Securing the independence of official statistics: Introductory remarks

from left to right: Hallgrímur, Pali, Andreas, Dennis, Gerry, Jean-Luc

The topic of securing the independence of official statistics was discussed in an invited paper session at the WSC 2017 in Marrakech under the heading “Does the production of official statistics need to be a separate brach of government or should it remain within the executive branch of government?” The session was organised by Andreas V. Georgiou, former President of the Hellenic Statistical Authority, ELSTAT, the national statistical institute (NSI) of Greece. The session included papers given and presented by Andreas V. Georgiou, Jean-Luc Tavernier, the DG of INSEE (the NSI of France), Dennis Trewin, former Australian Statistician and head of the Australian Bureau of Statistics, ABS, and Pali Lehohla, Statistician General of South Africa. Gerry O’Hanlon, former Director of the Central Statistical Office, CSO, of Ireland acted as discussant. I chaired the session. All the papers presented at the session appear here below.

The preamble to the session states that there are no more important issues in official statistics than their independence, impartiality and objectivity. These issues, it is argued, may be at risk when the production of offical statistics is carried out within the executive branch of government as official statistics are partly compiled for monitoring the operations of government and outcomes of government policies. Hence, a case can be made for the production of such information being kept outside of the government executive. Otherwise, the information may not fulfill the conditions of being impartial and objective. The risk of conflict may, however, be mitigated by a framework of legal and institutional safeguards and in many countries the autonomy of the main producer(s) of official statistics is supported by the prevailing culture, traditions, and good practices. But it is uncertain if the traditions and institutional settings are robust enough to ensure sustainable statistical independence when official statistics are part of the executive.

These issues are likely to be viewed differently in different countries; in some cases the situation may be considered robust and sustainable, in other cases there may be a need for strengthening the safeguards for the independence of the NSI and other main producers of official statistics, perhaps even to the extent of moving the official statistics out of the government executive. In that case, it will have to be shown clearly what would be the net real gain of having the official statistics produced outside of the government executive. Or indeed if there would be a real difference between these two situations, bearing in mind that the funding of the official statistics would still be dependent on or at least heavily influenced by the government executive.

All this boils down to the main issue – how to best ensure independence of official statistics. And this was the issue discussed at the Marrakech session. The issue was not selected at random; it may be thought to be farfetched in some countries, but experience has shown that it is both real and troublesome. There is recent evidence for this. In Argentina, the government insisted on interfering with the compilation of the CPI which resulted in erosion of the credibility of official statistics, at home and internationally. In Kazahkstan, the government found it opportune to have the head and two main managers of the NSI take the blame for shortage of funds for the population and housing census and had them jailed for several years. In Greece, the former head of the NSI is being prosecuted for complying with the rules of both Greek and European regulations and practices in carrying out his functions as the cleaning up of the Greek official statistics resulted in outcomes that clearly showed up bad policies and practices (as well as considerable statistical misinformation) of the government at the time. These are extreme cases but there may have been others less extreme but still disturbing ones that have occurred in some countries, related to specific sectors or particular policies. Government finance statistics are a case in point as are measurements of outcomes of social policies.

It is against this background that the WSC session was organised and conducted. All the presenters were aware of the background and the issues, being or having been the main upholders of the principles and practices of independent official statistics, each in their own country. They all agreed that the independence, impartiality and objectivity is the fundamental issue for official statistics. But their approaches and their conclusions differed, reflecting their background and experience, and their views of the robustness of statistical systems, international principles and guidelines, and their abilities to withstand political pressures.

Andreas Georgiou argues that in order to avoid conflict and ensure independence it is necessary that the production of official statistics is not carried out by one of the branches of the government whose performance is assessed on the basis of these statistics. One of the fundamental conditions for sutainable and robust professional independence of official statistics, their impartiality and objectivity, is the institutional independence of official statistics. The situation in most countries with official statistics being produced within the executive branch of government is, in his view, inherently not appropriate and should be changed with the official statistics becoming a separate part of government. Thus, the institutional independence of official statistics is a fundamental condition for their sustainable professional independence, impartiality and objectivity. This is partly the case as the safeguards, checks and balances applied at present are insufficient. Hence, Georgiou advocates a paradigm shift in this respect; that the official statistics are formally declared in law as separate from the executive and established as an independent branch of government. This arrangement would need to be supported by a number of legislative provisions and policy commitments to ensure the independent operations of the statistical authorities. These would need to involve all aspects of professional independence including the selection of the head of the statistical office, recruitment and human resource issues, use of budgetary funds, and independent reviews of the operations. In conclusion, it is emphasised that to fully and sustainably apply the principle of professional independence of producers of official statistics, these need to be institutionally independent of the executive branch of government.

Georgiou’s view is not supported by the other authors. Jean-Luc Tavernier and Dennis Trewin maintain that the experience in their countries, France and Australia, shows that conducting official statistics within the executive branch of government has not invited or encouraged political interference. They both emphasize the importance of official statistics being carried out by government institutions that enjoy professional independence, partly de jure but no less de facto; institutions that are held in high esteem and are generally trusted. They acknowledge that the independence and the trust cannot be taken for granted; it has to be worked at and earned through professional and prudent operations. These authors both assign much importance to close working relations with the main user of official statistics, the government, for supporting the professional independence of the NSI and for reasons of funding. Trewin, in particular, emphasises the importance of transparency for upholding the trust in the official statistics. Both Tavernier and Trewin conclude that moving the NSIs in their countries out of the executive branch of governement would most likely result in weakening and marginalising their stature and operations.

Padi Lehohla is less occupied by the formalistic or institutional arrangements of the official statistics but more so with their real role producing the evidence on which political decisions are based. He wonders if there is still space for the national statistician to carry out his/her role. This role is being challenged and the traditional demarkation between statistics and policy has been diffused by technology. As a consequence, our conventional statistics are seen to be slow and cumbersome and not able to fulfill expectations for real time statistical evidence. Lehohla maintains that it is necessary for the official statisticians to abandon their safe ride on the smooth side of the saw which will inevitably lead to their extinction, and risk riding on the serrated side, taking on the challenges of technologies superseeding measurements and of producing real time evidence for political decisions. Do the UN Fundamental Principles of Official Statistics continue to hold in this environment, he asks. The answer is affirmative; more than ever are they needed as the deluge of data and new ideas can soon dissipate in the absence of standards and ethics.

In discussing the four papers, Gerry O’Hanlon is not convinced by Georgiou’s description of the situation of official statistics and finds that cases of government interference with official statistics have been few and far between. He also sees weaknesses in placing the NSI outside the executive branch of government. He points out that in most countries a considerable part of the

official statistics is produced by other institutions than the NSI. Moving these outside the executive branch is hardly feasible and having the NSI outside the executive but the other producers within it may create problems of coordination. As argued by Georgiu, the institutional independence would need to be supported by legislation but O’Hanlon doubts that these can guarantee independence as ways of side-stepping the legal provisions can always be found if there is will to do so. O’Hanlon agrees with Trewin regarding the usefulness of identifying in advance various scenarios that may arise and threathen the independence of and trust in official statistics and of planning how to act in such situations.

Thus, Georgiou’s arguments, main conclusions and recommendations are not shared by the authors of the other papers presented at the Marrakech session. Nevertheless, the posing of the main question in the session’s heading and his arguments have given rise to well-considered papers and lively discussions, both among the five authors and by others during the Marrakech session. The large attendance at the session and the quality of the discussion show clearly the importance the international statistical community attaches to the independence of official statistics.