Economic Explainer
A scenario reports what follows from specified assumptions; a forecast adds a judgment about likely outcomes. An explicit fiscal calculation shows how sensitivity, consistency and uncertainty should be presented without assigning unsupported probabilities.
A scenario states what follows if a specified set of conditions holds. A forecast adds a judgment about which conditions and outcomes are likely. A stress test may deliberately examine an unfavorable combination without describing it as the most likely path. The numerical precision of a calculated result does not convert an assumption into a prediction.
The Federal Reserve made this distinction explicit in its December 2020 supervisory stress-test scenarios, describing the adverse paths as hypothetical exercises to assess bank resilience rather than forecasts. That historical example establishes the purpose of the exercise; its assumed economic paths should not be read as current projections. Federal Reserve, December 2020 supervisory scenarios.
Consider a hypothetical one-year budget. Assume nominal GDP of one thousand monetary units, a constant tax-revenue ratio of twenty percent, primary expenditure of one hundred ninety and interest expense of twenty. Ignore all other revenue and expenditure. The overall balance is revenue less primary expenditure and interest.
| Assumption or result | Reference case | Lower nominal activity | Joint adverse conditions |
|---|---|---|---|
| Nominal GDP | 1,000 | 950 | 950 |
| Revenue ratio | 20% | 20% | 20% |
| Calculated revenue | 200 | 190 | 190 |
| Primary expenditure | 190 | 190 | 198 |
| Interest expense | 20 | 20 | 25 |
| Overall balance | −10 | −20 | −33 |
The reference calculation is 0.20 × 1,000 − 190 − 20 = −10. Reducing nominal GDP to nine hundred fifty lowers revenue to one hundred ninety under the unchanged revenue-ratio assumption, producing a balance of minus twenty. In the joint adverse case, primary expenditure and interest also rise, so the balance becomes 190 − 198 − 25 = −33.
The table is an original illustration with invented inputs. It does not estimate the finances of any government. The reference case is merely the comparison point chosen for the exercise, not a central forecast. The two alternatives have no assigned probabilities.
The second column changes nominal GDP while holding the other inputs fixed. This reveals the arithmetic exposure to that single input under the stipulated tax rule. It does not claim that expenditure and interest would actually remain constant during a downturn.
The GDP concept is important. The tax calculation uses nominal activity. A fall in real output accompanied by higher prices would not necessarily produce the same nominal revenue change. Calling this a real-growth scenario without specifying prices would hide a material assumption.
The constant revenue ratio is also restrictive. Different tax bases respond differently to profits, wages, consumption and asset prices. A more detailed model could allow those bases and collection lags to vary. Until that work is done, the simple ratio should remain visible as the device that generates the result, rather than being described as an estimated fiscal response.
Sensitivity can also be reported around a threshold. In this example, with primary expenditure and interest fixed at their reference values, balance requires revenue of two hundred ten. At the assumed tax ratio, that corresponds to nominal GDP of one thousand fifty. This threshold is an accounting result. It does not establish whether that output level is economically attainable or how policy could produce it.
The third column moves several inputs together. To become a substantive scenario rather than a collection of unfavorable numbers, it needs an explanation of the joint conditions. Higher primary spending could represent an assumed support response; higher interest could reflect a specified refinancing exposure. These possibilities are hypotheses requiring evidence, not mechanisms proven by the table.
Consistency also requires attention to timing. New borrowing may increase interest only as debt is issued or repriced. An immediate increase applied to the entire stock would overstate near-term exposure when much of it carries fixed coupons. The difference between debt levels and financing schedules is examined in debt-service exposure.
Feedback can alter the result further. A spending response may influence activity, and a change in financing costs may affect private investment. The illustrative table does not model either effect. Its conclusion is consequently limited to the direct arithmetic under the stated inputs. Adding the word scenario does not make omitted feedback disappear.
A range of calculated outcomes can communicate sensitivity to selected assumptions. It is not a confidence interval unless a statistical procedure and probability interpretation support it. Three columns do not imply that outcomes between them are equally likely, nor that events outside them are impossible.
An operational presentation should state the information date, horizon, policy assumptions and variables held fixed. It should distinguish observed starting values from judgmental inputs and show which equations connect them. Alternative paths should address a meaningful decision, such as liquidity needs or the scale of a possible budget response.
A practical review should also identify the assumption that would change the decision. If a funding shortfall appears only beyond a stated rate or revenue threshold, monitoring that threshold is more informative than repeatedly publishing the same scenario table. This connects the conditional exercise to a specific observation that can be checked.
The final conclusion should describe the exposure the calculation establishes. In this exercise, the budget's assumed revenue rule transmits lower nominal activity into borrowing, with additional spending and interest assumptions enlarging that requirement. The table provides a transparent conditional result. A forecast would require a separate, defensible judgment about the likely path of each input and the interactions among them.