Audit

They overlap in the records they read and differ in what they are testing. The financial audit cares whether the gratuity provision is adequate; the inspector cares whether gratuity was paid on time. An organisation can satisfy one and fail the other.

A fourth sense of the word, an HR audit reviewing policies, capability and practice against good practice rather than against law, is a management exercise with no external standard behind it. It is useful and it should not be confused with the three above, because nothing turns on it.

Inspections proceed from records outward, and the sequence is fairly consistent.

• Are the prescribed registers maintained, in the prescribed form, with the prescribed particulars, and retained for the prescribed period? Electronic maintenance is permitted under the Codes; incompleteness is not.
• Do attendance, the wage register and the wage slips reconcile for a sample of employees and periods? A mismatch means one of the three was produced independently of the others.
• Are wages at or above the applicable minimum for the category of work, tested against the statutory definition of wages rather than against gross pay?
• Were wages paid within the prescribed period, and dues on termination within theirs?
• Are deductions within the permitted categories and the ceiling, and where fines or damage recoveries appear, was the procedure followed and recorded?
• Are contributions computed on the right base, deposited by the due date, and reconciled to what was reported in the returns?
• Do contract workers at the premises appear where they should, and does the principal employer's own record reflect them?

The Codes designate the inspecting officer an inspector-cum-facilitator, with a function of advising employers on compliance alongside the enforcement role, and provide for web-based inspection schemes. Verify the scheme applicable to the establishment and state, since the inspection regime and its randomisation differ.

Three of those tests can be tied to specific provisions, which is worth doing because they are the ones that produce findings. The registers are required by section 50(1) of the Code on Wages, which names the register of persons employed, the muster roll and wages, and by section 33 of the OSH Code, which sets out the particulars and permits the register to be kept electronically or otherwise in the form the appropriate Government prescribes. Wage slips are required every wage period by section 50(3), and the notice board abstract under section 50(2) has to carry category-wise wage rates, the wage period, the day and time of payment and the Inspector-cum-Facilitator's name and address. Deductions are tested against section 18(2), which is a closed list of permitted purposes, and section 18(3), which caps the total in any wage period at fifty per cent of wages. The point about testing wage floors against the statutory definition rather than gross pay comes from section 2(y), whose first proviso sets a one-half proportion rule between the excluded components and total remuneration.

Deliberate underpayment is rare. The findings that actually recur are process failures.

• Late deposit of contributions or tax. The computation was right and the money moved a week late, attracting interest and damages.
• Challans that do not reconcile to returns, so money is with the authority and the credit is not with the employee.
• Registers missing prescribed particulars, usually a column nobody knew was required.
• Records unavailable for earlier periods after a system migration.
• Contract labour treated as wholly the contractor's concern, with no principal employer records at all.
• Wage floors tested against gross pay rather than the statutory wage base, which produces a shortfall in structures heavy on excluded allowances.
• Wage slips not issued for every wage period, or issued only on request.
• Fines applied without the procedure, and no register of fines.

Every one of these is cheaper to prevent than to remediate, and none requires legal expertise to avoid. They require the calendar to be complete and the records to be generated from the run rather than assembled afterwards.

The common preparation, a scramble to assemble registers before a visit, is both expensive and self-defeating: records assembled for an inspection are exactly the records that will not reconcile to anything.

The alternative is to make the records a by-product.

• Generate registers, challans and returns from the payroll run rather than preparing them separately. Two people building the same numbers from different sources is the origin of most reconciliation findings.
• Reconcile deposit to return every period rather than annually, so a gap costs a correction rather than a correction statement.
• Hold the prescribed particulars as fields rather than free text, so a missing column is visible before an inspector finds it.
• Test retrievability for old periods occasionally, rather than discovering during an inspection that a migration took three years with it.
• Keep a compliance calendar per state and per entity, with an owner against each obligation.
• Run an internal review on the inspection's own sequence: pick five employees and five periods and follow attendance through to the return.

That last exercise takes an afternoon and finds more than a policy review does, because it tests the joins between systems, which is where the failures live.

If an organisation runs one HR audit a year, the highest-yield version is narrow and specific.

• Sample-based, following individual employees end to end rather than reviewing documents by category.
• Focused on joins: attendance to payroll, payroll to challan, challan to return, return to the employee's annual statement.
• Including the edges, since joiners, leavers and mid-year changes are where errors concentrate.
• Covering contract labour at the premises, which is the most commonly unexamined population.
• Reported with owners and dates rather than as observations.

A broad policy review reads well and finds little. Following five people through the system finds the things that would actually appear as findings, and it can be done without anyone external.