Economy pulse: methods
Page: /macro/pulse/.
Code: pipeline/tipsheet/compute/pulse.py (the indicator list, readings and colours), publish/pulse.py, step steps/macro_zz_pulse.py; heat map apps/site/src/components/dense/PulseHeat.astro.
Tests: pipeline/tests/test_pulse.py.
Bundles: pulse/heat, pulse/s/<indicator> (monthly), pulse/d/<indicator> (daily, where the source is daily).
Build record: docs/status/pulse.md.
What the page is for
GDP arrives two months after a quarter ends and is revised later. Payments, tax collections, fuel sales, power demand, vehicle registrations, toll traffic, freight and credit arrive within days or weeks. Read together, they tell a reader how the economy is doing this month. The page puts them on one heat map and gives each its own chart with its full history.
Sources
Every series is read from its publisher, through the Data bank, and none comes from a data vendor. The indicator list, with each series’ source, is in INDICATORS in compute/pulse.py; the page’s last chapter prints the source and latest month of each.
| Group | Indicators | Publisher (Data bank dataset) | Starts |
|---|---|---|---|
| Payments | UPI value (monthly and daily), IMPS value | NPCI’s product statistics (npci_product_statistics, npci_daily_statistics) | UPI April 2016 (daily May 2021), IMPS September 2013 |
| Payments | UPI payments to merchants (person to merchant) | NPCI’s UPI ecosystem statistics (npci_upi_ecosystem) | April 2020 |
| Payments | Credit and debit card spending, cash withdrawn at ATMs | RBI, Payment System Indicators (RBI Bulletin Table 45, DBIE report 61) | November 2019 (RBI’s current table; its 2004-2019 table defines cards differently and is not joined on) |
| Travel and transport | FASTag toll trips | NPCI (NETC): toll and parking transactions plus annual-pass trips from August 2025 | December 2016 |
| Travel and transport | Petrol, diesel and jet fuel consumption | PPAC (ppac_consumption); PPAC’s earlier history is behind a sign-in and is not used | April 2020 |
| Travel and transport | Domestic air passengers | DGCA, traffic of Indian carriers, scheduled domestic (dgca_domestic_traffic) | January 2009 |
| Travel and transport | Cargo at the twelve major ports | Ministry of Ports, Shipping and Waterways, monthly report (major_port_traffic); the newest report’s figure for each month | April 2023 |
| Spending | Two-wheeler, car and four-wheeler, and tractor registrations | MoRTH’s Vahan analytics dashboard (vahan_analytics_monthly), all registrations including vehicles later scrapped, national, complete months only; each group is its own series (Vahan’s groups overlap and are never added up) | January 2003 (registrations before about 2012 are incomplete: states moved onto Vahan over the decade) |
| Travel and transport | Goods vehicle registrations | Same | January 2003 |
| Industry and energy | Peak power demand met (average of daily peaks), electricity supplied | Grid-India, daily power supply position report (grid_india_psp_daily); a month counts only when every day’s report is present. Grid-India posted no report for 19 February 2023, 21 March 2023, 30 and 31 December 2024 and 10 November 2025, so those months are blank, and so is the growth of the same months a year later | January 2023 (earlier reports are PDF only) |
| Taxes | GST collections (CGST + SGST + IGST, gross, cess left out), e-way bills | GST Network statistics (gst_state_collections, gst_eway_bills) | August 2017; e-way bills July 2018 (October 2021 left out: defective at source) |
| Industry and energy | Core industries, cement, steel | Office of the Economic Adviser (core_industries_index); growth within each base year (2004-05, 2011-12, 2022-23), the newer base where two overlap | April 2005 |
| Industry and energy | Industrial production, manufacturing | MoSPI, IIP, base years linked as on /macro/ | 1995 (IIP), 2013 (manufacturing) |
| Trade | Merchandise exports and imports, US dollars | RBI, India’s foreign trade (DBIE report 208); RBI’s final figure where it exists, else the provisional one | April 1990 |
| Trade | Crude oil imports, tonnes | PPAC (ppac_import_export) | April 2020 |
| Credit | Bank credit and deposits of scheduled commercial banks | RBI, Section 42 returns (DBIE), as computed for /macro/credit/ | 1998 |
Looked at and not used: Indian Railways’ monthly freight loading (no table or file after September 2023, only press releases), the Ministry of Coal (its site’s security certificate had expired), the Airports Authority of India (its traffic files sit behind a name, e-mail and OTP form), the Ministry of Civil Aviation’s daily traffic (refused our requests, and shows only the latest day). PMI is licensed by S&P Global and is not used.
The reading in each cell
- For an amount (rupees, dollars, counts, tonnes): growth on the same month a year earlier, in per cent. Comparing with the same month removes the seasons and the number of days in the month (except February in leap years). Blank when that month or the same month a year earlier is missing or incomplete at source.
- For a series the publisher already states as growth (IIP, bank credit and deposits): that rate as published.
- Bank credit and deposits are fortnightly (twice a month from December 2025); a month takes its last reporting date. From July 2023 to August 2025 they use RBI’s series adjusted for HDFC’s merger into HDFC Bank, which would otherwise lift growth for a year: HDFC Ltd’s loans and deposits joined the banks’ totals in one step.
- Daily series are summed to months (flows) or averaged (peaks and rates). A month counts only when every day is present.
- So far this month (daily series only): the days of the current month to the latest day, against the same calendar days a year earlier. It is shown only when no day is missing on either side.
The colour
A cell’s colour compares its reading with the same indicator’s 36 previous monthly readings:
- the reference is the median of those months, and the spread is the median absolute deviation, scaled by 1.4826 so that it is comparable to a standard deviation;
- April 2020 to March 2022 are skipped, and the window reaches further back instead: the lockdowns, and a year later their base effects, swung every series far outside its usual range;
- the reading’s distance from the median, in spreads, sets the colour: within half a spread is grey (“about usual”); beyond it, blue for faster and red for slower, darker up to two and a half spreads;
- at least 18 reference months are needed, so a series’ first years are grey.
The colour describes an indicator against its own recent past, not whether a reading is good. UPI has grown fast for years but more slowly each year, so it can be red while growing at 20%. Debit card spending has been shrinking as payments move to UPI, so a month in which it shrank less than usual is blue.
Only the past is used: each month’s colour uses readings before it, never later data.
Checks
- Each loader reads the newest promoted Data bank snapshot; the Data bank validates schema, keys, units and the publisher’s own identities before promotion.
- Bank credit and deposit growth are the
/macro/credit/computations, which reproduce RBI’s printed growth rates. - NPCI against RBI and India Data Hub: UPI monthly values agree within 0.1% in 75 of 78 months; the misses are months a vendor shifted by one. NPCI’s daily UPI figures add up to its monthly totals.
- GST tax heads and e-way bills agree exactly with India Data Hub’s copies; PPAC’s figures are newer revisions than the vendor’s. DGCA’s domestic passengers are consistently about half of AAI’s airport counts (each trip is counted at two airports).
Limits
- Indicators are published at different lags (days for daily series, two to three months for RBI’s trade and payment tables), so the newest column of the heat map is never full.
- Festivals move between months (Diwali, Navratri, Onam), and so do vehicle purchases, fuel use and payments around them. A month that holds a festival this year but did not last year shows high growth, and the next month low. Read two months together around festivals.
- Growth on a year earlier carries base effects: a weak month a year ago makes this month look strong. The level panel of each chart shows whether that is happening.
- The page describes the economy’s recent past. It is not a forecast.