Commerce Commission v Bunnings Limited
[2020] NZCA 310
•27 July 2020
| IN THE COURT OF APPEAL OF NEW ZEALAND I TE KŌTI PĪRA O AOTEAROA |
| CA59/2020 [2020] NZCA 310 |
| BETWEEN | COMMERCE COMMISSION |
| AND | BUNNINGS LIMITED |
| CA61/2020 | ||
| BETWEEN | BUNNINGS LIMITED | |
| AND | AUCKLAND DISTRICT COURT | |
| CA62/2020 | ||
| BETWEEN | BUNNINGS LIMITED | |
| AND | COMMERCE COMMISSION | |
| Hearing: | 14 and 15 July 2020 |
Court: | Kós P, Gilbert and Courtney JJ |
Counsel: | J C L Dixon QC, I M Brookie and C G Farquhar for Commerce Commission |
Judgment: | 27 July 2020 at 11 am |
JUDGMENT OF THE COURT
ALeave to appeal is granted in CA59/2020 and CA62/2020.
BThe Commission’s appeal in CA59/2020 is dismissed and Bunnings’ appeal in CA62/2020 is allowed. The statement of Murray Snowden of 30 October 2017 is inadmissible.
CBunnings’ appeal in CA61/2020 is dismissed. The summons issued by the District Court is valid.
____________________________________________________________________
REASONS OF THE COURT
(Given by Kós P)
This is a category 1 summary prosecution. Charges were laid in December 2016. They have not yet been heard. Trial is now scheduled for September of this year.
Bunnings Ltd is charged with misleading advertising in claiming it offers the “lowest prices”.[1] It pleaded not guilty in March 2017. The Commerce Commission has summoned a Mr Murray Snowden. He is the Business Efficiency Manager at Bunnings’ main competitor, Mitre 10 (New Zealand) Ltd. It is proposed he give evidence of an automated price comparison survey undertaken by Mitre 10 between June 2015 and May 2016. He produced a formal statement of evidence in October 2017. Bunnings promptly opposed the giving of that evidence on the basis that it was “(i) not relevant evidence; (ii) not reliable (given the absence of any means to test reliability); (iii) lacks probative value; (iv) is hearsay; and (v) is otherwise inadmissible under the [Evidence Act 2006]”. The Commission immediately sought a pre-trial admissibility hearing under s 78 of the Criminal Procedure Act 2011. That hearing took place in the District Court in March 2018. The hearing also involved challenges to the validity of the summons. In a judgment delivered five months later, in August 2018, the District Court Judge upheld the summons in more restricted terms and dismissed the challenge to the admissibility of Mr Snowden’s evidence.[2]
[1]Fair Trading Act 1986, ss 10 and 13. The relevant charge period is June 2014 to February 2016.
[2]Commerce Commission v Bunnings Ltd [2018] NZDC 15629 [District Court judgment].
Bunnings challenged those conclusions by appeal and judicial review in the High Court. The hearing took place in that forum in December 2018, and judgment was delivered, after numerous intervening submissions, a year later in December 2019. The High Court Judge dismissed Bunnings’ challenges to the summons, and formally to the admissibility of Mr Snowden’s statement. But she held that in the form tendered, Mr Snowden’s statement would be inadmissible, as hearsay or inadmissible opinion, unless supported by evidence “either in electronic or hard copy form, from Mitre 10’s electronic records that shows the items being compared, the dates on which they were for sale and the prices at which they were being sold”.[3]
[3]Bunnings Ltd v Commerce Commission [2019] NZHC 3340 [High Court judgment] at [59].
Now Bunnings and the Commission both seek to appeal to this Court. CA59/2020 is the Commission’s appeal on the admissibility of Mr Snowden’s statement. CA62/2020 is Bunnings’ appeal on the relevance of that statement. CA61/2020 is Bunnings’ appeal on the summons. Leave is required in the first two. Leave is not opposed by either party. We are satisfied the appeals raise a matter of general or public importance, concerning as they do the admissibility of automated electronic price comparison survey evidence. We grant leave.
The pre-trial admissibility application was made in November 2017 on the basis that it was more convenient the matter be dealt with prior to trial. It may well be thought, no doubt with the benefit of hindsight, that that was not the case after all. And that it might have been much better for all concerned, in this category 1 summary prosecution, launched in 2016 and still unheard, if the admissibility of Mr Snowden’s evidence had simply been dealt with at trial.
Because trial is due to commence in September, and because of this Court’s other commitments, we shall get straight to the nub of what really matters in this case. We shall combine to an extent the various issues tendered by counsel for decision.
The admissibility of Mr Snowden’s statement of 30 October 2017
We will start with what Mr Snowden says in his statement. What is of primary importance is the extent to which he was responsible for, and undertook, Mitre 10’s automated price comparison survey between June 2015 and May 2016.[4] In para 7.9 of his statement Mr Snowden gives a summary of the results of that survey. But for para 7.9, he would not be called.
What Mr Snowden says
[4]That is, beginning partway through the charge period, and continuing past it: see above n 1.
Mr Snowden’s evidence is offered as by a general witness of fact. He is not presented by the Commission, or qualified, as an expert. He has been with Mitre 10 since February 2014. His role includes the “development of enhanced business processes, including processes for tracking and monitoring competitors’ prices”.
He explains that the traditional method of price comparison between competitors is a manual survey. That is, employees or third party contractors identify competitors’ prices by making phone inquiries, looking up individual prices on the internet, reviewing print advertisements or visiting the competitors’ stores. In early 2015 Mitre 10 began to use a “more advanced technique” and began to “automatically collect publicly available website information to conduct price surveys. This automatic technique is sometimes referred to as ‘web data collection’ or ‘price scraping’”. Mr Snowden refers to “specific software” the companies can use to undertake data collection themselves, and there are also external IT providers who will perform this service on request.
The method allows an enormous amount of data to be obtained and reviewed. It enabled Mitre 10 to review prices for all products listed on Bunnings’ publicly available website. It is, Mr Snowden says “a vastly more comprehensive data set than what would generally be obtained using manual price surveys”. He goes on to say:
Another advantage of this approach is its reliability. Data stored on the website is automatically downloaded, so there is little room for human error. We have tested the reliability of web data collection on numerous occasions in a variety of ways in our business and have found it to be reliable.
He says the testing process has included manual checks of website information as well as store visits. The reliability of the information is important to pricing decisions that Mitre 10 has to make.
Mitre 10 used an external IT provider to compile a database of its competitor Bunnings’ product description and prices. That provider was PriceTech Ltd, and its CEO Andrew Pascoe, is also giving evidence. In his statement, which to an extent overlaps with Mr Snowden’s, he says:
PriceTech used its resources and technology to collect product information from the Bunnings website on numerous occasions from 1 June 2015 to 30 May 2016.
We collect information (including via automated processes) from the Bunnings website based upon parameters provided by M10. Those parameters were mutually agreed to be confidential. The nature of the systems and processes used by PriceTech in collecting information are also confidential in nature.
We provided that data to Mitre 10 in an [excel spreadsheet]. The data provided contained a timestamp and metadata indicating the source and date of the data download.
Mr Snowden goes on to state that the number of products on the Bunnings website varied from time to time but was generally in the approximate region of 35,000. The information he said was downloaded into an Excel spreadsheet on a regular basis. So that is the Bunnings product and pricing database.
What about the Mitre 10 product and pricing database? Mr Snowden goes on to say:
On one day each month, Mitre 10 compared that product information with its own information about Mitre 10’s products and recommended retail prices for the same day taken from Mitre 10’s internal product database.
Mr Snowden also observes in his statement that because Mitre 10 has individual owner-operators, prices vary by location. But he says “[i]n the vast majority of cases where there is a variation, the store price will be lower than the RRP”. So those are the two databases that need to be compared.
What, then about the comparison (matching) process? Mr Snowden says (at para 7.6):
Although it would have been possible to compare products and prices manually, to assist in matching products, Mitre 10 spent considerable time and resources in developing and evolving a series of computer-based comparison algorithms or filters that essentially automated this process. This information included:
(a) Product description.
(b) Product codes.
(c) Price.
It follows from this that the comparison (matching) process between the two product/price databases was automated. The result of this methodology was a database of products that both Mitre 10 and Bunnings sold that he says were either exactly identical (at least 95 per cent of the products) or virtually identical (where there are inconsequential differences such as packaging), along with the prices at which they are sold in each chain. Mr Snowden then says (at para 7.8):
Although it is impossible to always guarantee 100% accuracy, because of the process that we went through to review and refine the algorithms – which included numerous manual cross-checks – I can say that I am confident that the products we determined were matching products were either exactly or virtually identical as defined above.
The nub of Mr Snowden’s evidence is para 7.9. Without that paragraph, his evidence serves no purpose. It presents a summary of the comparison described earlier in this evidence, for the first day of each month from June 2015 to May 2016. It will suffice to quote the first:
In June 2015, there were 3095 matching products, 884 of which were cheaper at Mitre 10, 1989 were cheaper at Bunnings and 222 were the same price.
Two things need to be noted at once. First, Mr Snowden is inexplicit about exactly what role he performed in describing the instructions given to the PriceTech side of the analysis, the instructions given in the Mitre 10 side of the analysis (including the comparison), and in checking the accuracy of the results. Apart from his overview assessments expressing confidence in the process and results (of the process) he presents at para 7.9, the human elements in the process itself are attributed to “Mitre 10” or “we”.
Secondly, Mr Snowden does not say it in his statement, but we were advised, that no physical record was kept by Mitre 10 of the month-by-month comparisons (matches). Fundamentally, Mitre 10 has the product/price databases for each side, it has the automated process (although we are unclear whether that has “evolved” further since June 2015) and it has the results for the comparisons actually done by someone, by the automated programme, or by both, each month from June 2015 to May 2016. But it lacks the actual comparative database underlying those results.
Mr Snowden then attempts to meet these potential problems in statements and affidavits subsequently filed. Although the question determined by the Courts below, and therefore appealed to us, is the admissibility of the October 2017 statement, we can take general cognisance of other evidence that may be called in determining disposition of the underlying application.[5]
[5]R v Fenton [2008] NZCA 163 at [20].
In an affidavit sworn in support of Mitre 10’s application for confidentiality orders in February 2018, Mr Snowden says that “I am responsible for … developing the tools I describe below”. Later in the affidavit he describes his perception of the accuracy of the process compared to other methods and says that he has a high degree of confidence in the accuracy of the data Mitre 10 uses. In a revised statement prepared in November 2018 he describes further analysis undertaken after the District Court judgment, as a result of which some additional data sets were made available to the Commission (and subject to confidentiality) to Bunnings’ counsel and experts. It is clear from the evidence Mr Snowden gives that he was closely involved in the preparation of those further data sets. But that does not tell us anything about what he did personally to create the results which are the subject of para 7.9 of his statement, the admissibility of which is the focus of the appeal.
The objection to that evidence
Mr Snowden’s statement was served by the Commission on Bunnings on 30 October 2017. By letter 10 November 2017, Bunnings’ solicitors wrote to the Commission to give formal notice that it objected to the admissibility of Mr Snowden’s statement. The focus of the objection was, unsurprisingly, para 7.9. Three grounds for objection were advanced. The first was that the content of para 7.9 failed the relevancy test in s 7 of the Evidence Act, because the survey results were “not probative per se”. Without being able to scrutinise how the survey was conducted, and the actual results, neither the defendant or the trier of fact could conclude whether the results were true or inaccurate. The evidence was therefore purely speculative. Secondly, the results were said to be inadmissible pursuant to s 8, although we need not detail that point further. Thirdly, it was asserted that the survey results were inadmissible hearsay statements. The survey results are dependent on the extraction of data by a third party, not at that stage giving evidence.[6]
[6]Mr Pascoe’s statement was prepared to meet this objection.
Bunnings’ solicitors suggested that a pre-trial admissibility hearing under s 78 of the Criminal Procedure Act was required and appropriate. On 13 November 2017 the Commission made application for a pre-trial admissibility hearing.
On 27 November 2017 Bunnings filed a notice of response to the application. In it Bunnings asserted that the Snowden statement was inadmissible because it was “(i) not relevant evidence; (ii) not reliable (given the absence of any means to test reliability); (iii) lacked probative value; (iv) is hearsay; and (v) is otherwise inadmissible under the [Evidence Act]”.
Judgments below
In the District Court Judge Cunningham held Mr Snowden’s statement was relevant, as it tended to prove something of consequence to the charges, and was not inadmissible hearsay.[7]
[7]District Court judgment, above n 2, at [206] and [210].
In the High Court Duffy J did not directly address the relevance of Mr Snowden’s evidence, though noted irrelevance was an argument put forward by Bunnings.[8] She held however that without the full comparison data to support Mr Snowden’s oral evidence, his statement would amount to him giving an inadmissible hearsay account of what the full comparison data revealed. And insofar as he purported to interpret the full comparison data, Mr Snowden would be offering inadmissible opinion evidence in circumstances where he is not independent and there is no basis to establish he has the necessary expertise to offer opinion evidence.[9]
Submissions
[8]High Court judgment, above n 3, at [50].
[9]At [50] and [59]–[60].
For Bunnings, Mr Ross QC submits that the Judge below was wrong to accept Mr Snowden’s statement as relevant evidence for the purposes of s 7 of the Evidence Act. If full comparison data is not disclosed, the results data cannot satisfy the minimum probative requirement under s 7(3) of that Act: it has no logical tendency to prove or disprove the price comparisons it is offered to prove. Mitre 10’s price comparison process is complex and there is no record of the comparisons and subjective decisions made during the process. Without the disclosure of the full comparison data and price comparison process, the weight to be given to the results data cannot be assessed. Bunnings supports the Judge’s decision on hearsay and opinion.
For the Commission, Mr Dixon QC submits, first, that relevance does not include a reliability threshold and simply requires a tendency to prove something of consequence of the proceeding. Reliability is to be assessed by the fact‑finder. Disclosure of the full comparison data was not required for the results data to be relevant. Disclosure provides the means to corroborate and test reliability; corroboration is unnecessary and reliability is to be assessed by the fact-finder.[10]
[10]Citing s 121 of the Evidence Act 2006.
Secondly, Mr Dixon submits Mr Snowden’s evidence about the price comparison process and results data are observable facts recorded by him, a witness at the trial, and therefore is not hearsay. Nor, thirdly, is it opinion evidence. Whether the automated process identified two products as matching, and one competitor offering a lower price, are statements of fact. But if it is opinion evidence, then it is admissible either as necessary to communicate what Mr Snowden saw, or because Mr Snowden is an expert in comparing products and prices.
Is Mr Snowden’s statement inadmissible?
We start with first principles, and the most fundamental inquiry of all: what is evidence? We then consider relevance and the rule against hearsay in the context of Mr Snowden’s statement. It proves unnecessary for us to traverse the objection based on opinion.
Evidence and relevance
It is of essence that material presented to a court as evidence have a tendency to prove or disprove a fact in issue. Most definitions of evidence draw on that central idea. Lawyers like to talk about “facts in issue”, which invests the concept with a somewhat spurious certainty. The reality is that evidence exists to prove or disprove hypotheses — about facts the existence of which is uncertain. To be evidence the material tendered must tend to do that. If it does not, it is mere distraction and has no place being in the courtroom.
Two subsidiary points follow. First, evidence creates probabilistic knowledge, or conviction, but need not generate certainty. Secondly, to have a tendency to prove or disprove a hypothesis, or fact in issue, the witness must generally have personal knowledge of the assertions of fact they are advancing. There is a general exception for experts, who may infer from the observations of others and from expert understanding. There are other, specific exceptions to the hearsay rule, such as for business records. But those exceptions aside, a witness of fact describes his or her actions and observations, rather than the observations of others.
Jeremy Bentham described evidence as “any matter of fact, the effect, tendency or design of which, when presented to the mind, is to produce a persuasion concerning the existence of some other matter of fact”.[11] The American evidence scholar, James Bradley Thayer, fixed the law of evidence on two crisp but coherent principles:[12]
(1) that nothing is to be received which is not logically probative of some matter requiring to be proved; and (2) that everything which is thus probative should come in, unless a clear ground of policy of law excludes it.
[11]Jeremy Bentham Rationale of Judicial Evidence: Specially Applied to English Practice (Hunt and Clarke, London, 1827) vol 1 at 17.
[12]James Bradley Thayer A Preliminary Treatise on Evidence at the Common Law (Little, Brown and Co, Boston, 1898) at 530.
It is thus inherent in evidence that it first must be relevant: that is, it is probative (by tendency to prove or disprove) of something material (a fact in issue). There follows a second and subsequent question as to whether the evidence is then inadmissible because it belongs to a class of evidence the law excludes. And then a third enquiry as to what weight is to be given to evidence that is relevant and admissible.
Section 7 of the Evidence Act essentially reflects that approach by making relevance a prerequisite to admissibility. “Evidence” tendered that is not relevant is not admissible, and it is only relevant if it “has a tendency to prove or disprove anything that is of consequence to the determination of the proceeding”: s 7(2) and (3). It also reflects the twin requirements of relevance just mentioned, with relevant evidence being both probative and material.[13]
[13]Elisabeth McDonald and Scott Optican (eds) Mahoney on Evidence: Act and Analysis (Thomson Reuters, Wellington, 2018) at [EV7.02].
In Bain v R Elias CJ and Blanchard J referred to relevance in terms of probative connection to the fact in issue:[14]
That connection or relationship is not a matter of assessing the probative weight of the evidence but of accepting its logical connection to the fact it is said to prove.
Thus, in order to be relevant, the evidence must be reasonably capable of influencing the judge or jury’s assessment of a material issue:[15]
Where relevance to a fact is questioned, the judge must determine whether the evidence “is reasonably capable of supporting the fact”. If it is, then the evidence is relevant and, subject to rules of exclusion on policy grounds, must be left to the jury to evaluate.
[14]Bain v R [2009] NZSC 16, [2010] 1 NZLR 1 at [42].
[15]At [43], quoting R v Thomas [1970] VR 674 (VSC) at 679.
The s 7 threshold is not a high one: as Tipping J observed in Wi v R:[16]
This is not an exacting test: nor should it be. Any definition of relevance has to accommodate all kinds of evidence and in particular circumstantial evidence, individual pieces of which are often of slender, and sometimes very slender, weight in themselves. The question is whether the evidence has some, that is any, probative tendency, not whether it has sufficient probative tendency. Evidence either has the necessary tendency or it does not. As Lord Steyn said in R v A:
[T]o be relevant the evidence need merely have some tendency in logic and common sense to advance the proposition in issue.
[16]Wi v R [2009] NZSC 121, [2010] 2 NZLR 11 at [8], citing R v A [2001] UKHL 25, [2002] 1 AC 45 at [31].
In Keil v Police this Court was dealing with evidence of a constable that he could recognise the defendant from CCTV stills taken at a crime scene. The constable had not witnessed the offending. Rather he was offering an opinion as to identity based on a photograph. But that photograph was also available to the finders of fact to undertake the same exercise for themselves. We said:[17]
Evidence will be relevant under s 7 if there is some rational basis for a judge or jury to infer that the offered proof is probative on a material issue in the case. … It is not enough that evidence is relevant on its face, as Constable Muir’s would be in a conversational context. The evidence must be relevant in the context of the Court setting in the sense of being capable of assisting the fact-finder.
The constable’s evidence was excluded; it did not meet the s 7 relevance threshold.[18]
[17]Keil v Police [2017] NZCA 430 at [20].
[18]Compare Shepherd v R [2011] NZCA 666, [2012] 2 NZLR 609, concerning expert evidence of facial mapping.
The fact in issue to which Mr Snowden’s evidence is directed is the Commission’s hypothesis that a substantial proportion of identical products are priced more cheaply at Mitre 10 than at Bunnings. Mr Snowden’s statement makes one direct, and one indirect, assertion of fact. The direct assertion is that the price comparison outputs stated in para 7.9 of his brief are reliable (that is, true). That is the core of his evidence; all else is scene-setting. Because the outputs are the product of an automated process, he also makes an indirect assertion: that the automated process reliably matches and compares products and prices offered by Bunnings and Mitre 10. It is the functional equivalent of a witness describing a manual comparison between identical products at each store, using identifiers such as brand, product type and description and bar codes.
It is convenient to deal with these assertions in the reverse order. Each assertion requires foundation to have a tendency to prove the fact in issue concerning comparative pricing. That foundation must come from observation, or (where permissible) a combination of opinion and observation. Let us take, first, the implied assertion: that the automated process reliably matches and compares products and prices offered by Bunnings and Mitre 10. A witness, suitably qualified, could lay foundation verifying that assertion ex ante by proving the accuracy of the algorithms embedded in the programme. That witness or another (who need not be an expert) might then lay further foundation ex post by demonstrating that the programme had worked in fact by auditing the outputs. The extent of the audit would affect weight; all of the foregoing evidence (had it existed) would easily meet the relevance threshold.
Mr Snowden is not presented as an expert. He is not qualified to verify the accuracy of the programme. Indeed, it is unclear whether he had any direct role in its development that might permit him to give evidence tending to verify its accuracy. Nor does the statement offer ex post non-expert foundation via Mr Snowden’s own observations of checking the outputs of the automated price comparison process, tending thereby to show them to be reliable. We infer that is because he has not undertaken that exercise.
We turn to the other, direct assertion: that the outputs stated in para 7.9 are true. This really is parasitic to the implied assertion. The comparative database of electronic matches that underlie the outputs in para 7.9 is lost. Because the two source product/price databases still exist, the exercise is capable of being replicated. In the statement at least, this has not been done, although other evidence is intended to be called about limited sampling that Mitre 10 has undertaken. That further evidence is not directly in issue in this appeal.
The outputs Mr Snowden presents in his para 7.9 are merely a record of a past and unaudited use of the automated process. In the absence of foundation by observation, the statement is simply conjecture based on an unproved automated process and does not tend to prove the fact in issue (that is, whether a substantial proportion of identical products are priced more cheaply at Mitre 10 than at Bunnings). It is accepted that the balance of the statement serves little independent purpose.
We therefore accept Bunnings’ submission that Mr Snowden’s statement of 30 October 2017 lacks relevance and is inadmissible under s 7.
Hearsay
In case we are wrong on relevance, we turn to the issue of hearsay. Para 7.6 of Mr Snowden’s evidence (quoted above at [13]) makes clear the outputs reported in para 7.9 are the product of an “essentially automated” process. The question then is whether, even if the s 7 threshold is cleared, the two inherent assertions of fact at [37] are assertions (1) made by a person other than the witness, and (2) offered in evidence to prove the truth of their contents.[19]
[19]Evidence Act, s 4(1) (definition of “hearsay statement”).
The law of evidence has found it difficult determining quite how to deal with evidence which is the product of automated processes. As Phipson on Evidence observes, if the document is created without the intervention of a human mind, it is real evidence and no hearsay issue arises.[20] But a digital document (1) dependent in whole or in part on information supplied by a person, and (2) relied on to prove the truth of its contents, is likely to involve hearsay. In fact, as the cases show, the more important enquiry is likely to be with the accuracy of the data. As the English Law Commission put it:[21]
The question is, on what basis should such evidence be excluded? One view is that it is hearsay, because it is tantamount to a statement made by the person who fed the data into the machine. An alternative view is that the statement by the machine, properly understood, is conditional on the accuracy of the data on which it is based; and that, if those data are not proved to have been accurate, the statement therefore has no probative value at all. The question of hearsay does not arise, because the statement is simply irrelevant.
As will be apparent, we have adopted the second analytical path, which is one of relevance. But in case it has led astray, we have circled back to tackle the first path.
[20]Hodge Malek (ed) Phipson on Evidence (19th ed, Sweet & Maxwell, London, 2018) at [28‑28]. Compare Mathew Downs (ed) Cross on Evidence (looseleaf ed, LexisNexis) at [EVA137.2] which offers a slightly different analysis. We prefer the more precautionary phrasing of principle in Phipson.
[21]Law Commission (England) Evidence in Criminal Proceedings: Hearsay and Related Topics (LC245, 1997) at [7.48].
In Holt v Auckland City Council this Court was concerned with an automated process to test a blood alcohol specimen.[22] No statutory presumption applied to it, and the result had to be proved by an expert. The expert called was the chemist who had undertaken the analysis, but part of the analytical process was automated via use of an “integrator” and computer. The computer had been programmed by two different scientists; the witness said she assumed it had been checked quite regularly and if something had been wrong with it, she would have noticed.[23] Richardson J observed:[24]
The computer print out of the blood alcohol analysis involved obvious hearsay. The role of the integrator and computer is not confined to the retrieval of information. The programming is by persons expert in that field and that is reflected in the computer print out. Miss Campbell had not programmed the computer and disclaimed any expertise in computer science. So there was no evidence from any expert in the field as to the functioning and reliability of integrators and computers in general and of the particular apparatus used in this case.
…
… In a technology based society the user of scientific apparatus is not likely to have made the calculations on which the capability and accuracy of the machine depends. Indeed, he may lack any real understanding of the scientific principles on which it is based. But his experience of the machine may have satisfied him as to its reliability for his needs. It does not follow that his reliance on the machine will make it unnecessary to demonstrate its functioning and reliability before the results are admissible in evidence.
[22]Holt v Auckland City Council [1980] 2 NZLR 124 (CA).
[23]At 126.
[24]At 126–127.
Statute may presume reliability for some devices (such as breath alcohol devices), reversing the onus of proof as to reliability.[25] And at common law, judicial notice may be taken of the reliability of commonplace instruments: there is a presumption of fact, in the absence of evidence to the contrary, that readings taken from watches, thermometers, speedometers and the like are correct.[26] This principle is now the subject of s 137 of the Evidence Act.
[25]See, for example, Land Transport Act 1998, s 77(1).
[26]Holt v Auckland City Council, above n 22, at 127–128, citing Zappia v Webb [1974] WAR 15 (WASC) at 17.
In R v Miller this Court revisited Holt after a gap of eight years.[28] In Miller the appellant was charged with using a false document to procure a loan. An accountant, Mr Hulst, gave evidence of the financial state of the appellant’s business. He produced documents prepared with the help of a company employee (Ms Dreher). She in turn had used the company’s business computer, on which the company’s books were loaded, to assist her. Trial counsel took a hearsay objection to Mr Hulst’s evidence, and renewed it on appeal. This Court said:[29]
Mr Hulst did not merely rely on Mrs Dreher’s evidence … He was at pains to establish the company’s true financial position from his own checks and inquiries. His evidence was not hearsay. It was evidence of a position he had established to his own satisfaction. He personally checked on figures supplied to him by Mrs Dreher.
[28]R v Miller (1988) 3 CRNZ 609 (CA).
[29]At 614.
The Court then referred to its earlier decision in Holt. It distinguished it. The integrator and computer in Holt were, the Court noted, “held to come within the category of recently invented devices which usually require expert evidence to establish what they can measure and accomplish”.[30] But the Court went on:[31]
But there is nothing to suggest that the computer Mrs Dreher used was in that category. There is no reason to suppose that a computer in a relatively modest freight forwarding business would be anything out of the ordinary or differ in its workings from those now in wide use or be at all complicated.
There was therefore no need for evidence to be called as to the workings or programming of that computer.
[30]At 614.
[31]At 614.
We turn now to the evidence tendered here. We make four points.
First, the automated price comparison system developed by Mitre 10 has two parts. The first is the establishment of the product/price databases (PriceTech supplying the Bunnings one). The second part uses an evolved system of algorithms to match and compare product descriptions and prices on each database. It is acknowledged to be novel and bespoke. The intellectual property is guarded jealously by Mitre 10. No statutory or common law presumption of accuracy applies to it. The Commission must prove the accuracy of its outputs.
Secondly, the implied assertion noted at [37] above (reliably matches and compares) must be the product of either observation or qualified expert opinion. Mr Snowden is not an expert and nor are the outputs confirmed by his own observation. The assertion is therefore the implied statement of another, namely the programmer who developed and tested the automated system. It is offered to prove the truth of the statement (reliability) and is hearsay.
Thirdly, the same may be said of the direct assertion at [37] above (the reliability of the outputs reported in para 7.9). These are the product of the automated process. Mr Snowden does not give evidence of having observed the process or checking the outputs. The outputs are in part the product of inputs of other persons. While the exact details are not in evidence before us, that will include the programmers of the web crawlers and other devices that generate the two product/price databases, and the programmer of the Mitre 10 automated price comparison system. On the same basis as in Holt, this assertion is hearsay.
Fourthly, and for completeness, we note that the Commission does not attempt to argue that the evidence is admissible under the business records exception to the hearsay exclusion.[32] For these reasons we find Mr Snowden’s statement of 30 October 2017, and para 7.9 in particular, infringes the rule against hearsay.
Conclusion
[32]Evidence Act, s 19. See for example Asgedom v R [2016] NZCA 334, (2016) 28 CRNZ 70 at [69].
Mr Dixon, reading the writing on the wall at the hearing, encouraged us to follow the course taken by this Court in R v Fenton and decline to make a ruling excluding Mr Snowden’s statement, on the basis that repair work can be undertaken on it before trial.[33] Such a course was appropriate in that case, where the Crown had been taken by surprise at the pre-trial admissibility hearing. It is not here. The Commission applied for a ruling as to admissibility, the defects in the statement are patent, and the ruling should be given. It is that Mr Snowden’s statement of 30 October 2017 is inadmissible both under s 7 and as infringing the rule against hearsay.
[33]R v Fenton, above n 5, at [20].
The receipt and admissibility of any substituted statement by Mr Snowden will be a matter for the trial Judge.
The validity of the Commission’s summons of 13 December 2017
This was very much a secondary argument in the appeals. It was accepted that it is essentially parasitic to the challenge to Mr Snowden’s statement. The summons was sought to support the admissibility of the statement. Having upheld the admissibility challenge, we can be brief on the summons.
The summons concerns further evidence Mr Snowden proposes to tender beyond the content of the October 2017 statement just addressed. The Commission sought further information from Mitre 10 after the charges were laid against Bunnings. Mitre 10 was willing to provide the information so long as confidentiality measures protected commercially sensitive information (in particular, the design and workings of its automated price comparison system). Bunnings, it appears, did not agree to voluntary restrictions, and nor did it apply for non-party disclosure against Mitre 10 under ss 24 to 29 of the Criminal Disclosure Act 2008. It was, of course, under no obligation to do either.[34]
[34]At the hearing we were informed that Bunnings has recently filed an application for non-party disclosure from Mitre 10.
On 28 November 2017 Mitre 10’s solicitors wrote to the Commission indicating willingness to provide the two product/price databases (that is, of Bunnings and Mitre 10 products and prices) and a sample of 400 matches from the comparison system database, on an open basis. As an alternative, it suggested the Commission seek a summons under s 159 of the Criminal Procedure Act. That would enable the Commission to obtain the information, while Mitre 10 would have standing to apply for confidentiality orders.[35]
[35]Under s 69 of the Evidence Act or s 29 of the Criminal Disclosure Act.
The summons sought and obtained by the Commission, ostensibly under s 159, required Mr Snowden to attend the hearing of the pre-trial admissibility application before Judge Cunningham. It was broader than Mitre 10 wanted. It sought Mr Snowden’s attendance with the two product/price datasets, and (1) “the datasets derived from each of Mitre 10’s comparisons of the above two datasets, comprising all matched products across the two companies” and (2) “[a]ll information as to how the comparison was undertaken, including any formulae or algorithms used by Mitre 10 to create the comparison dataset”.
Mitre 10 then negotiated with the Commission the amendment of the summons, which now conformed essentially to what Mitre 10 had said it would volunteer in its letter of 28 November 2017. However, service of the original summons was essential to enable Mitre 10 to seek confidentiality orders for all the material tendered apart from that obtained from Bunnings.
Judge Cunningham dismissed Bunnings’ challenge to the validity of the summons. She granted Mitre 10’s applications for amendment of the summons and confidentiality orders.[36]
Judgment below
[36]District Court judgment, above n 2.
Duffy J considered the lawfulness of the Commission’s use of s 159(2) of the Criminal Procedure Act could not be assessed in a vacuum. She found Mitre 10 had provided the requested information to the Commission voluntarily.[37] The Judge therefore considered the contrasting scope of the Commission’s powers under ss 47G of the Fair Trading Act 1986 and 98 of the Commerce Act 1986 irrelevant as nothing prevents a prosecuting authority accepting information provided to it voluntarily by a third party.[38] If this required a statutory basis it could be regarded as within the Commission’s implied incidental powers and functions available to it when carrying out its prosecutorial powers and functions.[39] Though a summons may not have been necessary, it was usual for a prosecuting authority to summons a witness meaning the Commission simply followed standard prosecutorial practice.[40] Section 159 of the Criminal Procedure Act permitted that course. The Judge accordingly upheld the summons.
Submissions
[37]High Court judgment, above n 3, at [39].
[38]At [40]–[41].
[39]At [45].
[40]At [47].
For Bunnings, Mr Ross submits the Judge failed to consider that the summons subverts the evidence-gathering and disclosure scheme of the Fair Trading Act and Criminal Disclosure Act. The former was said to provide a complete code for the Commission’s evidence-gathering powers for prosecutions brought under that Act, and those powers are extinguished upon the filing of charges. The argument is based on a contrast of s 98G of the Commerce Act, which expressly permits the use of equivalent search and compulsion powers in that Act notwithstanding the filing of charges, with the Fair Trading Act’s lack of an equivalent provision.
It is submitted, further, that the Commission cannot use summons issued under s 159 of the Criminal Procedure Act to gather evidence it knew about, but did not collect during the investigative phase. To do so would, it was said, undermine the Criminal Disclosure Act scheme, which does not empower a prosecutor to seek disclosure form non-parties.
Was the summons invalid?
We are satisfied the summons was validly issued. We make four points.
First, we accept that the summons was a material step in the obtaining of the evidence. While Mitre 10 was willing to volunteer the evidence tendered under the amended summons, ultimately it did not do so and the compulsory process described above was adopted, giving it standing to argue for confidentiality orders over its own product/price dataset and the comparison dataset (and to exclude the wider range of material sought in the original summons). We do not think it right in these circumstances to reason that Mitre 10 volunteered the evidence apart from the summons.
Secondly, we do not accept the thrust of Mr Ross’s submissions on the validity of the summons. The Commission was entitled to continue to investigate despite the laying of charges.[41] In conceptual terms the Commission’s case is not complex: its factual foundation is the hypothesis that Bunnings is frequently not the cheapest‑priced retailer of the products it sells. The Commission was well aware of the existence of the three datasets, and of the automated process. Much of this material it had seen, because Mitre 10 (who were keen to see Bunnings investigated) had disclosed it. This is evident from the statement of the Commission’s lead investigator, Mr Allan. The first problem for it was Mitre 10’s reticence in placing all that material before the Court (and Bunnings in particular). The second was Bunnings’ challenge to the relevance and admissibility of Mr Snowden’s statement.
[41]Commerce Commission v Air New Zealand Ltd [2011] NZCA 64, [2011] 2 NZLR 194 at [107]–[108]. That decision concerned powers under the Commerce Act, but it applies equally to a prosecution under the Fair Trading Act in our view.
Thirdly, it is unnecessary for us to resolve the significance of the drafting distinction between the search and compulsion powers of the Commerce and Fair Trading Acts. It may or may not be a deliberate distinction, inferentially restricting post-prosecution search and compulsion powers under the latter Act, although Bunnings’ argument is not appealing at first impression. But it is unnecessary to decide it because of the next point.
Fourthly, we accept the argument made by Mr Dixon that the relevant source of power here is s 159 of the Criminal Procedure Act. That is the successor of s 20 of the Summary Proceedings Act 1957. Its purpose is forensic: to require the attendance of a witness (and to produce documents) at a hearing, at the behest of either defendant or prosecutor. It is by no means unusual for a prosecutor to have used the power in s 159 (or s 20, its predecessor) without having seen the whole array of documents then brought to court. The introduction of the Criminal Disclosure Act three years before the Criminal Procedure Act did not alter s 20. That Act serves a different purpose, given the inability of the summons process to be used for pre-trial disclosure (the premise for the summons being that the content is intended to be evidential).[42] Similarly s 24 of the Criminal Disclosure Act enables a defendant to obtain non-party disclosure which the summons process does not, for that very reason.
[42]Livingston v Institute of Environmental Science and Research Ltd (2003) 20 CRNZ 253 (CA) at [63].
Here the summons was issued to meet a challenge to the relevance and admissibility of Mr Snowden’s statement, to be dealt with pre-trial. The use of the power in s 159 was authorised and orthodox in that context. The documents produced have now been disclosed to Bunnings under confidentiality orders and there can be no complaint of unfairness or surprise. The challenge by Bunnings to the relevance and admissibility of Mr Snowden’s statement has succeeded.
We conclude that the summons in question was validly issued.
Result
Leave to appeal is granted in CA59/2020 and CA62/2020.
The Commission’s appeal in CA59/2020 is dismissed and Bunnings’ appeal in CA62/2020 is allowed. The statement of Murray Snowden of 30 October 2017 is inadmissible.
Bunnings’ appeal in CA61/2020 is dismissed. The summons issued by the District Court is valid.
Solicitors:
Kayes Fletcher Walker Ltd, Manukau for Commerce Commission
Lindsay LA, Auckland for Bunnings Ltd
Meredith Connell, Auckland for Mitre 10 (New Zealand) Ltd
[27]At 128.
4
0