Logout / Access Other products Drop Down Arrow
Get live help Monday-Friday from 7AM-7PM ET (12PM-12PM GMT)  •  Contact Us
Check out our new FAQ section!
RSS Feed
FAQ: Forecasts (v.2)
Friday, 08 May 2009 14:40 ET
By Anthony Cione, Phillip Thorne
Summary
May 2009 -- Moody's Economy.com produces forecast time series, available through DataBuffet.com. We receive many client questions about how to purchase, navigate, manipulate and interpret them.
Detail

This FAQ document will change and expand as we respond to additional questions from clients. You can expect it to appear periodically in the DataBuffet News feed.

Identification

1. How do I know it's a forecast time series?
If the series mnemonic (alphanumeric identifier) begins with “F” it's a forecast.
If it's an “R” it may be. Usually, “R” denotes a Moody’s Economy.com estimated series, which we produce when official data are sparse in time, space or detail; but a minority of these series contain forecasted observations.
2. How do I know if the forecast series is seasonally adjusted?
There are two ways: you can examine the mnemonic or the description.
If the first two letters of the mnemonic are “FX” the series is NSA (not seasonally adjusted). If the second letter is notX” it's SA (seasonally adjusted) or SAAR (seasonally adjusted, annualized rate).
In the online catalog, you can right-click a series and pick More Info. The Summary tab summarizes the contents of the folder, and the Units field may indicate if the series is NSA, SA, or SAAR. If it reads “As specified,” that means the folder contains series with a mix of units.
If you open the series in View or Basket mode, you'll see a series-specific description. The last part of the description, enclosed in parentheses, describes the measurement units. The NSA, SA or SAAR indicator will appear here.
Note: Some historical data series employ a “working day adjustment,” abbreviated “WDA.”
3. What's a macro forecast?
We use the term “macro” to describe our U.S.national-level forecast.  It is labeled “United States” in the online catalog.
4. What's a regional forecast?
We use the term “regional” to describe our U.S. subnational forecasts, as opposed to our “macro” (U.S. national) forecast.  We produce these at the state, metropolitan area, and county levels.

Runs and Revisions

5. How frequently are the forecasts updated?
Our forecasts are updated each month with the latest available historical data.
6. When will the next update be published?
Update schedules are posted in the DataBuffet.com catalog.  Open the forecast in the Catalogs tab, right-click any time series, choose MoreInfo, then the Release Schedule tab.
Please note that all forecast release dates are approximate (marked with an asterisk) and subject to delays.
7. What does “quarterly, updated monthly” mean?
Most of our forecast time series have a quarterly native frequency, meaning that an observationis computed for each calendar quarter. However, that computation is performed once per month, utilizing the latest available input data (drivers). Hence, you can expect the forecast observations to change repeatedly.
Note: A minority of forecast time series have an annual native frequency, not quarterly.
8. When are the monthly updates performed?
For the U.S., the national, state and metro forecasts are updated during the first, second and third weeks of the month, respectively, unless special circumstances intervene.
Specialized (consumer, housing stock, industry, Case-Shiller® home price index) and international forecasts have other schedules.
The staggered U.S. release schedule was announced in early February on DataBuffet News. In March 2009, regional forecasts were delayed to incorporate the BLS 2008 benchmark.
9. Can you notify me when the forecast is updated?
You can notify yourself by creating a scheduled basket with a triggered delivery to automatically run and transmit a file containing forecast series of interest.
Open a basket, click Schedules on the blue menu bar, and proceed through the Schedule Wizard. In step two, for Frequency pick “Any change to series” and name a trigger series. To pick a series from the basket, click Toggle series list.
The “New data point” trigger option is useful for historical time series.
Note: Occasionally a forecast time series will undergo changes unrelated to model updates, such as corrections to metadata. These changes will trigger deliveries that do not contain updated values.
10. Can I obtain a copy of an older forecast?
Previous runs (vintages) of the forecast model are not available through DataBuffet.com. If you have questions, please contact your sales representative.
11. Which vintage is available now?
To find out, open the forecast in the DataBuffet.com online catalog.  Select any time series, left-click for the Geography Wizard, and look for the parenthetical note alongside each checkbox.
Note: This feature is not available for the Housing Stock, Consumer, County, Employment Output & Wages, orIndustry forecasts.

The Historical Portion of Forecasts

12. Do the forecast series include historical actuals?
Yes, with caveats. Each forecast series does have a historical portion and is based on one or more historical driver series. If the forecast is a composite there will be nothing to compare. If it has a single driver, but a nonquarterly one, the two series won't be directly comparable (see #16).
13. Where do the observations change from historical to forecast?
There's no indicator in DataBuffet.com, but you can include an optional header field in a downloaded basket.
Open the basket editor, click Options from the blue menu bar, and then Option Group “Fields.” In the Data Values group, select the “Historical end dates” checkbox. Press Apply to close the Options pane, and save the basket.
Note: This feature is not available for all forecast datasets.
14. Why does the historical portion change? Shouldn't it be immutable?
This occurs if the forecast series has a historical driver that's subject to revision, such as the house price indices from FHFA or Case-Shiller®.
15. A driver was updated today. Why isn't it part of the forecast immediately?
We have a fixed update schedule for the forecasts. For example, the historical data may be published on Day 3 of the month, but we don’t update the forecast until Day 7.
The subsequent forecast points are based on all available historical data at the time of the monthly model update.
16. Is there a way to compare a forecast with its historical driver?
Yes, but you'll need to convert both series to their lowest common frequency. For example:
  • If the historical series is monthly, convert it to quarterly to match the forecast.
  • If the historical is annual, convert the forecast to annual to match it.
Converting in the other direction won’t work because frequency conversion is nonreversible. If you up-convert a quarterly forecast to a monthly frequency, it’s just an interpolation, which is inherently smooth and cannot be expected to match the jagged monthly driver.
In some cases, and for various reasons, the two series will not match point-for-point, no matter what manipulation you apply.
17. How exactly does frequency conversion work?
A forthcoming article will explain this topic in detail.

Scenarios

18. What are the baseline and alternatives?
Our forecast models are based on assumptions. From the infinite number of possible assumptions, our analysts pick five that cover a range of illustrative outcomes. The “baseline” is the one we deem most likely.
For pricing flexibility, we offer two forecast packages: one with just the baseline (called the Current Forecast) and another with five (the Alternative Scenarios, consisting of the baseline plus four alternates).
Note: We produce alternative scenarios only for the U.S. macro, state, and metro forecasts.
19. How do the baseline and current forecasts differ?
The current and baseline forecast datasets are computed with the same assumptions. However, they're computed at different points in the month.  Specifically, the baseline is updated later, along with the alternative scenarios.
There are alternating periods when the current and baseline forecasts will have equal values (when they have the same vintage) and when the differ (when the vintage of the baseline lags).
You can compare the vintages through the Geography Wizard.
20. How are the alternative time series named?
Each set is denoted by a concept code suffix. For example, with our forecast of U.S.-level nominal GDP, the current forecast series is named FGDP.US (no suffix), the baseline is FGDP_B.US, and the four alternatives are FGDP_S1.US through FGDP_S4.US.
21. Where are scenario probabilities located?
You will need to open the individual scenario assumption documents.
22. What do the probabilities on each scenario mean?
It's impossible to claim that “There's a 26% chance the economy will unfold this particular way.” Instead, you can read the numbers as “Under these assumptions, out of all possible futures, 26% are better than this forecast, and 74% are worse.”

Scenario Assumptions

23. Do your assumptions change?
Yes, as economic conditions warrant. Accordingly, two successive vintages of a forecast series may show very different values, even if the driver values haven't changed much.
24. Where do I get the scenario assumptions?
Each month, our analysts write explanations of the five sets of assumptions. These documents are available through DataBuffet.com,or we can email them to you.
On DataBuffet.com, pick a forecast catalog (labeled with a blue book icon), right-click, and pick View Assumptions.  The resulting pop-up window has a drop-down box to choose between the five documents.
Or, you can open a series with the Geography Wizard,then pick the View Assumptions link.

Forecast Documentation and Support

25. Where can I find assumptions?
Please see above.
26. Do you have methodologies?
Methodology documents that explain our forecast process in detail are attached to the series in selected forecasts.  Left-click, and in the More Info window, switch to the Background tab.  If necessary, click the Additional Information hyperlink.
You can also check in the Reference Files section of DataBuffet.com, accessible through the Tools menu.  Scroll down to the Forecast Documentation section.
27. Do you have other documentation?
We produce external documentation in response to client requests as circumstances and resources permit. Please check in these locations:
  • The Reference Files section of DataBuffet.com, accessible through the Tools menu.
  • The daily searchable DataBuffet News feed.
28. Who can I contact with additional questions?
If you have questions about forecasts, historical data, your account, or any other aspect of DataBuffet.com, you may contact us by email at help@economy.com, or by LiveHelp internet chat.
LiveHelp is available during U.S. business hours (Monday to Friday, 7:00 a.m. to 7:00 p.m. ET; to 6:00 p.m. during the summer).  To start a chat, use the “Get live help” hyperlink in the upper right corner of the DataBuffet.com window, or from the Help menu.
Each forecast has a chief researcher.  We may provide direct contact details so that you can ask specific questions; please keep these for future reference.
Other