You are viewing a javascript disabled version of the site. Please enable Javascript for this site to function properly.
Go to headerGo to navigationGo to searchGo to contentsGo to footer
In content section. Select this link to jump to navigation

Prerelease access to official statistics is not consistent with professional ethics

Abstract

Arguments for prerelease access recall the ‘public interest’, but the interests served are, at the end of the day, not the ones of the public. Prerelease access is inconsistent with first principles of statistical ethics: impartiality, objectivity, independence. It is actually a potent enabling condition for undermining the integrity of official statistics and undercutting the public’s trust in these statistics. It serves as a vehicle for unfair gains – whether political, economic, or career. Prerelease access is not necessary either for policy conduct or for the public’s understanding of the statistics; there are appropriate substitutes. In this paper we discuss in some detail arguments for and against prerelease access to official statistics by the government and/or the press. We recommend complete elimination of prerelease access by both the government in power and members of the press. We also advocate that contradictions in codified principles regarding prerelease access under the principle of impartiality must also be addressed: there should be no place there for ‘double think’.

1.Introduction

While equal access to official statistics by users at the same time1 is broadly recognized as an essential guideline to support the statistical principles of professional independence and of impartiality and objectivity, prerelease access by the government and the press is practiced by many advanced national statistical systems. Even if prerelease is not practiced by a national statistics office, or is relatively restricted, other producers in the same national statistical system are often more relaxed about such access. There has recently been some movement away from some forms of prerelease access in some countries.

We argue that prerelease access to official statistics is contrary to ethics in the production of official statistics. Eliminating prerelease access is in the long term interest of participants in the political process (politicians and political parties), market participants and the broader public. Absence of prerelease access is also best for international collaboration and cooperation. Prerelease access is a legacy of older times, when official statistics were the statistics ‘of the government’. However, this is not the case anymore, despite the lagging institutional independence of official statistics. Official statistics are a (global) public good and they are not the statistics of the government; official statistics belong to all users.

We recommend the complete elimination of all prerelease access. Progress in recent years in moving away from prerelease access should be consolidated and further strengthened. There is important progress that needs to be done by major advanced economies. It would be catalytic for progress in other countries.

The present paper was produced with the hope of providing some thoughts for the discussions of official statisticians and policy makers at national and regional levels regarding the future of prerelease access to official statistics by government and/or the press. It could also inform discussions about addressing prerelease access in the context of any review of codes of practice/ethics for official statistics.

The paper is organized as follows: Section 2 discusses the state of play with prerelease access and notes the wide variability of practices among countries, within national statistical systems and even among statistical products. Section 3 presents various arguments in favor of prerelease access. Section 4 presents arguments against prerelease access and also comments on the arguments for prerelease access presented in Section 3. This section also examines what first principles of statistical ethics indicate about prerelease access. Section 5 provides conclusions and recommendations.

2.State of play with prerelease access

Prelease access to official statistics is the situation whereby someone who is not involved in the production or dissemination of the statistics, i.e., someone outside the statistical perimeter,2 has access to them before the statistics are officially released to all other users.

Practice regarding prerelease access can vary widely from one official statistics producer to another within a given national statistical system (NSS), from one country’s NSS to another country’s NSS, and from one official statistical product to another. The individuals who are outside the statistical perimeter and get prerelease access can be members of the government and their staff; other policy officials, such as central bank officials or officials and administrators in ministries; members of the press; and even analysts working in research environments.3

Prerelease access is practiced more widely than no prelease. Prerelease access by government is explicitly practiced in most G7 market economies – e.g., US, Germany, France, Canada, and – still to a notable extent – the UK.4 Other OECD countries, such as Australia, Israel, Korea, New Zealand, also allow prerelease access by government officials.5

However, smaller advanced economies, with several of them in the European Union, have led the way in providing no prelease access. The Nordic countries (Denmark, Iceland, Norway, Finland, Sweden) are clear examples. Among larger economies, Italy’s NSS is not allowing prerelease access by the government.

If prerelease access by government officials is practiced widely by national statistical offices (NSOs), it is practiced even more widely by other national authorities (ONAs) that produce official statistics, such as statistical producers that are part of ministries or central banks. Such statistical producers are much more likely than NSOs to ‘show’ the statistics they produce to the policy hierarchy of the institution they are a part of before release to all other users.

Moreover, prelease access to official statistics by the press is quite common. For example, it is practiced in countries such as the US, Canada, and Italy. Some statistical producers (such as the NSO in Italy) that do not allow prerelease access by government opt to allow some form of prerelease access by the press.

Turning to prerelease access that is informal and ‘under the radar’, anecdotal evidence indicates that probably quite a bit of that is going on around the world even in systems that declare they have no prerelease access. This is an important phenomenon to note, as this kind of prerelease access is taking place usually in the context of power and exchange relationships: hierarchical/dependency, clientele and favoritism relationships. Examples would inter alia include situations whereby a minister calls up the head of the statistics office to see ‘how the numbers are coming’ or a journalist who is ‘close’ to some official in the statistics office calls to have ‘a heads up’ about the main number(s) of an upcoming statistical release.

3.Arguments in favor of prerelease access

In this section we present various arguments in favor of prerelease access. We do not comment on the arguments here, as we reserve this discussion for the following section. Instead, the intention here is to give the reader a sense of the arguments as presented by their proponents. The section also presents some arguments that may not usually get presented in the public domain, but which are presented in other domains, especially in private conversations of the relevant stakeholders.6

  • It is in the public interest

The most basic and broadest argument for prerelease access is that it is ‘in the public interest’.7 The issue, therefore, is what is the ‘public interest’ that is served and this is where the more specific arguments for prerelease access are made.

  • It improves the workings of the government in office and the ministries/departments to have the statistics as soon as possible8

The gains in efficiency and effectiveness are said to be of various sorts.

  • Minimizing policy response time to news in statistics

    One argument is that the government can take action as soon as possible, especially in areas where time is of the essence – where a day or a few hours make a material difference in the efficiency and effectiveness of policy. With prerelease access, the government and its ministers can take action at the time of the statistical release minimizing the time and maximizing the usefulness of the policy response.9

  • Taking measures ahead of market knowledge of the news in statistics

    A related argument is that the government is given the opportunity to think through necessary actions and even take some actions in response to the news in the statistics in advance of actions that other users of the statistics may take when they see the statistical release. This argument is thought to be particularly strong when it comes to market sensitive data, i.e., data that can move the markets, even in extreme ways, with the result of market panics and crashes. Having the statistics as soon as possible, and specifically earlier than the market participants that are said to be subject to behavior that leads to market instability, provides an opportunity that would not exist otherwise to take measures (do and say things) that preempt market instability.10,11,12

  • Have adequate time to prepare for comment on news in statistics

    Another argument is that the government is able to prepare for comment on the statistics, by verbal or written interventions (e.g., a press release), at the same time that the statistics will be known by the public. Thus, the staff of a Minister or the Prime Minister/President can prepare the reaction of the policy maker to the statistics, offering a specific and well thought out interpretation of the news that the statistics bring.13 This argument is sometimes supplemented by the one that the government needs time to process and analyze complex information provided by the statistics.14

Other arguments are also offered in favor of prerelease access.

  • It allows for tighter control of official statistics that are potentially damaging to the national interest

According to this argument, (some) political leaders and policy makers may feel that official statistics may become falsified and damage the ‘national interest’. Thus, it is argued, prerelease access by a cabinet minister may help prevent such statistics from being disseminated and damaging the ‘national interest’.15,16

  • It helps protect the interest and reputation of the persons in the government and of the political party in power

The politicians/policy makers involved feel that if they have no knowledge of the official statistics before they are released they ‘look bad’ for two reasons: First, because they are not adequately prepared to respond to questions and comments by the press and the opposition, and thus do not look intelligent and effective. Second, because they give the appearance to the public of being weak as politicians/policy makers as they are expected by (some in) the electorate/population to be able (to order) to have early access to such information and to control the parts of the state administration that produce official statistics.

  • It is a right of the government. The government should be able to see the statistics early because it pays for them

The government makes decisions to finance with scarce economic resources the production of official statistics in the country and may feel that it has the right to see, as soon as it is available, what it has paid for.

  • It allows the government to have privileged access to statistical information it can use when interacting with international partners

The argument for prerelease access here is that the government needs to have the information early when it is involved in discussions or negotiations with international partners (e.g., with the International Monetary Fund on an adjustment program, or with EU partners in the context of the European Semester). In this argument, the government feels that it can potentially achieve a ‘better outcome’ in its discussions with the other party if it has the information available (and the other party a priori does not) and can then decide whether and how to use the information. Sometimes the outcome of having this privileged access to statistical information before its release to the public could be – at least in the first instance – a policy commitment that is less constraining than would otherwise be or the creation of more positive impressions in the minds of international interlocutors (e.g., foreign investors).

  • It allows the government to take actions or avoid taking actions, whether economic/financial or political, on a matter for which it would have proceeded differently in the absence of the early access to statistics

In terms of economic/financial actions, for example, the government may decide to move the issuing of government paper to an earlier date to avoid the punishment by markets in terms of higher borrowing rates for impending bad economic news. Or the government may decide to undertake political actions of various sorts in anticipation of the political/psychological impact of the statistical release on the broader public. For example, the government could preemptively verbally attack its political opposition, or highlight problems with a difficult geographic neighbor, or even initiate military action. These possibilities become more feasible as the time allowed for access prior to release to all users increases; that is why usually these arguments are accompanied by demands to have access that is days and not hours before the official dissemination of the statistics. However, even with a few hours’ advance knowledge of the data, significant actions can be taken (or cancelled or postponed) both on the political and the economic front.

We now turn to arguments for prerelease access by nongovernment users. The basic argument for this kind of access is again that it is in the public interest.

  • Prerelease access by the press helps the public understand the official statistics and reduces the probability that they will be intentionally or unintentionally misused or abused

The argument is that prerelease access by the press and other potential analysts and commentators enhances the information available to the public when the statistics are officially released. This helps avoid misunderstandings that could occur due to quick and superficial study of the statistical releases.

  • It facilitates the work of the press

The argument goes that when the work of the press is facilitated it is not only the public that benefits but also the statistical producer, which earns goodwill and in addition avoids some of the need and cost to ex post respond to misunderstandings or even criticisms of the statistics.

4.Arguments against prerelease access

In this section we present arguments against prerelease access. In this context, we also comment on the arguments in favor of prerelease access presented in Section 3.

4.1First principles of statistical ethics

We first turn to arguments against prerelease access that hark back to first principles of statistical ethics. Prelease access to official statistics is assessed and found wanting on the basis of the principles of impartiality and objectivity, as well as on the basis of the principle of independence.

  • It is incompatible with the statistical ethics principle of impartiality

Impartiality means being not partial or biased: treating or affecting all equally [9]. Thus, if official statistics are to be released/made available to the public with impartiality, then all users must be treated equally. This means equal access to all users at the same time and no prerelease access by any users. Thus, Principle 1 of the UN Fundamental Principles of Official Statistics (UNFP) [10] “official statistics …are to be compiled and made available on an impartial basis by official statistical agencies” cannot but mean that for all users to be treated equally, official statistics are to be made available on an equal basis to all users.

  • It is incompatible with the statistical ethics principle of objectivity

Objectivity is defined as the quality or character of being objective, lack of favoritism toward one side or another and freedom from bias [11]. This reinforces the idea that the statistics cannot be characterized by favoritism toward one side or another; thus no side can be favored and have prerelease access.

Furthermore, impartiality and objectivity in official statistics is indicated, according, for example, to the European Statistics Code of Practice, Indicator 6.1, by statistics being compiled on an objective basis determined by statistical considerations [12]. Of course, compilation here cannot possibly denote just the production phase but all stages of the statistical process, including the release phase.17 Impartiality and objectivity can also be expressed as in Principle 2 of the UNFP: the statistical agencies need to “decide according to strictly professional considerations, including scientific principles and professional ethics, on …the presentation of statistical data” [10].

The statements in the paragraph above mean that if there is no statistical reason/consideration or no strictly professional consideration to differentiate by user the access to statistics, then there should not be such differentiation. And in deed there is no such statistical/professional reason and there cannot be such reason to differentiate access to official statistics by user.

Moreover, the principle of pursuing objectivity means that “statisticians should pursue objectivity without fear or favor” according to the ISI Declaration of Professional Ethics [13]. Pursuing objectivity in the dissemination of statistics means, therefore, showing no favor to any user, including the government (as well as the central bank, the press, or anyone else) by providing to it prerelease access to the statistics.

  • It is incongruent with the statistical ethics principle of professional independence

The principle of professional independence means inter alia “independence [of the producers of official statistics] from political and other external interference in developing, producing and disseminating statistics” according to the European Statistics Code of Practice [12]. What independence from government in the dissemination/release of statistics can there be when official statistics have to be disseminated first to the government?

Moreover, according to the same Code of Practice [12], “the heads of the National Statistical Institutes …have the sole responsibility for deciding on statistical methods, standards and procedures, and on the content and timing of statistical releases.” What independence can there be when the decision on prerelease access is taken by somebody other than the head of the NSO?

To demonstrate independence from government or any other actor in the process of dissemination of official statistics, there cannot be privileged prerelease access to the statistics by the government or any other actor. Stating that official statistics producers are independent but they have to release the statistics first to the government or that the government can decide itself to have prerelease access is self-contradictory and paradoxical, and in our view points to a fundamental incongruence.

In conclusion, first principles of statistical ethics tell us that prerelease access to official statistics by the government or anyone else is not consistent with professional statistical ethics.

We now turn to more specific arguments against prerelease access.

4.2Integrity and credibility of statistics

  • It increases the possibility of pressure on the independent production and dissemination of statistics to serve political/policy interests

Prerelease access opens the door for last minute influence on statistical information, whether in the form of changing the statistical figures themselves or their presentation (language, graphs, ‘what’s in and what’s out’), or in the form of a temporary or permanent suppression of the statistical results. The influence on the statistics can be exerted directly by having the government politicians/policy makers intervening directly to alter the content of the statistical releases.18

Alternatively, prerelease access can give rise to the preemption of the statistical results by politicians/policy makers. The latter may in some cases use the information they get through prerelease access to influence the finalization of figures by preannouncing a ‘desirable’ statistical result, which signals to the statisticians what they have to do to meet the expectations of the political/policy side. Thus, prerelease access creates possibilities of giving guidance to the NSO and the statisticians on what statistical result (figure) is acceptable to finally produce.

Prerelease access to statistical results leading to preemption of the results through preannouncing of the figures by politicians/policy makers create significant credibility problems for the statistical office when it happens (or is thought by observers to happen), even if statistical results are not actually modified as a result.19

Very importantly, in the context of prerelease access, the influence of politicians/policy makers can also be more indirect, yet powerful. This can take place through the self-censorship of the statisticians, who may feel pressure to show, in what is effectively a ‘private’ setting, to individuals they consider their ‘superiors’ statistical results and presentations of these results that they feel would please their ‘superiors’.20

It should be noted here that arguments that “prerelease access by government allows for tighter control of official statistics that are potentially damaging to the national interest” (see Section 3 above) should be seen as a ‘red flag’ that political interests are likely intent on censoring, suppressing or manipulating the official statistics. If there are any issues with the integrity and broader quality of official statistics, the solution cannot and should not be to have a government minister check them before release. Involving international processes and institutions in carrying out an assessment of the quality of statistics, including the implementation of statistical principles in their production, would be the appropriate way to proceed. In this context, it could be mentioned that in order to carry out such assessments effectively and efficiently there is a need to set up an independent international institution with the mandate and resources to carry out regular as well as ad hoc evaluation, verification, follow-up and publish reports in the form of audits [16].

Thus, prerelease access is one potential gateway – enabling condition – to suppressing or manipulating official statistics in direct and indirect ways. It should be highlighted that the mechanism may not be used for a long time, but a time can come sooner or later when, under the right circumstances and with the right personalities in place, the weakness entailed by prerelease access will be exploited against the integrity of official statistics. This issue is very clear to politicians/policy makers around the world.21

  • It increases the possibility of the perception of pressure on the independent production and dissemination of statistics to serve political/policy interests

Prerelease access, even if not abused, gives rise to significant credibility problems for official statistics on account of the public perception of the practice. Truly, this is a pretty high price to pay for any benefits of prerelease access – if there are any. The fact that prerelease access enables simultaneous commentary by the government on the statistical results when they are officially announced to the public by the official statistics producer – touted as one of the benefits of prerelease access (see Section 3 above) – is actually damaging because of the perceptions it is feeding. As the Royal Statistical Society succinctly points out [4], it “perpetuates the impression that ministers control the data.” The implications of this can be very damaging for the credibility of and confidence in official statistics regarding two things: whether they are accurate/reliable and whether they are properly used and not used for political and economic gain.

The points just above explain why the argument for prelease (made in Section 3) that such “access by government helps protect the interest and reputation of the persons in the government and of the political party in power” is false and unprincipled, with adverse consequences over time.22 Actually, if some minister gives the impression that he can order the statistics producers to hand him the data early, and that he prepares thoroughly regarding every detail of the statistics under his control, then his reputation as well as that of his government and political party can be thoroughly tainted. It is better for the minister to advertise the independence of the statistics office and the arms-length relationship he has with it and to reserve his right to study the statistics at the same time as everybody else does it before commenting in detail on the statistical results.

Regarding the argument (made in Section 4) that prerelease access allows the government to have privileged access to information it can use when interacting with international partners, it is also false and unprincipled with adverse consequences over time. The use of official statistics as a tool of the government that can be taken out of the ‘tool box’ or hidden out of sight surely will negatively impress any international partners (whether foreign governments, international organizations or foreign investors). The credibility and image of the government will suffer from such behavior. The international perception of the integrity of the country’s official statistics will also suffer on account of the perception that the government controls the statistics.

  • It is not necessary as the government does not need to know the exact statistics from the statistical office; there are alternative sources of data (administrative data) and other governmental information that provide a picture to policy makers even on a day to day basis and well in advance of any statistical release23

The general point is that administrative information available to policy makers is in the big majority of cases adequate to make all the necessary policy decisions and to prepare any informed responses for the ministers when the actual statistics come out. Anyway, often policy makers go ahead and inform the public of their own estimates of what the outcome will be.24 Thus, either the information that the government needs actually exists within the policy structures controlled by the government and is already available to it or it can be made available with a relatively small investment in human resources and processes within policy structures.

Let us consider in more detail the argument for prerelease access that having the statistics as soon as possible, and specifically earlier than market participants get to see them, provides an opportunity that would not exist otherwise to take measures (do and say things) that preempt market instability. This argument is based on the assumption that the markets can be irrational as a whole in their reaction to news, especially when they are left to interpret the news by themselves with no guidance from the government, leading to excessive financial/economic instability. Thus, the role of the government is to help the markets react in ways that do not lead to instability, by offering at the time of statistical release interpretations of developments and plans of actions, as well as tangible actions, that can help avert extreme market reactions.

First, we note that there may be an issue here regarding whether the markets really are waiting for the government to interpret the statistics for them. We would argue that this is probably an exaggerated view of the ‘gullibility’ of markets. Second, regarding taking action earlier rather than later, there is, in our view, in principle, a point there. In some cases, it is sensible to think that quick preemptive policy actions/measures can lead to reducing real welfare losses. However, the critical issue here is why the government or the central bank have to wait and rely on the official statistics producers to become aware that they need to take measures. Policy makers do have access to rich administrative data sources all along the way. If the ministries and the central bank are run with any level of effectiveness, their decision makers should have advance warning on problems well before official statisticians get to produce their statistics, many of which are sourced from these administrative data. On the basis of such early administrative information, necessary actions can be contemplated and decided upon soon enough so as to minimize market instability. Admittedly, in most cases, the figures produced by statisticians will be more accurate and complete than those of the policy analysts who are relying on administrative data, but these differences would not be the kind of differences that would require super urgent measures to be taken by fiscal and monetary authorities. Thus, while the argument in the previous paragraph appears to be a sensible one in favor of prerelease access, it is not very convincing as it does not take into account the reality of information gathering and use by modern policy making institutions entrusted with preserving market stability.

4.3Unfair political, economic and career gains from prerelease of statistics

An important broad argument against prerelease access to official statistics by government, the press or anyone else is the following:

  • It allows for benefits to accrue to some users and not others and these benefits can be of all sorts – political, economic, career – and they are all equally unacceptable

An important point that is often not in the discourse on prerelease access is that, as one would readily consider unacceptable that a trader makes money out of getting hold of the statistics early, one should also consider it equally unacceptable if a politician gains politically because he got hold of the statistics earlier than his opponents. Similarly, a journalist that has prelease access gets to gain in terms of his career (and whatever that brings to him), while his colleagues and competitors that do not have prerelease access loose in this respect. And this is not fundamentally different from (‘insider’ style) trading on the basis of statistics accessed early.

Below, we look into some of the more specific arguments against prerelease access under this broad argument.

  • It treats users in the political sphere unequally and creates an uneven playing field in political competition

Prerelease access by government infringes on the statistical principle of impartiality when it gives a head start to the party in power relative to its political opposition and the electorate.

The presumption that those in power have to have an opportunity to study the statistical results before anyone else is false. Moreover, the argument presented in favor of prerelease access by government officials – that such prerelease access allows the government to be ready to make a comment or put out a press release exactly at the time of (to coincide with) the statistical press release– actually points to a fundamental argument against prerelease access. It is not desirable to have such simultaneous commentary and policy pronouncements, and it is not desirable for a number of reasons.

Prerelease access by the government does not give an equal opportunity to others in the opposition or the press to study the data and reduces their ability to ask meaningful questions or make informed comments and criticisms of the relevant policies in a timely manner. Instead, the government is the one prepared with sophisticated interpretations and analyses. It should be noted that the government in power gets a leg up on the opposition by taking advantage of access to statistics that do not really belong to it but to the public as a whole, including of course the political opposition.25 In this context, it could be noted that arguments of the sort that “government should be able to see the statistics early because it pays for them” are wrong as it is not the government in power that pays for the statistics but the people of the country.

Government politicians/policy makers also get a leg up on the public/electorate, which may be manipulated more readily by politicians/policy makers who have prerelease access. Prelease access enables the government in power to ‘spin’ the news and also to ‘run ahead of’ or preempt potential sceptics and critics by preparing and taking actions. Moreover, the government may potentially choose to offer to the public ‘alternative news and headlines’ and take diversionary actions, all with a view to divert the attention of the public from the news that the released statistics bring.

In other words, the government is making ‘political money’ unfairly out of this arrangement in a similar way as traders that may have the statistical information before others make money unfairly out of ‘out-positioning’ traders that do not. Government ministers do not have to look smart and good handling the news in the statistics, while the opposition looks inept and weak. If this happens, then this is an argument against prerelease access, not for it.

  • It allows the government to take actions on a matter for which it would have proceeded differently in the absence of the early access to statistics and this can lead to troubling situations and outcomes

The issue here is what kind of actions the government will take if it has access to official statistics before everyone else does. The government could, for example, in advance of bad economic news from an impending statistical release take action to move the timing of an international bond issue to avoid paying more for government borrowing. Alternatively, the government may decide to engage in diversionary foreign policy or in preemptive attacks on the previous government branding it responsible for the unfavorable figures that will come out soon. These do not appear to be legitimate actions that should be taken on the basis of the privileged knowledge gained by prerelease access. In addition, it is not clear where the line of legitimate actions resulting from prerelease action is. However, it is clear that some politically and economically unfair actions towards some people and groups will at some point arise from the prerelease access official statisticians provide to the government in power. There will be actions that are politically unfair to the political parties not in power. There will also be actions that are economically unfair to market participants (investors or workers) that are ‘blindsided’ by the government acting on the basis of privileged early access to official statistics, although these statistics truly belong to everyone. And the ‘blindsiding’ will not always be for the ‘public good’, at least from the point of view of some of these market participants. So the question is: knowing that prerelease access can and does lead to redistribution of political and economic power between political and economic groupings and populations, should one consider prerelease access consistent with statistical ethics?

  • It opens the door to opportunities for profiteering behavior in economic terms by some of those that get early access to the statistics

People with prerelease access, either in the government or in the press, and/or their associates could potentially (act illegally and) take positions in markets informed by the news in the statistics. Research has provided evidence that something like this is going on in some cases and the press has highlighted it in reporting on prerelease access issues.26,27 We are definitely not saying that all those receiving prerelease access use the information to reap economic benefits. However, there is a positive probability that some individuals (and their associates) will reap economic benefits when prerelease access is allowed. And this probability becomes larger, the larger the number of individuals having prerelease access, ceteris paribus. It should be noted that there are big incentives to abuse information from prerelease access and some individuals may succumb. And these incentives have become increasingly bigger in recent years with the spread of algorithmic trading. As testimony to the US House of Representatives has highlighted [19]: “A few years ago, a few seconds here or there would not have had much of an impact. Today, fractions of a second can equate to millions or even billions of dollars in market movements”. The magnitude of the stakes given the risks of prelease access is clear to market participants.28

Of course, technologies and processes can be used in some cases, such as those employed in the context of access offered to journalists in lock down facilities, to reduce the probability of information reaching some market participants before the rest of them. However, processes and technologies can fail or they can be circumvented by other (ever evolving) technologies. In the end, prerelease access by the press is not and cannot be fully consistent with protecting the level-playing field for market participants and can lead to profiteering by some, again undermining the principle of impartiality.

Having said that, the problem of leaks to some market participants may actually be an equal or greater problem when politicians/policy makers and their staff have an early peek at the statistics than when journalists get prerelease access. And this is because members of the press receiving prerelease access are usually monitored more closely and are more constrained than those persons that have prerelease access on the political/policy side. There is evidence, for example, from the case of the UK that when the government had significant prerelease access, markets also behaved as if some participants were getting the figures early also; but when prerelease access by the government was significantly curtailed in the UK in 2017, the indications that market participants were getting the statistics early also decreased [21]. To avoid the problem, there should be no prerelease access either by the press or the government.

If prerelease access increases the probability that some will profit at the expense of others in the market, and there is evidence that this does happen even in sophisticated advanced economy settings, then this implies that less advanced economies may find it even more difficult to contain the problem of profiteering from prerelease access. It argues for eliminating prerelease access anywhere in the world.

  • It can lead to market behavior that leads to suboptimal allocation of resources and welfare losses for society

The perception of the markets that prerelease access is leading, or may lead, to leaks to some market participants is something that affects the behavior of markets. Such changes in behavior are real29 and affect resource allocation in the economy, leading over time to lower welfare than could otherwise be achieved.30 Thus, over the long run market participants would tend to favor restriction instead of permissiveness in prerelease access. As noted above, in order to avoid the problem of market perception, a policy of ‘no prerelease access’ should apply both to the press and the government.

  • Prelease access begets more prerelease access, increasing the probability that the statistical information will be used unethically

Prerelease access is a ‘slippery slope’ in that some initially restricted number of individuals having such access increases over time, and as that number increases so does the probability of leaks to market participants.31 Prelease access is also a ‘slippery slope’ when it comes to giving access to different groups of users: if you give access to the government then the press wants it and is more likely to get it and, of course, vice versa. Moreover, prerelease access is a ‘slippery slope’ when it comes to various producers within a national statistical system. If one producer provides such access, then this serves as an argument made to – and as leverage on – other statistical producers to also provide prerelease access. Eventually, prerelease access allowed in foreign national statistical systems (in the region or beyond) is used as an argument to adopt or expand prerelease access in the national statistical system, usually with pressure to adopt the most ‘permissive’ model.

  • Prerelease access to the press and other nongovernment actors entails significant risks while the limited help it provides for the public to better understand official statistics can be achieved to a large extent via alternative means

The issues with prerelease access by journalists – whether via lock-up rooms or embargo arrangements – are whether such access is truly secure, whether it is fair, and whether it can be contained so that it does not lead to more journalists as well as the government politicians/policy makers demanding and receiving equal treatment. The answers to these questions in our view should be in the negative on the basis of actual experience.32 Moreover, we would argue that prerelease access is not the appropriate way to earn goodwill with members of the press. Admittedly, without prerelease access by journalists there may be a greater need for ex post clarifications and interventions by the official statistics producer when the statistics are being wrongly presented by some in the press. However, if members of the press are well prepared over time with briefings/workshops by the official statistics producer on the methodologies used in the production of statistics and if they are offered a technical briefing immediately after a release of statistics, then the above issues would be minimized, at least for those commentators that do not misrepresent the statistics intentionally.

5.Conclusions and recommendations

We have presented arguments for and against prerelease access to official statistics and we firmly believe that the arguments against such access outweigh the arguments for it. Prerelease access to official statistics is contrary to ethics in the production of official statistics, putting at risk the integrity of statistics and the trust of the public. We have also aimed to show that actually the arguments for prerelease access have truly little merit, surely from a statistical point of view, but also from the point of view of those who receive such access – the government in power and some members of the press. It is in the long term interest of all participating in the political process (politicians and political parties) and in the markets that there is no prerelease access. No prerelease access is also in the long term interest of the broader public in the country where the official statistics are produced, as such practice best safeguards the public’s right to impartial and objective information from official statistics at every stage of the statistical process and thus it safeguards the public’s democratic and human rights. Absence of prerelease access is more consistent with optimal allocation of resources in an economy and welfare maximization. Finally, it is also best for international collaboration and cooperation.

It is no wonder that equal access to official statistics by all users at the same time is recognized as ‘first best’ practice for adhering to the statistical principles of impartiality and objectivity as well as of independence. Yet, practice varies across countries and from one statistical producer to another within a country, with the situation being particularly concerning in the case of official statistics production outside the national statistics office. We believe prelease access is a legacy of older times, when official statistics were the statistics ‘of the government’. However, this is not the case anymore. Official statistics are a (global) public good and they are not the statistics of the government, although they are produced within the public administrations of nation states [16]. Official statistics belong to all users.

Thus we recommend the complete elimination of prerelease access. This includes prerelease access to politicians and policy officials of the government in power as well as members of the press or any other user. It is the cleanest/easiest approach to implement, as well the most just and the most stable approach.

We note movement in recent years away from prerelease access, for example in large parts of the EU. We would argue progress in this area should be consolidated and further strengthened, ensuring that not only formal but also any occasional informal prerelease access is eliminated. There is important progress that needs to be done by major advanced economies; it would be catalytic for progress in smaller countries and less advanced economies around the world.

We call for a strengthening of codes of principles and ethics for official statistics in this area to support this movement away from prerelease access. The reason is that a number of the main codes of ethics may display a certain degree of “double think”33 regarding the specific issue of access by all users and prerelease access when they discuss the statistical ethics principle of impartiality! For example, the European Statistics Code of Practice (COP) [12] first notes under the principle of impartiality and objectivity: “All users have equal access to statistical releases at the same time.” Yet in the very next sentence it says: “Any privileged pre-release access to any outside user is limited, well-justified, controlled and publicised.” Thus, first, it says unequivocally that all users have equal access to statistical releases at the same time, which is a strong and clear statement, but it follows it up by describing privileged prerelease access, seemingly as an acceptable variation of the first sentence. In a way, it seems to say that ‘all users are equal, but some users are more equal than others’.34

On the other hand, UN codified principles such as the UNFP and the Principles Governing International Statistical Activities [22] are actually not contradictory in this respect, with the latter stating clearly: “Providing equal access to statistics for all users.”

The persistence of clear contradictions in the language of some major codifications of ethical principles for official statistics is the result of significant ties still tethering official statistical production to the governments statistical producers are formally part of. As the institutional independence of official statistics production gains strength [1, 24], such contradictions in the codifications of ethics for official statistics will be resolved. However, it is our hope and recommendation that the contradictions regarding prerelease access in the codifications of ethics of official statistics would be resolved as a matter of priority and as early as possible so as to be used as a tool in the effort to eliminate prerelease access. Otherwise, the relevant language will continue to be used as an excuse to not make substantive progress in this very important area for the integrity of official statistics and for the trust that these statistics should be accorded.

Notes

1 Users are understood as all but the official statisticians involved in the production, quality assurance and release process of the final statistics. Users are supposed to have equal access to the statistics at the same time i.e., have access simultaneously/concurrently.

2 Statistical perimeter is defined [1] as the line between those outside the statistical perimeter and those inside, whereby outside the statistical perimeter are users of official statistics such as policy makers, legislators, civil servants/administrators as well as the press, market participants, academic researchers and the general public, and inside the statistical perimeter are official statisticians directly involved at a given point in time/stage of statistical production (including quality assurance and release). It should be noted that administrative data or other upstream data providers are not within the statistical perimeter for the production of the downstream statistics but outside it.

3 The list of individuals granted prerelease access may stay stable or vary between releases of a given statistical product. The list may be formally and explicitly justified for each specific individual or it may be justified in broader terms for categories of users or it may not be justified at all. The list and its justification, if it exists, may be public or it may not be. The formal author of the list may be the statistical producer, e.g., the National Statistics Office, or it may be a cabinet member or some other government official. The actual decision maker on the contents of the list may also be any of the above. The time before global release during which statistics are provided under prerelease access may be of various durations and may be stable or vary across statistical products or even from one release of a given product to another; the way this is decided upon may also vary. In general, decisions on prerelease access and its modalities may be taken every time there is a release or they may be taken for a longer period of time and enshrined in some form of regulation; decisions may be publicized and transparent or not. Very importantly, prerelease access can be and often is completely informal and not transparent or subject to any accountability.

4 In 2017, the UK eliminated some of the prerelease access previously available. Specifically, following implementation of the Statistics Act (SRSA) in 2008, England and Northern Ireland adopted 24 hours’ prerelease access, whereas Scotland and Wales retained the previous 5 days’ prerelease access but introduced the 24 hours for market sensitive data. The significant curtailment of prerelease access that took place in 2017 only related to Office of National Statistics data, leaving other departments in England (ONAs) and Northern Ireland still on 24 hours and no change in Scotland and Wales.

5 For example, according to OECD country self-assessment documents [2].

6 Based on anecdotal information collected by the author in his research.

7 See, for example, the Statistics New Zealand relevant policy [3].

8 The Royal Statistical Society notes: “It is sometimes argued that access to statistics at the earliest opportunity is necessary for ministers to manage the affairs of their department” [4].

9 The policy in the UK that was in place until 2017 specified: “Pre-release may only be granted to ensure that Ministers and others with similar responsibilities are able to respond or take action at the time of release; Pre-release may only be granted where to deny such access would significantly impede the response” [5].

10 Thus, for example, prerelease access used to be provided until June 2017 to “certain members of the Bank of England and HM Treasury for longer than 24 hours to facilitate the effective coordination of fiscal and monetary policy” [5].

11 And even now, after the UK has eliminated a good part of prerelease access, “a limited number of officials at the Bank of England would still be granted prerelease access when an economic data release coincides with a BOE interest-rate decision” [6].

12 In Canada “a pre-release of less than 24 hours is made of a handful of key series to designated senior officials in the departments for purposes such as the orderly management of money markets” [2].

13 The necessity for comment or a press release by the government at the same time as the release of the statistics is indicated in the UK policy in place until 2017 [5].

14 The Australian policy on the matter [2] reflects this argument: “The provision of pre-release access may be appropriate where either: a. the interpretation of the release is likely to be complex enough to require analysis in advance to support informed responses and commentaries; or b. on the day of release a relevant Australian government official, Australian government entity or Australian government minister is likely to be asked to provide public comment on the release.”

15 For example, the Brazilian President, according to press reports [7], on July 31, 2019 “called for tighter control of official deforestation data, resuming attacks on government space research agency INPE, whose satellite imaging shows a jump in clearing of the Amazon rainforest this year. Bolsonaro, who on Friday accused INPE of falsifying data, added that data should be shown first to a member of his cabinet before being disclosed publicly. Mid–July [2019] INPE data showed Amazon deforestation hitting the fastest pace in years, raising red flags for a trade deal with the European Union.”

16 According to press reports [8], “Mr Bolsonaro said the institute [INPE} was smearing Brazil’s reputation.” Incidentally, the head of INPE was fired on August 2, 2019.

17 To see that this must be true, one can consider the case where official statistics are produced on an objective basis, i.e., without taking into account the desires of certain political interests for a certain statistical result, but then the statistical office does not release the statistics to the public as its results offend the political interests. One would not consider that these statistics are still compiled on an objective basis determined by statistical considerations. The notion of compilation is much broader and it would not make any sense to be otherwise.

18 An example is the case of the leader of the Central African Republic (then Central African Empire) changing the population figures in the press release on the ca. 1975 national census results. Another example is the 2005 case in the United States when prerelease access led to the management and at least partial suppression of Bureau of Justice Statistics (BJS) data on traffic stops and racial profiling, which were showing disparities in how racial groups were treated once they were stopped by the police. According to press reports [14], political supervisors within the Department of Justice of which the BJS is part, having prerelease access, ordered the head of BJS to delete certain references to the disparities from a news release that was drafted to announce the findings. The head of the BJS refused to delete the racial references, arguing to his supervisors that the omissions would make the public announcement incomplete and misleading. Instead, the Justice Department opted not to issue the news release on the findings and posted the report online (which could not be found by the journalists), thus effectively suppressing to some significant extent indefinitely the statistical results. The head of BJS was removed and he was demoted to a lower position in another institution for the last six months of his civil service career.

19 Some politicians/policy makers preempting the statistical release on the basis of prerelease access do not do it to alter the final statistical result, but just to gain something politically. This of course has repercussions for the perception of the statistics as well as for unfair gains that can be made both economically and politically by some. Note the case of the US President, who in June 2018 effectively preannounced unemployment statistics news provided to him under prerelease access. He posted on Twitter at 7:21 a.m.: “Looking forward to seeing the unemployment numbers at 8:30 this morning” [15].

20 Thus, prerelease access enables greater self-censorship by statistical producers at earlier stages of the statistical process. The mere expectation of showing the statistical release to a nonstatistician with significant authority before publication, when there is some sense of what the nonstatistician would find comfortable and what not, will tend to bias not only the presentation of the statistical release but also the earlier stages of the production of the statistical results. This holds true whether the statisticians are aware or not of this psychological mechanism.

21 For example, in July 2019 Brazil’s President announced the introduction of prerelease access to, apparently, be able to activate these options, when he said that “data should be shown first to a member of his cabinet before being disclosed publicly” [4]. Such options were evidently used, when it was felt by the politicians/policy makers in charge that it was necessary, in the 2005 US case of the significant suppression of certain Bureau of Justice Statistics results [14].

22 Admittedly, this point depends on the public in the country having a good political/democratic culture and not being attracted by ‘strongman’ overreaching approaches. This is why, the argument we make here should be seen as a ‘long run’ argument.

23 The Royal Statistical Society notes [4]: “It is sometimes argued that access to statistics at the earliest opportunity is necessary for ministers to manage the affairs of their department. It is however the case that where statistics are derived from administrative data and management information, managers and ministers can access those data directly.”

24 For example, the government may provide early estimates of the general government deficit.

25 This kind of problem of uneven playing field has prompted countermeasures in some statistical systems, such as the one in the US. Office of Management and Budget (OMB) Directive 3 from 1985, which provides for prohibiting “employees of the Executive Branch from commenting publicly about the release of PFEIs [Principal Federal Economic Indicators] until at least one hour following their release.” Thus, while prerelease access is maintained (and usually is significant in terms of time before official release) there is a rule of no public comments by government officials on the figures for a certain period of time after the statistical release. This tends to even the field to some extent, regarding the political advantage imparted by prerelease access by government. It could also be noted that the current US administration has formally sought since April 2019 to change the Directive in this aspect [17].

26 For example, the press has reported [6]: “Analysis for the Journal showed that U.K. markets often anticipate that economic data is going to be better or worse than analysts expect ahead of its release. In 59.5% of 172 U.K. economic data releases between April 2011 and December 2016, British government-bond futures correctly anticipated the rise or fall that ultimately happened when economic data were published, according to an analysis prepared for the Journal by Alexander Kurov, associate professor of finance at West Virginia University. The analysis for the Journal also showed that the British pound often moves sharply in the hour before data are released. But in Sweden, where there is no pre-access, the krona shows no signs of moving ahead of official numbers, an analysis of trading data between January 2011 and March 2017 suggests.”

27 In a 2016 working paper Kurov et al. wrote [18]: “We examine stock index and Treasury futures markets around releases of U.S. macroeconomic announcements. Seven out of 21 market-moving announcements show evidence of substantial informed trading before the official release time. Prices begin to move in the “correct” direction about 30 minutes before the release time. The pre-announcement price drift accounts on average for about half of the total price adjustment. These results imply that some traders have private information about macroeconomic fundamentals.” We note that this movement about 30 minutes before the time of official release aligns with the 30-minute prerelease ‘lock-up’ period allowed by US federal statistical agencies, when journalists get prerelease access in special facilities.

28 See for example [20]: “The data should be kept so tight,” Love [a trader of interest-rate derivatives at Le Groupe Jitney Inc. brokerage in Montreal] said in a telephone interview. “Do you know how much money you could make by having the data one day before? It’s crazy.”

29 See for example [20]: “David Love, a trader of interest-rate derivatives at Le Groupe Jitney Inc. brokerage in Montreal, said he’s wary of taking market positions ahead of key releases. “You have to play safe,” Love said.”

30 Manove [22] shows the effects of insider trading on company shares. A basic result is that corporate investment would be lower in the presence of such insider trading. He argues that this also holds true for the entire securities market.

31 This problem had been noted in the case of the UK before its decision to significantly reduce prerelease access in 2017. Observers reported [6]: “In 2008, the number of recipients of major economic data releases was cut after a concerted by senior statisticians and lawmakers, including Jack Straw the then Labour Party government’s then Justice Secretary. But the numbers soon rose again. For instance, at the end of 2011, 77 people received U.K. labor market data 24 hours before its release. By the end of 2016, that number had risen to 122.”

32 Some examples of breaches of the protocols of the US Bureau of Labor Statistics were provided in the 2012 testimony to the US House of Representatives of former Senior Advisor for Communications and Public Affairs, C. Fillichio [19]: “Over the years, there have been different types of violations of security protocols, technological and otherwise. For example: A news organization installed a fiber optic line through one of the department’s fiber optic hook ups, located in a DOL [Department of Labor] phone closet, despite having been told repeatedly that use of fiber optic cable was not allowed. A news organization asked to disconnect and replace the “black box” supplied by DOL to disable Internet activity during the embargo period. While we explicitly denied this request, it was later discovered that the news organization had ignored this directive and replaced the “black box” without permission. Numerous media organizations failed to comply with DOL’s requirement that they inventory and label their data lines located in the lock-up facility. On two separate occasions, a wire service inadvertently transmitted sensitive employment and productivity data to its subscribers during the embargo period. The problem was related to a computer cable connection. A lock-up participant was sanctioned for retrieving his BlackBerry from a storage container and using it during the embargo period. Following that incident, the containers used to hold electronics were replaced with lockboxes. A reporter had his credentials temporarily revoked for using a mobile device to take a photograph during the lock-up. Participating media organizations have contacted the department frequently to voice concerns that competitors may have gained unfair advantage in speed of transmission or have surreptitiously broken embargoes.”

33 Employing the concept of “doublethink” the ministries in Orwell’s 1984 had names that conveyed the opposite of their true nature: “The Ministry of Truth, which concerned itself with news, entertainment, education, and the fine arts. The Ministry of Peace, which concerned itself with war.”

34 Similar contradictions can be found, for example, in the UK’s Code of Practice for Statistics, the OECD’s Recommendation on Good Statistical Practice, and the European System of Central Banks Public Commitment.

References

[1] 

Georgiou A. The production of official statistics needs to be a separate branch of government. Statistical Journal of the IAOS 34 (2018), 149–160. https://content.iospress.com/articles/statistical-journal-of-the-iaos/sji170399.

[2] 

OECD. Good statistical practice toolkit. Country assessments. Available from: http://www.oecd.org/statistics/good-practice-toolkit/countryassessments/ (accessed 27 January 2020).

[3] 

UN Statistics Division. Development of National Statistical Systems. Statistics New Zealand Release Practices Policy. Available from: https://unstats.un.org/unsd/dnss/docViewer.aspx?docID=135#start (accessed 27 January 2020).

[4] 

Royal Statistical Society. Influencing Change. Data Manifesto. Public trust and pre-release access. Available from: https://www.rss.org.uk/RSS/Influencing_Change/Data_manifesto/Public_trust_and_pre-release_access_of_statistics/RSS/Influencing_Change/Data_democracy_sub/Public_trust_and_pre-release_access_of_statistics.aspx?hkey=e890776e-aa19-4da3-bae6-def16454c9c7 (accessed 27 January 2020).

[5] 

UK Government. Cabinet Office. Review of prerelease access to official statistics. Available from: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/62084/pre-release-stats.pdf.

[6] 

Bird M. U.K. Government’s Controversial Early Peeks at Economic Stats to End. Wall Street Journal. 15 June 2017. https://www.wsj.com/articles/u-k-statistics-body-to-end-government-prerelease-access-1497535987 (accessed 27 January 2020).

[7] 

Reuters. Bolsonaro calls for tighter control of Brazil deforestation data. 22 July 2019. https://www.reuters.com/article/us-brazil-environment-idUSKCN1UH22D (accessed 27 January 2020).

[8] 

BBC. Brazil deforestation row: Space research head Galvão out. 2 August 2019. https://www.bbc.com/news/world-latin-america-49212115 (accessed 27 January 2020).

[9] 

The Merriam-Webster.com Dictionary, s.v. “impartial (adj.),” https://www.merriam-webster.com/dictionary/impartiality?src=search-dict-box (accessed 27 January 2020).

[10] 

United Nations Statistics Division. Fundamental Principles of Official Statistics. Available from: https://unstats.un.org/unsd/dnss/gp/fundprinciples.aspx.

[11] 

The Merriam-Webster.com Dictionary, s.v. “objectivity (n.),” https://www.merriam-webster.com/dictionary/objectivity (accessed 27 January 2020).

[12] 

European Statistics Code of Practice. Eurostat website. Available from: https://ec.europa.eu/eurostat/web/products-catalogues/-/KS-02-18-142.

[13] 

International Statistical Institute Declaration on Professional Ethics. International Statistical Institute website. Available from: https://www.isi-web.org/images/about/Declaration-EN2010.pdf.

[14] 

Lichtblau E. Profiling Report Leads to a Clash and a Demotion. NY Times. 24 August 2005. Available from: https://www.prisonpolicy.org/scans/disturbing/2005/profiling.shtml.

[15] 

Tankersley J, Phillips M. Trump Touts Jobs Report Before Official Release, Breaking Protocol. NY Times. 1 June 2018. Available from: https://www.nytimes.com/2018/06/01/us/politics/trump-jobs-twitter.html.

[16] 

Georgiou A. Towards a global system of monitoring the implementation of UN fundamental principles in national official statistics. Statistical Journal of the IAOS 33 (2017). https://goo.gl/R8NKzU.

[17] 

Federal Register. Statistical Policy Directive No. 3: Compilation, Release, and Evaluation of Principal Federal Economic Indicators. Timing of Public Comments by Employees of the Executive Branch. A Notice by the Management and Budget Office on 04/11/2019. Available from: https://www.federalregister.gov/documents/2019/04/11/2019-07172/statistical-policy-directive-no-3-compilation-release-and-evaluation-of-principal-federal-economic (accessed January 27, 2020).

[18] 

Kurov A, Sancetta A, Strasser G, Halova Wolfe M. Price drift before U.S. macroeconomic news: private information about public announcements? European Central Bank Working Paper Series No. 1901/May 2016. https://www.ecb.europa.eu/pub/pdf/scpwps/ecbwp1901.en.pdf?ca0947cb7c6358aed9180ca2976160bf.

[19] 

Testimony in 2012 of Senior Advisor for Communications and Public Affairs of the US Bureau of Labor Statistics C. Fillichio to the US House of Representatives. Available from: https://www.dol.gov/newsroom/lockups/testimony.

[20] 

Mayeda A, Quinn G. Policy aides getting Canada data early. Financial Post. 17 October 2011. Available from: https://business.financialpost.com/news/economy/policy-aides-getting-canada-data-early (accessed 29 January 2020).

[21] 

Bird M. Suspicious market movements wane after UK bans early peeks at data. Wall Street Journal. 23 December 2019. Available from: https://www.wsj.com/articles/suspicious-market-movements-wane-after-u-k-bans-early-peeks-at-data-11577096078 (accessed 29 January 2020).

[22] 

Manove M. The harm from insider trading and informed speculation. The Quarterly Journal of Economics. November 1989. Available from: http://sites.bu.edu/manove/files/2013/05/ManoveHarmfromInsiderTradingQje1989.pdf.

[23] 

UN Statistics Division. Principles governing international statistical activities. Available from: https://unstats.un.org/unsd/methods/statorg/Principles_stat_activities/principles_stat_activities.pdf.

[24] 

Georgiou A. Extracting statistical offices from policy-making bodies to buttress official statistical production. Journal of Official Statistics 35(1) (2019). Available from: https://search.proquest.com/openview/aa00cd2339e01bd5b3f29f0023524fcf/1?pq-origsite=gscholar&cbl=105444.