Depreciation

Depreciation spreads an asset's cost across its useful life. The important design decision is that it is written for each period, not recomputed whenever somebody looks.

Why written, not recomputed

If depreciation were calculated on read, then editing a useful life or a residual value would restate every period already reported. Somebody adjusting an estimate in October would change what was reported in March, without intending to and without anyone noticing.

Writing the charge per period means posted figures stay posted. A change to an estimate affects future periods, which is also how accounting standards expect a change in estimate to be treated.

Re-running a period is refused

Running depreciation for a period that has already been run is rejected rather than silently doubling the charge. Depreciation runs are usually scheduled, and a scheduled job that runs twice because of a retry or an overlap is not unusual — refusing the second run is what makes the schedule safe.

If a period genuinely needs correcting, the answer is an adjusting posting, visible as such.

Setting up

Depreciation needs the method, the useful life and any residual value, usually held at class level so a policy is applied consistently rather than decided per asset.

Running it

Run depreciation as part of period close, before closing. The charge posts to the ledger like any other journal, so it appears in the period's figures and in the trial balance you review before closing.

Assets acquired mid-period

Decide your convention — full period in the month of acquisition, pro-rata by days, or none until the following period — and apply it consistently. Any of them is defensible; changing between them is not.

Checking

Reconcile accumulated depreciation to the register periodically. A divergence usually means an asset was added with a wrong acquisition date, or a disposal was recorded without stopping the depreciation.