Why governments raise public pay in steps—and cut it during fiscal stress

The question behind a pay revision

When public pay is fixed by law in nominal terms—the amount stated in currency—rising prices reduce what it buys. Governments must decide when to revise that schedule and by how much. Under financial pressure, they may instead legislate a reduction.

Faisal Quaiyyum’s November 2026 arXiv preprint, That Time of the Year Again: Government Pay Revisions in the Aggregate Economy, studies these decisions themselves, rather than treating public wages as something governments adjust continuously.

Its central explanation involves two opposing costs. A fixed cost of arranging legislation, bargaining or a pay commission encourages governments to act infrequently and make larger changes. A financing cost that increases disproportionately with the size of a raise encourages smaller, more frequent adjustments.

Reading the legal decisions

Quaiyyum builds detailed decision histories for eleven countries: eight European countries, the United States, Bangladesh and India. The records draw on legislation and official notices, distinguishing decisions from the later implementation of previously agreed changes.

Changes are dated by their first payment, when employees receive income and the Treasury pays it. Missing documentation is not treated as evidence that pay stayed unchanged.

The study follows general legal pay for a fixed position, excluding promotions and within-grade increments. This separates revisions to the schedule from changes in the total wage bill caused by employment or career progression. It compares pay with consumer prices and, where available, private earnings. For Bangladesh and India, income per person provides the second benchmark because long private-earnings series are unavailable.

Three patterns emerge. Discretionary pay loses value between revisions and is restored in discrete steps. Across countries, larger raises accompany longer intervals between them. All fifteen cuts in the eight European decision histories were decided during market stress or an adjustment program. Finally, court rulings can separate the cut enacted in law from the cut employees experience and the reduction that remains afterward.

A model connecting pay to the budget

The author estimates a model separately for each country. Each year, the government chooses among the actions its institutions permit, considering current costs and expected future consequences.

The model tracks a gap between legal pay and an unobserved reference schedule, defined as pay consistent with efficient public-service quality. That reference is a model construct, not a directly measured salary target.

In the European models, payroll decisions affect deficits and debt. These influence borrowing spreads—the extra borrowing interest relative to Germany—and the risk of fiscal stress. During stress, the Treasury requires payroll savings. Debt remaining after a crisis can keep pay subdued.

Simulations reproduce many features of the documented histories when observed and simulated pay are measured consistently. Portugal’s real-pay path falls within the model’s simulation interval throughout the studied period. Bangladesh’s simulated new scales, however, are generally larger and less frequent than documented ones.

Removing individual features at unchanged parameter values identifies their roles inside the model. Portugal’s simulated cuts disappear without the Treasury’s savings requirement. Removing the debt-overhang mechanism brings an earlier recovery in pay. These exercises describe model mechanisms, not independently established causal effects.

Courts and automatic adjustments

Portugal illustrates why the timing of judicial review matters. A cut overturned after taking effect can still produce temporary savings. In the model, this gives governments a reason to propose cuts they expect a court to strike.

Simulated review before implementation removes those temporary savings and reduces subsequent pay restorations. It has a lower modeled social cost than review afterward, principally because restoring pay requires financing an increase at once.

The welfare calculation does not assign an independent value to constitutional protections. Its court comparisons therefore concern specified effects on pay, public finances and output, not the overall worth of judicial review.

For Bangladesh, the study considers indexation: automatically adjusting pay for inflation between commission awards. Such rules reduce deep erosion, but their benefits depend on the cost of financing automatic increases. Partial indexation remains beneficial across a wider range of financing assumptions than full indexation. These are conditional simulation results, not measurements of what an automatic rule would actually cost.

Explanation is not prediction

The paper explicitly reports forecasting weaknesses. When blocks of years are withheld and the model is re-estimated, it predicts those decisions worse than constant probabilities based on the remaining record. Outside Portugal and Spain, a government modeled without looking ahead fits decisions as well or better.

The legal records also do not identify how pay revisions causally affect aggregate output. Output effects enter through a calibrated model and do not feed back into pay decisions or fiscal conditions. Together with uncertain parameter estimates, these boundaries make the preprint an account of documented histories and conditional policy experiments—not a validated forecasting tool.