We shall work with the State governments to usher in the GST in an appropriate timeframe.

The 2014 manifesto's commitment on GST is a single line, and its most important word is the qualifier: "We shall work with the State governments to usher in the GST in an appropriate timeframe." Not a date, and not a unilateral act — a commitment to bring the states along.

Both halves happened. The Constitution (One Hundred and First Amendment) Act, which created the concurrent power to tax goods and services and the GST Council on which the Union and every state sits, was passed by Parliament in 2016 and ratified by a majority of state legislatures. GST came into force on 1 July 2017, replacing a stack of central and state indirect taxes — excise, service tax, VAT, octroi and others — with one tax and one interstate mechanism. Whatever one thinks of its rate structure, the reform itself was delivered, and delivered through the states.

The promise said nothing about revenue, rate slabs or compliance cost, so this fact-check does not grade those. What can be checked is whether the machinery built to make the new tax work actually works, and there the CAG has audited one of its central pieces.

The e-way bill — the electronic document that must accompany goods in transit — was introduced on 1 April 2018 precisely so that movement of goods could be matched against tax paid. The CAG's performance audit of it, tabled on 12 August 2025, took a risk-based sample of 2,244 bills belonging to 956 taxpayers, and the matching failed repeatedly. "103 taxpayers… who had effected outward supplies for ₹2,285.23 crore from April 2018 to March 2022 supported by EWBs either did not file their return or did not report the turnover in their returns", involving ₹307.37 crore of tax. Another 174 moved ₹1,750.87 crore of goods "without discharging tax liability of ₹168.06 crore". Seventy-one taxpayers declared nil turnover while their own e-way bills showed ₹340.42 crore of supplies.

Some of it the audit calls deliberate: 18 taxpayers in three commissionerates "had generated 3,137 EWBs for outward supplies of ₹168.21 crore" and paid none of the ₹81.11 crore due — a paragraph headed "Wilful suppression of turnover".

Two findings show the system missing things it was built to catch. Forty-three taxpayers kept generating e-way bills "after the effective date of cancellation of their registration", for ₹152.60 crore of supplies. And unregistered persons generated 3,585 bills each above the ₹40 lakh threshold at which registration is compulsory — a state of affairs that a validation rule would end, and the audit recommends exactly that.

Enforcement on the ground was not using what it had. The National Informatics Centre produces 97 analytical reports on e-way bill traffic for tax officers; of 58 commissionerates audited, 46 "stated that they were not using the reports for the purpose of planning of interception of vehicles". And where vehicles were intercepted, the case did not always follow: in 200 of 1,405 booked cases "vehicles were released without creating demand of ₹2.60 crore".

One limitation deserves noting because it constrains the audit's own numbers. For 108 of the 956 taxpayers whose records were called for, nothing was produced; for 334 more, only part. Thirteen commissionerates did not produce the files for 153 booked cases. The figures above are therefore a floor on what the sample contained, not a ceiling.

The promise was to bring in GST with the states, in an appropriate timeframe. That was done, and our verdict is accurate. The audit belongs in the same entry because a tax is only as unified as its enforcement, and on the piece of enforcement that has been audited, the record is uneven.