When numbers lie: Towards reliable and robust statistics

There is a saying that numbers do not lie - a statement that is incorrect. When information and data are not reliable, when statistics are tampered with some hidden agenda, numbers do not reflect the truth. This may happen in the case of an individual's tax evasion or official statistics at the macroeconomic level. 

People sometimes wonder why there is so much fuss about the purity of statistics. Is it a simple moral issue, or does the concern have a larger perspective?

The bigger picture

At the national level, data has three specific objectives. First, reliable statistics are essential for policymaking and drawing development strategies of a country. Plans and strategies cannot be formulated totally from a void, independent of realities.  If policies are strategies formulated based on faulty data, desired outcomes will not be achieved. 

Second, data and statistics are also required to evaluate the development results and outcomes. If the baseline data are incorrect, statistics used for evaluations will be biased, and development outcomes will be wrongly assessed. Similarly, if inflated data are used for monitoring, the results obtained will provide wrong signals to the policymakers. 

Third, academia, research organisations, civil society and development partners use information and statistics for public advocacy to raise awareness of the common people about development issues, and also to put pressure on the government. Reliable and credible information, data and statistics are prerequisites for objective development debates and policy discussions in a society. 

But despite all these specific goals, unfortunately, true statistics and correct data are not generated in many countries, particularly in the developing world. There are three reasons for this phenomenon. 

The flaws and weaknesses

One is structural weaknesses of or lapses in national statistical agencies, followed by ignorance, apathy and indifference on the part of the policymakers, on the importance of the data. And thirdly, data is inflated to create an impressive development picture, either because of political pressure from the government or to please the political process.

However, in many instances, statistics are also downgraded in order to get benefits in external trade or foreign grants or loans.

Often, the reliability and credibility of official statistics of Bangladesh have been questioned. At the same time, discrepancies are observed among data published by different relevant organisations. Academics, development researchers, news media, private organisations and development partners have raised concerns on this issue. 

The recently proposed White Paper on Economic Reforms, initiated by the Interim Government of Bangladesh, has also emphasised again the importance of objective, reliable and robust data and statistics for policy formulation, monitoring and evaluation of development results. 

In the context of the process and structure of official statistics, some specific weaknesses have been identified in Bangladesh for quite some time. The first is the politicisation of the statistical process, which has led to the publication of data and statistics devoid of honesty and objectivity. 

Data were inflated because of political pressure or to please the political process. The baseline data were also influenced so that the development outcomes are upwardly biased. This has happened at the programme and the project level also. The projections were also influenced.

Second is, a series of correct data and true statistics were not brought to the public domain. Statistics which had reflected the failures of the government were permanently kept in cold storage. In many cases, this was possible for the simple reason that instead of competent statistical professionals, officials were appointed and promoted based on political favours. As a result, the emerged structure was devoid of transparency and accountability, and the reliability and credibility of the published data and statistics became secondary.  

Third is an alleged lack of coordination among the official data-producing entities. For example, there are discrepancies in the national income data published by the Bangladesh Bureau of Statistics (BBS), the Bangladesh Bank and the Planning Commission. It may be because the methodologies of the three agencies are different, or they depend on different sets of basic data or the baselines pursued are not the same. 

But whatever the reasons, in the absence of transparency and accountability, confusions loom large. On the goals, structures and activities of the national statistical organisation, the future economic reforms of Bangladesh should focus on the following issues.

The ways forward

The official statistical agency must undergo some fundamental reforms. The goal of this entity will have to rise above political considerations, publishing unbiased, reliable and robust data.  

Accordingly, with the appointment of honest and competent professionals, the organisation must be put on a solid base, where transparency and accountability of statistics and data will be ensured. For all these, a 'data availability and transparency' law may be adopted. 

Second, this agency must be independent of political influence and pressure. Otherwise, true and objective data cannot be generated. 

One example from the United States may be of relevance. In the US, the Labour Department publishes the country's US unemployment rate. These data have a significant impact on other indicators such as share market prices. 

Because of such sensitivities, before its official publication, even the President of the United States, let alone the US Labour Secretary, cannot have a peek at the US unemployment numbers. This strict procedure is followed to ensure that such sensitive data remains above political influence. 

Third, the capabilities of the statistical agency must be enhanced. This is not only a resource issue, but it also requires training, inter-agency coordination, particularly among data-producing entities, as well as decentralisation of data-generating activities. 

Coordination is also needed with the UN and the international organisations, whose mandate is international data generation. Regular publication of statistics generated is also needed for transparency and trust of the people in the numbers published. 

The use of modern information and communication technology (ICT) will improve the institutionalisation of data collection and preservation. In that respect, attention should be provided to required financing, training and infrastructure development. 

Fourth, statistics have two aspects – supply and demand. From the demand side, policymakers should be equipped with 'data literacy' along with a solid understanding and appreciation of the importance of statistics in their work. 

From time to time, the statistical agency can organise training for them, undertake surveys among data users to get their feedback and use those to improve the effectiveness of the agency's work. The agency can also identify the best examples of statistical processes of other countries and use the lessons learnt to improve their work.

Fifth, the analytical content of the statistical work is also critical. For example, all data must have a strong analytical foundation. Data are not merely simple numbers, but they also have theories behind them. They require a solid analytical foundation for the purity and honesty of the number, which is needed both to reach correct conclusions and to gain the trust of the people. 

Data, which are not credible to people, do not have utility to them either. Statistics must be disseminated widely among various populations so that people have access to them. Similarly, serious efforts must be made to mobilise disaggregated data – disaggregated in terms of regions, socio-economic groups, gender, environmental fragility etc. 

Inequalities are not properly reflected in average numbers. Only disaggregated data can reveal disparities. In addition, the statistical agency must pursue research, new ideas and knowledge, and exchanges of experiences. A statistical structure cannot be static, it must be dynamic, flexible and agile enough to rise to the new demands. 

Finally, to build a society without inequalities and disparities, there is no alternative to economic reforms. A blue-print of such reforms requires an honest, objective and correct statistical structure for people, which would earn the trust of all sections of the society. Because in the ultimate analysis, information and statistics are the power.