Official information kept secret too long

Correction 2020-06-02 When creating the codebook for this article almost two years after it was published, I had to re-write the code to generate all of the graphs because I no longer had the original code. In this process, I found that I had made mistakes when looking at how many requests were extended by the Ministry of Social Development. These mistakes extended to showing MSD to have had more overdue requests than they actually had, because I had not correctly recognised that some requests had been extended. When re-creating the graphs, there were also some small differences due to different choices being made when cleaning the data. For all details of the changes made when this article was corrected, please see the log of changes made to this page on GitHub.

For the past few years, I have been keeping track of how long it takes for government agencies to respond to Official Information Act requests.

The OIA sets a time limit of 20 working days to complete Agencies can extend this, but they have to tell the requester how much longer they need and give a reason why the extension is necessary. Technically the time limit is for sending a notice of the agency's decision about whether or not to release the requested information, rather than applying to the release of the information itself. Section 28(5) of the OIA essentially requires that, when a decision is made to grant a request, the information must be released without "undue delay". However, in the vast majority of cases, information will be released at the same time as a requester is told the agency has decided to grant their request. these requests. As well as this due date, the OIA also requires that requests must be completed "as soon as reasonably practicable". However, I've found that it's been very common for responses to my requests to arrive on the day they are due.

Of course, this might be because it wasn't reasonably practicable for a response to be sent any sooner than the due date. But when less than half of the responses arrive before the due date, things start to look a bit suspicious.

In late 2014, then Prime Minister John Key admitted that, when handling OIA requests:

Sometimes we wait the 20 days because, in the end, Government might take the view that's in our best interest to do that

John Key

However, it's been a few years since then, and things might have changed. Later that year, then Chief Ombudsman Beverley Wakem started a year-long review of the operation of the OIA across 12 government agencies, which resulted in the report Not a game of hide and seek. In response to Key's admission of the government gaming the OIA, she said:

It's pretty clear. It couldn't be much clearer than that... As soon as you have made a decision as to whether you're going to respond to the request or how you're going to respond to it, you ought to convey that.

Beverley Wakem

This view was backed up in April 2018 by the current Chief Ombudsman, Peter Boshier, who referred to the report in a speech given to managers at MBIE:

Some [agencies] were frank about their practice of 'gaming' the OIA, particularly by interpreting the 20-day time limit as an opportunity [to] 'put it off until the 20-day deadline' to either delay release or delay having to deal with the request.

This is clearly not acceptable in an open and participatory democracy.

Peter Boshier

One thing that has changed since 2014 is that the State Services Commission has started collecting and publishing statistics on OIA compliance.

This includes a measure of how many requests were completed "within the legislated timeframe". Most agencies seem to be doing fairly well, with few falling below 90% for this measure of compliance. But were those responses sent "within the legislated timeframe" also sent as soon as reasonably practicable, or did they tend to be put off until the due date arrived?

To find out, I sent requests for six months' of data on OIA response times to 11 government agencies, many of which had also been included in the review by the previous Chief Ombudsman.

I also sent an OIA request to the State Services Commission beforehand, in the hope of both refining my request and getting some basis for comparison, asking how quickly they responded to OIA requests in 2016/17.

There are various ways in which agencies can delay OIA requests, or deter requesters. Some of these can be examined by looking at the data that has been released, others I've observed anecdotally but don't have data to show how common or rare they are. Whether or not it's ever an agency's intention to delay or prevent the release of official information isn't something I can tell from that, but in either case the effect is the same.

You can find links to the data released by each agency at the start of their section. There's more information there than I have analysed in this article, and I'd encourage you to explore the data yourself if you're interested.

State Services Commission

Data

The day before my request to them was due, the SSC extended it by eight working days. I received the data the day before that extended due date.

The SSC told me they dealt with The criteria used for my requests weren't identical to those used in the statistics published by the SSC, so the numbers won't always line up perfectly. For example, the 2016/17 OIA statistics published for the SSC show they completed 97 requests in that timeframe, all of which were completed on time. 96 OIA requests in 2016/17. Looking at the number of working days remaining when they completed these requests, a very clear pattern emerges:

SSC OIA days remaining 2016-07-01 – 2017-06-30
Extended requests
Unextended requests
Requests
  • 0
  • 10
  • 20
  • 30
  • 40
Working days
    • Extended requests 25: 0
    • Unextended requests 25: 0
    25
    • Extended requests 24: 0
    • Unextended requests 24: 0
    • Extended requests 23: 1
    • Unextended requests 23: 0
    • Extended requests 22: 0
    • Unextended requests 22: 0
    • Extended requests 21: 0
    • Unextended requests 21: 0
    • Extended requests 20: 0
    • Unextended requests 20: 2
    20
    • Extended requests 19: 0
    • Unextended requests 19: 4
    • Extended requests 18: 0
    • Unextended requests 18: 4
    • Extended requests 17: 0
    • Unextended requests 17: 3
    • Extended requests 16: 0
    • Unextended requests 16: 5
    • Extended requests 15: 0
    • Unextended requests 15: 5
    15
    • Extended requests 14: 0
    • Unextended requests 14: 2
    • Extended requests 13: 0
    • Unextended requests 13: 3
    • Extended requests 12: 0
    • Unextended requests 12: 3
    • Extended requests 11: 0
    • Unextended requests 11: 1
    • Extended requests 10: 0
    • Unextended requests 10: 0
    10
    • Extended requests 9: 0
    • Unextended requests 9: 3
    • Extended requests 8: 0
    • Unextended requests 8: 1
    • Extended requests 7: 0
    • Unextended requests 7: 2
    • Extended requests 6: 0
    • Unextended requests 6: 3
    • Extended requests 5: 0
    • Unextended requests 5: 2
    5
    • Extended requests 4: 0
    • Unextended requests 4: 3
    • Extended requests 3: 0
    • Unextended requests 3: 4
    • Extended requests 2: 0
    • Unextended requests 2: 2
    • Extended requests 1: 30
    • Unextended requests 1: 5
    • Extended requests 0: 3
    • Unextended requests 0: 4
    0
    • Extended requests -1: 0
    • Unextended requests -1: 1
    • Extended requests -2: 0
    • Unextended requests -2: 0
    • Extended requests -3: 0
    • Unextended requests -3: 0
    • Extended requests -4: 0
    • Unextended requests -4: 0
    • Extended requests -5: 0
    • Unextended requests -5: 0
    -5

Requests that didn't have their due date extended have a fairly even spread, with all but one being completed before the due date. But over a third of all requests were extended, and all but four of those were completed on the day before the due date.

However, looking at the lengths of those extensions the spread is also fairly even:

SSC OIA extension length 2016-07-01 – 2017-06-30
Requests
  • 0
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
Working days
    • Extension duration 0: 0
    0
    • Extension duration 1: 0
    • Extension duration 2: 4
    • Extension duration 3: 3
    • Extension duration 4: 2
    • Extension duration 5: 6
    5
    • Extension duration 6: 0
    • Extension duration 7: 2
    • Extension duration 8: 2
    • Extension duration 9: 2
    • Extension duration 10: 1
    10
    • Extension duration 11: 0
    • Extension duration 12: 1
    • Extension duration 13: 1
    • Extension duration 14: 1
    • Extension duration 15: 0
    15
    • Extension duration 16: 1
    • Extension duration 17: 1
    • Extension duration 18: 1
    • Extension duration 19: 1
    • Extension duration 20: 1
    20
    • Extension duration 21: 1
    • Extension duration 22: 0
    • Extension duration 23: 0
    • Extension duration 24: 0
    • Extension duration 25: 1
    25
    • Extension duration 26: 0
    • Extension duration 27: 0
    • Extension duration 28: 0
    • Extension duration 29: 0
    • Extension duration 30: 0
    30
    • Extension duration 31: 0
    • Extension duration 32: 0
    • Extension duration 33: 0
    • Extension duration 34: 0
    • Extension duration 35: 0
    35
    • Extension duration 36: 0
    • Extension duration 37: 1
    • Extension duration 38: 0
    • Extension duration 39: 0
    • Extension duration 40: 0
    40
    • Extension duration 41: 0
    • Extension duration 42: 0
    • Extension duration 43: 0
    • Extension duration 44: 0
    • Extension duration 45: 0
    45
    • Extension duration 46: 1
    • Extension duration 47: 0
    • Extension duration 48: 0
    • Extension duration 49: 0
    • Extension duration 50: 0
    50

So instead of waiting until the day before requests are due before responding, it looks like the SSC is probably just good at estimating the amount of work remaining, and they tend to add an extra day just in case.

There is one more thing worth checking, however, and that's whether or not the SSC has a practice of not sending notices of extension until a request's original due date arrives:

SSC OIA days remaining when extended 2016-07-01 – 2017-06-30
Requests
  • 0
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
Working days
    • Working days remaining when extended 10: 0
    10
    • Working days remaining when extended 9: 3
    • Working days remaining when extended 8: 3
    • Working days remaining when extended 7: 2
    • Working days remaining when extended 6: 6
    • Working days remaining when extended 5: 2
    5
    • Working days remaining when extended 4: 3
    • Working days remaining when extended 3: 6
    • Working days remaining when extended 2: 3
    • Working days remaining when extended 1: 2
    • Working days remaining when extended 0: 3
    0
    • Working days remaining when extended -1: 1
    • Working days remaining when extended -2: 0
    • Working days remaining when extended -3: 0
    • Working days remaining when extended -4: 0
    • Working days remaining when extended -5: 0
    -5

There's a fairly even spread here as well, though there was one request where a notice of extension wasn't sent until the day after the request was originally due. That request was extended by 16 working days, and was one of those that received its final response the day before its extended due date.

Given the SSC's involvement in publishing OIA guidance for other agencies and monitoring OIA compliance, I'm not surprised to find that they seem to be doing pretty well here.

Ministry for Primary Industries

Data

I was pleasantly surprised by the quick response I received from MPI. In contrast with the other agencies I contacted, MPI was quick to get in touch with me to discuss refining my request. In the end, they took only 6 working days to respond, which was one of the fastest responses I've ever received to an OIA request.

We reached a compromise where they would release only some of the data I had requested, because the rest would be difficult to retrieve. However, the quick turnaround and the fact that they'd talked to me about how best to manage my request were both refreshing changes from my typical experience with OIA requests.

Unfortunately, the data are not so complimentary:

MPI OIA days remaining 2017-07-01 – 2017-12-30
Extended requests
Unextended requests
Requests
  • 0
  • 50
  • 100
  • 150
Working days
    • Extended requests 20: 0
    • Unextended requests 20: 2
    20
    • Extended requests 19: 0
    • Unextended requests 19: 0
    • Extended requests 18: 0
    • Unextended requests 18: 1
    • Extended requests 17: 0
    • Unextended requests 17: 4
    • Extended requests 16: 0
    • Unextended requests 16: 2
    • Extended requests 15: 0
    • Unextended requests 15: 5
    • Extended requests 14: 2
    • Unextended requests 14: 2
    • Extended requests 13: 0
    • Unextended requests 13: 8
    • Extended requests 12: 2
    • Unextended requests 12: 6
    • Extended requests 11: 1
    • Unextended requests 11: 3
    • Extended requests 10: 3
    • Unextended requests 10: 3
    10
    • Extended requests 9: 2
    • Unextended requests 9: 2
    • Extended requests 8: 0
    • Unextended requests 8: 8
    • Extended requests 7: 1
    • Unextended requests 7: 4
    • Extended requests 6: 2
    • Unextended requests 6: 8
    • Extended requests 5: 2
    • Unextended requests 5: 7
    • Extended requests 4: 1
    • Unextended requests 4: 12
    • Extended requests 3: 0
    • Unextended requests 3: 6
    • Extended requests 2: 6
    • Unextended requests 2: 6
    • Extended requests 1: 3
    • Unextended requests 1: 24
    • Extended requests 0: 69
    • Unextended requests 0: 70
    0
    • Extended requests -1: 1
    • Unextended requests -1: 0
    • Extended requests -2: 3
    • Unextended requests -2: 0
    • Extended requests -3: 3
    • Unextended requests -3: 0
    • Extended requests -4: 1
    • Unextended requests -4: 1
    • Extended requests -5: 2
    • Unextended requests -5: 0
    • Extended requests -6: 2
    • Unextended requests -6: 1
    • Extended requests -7: 0
    • Unextended requests -7: 0
    • Extended requests -8: 0
    • Unextended requests -8: 1
    • Extended requests -9: 2
    • Unextended requests -9: 0
    • Extended requests -10: 0
    • Unextended requests -10: 0
    -10
    • Extended requests -11: 0
    • Unextended requests -11: 0
    • Extended requests -12: 1
    • Unextended requests -12: 0
    • Extended requests -13: 1
    • Unextended requests -13: 0
    • Extended requests -14: 0
    • Unextended requests -14: 1
    • Extended requests -15: 0
    • Unextended requests -15: 0
    • Extended requests -16: 0
    • Unextended requests -16: 0
    • Extended requests -17: 1
    • Unextended requests -17: 0
    • Extended requests -18: 0
    • Unextended requests -18: 0
    • Extended requests -19: 0
    • Unextended requests -19: 0
    • Extended requests -20: 0
    • Unextended requests -20: 0
    -20

There is a huge spike in responses sent on the due date. There is also a long tail of requests that were not responded to in time. Nineteen more late requests don't fit on the graph, with the latest being over five months overdue.

Though the SSC's stats report that MPI completed 87% of requests in this period within the legislated timeframe, that number obscures the fact that fewer than half of their responses were sent before the due date.

Of those requests that were late, they weren't clustered soon after the due date as you might expect. Rather they were spread out over the next few months. Of the overdue requests, the median number of working days late they were was 15.5 — over three weeks.

MPI didn't tell me how many days were remaining when they sent notices of extension, so I can't tell if these were also left until the last minute. I can, however, look at the durations of these extensions:

MPI OIA extension length 2017-07-01 – 2017-12-30
Requests
  • 0
  • 5
  • 10
  • 15
  • 20
Working days
    • Extension duration 0: 1
    0
    • Extension duration 1: 0
    • Extension duration 2: 0
    • Extension duration 3: 0
    • Extension duration 4: 3
    • Extension duration 5: 3
    5
    • Extension duration 6: 7
    • Extension duration 7: 3
    • Extension duration 8: 1
    • Extension duration 9: 3
    • Extension duration 10: 11
    10
    • Extension duration 11: 5
    • Extension duration 12: 3
    • Extension duration 13: 3
    • Extension duration 14: 5
    • Extension duration 15: 8
    15
    • Extension duration 16: 5
    • Extension duration 17: 4
    • Extension duration 18: 3
    • Extension duration 19: 7
    • Extension duration 20: 18
    20
    • Extension duration 21: 4
    • Extension duration 22: 2
    • Extension duration 23: 3
    • Extension duration 24: 2
    • Extension duration 25: 7
    25
    • Extension duration 26: 2
    • Extension duration 27: 1
    • Extension duration 28: 0
    • Extension duration 29: 2
    • Extension duration 30: 3
    30
    • Extension duration 31: 2
    • Extension duration 32: 0
    • Extension duration 33: 0
    • Extension duration 34: 0
    • Extension duration 35: 0
    35
    • Extension duration 36: 0
    • Extension duration 37: 0
    • Extension duration 38: 1
    • Extension duration 39: 0
    • Extension duration 40: 1
    40

(The request which was apparently extended by 0 days was reported by MPI as having been extended, but the due date they reported for it was 20 working days after the date on which it was received.)

Unlike the SSC, MPI's extensions are clustered around multiples of five. Extensions of 20 working days, in particular, are very common.

Those requests extended by a multiple of 5 working days did tend to be answered a little less late On average, requests extended by a multiple of five working days were 2.1 working days overdue, whereas other extended requests were 6.4 working days overdue on average. There were some outliers though, requests that were months overdue which can skew the average. By far the most common outcome for each group of extended requests was for the response to be sent on the day it was due. on average.

Whether they intend to delay OIA requests or not, that certainly seems to be the effect of the way MPI handles them.

MPI have also recently made the news for dragging their feet in releasing information to the Herald. MPI refused a request the Herald made in March, but reconsidered their decision following an investigation by the Ombudsman. MPI also extended a separate request from the Herald by a further 25 working days, only notifying the Herald of the extension once the due date had arrived.

Ministry of Justice

Data

The Ministry of Justice handled 446 OIA requests The numbers reported by the SSC — 436 requests completed with 85% being on time — are slightly different due to different interpretations of the criteria for which requests to include. within Q3-4 2017, and 88% of them were answered within the legislated timeframe.

Of those 87%, about a third of them were only completed either on the due date or the day before it:

Ministry of Justice OIA days remaining 2017-07-01 – 2017-12-30
Extended requests
Unextended requests
Requests
  • 0
  • 30
  • 60
  • 90
Working days
    • Extended requests 20: 0
    • Unextended requests 20: 19
    20
    • Extended requests 19: 0
    • Unextended requests 19: 4
    • Extended requests 18: 0
    • Unextended requests 18: 10
    • Extended requests 17: 0
    • Unextended requests 17: 11
    • Extended requests 16: 0
    • Unextended requests 16: 10
    • Extended requests 15: 0
    • Unextended requests 15: 12
    • Extended requests 14: 1
    • Unextended requests 14: 19
    • Extended requests 13: 0
    • Unextended requests 13: 17
    • Extended requests 12: 0
    • Unextended requests 12: 14
    • Extended requests 11: 0
    • Unextended requests 11: 9
    • Extended requests 10: 1
    • Unextended requests 10: 4
    10
    • Extended requests 9: 1
    • Unextended requests 9: 10
    • Extended requests 8: 1
    • Unextended requests 8: 13
    • Extended requests 7: 0
    • Unextended requests 7: 15
    • Extended requests 6: 1
    • Unextended requests 6: 16
    • Extended requests 5: 2
    • Unextended requests 5: 17
    • Extended requests 4: 5
    • Unextended requests 4: 16
    • Extended requests 3: 0
    • Unextended requests 3: 14
    • Extended requests 2: 3
    • Unextended requests 2: 17
    • Extended requests 1: 6
    • Unextended requests 1: 33
    • Extended requests 0: 15
    • Unextended requests 0: 71
    0
    • Extended requests -1: 3
    • Unextended requests -1: 8
    • Extended requests -2: 1
    • Unextended requests -2: 12
    • Extended requests -3: 0
    • Unextended requests -3: 7
    • Extended requests -4: 1
    • Unextended requests -4: 3
    • Extended requests -5: 2
    • Unextended requests -5: 0
    • Extended requests -6: 0
    • Unextended requests -6: 0
    • Extended requests -7: 0
    • Unextended requests -7: 4
    • Extended requests -8: 0
    • Unextended requests -8: 1
    • Extended requests -9: 0
    • Unextended requests -9: 3
    • Extended requests -10: 0
    • Unextended requests -10: 2
    -10
    • Extended requests -11: 0
    • Unextended requests -11: 1
    • Extended requests -12: 0
    • Unextended requests -12: 1
    • Extended requests -13: 0
    • Unextended requests -13: 1
    • Extended requests -14: 0
    • Unextended requests -14: 0
    • Extended requests -15: 0
    • Unextended requests -15: 0
    • Extended requests -16: 0
    • Unextended requests -16: 1
    • Extended requests -17: 0
    • Unextended requests -17: 0
    • Extended requests -18: 0
    • Unextended requests -18: 0
    • Extended requests -19: 0
    • Unextended requests -19: 0
    • Extended requests -20: 0
    • Unextended requests -20: 0
    -20

The Ministry of Justice only extended a relatively small number of their requests compared with agencies like MPI and the SSC. It's quite possible that different agencies will typically receive OIA requests with different degrees of complexity, so without more information I wouldn't want to read much into that. The Ministry of Justice may not have extended many OIA requests because it just didn't need to.

However, it seems the Ministry of Justice also avoided sending notifications of an extension after a request's due date. They didn't include the date on which extension notices were sent in the data they released, but many of their overdue requests The Ministry of Justice extended 44 OIA requests within this period, whereas 36 unextended requests were at least two working days late. had not been extended. This is despite there being time between the due date and when the request was completed during which the requester could have been told how much longer was still necessary.

The Ministry of Justice was about twice as likely to respond to extended requests on the due date (34%) as they were for unextended requests (18%). This could be in large part due to the durations of the extensions being accurate reflections of the time remaining, but looking at the distribution of these durations there is a pattern of extensions being most often taken for whole numbers of weeks:

Ministry of Justice OIA extension length 2017-07-01 – 2017-12-30
Requests
  • 0
  • 2
  • 4
  • 6
  • 8
Working days
    • Extension duration 0: 0
    0
    • Extension duration 1: 0
    • Extension duration 2: 0
    • Extension duration 3: 0
    • Extension duration 4: 1
    • Extension duration 5: 4
    5
    • Extension duration 6: 0
    • Extension duration 7: 1
    • Extension duration 8: 1
    • Extension duration 9: 1
    • Extension duration 10: 4
    10
    • Extension duration 11: 0
    • Extension duration 12: 0
    • Extension duration 13: 0
    • Extension duration 14: 1
    • Extension duration 15: 2
    15
    • Extension duration 16: 2
    • Extension duration 17: 3
    • Extension duration 18: 1
    • Extension duration 19: 2
    • Extension duration 20: 8
    20
    • Extension duration 21: 0
    • Extension duration 22: 1
    • Extension duration 23: 0
    • Extension duration 24: 0
    • Extension duration 25: 3
    25
    • Extension duration 26: 0
    • Extension duration 27: 0
    • Extension duration 28: 1
    • Extension duration 29: 0
    • Extension duration 30: 1
    30
    • Extension duration 31: 0
    • Extension duration 32: 0
    • Extension duration 33: 0
    • Extension duration 34: 1
    • Extension duration 35: 1
    35
    • Extension duration 36: 0
    • Extension duration 37: 0
    • Extension duration 38: 2
    • Extension duration 39: 1
    • Extension duration 40: 0
    40
    • Extension duration 41: 1
    • Extension duration 42: 0
    • Extension duration 43: 0
    • Extension duration 44: 1
    • Extension duration 45: 0
    45

With such a small number of extended requests, it's hard to conclude much from this pattern for the Ministry of Justice. The largest spikes are at 5, 10, and 20 working days. Those extensions account for just over a third of all extensions (36%). This could be worse, but it still seems likely that this practice of often extending requests by a number of weeks rather than a number of days may be leading to responses being unnecessarily delayed.

Ministry of Foreign Affairs and Trade

Data

The Ministry of Foreign Affairs and Trade handled just over 100 OIA requests in this timeframe, and the SSC's statistics report that 93% of them were completed within the legislated timeframe. However, as with other agencies, a large number of those requests were completed on the due date. In the case of MFAT, about a quarter of OIA requests they completed received a response on the due date:

Ministry of Foreign Affairs and Trade OIA days remaining 2017-07-01 – 2017-12-30
Extended requests
Unextended requests
Requests
  • 0
  • 10
  • 20
  • 30
Working days
    • Extended requests 25: 0
    • Unextended requests 25: 0
    25
    • Extended requests 24: 0
    • Unextended requests 24: 0
    • Extended requests 23: 0
    • Unextended requests 23: 0
    • Extended requests 22: 3
    • Unextended requests 22: 0
    • Extended requests 21: 0
    • Unextended requests 21: 0
    • Extended requests 20: 0
    • Unextended requests 20: 1
    • Extended requests 19: 1
    • Unextended requests 19: 0
    • Extended requests 18: 2
    • Unextended requests 18: 0
    • Extended requests 17: 0
    • Unextended requests 17: 1
    • Extended requests 16: 4
    • Unextended requests 16: 1
    • Extended requests 15: 1
    • Unextended requests 15: 0
    15
    • Extended requests 14: 1
    • Unextended requests 14: 2
    • Extended requests 13: 0
    • Unextended requests 13: 4
    • Extended requests 12: 1
    • Unextended requests 12: 2
    • Extended requests 11: 0
    • Unextended requests 11: 2
    • Extended requests 10: 1
    • Unextended requests 10: 0
    • Extended requests 9: 2
    • Unextended requests 9: 2
    • Extended requests 8: 2
    • Unextended requests 8: 2
    • Extended requests 7: 1
    • Unextended requests 7: 1
    • Extended requests 6: 1
    • Unextended requests 6: 1
    • Extended requests 5: 1
    • Unextended requests 5: 2
    5
    • Extended requests 4: 4
    • Unextended requests 4: 5
    • Extended requests 3: 3
    • Unextended requests 3: 1
    • Extended requests 2: 2
    • Unextended requests 2: 6
    • Extended requests 1: 5
    • Unextended requests 1: 2
    • Extended requests 0: 17
    • Unextended requests 0: 9
    • Extended requests -1: 2
    • Unextended requests -1: 1
    • Extended requests -2: 0
    • Unextended requests -2: 0
    • Extended requests -3: 0
    • Unextended requests -3: 0
    • Extended requests -4: 0
    • Unextended requests -4: 0
    • Extended requests -5: 1
    • Unextended requests -5: 0
    -5
    • Extended requests -6: 1
    • Unextended requests -6: 0
    • Extended requests -7: 0
    • Unextended requests -7: 0
    • Extended requests -8: 0
    • Unextended requests -8: 0
    • Extended requests -9: 0
    • Unextended requests -9: 0
    • Extended requests -10: 0
    • Unextended requests -10: 0
    • Extended requests -11: 0
    • Unextended requests -11: 0
    • Extended requests -12: 0
    • Unextended requests -12: 0
    • Extended requests -13: 0
    • Unextended requests -13: 0
    • Extended requests -14: 0
    • Unextended requests -14: 0
    • Extended requests -15: 1
    • Unextended requests -15: 0
    -15
    • Extended requests -16: 0
    • Unextended requests -16: 0
    • Extended requests -17: 0
    • Unextended requests -17: 0
    • Extended requests -18: 0
    • Unextended requests -18: 0
    • Extended requests -19: 0
    • Unextended requests -19: 0
    • Extended requests -20: 0
    • Unextended requests -20: 0
    • Extended requests -21: 0
    • Unextended requests -21: 0
    • Extended requests -22: 0
    • Unextended requests -22: 0
    • Extended requests -23: 0
    • Unextended requests -23: 0
    • Extended requests -24: 0
    • Unextended requests -24: 0
    • Extended requests -25: 2
    • Unextended requests -25: 0
    -25
    • Extended requests -26: 0
    • Unextended requests -26: 0
    • Extended requests -27: 0
    • Unextended requests -27: 0
    • Extended requests -28: 1
    • Unextended requests -28: 0
    • Extended requests -29: 0
    • Unextended requests -29: 0
    • Extended requests -30: 0
    • Unextended requests -30: 0

When I asked on Twitter about agencies with a history of handling OIA requests poorly, I was told that the Ministry of Foreign Affairs and Trade were "serial delayers". Sure enough, more than half of the requests MFAT handled in this period had their original deadlines extended.

Unfortunately, MFAT didn't release information on how much time was remaining when a notice of extension was sent, so I can't tell if it was also common for this to happen on the due date.

Those requests that were extended were about as likely to receive a response on the due date as unextended requests. However, looking at the lengths of the extensions taken by MFAT, every one of them was a multiple of 5 working days:

Ministry of Foreign Affairs and Trade OIA extension length 2017-07-01 – 2017-12-30
Requests
  • 0
  • 10
  • 20
  • 30
Working days
    • Extension duration 0: 0
    0
    • Extension duration 1: 0
    • Extension duration 2: 0
    • Extension duration 3: 0
    • Extension duration 4: 0
    • Extension duration 5: 0
    5
    • Extension duration 6: 0
    • Extension duration 7: 0
    • Extension duration 8: 0
    • Extension duration 9: 0
    • Extension duration 10: 2
    10
    • Extension duration 11: 0
    • Extension duration 12: 0
    • Extension duration 13: 0
    • Extension duration 14: 0
    • Extension duration 15: 3
    15
    • Extension duration 16: 0
    • Extension duration 17: 0
    • Extension duration 18: 0
    • Extension duration 19: 0
    • Extension duration 20: 16
    20
    • Extension duration 21: 0
    • Extension duration 22: 0
    • Extension duration 23: 0
    • Extension duration 24: 0
    • Extension duration 25: 10
    25
    • Extension duration 26: 0
    • Extension duration 27: 0
    • Extension duration 28: 0
    • Extension duration 29: 0
    • Extension duration 30: 22
    30
    • Extension duration 31: 0
    • Extension duration 32: 0
    • Extension duration 33: 0
    • Extension duration 34: 0
    • Extension duration 35: 0
    35
    • Extension duration 36: 0
    • Extension duration 37: 0
    • Extension duration 38: 0
    • Extension duration 39: 0
    • Extension duration 40: 1
    40
    • Extension duration 41: 0
    • Extension duration 42: 0
    • Extension duration 43: 0
    • Extension duration 44: 0
    • Extension duration 45: 5
    45
    • Extension duration 46: 0
    • Extension duration 47: 0
    • Extension duration 48: 0
    • Extension duration 49: 0
    • Extension duration 50: 1
    50

This implies that MFAT was only estimating the remaining time in terms of weeks, rather than days. Their extended requests were no more likely to be responded to before the due date than those that weren't extended, so it seems either some of these requests were delayed or Parkinson's Law may have been at play: "work expands so as to fill the time available for its completion".

If that was the case, then MFAT's practice of always extending OIA requests by a number of weeks instead of a specific number of days may be effectively delaying these OIA requests more than is necessary.

MFAT's OIA practices underwent a fair amount of criticism during the negotiations of the Trans-Pacific Partnership Agreement (TPPA). In 2015, they lost a court case to Jane Kelsey and several others, where it was found that their blanket refusal to release information was unlawful.

Accident Compensation Corporation

Data

I had some trouble with ACC's response, at first. The reason being that, though I had explicitly requested that the data be released as a spreadsheet, their response (which arrived on the afternoon of the due date) only included a PDF of the data. Though it thankfully wasn't one of the dreaded image-based PDFs that are so commonly used, it still wasn't useful to me until it was converted into a spreadsheet.

Section 16(2) of the OIA requires that agencies must release data in the format preferred by the requester unless there is a specific good reason not to. ACC had not done this, and when I asked them again to release it as a spreadsheet they took a further 8 working days to do so.

This is an example of the kind of response that, while it could be considered overdue, will certainly have been recorded as having been completed within the legislated timeframe. Because of how it avoids being recorded, there's no real way to know how common or how rare this sort of case is, within ACC or across other agencies.

Looking at the data they have recorded, however, it looks like ACC have generally done very well at responding to requests within the legislated timeframe.

ACC OIA days remaining 2017-07-01 – 2017-12-30
Extended requests
Unextended requests
Requests
  • 0
  • 40
  • 80
  • 120
Working days
    • Extended requests 20: 0
    • Unextended requests 20: 0
    20
    • Extended requests 19: 0
    • Unextended requests 19: 4
    • Extended requests 18: 0
    • Unextended requests 18: 7
    • Extended requests 17: 0
    • Unextended requests 17: 7
    • Extended requests 16: 0
    • Unextended requests 16: 5
    • Extended requests 15: 0
    • Unextended requests 15: 17
    15
    • Extended requests 14: 0
    • Unextended requests 14: 11
    • Extended requests 13: 0
    • Unextended requests 13: 17
    • Extended requests 12: 0
    • Unextended requests 12: 14
    • Extended requests 11: 0
    • Unextended requests 11: 7
    • Extended requests 10: 0
    • Unextended requests 10: 16
    10
    • Extended requests 9: 0
    • Unextended requests 9: 15
    • Extended requests 8: 0
    • Unextended requests 8: 11
    • Extended requests 7: 1
    • Unextended requests 7: 15
    • Extended requests 6: 0
    • Unextended requests 6: 11
    • Extended requests 5: 3
    • Unextended requests 5: 20
    5
    • Extended requests 4: 0
    • Unextended requests 4: 15
    • Extended requests 3: 0
    • Unextended requests 3: 30
    • Extended requests 2: 6
    • Unextended requests 2: 32
    • Extended requests 1: 6
    • Unextended requests 1: 22
    • Extended requests 0: 19
    • Unextended requests 0: 93
    0
    • Extended requests -1: 0
    • Unextended requests -1: 0
    • Extended requests -2: 0
    • Unextended requests -2: 1
    • Extended requests -3: 0
    • Unextended requests -3: 0
    • Extended requests -4: 1
    • Unextended requests -4: 0
    • Extended requests -5: 0
    • Unextended requests -5: 0
    -5
    • Extended requests -6: 0
    • Unextended requests -6: 0
    • Extended requests -7: 0
    • Unextended requests -7: 0
    • Extended requests -8: 0
    • Unextended requests -8: 1
    • Extended requests -9: 0
    • Unextended requests -9: 0
    • Extended requests -10: 0
    • Unextended requests -10: 0
    -10
    • Extended requests -11: 0
    • Unextended requests -11: 0
    • Extended requests -12: 0
    • Unextended requests -12: 0
    • Extended requests -13: 0
    • Unextended requests -13: 0
    • Extended requests -14: 0
    • Unextended requests -14: 0
    • Extended requests -15: 0
    • Unextended requests -15: 0
    -15
    • Extended requests -16: 0
    • Unextended requests -16: 0
    • Extended requests -17: 0
    • Unextended requests -17: 0
    • Extended requests -18: 0
    • Unextended requests -18: 0
    • Extended requests -19: 0
    • Unextended requests -19: 1
    • Extended requests -20: 0
    • Unextended requests -20: 1
    -20
    • Extended requests -21: 0
    • Unextended requests -21: 0
    • Extended requests -22: 0
    • Unextended requests -22: 0
    • Extended requests -23: 0
    • Unextended requests -23: 0
    • Extended requests -24: 0
    • Unextended requests -24: 1
    • Extended requests -25: 0
    • Unextended requests -25: 0
    -25
    • Extended requests -26: 0
    • Unextended requests -26: 0
    • Extended requests -27: 0
    • Unextended requests -27: 0
    • Extended requests -28: 0
    • Unextended requests -28: 0
    • Extended requests -29: 0
    • Unextended requests -29: 0
    • Extended requests -30: 0
    • Unextended requests -30: 0
    -30
    • Extended requests -31: 0
    • Unextended requests -31: 0
    • Extended requests -32: 0
    • Unextended requests -32: 1
    • Extended requests -33: 0
    • Unextended requests -33: 0
    • Extended requests -34: 0
    • Unextended requests -34: 0
    • Extended requests -35: 0
    • Unextended requests -35: 0
    -35

Out of over 400 OIA requests, only seven were completed outside the legislated timeframe. Unfortunately, some of those seven were significantly outside the legislated timeframe. The worst of these requests was over six weeks overdue, and ACC did not record sending a notice of extension to the requester to let them know it would be late and how much more time was necessary.

ACC extended about 11% of requests within this period. Looking at the pattern of extension lengths, there are clear spikes at 10 and 15 working days. Combined, these account for about half of all extensions within this period:

ACC OIA extension length 2017-07-01 – 2017-12-30
Requests
  • 0
  • 5
  • 10
  • 15
  • 20
Working days
    • Extension duration 0: 0
    0
    • Extension duration 1: 0
    • Extension duration 2: 0
    • Extension duration 3: 0
    • Extension duration 4: 0
    • Extension duration 5: 1
    5
    • Extension duration 6: 1
    • Extension duration 7: 1
    • Extension duration 8: 1
    • Extension duration 9: 2
    • Extension duration 10: 15
    10
    • Extension duration 11: 1
    • Extension duration 12: 1
    • Extension duration 13: 0
    • Extension duration 14: 2
    • Extension duration 15: 4
    15
    • Extension duration 16: 2
    • Extension duration 17: 1
    • Extension duration 18: 0
    • Extension duration 19: 0
    • Extension duration 20: 0
    20
    • Extension duration 21: 2
    • Extension duration 22: 0
    • Extension duration 23: 0
    • Extension duration 24: 0
    • Extension duration 25: 0
    25
    • Extension duration 26: 0
    • Extension duration 27: 0
    • Extension duration 28: 0
    • Extension duration 29: 0
    • Extension duration 30: 1
    30

ACC was far more likely to complete extended requests on their due date (53%) than unextended requests (25%). This could reflect well-estimated extensions, like in the case of the SSC, but given the spikes at extensions of two and three weeks long it seems more likely to me that ACC has similar issues to MFAT in this regard. If the duration of each extension better reflected the actual amount of work remaining, it's quite possible that would have led to responses being sent earlier.

Like the Ministry of Justice, ACC never sent a notice of extension after any request's original due date had passed. Though to be fair only three unextended requests passed their due date without being extended:

ACC OIA days remaining when extended 2017-07-01 – 2017-12-30
Requests
  • 0
  • 5
  • 10
  • 15
  • 20
Working days
    • Working days remaining when extended 15: 0
    15
    • Working days remaining when extended 14: 0
    • Working days remaining when extended 13: 0
    • Working days remaining when extended 12: 1
    • Working days remaining when extended 11: 0
    • Working days remaining when extended 10: 0
    10
    • Working days remaining when extended 9: 1
    • Working days remaining when extended 8: 0
    • Working days remaining when extended 7: 0
    • Working days remaining when extended 6: 0
    • Working days remaining when extended 5: 2
    5
    • Working days remaining when extended 4: 1
    • Working days remaining when extended 3: 4
    • Working days remaining when extended 2: 4
    • Working days remaining when extended 1: 11
    • Working days remaining when extended 0: 12
    0

In July 2016, Mark Honeychurch from the Society for Science Based Healthcare I am the chair of the Society for Science Based Healthcare, and have made various OIA requests regarding issues such as ACC's funding of acupuncture, though I wasn't directly involved in these particular discussions with ACC. was told by ACC that, instead of asking for information through an Official Information Act request, he should make a "media request".

This revealed a lack of understanding on ACC's part of what the Official Information Act does, which likely stemmed from the fact that they have a specific internal process for correspondence labelled as "OIA requests".

The OIA applies to every request for information received by a government agency. It doesn't matter who asked for it, how they asked for it, or whether or not they mentioned the OIA. In all cases, the agency has exactly the same legal obligations.

In the Not a game of hide and seek report, it was noted that many agencies treated requests differently depending on who the requester was. Recommendation 27 of the report was that:

Agencies should review their practices to ensure that the identity of the requester, their mode of engagement, or any practices do not impinge on the requirements to make a decision that is appropriate under the OIA and communicate it to the requester 'as soon as reasonably practicable'.

When ACC was asked for information as a "media request", their media advisor responded with the requested information within an hour. While that response time is certainly laudable, there's no reason why such a quick response should be restricted to only certain classes of requesters. It's also not clear whether or not ACC includes these requests as OIA requests for the purpose of their records and the statistics published by the SSC.

Ministry of Business, Innovation and Employment

Data

MBIE extended my request by 20 working days, and sent the eventual response on the afternoon of the due date. The SSC's data showed they completed 782 OIA requests within Q3-4 of 2017, 97% of which were completed within the legislated timeframe.

In their response to me, over 100 requests didn't have a completion date recorded, due to being either transferred (24 requests) or withdrawn (79 requests). There were also a dozen or so rows where there had clearly been errors in data entry (e.g. requests that were reported as being extended before they had been received, or after they had been completed). These requests, which I've had to exclude from my analysis due to the missing information, made up 14% of the data reported by MBIE.

Of the remaining 710 requests, most of them were completed no earlier than the day before the due date. The spike on the due date itself represents over a third of all MBIE's responses.

Ministry of Business, Innovation and Employment OIA days remaining 2017-07-01 – 2017-12-30
Extended requests
Unextended requests
Requests
  • 0
  • 100
  • 200
  • 300
Working days
    • Extended requests 20: 0
    • Unextended requests 20: 3
    20
    • Extended requests 19: 0
    • Unextended requests 19: 8
    • Extended requests 18: 1
    • Unextended requests 18: 7
    • Extended requests 17: 0
    • Unextended requests 17: 2
    • Extended requests 16: 2
    • Unextended requests 16: 10
    • Extended requests 15: 0
    • Unextended requests 15: 9
    • Extended requests 14: 3
    • Unextended requests 14: 7
    • Extended requests 13: 1
    • Unextended requests 13: 6
    • Extended requests 12: 2
    • Unextended requests 12: 14
    • Extended requests 11: 3
    • Unextended requests 11: 12
    • Extended requests 10: 3
    • Unextended requests 10: 15
    10
    • Extended requests 9: 3
    • Unextended requests 9: 18
    • Extended requests 8: 5
    • Unextended requests 8: 19
    • Extended requests 7: 3
    • Unextended requests 7: 15
    • Extended requests 6: 5
    • Unextended requests 6: 29
    • Extended requests 5: 4
    • Unextended requests 5: 20
    • Extended requests 4: 5
    • Unextended requests 4: 29
    • Extended requests 3: 7
    • Unextended requests 3: 31
    • Extended requests 2: 7
    • Unextended requests 2: 29
    • Extended requests 1: 17
    • Unextended requests 1: 74
    • Extended requests 0: 96
    • Unextended requests 0: 158
    0
    • Extended requests -1: 3
    • Unextended requests -1: 5
    • Extended requests -2: 2
    • Unextended requests -2: 1
    • Extended requests -3: 1
    • Unextended requests -3: 0
    • Extended requests -4: 1
    • Unextended requests -4: 0
    • Extended requests -5: 0
    • Unextended requests -5: 2
    • Extended requests -6: 0
    • Unextended requests -6: 0
    • Extended requests -7: 0
    • Unextended requests -7: 0
    • Extended requests -8: 1
    • Unextended requests -8: 0
    • Extended requests -9: 0
    • Unextended requests -9: 0
    • Extended requests -10: 0
    • Unextended requests -10: 1
    -10
    • Extended requests -11: 1
    • Unextended requests -11: 0
    • Extended requests -12: 0
    • Unextended requests -12: 0
    • Extended requests -13: 1
    • Unextended requests -13: 0
    • Extended requests -14: 0
    • Unextended requests -14: 0
    • Extended requests -15: 0
    • Unextended requests -15: 0
    • Extended requests -16: 0
    • Unextended requests -16: 0
    • Extended requests -17: 0
    • Unextended requests -17: 0
    • Extended requests -18: 1
    • Unextended requests -18: 1
    • Extended requests -19: 0
    • Unextended requests -19: 1
    • Extended requests -20: 0
    • Unextended requests -20: 0
    -20

Very few requests — only 28 — were overdue. However, some of those requests appear to have been very overdue. The worst offender in that regard is a request that was received in October 2016, and extended by 33 working days in November that same year so that its new due date The Official Information Act effectively takes a break over summer, as dates from the 25th of December to the 15th of January aren't considered to be "working days". This means requests received near the end of the year typically take longer to be completed. was the 20th of January 2017. The actual response didn't arrive until the 29th of August, over seven months overdue.

As well as this rather extreme example, MBIE also had requests overdue by 117, 62, 51, 44, and 27 working days. The three requests that were the least overdue were not extended, even though MBIE had plenty of time after their due date during which they could have told the requester how much time was remaining.

MBIE extended a quarter of all their requests, and extended requests were far more likely to receive a response on the due date itself: 53%, compared with 30% of unextended requests.

Of course, this could just be due to MBIE being good at accurately estimating the time remaining when extending a request. Here's the distribution of the lengths of extensions they granted:

Ministry of Business, Innovation and Employment OIA extension length 2017-07-01 – 2017-12-30
Requests
  • 0
  • 20
  • 40
  • 60
Working days
    • Extension duration 0: 2
    0
    • Extension duration 1: 0
    • Extension duration 2: 1
    • Extension duration 3: 2
    • Extension duration 4: 1
    • Extension duration 5: 6
    • Extension duration 6: 3
    • Extension duration 7: 1
    • Extension duration 8: 1
    • Extension duration 9: 4
    • Extension duration 10: 26
    10
    • Extension duration 11: 3
    • Extension duration 12: 3
    • Extension duration 13: 8
    • Extension duration 14: 3
    • Extension duration 15: 10
    • Extension duration 16: 2
    • Extension duration 17: 1
    • Extension duration 18: 4
    • Extension duration 19: 4
    • Extension duration 20: 50
    20
    • Extension duration 21: 1
    • Extension duration 22: 10
    • Extension duration 23: 2
    • Extension duration 24: 1
    • Extension duration 25: 3
    • Extension duration 26: 3
    • Extension duration 27: 3
    • Extension duration 28: 1
    • Extension duration 29: 3
    • Extension duration 30: 8
    30
    • Extension duration 31: 1
    • Extension duration 32: 0
    • Extension duration 33: 2
    • Extension duration 34: 3
    • Extension duration 35: 2
    • Extension duration 36: 0
    • Extension duration 37: 0
    • Extension duration 38: 0
    • Extension duration 39: 1
    • Extension duration 40: 1
    40

As seen with many other agencies, there are large spikes at extensions of a whole number of weeks. Extensions of 20 and 10 working days were by far the most common, comprising 42% of all extensions. Requests extended by these durations were less likely than other extended requests to receive a response on the due date (40% vs 60%).

As with the Ministry of Foreign Affairs and Trade, it seems likely that MBIE's practices of often extending OIA requests by a whole number of weeks is delaying their responses more than necessary.

Looking at the amount of time remaining when MBIE sent requesters a notification of extension, it's clear that this is also often left until the last day:

MBIE OIA days remaining when extended 2017-07-01 – 2017-12-30
Requests
  • 0
  • 20
  • 40
  • 60
  • 80
Working days
    • Working days remaining when extended 20: 0
    20
    • Working days remaining when extended 19: 0
    • Working days remaining when extended 18: 1
    • Working days remaining when extended 17: 0
    • Working days remaining when extended 16: 0
    • Working days remaining when extended 15: 0
    15
    • Working days remaining when extended 14: 3
    • Working days remaining when extended 13: 1
    • Working days remaining when extended 12: 1
    • Working days remaining when extended 11: 5
    • Working days remaining when extended 10: 4
    10
    • Working days remaining when extended 9: 3
    • Working days remaining when extended 8: 5
    • Working days remaining when extended 7: 8
    • Working days remaining when extended 6: 3
    • Working days remaining when extended 5: 3
    5
    • Working days remaining when extended 4: 6
    • Working days remaining when extended 3: 10
    • Working days remaining when extended 2: 13
    • Working days remaining when extended 1: 24
    • Working days remaining when extended 0: 63
    0
    • Working days remaining when extended -1: 4
    • Working days remaining when extended -2: 1
    • Working days remaining when extended -3: 1
    • Working days remaining when extended -4: 0
    • Working days remaining when extended -5: 0
    -5
    • Working days remaining when extended -6: 0
    • Working days remaining when extended -7: 1
    • Working days remaining when extended -8: 0
    • Working days remaining when extended -9: 0
    • Working days remaining when extended -10: 0
    -10
    • Working days remaining when extended -11: 0
    • Working days remaining when extended -12: 0
    • Working days remaining when extended -13: 1
    • Working days remaining when extended -14: 1
    • Working days remaining when extended -15: 0
    -15
    • Working days remaining when extended -16: 0
    • Working days remaining when extended -17: 0
    • Working days remaining when extended -18: 0
    • Working days remaining when extended -19: 1
    • Working days remaining when extended -20: 0
    -20

As with the final responses to requests, MBIE sent more than half of their notices of extension no earlier than the day before the due date.

With requests requiring extensions almost always requiring many weeks more time to complete, it seems very unlikely that MBIE has been notifying requesters of these extensions as soon as reasonably practicable, as might be expected of them. If you need another four weeks to complete something when the due date arrives, either you've known for a while that you'll need more time or you've just let it sit until the last minute.

Ministry of Social Development

Data

The statistics published by the State Services Commission show that MSD completed 747 OIA requests within Q3-4 of 2017, 97% of which were completed within the legislated timeframe.

Unfortunately, they only record detailed statistics for one third of these requests, as the remainder are considered "media requests" by MSD and handled through a different internal process. They also do not record data on OIA requests that were answered within 24 hours, so they only released information on 241 OIA requests to me.

In analysing this data, I found that MSD uses an unusual method of responding to OIA requests. The approach is uncommon and unexpected enough that I've seen many people, including journalists familiar with the process of making OIA requests, question its legality when they first encounter it.

The thing is, the OIA's 20 working days due date technically doesn't apply to the information that's been requested. Instead, it applies to sending the requester a notice of the decision made on the request. This means it's legal for agencies to tell the requester that they have decided to provide the information they requested, but not actually provide that information until a later date.

Just over a third of the OIA requests included in the data released by MSD seem to have used this method. Technically, the legal countdown to the due date stops once a notice of decision is sent, so keep in mind when looking at this graph of the number of days remaining when MSD "completed" their requests that it doesn't entirely reflect when they released information that had been requested of them:

MSD OIA days remaining 2017-07-01 – 2017-12-30
Extended requests
Unextended requests
Requests
  • 0
  • 30
  • 60
  • 90
Working days
    • Extended requests 20: 1
    • Unextended requests 20: 2
    20
    • Extended requests 19: 0
    • Unextended requests 19: 3
    • Extended requests 18: 0
    • Unextended requests 18: 6
    • Extended requests 17: 0
    • Unextended requests 17: 3
    • Extended requests 16: 0
    • Unextended requests 16: 6
    • Extended requests 15: 3
    • Unextended requests 15: 7
    • Extended requests 14: 1
    • Unextended requests 14: 5
    • Extended requests 13: 0
    • Unextended requests 13: 2
    • Extended requests 12: 0
    • Unextended requests 12: 2
    • Extended requests 11: 1
    • Unextended requests 11: 6
    • Extended requests 10: 1
    • Unextended requests 10: 1
    10
    • Extended requests 9: 0
    • Unextended requests 9: 3
    • Extended requests 8: 0
    • Unextended requests 8: 9
    • Extended requests 7: 1
    • Unextended requests 7: 3
    • Extended requests 6: 0
    • Unextended requests 6: 5
    • Extended requests 5: 0
    • Unextended requests 5: 6
    • Extended requests 4: 1
    • Unextended requests 4: 6
    • Extended requests 3: 3
    • Unextended requests 3: 3
    • Extended requests 2: 1
    • Unextended requests 2: 8
    • Extended requests 1: 4
    • Unextended requests 1: 17
    • Extended requests 0: 29
    • Unextended requests 0: 53
    0
    • Extended requests -1: 2
    • Unextended requests -1: 1
    • Extended requests -2: 2
    • Unextended requests -2: 0
    • Extended requests -3: 0
    • Unextended requests -3: 0
    • Extended requests -4: 0
    • Unextended requests -4: 1
    • Extended requests -5: 0
    • Unextended requests -5: 1
    • Extended requests -6: 0
    • Unextended requests -6: 0
    • Extended requests -7: 0
    • Unextended requests -7: 1
    • Extended requests -8: 0
    • Unextended requests -8: 2
    • Extended requests -9: 0
    • Unextended requests -9: 1
    • Extended requests -10: 0
    • Unextended requests -10: 0
    -10
    • Extended requests -11: 0
    • Unextended requests -11: 0
    • Extended requests -12: 0
    • Unextended requests -12: 0
    • Extended requests -13: 0
    • Unextended requests -13: 0
    • Extended requests -14: 0
    • Unextended requests -14: 0
    • Extended requests -15: 0
    • Unextended requests -15: 0
    • Extended requests -16: 0
    • Unextended requests -16: 0
    • Extended requests -17: 0
    • Unextended requests -17: 0
    • Extended requests -18: 0
    • Unextended requests -18: 0
    • Extended requests -19: 0
    • Unextended requests -19: 0
    • Extended requests -20: 0
    • Unextended requests -20: 0
    -20

There are 11 more overdue requests that don't fit on the graph, the worst of which was 99 working days overdue. That request, which wasn't extended, was due for a response in early November 2017 but didn't receive a response until over five months later on the 20th of April 2018.

Also, despite the figure of 97% of requests completed within the legislated timeframe reported to the SSC, MSD responded to only 56% of their OIA requests before the due date.

In those cases where MSD sent a decision before they released the requested information, they set themselves an internal due date for when they aimed to send the information. There is no due date requirement in the legislation for this, but they did include their internal due dates in the information they released.

Here's how MSD performed when it came to sending that information before their own internal due date:

MSD OIA days remaining (internal deadline) 2017-07-01 – 2017-12-30
Extended requests
Unextended requests
Requests
  • 0
  • 4
  • 8
  • 12
  • 16
  • 20
Working days
    • Extended requests 15: 0
    • Unextended requests 15: 0
    15
    • Extended requests 14: 0
    • Unextended requests 14: 0
    • Extended requests 13: 0
    • Unextended requests 13: 0
    • Extended requests 12: 0
    • Unextended requests 12: 0
    • Extended requests 11: 0
    • Unextended requests 11: 1
    • Extended requests 10: 0
    • Unextended requests 10: 0
    10
    • Extended requests 9: 0
    • Unextended requests 9: 0
    • Extended requests 8: 1
    • Unextended requests 8: 2
    • Extended requests 7: 0
    • Unextended requests 7: 3
    • Extended requests 6: 0
    • Unextended requests 6: 2
    • Extended requests 5: 0
    • Unextended requests 5: 3
    5
    • Extended requests 4: 0
    • Unextended requests 4: 3
    • Extended requests 3: 0
    • Unextended requests 3: 5
    • Extended requests 2: 4
    • Unextended requests 2: 4
    • Extended requests 1: 4
    • Unextended requests 1: 8
    • Extended requests 0: 5
    • Unextended requests 0: 8
    0
    • Extended requests -1: 5
    • Unextended requests -1: 1
    • Extended requests -2: 0
    • Unextended requests -2: 0
    • Extended requests -3: 1
    • Unextended requests -3: 0
    • Extended requests -4: 0
    • Unextended requests -4: 2
    • Extended requests -5: 0
    • Unextended requests -5: 1
    -5
    • Extended requests -6: 1
    • Unextended requests -6: 1
    • Extended requests -7: 0
    • Unextended requests -7: 1
    • Extended requests -8: 0
    • Unextended requests -8: 0
    • Extended requests -9: 1
    • Unextended requests -9: 0
    • Extended requests -10: 1
    • Unextended requests -10: 0
    -10
    • Extended requests -11: 0
    • Unextended requests -11: 0
    • Extended requests -12: 0
    • Unextended requests -12: 0
    • Extended requests -13: 1
    • Unextended requests -13: 0
    • Extended requests -14: 0
    • Unextended requests -14: 0
    • Extended requests -15: 0
    • Unextended requests -15: 0
    -15
    • Extended requests -16: 0
    • Unextended requests -16: 0
    • Extended requests -17: 2
    • Unextended requests -17: 1
    • Extended requests -18: 0
    • Unextended requests -18: 0
    • Extended requests -19: 0
    • Unextended requests -19: 0
    • Extended requests -20: 0
    • Unextended requests -20: 0
    -20
    • Extended requests -21: 0
    • Unextended requests -21: 1
    • Extended requests -22: 1
    • Unextended requests -22: 0
    • Extended requests -23: 0
    • Unextended requests -23: 0
    • Extended requests -24: 0
    • Unextended requests -24: 0
    • Extended requests -25: 0
    • Unextended requests -25: 0
    -25

As with the graph of how quickly MSD sent decision notices, there are a number of requests that are so overdue they don't fit on the graph. For seven requests, MSD took more than 20 working days longer than they had allowed themselves before finally releasing the information. The most overdue of these requests took 77 working days longer than they had expected.

MSD had received that request in July 2017, and technically answered it before the due date in August. However, they didn't actually send the requested information until December.

When I asked on Twitter about bad OIA experiences people have had with agencies, a producer for RNZ told me she didn't think she'd received a single OIA from MSD within 20 days for the past couple of years.

Looking at how long it took for MSD to fully complete OIA requests (i.e. if they decide to grant a request, the time it took for them to release the information), it turns out MSD were just as likely to release information within 20 working days as they were to take longer.

Ministry of Health

Data

The Ministry of Health extended my request at 4:55pm on its original due date, pushing the deadline out another 21 working days. The eventual response squeezed in at 4:39pm on the new due date.

In Q3-4 2017, the Ministry received 354 requests, and they replied to 89% of them within the legislated timeframe. Looking at the distribution of response times, the spike at the due date is less pronounced than for many other agencies:

Ministry of Health OIA days remaining 2017-07-01 – 2017-12-30
Extended requests
Unextended requests
Requests
  • 0
  • 20
  • 40
  • 60
  • 80
Working days
    • Extended requests 20: 0
    • Unextended requests 20: 8
    20
    • Extended requests 19: 0
    • Unextended requests 19: 4
    • Extended requests 18: 0
    • Unextended requests 18: 17
    • Extended requests 17: 0
    • Unextended requests 17: 6
    • Extended requests 16: 2
    • Unextended requests 16: 8
    • Extended requests 15: 1
    • Unextended requests 15: 9
    • Extended requests 14: 1
    • Unextended requests 14: 6
    • Extended requests 13: 1
    • Unextended requests 13: 3
    • Extended requests 12: 1
    • Unextended requests 12: 11
    • Extended requests 11: 1
    • Unextended requests 11: 4
    • Extended requests 10: 0
    • Unextended requests 10: 9
    10
    • Extended requests 9: 0
    • Unextended requests 9: 15
    • Extended requests 8: 0
    • Unextended requests 8: 5
    • Extended requests 7: 1
    • Unextended requests 7: 14
    • Extended requests 6: 1
    • Unextended requests 6: 11
    • Extended requests 5: 3
    • Unextended requests 5: 11
    • Extended requests 4: 2
    • Unextended requests 4: 9
    • Extended requests 3: 1
    • Unextended requests 3: 19
    • Extended requests 2: 1
    • Unextended requests 2: 24
    • Extended requests 1: 4
    • Unextended requests 1: 37
    • Extended requests 0: 11
    • Unextended requests 0: 53
    0
    • Extended requests -1: 8
    • Unextended requests -1: 5
    • Extended requests -2: 1
    • Unextended requests -2: 5
    • Extended requests -3: 0
    • Unextended requests -3: 2
    • Extended requests -4: 0
    • Unextended requests -4: 2
    • Extended requests -5: 0
    • Unextended requests -5: 2
    • Extended requests -6: 0
    • Unextended requests -6: 0
    • Extended requests -7: 0
    • Unextended requests -7: 0
    • Extended requests -8: 0
    • Unextended requests -8: 0
    • Extended requests -9: 2
    • Unextended requests -9: 2
    • Extended requests -10: 1
    • Unextended requests -10: 2
    -10

They also extended a relatively small number of requests, only 12% of the total. Looking at the breakdown of the lengths of these extensions there's a very clear spike at 10 working days, corresponding to a quarter of these extensions:

Ministry of Health OIA extension length 2017-07-01 – 2017-12-30
Requests
  • 0
  • 5
  • 10
  • 15
Working days
    • Extension duration 0: 0
    0
    • Extension duration 1: 0
    • Extension duration 2: 0
    • Extension duration 3: 0
    • Extension duration 4: 1
    • Extension duration 5: 3
    5
    • Extension duration 6: 2
    • Extension duration 7: 1
    • Extension duration 8: 2
    • Extension duration 9: 0
    • Extension duration 10: 12
    10
    • Extension duration 11: 2
    • Extension duration 12: 1
    • Extension duration 13: 0
    • Extension duration 14: 2
    • Extension duration 15: 1
    15
    • Extension duration 16: 1
    • Extension duration 17: 0
    • Extension duration 18: 2
    • Extension duration 19: 1
    • Extension duration 20: 4
    20
    • Extension duration 21: 1
    • Extension duration 22: 2
    • Extension duration 23: 1
    • Extension duration 24: 0
    • Extension duration 25: 1
    25
    • Extension duration 26: 1
    • Extension duration 27: 2
    • Extension duration 28: 0
    • Extension duration 29: 0
    • Extension duration 30: 0
    30

A quarter of those notices of extension also came on the original due date:

Ministry of Health OIA days remaining when extended 2017-07-01 – 2017-12-30
Requests
  • 0
  • 5
  • 10
  • 15
Working days
    • Working days remaining when extended 15: 1
    15
    • Working days remaining when extended 14: 1
    • Working days remaining when extended 13: 1
    • Working days remaining when extended 12: 0
    • Working days remaining when extended 11: 1
    • Working days remaining when extended 10: 1
    10
    • Working days remaining when extended 9: 1
    • Working days remaining when extended 8: 4
    • Working days remaining when extended 7: 1
    • Working days remaining when extended 6: 2
    • Working days remaining when extended 5: 4
    5
    • Working days remaining when extended 4: 5
    • Working days remaining when extended 3: 4
    • Working days remaining when extended 2: 1
    • Working days remaining when extended 1: 5
    • Working days remaining when extended 0: 11
    0

The Ministry of Health's data struck me as very middle of the road. They aren't exemplary like the State Services Commission appears to be, but the issues they do have aren't at the same magnitude as many other agencies.

Yes, there's a spike of requests being received on the due date, but it's small when compared with agencies like MPI and MFAT. Lots of extended requests had the same extension length, and extension notifications were sent on the last allowable day, but it's no where near as big a problem at the Ministry of Health as it is at agencies such as MBIE.

Personally, I have experienced issues with certain parts of the Ministry of Health, such as Medsafe, not appearing to understand their obligations regarding responding to requests for information. However issues like this don't tend to get recorded, so I'm not sure how wide-reaching those problems might be within the Ministry.

Ministry of Education

Data

The Ministry of Education initially extended my request by 13 working days, notifying me of the extension the day before the initial due date. The response eventually arrived on the afternoon of the new due date.

When analysing their data, I was struck by apparent discrepancies with the statistics reported by the SSC. Those statistics say the Ministry of Education completed 285 OIA requests in Q3-4 2017, and all but four of them — just 1.4% — were completed within the legislated timeframe.

However, the data they initially released to me described 26% of the requests they handled in this period as not being completed before the due date. I also found several data entry errors, so I sent them a follow-up request to ask about these issues.

Though the Ministry of Education did respond to this second request, I didn't receive that response until six working days after it was due. They also didn't send any notice of extension to tell me that it would be late, or how much longer it would take. Strangely, they also seemed to have miscalculated the date on which their response was due, though not by enough to explain how late it was.

In explaining why their data appeared to show more requests receiving late responses than the SSC had reported, this is what I was told:

When considering requests for official information we regularly engage with requesters on their requests which can result in requests being refined/clarified. When this occurs, the statutory timeframe for providing a decision is reset, so the due date will not always be 20 working days from the date the request is received. We do this in accordance with the OIA.

Additionally, when a large volume of information is involved in a request, a decision can be made to grant the request, which satisfies the legislative requirement, and the relevant information can be released over subsequent days/weeks.

Ministry of Education

They're correct about the OIA allowing deadlines being "reset" if a request is clarified or refined, under certain conditions. Section 15(1AA) of the OIA states that a new 20 working day deadline will replace the initial one either if the requester updates a request of their own accord, or if it is updated as a result of the agency that received the request asking for an amendment or clarification within seven working days of having received the request.

I've never experienced an agency using this allowance in response to one of my requests. In those few cases where an agency has asked me to refine or clarify my request within seven working days of having received it, like MPI did with my request for their OIA response time data, they've tended to also respond well ahead of the due date.

So it sounds like the apparent discrepancy with the SSC's data is due partly to the Ministry of Education using this approach, and partly due to them using the same workaround described by MSD of not releasing information until days or weeks after they told the requester that they will grant their request.

Keeping those caveats in mind, this is how the Ministry of Education's OIA response times look with the assumption that 20 working days from the date on which they were received are allowed for unextended requests:

Ministry of Education OIA days remaining 2017-07-01 – 2017-12-30
Extended requests
Unextended requests
Requests
  • 0
  • 10
  • 20
  • 30
  • 40
  • 50
Working days
    • Extended requests 30: 0
    • Unextended requests 30: 0
    30
    • Extended requests 29: 0
    • Unextended requests 29: 0
    • Extended requests 28: 0
    • Unextended requests 28: 0
    • Extended requests 27: 0
    • Unextended requests 27: 0
    • Extended requests 26: 0
    • Unextended requests 26: 0
    • Extended requests 25: 0
    • Unextended requests 25: 0
    • Extended requests 24: 0
    • Unextended requests 24: 0
    • Extended requests 23: 0
    • Unextended requests 23: 0
    • Extended requests 22: 2
    • Unextended requests 22: 0
    • Extended requests 21: 1
    • Unextended requests 21: 0
    • Extended requests 20: 0
    • Unextended requests 20: 3
    20
    • Extended requests 19: 0
    • Unextended requests 19: 4
    • Extended requests 18: 1
    • Unextended requests 18: 7
    • Extended requests 17: 0
    • Unextended requests 17: 3
    • Extended requests 16: 0
    • Unextended requests 16: 3
    • Extended requests 15: 0
    • Unextended requests 15: 4
    • Extended requests 14: 0
    • Unextended requests 14: 5
    • Extended requests 13: 0
    • Unextended requests 13: 4
    • Extended requests 12: 0
    • Unextended requests 12: 4
    • Extended requests 11: 2
    • Unextended requests 11: 5
    • Extended requests 10: 1
    • Unextended requests 10: 3
    10
    • Extended requests 9: 4
    • Unextended requests 9: 5
    • Extended requests 8: 2
    • Unextended requests 8: 5
    • Extended requests 7: 1
    • Unextended requests 7: 1
    • Extended requests 6: 6
    • Unextended requests 6: 2
    • Extended requests 5: 5
    • Unextended requests 5: 11
    • Extended requests 4: 3
    • Unextended requests 4: 6
    • Extended requests 3: 13
    • Unextended requests 3: 12
    • Extended requests 2: 6
    • Unextended requests 2: 15
    • Extended requests 1: 8
    • Unextended requests 1: 19
    • Extended requests 0: 11
    • Unextended requests 0: 39
    0
    • Extended requests -1: 8
    • Unextended requests -1: 10
    • Extended requests -2: 5
    • Unextended requests -2: 12
    • Extended requests -3: 4
    • Unextended requests -3: 2
    • Extended requests -4: 3
    • Unextended requests -4: 2
    • Extended requests -5: 0
    • Unextended requests -5: 5
    • Extended requests -6: 2
    • Unextended requests -6: 1
    • Extended requests -7: 0
    • Unextended requests -7: 2
    • Extended requests -8: 3
    • Unextended requests -8: 0
    • Extended requests -9: 1
    • Unextended requests -9: 0
    • Extended requests -10: 1
    • Unextended requests -10: 1
    -10
    • Extended requests -11: 0
    • Unextended requests -11: 1
    • Extended requests -12: 1
    • Unextended requests -12: 3
    • Extended requests -13: 0
    • Unextended requests -13: 1
    • Extended requests -14: 0
    • Unextended requests -14: 1
    • Extended requests -15: 0
    • Unextended requests -15: 0
    • Extended requests -16: 0
    • Unextended requests -16: 0
    • Extended requests -17: 2
    • Unextended requests -17: 0
    • Extended requests -18: 0
    • Unextended requests -18: 1
    • Extended requests -19: 1
    • Unextended requests -19: 0
    • Extended requests -20: 0
    • Unextended requests -20: 0
    -20
    • Extended requests -21: 0
    • Unextended requests -21: 1
    • Extended requests -22: 1
    • Unextended requests -22: 0
    • Extended requests -23: 1
    • Unextended requests -23: 0
    • Extended requests -24: 2
    • Unextended requests -24: 0
    • Extended requests -25: 0
    • Unextended requests -25: 0
    • Extended requests -26: 0
    • Unextended requests -26: 0
    • Extended requests -27: 1
    • Unextended requests -27: 1
    • Extended requests -28: 0
    • Unextended requests -28: 0
    • Extended requests -29: 0
    • Unextended requests -29: 2
    • Extended requests -30: 0
    • Unextended requests -30: 0
    -30

Like with other agencies, there is a large spike of responses with 0 days remaining. However, the drop off after this date is slower than seen with other agencies, which I expect is due to the Ministry of Education extending deadlines under section 15(1AA) of the OIA.

Even taking that into account, though, there are still a lot of requests that appear to be more than seven working days overdue. If I were to assume that those requests are the only ones that were actually overdue, that's still 29 requests. That's a much higher proportion than the four late requests reported for this period by the SSC.

Looking at the durations of the extensions granted by the Ministry of Education, there is a significant spike at 20 working days, with smaller spikes at and around other whole numbers of weeks:

Ministry of Education OIA extension length 2017-07-01 – 2017-12-30
Requests
  • 0
  • 10
  • 20
  • 30
  • 40
Working days
    • Extension duration 0: 0
    0
    • Extension duration 1: 3
    • Extension duration 2: 0
    • Extension duration 3: 0
    • Extension duration 4: 0
    • Extension duration 5: 5
    5
    • Extension duration 6: 3
    • Extension duration 7: 3
    • Extension duration 8: 0
    • Extension duration 9: 2
    • Extension duration 10: 7
    10
    • Extension duration 11: 0
    • Extension duration 12: 0
    • Extension duration 13: 0
    • Extension duration 14: 6
    • Extension duration 15: 5
    15
    • Extension duration 16: 1
    • Extension duration 17: 1
    • Extension duration 18: 0
    • Extension duration 19: 1
    • Extension duration 20: 34
    20
    • Extension duration 21: 7
    • Extension duration 22: 10
    • Extension duration 23: 6
    • Extension duration 24: 0
    • Extension duration 25: 7
    25
    • Extension duration 26: 0
    • Extension duration 27: 1
    • Extension duration 28: 0
    • Extension duration 29: 0
    • Extension duration 30: 1
    30

The Ministry of Education seems to handle its extensions slightly differently to other agencies. In the data they released, some extensions were described in terms of weeks or months instead of days, whereas every other agency I asked for this information appeared to only record extension durations as a number of working days.

That spike at 20 working days represents just under a third of all 104 requests the Ministry of Education extended within this period. Those requests extended by 20 working days were more likely (21%) to receive a response on the last allowable day than requests extended by a different duration (15%), which seems to imply that the responses may have been delayed unnecessarily just because 20 working days was a nice round number of days to extend the request by.

The fact that many extensions were recorded as a whole number of weeks or months is consistent with the idea that the Ministry of Education preferentially selects extensions that are neat, rather than estimating the precise amount of time still required to complete a request.

The Ministry of Education also often sent notices of extension at the last minute:

Ministry of Education OIA days remaining when extended 2017-07-01 – 2017-12-30
Requests
  • 0
  • 10
  • 20
  • 30
  • 40
  • 50
Working days
    • Working days remaining when extended 15: 0
    15
    • Working days remaining when extended 14: 0
    • Working days remaining when extended 13: 0
    • Working days remaining when extended 12: 0
    • Working days remaining when extended 11: 1
    • Working days remaining when extended 10: 2
    10
    • Working days remaining when extended 9: 0
    • Working days remaining when extended 8: 0
    • Working days remaining when extended 7: 0
    • Working days remaining when extended 6: 2
    • Working days remaining when extended 5: 13
    5
    • Working days remaining when extended 4: 3
    • Working days remaining when extended 3: 6
    • Working days remaining when extended 2: 6
    • Working days remaining when extended 1: 12
    • Working days remaining when extended 0: 43
    0
    • Working days remaining when extended -1: 3
    • Working days remaining when extended -2: 2
    • Working days remaining when extended -3: 2
    • Working days remaining when extended -4: 1
    • Working days remaining when extended -5: 0
    -5
    • Working days remaining when extended -6: 0
    • Working days remaining when extended -7: 0
    • Working days remaining when extended -8: 2
    • Working days remaining when extended -9: 1
    • Working days remaining when extended -10: 0
    -10
    • Working days remaining when extended -11: 0
    • Working days remaining when extended -12: 0
    • Working days remaining when extended -13: 0
    • Working days remaining when extended -14: 0
    • Working days remaining when extended -15: 1
    -15
    • Working days remaining when extended -16: 0
    • Working days remaining when extended -17: 0
    • Working days remaining when extended -18: 0
    • Working days remaining when extended -19: 1
    • Working days remaining when extended -20: 0
    -20

It's worth remembering that some of those notices of extensions that this graph shows as being late, but less than seven working days late, may actually have been on time due to section 15(1AA) of the OIA. Even taking that into account, though, the spike at 0 working days is significant (41% of all extended requests), and a few notices of extension were not sent until weeks after the original due date.

That said, given how many unextended requests appear to have been responded to after the due date, the Ministry of Education could perhaps have sent more late notices of extension to let the requester know when exactly they should expect to receive a response that was already overdue.

Unlike MSD, the Ministry of Education didn't release any information to me about how long they took to released information in cases where it was not sent alongside the notice of the decision they had made. Unfortunately, this means I can't tell how long they typically took to release information that was delayed in this way.

Department of Corrections

Data

The Department of Corrections took quite some time to get the correct statistics to me. They extended my request by 20 working days, naturally sending the notice of extension at 4:19pm on the due date. When they finally did reply, they sent me information on more requests than they reported having handled.

After a follow-up request asking for clarification, they revealed that they'd accidentally sent me information on all their correspondence, instead of just OIA requests.

Even now, the picture painted by the Department of Corrections strikes me as odd and confusing. The distribution of response times looks very different to any other agency:

Department of Corrections OIA days remaining 2017-07-01 – 2017-12-30
Extended requests
Unextended requests
Requests
  • 0
  • 75
  • 150
  • 225
  • 300
Working days
    • Extended requests 20: 0
    • Unextended requests 20: 223
    20
    • Extended requests 19: 0
    • Unextended requests 19: 160
    • Extended requests 18: 2
    • Unextended requests 18: 117
    • Extended requests 17: 3
    • Unextended requests 17: 107
    • Extended requests 16: 4
    • Unextended requests 16: 132
    • Extended requests 15: 2
    • Unextended requests 15: 143
    • Extended requests 14: 2
    • Unextended requests 14: 113
    • Extended requests 13: 2
    • Unextended requests 13: 83
    • Extended requests 12: 0
    • Unextended requests 12: 56
    • Extended requests 11: 0
    • Unextended requests 11: 61
    • Extended requests 10: 2
    • Unextended requests 10: 44
    10
    • Extended requests 9: 1
    • Unextended requests 9: 54
    • Extended requests 8: 3
    • Unextended requests 8: 39
    • Extended requests 7: 2
    • Unextended requests 7: 38
    • Extended requests 6: 2
    • Unextended requests 6: 39
    • Extended requests 5: 3
    • Unextended requests 5: 38
    • Extended requests 4: 2
    • Unextended requests 4: 48
    • Extended requests 3: 1
    • Unextended requests 3: 49
    • Extended requests 2: 2
    • Unextended requests 2: 64
    • Extended requests 1: 1
    • Unextended requests 1: 79
    • Extended requests 0: 11
    • Unextended requests 0: 172
    0
    • Extended requests -1: 0
    • Unextended requests -1: 6
    • Extended requests -2: 1
    • Unextended requests -2: 5
    • Extended requests -3: 1
    • Unextended requests -3: 0
    • Extended requests -4: 0
    • Unextended requests -4: 0
    • Extended requests -5: 0
    • Unextended requests -5: 0
    • Extended requests -6: 0
    • Unextended requests -6: 1
    • Extended requests -7: 1
    • Unextended requests -7: 0
    • Extended requests -8: 0
    • Unextended requests -8: 0
    • Extended requests -9: 0
    • Unextended requests -9: 1
    • Extended requests -10: 1
    • Unextended requests -10: 0
    -10
    • Extended requests -11: 0
    • Unextended requests -11: 0
    • Extended requests -12: 0
    • Unextended requests -12: 1
    • Extended requests -13: 1
    • Unextended requests -13: 0
    • Extended requests -14: 0
    • Unextended requests -14: 0
    • Extended requests -15: 0
    • Unextended requests -15: 0
    • Extended requests -16: 0
    • Unextended requests -16: 0
    • Extended requests -17: 2
    • Unextended requests -17: 1
    • Extended requests -18: 0
    • Unextended requests -18: 0
    • Extended requests -19: 1
    • Unextended requests -19: 0
    • Extended requests -20: 0
    • Unextended requests -20: 0
    -20

Of course, this could just mean that they're very good at responding to OIA requests quickly. The strange thing, though, is that the distribution for their non-OIA correspondence I obtained this by calculating the difference between the initial data released and the OIA-specific data released later. These data sets don't line up perfectly, as the OIA data shows many more responses with 20 days remaining than in the data that supposedly included both OIA and general correspondence. In the graph below, that is shown as 0 instead of the actual value calculated, which was -186. looks a lot like other agencies' OIA response data:

Department of Corrections non-OIA correspondence days remaining 2017-07-01 – 2017-12-30
Extended requests
Unextended requests
Requests
  • 0
  • 100
  • 200
  • 300
  • 400
Working days
    • Extended requests 20: 1
    • Unextended requests 20: 42
    20
    • Extended requests 19: 1
    • Unextended requests 19: 27
    • Extended requests 18: 1
    • Unextended requests 18: 27
    • Extended requests 17: 1
    • Unextended requests 17: 41
    • Extended requests 16: 0
    • Unextended requests 16: 54
    • Extended requests 15: 0
    • Unextended requests 15: 44
    • Extended requests 14: 0
    • Unextended requests 14: 70
    • Extended requests 13: 1
    • Unextended requests 13: 65
    • Extended requests 12: 1
    • Unextended requests 12: 62
    • Extended requests 11: 0
    • Unextended requests 11: 66
    • Extended requests 10: 0
    • Unextended requests 10: 57
    10
    • Extended requests 9: 0
    • Unextended requests 9: 49
    • Extended requests 8: 1
    • Unextended requests 8: 53
    • Extended requests 7: 2
    • Unextended requests 7: 40
    • Extended requests 6: 0
    • Unextended requests 6: 52
    • Extended requests 5: 1
    • Unextended requests 5: 76
    • Extended requests 4: 5
    • Unextended requests 4: 92
    • Extended requests 3: 3
    • Unextended requests 3: 100
    • Extended requests 2: 5
    • Unextended requests 2: 115
    • Extended requests 1: 5
    • Unextended requests 1: 151
    • Extended requests 0: 14
    • Unextended requests 0: 329
    0
    • Extended requests -1: 1
    • Unextended requests -1: 24
    • Extended requests -2: 1
    • Unextended requests -2: 6
    • Extended requests -3: 0
    • Unextended requests -3: 7
    • Extended requests -4: 1
    • Unextended requests -4: 1
    • Extended requests -5: 0
    • Unextended requests -5: 8
    • Extended requests -6: 0
    • Unextended requests -6: 4
    • Extended requests -7: 1
    • Unextended requests -7: 6
    • Extended requests -8: 2
    • Unextended requests -8: 1
    • Extended requests -9: 1
    • Unextended requests -9: 0
    • Extended requests -10: 0
    • Unextended requests -10: 2
    -10
    • Extended requests -11: 0
    • Unextended requests -11: 0
    • Extended requests -12: 0
    • Unextended requests -12: 0
    • Extended requests -13: 0
    • Unextended requests -13: 1
    • Extended requests -14: 1
    • Unextended requests -14: 1
    • Extended requests -15: 0
    • Unextended requests -15: 1
    • Extended requests -16: 0
    • Unextended requests -16: 0
    • Extended requests -17: 0
    • Unextended requests -17: 0
    • Extended requests -18: 0
    • Unextended requests -18: 0
    • Extended requests -19: 0
    • Unextended requests -19: 0
    • Extended requests -20: 0
    • Unextended requests -20: 0
    -20

It remains a mystery to me why the Department of Corrections seems to respond to so much of their general correspondence after 20 working days. When I asked them about it, they cited Privacy Act requests as an example of the type of correspondence this could apply to.

I've made a Privacy Act request to the Department of Corrections before, and received a response at 4:09pm on the last day allowable, so this sort of behaviour is consistent with my experience. However, the Department's annual report for 2016/17 shows that Privacy Act requests were only 6% of the correspondence they received.

In any case, according to the data they released to me it looks like the Department of Corrections has been doing very well in responding quickly to OIA requests. They apparently responded to most of the requests they received with at least 13 working days remaining before the due date.

They didn't extend many requests, but those they did extend were almost always extended by 20 working days:

Department of Corrections OIA extension length 2017-07-01 – 2017-12-30
Requests
  • 0
  • 10
  • 20
  • 30
  • 40
Working days
    • Extension duration 0: 1
    0
    • Extension duration 1: 0
    • Extension duration 2: 0
    • Extension duration 3: 0
    • Extension duration 4: 0
    • Extension duration 5: 0
    5
    • Extension duration 6: 1
    • Extension duration 7: 0
    • Extension duration 8: 0
    • Extension duration 9: 0
    • Extension duration 10: 0
    10
    • Extension duration 11: 0
    • Extension duration 12: 0
    • Extension duration 13: 0
    • Extension duration 14: 0
    • Extension duration 15: 0
    15
    • Extension duration 16: 0
    • Extension duration 17: 1
    • Extension duration 18: 1
    • Extension duration 19: 5
    • Extension duration 20: 40
    20
    • Extension duration 21: 3
    • Extension duration 22: 0
    • Extension duration 23: 0
    • Extension duration 24: 0
    • Extension duration 25: 0
    25
    • Extension duration 26: 0
    • Extension duration 27: 0
    • Extension duration 28: 0
    • Extension duration 29: 0
    • Extension duration 30: 0
    30
    • Extension duration 31: 0
    • Extension duration 32: 0
    • Extension duration 33: 0
    • Extension duration 34: 0
    • Extension duration 35: 0
    35
    • Extension duration 36: 0
    • Extension duration 37: 0
    • Extension duration 38: 0
    • Extension duration 39: 0
    • Extension duration 40: 1
    40

Those notices of extension were also almost always sent on the last allowable day:

Department of Corrections OIA days remaining when extended 2017-07-01 – 2017-12-30
Requests
  • 0
  • 10
  • 20
  • 30
  • 40
Working days
    • Working days remaining when extended 10: 1
    10
    • Working days remaining when extended 9: 0
    • Working days remaining when extended 8: 0
    • Working days remaining when extended 7: 0
    • Working days remaining when extended 6: 0
    • Working days remaining when extended 5: 0
    5
    • Working days remaining when extended 4: 0
    • Working days remaining when extended 3: 0
    • Working days remaining when extended 2: 5
    • Working days remaining when extended 1: 8
    • Working days remaining when extended 0: 34
    0
    • Working days remaining when extended -1: 0
    • Working days remaining when extended -2: 0
    • Working days remaining when extended -3: 0
    • Working days remaining when extended -4: 0
    • Working days remaining when extended -5: 0
    -5
    • Working days remaining when extended -6: 1
    • Working days remaining when extended -7: 0
    • Working days remaining when extended -8: 0
    • Working days remaining when extended -9: 0
    • Working days remaining when extended -10: 0
    -10
    • Working days remaining when extended -11: 0
    • Working days remaining when extended -12: 0
    • Working days remaining when extended -13: 0
    • Working days remaining when extended -14: 0
    • Working days remaining when extended -15: 0
    -15
    • Working days remaining when extended -16: 0
    • Working days remaining when extended -17: 0
    • Working days remaining when extended -18: 0
    • Working days remaining when extended -19: 0
    • Working days remaining when extended -20: 3
    -20
    • Working days remaining when extended -21: 0
    • Working days remaining when extended -22: 1
    • Working days remaining when extended -23: 0
    • Working days remaining when extended -24: 0
    • Working days remaining when extended -25: 0
    -25

Given they are responsible for the care of some of the most disenfranchised people in New Zealand, and that they are one of the few branches of the state authorised to use force, the Department of Corrections is perhaps one of most important agencies to hold to account. Looking aside from general patterns of behaviour, the Department of Corrections has severely mishandled some OIA requests in the past.

On 8 December 2016, the Department of Corrections published a set of four reports produced by the Office of the Ombudsman, as the result of unscheduled prison visits from Crimes of Torture Act (COTA) inspectors. Though Corrections did not mention it in their press release, this was the result of a request from NZ Herald reporters in April 2016. This request had initially been refused, but the Ombudsman intervened after a complaint was lodged and recommended that the reports be released.

These reports revealed "cruel, inhuman or degrading treatment" in New Zealand prisons.

Two days later, Corrections received a request from Ti Lamusse, a researcher for advocacy group No Pride in Prisons (now People Against Prisons Aotearoa) asking them to release all of these COTA reports that had been produced from 2007 until the end of 2016. Given the conditions revealed by the other reports, there was a very clear public interest in previous reports also being made public.

Corrections initially declined the request, saying it would be too much work. When Lamusse asked that they reconsider, Corrections took a full 20 working days to decide that they could only release the reports if they were first paid $9,956.

The OIA has provisions that allows agencies to charge for OIA requests that would be particularly time consuming. There are charging guidelines that specify an acceptable rate as well as listing activities that can and cannot be charged for. For example, an agency could charge for time taken to physically redact documents, but not for the time taken to decide which parts of documents should be redacted.

Corrections justified their charge by saying it would take 132 hours for their staff to fulfil the request. This included time for "departmental consultation", which is an activity that cannot be charged for under the guidelines.

Clearly, a $9,956 charge is a dead end for an advocacy organisation.

In June 2017, Corrections published 9 of the requested reports on their website. They didn't inform Lamusse.

Following that, on the 29th and 30th of July, Corrections received eight smaller requests, each asking for the COTA reports produced for a single year. Unlike a single request for 44 reports, Corrections couldn't justify refusing a smaller request under section 18(f) of the OIA as requiring "substantial or collation or research".

One of these requests was quickly refused, as no reports had been created in that period. Another didn't receive a response until February 2018.

Though they sent no notices of extension, Corrections took 21 working days to respond to six of the remaining seven requests. Each requester was told by Corrections

As a result of a large number of requests for COTA reports over recent months, the Department has determined that it would be appropriate to publish these reports on our website. We have identified 38 COTA reports to be released and we are in the process of preparing these for publication. Is is expected all 38 of these COTA reports will be published in the near future.

As we will be publishing these reports on our website, your request is declined under section 18(e) [sic] Section 18(e) of the OIA allows requests to be refused if the requested information does not exist. The section Corrections meant to invoke was 18(d). of the OIA, because the information will soon be publicly available.

Three requesters asked when the information would be released. These questions were treated by Corrections as new OIA requests, and again they took a full 20 days to respond. The requesters were told

These reports will be released on the Department's website as soon as this process has been completed. It is currently estimated that they will be published by the end of November 2017.

There was no new word from Corrections until February 16 2018, two and a half months after they estimated and a year since they said they wouldn't release the reports unless they were paid $9,956 up front. They told each requester that the requested reports had all been made publicly available on their website.

In their press release, Corrections didn't mention that the reports had been requested a year earlier under the OIA. Rather, they said

In the interests of transparency, Corrections has today released the reports relating to inspections of these prisons.

Thankfully, as part of this process, the Office of the Ombudsman undertook that they would publish COTA reports on their website in the future. Arguably, the reports should never have been kept secret in the first place.

New Zealand Police

Request

New Zealand Police have a more complicated relationship with the OIA than other agencies covered under it. Though the Ombudsmen Act allows the Office of the Ombudsman to investigate most public sector agencies, Police are effectively exempt from this power.

This meant they could not be included in previous Chief Ombudsman Beverley Wakem's review. It also means that they cannot be investigated for patterns of behaviour regarding the Official Information Act. Instead, the Office of the Ombudsman can only investigate complaints about specific decisions made by Police about requests for information.

New Zealand Police also processes far more OIA requests than any other agency. The SSC's statistics for Q3-4 2017 show that Police completed 5,650 requests in that time, 7% of which were completed late. In terms of the number of requests handled, there's a big gap before the next agency, The Earthquake Commission processed 2,980 requests in the same period, just over half the number processed by Police. and only four agencies As well as Police and the Earthquake Commission, the Department of Corrections processed 1,906 requests in this period and Fire and Emergency New Zealand processed 1,071. The next agency is the Ministry of Business, Innovation, and Employment at 810 requests. processed over 1,000 requests in that period.

However, Police were also one of two agencies who refused my request for data on their OIA response times. The reason given for this was that:

The database we use to track requests has only a limited reporting capability that does not record sufficient information to answer your questions. It can provide a count of requests received, completed requests and the number that were completed within legislated timeframe only. It is not able to extract specific information about requests such as received dates, answered dates i.e. information you request.

I was quite surprised to hear this. Some of the recommendations in Not a Game of Hide and Seek revolved around agencies recording and monitoring their OIA compliance. Recommendations 38 and 40 of the report were:

Agencies should ensure they are counting their OIA workload and compliance rates accurately.
Agencies should ensure they have a fit for purpose OIA logging and tracking system which is easy to use and actively monitored.

As NZ Police process far more requests than any other agency I would have expected them to have a robust system in place for this. I had also been told by NZ Police in March 2017 that:

we are intending to replace Police's current information request system with a new OIA management platform. The ability to provide more comprehensive 'back end' reporting is an element of the design work which is going into planning for the new platform.

I don't know if Police are still working on the implementation of this new platform, or if the "comprehensive 'back end' reporting" described to me last year is the same "limited reporting capability" that led to them refusing my request for OIA response time data this year.

Even without a large dataset, I've observed certain patterns in my own experience with making OIA requests to Police, and seeing how they respond to others' public requests on the website FYI.org.nz. Stories of Police mishandling OIA requests are a dime a dozen, so here are a few which illustrate the patterns of behaviour I've observed over the years:

Police seem far more likely than other agencies to demand evidence that a first-time requester is eligible Section 12(1) of the OIA states that anyone who is in New Zealand, or is a citizen or permanent resident of New Zealand, is eligible to make a request under the OIA. Certain body corporates are also eligible, but requests made publicly on FYI tend to be from individuals. to make a request under the OIA. Police typically do this by demanding the requester provide them with private information before the request will be processed. For example:

In order for NZ Police to satisfy itself that you are eligible under section 12 of the Official Information Act 1982, Police requests that you provide supporting information as to eligibility, such as: your New Zealand postal address; personal email address indicating a NZ domain; NZ phone number on which I can contact you; or any identity documentation.

Though the OIA does have eligibility requirements, the guidelines are also clear that there is no need to establish eligibility:

Agencies are entitled to make reasonable enquiries to satisfy themselves that a requester is eligible to make a request under the OIA. However, the eligibility requirement is not about imposing unnecessary barriers to legitimate requests. Agencies should only query eligibility if there is a genuine need to do so, and they should be mindful of their obligation to provide reasonably assistance to requesters.

Even if the requester turns out to be ineligible to make a request under the OIA agencies should still provide a reasonable response… Agencies may also decide it doesn't really matter whether the requester is eligible or not, because they'd be perfectly happy to supply the information to any person regardless of eligibility.

Requests made online | Office of the Ombudsman

Clearly, having Police ask you for your address or identification documents can be an intimidating experience. The requester in the above example clearly expressed that "I don't feel safe giving out my personal details in this correspondence". In that case, even after they contacted Police through other channels to provide this information, Police refused their request.

In other cases, Police have used the contact details sent to them to prove eligibility to move correspondence for a request made on FYI away from public view, and to question the requester's motives and associations. Here is an example of Police doing this in 2015, when a requester asked them about their policies for dealing with transgender and intersex people. Thankfully, the requester recorded a summary of the phone conversation:

Provision of your home address and a landline phone number would certainly resolve the matter of confirming your eligibility status required under the Act and in order to keep those details private, you could phone me at…

NZ Police

[Police Inspector Roly] Williams then asked me what I wanted the information for. He asked me if it was going to have an academic use, if I was part of a group, and what my motivation was.

Sophie Buchanan

The OIA does not allow agencies to use information such as a requester's motivations, intentions, or associations to influence their decision on a request.

Another problem Police have had with responding to OIA requests is poorly trained staff telling requesters that OIA requests must be lodged in a particular way, such as by submitting a form on the NZ Police website or by going to a police station.

This is incorrect — the OIA is very explicit and unambiguous that a request "may be made in any form and communicated by any means".

When I raised this issue with the Office of the Ombudsman, I was told the legal restriction on them investigating the Police prevented them from doing anything about the issue.

However, Police did offer me an assurance in March 2017 that they would "endeavour to put matters right, and to ensure appropriate adherence with our obligations under the OIA". However, as recently as June 2018 they have still been telling requesters to submit OIA requests via a specific online form.

As with practically all agencies, it's also been common for Police to send responses to OIA requests at the last minute. In some cases, this could be justified as rushing a response to a particularly complex request instead of extending the deadline. However, I've experienced it with some requests that should have been very straightforward.

In January 2017 I asked for Police's internal OIA request processing guidelines. This is a document that anyone handling my request should have been familiar with, so that it could have quickly been found and any necessary redactions — if any — could have been made. However, I received the unredacted document at 4:52pm on the request's due date.

In February 2017, I asked about research Police had referred to in a previous request. They had stated only that it "is freely available on the internet", so I asked them to specify the research they had referred to and tell me where it was available online. At 4:03pm on the due date, Police sent this response:

The research referred to by Detective Superintendent Anderson is publicly available and widely reported on the internet. If you undertake a simple Google search of 'NZ Drug Harm Index', you will find a significant number of information sources; but more especially you should search the Ministry of Health website.

Therefore your request is refused pursuant to Section 18(d) of the Official Information Act as the information is already publicly available.

It took a follow-up request, and 13 more working days, for NZ Police to provide a picture of the URL Due to the practice of printing, signing, then scanning responses before sending them, Police didn't actually send a link. Rather, they sent an image-based PDF file containing a picture of the URL. Because it was just a picture, I couldn't click on the URL, though as I don't rely on a screen reader I could still type it out by hand. The use of inaccessible formats for responses to OIA requests is a significant problem, which I have written about previously. where I could find the specific research they had been referring to.

Police also didn't tell me that they would refuse my request for OIA response time data until the afternoon of the day the response was due.

Though I haven't been asked by Police to pay for information I've requested, Dunedin-based journalist for Stuff Hamish McNeilly told me on Twitter that when he requested information about an old case from Police they refused to release it unless they were first paid $24,000. Using the rates prescribed in the charging guidelines, this amounts to 40 full days of chargeable work in order to produce the information.

New Zealand Transport Agency

Request

Like NZ Police, NZTA also refused my request for OIA response time data, saying it would be too much work:

To provide information on individual OIA requests would require a staff member to access each individual request folder, and review and analyse all documents and correspondence relating to that request. Refining your request would not alter the time necessary to complete this task eg a shorter time range or less detail required.

NZTA

NZTA completed 307 OIA requests in Q3-4 2017. Only two of the 10 other agencies I asked for this information completed fewer requests in that period.

Like Police, it sounds as though the system NZTA has in place for recording information on how they handle OIA requests has very restricted reporting capabilities. Whether it involved manually checking OIA requests or extracting information from a database, seven agencies who handled more OIA requests The agency with the next highest number of requests was MPI, with 328. Corrections had 1,906 requests, more than six times the number NZTA completed, and was able to provide me with data on each of them. than NZTA were able to provide me with the data I had requested.

Unfortunately, without granular data, it's difficult to say much about how NZTA performs. They reported to the SSC that only three of the 307 OIA requests completed in Q3-4 2017 were not completed on time. It's not at all clear how many of those 304 remaining requests had a response sent on the last allowable day.

Comparing with data for the 2015/16 financial year, it looks as though NZTA's OIA workload and the rate at which they respond to requests on time have both increased substantially. During the full year from 1 July 2015 to 30 June 2016, NZTA had completed 495 OIA requests, and only 396 of them (80%) were completed on time.

For what it's worth, I asked a few people who make a lot of OIA requests about their experiences with NZTA, and the responses I got were all pretty positive.

Fixing the OIA

The hashtag #FixTheOIA has been a rallying point on social media for journalists, activists, and others frustrated with mishandled OIA requests. It's been common to see complaints of overdue requests, excessive redaction, and last minute extensions.

It's clear to many of us that the OIA needs reform to fix these common issues. But the specifics of that reform aren't often discussed.

How success is measured is important. Even without any negative intentions, people's behaviour invariably shapes itself to match the way we measure success. In this case, I think the way success has been measured has been a contributing factor to the issues I've shown.

For example, reporting only the proportion of requests that were completed "within the legislated timeframe" makes it look like MPI (87.2% of requests) is doing better than the Ministry of Justice (85.1%). But if you're able look closer you'll see that the Ministry of Justice responds to most of its requests with at least four working days remaining, whereas MPI responds to most of its requests on or after the due date.

If we're going to fix the Official Information Act, part of that fix has to be adequate and appropriate monitoring of behaviour like this.

The New Zealand Council for Civil Liberties recently published a proposal for how to reform the Official Information Act. Their suggestions included improved requirements for recording how OIA requests are handled.

The have also made other strong suggestions, such as removing eligibility requirements to bring the OIA in line with the related Local Government Official Information and Meetings Act (LGOIMA), and better requirements around accessibility of information and proactive publishing

Unfortunately, the current government doesn't appear to be as keen on reforming the Official Information Act as they were while in opposition. So it looks like we can expect current behaviour to continue a while longer.

In the absence of actual legal reform, however, it's still possible for agencies to improve their behaviour. I hope you'll join me in drawing attention to the types of bad behaviour I've described here, and laying complaints to the Ombudsman when you're on the receiving end of it.

Let's also hold up examples of agencies handling OIA requests well, such as the quick response I received from MPI when gathering data on this request.

If you want to make an OIA requests, but are unsure where to start, I've published a guide to making OIA requests.