Receivables & payables × Research note
Getting paid in construction: payment delays, the MSME rules and evidence you can act on
Cash, not profit, is what sinks contractors, and construction has some of the slowest payment cycles of any industry. What the evidence and statutory payment rules say about late payment, and why a collections worklist should keep confirmed risk and suspicion apart.
Key takeaways
- One 2024 industry report estimates the average construction payment cycle at about 90 days and the cost of slow payment at $280bn a year in the market it covers.
- Some laws set hard limits. Under India’s MSMED Act, for example, buyers must pay registered small suppliers within 45 days at most, or owe compound interest at three times the central bank rate, whatever the contract says.
- Receivables and payables are two sides of the same cash problem, but they should never be netted in a review list.
- A worklist should keep confirmed settlement risk separate from patterns worth a look. Blending them into one score hides what is actually known.
- Predicting who will pay late needs point-in-time data and a time-based test; a snapshot of today’s statuses is not a forecast.
01The cash problem
A contractor pays for labour, materials and plant weeks or months before being paid for the work. The gap is financed from cash reserves, overdrafts or suppliers’ patience. When payment is slow, profitable firms can still fail.
Rabbet’s 2024 Construction Payments Report, a survey of construction firms by a construction-finance software company, estimated the average payment cycle at about 90 days and put the industry-wide cost of slow payment at about $280 billion in 2024[1]. As a vendor survey, its exact figures deserve caution; the direction matches what anyone who has run a project’s cash flow will recognise.
Profitable contractors still fail when cash arrives three months after the work.
02What the law can say about paying small suppliers
Several countries set statutory payment terms to protect small suppliers. India’s Micro, Small and Medium Enterprises Development (MSMED) Act, 2006 is a clear example: it sets hard limits on how long a buyer may take to pay a registered micro or small enterprise[2]:
- Section 15: pay within the agreed period, and in any case within 45 days of accepting the goods or services. Where there is no agreed period, the limit is 15 days.
- Section 16: if payment is late, the buyer owes compound interest with monthly rests at three times the bank rate notified by the Reserve Bank of India. This rate applies notwithstanding any agreement to the contrary.
For a main contractor, many subcontractors and material suppliers fall within this Act. That turns the payables ledger into a compliance register: an invoice from a registered small supplier that is 60 days old is not just slow, it is accruing statutory interest.
03Case study: Accounts Receivable & Payable
From my portfolio · project 10 of 15
Synthetic data- Problem
- Unsettled customer and vendor invoices need follow-up, but not all have the same evidence behind them.
- Decision supported
- Which receivables to chase and which payables to resolve first.
- Data
- 1,000 invoices (500 AR, 500 AP), 110 of them unpaid.
- Method
- Two separate worklists: confirmed settlement risk, and repeat-pattern review; no blended score.
- Result
- 22 confirmed (12 AR, 10 AP) and 88 repeat-pattern invoices from two customers.
- Limits
- An evidence viewer, not a predictive model; label agreement is by construction.
04Evidence, not a blended score
The tempting design for a collections tool is one “risk score” per invoice. My Accounts Receivable & Payable project deliberately avoids that, because the invoices in a review list do not all carry the same kind of evidence[3]:
Figure
Unpaid invoices by worklist (1,000 invoices, 110 unpaid)
- Confirmed settlement risk is an invoice already labelled as a settlement problem: unpaid and overdue or disputed.
- Repeat-pattern review is an unpaid invoice from a party that has at least two unsettled invoices. It is worth a look, but it is a pattern, not a finding.
Keeping these apart means a collections team knows which calls are about a confirmed problem and which are about a customer’s behaviour. Receivables and payables are shown in separate views and never netted, because a customer who owes you money and a vendor you owe are different conversations, even when they are the same company.
05From review to prediction
The project is honest about what it is: an evidence viewer on a snapshot, not a predictive model. Its agreement with the labelled data is 100% by construction, because the confirmed category comes from the labels themselves[3]. Predicting which invoices will be paid late is a different problem, and it needs different data:
- Point-in-time history. For each invoice, what was known on the day it was raised: the customer’s payment record up to that date, not including later invoices.
- Features that exist on day one: amount, terms, project stage, whether the work is certified, dispute history, client type.
- A time-based test. Train on invoices raised earlier, test on invoices raised later, and compare with a simple baseline such as “customer paid late last time”.
06How strong is the evidence?
Not every finding in this note rests on the same kind of evidence. This is how I would weigh each one before acting on it.
| Finding | Evidence | Strength | Main caveat |
|---|---|---|---|
| Construction payment cycles are long | Rabbet 2024 industry survey | Indicative | Vendor-sponsored survey |
| Small suppliers must be paid within 45 days | MSMED Act 2006, s.15–16 | Strong | Applies to registered micro and small enterprises |
| Confirmed risk and patterns should stay separate | Design principle shown in my AR/AP app | Moderate | Evidence viewer, not a tested predictor |
07Controls for cash
- Record MSME status in the vendor master and report payables older than 45 days for registered suppliers.
- Separate confirmed issues from patterns in every collections and payables list, and label which is which.
- Never net receivables against payables in a review view, even for the same counterparty.
- Chase certification, not just invoices. Many receivables are late because the underlying work has not been certified.
- Keep dated history of every status change, so you can later build and test a genuine late-payment forecast.
NotesSources
- Rabbet (2024). 2024 Construction Payments Report.
- Micro, Small and Medium Enterprises Development Act, 2006, sections 15–16. Summary in SCC Online: when is interest payable under section 16 of the MSMED Act.
- Iwale, A. (2026). Accounts Receivable & Payable. GitHub.
Figures are quoted from the sources above as published; where a source reports a range or a survey estimate, it is described that way. Results from my own projects say whether they use real public data or synthetic data.