This presentation is based on a presentation from World Bank looks at why governments set up statistical systems, how these systems operate and what they do. It analyzes the nature of official statistics and the problems they face as public goods. It also looks at how statistics are used, who uses them and the problems of maintaining a balance between supply and demand. The presentation will lead to a discussion of how different statistical systems are organized, some key management issues and how best to maintain the integrity of the data and ensure independence from political manipulation. The role of the international community is also touched on, including the use of data frameworks and standards, the need for coordination and the the need to support statistical capacity building, especially in poor developing countries. The UN’s fundamental principles of official statistics are introduced and these will form the basis for assessing progress and performance in the next part of the course.
The presentation is divided into five parts: A discussion of why countries need statistics and why governments need to be involved; A description of statistical systems in general, how they work in countries and links to the international level; A presentation of some basic principles underlying the operation of official statistical systems, why these are needed and what they imply for both governments and users; A review of some of the main challenges and opportunities facing official statistical systems today, especially those in developing countries; and A final section that leads into the later parts of the course outlining what needs to be done to build capacity on a sustained basis.
Almost all countries in the world, more or less regardless of their political persuasion, have identified the need to collect and use statistical data to support the business of government and have established statistical systems and set up specialized agencies for this purpose. Official statistics have a long history and there are documented examples from more than two thousand years ago of censuses and other types of enquiry to support revenue collection and to respond to external threats. Some of the earliest uses of the word “statistics” in Italian and German relate to the management of the state. Modern governments need a very large amount of statistical data to manage their affairs, to promote economic stability, to deliver effective services to their citizens and to promote investment and development. Official statistics serve to identify and quantify issues and problems, they provide a picture of the current economic, social and environmental situation of a country. They help to identify the most effective courses of action and they are essential for monitoring what progress is being made and for evaluating the impact of past actions. Although we all access and use a wide variety of information in many different forms, statistical data are especially important because they are provided as numbers. The use of numbers provides for a certain degree of objectivity and precision, although care is needed not to imply that all numbers are either objective or precise. In particular numbers can be manipulated using the powerful analytical tools of mathematics and statistics and, most importantly, results can be replicated.
Official statistics are public goods. The use of them by one person or agency does not detract from their use by another. While they are often costly to produce, they are readily disseminated and once they are publicly available, it is difficult to exclude other users. But the value of statistics depends upon their quality. Because it is not easy to ascertain the quality of statistics directly, users must have confidence in the producer and in the methods and standards employed in the production of statistics. For all these reasons, it is difficult to establish functioning markets in statistics, leaving national governments and international agencies to produce and disseminate statistical information. The fact that the large majority of official statistics are public goods has a number of important implications for the management and organization of statistical systems. Their collection and dissemination is undertaken by governments and is financed from general tax revenue. This is true whether or not the statistical agency is formally part of central government or has some other, more independent status. The lack of a functioning market also has implications. There is no price mechanism to provide signals about investments and managers need to identify other ways to identify what data are needed and where investment should be directed in the future.
Statistics produced and financed by governments have a dual role. They must serve the need of government for efficient administration and management as well as for longer-term policy making. They must also serve the need of the public to monitor the activities of government and the changes in public well being. To be effective, statistical systems must have legitimacy backed up by legislation that provides both the safeguards of confidentiality for the providers of raw data and assurances of integrity and accessibility for users. The nature and organization of national statistical agencies vary according to the political system, the demand for data, and the organization of local and central governments. In decentralized systems, separate agencies have independent mandates to gather and disseminate statistics in particular areas. But even in highly centralized systems responsibilities may be divided. Central banks, for example, usually collect data on money and banking and may be responsible for other areas such as balance of payments. Line ministries may compile and disseminate data derived from their administrative processes. Thus a national statistical system is often a network of agencies whose activities are ideally coordinated by legislation, administrative practices, and professional standards, essential if data discrepancies are to be avoided. Throughout this presentation references to the national statistical agency or statistical system should be understood to refer to all of those agencies or departments of government that are responsible for the collection and dissemination of official statistics.
Although the organization of statistical agencies and systems differs, the overall process they operate is much the same in all environments. Statistical agencies collect information in the form of different kinds of data from a number of different providers. These data are then processed, checked, summarized and analyzed and are then provided to different users in different forms or products. Although agencies need to understand what data are needed and how they are used, the primary function is normally one of collection, processing and dissemination. Because official statistics are mostly financed from general tax revenue, the decision of what data to collect and disseminate is basically a political one. The decision as to what kinds of data should have priority is generally one that belongs in the political arena. Professional considerations can influence the decision, but will usually not be the deciding factor. The question of how to collect the data, how they should be processed, analyzed and disseminated is mainly a technical one and should be decided on professional grounds. To be effective, statistical agencies need to have independence from political or any other source of influence on how they fulfill their mandates.
Statistical agencies collect data from a variety of data providers, using a number of different methods. In order to provide the information needed to manage development, agencies need to have contact with a large number of different kinds of data provider. These will include individuals and the households they live in and the owners and operators of a wide range of business enterprises, including both those that are formally registered and those operating in the informal sector. Other sources of information include the managers of public enterprises and service providers as well as representatives of civil society organizations and other groups. Data collection can be, and often is, quite intrusive. It is inherently a process where the costs are born by individuals, but the benefits accrue to society as a whole. In most cases providers are not recompensed for their time or the disruption caused by the whole process. While many countries have legislative provision to enforce the requirement for data provision, in practice, data quality depends crucially on the extent to which providers are willing to cooperate with the data collection process willingly. This in turn means that they need to be persuaded that the information is important and will be properly used. A key requirement of data agencies is to minimize the level of intrusion and disruption to providers. It is also essential that providers are convinced that the information they give cannot be used to their disadvantage. Data collected for statistical purposes must be and be seen to be confidential and information about individuals must never be disclosed to third parties, including other parts of government.
Statistical agencies use a number of different methods to collect the raw data that forms the basis of the statistics they produce. Censuses are usually complete enumerations of all the units in some population, such as all the people in a country (population census), all agricultural enterprises (agricultural census) or all business establishments in specified industries (economic census, or a census of business activity). They are usually large, expensive and complex data collection exercises carried out at fairly infrequent intervals. Most countries, for example, carry out population censuses only once every ten years. Surveys are usually administered to a random sample of the population of interest. Household surveys are used to collect data about the economic and social condition of the people and are important for monitoring poverty, for example. Business surveys are important for measuring production, employment and the collection of other economic data. As well as specialist statistical activities, governments also generate a substantial amount of data as a by-product of different administrative purposes. Examples include information collected about service delivery, licensing and the collection of revenue. Administrative data is generally cheaper to collect, but may provide the detail required. Other data sources include remote sensing (weather and traffic records, for example) as well as a wide range of participatory and qualitative methods. These can be very valuable in adding context to data derived from other sources.
Raw data are transformed into statistics for publication by a number of statistical processes. These include validation and editing, where the data as collected are checked for completeness and consistency and where, as far as possible, errors are corrected. Summary statistics, such as means, proportions, ratios, percentages and totals are calculated. Where the data are derived from a sample, then population estimates are calculated, together with estimates of the sampling error. More complex procedures may also be used, for example to allow for the effects of seasonality or to estimate for the effect of missing data. Typically, the data will be accompanied by commentary and explanation, to provide sufficient information for users to make judgments about the validity and utility of the results. The statistics may be disseminated in number of different products. The traditional output from statistical agencies has been printed reports and summary publications such as abstracts and digests.Increasingly, however, other media are also being used, including CD-ROMs and diskettes and release on the Internet through a dedicated web-site. Statistical agencies may also, from time to time produce reports on methods and procedures.
To be effective, and to ensure that the products and services they are providing continue to be effective, statistical agencies need to have a clear understanding of who their main users and customers are. Government at all levels, central line ministries and local administrations includes a major group of users, including policy makers, planners, analysts and managers. Politicians and legislators are also key users, needing data not only to support the legislative process, but also to hold the executive to account. Outside government, market agents of all kinds, both domestic and international use official statistics to inform their business decisions. They need to evaluate both the macroeconomic climate as well as conditions in their specific market. The public, often reached through the press, radio, TV and other media also form an important class of users. Civil society organizations (CSOs) and other lobbyists use data to monitor progress and also to advocate for action. Official statistics are also an important part of the democratic process, providing objective data for voters to review the effects of past policies and to assess competing platforms. The international community are also major users of statistics. Countries have a number of obligations to report on key indicators to different agencies, but development agencies and donors often require much more detail to identify the need for support and to monitor the impact of aid. Increasingly agencies are becoming more involved in policy dialogues with governments.
The activities of the international statistical system are an important complement to those of national systems. Developing countries are increasingly being integrated into world markets through the globalization process and it s important therefore, that national statistics can be compared and aggregated across countries. At the same time, national systems have obligations to report on basic indicators to different international agencies. Since the effectiveness of the international system depends crucially on how national systems operate, the international community has a direct interest in capacity building and in improving efficiency. Key areas where support is needed include technical advice, training and financial investment, the development of appropriate frameworks and standards, and overall coordination. Continuing support is needed for improving international frameworks and standards, encouraging international networking and coordination efforts, and improving the functioning of agencies that collect and produce global and transnational statistics. National data enter the international statistical system in a process through which specialized agencies review and further standardize national data to produce consistent, international data sets. Transnational data are data that transcend national boundaries in their production or coverage, for example the purchasing power parity statistics produced through the International Comparison Program and many environmental indicators. Beside compiling and disseminating data, an important function of the international statistical system is to agree on standard definitions and methods of data collection and to provide opportunities for the exchange of information between national and international statistical agencies. Although there is no complete description of the extent of the international statistical system, and no single authority to oversee it, the UN Statistical Commission is the agreed forum for the coordination of international statistical activities. Just as national statistics support national processes, international statistics support global policy formulation, decision making, and analysis by international agencies, businesses, researchers, the press, and private citizens.
Users official statistics are generally not able to determine easily the quality of statistics or statistical products. Without additional information on how the data were collected, what procedures were used in processing and what methods of classification and analysis have been employed, users cannot tell if published data meets their needs. A key requirement for any statistical agency, therefore, is to create trust in their products and to ensure that users have the information they need to assess data quality. Over the years a body of good practice and experience has been built up on what statistical agencies need to ensure the integrity of their data and statistical products and to provide assurance on data quality. This has been summarized as a set of basic principles by the United Nations. These ten fundamental principles of official statistics have been discussed and agreed by the United Nations Statistical Commission and provide a framework for assessing the performance of statistical agencies.
The first principle requires statistical agencies to collect, process and compile statistical data that is needed and can be used. At the same time, agencies are required to make their data and products available to all users on an impartial basis. In practice, this means that all users should have equal and simultaneous access. The second principle relates to the methods and procedures used to collect and compile statistical data. Statistical agencies are required to select the methods they use on professional and technical considerations only. The purpose of this is to ensure that political considerations play no part in deciding what methods to use. The third principle requires agencies to publish information on the sources, methods and procedures they use and to make these available to all users. In this way users can make judgments about the data and how they can best be used. The fourth principle ensures that agencies are able to comment on the erroneous interpretation or misuse of statistics. Many data items are politically sensitive and agencies have the right, not to enter into political debate, but to indicate when they think erroneous interpretations have been made or where they consider data have been misused.
Many statistical systems, especially those in poor countries are under stress and are under-performing. There are many reasons for this, but one that seems to be fundamental is that of a lack of demand for good statistics and consequently limited political support. This is often manifest in inadequate and declining real budgets, where especially the non-staff costs are inadequate even to maintain data systems, let alone extend and improve them. Other common problems include: inadequate human resources, and limited incentives or motivation for staff to work well; poorly designed and disseminated statistical products so that even when data are produced they are not widely used; limited feedback or communications between providers and users, so that data that are collected are not used and the statistics users want are not available; and ineffective coordination and management. A result of these constraints is that many systems are caught in a vicious cycle where inadequate finance and resources lead to poor quality output, which results in lower demand for statistics and hence low priority in the allocation of scarce resources.
The weaknesses of the statistical systems in many developing countries have long been recognized. Capacity building activities, including training, technical assistance, and sponsorship of surveys and censuses, have helped to improve the availability of important economic and social indicators. But much of this improvement has proved difficult to sustain and perhaps the key message that has emerged has been that sustained development depends as much on increasing the demand for statistics within a country as it does on improving supply. This is why the PRSP process, which generates substantial demand for many different kinds of data, is so important. Balancing supply and demand and ensuring the statistical systems are actually producing the data and products users want is not easy. Because there is no formal market for official statistics, the managers of statistical systems have no price signals indicating where investment should be directed. Other mechanisms are needed, including regular consultation with users and obtaining feedback from all stakeholders. Statistical agencies need to be much more actively involved in marketing their output and services.
Breaking out of the vicious cycle of under-performance and under-funding requires short-term as well as long-term action. Stakeholders need to see improvements straight away if they are to be convinced to support long-term capacity building. Managers need to address the trade-off between short-term responses to demands for more data targeted at specific programmatic questions and the need for long-term investments in strengthening statistical systems. In the short term, better use could be made of data that have been collected but not widely analyzed or disseminated – for example, survey databases and administrative statistics. Over the long run, decisions have to be made on priorities for data collection and investment in infrastructure, equipment, and human resources. Agencies need to build trust and to demonstrate that they are focused on improving data quality and on increasing efficiency. This may include using new technology, especially in information and communications to reduce costs and especially to improve timeliness of data release. Other areas of focus include strengthening management and coordination and making statistical systems themselves more transparent, responsive and accountable. It is ironic that many statistical systems are keen to collect and report on the performance of just about the whole of the rest of the economy, but are remarkably reluctant to measure and report on their own performance.
Support for statistical development should focus on country-owned and developed strategies. To guide their efforts, the World Bank and other partners in PARIS21 have encouraged countries to develop or update master plans for statistical development. Experience suggests that this process should: involve all stakeholders, including donors, so that they support locally owned statistical development strategies, with an agreed program for data collection and long-term capacity building; use the Poverty Reduction Strategy Paper process, where appropriate, as a framework for articulating needs and prioritizing statistical activities; move from a project approach toward program funding for statistics and, ultimately, to providing finance through budget support; take into account ongoing programs especially in sectors, or for specific themes, already in place; and develop mechanisms for regular consultations between governments and donors and for monitoring progress of the plans.
Official Statistics & Statistical Systems
UNSD Workshop on CountryData – Technologies for Data Exchange Bangkok, Thailand, 18-22 March 2013 Official Statistics & Statistical Systems The case for capacity buildingZoltan Nagy – Statistics Division, Department of Economic and Social affairs, United Nations
OutlineWhat is official statistics?National and international statistical systemsPrinciples of official statisticsChallenges and building capacity in exchanging data
The need for official statistics?All countries need statistical data To identify issues and problems To provide a picture of the current situation To provide the evidence for policy making To monitor progress Numerical data Provide for comparisons Can be seen to be precise and objective Can be manipulated Give access to powerful tools of statistical and mathematical analysis
Official statistics as a public goodOfficial statistics Collected and published by governments Mandate generally set out in legislation Financed from general tax revenue Public good Use by one person does not affect others Costly to produce, but easily disseminated Value depends on quality, but difficult for users to determine this
Statistical systemsAlmost all countries have set up statistical systems Part of central government Organization varies, but even when centralized may involve several agenciesNational systems have a dual role To serve the needs of government To provide information to the public
The statistical processStatistical agencies apply data collection methods to obtain data from data providersData are processed, summarized and disseminated in different statistical products to usersWhat data to collect is a political issueWhat methods to use is a technical question
Data providersData sources Households and individuals Business enterprises, both formal and informal Public enterprises and service providers Civil society organizations and other groupsTrade-off between need for data and level of intrusion and cost to providersImportant to maintain confidentiality
Data collection Formal censuses and surveys Population censuses and household surveys Business enquiries Administrative records and MIS Service delivery Economic transactions with government Requirements of legislation Other methods Remote sensing Participatory methods
Statistical processes and products Statistical processes Data validation and editing Compilation of summary statistics Analysis and estimation Provision of commentary and explanation Statistical products Publications, abstracts, digests, reports Electronic media, including data for further analysis Publication through the Internet Reports on methods
Data users Government Policy makers, planners, analysts and managers Politicians and legislators Markets Domestic and international The public Lobbyists, CSOs, individuals etc. The media International community Development agencies and donors
The international statistical system Implementation Policies, resources, Measuring development programs, projects progress - MDGs Assessment and analysis Supply of data National data International Global data frameworks Data processes Consistent Statistical international Methods and data infrastructure standards UN specialized Financed by: agencies, IMF, WB, Government budgets Review processes , regional agenciesmultilateral trust funds, Transnational and bilateral donors Coordination, ; advocacy, dataSupported by technical ICP, environmental informationassistance and training data, etc.
The theory of official statisticsUsers are not easily able to determine the quality of statisticsThe UN have developed 10 fundamental principles of official statisticsBy following these principles statistical agencies are able to demonstrate their integrity and to build trust and confidence in their products
Key principles of official statistics 1. Relevance, impartiality and equal access 2. Professional standards and ethics 3. Accountability and transparency 4. Prevention of misuse 5. Sources of official statistics 6. Confidentiality 7. Legislation 8. National coordination 9. Use of international standards 10. International cooperation
Challenges and constraints Many statistical systems are under stress and under- performing Lack of demand and limited political support Inadequate and declining real budgets and over-dependence on donor funding Staff lack incentives and skills Products difficult to access and use Limited feedback from users Ineffective coordination and management Many systems are caught in a vicious cycle Inadequate finance ⇒ Poor quality output ⇒ Lower demand ⇒ Low priority for scarce resources
Balancing supply and demandCapacity building requires building demand as much as improving supplyNo formal market for statistics No price mechanism to provide signals for investmentNeed for mechanisms to identify and respond to new demands for dataNeed for active marketing of output
Improving performance Improving communications with users Building trust in products and focusing on data quality Making better use of existing data Using new technology to reduce costs and improve efficiency Improving coordination and management Making statistical systems more transparent, responsive and accountable
Developing a strategyIdentify stakeholdersBroad consultation to build ownershipAssess strengths and weaknessesIdentify investment needsPrioritize actionsDevelop a time-bound planMonitor and evaluate progress