Q) The FPI (forecast period indicator) variable in IBES is confusing.
One company with ticker “ADRX” has the earnings announcement date (REPDATS) on March 5, 2003 for Dec. 31, 2002 earnings. Also analyst code = 047290, has ESTDATS = REVDATS = 20030306, FPEDATS = 20031231, value = 0.66 and fpi = 1.
At the same time, a different analyst code with the code 049483 has ESTDATS = 20030306, REVDATS = 20030324, FPEDATS = 20031231, value = 0.45 and fpi = 2. Do you know why fpi is different for these two entries? Shouldn’t the second one also fpi = 1?
A) FPI is best thought of as the way to distinguish two or more estimates by the broker(s) (for a particular company) around the same estimate date that may have the same or similar forecast period end dates. You can use it to determine whether the estimates are for a quarter, a current fiscal year, the next year, or 'long-term'.
For example, if FPI=”1”, the estimate is for the current fiscal year (“Fiscal Year 1”) and if FPI=”2”, it’s for the next fiscal year (“Fiscal Year 2”). Quarterly estimates for the current and next quarters are identified by FPI=”6” and FPI=”7”, respectively. See the IBES Manuals for more information and our Overview of IBES.
In one sense, FPI is independent of review dates because it refers to a type forecast period. However, there is review date & FPI relationship in transferring an unchanged estimate to an earlier period. Generally, the announcement of earnings by a company will decrement the FPI variable by 1 in all IBES records for that company and that fiscal period which is dated after the announcement date (e.g., REVDATS > REPDATS). For example, estimates in Jan. 2002 for FY 2002 (assume the calendar year = the fiscal year) will have an FPI value of "2" if FY 2001 earnings have not yet been announced. Once earnings for FY 2001 are announced (say, in Feb. 2002), the records for FY 2002 estimates would have an FPI value of "1" for all Review Dates (REVDATS) after the announcement date. No other variable values are affected by earnings announcements, however.
FPI does not resolve all ambiguities in all cases, however. IBES tables may have entry errors or repeated records that were missed in the review process and have not been cleaned by Thomson.
For your example: the company with ticker “ADRX” and earnings estimates ending Dec 31, 2003 (FPEDATS 20031231) has many anomalies where the FPI=2 seems to conflict with the ESTDATS and FPEDATS values.