Collection Without Reconciliation
Why collection at source fails in low-capacity administrations, and what should replace it
Withholding takes the tax before the payee holds the money and distributes collection across thousands of independent payers at no administrative cost beyond oversight. It should leak least. It frequently leaks most. This study asks why.
Expected against remitted
Transaction data implies a liability; a payment arrives. Neither figure is wrong. Nothing sets one against the other.
Overview
Withholding is, by design, the most assured collection mechanism available to a revenue authority. It takes the tax before the payee ever holds the money, and it distributes the work of collection across thousands of independent payers who deduct and remit on the state’s behalf at no administrative cost beyond oversight.
It should therefore be the revenue stream that leaks least. In practice it is frequently among the least certain. The failure is not one of payer honesty, rate, or enforcement will. It is a failure of reconciliation.
The distinguishing feature of this failure, and the reason it merits separate treatment, is that the two quantities to be compared are held by different parts of the administration and arrive on different clocks. Liability is implied by transaction data; remittance appears in a payment ledger. Neither is wrong. Neither is missing. They are simply never brought together — and so the difference between them, which is the entire compliance position, exists nowhere.
The central claim
The revenue lost to withholding non-compliance is, in the main, not evaded revenue. It is unreconciled revenue — under-deduction, partial remittance, late lodgement and misallocation absorbed silently into normal operations because nothing systematically compares what was owed against what arrived.
Diagnostic framework
Process reconstruction of the as-is cycle, failure-mode decomposition against a problem tree, and design prescription derived by inverting the root condition.
Key diagnostic insights
- 01 Expected and remitted amounts are compared at reporting intervals rather than as they arrive, so a shortfall persists across periods before anyone looks — by which time recovery has weakened and the same error has recurred.
- 02 What is matched is what there was time to match. What was not reached is not recorded as unreached; its absence is indistinguishable from its having reconciled.
- 03 Management receives the total remitted, because that is what the payment ledger produces without further work. Without the expected figure beside it, the remitted figure is uninterpretable.
The root condition
Withholding administration is treated as a series of disconnected handoffs — registration, then submission, then payment, then case — each performed competently in isolation by a different part of the institution, rather than as a single continuous comparison between expected liability and remitted amount for which the institution holds one defensible record.
The reference design
The prescription is organised around a single invariant, from which the architecture follows.
The invariant
The expected liability, computed from validated transaction data under the rate and threshold in force, is never adjusted to match what arrived. What was remitted is recorded alongside it, never in place of it. The difference between them is itself a record, not a transient calculation. Reconciliation is the act of explaining that difference — not of making it disappear.
The three-record model
Expected · D
What the law required
Derived from the validated transaction by applying the rate and threshold in force, with the rate version stamped on the record. Immutable once computed.
Remitted · C
What arrived
The payment matched to this payer, period and expectation — including the basis on which the match was made, because a match is itself an inference.
Difference · V
The case
V = D − C, retained as a record rather than derived on request, with its sign, magnitude, age and classification.
What should be measured
An implementation should be evaluated against outcomes rather than delivery milestones. Each indicator below is observable, and each degrades visibly when the design is failing.
| Indicator | What it measures | Healthy trend |
|---|---|---|
| Reconciliation coverage | The share of expectations matched against remittance, as against those never reached. | The figure to watch is the never-reached share, which should approach zero — under the manual process it was never measured at all. |
| Detection latency | Elapsed time between a difference arising and its being examined. | Falls sharply, then stabilises. Latency that stops falling while volume rises indicates the queue is outpacing capacity. |
| Disposition mix | What differences turn out to be. | Varied and moving. A mix that is stable and overwhelmingly one category indicates a defect in the design, not a fact about payers. |
| Approval amendment rate | Whether the approval checkpoint is exercising judgement. | Low but non-zero. Zero means approval is a formality. |
| Traceability | Whether any closed difference can be reconstructed on demand. | Complete, and tested by reproducing a historical case end to end. |
Study reference & lineage
Unit of analysis: The withholding reconciliation cycle of a national revenue authority.
Method: Process reconstruction, failure-mode decomposition, design prescription.
Applicability: Revenue administrations that rely on independent payers to deduct and remit on the state’s behalf.
This study advances a general argument about administrative design. It does not describe, endorse, or evaluate any commercially available system.
© 2026 Collection Without Reconciliation — concept study prepared by Sujoy Maitra.