Why Money Supply Data Is Revised: How to Track M2 Vintages
Money-supply data are revised when source reports, estimation methods, or seasonal factors change. A past observation can receive a new value without changing its observation date. To reproduce an older chart, you need the data vintage available when it was made, not simply the latest history for the same months.
Why money supply data is revised
A money figure can change when banks send better data or a source changes its method. A source can also correct how it groups accounts. The Federal Reserve’s H.6 history records benchmark updates to deposit data and changes to seasonal factors. Read the release note to find out which change took place. Source: Federal Reserve H.6 announcements.
US M2 provides a dated example. On 28 July 2026, the Fed changed where it subtracts IRA and Keogh balances: at the aggregate M2 level rather than within individual components. Unadjusted total M2 stayed unchanged, while seasonally adjusted M2 received minor revisions. The change raised some component values without the same rise in total M2. Source: H.6 technical Q&A, July 2026.
Find out what changed first. Did the source revise the balance? Did it update a seasonal factor? Did it just change how it shows one part of the total? A chart of the total cannot answer a question about one part on its own.
We explain adjusted and unadjusted data in our seasonally adjusted money supply guide. Neither label means that past values must stay fixed.
Separate the observation date from the vintage
The observation date tells you which period a value measures. The release date tells you when the source published it. A vintage is the version available at a point in time. The St. Louis Fed’s ALFRED service saves past vintages so you can look up older versions. Source: ALFRED help.
Consider a hypothetical March value. An April release reports it for the first time. A June release then changes it. Both files still label the value as March, but their release dates and vintages differ. You need those details to rebuild the old chart.
The retrieval date is when you downloaded the file. It may differ from the source’s release date. ALFRED also says that it may update its database after the source releases new data. Check which dates and vintages the service holds for your series.
This matters when you ask what someone could have known at the time. Today’s history gives the current estimate of the past. A dated vintage shows what data were available then. Mixing them can make an old result look wrong even when it used the right inputs at the time.
A revision can change a growth rate
A change to the earlier value can reverse the growth rate, even if the later value stays the same. Our hypothetical table shows this effect. All balances are invented teaching values in currency units. “Earlier” and “later” name data versions, not actual releases.
| Hypothetical data vintage | Month A balance | Month B balance | Calculated A-to-B growth |
|---|---|---|---|
| Earlier vintage | 100.00 | 101.00 | 1.00% |
| Later vintage | 102.00 | 101.00 | −0.98% |
Scroll to see all columns
For the earlier vintage, we calculate (101 / 100 − 1) × 100 = 1.00%. For the later vintage, we calculate (101 / 102 − 1) × 100 = −0.98%. Month B never changes. The denominator does.
To check the math in a spreadsheet, enter each row’s balances in cells side by side. Divide Month B by Month A, subtract one, and format as a percentage. Round only for display. Use both values from the same vintage; do not mix old and new downloads.
The same rule applies to year-over-year growth if the value twelve months earlier changes. A new growth rate does not tell you which input changed. Check both levels before you explain the percentage.
Breaks and reclassifications need a separate check
A series break changes what you can compare across periods. A revision changes an already published value. They can happen together, but they are not the same thing. US M1 has a documented break beginning in May 2020, when savings deposits entered its definition. That classification change left M2 unchanged. See the Federal Reserve’s explanation and our M1 glossary entry.
Some published growth rates also strip out changes that do not come from transactions. The Bank of England separates those flows from changes in classification or value. Its growth rate can thus differ from the simple change between two published balances. Source: Bank of England flow and growth-rate methodology.
For example, suppose a hypothetical series starts at 100 currency units. It records 5 units of transactions and 10 units from a reclassification. Its closing balance is 115. The balance grows by 15.00%. Strip out the assumed reclassification and growth from transactions is 5.00% for that period. We built this example to show the difference. These are not UK data.
If your rate differs from the source’s headline, check how the source defines growth before calling either result wrong. Our methodology explains the calculations used on this site. Use the label attached to the actual calculation.
Keep a reproducible record of your chart
Save the data and its meaning together. A screenshot keeps the chart’s look, but may lose its inputs and formula. Our record below splits out the facts that a single “updated” label can hide.
| Record field | What to save |
|---|---|
| Series identity | Provider, identifier, monetary aggregate, geography |
| Measurement | Currency, units, frequency, seasonal treatment, average or end-period convention |
| Observation window | First and last measured periods |
| Publication history | Source release date and selected vintage, where available |
| Retrieval | Download date and the original data file |
| Calculation | Exact formula, comparison periods, and rounding rule |
| Caveats | Break flags, revision notes, and any missing observations |
Scroll to see all columns
For a chart based on our data downloads, save the file you used. A link helps readers find current data. It cannot keep an older version by itself. A file checksum can show which saved copy you used if you later compare archives.
When reviewing the money-supply growth ranking, keep the data version and ranking rules. A rank can change when a value changes, when its growth window changes, or when rows enter or leave the ranking. The rank alone cannot tell you which.
Frequently asked questions
Does a revision mean the original release was fabricated?
No. A revision alone shows only that a published value changed. Read the source’s notes to find the cause. It could be new reports, a corrected error, revised seasonal factors, or a new method. The size and reason need evidence.
Which vintage should I use for an old chart?
Use the vintage that fits your question. To rebuild an old chart, use its inputs or a matching vintage. To show the current estimate of a past period, use the latest data for it. State which version you chose.
Can unadjusted money-supply data be revised too?
Yes. “Not seasonally adjusted” names a method; it does not mean the data are final. Check source corrections, benchmark updates, and notes on account groups. If only seasonal factors change, only the adjusted series may change.