Measurement in Economics, Nowcasting – Beyond GDP, Chair, Insee – QuantCube – PSE – CANDRIAM – Société Générale
The Measurement in Economics, Nowcasting – Beyond GDP Chair is created at PSE (October 27, 2021 - Press release). It aims to contribute to the progress of economic statistics methods, by promoting the mobilization of new sources and the development of very short-term forecasting tools (Nowcasting), and by continuing the reflections initiated from the Stiglitz Commission on a deepening of the statistical measurement of economic performance and well-being (Beyond GDP). On these two components, the objective of the Chair is to better meet the expectations of private and public decision-makers, and more broadly of the social demand on these subjects.
This Chair is the result of an exceptional and original partnership, thanks to the diversity and the complementarity of the entities who are committed to sharing and carrying out their ambitions in these fields of research. It brings together the French National Institute of Statistics and Economic Studies (INSEE), whose work in these fields is authoritative, a start-up offering macroeconomic forecasts based on Big Data and Artificial Intelligence (QuantCube Technology), an international asset management company (CANDRIAM) and a major French bank (Société Générale), that together intend to develop their methods of economic measurement and analysis, and a research and teaching center (Paris School of Economics) that has always been committed to a quantified and statistical approach in economics. The Chair will be able to develop scientific collaborations and mobilize diverse expertise. For example, it will collaborate with the OECD in the framework of its WISE (Well-being, Inclusion, Sustainability and Equal Opportunity) program.
Nowcasting – Relevance, timeliness and speed are three major objectives in the production of economic statistics: to provide indicators that shed the best possible light on the current state and prospects of the economy, and to do so in the shortest possible time, without sacrificing the quality of the information. This dual expectation was further reinforced by the health crisis of 2020. It has increased the demand for speed in the production of the usual indicators. This axis is led by Catherine Doz, PSE and University Paris 1 Professor.
Beyond GDP - Reflections on what the “next world” could or should be reinforce questions about the limits of these indicators, mainly those produced by the national accounts. In the face of these questions, it is necessary to communicate better on what their contributions are, to work on further improving them, and to continue the search for complementary indicators, for which the same question of time will arise. This axis is led by Marc Fleurbaey, PSE Chaired Professor and CNRS Senior Researcher.
Nowcasting & Beyond GDP - The link between the two approaches of the Chair lies in their joint contribution to the effort to increase the relevance of available statistics and data for decision-makers and the public, whether in the temporal dimension (speed, visibility) or in the field of measured objects (economic activity, well-being, sustainable development). In short, it is a question of producing advanced indicators on the one hand, and of making progress in the construction of relevant indicators for society on the other.
Co-Directors of the Chair:
- Catherine Doz – PSE and University Paris 1 Panthéon Sorbonne Professor
- Marc Fleurbaey – PSE Chaired Professor and CNRS Senior Researcher
- The Steering Committee
- Presentation of the Partners
- The Chair Activities
- A Scientific Program in 2 axis
The Steering Committee
Composed of the co-directors of the Chair, one or two representatives of each Partner, and a representative of the PSE Direction, the Steering Committee defines the strategic orientations of the Chair, the means necessary to implement them, and their implementation in terms of recruitment or research themes. It meets at least three times a year.
The mission of Insee is to collect, analyze and disseminate information on the French economy and society throughout its territory. It conducts its work in complete professional independence. With a nationwide presence through its establishments and regional offices, INSEE conducts population censuses in partnership with municipalities and draws up an economic and social overview of France each year through some fifty surveys. In the macroeconomic field, INSEE is responsible for producing annual and quarterly national accounts in accordance with international standards. It prepares and disseminates cyclical information and periodically summarizes it. INSEE conducts studies aimed at shedding light on trends in the French economy and its prospects, and promotes the development of methods and statistics in the fields within its competence.
QuantCube Technology uses artificial intelligence and big data analytics to deliver real-time macro-economic insights. The firm operates one of the largest alternative data lakes in the world, processing more than 14 billion data end points. Sources encompass news, social media, satellite data, professional networks and consumer reviews, as well as international trade, shipping, real-estate, hospitality and telecoms data. QuantCube’s macro nowcast indices, on variables including economic growth, inflation, employment and international trade, correlate highly with official data and significantly beat the consensus. Financial institutions using QuantCube data benefit from real-time insight, often ahead of official numbers, which they can use to inform their investment strategies. Headquartered in Paris, QuantCube employs a diverse international team of data scientists with expertise in multilingual NLP, deep learning and machine learning techniques. The company’s shareholders include Moody’s and Caisse des Dépôts and its R&D in computer vision has been partially funded by the European Space Agency (ESA) and French government space agency CNES.
CANDRIAM, which stands for “Conviction AND Responsibility In Asset Management”, is a European multi-specialist asset manager. A pioneer and leader in the field of sustainable investment since 1996, CANDRIAM manages approximately €150 billion of assets and relies on a team of more than 600 professionals. The company has management centers in Luxembourg, Brussels, Paris, and London, and its client managers cover more than 20 countries throughout continental Europe, the United Kingdom, the United States, and the Middle East. CANDRIAM offers innovative and diversified investment solutions in several key areas: bonds, equities, absolute return strategies and asset allocation, a broad and innovative range covering all its asset classes. CANDRIAM is a New York Life company. New York Life Investments ranks among the world’s leading asset managers.
Société Générale. Active in the real economy for over 150 years, with a solid position in Europe and connected to the rest of the world, Societe Generale has over 133,000 members of staff in 61 countries and supports on a daily basis 30 million individual clients, businesses and institutional investors around the world by offering a wide range of advisory services and tailored financial solutions. The Group is built on three complementary core businesses: French Retail Banking which encompasses the Societe Generale, Credit du Nord and Boursorama brands - each offers a full range of financial services with omnichannel products at the cutting edge of digital innovation; International Retail Banking, Insurance and Financial Services to Corporates, with networks in Africa, Russia, Central and Eastern Europe and specialised businesses that are leaders in their markets; Global Banking and Investor Solutions, which offers recognised expertise, key international locations and integrated solutions. Societe Generale is included in the principal socially responsible investment indices: DJSI (World and Europe), FTSE4Good (Global and Europe), Bloomberg Gender-Equality Index, Refinitiv Diversity and Inclusion Index, Euronext Vigeo (Europe and Eurozone), STOXX Global ESG Leaders indexes, and the MSCI Low Carbon Leaders Index (World and Europe).
- Organization of scientific or general public events Research seminars, international conferences and thematic workshops will punctuate the scientific life of the Chair, bringing together an academic audience and professionals with expertise in these issues. Each year, a major conference bringing together researchers and partners, but also a wider public, will report on the latest work (including a round-table format conference).
- Invitations of researchers The Chair brings together researchers from all over the world, for short or long periods, in order to enrich the work carried out and initiate new collaborations. These guests participate in the events and/or teachings of the Chair.
- Funding of doctoral and postdoctoral grants On topics specific to the Chair and defined by the partners, doctoral and post-doctoral fellowships are regularly offered through an international call for applications. After review by an appropriate jury, the selected candidates will join the doctoral program or the PSE scientific community.
- Implementation of Executive Education sessions Based on the most recent developments in research, the Chair can contribute to the strategic thinking of professionals and help them respond to the challenges they face through short, tailor-made training sessions, either face-to-face or through distance learning. For example, through a “mini-course” on dynamic factor models or on the equivalent income approach.
- Exchanges with partners The Chair is a place for reflection and exchange between the partners. A transversal seminar is organized, bringing together all the researchers involved and the employees of the partners interested in the work in progress. Regular collaborations can be set up between the different people involved, on specific subjects or longer term projects.
A Scientific Program in 2 axis
>> Improve measurement and forecasting in the very short term (nowcasting)
Nowcasting is the term used to describe very short-term forecasting (also known as “nowcasting”), i.e., the forecast that can be made of a quantity (e.g., GDP) before its first official publication. However, the techniques mobilized can also be used for short-term forecasting, for example the one-year forecast. These very short-term forecasts are useful for public decision-makers in charge of national or international economic policy. They are also useful for private investors whose choices may be partly guided by this type of forecast. Indeed, they provide an estimate of the current values of the macroeconomic variables of interest at a time when they are not yet known, due to the publication deadlines to which these variables are subject.
Acknowledged statistical techniques
The models used for nowcasting are notably those called “Dynamic Factor Models”, and have been the subject of an abundant literature over the last twenty years. Many public actors use this type of model: for example, the New York Fed, the St Louis and Atlanta Fed, the ECB, among many others. The principle of these models is to summarize the dynamic behavior of a large number of variables by assuming that they are linked to a small number of unobservable latent variables called factors.
Statistical techniques to be developed
First, variants of these models have been developed in recent years and may deserve to be evaluated for their contribution to forecasting with French data. Examples include the use of level variables rather than growth rate variables to forecast the trend and not just the cycle, methods of prior selection of relevant variables to construct the forecast, and regime change models in which the parameters describing the dynamics of the factors are different depending on whether we are in an expansion or recession phase, and which allow us to estimate the probabilities of entering (or exiting) a recession. This type of model, which makes it possible to quickly identify cyclical reversals, is not necessarily adapted to extreme periods such as the one we are currently experiencing, but it proves to be a fairly effective tool in periods where cyclical fluctuations have a more normal amplitude.
Second, the flexibility of these models to handle data of different periodicity is likely to be exploited more intensively than it has been to date, as the uncertainties of the current period create the need for more frequent forecasts based on more frequent data. The current period has spawned a number of new approaches to data and forecast update frequency that will continue beyond the health crisis. The goal is to identify more frequent and readily available data that can provide quality forecasts. It would be particularly interesting to study whether factor models, used with high-frequency data, can be used to construct reliable activity indices, both at the French level and at the level of the Euro zone.
Finally, we can envisage that the use of this type of data of higher frequency (approaching “Big Data”) will lead to the exploration of different techniques, stemming from machine learning. In a second phase of the Chair’s research agenda, these new techniques will be studied in order to identify their potential but also their limits, and thus progress on the path of an ever faster and more efficient forecast. Identifying the most useful type of very high frequency data, progressing in the techniques of production, collection and rigorous use of these data are promising avenues of research that the Chair will put on its agenda.
>> Measuring « around and beyond GDP »
Over the past decade, questions about the “mismeasurement” of an increasingly polymorphic and dematerialized growth have been added to the more traditional criticisms of the “beyond GDP” literature. The crisis has offered new variants of these different questions, concerning, for example, the question of the relative value of public services or other so-called “essential” activities, or that of the measurement of prices and living standards in a context of brutal and forced deformation of consumption structures.
All these questions require that the conceptual framework of national accounting be confronted with the broader issue of welfare measurement. Many of the concepts proposed by national accounting should be articulated with the objective of measuring well-being.
Work based on national accounts could also make a more meaningful contribution to the assessment of development sustainability. While it is not necessarily a question of integrating such indicators into the central accounting framework, particularly given that this framework is set by international standards and that these indicators are based on assumptions that are different from retrospective statistics, it is still necessary to find extensions to the central framework that integrate this prospective dimension: it is necessary to prevent official statistics from appearing to be behind a concern that has become major.
In the area of corporate social responsibility (CSR) reporting, many initiatives have emerged among large companies as well as investment funds, rating agencies and associated consulting firms. While there is a consensus on the need to cover environmental, social and governance (ESG) dimensions, there is a lack of clear understanding of what corporate social responsibility means in a division of labor between governments, households, civil society and business organizations. The increasingly popular idea of serving the interests of all stakeholders actually clashes with conventional thinking in economic and management theory about the need to rely on objective measurement of goals and achievements, and the need for unified governance of the organization under a homogeneous group of ultimate owners. And without a good tool to measure its broader impact on societal well-being, companies take initiatives and communicate about them in multiple ways that lack macro-level coordination.
Proven statistics and solid theoretical approaches
National accounting provides a coherent framework for describing economic activity and the flow of resources between sectors of activity and between different types of agents. It is based on a number of conventions that aim to make the published figures more relevant for policy makers and public debate. Some of these conventions can be revised to take account of changes in the economy and society and to better define the link between certain quantities and measures of well-being.
Three approaches to measuring well-being can be identified, based on theoretical foundations derived from the academic literature (as opposed to synthetic indicators, which are more pragmatic in nature). The most objective approach, proposed by Amartya Sen, proposes to rely on living conditions and, more precisely, on the opportunities (“capabilities”) that the population enjoys in this area. The more subjective approach, driven by empirical work on subjective well-being, proposes to measure directly the answers to satisfaction or happiness questionnaires. Between the two lies the equivalent income approach, which starts from living conditions but reduces them to a monetary standard by relying on the preferences of the population (through the use of consents to pay). This last approach is the closest to accounting, since it seeks to correct income for the costs and benefits, felt by the population, of non-income aspects of living conditions. All three approaches can be used, with varying degrees of robustness, to measure inequality and to estimate the specific impact of various aspects of living conditions.
Statistical techniques to be developed
Several avenues of research deserve to be pursued, particularly for the last two approaches.
Numerous issues around the measurement of GDP arise for national accountants: the volume-price split, especially in times of large fluctuations in the structure of production and consumption; the inclusion of quality and new products in a context of accelerating obsolescence; the emergence of quasi-free services in the digital economy; the contour of household production; the valuation of government production; the location of production and income-generating activities in the globalized economy; the place of housing in purchasing power indicators; and the measurement of human capital investment. By examining these issues, a dialogue between researchers and national accountants will be able to identify ways to modify certain accounting conventions at the margin and to develop complementary measures that are closer to theoretical concepts and measures of well-being.
In particular, the equivalent income approach can be used as an inspiration in considering these different issues. This approach offers measurement principles that overcome the limitations imposed by market prices to weight different aspects of economic activity, including non-market activity. Although in theory this approach requires an estimation of people’s preferences, which is generally costly and difficult, it is possible to develop more easily implemented proxies.
Beyond these accounting issues, the equivalent income approach can also be used to estimate the impact on the population of certain externalities (environmental quality, social relations) induced by production, investment and consumption behaviors, and thus contribute to calibrating not only incentive and regulatory responses, but also ESG initiatives by the private and financial sectors that are concerned with internalizing the positive and negative externalities created by the activity of companies. It can also, in this way, offer a decomposition of the contributions to well-being of the various market and non-market factors that influence well-being.
More thought needs to be given to the measurement of sustainable development. The theory indicates how measures of net savings can be relevant, but the conditions for applying the theory are far from being met, and reflection on the development or use of long-term scenarios for estimating the value of current stocks would be worthwhile, in conjunction with the efforts of climate researchers. Furthermore, the cost-effectiveness approach favored in France to estimate the values of environmental protection actions (such as the “shadow price” of carbon) is opposed to the cost-benefit approach used elsewhere to estimate these same values, and this methodological debate deserves to be pushed to its conclusion. Taking biodiversity into account also raises major difficulties, and its purely “instrumental” integration into the anthropocentric theory of sustainable development is questioned. Recent work has demonstrated the possibility of making exploratory estimates of a global well-being integrating multiple species and type of living organisms.
>> Towards faster measurement of indicators around and beyond GDP?
Eventually, and if the data are suitable, it is conceivable that synergies could be created between the two parts of the Chair. This perspective of work, which would be nourished by the two parts of the Chair, is today at the stage of reflection, and its feasibility depends in particular on the availability of suitable data. However, it would constitute an important research and application opportunity, which could be put on the Chair’s agenda and lead to collaboration between the two axes. For example, the fact that social and welfare data often attract less media and political attention is often lamented by analysts, and attributed to their too episodic availability, as well as the long delays that characterize their publication. Increasing the frequency and speed of publication of indicators of well-being, living conditions, inequality and poverty can come not only from a better mastery of the concepts and measurement methods that can be used, but also from the mobilization of original data and adapted statistical processing.