The Test Year and Adjustments to It

Rates are set prospectively from a retrospective measurement. The convention that makes this possible is the test year: a twelve-month period whose revenues, expenses and rate base are treated as representative of the operation the new rates will govern. No actual period is ever representative, which is why the choice of period matters less than the adjustments made to it.

Why the period is a year, and which year

Twelve months is the shortest period that captures a full seasonal cycle in a business whose sales are driven by weather and whose maintenance follows annual patterns. Anything shorter would embed a season. Anything longer would blur changes the rates need to reflect. The convention is universal in American ratemaking even though the statutes rarely compel it in terms.

Which twelve months is a jurisdictional choice. A historic test year uses a recently completed period for which recorded data exist, and its virtue is verifiability. A fully forecast test year projects the period during which the new rates will actually be in effect, and its virtue is currency. Between the two sits the common compromise: a historic year updated through a later date for changes that have occurred and are documented.

Federal wholesale practice illustrates the hybrid. A filing must present data for a recorded twelve-month period ending shortly before the filing, and may present a second twelve-month period beginning and ending within stated windows around the proposed effective date. If the projected period is not submitted, the recorded period becomes the test period by default.

The choice is also constrained by data. A jurisdiction using forecast periods requires utilities to maintain budgeting systems capable of producing auditable projections at the level of detail the case demands, and requires commission staff able to test them. Where neither capability exists, a historic period is the only workable convention regardless of its analytic disadvantages.

The known and measurable constraint

Because the test period stands in for the future, adjustments to it are inevitable. The traditional boundary on those adjustments is that a change must be known and measurable. Known means reasonably certain to occur rather than merely possible. Measurable means quantifiable from evidence rather than estimated from judgment. A wage increase already agreed in a contract satisfies both. A wage increase management expects to negotiate satisfies neither.

The constraint is applied more strictly in historic test year jurisdictions than in forecast ones, for an obvious reason: a forecast test year is by construction a set of projections, so applying a known and measurable filter to it would defeat the purpose. Forecast jurisdictions substitute a reasonableness inquiry, testing whether the projection method is sound and whether the assumptions are supported.

Timing is part of the test. Most jurisdictions require that a change occur within a defined window after the end of the test year, often six or twelve months, so that a utility cannot include an expense increase expected several years out. The cutoff date is itself contested, since a plant addition placed in service just after the cutoff represents real investment on which no return will be allowed until the next case.

The adjustments made in almost every case

Weather normalization comes first. Sales during the test period are adjusted to reflect normal weather, computed from a long-run average of degree days, so that a mild winter does not produce an inflated revenue deficiency and a severe one does not produce an inflated revenue base. Customer counts are annualized, using either an average or a year-end figure depending on local convention.

Payroll is annualized to the level in effect at the end of the period, including contractual increases already agreed. Depreciation is recomputed on the adjusted plant balances at approved rates. Property taxes are updated to the most recent assessment. Interest on debt is recomputed for issues completed or retired. Non-recurring items are removed, and items that recur on a cycle longer than a year are normalized to an annual level.

Plant additions receive particular attention. A utility that completed a major project shortly after the test year has an obvious interest in including it, and customers have an equally obvious interest in the offsetting revenue and expense effects being included too. Most commissions permit post-test-year plant that is in service by a stated date, on the condition that all related effects are reflected.

Revenue adjustments follow the same logic in reverse. Rate schedules changed during the test period are applied to the full period as though they had been in effect throughout, so that the revenue figure reflects the tariff the utility actually charges. Unbilled revenue at each end of the period is trued up, and revenues from discontinued services are removed along with the costs that produced them.

Attrition, lag and the next case

Where costs rise faster than revenue, a utility earns less than its authorized return as the period between cases lengthens. That erosion is called attrition, and it is the principal argument for forecast test years and for the tracking mechanisms that operate between cases. The counterargument is that lag disciplines cost: a utility that reduces expense after rates are set keeps the difference until the next case, and removing that opportunity removes the incentive.

Statutory deadlines interact with the choice. Where a commission must act within a stated number of days, and the utility may otherwise place rates into effect subject to bond and refund, the test year's currency at the time of decision depends partly on how long the case actually runs. A period that was current at filing may be a year and a half old when the order issues.

The period selected determines the figures used throughout the rest of the case, including the balances discussed in what enters the rate base and the expense levels in the revenue requirement and how it is built. Costs that are excluded from the test year because they are volatile are often recovered instead through the mechanisms described in fuel and purchased power adjustments and riders and trackers outside a rate case.