Seven Rounds of Senate Estimates, 2019–2022: The Other 25 Portfolios Left 2 Questions Unanswered Between Them. Prime Minister and Cabinet Left 358.

When a senator asks a question at a Senate estimates hearing and the official at the table cannot answer it, the question is “taken on notice”: the department owes a written answer, and the committee sets the date it is due. The Senate keeps a database of every such question — 176,568 of them since October 2017, spanning 26 estimates rounds and eight committees. It records 172,321 as answered.

The 4,247 that remain are all past the date their committee set. Most of that number is unremarkable. 3,750 of the unanswered questions belong to the round whose deadlines fell on 17 and 24 July 2026 — two to three weeks before this analysis — and answers were arriving in bulk on the day the data was pulled. That is a queue, not a backlog.

Set the fresh round aside and what is left is 497 questions. They are not spread thinly across a tardy public service. They sit in four portfolios, and 97.4% of them sit in two: Prime Minister and Cabinet, with 358, and Home Affairs, with 126. The gap between those two stocks and everything else is not a matter of degree. Over the seven estimates rounds from October 2019 to February 2022 — the rounds the PM&C backlog comes from — the other 25 portfolios were asked 30,939 questions and have two still unanswered between them, both in the Defence portfolio. The Prime Minister and Cabinet portfolio was asked 6,117 and has 358.

The empty band in the middle of that chart is the finding in miniature. There is a large population of questions running two to three weeks late, which the system will mostly absorb — and a separate population, starting at 90 days and stretching to 2,436, that the system has all but stopped absorbing. 360 of the 497 standing questions are more than four years past due. The median question in the standing backlog is 1,596 days — four years and four months — beyond its date.

The deadline is not a convention. It is set, per question, by the committee.#

“Overdue” here is not measured against an assumed grace period. Each committee sets a return date for each round, the date is printed on the answer documents themselves, and the database carries it per question. Six of those committee-set dates were verified against the dates printed in tabled committee reports and on an answer document before any lateness was computed; the analysis script refuses to run if the reproduction fails.

The Senate also maintains a standing requirement about these answers. Procedural order of continuing effect no. 21, agreed 25 June 2014, requires every minister in the Senate to table — at least ten days before each estimates round — a statement showing how many questions their departments took on notice at the previous round, how many were answered by the committee’s date, and, for the rest, when the answers reached the minister’s office. Lateness is not an informal lapse; it is a thing the Senate has ordered ministers to account for, in writing, three times a year.

Against that structure, the system’s overall record is close to what the Senate designed. Of the 23 rounds whose deadlines passed more than a year ago, twelve have nothing outstanding at all, and twenty are at or above 99% answered.

The pre-registered version of that claim was that every settled round clears at 99% or better. It failed, and the failure is reported here: three rounds sit below the bar, at 98.81%, 98.95% and 94.55%. But the failure does not soften the concentration — it restates it. Of the 245 unanswered questions in those three rounds, 243 are in the Prime Minister and Cabinet portfolio.

One department, seven rounds, one cliff#

The Prime Minister and Cabinet stock is not a story of a department running late everywhere, all the time. It is bounded, datable, and it ends at an election.

Across the seven rounds from October 2019 to February 2022, the Department of the Prime Minister and Cabinet itself was asked 1,429 questions and left 350 unanswered — 24.5%. The per-round path is a deterioration, then a cliff: 8.2% unanswered, then 4.1%, then 26.4%, 33.7%, 12.6%, 16.8% — and then, at the final round before the 2022 election, 91.1%. Senators asked the department 135 questions at the February 2022 hearings, due 25 March 2022. Twelve answers came back. The 123 others are still owed more than four years later.

The volume defence — that PM&C is the coordination department and receives an outsized share of questions — has a true premise. Over those rounds the PM&C portfolio was the highest-volume portfolio of the 26. The conclusion does not follow. Four departments were each asked more questions than PM&C’s 1,429 over the same rounds — Health with 2,076, Home Affairs with 2,057, Infrastructure with 1,746, and Agriculture, Water and the Environment with 1,452 — and all four have nothing outstanding from them. The Finance portfolio, appearing before the same committee on the same days, was asked 1,185 and has nothing outstanding from those rounds. And PM&C’s own later record makes the same point: in every round since the 2022 election whose deadline fell before this year’s just-closed round, the portfolio’s unanswered count today stands at zero — including a 1,697-question round in late 2025, heavier than any it faced in 2019–22. (Its 155 questions still open from the just-closed round sit in the same fresh queue as every other portfolio’s.)

What the record cannot say is why. The database carries no staffing, no workload, no reasons. Whether the non-answers of 2019–22 reflect a decision, a capacity failure, or something else is not knowable from this data, and this analysis does not claim otherwise. What it can say is that the same department, before the same committee, under the same rules, has nothing outstanding from any settled round since the 2022 election, including rounds heavier than any it faced in 2019–22.

What is in the unanswered pile#

These are not administrative trivia. The question documents are public, and the analysis retrieved and read fifteen of them across the two portfolios — separately from the eighteen sampled for the records check described below. Among the questions the Department of the Prime Minister and Cabinet has not answered:

• The drought envoy’s correspondence. Asked 21 October 2019, due 6 December 2019: how many letters did Barnaby Joyce, in his capacity as the government’s special drought envoy, send the Prime Minister, and on what dates? Finance minister Mathias Cormann took the number and dates on notice, noting he would consider what could appropriately be provided. Senator Jenny McAllister, in the hearing: “We are just trying to understand what people got for their $875,000.” Cormann rejected the question’s implication, telling the committee “the job that Barnaby Joyce did as special envoy was incredibly important and he worked incredibly hard”, providing feedback that fed into cabinet’s drought response. The question is 2,436 days past due — the oldest cohort in the database, 18 questions from that one hearing, all PM&C.
• The Prime Minister’s diary and Adani. Same hearing, same due date: did the Prime Minister meet Adani Mining’s chief executive between 1 and 5 April 2019, and if so, could the agenda and minutes be produced?
• A minister’s blind trust. Asked 12 November 2021: did Christian Porter seek advice from the Prime Minister, his office, or the department before establishing a blind trust to receive donations toward his legal fees? Asked again from another angle on 14 February 2022: on what dates did PM&C advise the Prime Minister about Stuart Robert’s blind trust, and what was done to comply with the Statement of Ministerial Standards? Both unanswered.
• Ukraine. Asked 14 February 2022 — ten days before the invasion — due 25 March 2022: what assistance was Australia providing Ukraine? The minister at the table said the government would “seek to get whatever additional detail around practical assistance we can to provide you with, noting that there may be constraints upon what we can share in that regard.” The question is recorded unanswered today.

The file also contains a question about the non-answering itself. On 4 March 2022, a question put to the department read: “Why has the department failed to answer Estimates question 74 relating to Mr Christensen taken on notice on 21 October 2019. Please provide an answer now.” The database records both questions — the 2019 original and the 2022 demand for it — as unanswered today. Two years earlier the same loop was already running: a March 2020 question noted that “Those questions (QONs 54 and 1155) remain unanswered” and asked why. All three are still open.

The committee asked. Eight came off the list.#

None of this went unnoticed inside the Senate. On 9 February 2022, Senator Katy Gallagher used standing order 74(5) — the Senate’s formal mechanism for demanding an explanation of an unanswered estimates question — over PM&C questions then 278 days overdue. Her account to the chamber, naming the department’s then secretary, Phil Gaetjens, was: “I wrote to the Secretary of Prime Minister and Cabinet on 13 October last year asking where the answers to these questions were … Secretary Gaetjens hasn’t even bothered to answer.” That is the senator’s assertion on the parliamentary record, reported here as such; nothing in the material examined for this analysis tests it. The minister representing the Prime Minister, Simon Birmingham, told the chamber that “over the last year” the government had received “not dozens, not hundreds, not thousands but tens of thousands of questions” across both chambers and their committees, and that “there are challenges in working through all of them.” All eighteen of the questions Gallagher named on the floor that day — Nos 1, 3, 5, 6, 501 and 519 to 531 — are still recorded unanswered.

Then the government changed, and the questions did not go away. In December 2022 — five months into the new Parliament — the Finance and Public Administration committee’s estimates report itemised the backlog, round by round: 134 outstanding answers from one round, 38 from another, 29, 60, 79, six and 20 from the rest. 366 in all. The committee wrote:

The committee acknowledges that the outstanding responses relate to answers not provided by the previous government. However, the committee reiterates the principle that outstanding responses be provided before the next estimates round.

That was more than three and a half years ago. Reconciling the committee’s list against the database today, round by round: 358 of the 366 are still unanswered. The outstanding count has fallen by eight — 2.2% — five against one round, three against another, and not at all against the other five. The reconciliation is by round-level counts; the database does not identify which eight.

Two features of the record hold this finding in proportion. First, the itemisation appears once. All eight F&PA estimates reports that the committee’s round pages and the Senate’s tabled-documents register record as tabled since August 2022 raise answer timeliness in some form, but only the December 2022 report lists the pre-election backlog, and the seven later reports do not refer back to it. The committee kept complaining about lateness; the itemised request was not repeated. Second, the itemisation was an observation, not a formal recommendation — the report’s only numbered recommendation concerned training for two land councils, and the government’s formal response answered that recommendation and, consistent with its scope, said nothing about the backlog.

Whether anything now compels those answers is genuinely unclear. The Senate’s procedural authority, Odgers’ Australian Senate Practice, records that “estimates questions on notice are regarded as continuing” across a prorogation, and that in 2016 and in 2019 the Senate passed resolutions setting fresh deadlines for the previous Parliament’s unanswered estimates questions. Whether an equivalent resolution exists for the questions in this backlog was not established by this analysis. What remains true either way: the committee that set the deadlines has asked for the answers, in writing, in a tabled report, and the outstanding list has since shortened by eight of 366.

There is one more turn in the record. Senator Murray Watt asked two of the still-unanswered blind-trust questions in February 2022, from opposition. By October 2024 he was the minister standing up in the Senate to explain a different department’s unanswered estimates questions.

The second stock is forming now#

That different department is Home Affairs, and its numbers are current. Outside the just-closed round, the Home Affairs portfolio has 126 questions past their dates — 106 of them asked in this Parliament, at the October–December 2025 and February 2026 rounds. At department level the shares — 7.0% of one round’s questions, 13.5% of another’s — are far below PM&C’s 24.5% across the seven 2019–22 rounds and nothing like its final round, but they are the largest figures accumulating in the current Parliament, and the older tail is aging: 20 Home Affairs questions are already more than a year past due.

The unanswered Home Affairs questions, like PM&C’s, are on contested ground: how many people in the NZYQ high-court-release cohort have particular offending histories and how many wear monitoring devices; how many visa applications from Palestinian document-holders were referred to ASIO and how many received adverse assessments; how many times ministers lifted the statutory bar to allow visa applications; how many people affected by the fast-track process are in detention. Each was lodged with the committee and taken on notice — in the bar-lift case, after an official at the table twice undertook to provide the figures; each is months past its date.

The Senate’s machinery has engaged here too, and recently. On 9 October 2024, Senator David Shoebridge demanded an explanation for ten unanswered questions; the minister representing Home Affairs, Murray Watt, answered that “the sheer volume of questions on notice” was the cause, citing more than 1,000 questions in a single round — of which, he told the chamber, 672 had already been answered and 376 remained outstanding. On 1 July 2026 — five weeks before this analysis — Shoebridge did it again, for 27 questions across three rounds. Watt’s explanation again rested on volume: he told the Senate that 12,294 estimates questions were submitted for the October 2025 sitting, the most since the database was established, that more than 24,000 were submitted across the supplementary round as a whole, and that a single senator accounted for 17,793 of them — “over 72 per cent”. Those figures are the minister’s account to the chamber, quoted here as such; they were not reconciled against the database for this analysis.

The volume argument deserves to be taken at full strength, and the department’s own tabled accounting is the best evidence for it. Under order 21, Home Affairs reported that of 860 questions from the February 2026 round, 240 were answered by the due date; of the 1,106 and 862 questions from the two late-2025 sittings, 480 and 60 respectively met the date. Most of the rest followed within weeks — the department’s statements show hundreds of answers moving to the minister’s office and on to the committee in the months after each deadline. In the department’s own accounting, the heaviest rounds are also the rounds with the most late answers — and most of what is late still arrives. The standing backlog is what is left after all of that: the questions that did not arrive weeks late, or months late, but have not arrived at all.

What could explain this instead#

The records could be wrong — “unanswered” might mean “never loaded”. This is the explanation that would kill the finding, so it got the most testing. Eighteen question documents were retrieved directly from the parliamentary endpoint: all twelve sampled “unanswered” questions return the question alone, and all six “answered” controls return the question with its answer attached. The committee’s own tabled reports reconcile against the database at every point checked: the 20 answers it recorded outstanding in March 2020 match, in number, the 20 the database shows for that round today; its December 2022 list of 366 matches the 358 now outstanding plus the eight that moved. And the database is demonstrably live — rows flipped to answered between two pulls on the day of this analysis. What could not be fully verified is the other direction: the spotless record of the earliest rounds (19,580 questions, zero unanswered) was confirmed against committee reports for only two of those rounds, so no claim is made here that the pre-2019 record was perfect.

The questions lapsed with the Parliament. Chamber questions on notice do lapse at prorogation and must be resubmitted. But estimates questions are lodged with committees, not the chamber, and Odgers records both that they “are regarded as continuing” and that the Senate has twice set new-Parliament deadlines for old estimates questions precisely so its explanation procedure could reach them. The committee itself plainly did not treat the 366 as extinguished — it itemised them after the election and asked for them. What survives of this objection is real, though: nothing retrieved for this analysis shows a current, enforceable deadline attached to the 2019–22 stock, and the formal remedy belongs to the senator who asked — and 46 of the 358 questions were asked by senators who no longer appear on the Senate’s current-senators list (matched by surname, so the figure is approximate).

The election calendar, not conduct, built the stock. Partly true, and conceded: 131 of the 358 — 37% — come from the February 2022 round, whose answers fell due eight weeks before the May 2022 election. A department whose answering simply stopped as the Parliament wound down would look like this at that round. But the other 227 questions fell due between December 2019 and December 2021, deep inside a sitting Parliament, and the two subsequent post-election governments have had every later round to clear the stock. The calendar explains the spike; it does not explain the years since.

PM&C’s questions were harder. Some of the unanswered questions touch cabinet records, intelligence briefings and ministerial conduct — territory where a full answer may never have been available. The Senate has a procedure for that: a public interest immunity claim, stated and contestable. As Senator James Paterson put it to Home Affairs officials on 4 November 2024, when a question was referred away to ASIO: “Saying again that it’s a matter for ASIO is not grounds, under the Senate’s resolutions, to refuse to answer a question. You have to make a public interest immunity claim.” Whether any of the 358 would have attracted such a claim is unknowable from this data — because a claim, unlike silence, would itself be an answer on the record.

Misattribution. Questions transferred between portfolios, or asked across several portfolios at once, could in principle park answers under the wrong department. In this database the flags for both are zero across all 4,247 unanswered questions. The multi-portfolio flag is demonstrably in use — 202 questions in the 2019–22 rounds carry it, every one of them answered — so its zero is informative. The transfer flag is false on every row inspected, and whether it is ever populated anywhere in the database was not established, so its zero carries less weight.

What this is worth#

The Senate estimates system is, by its own numbers, in working order: 97.6% of everything ever asked in this database has been answered, twelve of the 23 settled rounds have nothing outstanding at all, and even its worst current strain — a Home Affairs portfolio fielding hundreds of questions a round — mostly delivers, weeks late. The standing exception is specific. Over seven consecutive rounds, one department left a quarter of its answers undelivered, ending at twelve of 135; among the questions that remain unanswered are the ministerial-conduct questions set out above; the committee recorded the fact, acknowledged the change of government, and asked its successor to supply the answers; and in the three years and eight months since, under a government whose own senators asked many of the questions, the outstanding list has shortened by eight of 366.

No causal story is offered here, because the data cannot support one. What the record establishes is narrower and, for that, harder to argue with: the questions were asked, the deadlines were set, the machinery for demanding answers was used, the request was tabled — and the answers, 2,436 days on, have not come.