Where the Numbers Come From
A wrong figure does not look wrong. A row in this archive said 15 commercial poultry farms had ongoing H5N1 outbreaks on 20 February 2026, taken from a trade-press restatement of a BAI farm count. WOAH’s WAHIS records 11 Philippine outbreaks starting in February 2026, 10 of them in backyard flocks and one on a commercial farm. The count was in the right neighbourhood. The word in front of it was not, and a keeper who reads “commercial” concludes that the problem belongs to somebody else.
Five rows here have been retracted. Four of them were errors, all four looked as reasonable as that one, and each was found by opening the primary document and rebuilding the row field by field. The fifth was not an error. It was read correctly, straight off a PSA table, and PSA moved the table two days later. Ang numero lang ang dumarating sa iyo, at wala nang kasama iyon: walang column heading, walang petsa ng pangyayari, walang sinasabi kung sino ang binilang.
How does a wrong number get in?
That February row is the one that did the most damage, and not because of the size of the error. Sa bakuran nangyari ang wave na iyon. A backyard-weighted outbreak pattern makes two things live decisions for somebody with 20 birds in the yard: who walks into the kulungan, and where the next batch of stock is bought. A commercial-only reading of the same month takes both off the table. The disease page works through what those decisions look like in practice.
The corrected version of that month is in the archive now as 8 rows, 7 of them backyard flocks, rebuilt from two WAHIS reports. That is 8 against WAHIS’s 11 because the archive carries one row per date and WAHIS counts each outbreak separately. Two counts of the same wave, neither of them in disagreement with the other.
WAHIS event 6473 records the country’s first H5N9 detection in poultry starting 15 April 2025, in a backyard flock at Brgy. Marupit, Camaligan, Camarines Sur, with 110 birds susceptible. This archive filed that detection under 15 April 2026. The claim was true. The year was not, and the news report the row was built from carried a BAI announcement dated 30 April 2025 in its own text, so the correct year never left the page the error came from. A first detection filed twelve months late reads to anybody scanning the archive as a fresh event rather than a settled one.
Two things were wrong in the Oriental Mindoro row, and the numbers were not among them. It was dated 10 July 2026 and placed at “Capalan, Oriental Mindoro”. WAHIS event 7687 puts that outbreak at 2 May 2026, at Brgy. Baruyan, Calapan City, with 39 birds susceptible, 3 cases, 0 deaths and 39 killed. The date was the day the notification reached WAHIS rather than the day the outbreak began, and the place name was misspelled. A third fault was logged against the row and then withdrawn two days later, which is the reason line in the table below.
The fourth is the only peso figure in the set, a Q1 2026 broiler farmgate reading of ₱120.51 per kilo taken from a trade-press restatement and read off the wrong PSA table, worked through in full elsewhere.
Read the four together and they have one shape. Every one came from a secondary source relaying a primary record, and every one was caught by opening the primary rather than by anybody’s instinct about the figure. An agency publishes a record, somebody writes about it keeping the number and dropping the headings around it, and what lands in a spreadsheet afterwards is a description of the write-up.
Rebuilding a row field by field is less work than it sounds. Open the primary document, fill in the date, the place, the population, the unit and the count straight from it, and only then look at what the row already claimed. Kung saan hindi magkatugma ang dalawa, doon ang mali. All four errors here sat in a field nobody checks: the year, what the date field actually meant, the flock type, the table the figure was lifted from. Hindi mismo ang numero ang mali sa apat na iyon kundi ang paligid nito.
All five withdrawn rows are below, struck out and left where they were, each one with the route it travelled and what the primary record said.
| The row, as this archive published it | Where the row came from |
|---|---|
In the order they are worked through above. The four errors were withdrawn on 13 August 2026 and the fifth row on 15 August. Nothing is deleted: a withdrawn row keeps its reason and stays where it was.
The fifth is not a fifth error. That dressed-bird quarter came off PSA OpenSTAT table 0022E4FDDP0 directly, with nobody’s write-up in between, and two days later the same table returned 3,921,174 birds more for it, 1.52% up. Walang namali sa pagbasa at walang naikaltas na heading. PSA revises this table as later returns from registered dressing plants arrive, so the quarter moved underneath a figure that was right when it was taken.
Which means the count of four was measuring the wrong thing. Every figure this archive got wrong did come through a relay, and five retractions in that has not been dented. What no longer holds is the sentence that treated a retraction and an error as the same event. A relayed figure is somebody at this end being careless, and opening the document fixes it. A revision is not fixable from this end at all: the country’s own statistics moved, and they will move again on any recent quarter you happen to be holding. Kaya hindi na sapat na tama ang numero. Kailangan may petsa ito, kung kailan mo binasa ang talahanayan at hindi lang kung ano ang nakita mo doon.
Sino ba talaga ang naglalathala ng datos na ito?
Eighteen of the 30 sources registered in this archive are cited by nothing at all: they are places a Philippine poultry figure might come from rather than places one came from. The other twelve sit behind an actual row. Eight of those carry price observations, WOAH’s WAHIS carries 251 outbreak rows, and three more appear here only on a row that had to be withdrawn.
Here are those eight, most-cited first. The shape of it is the argument: two bars carry the page and the bottom four are nearly invisible, which is what “almost all of it” looks like when you draw it instead of writing it.
| Source | Rows citing it |
|---|---|
| BAI price monitoring | 1,744 |
| PSA OpenSTAT (farmgate) | 1,194 |
| PSA Livestock and Poultry Bulletin | 718 |
| PSA OpenSTAT (retail) | 515 |
| USDA FAS (PSD Online) | 81 |
| PSA OpenSTAT (indicators) | 76 |
| BAI avian influenza status | 72 |
Rows in the price archive, 4,401 in all, live and withdrawn together. The outbreak record is counted apart from this: WOAH WAHIS carries 251 rows there, and three news sources carry one row each, all three of them withdrawn.
Dalawang ahensya ang bumubuhat ng halos lahat nito. Counted by source across all 4,401 price observations, PSA and BAI account for everything except 81 rows of USDA supply-and-demand estimates and a single retracted trade-press row. The eleven national prices those two agencies publish are worked through in the price archive article.
The two are not interchangeable, and knowing which of them a presyo came from tells you what kind of number you are holding before you check anything else. BAI monitors weekly and publishes each week as its own PDF report, going back to January 2023, and it is the only official Philippine series for day-old chick farmgate and for feed ingredient wholesale prices. Doon lang lumalabas ang presyo ng sisiw at ng sangkap ng feeds. PSA works monthly and quarterly instead: the farmgate and retail tables in OpenSTAT are monthly national means, and the Livestock and Poultry Bulletin is quarterly inventory and volume of production. So a weekly BAI figure and a monthly PSA figure for the same manok are not two opinions about one number. Magkaibang bagay ang sinusukat nila, sa magkaibang bilis.
Two of the eight are not Philippine measurements at all. The 81 USDA FAS rows are PSD Online supply-and-demand estimates for the Philippines, produced abroad, and an estimate of this country’s poultry balance is a different object from a count taken inside it. WOAH’s WAHIS is the outbreak record, and it is the only source here that publishes at the level of a single event, with a start date, a location down to the barangay, birds susceptible, cases, deaths and the number killed. Every avian influenza correction above came out of that structure. Kapag wala ang mga detalyeng iyon, wala kang maihahambing sa isang restatement, at tiwala na lang ang natitira sa iyo.
The eighteen that nothing cites make up most of the registry, and they are the more interesting part of it. They include DA’s Bantay Presyo retail monitoring, PSA’s Value of Production tables, PCIC’s poultry insurance schedules, PCAARRD’s native chicken programme and BAFS, which sets the Philippine National Standards for free-range and organic poultry. Not one of them is behind a row today. A registry entry is permission to cite, not evidence that anybody has. Kaya kapag may narinig kang numero tungkol sa gastos sa bakuran at may pangalan ng ahensya sa likod nito, hindi ibig sabihin niyon na may talaan kang mabubuksan.
Which leads to a rule I would give somebody else, and it is stronger than “check your sources”. Do not take a Philippine poultry figure from a news report at all. Use the news report to learn that a release exists, then go and open the release. Exactly one price observation here came from the trade press and it is one of the four errors; the other three arrived through a news restatement of a WOAH or BAI record. That is not a verdict on Philippine journalism, which is where most people hear that a PSA release exists at all. A restatement is a different document from the record, and it does not travel with the record’s column headings.
What the sources themselves cannot do
Provenance discipline fixes the relay, not the source. Five limits sit under the figures on this site whatever route they took to get here, and none of them is fixable from this end.
| Limit | What it breaks | What you do about it |
|---|---|---|
| PSA revises published quarters and published years, not only new ones | The dressed-volume quarter above moved 1.52% between two readings two days apart. The self-sufficiency table read chicken meat 2024 at 92.5 in the vintage published on 13 August 2026, while an earlier reading of the same table put 2024 at 89.1, and 89.1 now sits against 2022 | Recheck a PSA figure against the live table before reusing it. A number you wrote down last year may not be in that table now |
| PSA publishes farmgate about a month behind retail | In the household farmgate table, June 2026 is the latest month carrying a value; July is null across every farmgate series | A “latest price” quoted to you in August is a June figure on the gate side and a July figure on the stall side. Two months of drift, presented as a market level gap |
| BAI’s avian influenza back catalogue is images | Of 137 status reports going back to December 2022, 128 carry no machine-readable text: 127 are a single 83 to 96 dpi JPEG wrapped in a PDF, one is a PNG served with a .pdf name | Area counts here start on 29 May 2026 and the earlier ones are recorded as absent. Any pre-2026 count in circulation was read off a low-resolution screenshot |
| BAI skips weeks in its price monitoring | Those weeks are gaps here, part of 346 nulls across the archive, never a zero and never the previous week carried forward | Ask how many weekly readings a “monthly average” from BAI monitoring is built on. Three is common |
| BAI’s Table 1 is headed “Weekly Price of Broiler per Kilo” and the day-old-chick column inside it is per head | Carry the heading onto the column and you publish a chick price per kilo without noticing | Read the column label, not the table title, on any BAI price you pull |
That last one deserves a box of its own, because the figure it damages is the only input price in the whole archive. Here is the day-old chick reading carrying the four things a title read as a heading takes off it: the unit, the week it describes, the BAI report it came from and the day this archive read it.
Day-old chick, BAI integrator and wholesale farmgate, national
₱33.50/head
BAI: Weekly Prices of Livestock, Poultry and Feed Ingredients (read on 2026-08-13)
Integrator and wholesale farmgate, not what a dealer charges over a counter for 20 sisiw. Kung kada kilo mo binasa, mali na ang gastos mo sa sisiw bago pa dumating ang unang sako ng feeds, at ang buong puhunan mo sa batch ang masisira.
Three of the five limits would disappear without anybody at this end doing anything clever. If BAI typeset its back catalogue, or published the underlying tables beside the images, the area counts before 29 May 2026 stop being a hole and start being rows. If PSA published its vintages alongside the live table, the 92.5 against 89.1 question answers itself instead of being logged twice. If the unit moved from the table title into the column header, the chick price stops being a trap for everybody who reads that report and not only for this archive. Wala pa akong nakikitang balak na gawin ang alinman sa tatlo, so plan around all five as permanent features of Philippine poultry data rather than as a backlog somebody is working through.
Anong itatanong mo sa numerong binanggit sa iyo?
Every question below comes out of a row that had to be withdrawn here. Anim na tanong lang, and they work on any figure anybody quotes you, including the ones on this site. None of them asks whether the number looks right, because all four errors passed that test comfortably.
- Does it name the table or the report, or only the agency? “PSA said” covers a household farmgate table, an establishment table and an average table, and the ₱120.51 row came from the wrong one of the three.
- Does it name the month the thing happened, rather than the month somebody wrote about it? The Oriental Mindoro row carried the date the notification reached WAHIS. The outbreak itself had started ten weeks earlier.
- Does it name the population? Household or commercial establishment on a price, backyard or commercial flock on an outbreak. February 2026 was overwhelmingly backyard and reached this archive as commercial.
- Does it name the unit? Per kilo liveweight, per kilo dressed, per head, per piece. BAI’s own Table 1 hides a per-head column under a per-kilo heading, so the source itself will not always tell you.
- Does the link land on PSA, BAI or WOAH, or on a news site? If it lands on a news site you are one document short of the record, and the missing document is the one with the headings on it.
- Does it say when the table was read? The 2026-Q1 dressed-bird figure was correct on 13 August and superseded on 15 August, and nothing about the number itself tells you which of the two you are holding. Recent quarters and provisional months are where this bites.
Three of those six cost you nothing but a glance at the sentence in front of you. The other three usually mean opening a document, and that is the part most people skip, kasi mabagal at nakakainip. It is also the only part that has ever caught anything. A figure that survives all six is not thereby correct, but you will at least know what it is a measurement of and when somebody looked, which is more than most Philippine poultry numbers in circulation can offer.
On 13 August 2026 this page said there would be a fifth retraction, and set a test on it: “if it ever turns out to have come straight from a primary document rather than through a relay, the finding on this page is wrong and the method needs rethinking rather than repeating.” Dalawang araw lang ang inabot. The fifth came off a PSA table with no relay anywhere near it, and the test fired. What replaced the finding is narrower and less flattering: a relay is how a figure gets wrong, a revision is a separate reason a right figure stops being current, and only the first of the two is yours to prevent.
What this archive does not know is listed beside what it does, in the gaps register, and every series comes out as CSV. What nobody publishes goes through the missing side properly, and it is the harder half: these six questions tell you what a figure is a measurement of, but they cannot conjure a figure where none exists. If you are still pricing your first batch, the five decisions before a first flock is where these numbers get spent.
Cite this page
Every figure here carries the source it came from and the date it was read. If you use one, this is the line that says where it came from.
A backyard poultry keeper. "Where the Numbers Come From". Manoknomics. Published August 22, 2026. https://manoknomics.com/articles/where-the-numbers-come-from
Select the text to copy it.
Questions people actually ask
Where do Philippine poultry price figures actually come from?
Almost all of them come from two agencies, the Philippine Statistics Authority and the Bureau of Animal Industry. Across the 4,401 price observations in this archive, BAI price monitoring carries 1,744, PSA OpenSTAT farmgate 1,194, the PSA Livestock and Poultry Bulletin 718 and PSA OpenSTAT retail 515. Outbreak records come from WOAH WAHIS rather than either agency. Thirty sources are registered in this archive; twelve of them are cited by any row at all.
Can I trust a chicken price I read in the news? Show the answer
Treat a news report as a signal that a release exists, then open the release. A restatement does not carry the column headings of the table it came from, and those headings are where a price gets its population, its market level and its unit. The single price observation this archive took from the trade press, a Q1 2026 broiler farmgate reading of ₱120.51 per kilo, reproduces to the centavo from a PSA table other than the one the series is built on, and it was retracted. Three further errors here arrived through news restatements of WOAH or BAI records.
Does PSA change figures after publishing them? Show the answer
Yes, and fast enough to catch a figure two days old. PSA OpenSTAT table 0022E4FDDP0 gave 257,577,671 chickens dressed in registered plants for the first quarter of 2026 when it was read on 13 August 2026, and 261,498,845 for that same quarter when the same table was read on 15 August: an upward revision of 3,921,174 birds, or 1.52%. PSA revises this table as later returns from registered dressing plants arrive, so the most recent quarters move the most. Carry the date you read a PSA figure alongside the figure, and recheck it against the live table before reusing it.
Where can I find Philippine avian influenza history before 2026? Show the answer
From BAI, but not in a form you can search or copy. BAI lists 137 avian influenza status reports going back to December 2022, and 128 of them carry no machine-readable text: 127 are a single 83 to 96 dpi JPEG wrapped in a PDF and one is a PNG served with a .pdf name. Only the nine most recent are typeset. WOAH WAHIS carries Philippine outbreak notifications in structured form and is the better route for anything older, because any pre-2026 area count in general circulation was transcribed off a low-resolution image by somebody.