Keep a Forecast Revision Ledger

An outlook is a dated assessment. When a new edition changes a growth, inflation, or balance projection, the older number does not become useless. It becomes the prior forecast vintage. Keeping both values shows how expectations evolved and prevents an analyst from presenting the latest estimate as if it had always been known.

Use a long-form revision table

Create one row for each indicator, economy or group, reference period, and publication edition. Include the value, observation state, release date, source URL, and a short source explanation. This structure avoids columns named “old” and “new,” which become unclear after the next update.

The World Bank maintains a MENAAP Economic Update series. The IMF maintains its Middle East and Central Asia outlook series. These institutional series can be tracked over time, but they should remain separate unless their definitions and groups are aligned.

Core fields include:

1. institution and report series;
2. publication edition and release date;
3. indicator, unit, and economy group;
4. reference year or quarter;
5. published value;
6. actual, estimate, or projection state;
7. change from the prior same-source vintage; and
8. cited reason or stated uncertainty.

Do not calculate a revision across two institutions and call it a revision. That is a difference between forecasts. Each institution can use different data cutoffs, assumptions, models, and country groups.

Separate number change from definition change

Before subtracting two values, compare their footnotes. A group can gain or lose a country. A national series can be rebased. A fiscal year can be handled differently. A change in published value after one of these events is not a clean forecast revision.

Add a break flag when definitions do not match. Preserve both values, but do not show a numerical revision without a comparable basis.

Record reasons without overclaiming

Use the report’s own explanation for the first reason field. Store the relevant passage as a short paraphrase with its page or section. If several causes are listed, do not select one as dominant unless the source does.

An analyst can add a second field for independent interpretation. That field must state the additional evidence and uncertainty. It should never silently replace the institution’s stated explanation.

Show the revision path

A line of forecast vintages for one reference year can be more informative than a standard time series. It shows when expectations changed and how large the changes were. Use a second marker for the later estimate or actual value.

The revision ledger supports honest updates. It preserves what was known at each date, makes changing assumptions visible, and helps readers distinguish an economic shock from a data or definition revision. The old forecast is not an error to delete. It is a dated research record.