The SORS Decision-Making Support System: A statistical tool for better policy making in the Republic of Serbia

2.1The rise of a post-truth society and crisis of truth

The overwhelming amount of information from a variety of official and non-official providers in combination with the urgent need for information for decision making creates a situation where the ‘quality’ of the information is under pressure. False news, misinformation and disinformation are circulating from social media to political party platforms and government. It seems that all data, news and statistics are distributed with a certain purpose and aim to generate required reaction or to benefit the owners of data. Rosenfeld [1] points out a recent English expression “that’s my truth”, which implies that people have lost out common ground about what is truth and what constitutes it, which can entail the existential crisis for democracy itself. In the circumstances of prevailing personal opinion over facts and various interpretations of data and facts, it seems that citizens have become alienated from political and civil institutions, and trust in institutions seems to have been permanently lost. As a typical example we can take the financial crisis from 2008 – amongst a data deluge, nobody saw and predicted the most severe crisis in the last fifty years and this lack of seeing and predicting undermined credibility in institutions and decision-makers [2, 3, 4].

The Oxford Dictionary Word of the Year 2016 was post-truth. It is explained as “an adjective defined as relating to or denoting circumstances in which objective facts are less influential in shaping public opinion than appeals to emotion and personal belief” [5]. Post – truth was chosen as a word that reflects “ethos, mood or preoccupations of that particular year and to have lasting potential as a word of cultural significance political and social discourse” [6], as well as the word which dominated by highly-charged political and social discourse in 2016. Therefore, while every epoch has its own specific characteristics and challenges, our epoch is faced with lots of information from diverse media mixed with personal beliefs that altogether make life more stressful and complicated instead of making it easier. To summarize the characteristics of our epoch, we can say that we can encounter many questions and assumptions everywhere, but no answers.

Indeed, in today’s socio – political environment, loaded with the multitude of data and sources of all kinds of information, dominant are the phenomena such as conspiracy theories, ill-information, ‘fake news’ stories, ‘clickbait’ etc. [7]. This is a time of political rhetoric, liberal social – economic paradigm, “freedom of the media”. Modern society is faced with “hearing everyone’s opinion”, from unprofessional journalists to politicians who lust for power. They are allowed to judge the accuracy and correctness of official statistics, regardless of their incompetence and lack of professional knowledge. These tendencies clearly undermine an NSO’s credibility in public just because someone has the right to say whatever he or she thinks.

Consequently, modern societies are experiencing an atmosphere of confusion and mistrust, and a mistrust on information from governments, institutions, politicians, and media. People become increasingly suspicious, skeptical and full of anger and resentment over the information they are given, constantly looking for a culprit to be blamed. “Citizen’s growing suspicion vis-à-vis the official statistics …is in line with our ‘post-truth’ and anti-intellectualist era” [8]. According to some authors, rather than diffusing controversy and polarization in society, statistics are stoking them. Antipathy to (state) statistics has become one of the hallmarks of the populist right [9].

For societies this creates the need to choose an approach for supporting decision-making that allows to create clarity in this overload of data and opinions. This choice ranges between two extremes:

2.2The precious opportunity of official statistics

It is within this context that official statistics have a precious opportunity to stand out as an objective and independent institution that produces information based on established methodologies, using tested, approved and supervised methods and calculations. This scientific approach is based on facts that can be tested and proven. Generating and using evidence to inform government is not a partisan issue – it requires complex strategy [10]. Professional communities are prominently insisting on evidence-based decision-making as the only functional and ethical response to the challenges that the modern society is faced with. And official statistics, during its long history, gained skills and potential to provide users with high-quality and reliable information.

Political power, transmitted through government ministries, along with the knowledge and expertise concentrated in professional institutions such as statistical offices, should work together, avoiding collisions. The ownership of data and information is not an exclusive right of statistical institutes; however, statistical institutes are not expected to be passive producers, deprived of the right to express its own opinion regarding the observed phenomena and tendencies in the current environment. Also, data and information are not the property of decision-makers and politicians; on the contrary – only joint work and cooperation between these two sides can be fruitful and enable the synergic process of maximizing the use of information for the purpose of quality decision-making and governance.

The huge amounts of data produced and disseminated every second take a risk of misuse, misunderstanding and manipulation. Whether true or false, they reach the public and transform themselves into a powerful tool that can be used (and misused) by politicians, interest groups and individuals, as the weapon of soft power for achieving certain goals. However, data around us are not self – explanatory; they need to be digested, processed and transformed into meaningful and useful information. For information to be used for official national or regional policy making this is supposed to be a role of official statisticians, having in mind their capacity and historical background to be regarded as the most engaged part in the process of keeping pace with the changing world in the domain of producing data and information.

However, in most countries the role of official statistics is still mostly limited to the traditional tasks: collecting, processing and dissemination of data, avoiding interpretations and explanations, let alone sharing opinion about trends in various phenomena. “Keeping objectivity” and absence of aspiration to influence public opinion is most often mentioned by statisticians as the reasons and explanations of this rigidity. However, does “keeping objectivity” imply intellectual passivity [11]?11 Does the desire to be a professional, in the traditional sense, imply staying apart and being neutral?

2.3A responsibility in processing the data but also in communicating the information

Policy-makers’ capacity to understand data, to make distinction between unprocessed “raw“ data and extracted information and to observe the interconnectivity among them, is often limited and their ability to use data might lag behind the pace of the changing world. In many cases, due to the nature of their work, it is not easy to recognize the real value and preciousness of the available data due to the lack of time, since the process of policy-making requires relatively quick reactions and the space for reaction is isolated and narrowed [12]. The challenge remains as how to define the policies in accordance with the respective information and how to bridge the gap between data and policymaking.

Moreover, among politicians a strong pressure is felt to keep the status quo in institutional life, with suspicion to innovation in the sense of reorganizations, changing traditional roles and overleaping responsibilities between institutions [13]. The rigidity of state administration is widely present and also well-known through all stages of state development. Furthermore, there is often (at least a subtle) attempt to impose political influence on the decision-making process, which can lead to undesirable solutions, rent-seeking behavior and opt for expensive and unprofitable projects. However, “people want government that solve problems” [10]. Given that, any democratic society should have the goal to perform the decision-making process on the basis of recorded data to the full extent (i.e. evidence-based decision-making). In that process, official statistics should be a necessary information-generating instrument which can significantly contribute to the relevance, efficiency and effectiveness of the decision-making process. To strengthen and regain trust in institutions and official statistics, NSOs more than ever need to adopt a pro-active role and to broaden their focus to the issues concerning everyday life. In our opinion in the current times this pro-active role has four elements.

The important first element is to make policy-makers aware of the value of information around them and to open debate with them about the proliferation of information in the modern world and the lack of quality and relevant data at the same time. Promoting data literacy, insisting on expertise and professionalism, favoring and emphasizing the advances and qualities of official statistics are other necessary steps in the process of shifting the role of the NSO and promoting it as an institution capable of providing decision-makers with high quality information.

The second element of this pro-active role is that official statistical institutes must also assure themselves that user communities recognize their relevance for understanding world around them or for informing decision made by themselves or others [14]. It means that in most cases they must work more on statistical promotion and dissemination, but also on updating statistical laws, modernizing statistical standards, introducing new technologies and data sources into their activities, and most importantly, improving data and statistical literacy in its users. In a time of lack of trust in government institutions coupled with the opportunity for data to impact and transform every part of governance, more flexible and open-minded models to correctly respond to the challenges of our era is an initiative that should be considered with the greatest attention. “Is statistical evidence used to stimulate political dialogue (opening up), to shorten it, or, in the worst case, to suppress it (closing down)? Depending on how these questions are answered, statistics will win or lose citizens’ trust [15]”.

The third element of the proactive role is related the challenges on authority and the leadership role of governments and statistical offices in the context of the new data ecosystem that provides many opportunities and enables more effective use of official statistics [16]. In such environment NSOs can assist public administrations as ‘data stewards’ and contribute with their expertise in data management, data governance and the generation of information derived from a variety of data. This is an opportunity to strengthen the role of NSOs and official statistics in the public domain [17]. The impact of data could be improved with combining separate datasets from different sources.

A fourth element of the proactive role goes even one step further outside the traditional role of official statistics organizations. A role that goes beyond being a steward managing data to facilitate coordination and collaboration between institutions that provide different data and ensure responsible, accountable data sharing in the public interest. This is the position of active contributor to trusted debate and decision-making.

3.1The Serbian statistical system

Serbia’s political decision-making structure is centralized around the Prime Minister with a variety of bodies connected to the Prime Ministers office. However, the communication between the connected bodies and institutions is limited. In such a structure, policy functions suffer from the lack of expert knowledge and without synergistic governance. The absence of coherence could be bridged by applying an integrated approach to cooperation between politics, politicians, decision-makers and experts since all of them represent different pieces of a policy-making process. The right balance of these “components” implies the division of roles among all parts and at the same time awareness that expertise cannot be separated from politics and policy making, and that professionals give specific value added to decision-making processes. In this way only the policy process could be made coherent, comprehensive and in touch with the needs of real world and real people. Or, in other words, we need to bring the policy process closer to the real world and bring the real world closer to the policy process [13].

For this reason the Government of the Republic of Serbia established in 2017 the Council for the coordination of activities and measures for the growth of GDP. The main role of the Council is to develop analytical and monitoring mechanism of GDP growth and to analyze the changes in trends of individual growth factors, with the aim to ensure timely consideration of possible growth support measures. The Council’s members are representatives of the government ministries, the Serbian Chamber of Commerce and SORS, which has a significant role not only as data producer, but also as an active member of the analytical team. At the very beginning of a month, the members of the Council gather to discuss and evaluate current economic situation based on the SORS analytical models and the latest information about each area of economy, completed with the ministries’ reports, the Chamber of Commerce information and the reports from public enterprises (such as Electric power industry of the Republic of Serbia that is responsible for the supply of electricity).

Specially designed questionnaires made by the Council staff are sent to government ministries to be completed with the corresponding planned projections. They are asked to identify problematic or opportune activities and areas or projects where those activities were detected. Afterwards, Council staff propose the relevant short-term or long-term measures or policies that could be implemented to spur the growth of production and gross value added.

3.2The DMSS as a specific tool within SORS

SORS, as an official producer of statistics, publishes a large amount of releases, indicators, bulletins etc., Users in the modern society utilize the statistics to evaluate their performances and the extent to which they are aligned with other business subjects and developments.

Considering that numerous users are statistically and economically illiterate and without adequate capacity to understand statistical methodologies and data, the diversity of data can lead to confusion, wrong conclusions, erroneous assessment of priorities, and even to reluctance towards information. Also, the statistical system is very complex, generalized and designed to meet the specific needs for information (e.g. agriculture, prices, manufacturing, etc.). The information provided is often not sufficient to decision-makers, because the macroeconomic environment is often only partially described.

As mentioned by Radermacher [15], the handy and popular formulation “Data for Policy” (D4P) covers numerous topics such as modernization of the public sector, evidence-informed policy making, new solutions, new technologies and indefinitely rich data sources. However, this raw material of data is not directly usable in politics – the statistical system needs to distill, refine and process valuable statistical knowledge from the flood of raw data into digestible information for politics. Hence, the purpose of the statistical system is also to find relevant, but often hidden or unnoticeable relations between different indicators, to extract key information from a large number of data and to define key indicators with the aim of augmenting the efficiency and reliability of the decision-making process.

In order to provide the users with sufficient, useful and reliable information to identify the situation in the economy including potential imbalances, risks to stability and their possible prevention, SORS has created a department dedicated to catalyzing pieces of information and transforming them into simple, reliable and widely usable indicators and accommodate SORS to fulfill the four proactive roles as sketched above.

To support the satisfactory fulfilling of the roles, SORS has also developed a Decision-making Support System (DMSS).22 The DMSS is designed as a set of tools essential for better understanding the economic position of a country, and therefore to facilitate a high-quality decision-making process. With this system SORS expands its main role from passive to active player in decision-making processes, from an institution responsible only for publishing sets of various data to a modern pro-active institution which provides explanations, offers alternatives, searches for interconnections in the economic environment and places them in order to facilitate complex process of decision-making.

3.3Main characteristics of the DMSS55, 66

The DMSS is designed as a set of tools, some already a regular part of the official statistical system and some subsequently introduced and designed to better illustrate and explain a particular phenomenon. These tools aim to statistically clarify the interaction between key economic indicators, to explain the economic position of the country and facilitate high-quality decision-making processes. Examples of these tools are a monthly projection updating system, a system of leading indicators, a system for quarterly GDP nowcasting, forecasts and a system of Economic Sentiment Indicators (ESI).

In common synergy, these tools can provide a multidimensional explanation of a phenomenon.

Many of the tools already a regular part of the official statistical system were, before DMSS, strictly related to only one statistical area (e.g. ESI was always published separately instead of being analyzed in interaction with other statistical indicators) and were never analyzed together with industrial production, tourism activity, etc. nor in the context of a broader picture composed of other economic indicators.

The DMSS has a three parts:

In Sections 4, 5 and 6 these three parts will be illustrated with specify examples of the DMSS tools.

In running the DMSS, there is cooperation between several actors. The role of SORS is to collect, process and analyze data with an emphasis on finding and highlighting traductive links between phenomena; the Chamber of Commerce of the Republic of Serbia (CCS) collects information and data of special interest via branch associations; the Department of Strategic Planning, Development and Analysis deals primarily with an analysis dedicated to the detection of issues and restricting them at micro level and levels of a homogenous group of producers and exporters. CCS has developed a Decision-Making Support System at lower levels of business (level of enterprises and business branches), whether by territory or classification category; experts from the other competent ministries evaluate projections based on the information they have; representatives of public companies present the problems and challenges they face, such as orders, sales, delivery etc. All actors together discuss short-term trends in the Serbian economy and consider possible measures to support economic growth.

The main output of the DMSS are specific statistics. Four categories can be distinguished:

4.1Monthly projection updating system

Projections indicate the level of an aggregate in future provided that a set of defined assumptions is realized. In the scope of the DMSS, the projection system has two aspects:

The discrepancies between the planning and statistical model based projections occur because the government ministries and other subjects observe the effects of certain measures (i.e. their quantitative impact) on the GVA, or due to the difference in the expressed factography (e.g. for monitoring construction works, SORS and the respective ministry use different methodologies). Where the discrepancies are rather expressive, the causes thereof should be thoroughly examined and discussed with the aim to provide as much as possible accurate estimations for the coming year.

4.2System of composite leading indicators

Leading indicators are used as analytical tool in predicting cyclical trends of economic activity. It includes detection of turning points, minimum and maximum that could result in anticipation of an economic cycle of the domestic economy in the future.

In the process of detection of the variables to be included in composite indicators, all macroeconomic areas and surveys on the expectations from the Serbian economic subjects were considered. The developed system of composite leading indicators leads ahead of the cycles of economic activity by six months on average and, combined with econometric models, enables a quantitative evaluation of the dynamics of the annual growth rate of economic activity on short term, quarterly and annual basis.

SORS produces the leading indicators for agriculture, industry, construction, trade, and services. The database formed for the calculation of leading indicators includes more than 3000 variables (monthly updated); however only variables with the best leading performance are included in the final composition of the composite indicators. Together with projections, forecasts and nowcasting, the leading indicators more clearly explain the movement of the most important macroeconomic indicators and GDP.

4.3Nowcasting quarterly GDP

Nowcasting is the evaluation of a macroeconomic variable obtained during a reference period and it is based on partial and indirect information about the targeted variable (in this case – a quarter). Nowcasting is fact – based, focused on the known and knowable, and therefore avoids forecasting.

The idea of nowcasting is to use the signals of the movement of the targeted variable based on a heterogeneous set of higher frequency information. According to Eurostat’s Handbook on Rapid Estimates [22] Statistical methods used in the nowcasting context do not necessarily replicate those used in the regular compilation process, they mainly aim to produce the most reliable results for the reference variable, that is the best estimate of the evolution of the reference variable.

Therefore, available data for the second month of the referent quarter is used for nowcasting of the quarterly GDP. Then, one month later, having data for the whole quarter, the flash estimation of GDP is being calculated. And finally, next month follows the regular estimation of quarterly GDP. For the nowcasting SORS uses MIDAS methodology [23] (Mixed Data Sampling as methodological frame) which explains low-frequency variables (GDP) as the function of high-frequency variables (monthly indicators on industry, prices, trade etc.).

4.4Indicator of the monthly GDP

This indicator reflects the monthly trend of the total economic activity (i.e. GDP) of the national economy. It is created and developed in SORS as synthetic macroeconomic indicator that should serve as the base for medium-term forecasts and estimate of monthly GDP. MK30 is obtained as the weighted average of corresponding monthly indicators: industrial production, construction, wholesale and retail trade turnover, deposits and loans, telecommunication services, number of employees and agricultural production.

The indicator is harmonized with the quarterly GDP trend using a benchmarking procedure which preserves the monthly dynamics of the economic activity with the constraints imposed by the data on quarterly GDP. Based on the movement of this indicator, it is easier for decision makers and other analysts to predict the GDP growth in in the coming quarter.

4.5Forecasts of major macroeconomic domains

SORS uses a forecast method based on time series ARIMA modeling processes developed by Box and Jenkins (1970) [24] to produce forecasts for major macroeconomic domains. The main stages in the modeling are: model identification, parameter evaluation, checking model adequacy and the use of the model for forecasts. Each of these stages has its own specific problems, and the whole structure can also be modified. On monthly basis, SORS makes forecasts for the following macroeconomic domains: Industrial production – total, Manufacturing, Consumer prices, Retail trade turnover, Exports and imports.

4.6Description of developments for major GDP components

SORS prepares short descriptions about the current evoluation of selected areas and sectors (industry, experts, prices, construction, etc.) which include the most important details related to a particular sector. These sectors are highly correlated with the development of the economic cycle, being therefore often used as predictors of the movements of the whole economy. E.g. industrial production, with a share of about 20% in GDP, has an important contribution in the Serbian economy andanalysis of trends includes the evaluation of the dynamics of the trends in the sections B – Manufacturing, C – Gas, steam and air conditioning supply, and D – Mining and quarrying, according to the Classification of Activities, 2010/NACE Rev 2, as well as in their major divisions with the highest shares in the formation of the Industrial Production Index. The brief information also considers enterprises that have low performances which contributeto falls in certain divisions. The descriptions are aimed at estimating the current evolutions of components such as the trend-cycle and seasonally adjusted indices. Based on the behavior of these components, the conclusion may be drawn about the (short-term) future trend, allowing thus the ex-post and ante-post validation of coincidental (current) trends, early detection of imbalances and key macroeconomic signals, as well as getting a global view of the whole sector, from all available aspects.

4.7Flash estimates – Economic Sentiment Indicator

The purpose of the composite Economic Sentiment Indicator (ESI) is to present producers’ and consumers’ perceptions about economic movements and economic stability. This indicator is used to assess the economic situation, make flash estimates, for scientific and analytical purpose, as well as for international comparisons. Therefore, economic agents’ expectations are indicative of the changes in economic movements, which contribute considerably to creating economic policies.

ESI has been developed by the General Directorate for Economic and Financial Affairs of the European Commission [25]. It is obtained through five different surveys of producers and consumers whose perceptions or expectations provide a reliable indication of economic movements, based on which confidence indicators are created. Confidence indicators of the analyzed sections are weighted in order to reflect their influence on economic activity: manufacturing -40%, service activities -30%, household consumption -20%, construction -5% and retail trade -5%. A value of ESI index exceeding 100 indicates improvement or economic activity, while that below 100 suggests decline.77

4.8Information on macroeconomic trends

The information on macroeconomic trends in the Republic of Serbia is a monthly dashboard/scoreboard-style publication that includes a selected set of indicators. Each indicator is accompanied by its visual presentation (graphs, maps, etc.) with brief explanations and estimation of the development presented as meteo signs. The information on macroeconomic developments is aimed at providing both quantitative and qualitative developments of the selected set of key indicators (employment, wages, public finances etc.).

SORS quarterly publication Trends uses advanced graphic solutions for presenting and disseminating a large set of statistics for displaying the most important economic signals.

4.9Visualization tools

For the purpose of achieving a quick insight into the position of the Serbian economy, without having to go into a deeper analysis or requiring expert knowledge, SORS publishes a monthly overview of the GDP cyclical component (and a couple of other indicators such as the age pyramid of the population) as an interactive graph on the SORS website, as well as many infographics (external trade, industrial production, construction, employment, wages).88