Is Government a Good Customer?

A government contract reads well on a slide. It is sticky, it rarely gets re-tendered on a whim, and it comes with the closest thing India’s corporate sector has to a credit-risk-free counterparty. What it does not come with, on average, is a customer who pays you back quickly.

Across 735 listed Indian companies whose annual reports we read for who actually buys from them, the ones with a government body as a dominant or major revenue source ran a median 74 days to collect their receivables in FY2025, against 59 days for companies with little or no government exposure. That fifteen-day gap has been there in nine of the last ten years, not just this one. It gets much wider at the extreme end: a government-exposed company is roughly twice as likely to be stuck above 90 days, and two to three times as likely to be stuck above 180 or 250, every single year back to FY2016.

That is a real, large, and durable pattern. Whether it survives asking the more careful question — is this about the government counterparty, or just about which industries happen to sell to government — turns out to depend on how far into the tail you look. We show both answers below, including the ones we could not settle.

The ten-year picture

debtor_days here is the standard efficiency ratio — average trade receivables divided by revenue, times 365 — computed the same way for every company in every year, sourced from screener.in’s published financial statements. We split the population into two groups using a government-exposure label we built separately (methodology below): companies where government counterparties are the dominant or a major source of revenue versus companies where government exposure is minor or none.

The gap is not a recent development and it is not shrinking. Government-exposed companies ran between 70 and 85 median debtor days in every one of the ten years shown; non-government companies ran between 54 and 68. Both series move together with the broader credit cycle — both dipped in FY2022–23 and both have crept back up since — but the spread between them has stayed roughly constant the entire time. We did not extend this further back than FY2016 even though some companies disclose debtor days earlier — coverage thins out fast before FY2015 and we wanted every year on the chart built from a comparable-sized population.

What we mean by “government exposure”

We had Gemini 2.5 Flash read the most recent annual report of each company on a working list of 742 NSE/BSE names — a hand-curated seed of companies known to sell heavily to government (defence, railways, state utilities) plus a broad sweep across the rest of our coverage universe — and answer one question: how much of this company’s revenue and receivables face a government counterparty, and which kind. Every label carries a verbatim quote from the report as its evidence, so the label is auditable against the source rather than taken on faith. We explicitly excluded paying taxes, receiving a PLI-type incentive, being regulated, or having a government shareholder from counting as exposure. The question is who buys from the company and who owes it money, not who taxes or owns it.

743 labels came back from the 742-ticker target list (a small number of tickers produced more than one usable filing). Of those, 735 matched a non-financial company in our FY2025 debtor-days panel — against a population of 2,463 eligible companies, that is 30% coverage, weighted somewhat toward companies likely to have a government customer because of the hand-curated seed. We treat the label as a snapshot of who a company’s customers are today, applied across the whole ten-year window; a company that only recently won government contracts would be mislabelled as government-exposed for years before that was true, which is a real limitation we have not corrected for.

Confidence in the labels: 483 high, 212 low, 40 medium. The strongest evidence basis was named customers quoted directly from the filing (304 companies), followed by a segment note in the financials (113) and an explicit disclosed government-revenue percentage (64, of which 20 companies disclosed 50% or higher). 244 labels rest on inference rather than an explicit statement, which is why we keep the confidence field and would weight a company-specific conclusion by it.

The gap widens at the tail

A median comparison understates what is actually happening. Look instead at what share of each group is stuck above a threshold — the companies actually taking a long time to get paid, not the typical one.

Almost 1 in 10 government-exposed companies ran over 180 debtor days in FY2025, against roughly 1 in 28 for the rest of the population. About 1 in 17 ran over 250 days — the better part of a year’s worth of revenue sitting uncollected — against about 1 in 50. These multiples (1.7x, 2.7x, 2.9x) have held within a fairly narrow band across all ten years in the panel; they are not a FY2025 artifact and not improving over time.

Not every government counterparty is the same kind of slow payer.

Companies whose government revenue comes from state power utilities and discoms carry the longest cycles by a clear margin — not surprising, given the sector’s own well-documented cash-flow problems. Central ministries, PSUs, and municipal bodies cluster together in the mid-70s. Companies collecting a central subsidy — a payment scheme rather than a buyer with its own balance sheet — run noticeably faster, which is a useful sanity check: it is specifically the customer relationship that drags, not government involvement of any kind.

Is this about government, or about which industries sell to government?

Government-heavy revenue clusters in specific industries — capital goods, infrastructure, engineering — that can carry long receivable cycles for reasons that have nothing to do with the counterparty. A fair test has to hold industry fixed and ask whether the gap survives inside a single sector, not across the whole market.

At the >90-day threshold, where we have enough companies per industry to say something, it does survive:

Five of eight industries with enough companies on both sides (n≥5 each) show government-exposed companies more likely to run over 90 days than their non-government peers in the same industry — Industrial Engineering, Business Services, Electronics Manufacturing, IT Services, and Cement. Two industries disagree by a wide margin: Pharmaceuticals and Steel Pipes & Products, where government-exposed companies in our sample actually collected faster than their non-government peers. We are not explaining these away — they are real counterexamples in the data we have, not noise we chose to discard.

At the >180 and >250-day thresholds — the part of the story with the largest headline multiples — the same test comes back essentially flat: a median difference of 0.0 percentage points, only 3 of 8 industries agreeing with the pooled direction. This is not evidence the severe-tail effect is fake. With 5 to 26 companies in a cell, a “share above 180 days” is often literally zero or one company crossing the line, and a single company can swing a cell by ten to twenty points. The honest statement is that we cannot yet confirm or rule out an industry-independent severe-tail effect with the data in hand — it would take either a larger labelled universe or coarser sector groupings to get a real read.

A note on the receivables-ageing schedule

An earlier version of this analysis used the Schedule III ageing note — the share of trade receivables outstanding for more than six months — instead of debtor_days, and appeared to show the gap widening steadily from FY2018 to FY2025. That apparent trend turned out to be almost entirely a reporting artifact: the ageing schedule changed format in FY2021–22, from a binary “over six months from the due date” disclosure to a multi-bucket schedule measured from the invoice date, phased in through FY2024. The share of companies reporting an exact 0% fell from 81% (FY2020) to 34% (FY2022) to 1% (FY2025) — that collapse tracks disclosure-standard adoption, not a change in how long anyone actually waited to get paid. We switched the primary metric to debtor_days, which is a single ratio unaffected by the schedule change and has clean coverage back to FY2015, and kept the ageing-share comparison only for the FY2023–25 window where the format is stable. It agrees in direction with the debtor_days result (+4.9 percentage points, permutation p=0.0005) but is not independent evidence — it carries the same sector-composition question as everything else in this article.

So, is government a good customer?

Two different answers, depending on which question you ask.

If the question is “does a government-exposed company typically wait longer to get paid, and is that a real, industry-independent pattern”, the answer at the moderate-delay end (running over 90 days) is yes: the effect is large (roughly double the incidence), stable across a decade, and survives being checked inside individual industries in five of eight cases we could test, with two honest exceptions.

If the question is “does government exposure specifically cause the most severe delays — the companies stuck over six months or over eight”, we do not have a confirmed answer. The pooled number looks dramatic (2–3x), it has been stable for ten years, and it is exactly the kind of thing a credit underwriter should ask about a government-heavy applicant. But we cannot yet separate it from sector composition with the data in hand, and we would rather say that plainly than round an unresolved result up to a finding.

What we would tell a credit analyst today: treat government revenue concentration as a real, moderate drag on collection speed worth pricing in, treat the specific counterparty type as informative (state discoms and power utilities materially worse than central ministries or a subsidy scheme), and treat any claim about severe multi-month delinquency being caused by the government relationship, rather than the industry the company happens to be in, as still open.

Data, code, and corrections

Financial data is scraped from screener.in’s consolidated statement pages and stored as dated snapshots; government-exposure labels come from Gemini 2.5 Flash reading each company’s own annual report PDF, with a verbatim evidence quote required for every label. The industry classification used for the sector control is our own mapping, built separately from this study. Code, the full label set, and the figures in this article are in the NAV Framework repository at research/studies/gov_receivables/.

Two things we are deliberately not claiming. We are not claiming causation — a slower-paying government customer could reflect the customer, the industry, or the kind of company that wins government work in the first place, and we have only partly separated these. And we are not claiming the label set is a random sample of the market — the seed list tilts toward companies already known to be government-heavy, though it is a small fraction (60 of 742) of the target list.

Corrections are the point. If you find an error in the labels, the panel, or the analysis, we would rather hear about it than have it stand.

This is research, not investment advice. Nothing here is a recommendation to buy or sell any security. The past collection performance of the companies discussed above is a historical fact and carries no guarantee about the future.