counter-intuitive inferences. As of FY18, total taxpayers were 5.9 crores (up from 3 crores in FY12), with a 6-year Compound Annual Growth Rate (CAGR) of 11 per cent. Yet, total incomes reported in the same period were up by 16 per cent. The story of this gap is within the income slabs that make up these aggregates.

In FY18, the share of tax filers in the income slab less than Rs 10 lakhs was 90 per cent, and those in the slab Rs 10 – 50 lakhs was 9 per cent. In contrast, the share of incomes reported in these two slabs were, respectively, 42 per cent and 18 per cent. Hence the top 1 per cent of tax filers by income reported nearly 40 per cent of total incomes.

Yet, going beyond this static FY18 snapshot, a more encouraging pattern is emerging. An equalizing trend had emerged in both metrics. Within the number of filers, a noticeable shift in the 6 years emerged between the sub-Rs 10 lakhs income bracket (94 per cent to 90 per cent) even as the share of filers in the Rs 10-50 lakhs bracket rose from 5 per cent to 9 per cent, a clear shift to a higher income bracket. Yet, the shift in reported incomes between the slabs was more skewed to a more unequal distribution; while the share of the lowest bracket rose from 40 to 42 per cent, the Rs 10-50 lakhs slab share rose from 13 per cent to 18 per cent. Most remarkably, the share of the taxpayers who reported incomes above Rs 1 crore (the richest) fell from 44 per cent in FY12 to 36 per cent in FY18. Hence, income distribution shifted up most significantly in the income slab of Rs 10-50 lakhs, which (do remember) was 9 per cent of tax filers.

Salaried Taxpayers Outnumber Business Professionals

Second, is a split in personal income taxpayers between personal income tax filers and professionals/businesses. Amongst personal income tax filers, salaried taxpayers far outnumber business professionals, except at the lowest income band of Rs 0-5 lakhs, where they outnumber the former almost 2:1. The Rs 0-5 lakhs income is gross, before deductions. These professionals and businesses are micro proprietorships, single-employee businesses that are just about breaking even and surviving. It is very unlikely that they are paying anything but negligible taxes.

The following rather arcane details are however important to get a sense of the nature of taxpayers, and hence, implicitly, of incomes. The taxpayers count reverses from the Rs 5 lakhs income bracket upwards, where the salaried taxpayers outnumber businesses, the latter are just half of the former in the Rs 5-10 lakh bracket. Note that a Rs 10 lakh gross income translates to Rs 83,000 a month. In the income bracket of Rs 10 – 50 lakhs, business taxpayers are less than a fifth of salaried. Overall, with incomes over Rs 1 crore, business tax filers are just a third of salaried.

In conclusion, income tax data can provide a validation of other consumer surveys on purchasing power, income distribution, source of income, and much else. While there is, expectedly, an overwhelming skew in incomes across income slabs, the heartening finding is that this had reduced over FY12-FY18. With the more limited data post that year, particularly with the onslaught of the COVID lockdowns and disruptions, there was a worry that the income generation capacity of lower-income slabs would have been disproportionately impacted, but the initial data releases suggest that this was not the case, and the income transition to higher income slabs has continued.