4 Expert testimony : Sir Charles Bean Professor of Economics at the London School of Economics

Transcription

1 Annual Report 2016

2 Contents 2 Editorial 4 Expert testimony : Sir Charles Bean Professor of Economics at the London School of Economics 8 Highlights of Producing short-term diagnoses 14 Studying the functioning of the French economy and the determinants of its major trends 16 Providing details about the productive system 18 Observation of the society 20 Conducting the population census and using the results 22 Describing and analysing regions and territories 24 Managing registers 26 Disseminating its statistics and studies 28 Improving its methods and innovating 30 Contributing actively, with the French Official Statistical System, to European and International statistics 32 The profession of systems integrator 33 INSEE in facts and figures Key figures Head Office organisation chart INSEE: a nationwide presence Workforce by region and by category The men and women at INSEE Training and budget Surveys carried out in 2016

3 Editorial for themselves while always providing explanatory comments with the statistics they deliver or producing ad-hoc studies. The aim is also to facilitate access to information for all audiences, from experts to the general public, notably by seeking to develop a range of educational products. Second, Innovating and being at the forefront with data sources refers to a dimension that is often underestimated. While the need for comparability of statistics over time results in a degree of stability in the methods used, it nevertheless does not preclude innovation, quite the contrary! Knowing how to make good use of technological progress and Big Data in particular, exploiting administrative data and, in the very near future, the data held by private companies too, reducing the response burden placed on individuals and companies, these are all challenges, and sources of constant innovation was the year of INSEE s 70th anniversary. The Institute s history took pride of place, highlighting its close links with the construction of French official statistics, and more widely with the economic and social history of France since the post-war period. The Institute s future was also set out in order to continue to shed light on the economic and social debate as it has done for the last 70 years was also the year when Horizon 2025, an ambitious strategy for INSEE was finalised. The result of strategic thinking in the well-established tradition of INSEE, this programme defined four major strategic directions. First, Making statistics meaningful and reaching out to all types of public focuses on one of INSEE s defining characteristics among National Statistical Institutes. INSEE s statisticians and economists have indeed developed an economic and social expertise enabling them to make the figures speak Third, Contributing actively, with the French Official Statistical System, to European statistics highlights the importance of the European context. Because economic and social facts and public policy performance can also be analysed in the light of comparisons with our neighbours, INSEE is careful to carry out its analyses in an international context and produce, with its partners, data of comparable quality from one country to another. This especially involves sharing good practice and promoting our methods. Finally, with the fourth strategic direction, Demonstrating collective agility and sobriety, INSEE is committed to changing its organisational framework and its methods. This mainly involves responding to the challenge of preserving its know-how when so many staff are about to retire, and being able to continue to meet the needs of users when resources are stretched. These four directions reaffirm the Institute s missions and values while also setting out our future plans. INSEE and all its staff are now focusing all their actions on Horizon 2025, as shown in this annual report for INSEE 2016 Annual Report

4 In response to the needs of political and economic decision-makers, INSEE was able to bring forward by two weeks the availability of three essential indicators for economic analysts: GDP, the Consumer Price Index and the unemployment rate. and the DREES 1 have made the source code for their INES model available, to make the figures from social and fiscal policies more meaningful and in this way they promote the flow of data and knowledge. After an initial experimental period in 2015, publication of advance indicators on poverty and inequalities was continued, to meet the high level of social demand on this subject. In the same area, the first indicators measuring disposable income and poverty among residents in urban policy priority districts were disseminated with the publication of a series of national and regional studies; in this instance too, INSEE sought to mobilise innovative projects to meet the newly defined needs for detailed analysis of territories in the light of urban policy reform. Concerning innovation and the use of data sources, especially administrative data, INSEE is making preparations to start using Nominative Social Declarations (DSN). The aim is to improve the information system on employment and earned income, and more broadly to gain greater knowledge of the labour market and study employment transfers. There has been an increasing number of projects and experiments surrounding Big Data, at both national and European levels. The Law for a Digital Republic, definitively adopted in October, defines data held by companies as the third database of the official statistical system, alongside surveys and administrative records. This major development enables INSEE to strengthen its project to produce part of the price index using scanner data from networks of large, predominantly food retail stores. INSEE AND ALL ITS STAFF ARE NOW FOCUSING ALL THEIR ACTIONS ON HORIZON 2025 More legislative news: the law revising the ways people register to vote has given INSEE responsibility for administering the new single electoral roll. The Institute is working on this with the many stakeholders involved to ensure that the requirements for the future electoral code are met. At European level and internationally, in close collaboration with the Ministerial Statistical Departments, INSEE continues to work on improving the quality of official statistics by becoming involved in new projects. Among those launched in 2016: twinning with Tunisian Statistics, an experimental European project to exploit the potential of Big Data, etc. Finally, so that all users of official statistics are able to access all publications more easily, and to make official statistics available to a greater number, INSEE has a new website which will be further improved throughout The provision of statistics and studies to inform public debate is becoming increasingly important, especially in a year like 2017 with so many strategic electoral events for our country. All this would not be possible without the commitment of the INSEE staff, for which I thank them all. Other provisions of this law are already in effect, such as the opening up of all data in the SIRENE Register since January INSEE prepared for this in 2016 and was therefore able to provide widespread access to this reference database of French enterprises and establishments. This opening up even extends to the models themselves: INSEE Jean-Luc Tavernier Director General 1. Directorate for Research, Studies, Evaluation and Statistics at the Ministry of Social Affairs and Health 3

5 Expert testimony Ensuring statistics accurately reflect a changing economy is one of the hardest challenges that NSIs face Sir Charles Bean is Professor of Economics at the London School of Economics and former Deputy Governor for Monetary Policy at the Bank of England. He was responsible for the United Kingdom s independent economic assessment mission for economic statistics, launched in 2015 in the framework of the UK government s productivity plan. Made public in March 2016, the Independent review of UK economic statistics final report 1 provides many lessons on the challenges faced by National Statistical Institutes. The review you have undertaken has been prompted by the growing difficulty of measuring output and productivity accurately in a modern, dynamic and increasingly diverse and digital economy. Why is it so difficult? The national accounting framework was originally developed in the years around World War II in order to measure the output, expenditure and incomes in economies that were still predominantly industrial in nature. Goods were relatively homogenous and delivered according to conventional business models. Measuring the amount of stuff produced in such an economy was certainly difficult, but economic developments over the past 70 years have made it even more of a challenge. Some of these challenges are long recognised. First, our economies have become increasingly services oriented in the UK, for instance, more than three quarters of GDP originates in the services sector, yet manufacturing still accounts for almost a half of the industries in the Standard Industrial Classification. Moreover, many services are bespoke and tailored to the needs of the buyer, making it harder to identify a representative service to which statisticians can attach an appropriate price (necessary for evaluating real magnitudes). And in the public sector, the provision of services such as health and education is often free at the point of delivery and so takes place without any prices attached at all. Measuring the economy has become even more challenging in recent times, however, in part as a consequence of the digital revolution. First, quality improvements and product innovation have been especially rapid in the field of information technology. Moreover, some products may provide an enormous array of services. For instance, a smartphone is not just a telephone, but also a mini-computer, a camera, a music player, a TV and cinema screen, a gaming console, a GPS navigational device and more, especially given the huge array of apps. Measuring the quality change as one vintage of phone succeeds another represents a significant challenge. And how does one take account of wholly new products? The existing national accounts framework is not well suited to capturing the impact of these. Second, the digital revolution has also made possible completely new ways of exchanging and providing services. Three particular characteristics of digital products products that are stored, delivered and/ or used in electronic format make capturing their economic value difficult: 1. Available on 4 INSEE 2016 Annual Report

6 Digital products are typically non-rival their consumption by one agent does not prevent their consumption by other agents. Indeed, the value of many digital products actually increases with the number of other users (a network effect ). Digital products can be replicated at negligible cost and are frequently indistinguishable from the original. Digital products are weightless and aspatial. They can be easily and freely stored on a computer and sent large distances over a network. A digital product may be costly to create in the first instance. But these features mean that if it can be easily imitated, without a barrier to entry, its price will be apt to be driven down to zero. Moreover, even when the original supplier can inhibit entry, he may still have an incentive to set the price very low in order to attract a large number of users, especially in the presence of network effects. As a result, there is no observable price for use for some of the most important digital products, excepting the fixed cost for general access to the internet. Hence it becomes hard to identify the value to users and to capture it in measures of GDP and of productivity. Firms can essentially generate revenues from online digital products in three ways: levy a conventional charge for access; sell information about their customers to third parties; or sell online advertising space. In the first case, the consumer pays for the product with money; in the second case, the customer offers (or provides unwittingly) their personal information; in the third case, the customer pays with their time, in the form of attention to the advertisements. The first of these is, in principle, amenable to measurement by conventional techniques. It is less obvious that the other two are, though. For instance, advertising expenditure is treated as an intermediate input in the national accounts. Consequently the value of digital products financed through selling advertising space will be accounted for in aggregate GDP only to the extent that it also translates into higher consumption of the goods and services being advertised. Third, advances in digital technology have dramatically reduced the transactions costs incurred when A FUNCTION PREVIOUSLY CARRIED OUT IN THE MARKET HAS BEEN DISINTERMEDIATED INTO HOME PRODUCTION, WHICH LIES OUTSIDE THE CONVENTIONAL NATIONAL ACCOUNTS BOUNDARY. individuals or organisations share the provision or use of an asset or service. Any business, by integrating its existing databases and applications with an internet interface, can now easily reach its customers, employees, suppliers and partners at any time of the day or night, no matter where they are. As a result, companies can engage in their core transactional activities much more efficiently. In particular, customer self-service electronic platforms offer end users 24/7 access to information and support without having to engage with a customer representative. Effectively, these electronic platforms greatly reduce the (marginal) costs of undertaking many information-intensive activities and have eliminated (or reduced) the need for a dedicated intermediary. Take, for instance, the travel industry. Previously, a consumer wishing to book travel or a holiday would visit (or phone) a travel agent to do this. Instead, consumers can now search online to find what they want and then book with hotels and airlines directly, or through online portals such as Expedia, with a significant reduction in the service commission. It is therefore hardly a surprise that the number of independent travel agents has fallen sharply over the past few years. In essence, an activity that was previously undertaken through the market the acquisition of information about travel options has now been outsourced to the consumer. In other words, a function previously carried out in the market has been disintermediated into home production, which lies outside the conventional national accounts boundary. Fourth, the increased availability of, and ability to process, data relevant to many users through the internet at negligible cost has unlocked online marketplaces and social networks that facilitate the purchase, hire, and sharing of assets (the sharing economy ). For instance, Airbnb allows individuals to rent out spare rooms or properties for short-term lets, while city bicycle-share schemes and ZipCar allow individuals to rent transport on-demand without facing the burdens of ownership. A characteristic feature here is the shift from ownership to rental on demand. Now the sharing of idle assets is not new think of purchasing a holiday timeshare, for instance but the sharing economy in its current digital incarnation is unprecedented in scale and 5

7 scope. While the existing national accounts framework might appear capable of capturing this through the payments that are made from the asset renter to the asset owner, in practice it seems likely that many transactions will slip through the net. It is not just assets, but also skills that can be shared (the gig economy ). The digital economy has made it much easier for people to have diversified sources for their labour income. Thus, Uber allows car owners to spend part of their time as, in effect, a taxi driver, while Fiverr and TaskRabbit provide marketplaces for small tasks and services. Again, while these activities in principle should be captured within the national accounts framework, in practice many of the transactions may not be picked up in tax returns and earnings surveys. You also underline the need to constantly improve the internationallyagreed statistical methodologies. How can a NSI face this kind of challenge? Ensuring statistics accurately reflect a changing economy is one of the hardest challenges that NSIs face. The basic conceptual framework underpinning the National Accounts when first devised was of an economy in which most businesses were engaged in the production of reasonably homogenous goods in a single country. The reality today is rather different, with many businesses operating across national borders and producing a range of heterogeneous goods and services that may be tailored to the tastes of individual consumers. Measuring this new economy poses particular challenges to established measurement approaches. Examples include: the provision of free services or content that is instead financed by bundling it with advertising; the displacement into home production of information-intensive activities previously undertaken by intermediaries, such as travel agents; the growth of the sharing economy ; and the increased importance of investment in intangible knowledge-based capital relative to that in physical capital. Although hard to capture, some back-of-the-envelope analysis reported in my Independent Review of UK Economic Statistics suggested that phenomena such as these could potentially lead official data on economic activity to be significantly understated and that further investigation was therefore warranted. Moreover, this is not a one-off challenge. As the economy evolves, so does the appropriate frame of reference for statistics: it is a constantly moving target. As a result, the internationally-agreed statistical methodologies will almost always be somewhat out of date or incomplete as they are bound to lag behind the changes in the economy. We should accept that this will be the normal state of affairs, but then look for a constructive way of handling the challenge. In my view, a progressive response to the challenge requires an NSI not only to be abreast in understanding (and explaining) the limitations of its statistics but also to lead the way in developing more appropriate measures after all they should know more about the data that anyone else! In partnership with relevant outside experts and institutions, NSIs, individually and collectively, should therefore establish a continuing programme of research into the measurement implications of emerging economic trends, conducting one-off studies at first to gauge their potential quantitative importance. Such studies may need to exploit novel sources of information that would be unsuitable for deriving regular statistical series, such as private big data. However, the aim is to find out more about the new developments and gauge their significance. In some cases, they may turn out to be quantitatively important, in others not. Once this has been determined, one can decide how, if at all, the new phenomena should be incorporated into the national accounting framework. Even where they are significant, the measurement challenges may be such as to vitiate against their inclusion. But the important thing is that statistics producers and users know roughly what sort of impact they might have for the conventional measures of economic activity. Among the recommendations you've made for the ONS, have you identified some strategic orientations that all NSI should adopt or share? I would identify three such strategic orientations. First, keeping economic statistics relevant to users in both public and private sectors means that NSIs need to move away from focusing largely on the production of statistics and see themselves more as a service provider, helping users answer their 6 INSEE 2016 Annual Report

8 questions about the economy. Moreover, by virtue of their access to the raw data, the NSIs are in the ideal position to understand and explain the limitations of their statistics and to develop alternative indicators when required. In some cases (though by no means all in carrying out my review, I came across examples of very good practice in some NSIs), that may require something of a cultural shift. Staff need to be proactive, rather than reactive, in engaging with users and responsive to those users needs. Staff also need to be curious about what is driving their statistics, self-critical in identifying shortcomings and open in explaining those limitations to users. Finally, staff need to be encouraged to identify ways to improve the production of their statistics, with ideas bubbling up from below. Second, NSIs should seek to take full advantage of the explosion of information and processing capability offered by the digital revolution. That certainly means exploiting so-called administrative data (information held within the public sector but obtained for purposes other than the construction of statistics) to the full. I believe that there should be a presumption that all publicly-held data is available to the respective NSI for the purpose of producing economic statistics, except where there is a very strong reason not to, for example on grounds of national security. But NSIs should look to go beyond the use of administrative data, and investigate the scope for using similar big data held by the private sector, for instance on payments. They should also explore the potential for using new techniques of collecting and analysing big data, such as web scraping, text-mining and machine learning. Although these may not always be suitable as core data sources for regularly produced economic statistics, they may be valuable for cross-checking data, filling in temporary gaps ( nowcasting ), and exploring the significance of new economic phenomena prior to undertaking more systematic measurement. Doing this may, of course, require some initial investment in both IT systems and in building staff capability in data-science techniques. THE INTERNATIONALLY- AGREED STATISTICAL METHODOLOGIES WILL ALMOST ALWAYS BE SOMEWHAT OUT OF DATE OR INCOMPLETE AS THEY ARE BOUND TO LAG BEHIND THE CHANGES IN THE ECONOMY Greater use of this data simultaneously holds out the prospect of more timely and accurate economic statistics, and a reduction in the reporting burdens on businesses and households. Unlocking this trove of information can extend statistical samples to near-census size, increasing their accuracy and allowing statistics, such as regional data, to be finely stratified to match individual user needs. Third, shedding light on hard-to-explain puzzles often requires digging down below the surface into the underlying microdata. Being adept at this is key to understanding the shortcomings and limitations in the data, and also identifying new trends. An enhanced capability within NSIs to interrogate the underlying microdata would better support the production of economic statistics and have the added benefit of making staff better able to sense-check statistics before their release. Realising this objective probably requires greater economic and analytic expertise to be embedded within NSIs. It also requires management to create sufficient space alongside the day-to-day production process to allow such exploratory investigation, something that may not always be the case at present. Finally, the supporting technology and data systems need to be sufficiently flexible and agile to permit effective interrogation of the microdata. An ability to access and interrogate the microdata is also important for some users. Consistent with the objective of providing a better service to users, NSIs ought to establish effective ways of allowing users to interrogate the microdata, either through the provision of direct access to anonymised data or else by carrying out investigations on their behalf. Of course, what is feasible will be determined by the available resources, and the top management of NSIs will no doubt need to make compromises and trade-offs to stay within their budgets. But I believe the digital revolution offers lots of opportunities as well as plenty of challenges! 7

9 Highlights of 2016 JANUARY J J-15 Faster indicators INSEE brought forward the publication calendar for two essential short-term indicators: GDP and the Consumer Price Index. They are now disseminated two weeks earlier. International cooperation with Tunisia Twinning between INSEE and Statistics Tunisia began. The aim of this two-year project is the modernisation of the Tunisian statistical system. FEBRUARY Population census: Internet works! The annual population census was completed. 39% of households, i.e. 46% of people surveyed, replied online, confirming the success of this response option. A new base for inflation INSEE changed the base for calculating the Price Index. The 2015 base is the 8 th generation of indices. The 1 st generation dates back to Every change of base is accompanied by methodological improvements and greater coverage of prices measured. MARCH Better monitoring of enterprises INSEE introduced a new methodology to calculate turnover indices. Major advantage: better monitoring of enterprise demographics. 8 INSEE 2016 Annual Report

10 Better monitoring of associations INSEE published Nine associations out of ten operate without employees, the first survey of associations by official statistics. Insee Première no Aims: sustainable development The United Nations Statistical Commission adopted an initial list of 241 indicators to monitor the Sustainable Development Goals. INSEE was part of the group of experts that drew up the list. APRIL The Institute celebrated 70 years 27 April April 2016: already 70 years that INSEE has been measuring the French economy and French society. Comparing prices: Paris, provinces, overseas INSEE published the results of its 2015 survey comparing prices between the Paris metropolitan area and the provinces, and between the five overseas departments and Metropolitan France. MAY CURRICULUM VITAE Unemployment figures appeared two weeks earlier INSEE made changes to the publication calendar for the unemployment rate. From now on, it is disseminated two weeks earlier than before. Analysis of urban policy districts INSEE delivered new data on income and profiles of inhabitants at district level and published The inhabitants of urban policy districts Poverty alongside other weaknesses. Insee Première no

11 The challenges of international comparisons On 26 May, in Paris, the National Council for Statistical Information (CNIS) organised a conference on the challenges of international statistical comparisons. JUNE The INES model accessible to all INSEE and the Directorate for Research, Studies, Evaluation and Statistics (DREES) made their microsimulation model INES available in open source to promote the wider circulation of data and knowledge, especially relating to poverty and inequalities. A conference to mark 70 years On 29 June, in Paris, INSEE and the Institute for Public Management and Economic Development (IGPDE) co-organised the conference History today, statistics tomorrow: different perspectives on the production and use of statistics. Guaranteeing the quality of European statistics From 31 May to 3 June, the 2016 edition of the conference on the quality of European statistics was held in Madrid, organised by Eurostat and INE, the Spanish National Statistical Institute. JULY White, future Head Office building INSEE signed the leasing agreement confirming the future move to Montrouge of 1,500 Head Office staff. Planned date: first quarter of SEPTEMBER What will INSEE be in 2025? INSEE published Horizon 2025, an ambitious strategy for INSEE presenting INSEE s directions, aims and actions for the decade to come. 10 INSEE 2016 Annual Report

12 OCTOBER Law for a Digital Republic On 7 October, the French Law for a Digital Republic was adopted. Among other things, it allows the official statistics authority to exploit data held by companies. European Statistics have their day On 20 October, Eurostat launched the first European Statistics Day. NOVEMBER Hackathon for SIRENE On 15 November, in Paris, INSEE and Etalab organised the #OpenSirene hackathon before the SIRENE database becomes freely available in January has a facelift On 21 November, INSEE s new-look website went online. DECEMBER Geography and statistics From 15 to 17 November, the 9th edition of the European Forum for Geography and Statistics was held in Paris, organised by INSEE, in partnership with IGN and SOeS. Earlier data on poverty and inequalities After an initial experiment in 2015, for the second year running INSEE published advance indicators for poverty and inequalities for the previous year. 11

13 Producing short-term diagnoses Every quarter, INSEE produces analyses and forecasts of the economic situation in France. In Conjoncture in France there are presentations of important sectors and macroeconomic topics. The publication French economy includes a report on the previous year along with structural analyses. The most recent edition discussed the impact of the Internet on growth. At the same time, INSEE consolidates and publishes many indicators that are used in these diagnoses, such as Gross Domestic Product, Consumer Price Index and unemployment rate. Since 2016, INSEE has published these three key indicators 15 days earlier than previously. Every quarter, INSEE produces a short-term forecast, looking six to nine months ahead. This analysis is published in three issues of Conjoncture in France which appear in March, June and December, and Point de conjoncture which appears in October. Forecasts mainly concern France, but also extend to the Eurozone, the United Kingdom, United States, Japan and the emerging economies. The forecasts for France are not restricted simply to a general analysis, but also include detailed forecasting data (by sector, by major demand item or on the labour market and inflation). To produce these forecasts, INSEE s economic analysts refer to business tendency surveys and detailed quantitative data, many of which are produced by INSEE. These include, in particular, consumer prices and production and import prices, consumption, industrial production, sector turnover indices, employment and unemployment. These data, like the results of the business tendency surveys, are published as soon as they become available in Informations Rapides, of which there are around 340 per year. Economic analysts look at the economic climate and produce specifically focused analyses and reports. For example, in Conjoncture in France in March 2016 they analysed the reasons why growth was stronger in the United Kingdom and measured the consequences of the terror attacks of November 2015 on market service consumption based on daily data on hotel room occupancy in Paris. In December 2016 they assessed the reasons for the difficulties experienced by the aeronautical sector. Every year, French economy produces reports to provide additional or more in-depth information on the previous year s analysis, or to explain INSEE s methods in more detail. In the 2016 edition, one report covered The role of the Internet in describing and analysing the economy (see Testimonial from Ronan Mahieu). Key indicators for the French economy available more quickly In 2016, INSEE managed to bring forward by two weeks the publication of three major economic indicators: Gross Domestic Product (GDP), the Consumer Price Index (CPI) and the unemployment rate. With regard to economic growth, INSEE now disseminates an initial estimate of quarterly GDP 30 days after the end of the quarter, compared with 45 days previously. A second estimate is issued at 60 days and the detailed quarterly accounts are consolidated by 85 days, as was the case before. To have this first estimate available after 30 days is particularly useful for those in the economic and social fields, as well as for Eurostat and the European Central Bank. On 29 January 2016, France became the first country in Europe to make available within 30 days not only the GDP growth rate, but also a complete goods and services account. Concerning inflation, INSEE now produces the CPI on the last working day of the current month. Lastly, the Institute publishes the quarterly unemployment rate about 1.5 months after the end of the quarter concerned. 12 INSEE 2016 Annual Report

14 To guarantee the quality of earlier GDP and CPI estimates, INSEE spent several years carrying out tests. It is mainly the earlier availability of many data sources that made this progress possible. To arrive at an estimate of GDP within 30 days, the Institute has managed to accelerate the production of other indicators used in its calculation. In particular, the Industrial Production Index is the subject of provisional estimates specifically taken into account for calculating the early GDP. This progress has been made possible by using many of INSEE s statistics-producing departments, in addition to the dedicated teams that calculate GDP. Concerning the earlier production of the unemployment rate, this is the result of a reorganisation in the processing of Labour Force Survey data. Providing indicators as close as possible to the period under consideration is part of the challenge faced by all the Institute s statistics-producing departments. In 2016, the time for producing the quarterly indicators for tourism occupancy in hotels, camp sites and other collective accommodation was also shortened by about ten days. At this point, the exact time saved still varies slightly from one quarter to another. It is unrealistic to attempt to isolate the value added generated by the Internet and the digital economy Testimonial from Ronan Mahieu, Head of the National Accounts Division at INSEE, co-author 3 of the study The role of the Internet in describing and analysing the economy 4 What impact does the Internet have on growth? For several years, this question has been a subject of some controversy among economists. Historically, any major disruption or technological innovation results in a marked increase in productivity. However, the emergence of the Internet has not coincided with such an acceleration. The question therefore arises as to whether this paradox is due to a measurement problem, some kind of shortcoming in the statistical system. This is the question that we tried to answer. Our study describes the various economic models that underpin the use of the Internet, from e-commerce to the collaborative economy, and shows how we integrate the associated activities into the national accounts. Use of the Internet has spread to all sectors of activity and it is unrealistic to attempt to isolate the value added generated by the Internet and the digital economy: for example a retailer who offers goods for sale online but also in-store is not necessarily able to distinguish between the turnover generated by one or the other of these channels. Customers may order online but then visit the store to pick up and pay for their products. However, new difficulties arise in identifying the country where an activity is carried out. Traditionally, a trader s activity is counted in the country where his store is located, which is the place where the transaction takes place: what happens when the transaction is dematerialised, and the choice of the location of platforms may be purely the result of tax optimisation? We offer some preliminary ideas on the impact of the Internet in terms of volume of activity, price and well-being. All in all, the impact of the Internet has perhaps less effect on GDP than on the wellbeing of individuals, giving easier access to goods or services that meet their needs exactly, but also to other people who share the same interests. 3. With Christophe Bellégo, research officer in the Economic Studies and National Accounts Directorate 4. Appeared in Insee Références French economy edition

15 Studying the functioning of the French economy and the determinants of its major trends INSEE measures change in the French economy, but also explores its overall functioning and its relations with its international environment, both in the short and long term. This study focuses on several models that the Institute develops or updates on a regular basis. To assist public decision-makers, INSEE also mobilises its databases to produce public policy evaluation studies. Among the subjects covered in 2016: the impact of the Hadopi law, the consequences of the 2010 pension reform, the effects of the hiring premium for SMEs, etc. As an essential and regular provider of key figures and trends in the French economic situation, INSEE relies on its expertise to analyse the global functioning of the French economy and take into account interactions and impacts linked with the international environment. The quarterly Conjoncture in France provides analysis reports on these complex phenomena, like the one entitled French growth in 2015: external and internal factors balance each other out, published i n the June 2016 issue. INSEE also takes a regular interest in the intense upheavals that shape the new trends in the French economy. For example, when public debate increasingly focuses on the impact of new technologies on employment, INSEE proposes Accounting for technology, trade and final consumption in employment: an Input-Output decomposition to answer the question What are the driving forces behind changes in employment in France over the last thirty years and how can we explain skill-biased changes in employment?. Informing the choices of public decision-makers In order to guide public decision-makers in directing their policies, INSEE produces evaluation studies on a wide variety of public policies. In 2016, during the international Cannes Film Festival, INSEE published an analysis entitled, W h at a re the effects of the Hadopi law? The case of cinema admissions. The study reveals the positive impact of the law on the market share of American films in cinemas, which has grown by 9%. Other analyses during the year include: The impact of transport infrastructure on local development, with the first study focusing on the use of the Île-de-France regional train network (RER). Key lessons learned were that employment appears to have increased more rapidly in the Îlede-France municipalities connected to Paris by an RER line than in those with similar characteristics, but served only by the commuter train network. The second study paves the way for an analysis of the High Speed Rail infrastructure as a competitiveness factor for French enterprises. The effect of the 2010 pension reform on older workers employment and activity. The labour force participation rate of employees concerned by the reform, between 2008 and 2014, appears to be between 19 and 22 points higher than that of employees not affected. An initial evaluation of the effects of the hiring bonus in SMEs concludes that in H1 2016, 54% of fixed-term contract hirings of at least 6 months and 34% of open-ended contract hirings in enterprises with less than 250 employees benefited from the bonus. However, by focusing on enterprises around this threshold, the evaluation does not reveal any significant effect of the bonus on the numbers hired. The need to use models INSEE uses sophisticated models to produce its analyses, such as the macroeconomic model MESANGE (Econometric Model of Simulation and General Analysis of the Economy) 1 or the multi-country model NiGEM (National Institute Global Econometric Model) Joint model by INSEE Ministry of the Economy and Finances 2. Developed by the National Institute of Economic and Social Research ( UK ) 14 INSEE 2016 Annual Report

16 In 2016, the Institute developed a new model for the economy of the Eurozone called MELEZE (Linearised Economic Model of Equilibrium in the Eurozone). Its aim is to understand how belonging to a monetary union can affect the implementation and the macroeconomic consequences of various economic and tax policies. To analyse the effects of redistributive and social policies, INSEE uses the INES model, which it created with DREES. Since 2015, INES has also been used to estimate the poverty rate over a reduced timescale. Another model, DESTINIE, which simulates individual retirement behaviour, was used for projections in the report Life expectancy post-retirement without severe disabilities should be stable for the generations born between 1960 and 1990, which appeared in the 2016 edition of France, social portrait. Opening up the INES model is a true public service action On 14 June 2016, INSEE and DREES opened the source code of the INES model (acronym for INSEE-DREES). Created in 1996, this model simulates direct taxes and social contributions and social benefits using data representative of the French population taken from the Tax and Social Revenues Survey. Testimonial from Xavier Timbeau, Principal Director, French Economic Observatory (OFCE). What does the opening up of the INES model represent for you? This is an important step in statistical information. By making this model available to all, INSEE and DREES are doing much more than simply making figures available. They are proposing a tool that provides an exhaustive description of the taxbenefit system, making it possible to study the way the system operates and obtain some very detailed information on monetary redistribution. This model is a valuable tool for us, for policy design, for simulating the impact of specific tax or social measures. With INES, for example, we are able to simulate the impact of a universal income and see who are the winners and the losers. We can use a macroeconomic approach, but this is not all. The model allows us to calibrate in great detail the parameters of the measurements we are analysing, and test, for example, the effect of awarding a certain amount for the old-age minimum income. In my opinion, opening up INES is an extremely useful move, and a true public service action. What do you plan to do with this model? We have used it to prepare a report on fiscal measures over the period , i.e. when the first budgetary constraints were introduced. Other projects include a forecast of the impact of measures before they are endorsed by the draft Finance Law. Every year, INSEE and DREES provide an analysis of the impact of benefit and contribution reforms introduced the previous year. We want to provide an a priori impact analysis. We also intend to add further dimensions to the model. For example, we have a research project funded by ADEME to examine the impact of a carbon tax on households, assessed according to the amount of CO² emissions they produce. 15

17 Providing details about the productive system Interviewing companies about their activities, their investments, their recruitment, and their organisation is essential in order to understand the productive system, monitor the economic outlook and enable companies to situate themselves in relation to other companies. To produce information that is always as relevant as possible and which better reflects changes in the productive system, especially due to globalisation, INSEE is working on a long-term project, group profiling. Among the key studies of 2016 are reports on SMEs and on city centre retail trade in Enterprises in France, a regional focus on enterprise creation and a report on the summer tourist season. The results of business surveys are of interest to several types of user. Professional bodies use them to monitor their sector and develop their strategies. Companies use them to compare themselves to their competitors and to gain a better understanding of sectors upstream or downstream from their own activities. Public stakeholders rely on these results for their decision-making. For economists, surveys fulfil many needs for sector-based or cross-sectional analyses. They are also essential for preparing the national accounts. To satisfy these many informational needs, INSEE has set up a very comprehensive survey system within the framework of European regulations. This system collects structural information (size, nature of the activity, characteristics of the workforce, type of organisation, etc.), short-term information, and data on specific topics (energy consumption, IT use, innovation, etc.). Several measures have been implemented to reduce the burden that these surveys represent for businesses. In particular, INSEE and the Ministerial Statistical Departments have improved their coordination in carrying out their surveys and make more regular use of the administrative data that companies have already provided. All in all, according to professional organisations, the burden of statistical surveys accounts for only 1% of the administrative constraints imposed on enterprises. Better awareness of companies economic reality INSEE is gradually defining enterprises in accordance with the economic definition introduced by the implementing decree of the Economic Modernisation Act of 2008 (LME) relating to categories of enterprise. Previously, the enterprise was defined in purely legal terms, and associated with the legal unit, i.e. an unincorporated enterprise or a company carrying out a production function. Now it is defined as the smallest combination of legal units that is an organisational unit producing goods and services, enjoying a certain decision-making autonomy, especially for the allocation of its current resources. This new definition gives a better understanding of how enterprises are organised into groups and thus a better insight into how the productive system is really organised. When the enterprise was assimilated to the legal unit, this did not give an accurate picture of the reality of companies owned by other companies within a group organisation, as they could have limited decision-making autonomy, or indeed none at all. To implement this new definition, INSEE is carrying out profiling operations which consist in identifying within the groups the relevant enterprise(s) within the meaning of the 2008 law, then recreating their consolidated accounts. Group profiling is a longterm operation, which INSEE began in Three categories of groups are defined: The first includes about a hundred large groups, designated as priority groups because of their size or complexity. Their profiling is carried out faceto-face in a dialogue with senior management. Only the group knows its operational structure and its information system, and knows whether it can provide information on enterprises created through profiling. By the end of 2016, the operation had been completed for 51 of these large groups; their 16 INSEE 2016 Annual Report

18 accounts are now processed at the level of the profiled enterprises, which are surveyed within the framework of the annual structural surveys; A second category includes 80,000 small or medium groups. There are too many of these groups for individual monitoring to be carried out, and they are considered as enterprises. Their consolidated accounts and their responses to the structural survey are constructed using an algorithm based on corporate accounts and survey questionnaire responses by their subsidiaries. This new method will be used for the first time in 2018 to calculate data for the 2017 financial year; The third category consists of around 3,900 groups of median size. They will be dealt with automatically over the short term, like the previous category. In June 2016, for the first time, the definitive results for the structural statistics for enterprises that were sent to Eurostat took into account 109 enterprises created through profiling 44 of the largest groups. Spotlight on SMEs and city centre retail trade Among the most significant publications of the year, the 2016 edition of Insee Références Enterprises in France, contained two reports on SMEs. The first analysed the link between financing conditions and export behaviour by SMEs. Payment deadlines seem to play a direct role in the export dynamics of SMEs, although the effects are modest. The second report looked at the size of SMEs and whether they were organised as part of a group. It does not really make sense to criticise the fact that the French productive fabric does not have many units with 50 or more employees, as most small SMEs tend to favour external growth, by combining with another unit: whether an SME with 30 employees doubles its workforce or acquires another SME also with 30 employees should not affect the analysis of the productive fabric. Organisation into groups is important, even for small units. The majority of legal units with 20 employees and over are part of a group. Organisation into groups increases at around 50 employees. A third report highlights the fact that in the last ten years, payroll employment in city centre retail trade has increased in the large regional metropolitan areas, but has decreased in the smaller urban areas, especially those of average size. Some of the factors that have affected this vitality are tourism, interest in heritage and proximity to the coast. National and regional snapshots of enterprise births and the results of summer tourism In addition to national analyses, INSEE also regularly carries out studies focusing on the regional level. The annual report on enterprise births includes a regional section. Findings for 2015: all types of enterprise combined, enterprise births fell in all regions, except Martinique and Île-de-France. The regional dimension is also important for the tourism results. For the summer 2016 tourist season, INSEE noted that although foreign tourists withdrew in large numbers to the Paris metropolitan area, they also visited the countryside and the mountains more than in

19 Observation of the society As a result of the regular household surveys it carries out, including the population census, and analysis of large numbers of administrative records, INSEE is able to observe developments in French society. Unemployment, inequalities, family structure all these are dimensions of the social portrait of France. Some of the issues under the spotlight in 2016: the challenge of measuring the halo of unemployment, and the difficulties young people experience in terms of access to housing, education, employment, violence, and also juvenile delinquency. Alongside the household surveys that INSEE conducts regularly, it also uses many administrative records to produce key figures and studies on French society. New prospects for measuring employment The Nominative Social Declaration (DSN), required by law, 1 is gradually replacing all declarations that employers must make to a variety of bodies (CPAM (public health insurance), URSSAF (Social security contribution collection office), Pôle emploi (Job Centre), Tax office, Funds for special tax systems, etc.). This change is accompanied by several major developments for INSEE, which uses these declarations to construct its information system on employment and earned income (SIERA). In 2016, INSEE received the first DSNs, and subsequently the declaration was widely used by enterprises in January INSEE introduced a new method for processing these data, starting with reception, checking, formatting statistical concepts, and including identification and coding of the profession, so that this source could be exploited to the full. In SIERA, the DSN replaces two administrative sources used previously: summary contributions forms (bordereaux récapitalutifs, BRC) for the general and agricultural schemes, and the Annual Declarations of Social Data (Déclarations annuelles de données sociales, DADS). INSEE began to make radical changes to its information system in order to adapt to the We were able to obtain valuable advice not only on processing the data but also on making them more meaningful Testimonial from Fabrice Gilles, University of Lille 1 - LEM and TEPP, co-author 2 of an assessment report on, The effects of the CICE (tax credit for encouraging competitiveness and jobs) on employment, wages and R&D requested by France Stratégie and published in September 2016, and Yannick L Horty, University of Paris-Est Marne-la- Vallée - ERUDITE and TEP In order to carry out our analysis we needed data from four producing services: ACOSS, 3 DGFIP, INSEE and the Ministry for Higher Education and Research. From data made available by INSEE, we used the information in the exhaustive version of DADS 4 and also FARE 5 data which provide companies entire accounts. Being able to use these data is the prerequisite for the work we carried out. The comprehensive and precise nature of the data was key. We were provided with all our statistical requests and conducted our econometric study after merging all the sources made available via secure access on the CASD. 6 In addition to providing data, INSEE also helped us with their technical expertise. We submitted a first version of the report to a panel of experts, some of whom were with INSEE, for an initial evaluation. As a result, we were able to obtain valuable advice not only on processing the data but also on making them more meaningful in response to the central question, What is the impact of the CICE? INSEE helped us select the best indicators, in particular for pinpointing the accounting implications for companies. For example, regarding the impact on employment, was it more appropriate to consider employment on 31 December or to take a rolling average over the year? Thanks to their valuable comments, we were able to refine our analysis and produce a reliable report. 1. Law no of 22 March 2012 concerning the simplification of the rules and the streamlining of administrative procedures 2. With three other researchers: Mathieu BUNEL, University of Bourgogne, LEDi and TEPP, Ferhat MIHOUBI, University of Paris-Est Créteil, ERUDITE and TEPP and Xi YANG, University of Paris-Ouest Nanterre la Défense, ECONOMIX 3. ACOSS: Central Agency of Social Security Associations (Agence centrale des organismes de sécurité sociale) 4. DADS: Annual declaration of social data (Déclaration annuelle des données sociales) which compiles the wage bill paid throughout the year by all employers, including general government, and comprises a list of staff employed, names and wages, with their position and the compensation received 5. FARE: fichier approché des résultats is a form providing results from the Annual Business Statistics Programme (Élaboration des statistiques annuelles d entreprises, ESANE) which gives detailed results from company balance sheets (turnover, investment, etc.) 6. Secure Data Access Centre (Centre d accès sécurisé aux données) 18 INSEE 2016 Annual Report

20 change in social declarations, make it more responsive to any future changes and to take advantage of new data provided in these declarations. While this change was happening, INSEE also started work on rationalising the production of quarterly employment estimates. This work was carried out in partnership with the Central Agency of Social Security Associations (ACOSS) and DARES. The aim was to produce these estimates jointly to avoid any duplication of studies on the private sector, and to publish only one quarterly indicator of change in employment, instead of the two or three previously. Now that the scope covers all employees, including the public sector and the agricultural sector, this co-production will ensure exhaustive monitoring of short-term changes in payroll employment. In the longer term, the statistical processing of the DSN, as well as other changes introduced into the Labour Force Survey, should shed light on other subjects such as particular forms of employment, in accordance with the recommendations of the CNIS in its Report of July Spotlight on the halo of unemployment and the difficulties faced by young people In the collection Insee Références, INSEE presents a new focus with each edition. Among the issues published in 2016, Employment, unemployment, earned income expands the topics touched on in the previous issue Employment and income. One of the four new reports that it contains looks in detail at the halo of unemployment, defining its different components. At individual level, the halo is often a more transitory category than unemployment: on average, between the end of 2013 and the end of 2015, only 32% of people in the halo were still there the following quarter, while unemployment was twice as stable. Another leading publication, France, social portrait, which appeared in November, contains five chapters throwing the spotlight on the problems encountered by young people in France in terms of access to housing, education, employment, violence, and juvenile delinquency. While 43% of young people aged have their own accommodation, only 17% live there exclusively and pay for it themselves. Regarding professional integration, the differential between young graduates from higher education and other young people has widened significantly over the last ten years. Concerning physical violence, young people aged 10 to 25 are twice as likely to be victims than the over 25 and four times more likely than children under 10. The challenge of increasing online participation in household surveys INSEE is continuing trialling household surveys online. The aim is to offer the online method of response for all household surveys within the next few years, as is already the case for the annual population census. In 2016, as part of the MUSE project, looking at responding to the Labour Force Survey online, two test operations were carried out on large samples to test this new protocol. These measures are planned to come into use by , to correspond with plans to modify the Labour Force Survey in line with European regulations. Some of the expected benefits: quality (possibility of reaching populations who are not often available by telephone or for face-to-face interviews), efficiency (dedicated survey interviewers will collect survey responses not carried out online), and greater reactivity (easier to organise surveys, therefore making it possible to respond to requests for new surveys). 19

21 Conducting the population census and using the results The French population census has historically been one of INSEE s activities, as the Institute organises and supervises the surveys, uses the data and disseminates the results. More and more households are now responding to surveys online. Using data from all the responses, INSEE provides statistics on the inhabitants, families and their housing, the number of people and their characteristics: sex, age, nationality, profession, housing conditions, travel to and from work, transport used for these trips, etc., all of which is essential information for many public and private stakeholders. Online response is gaining ground Since 2015, every household surveyed for the census has been able to choose between replying using a paper form or online. When the census agent contacts a household, he or she suggests they reply online. In 2015, one in three of the households surveyed opted to respond in this way. In 2016, 39% of households and 46% of individuals surveyed responded online. The end of the paper census means a reduction in the cost of data collection: fewer trips for the census agents who no longer need to return to people s homes to retrieve the completed questionnaires, and less paper to file, store and transport. Online responses do not need to be entered by INSEE. All this represents a saving in both human and financial resources. Data collection is easier both for municipalities and for INSEE. Those working in the field, especially the coordinators in the municipalities and the supervisors, can follow the progress of data collection in real time using the tool specifically designed by INSEE. Measuring two-wheeled travel in more detail To enable comparisons to be made over time, it is essential that the census questionnaires change as little as possible. Since the new annual data collection method was introduced in 2004, it is essential that all questions remain identical for at least five years so that the entire territory is covered. However, this questionnaire is not set in stone as society changes along with the social demand for statistics to shed light on these changes. There were considerable constraints in the preparation of the survey, in addition to those linked with the stability of the results over time and space, because the total size of the questionnaire could not be increased, so as not to penalise respondents: any additional question had to be matched by the removal of another. Among the other technical imperatives: as the questionnaire was to be filled in by the residents themselves, with no help from a survey worker, the questions had to be simple, clear and understood by everyone. The requirements of international harmonisation also had to be considered, as they required certain topics to be included and the corresponding response modalities. The National Council for Statistical Information ( Conseil national de l information statistique, CNIS) has addressed this question in order to best respond to the expectations of civil society and public stakeholders. Among the topics requiring more exhaustive statistical information was that of mobility. Those surveyed gave the mode of transport that they usually used to go to work. Since the 2015 survey, the response two-wheeled transport has been split in two: bicycle (including electricallyassisted) and two-wheeled motor vehicle. During 2016, experts from INSEE were therefore able to exploit the new survey data and carry out initial analyses, based on a single annual survey, of the use of the bicycle in home-to-work travel. An initial study has been published, Setting off in the morning, by bicycle, published on 17 January INSEE 2016 Annual Report

22 The census is an essential tool for planning the city and developing metropolitan areas Here we learn, in particular, that in 2015, only 2% of the occupied labour force travelled to work by bike. INSEE will be able to conduct more in-depth studies from 2020 onwards, when the full results from five census surveys will be available. More and more people work outside the municipality in which they live Home-to-work travel has given rise to other analyses, in addition to mode of transport. On 30 June 2016, INSEE published a study that described this situation, highlighting major trends across France but also the results for each region. Called An increasing number of people work outside their municipality of residence, the study showed that every day 16.7 million people leave the municipality where they live to travel to work, i.e. two out of three people who have a job. These commuters are proportionally most numerous in Hauts-de-France (71%) and Île-de- France (69%). We also learn of disparities according to professional category: three quarters of engineers and company managers are commuters compared with only a quarter of the self-employed or home helps. The survey also highlights the fact that the car is still the main means of transport with 80% of commuters using this method. Testimonial from Patrick Brun, Project Director Observation of Territories and Observatories & Head of production monitoring, UrbaLyon, urban planning agency in the Lyon metropolitan area Whatever the study we need to carry out, the census data form the basis of our work. When we need to shed some light on a sectoral policy, with some very specific questions, we have to use the detailed census files where practically every line is an anonymised individual (or dwelling). For Lyon, for example, in the context of housing policy, we needed to isolate all the characteristics of dwellings that had a lift during the construction period, cross-referenced with elderly people to follow a public action indicator over time. Having data at a detailed geographical level is crucial in order to produce analyses with which to consolidate accurate urban development projects. This enables us to quantify and identify the population on either side of an infrastructure project such as a road, a subway or a tramway. The census is truly a rich source of data for research and development. For one study, we isolated the buildings used for housing and estimated the volume of each building using sources giving the building heights. We were then able to calculate a habitable volume for every district. By geographically assigning the population of the district to this volume, we were able to distribute the population in the buildings. Next, accessibility indicators were calculated. For example, people s accessibility to green spaces. This type of work enables us to optimise our spatial development in terms of quality of life. Planning and building the city without the census would be impossible! 21

23 Describing and analysing regions and territories With its regional presence, INSEE is able to build up a detailed knowledge of the different French territories. The Institute s Regional Offices produce many studies, very often in partnership with local public stakeholders. INSEE also regularly offers studies with a dual perspective, which are both national and regional in scope. Among the highlights of 2016 was the report, The inhabitants of urban policy districts. Another event during the year was a double anniversary in Reunion Island which celebrated 70 years as a department and 50 years of INSEE on the Island. From a general perspective Every year, INSEE s Regional Offices produce hundreds of studies. No fewer than 743 in 2016! Of course this work primarily concerns all the local players, but their general scope means that they are also useful to civil society and the general public. Just like the national publications, these studies use the large amounts of statistical data that the Regional Offices help to produce and consolidate, via the annual census survey or business surveys, for example. Some of the topics that are constantly being reviewed are the economic outlook for the region, especially in employment, the annual economic balance sheet, changes in population, etc. to advising and analysing a territory or an industry The Regional Offices provide advice and expertise for regional public players in response to a demand linked with the introduction of a public policy, they participate in think tanks and regional committees and they support local stakeholders as they use the data. Local players regularly need analyses to help them define the specific features of a territory or an industry, to prepare an assessment at a given point in time, to define strategic directions for action then carry out an evaluation at a later date. This is the case when they want to anticipate economic and demographic change, build new inter-municipal schemes linked with territorial reform or monitor and evaluate any European funding to which they may be entitled. The Regional Offices then produce specific studies, in partnership with public players at regional level. They may be regional prefectures, government departments or regional or departmental councils, public intermunicipal cooperation establishments (Établissements publics de coopération intercommunale, EPCI), regional or departmental tourism committees, metropolitan areas, urban planning agencies, etc. In 2016, of the 743 regional studies, 269 were produced in the framework of a partnership with regional public players. The territorial reform certainly generated demand for INSEE to assist with descriptive studies, such as portraits of departments, detailed overviews of the strengths and weakness of the regions and their territories. These studies are always published. Among those produced in 2016 were two regional atlases, Hauts-de-France: the population of the territories of the Hauts-de-France region on 1 st January 2014 («Hauts-de-France : La population des territoires de la région Hauts-de-France au 1 er janvier») and Auvergne-Rhône-Alpes: a new regional space and metropolitan dynamics («Auvergne-Rhône- Alpes : Nouvel espace régional et dynamiques métropolitaines»), in addition to reports such as that on the metropolisation of Nouvelle Aquitaine entitled Territorial dynamics, metropolisation and networks in Aquitaine-Limousin-Poitou-Charentes («Dynamiques territoriales, métropolisation et réseaux en Aquitaine-Limousin-Poitou-Charentes» ) a n d another on the Grand Est about regional and crossborder dynamics, Socio-dynamics of the Grand Est in its regional and cross-border environment («Les dynamiques socio-économiques du Grand Est dans son environnement régional et transfrontalier»). 22 INSEE 2016 Annual Report

24 Statistics serving the future: the example of Reunion Island Testimonial from Valérie Roux, Regional Manager of INSEE La Réunion Mayotte 2016 for us was marked by a double anniversary: the 50 th anniversary of our Regional Office and the 70 th anniversary of Reunion Island as a department. These two important moments were the opportunity to highlight our expertise with figures, key data and studies, produced for all types of public interested in knowing more about the economic and social situation in Reunion Island. The main focus of our communication campaign for our 50th anniversary was to show how this situation has developed since we were first established via 6 short videos that show different aspects of the change in the island s economy, its population and in living conditions. Alongside the French Development Agency (Agence française de développement, AFD), the Economic, Social and Environmental Regional Council (Conseil économique, social et environnemental régional, CESER), the University of Reunion Island, the Overseas Department Issuing Institute (Institut d émission des départements d outremer, IEDOM) and the Departmental Council, we were closely involved with organising a conference on the department s 70 years. It took place on 5 and 6 October in the Departmental Council chamber, and brought together almost 200 people, key players and political and economic decision-makers representing civil society. The title of the conference was What lessons for the future, and it provided the opportunity to work together to identify the forces we need to muster to contribute to the future development of Reunion Island. On one point there was unanimous agreement: the current model for growth and development seems to have reached its limits. A new economic model must be consolidated, and INSEE intends to make its contribution, statistics at the ready! Expertise is shared as far away as Athens To carry out their work, the researchers and project managers in the Regional Offices draw on the expertise of the five Regional Action Service Clusters (Pôles de services de l action régionale), each specialising in a specific field: territorial analysis, urban analysis, regional economic studies, employment and population, and local summary reports. The sharing of skills, tools and methods is a valuable practice to guarantee that inter-regional comparisons remain relevant or when extending beyond the local territories to national level. Every year, INSEE and the Regional Offices produce around a dozen studies involving coordinated operations, i.e. associating a regional publication with a national publication. In 2016, 11 studies were produced using this dual approach. These included, The inhabitants of urban policy districts. Poverty alongside other weaknesses which was published in May 2016 by the Urban Analysis Cluster and then appeared in 12 regional versions: Alsace- Champagne-Ardenne-Lorraine, Nouvelle-Aquitaine, Auvergne-Rhône-Alpes, Bourgogne-Franche-Comté, Bretagne, Corse, Île-de-France, Occitanie, Hautsde-France, PACA and Pays de la Loire. These results were presented by INSEE at a seminar on urban poverty organised by the European Commission in Athens in September Filosofi, the social and fiscal localised incomes system (Fichier localisé social et fiscal), which is the source of the data used by INSEE for this type of study, also caught the attention of our European partners and the OECD at a Eurostat conference on urban and regional statistics, held in Lisbon in June

25 Managing registers INSEE is responsible by law for regularly updating several registers: the Computerised System for the Register of Companies (Système informatisé du répertoire des entreprises et des établissements, SIRENE) used to identify enterprises; the National Directory for the Identification of Natural Persons (Répertoire national d identification des personnes physiques, RNIPP) and the general electoral roll. Among the highlights of 2016 were improvements to SIRENE and the launch of a project to create a single electoral roll. Under French law, INSEE is responsible for managing SIRENE, the RNIPP and the general electoral roll, even though they are not used solely for statistical purposes. As INSEE has considerable experience in handling large databases, it is able to guarantee their security and confidentiality. In so doing the Institute is carrying out a sovereign administrative mission in managing registers of enterprises, establishments and individuals. Managing and improving the reference database for enterprises and establishments SIRENE includes all the enterprises (legal entities and sole proprietorships), and all the establishments to be found on French territory. Every enterprise is identified by a SIREN number, and every establishment by a SIRET number which begins with the SIREN number of the enterprise with which it is associated. SIRENE has covered government agencies since 1983 and the agricultural sector since In 2016, 9 million enterprises and 10 million active establishments were registered. INSEE updates SIRENE from the declarations completed by companies when they start up and when other events occur that affect them, such as a change of address or activity or a cessation of activity. This information is recorded by the Centres for administrative formalities for businesses who register these declarations before sending them to INSEE. Almost all enterprise births, which INSEE mainly receives electronically, are processed in under 24 hours. Every day, 15,000 modifications are dealt with, of which 20% are enterprise births. From the information available on each entity, SIRENE records the registered address. The substantial updates in 2016 also involved the inclusion of more than 300 new municipalities as a result of the mergers of 1,200 former municipalities. In 2016, the database also added new information categories. Enterprises and establishments working in the field of the social and solidarity economy are now more easily identified, since a new dedicated field has been added. Also, for every association registered as an establishment, SIRENE now includes its National Directory of Associations (RNA) number, managed by the Ministry of the Interior. On the one hand, this improves the administrative information system for these entities and, on the other hand, it simplifies procedures for associations as they can now use their SIRET number to apply for State or territorial authority subsidies, or to hire employees. Two reference databases for French citizens The National Directory for the Identification of Natural Persons (Répertoire national d identification des personnes physiques, RNIPP) includes everyone born in France and also everyone born abroad who needs to be registered within their social sphere, i.e. anyone who requires a registration number (NIR), usually known as the social security number (or INSEE number). For each person recorded, the RNIPP includes the NIR and information on civil status: full name, date and place of birth and, where applicable, date and place of death. INSEE updates the directory mainly from registrations of births and deaths and any other records of change in a person s civil status. These documents are sent in by local municipalities, and 91% are now sent electronically. The RNIPP is widely used to check the accuracy of personal data that appear in different 24 INSEE 2016 Annual Report

26 files. Many government agencies, especially the tax authorities, need an unambiguous identifier for the people they are dealing with; at INSEE too, several statistical operations require highly accurate personal identification. All these applications are controlled by the National Commission on Information Technology and Freedoms (Commission nationale informatique et libertés, CNIL). For each person currently or formerly on a voting list, the general electoral roll gives their civil status, the municipality where currently registered and where registered in the past, and where applicable, it mentions any disqualification from voting. INSEE manages this file to check that no one is registered on more than one electoral roll, and to ensure that a disqualified person does not vote. It is from this register that updates are sent to the municipalities for their electoral lists. Single electoral roll: greater responsibility for INSEE to maintain the quality of electoral lists The Law of 1 st August 2016 overhauled procedures for registering on electoral lists and established a single electoral roll, which INSEE will be responsible for maintaining. It also replaced the annual revision of electoral lists with continuous updating. The main challenge is to enable citizens to register to vote as close as possible to a ballot. The end of the year will no longer be synonymous with the registration deadline. This new system makes profound changes to INSEE s role in maintaining electoral lists. Currently, while INSEE is the hub for information used to update electoral lists, each municipality is responsible for its own list. In the new system, INSEE will be responsible for the quality of all lists. In consultation with all the stakeholders involved in this change (municipalities, Ministries of Foreign Affairs, Overseas Territories, Defence and Justice, public-service portal), and as part of a wider-reaching project steered by the Ministry of the Interior and covering other aspects of the modernisation of the electoral process, INSEE is in charge of constructing the electoral roll and creating a new tool so that there can be a continuous feed, in conditions that are both robust and user-friendly. In autumn 2016, INSEE launched the project ELIRE (Élaboration des listes à partir du répertoire électoral) to prepare lists based on the electoral register. The objective is two-fold: to construct the management system for the future process and to initialise the register based on municipality and consular lists and the current general electoral roll for which INSEE is responsible. In the words of Magali Demotes-Mainard, Project Manager, Through ELIRE, INSEE can make its own contribution to improving the way our democracy works. The new system for producing electoral lists is expected to be launched in

27 Disseminating its statistics and studies INSEE has chosen to disseminate its statistics and studies as widely as possible. Since 2003, the Institute has chosen the Open Data option, by offering free online access to millions of data and all its publications on the INSEE website. To simplify access to these publications, a new version of the insee.fr website was launched in November More key actions during the year included greater visibility for all official statistics publications and preparations for the dissemination of the SIRENE database free of charge. Statistics are a public good, for the benefit of all. As a pioneer among National Statistical Institutes, INSEE has taken the decision to offer free online access on the website to all data produced since 2003, and to all publications. This is truly a goldmine for many categories of users, economists, researchers, journalists, curious citizens; the insee.fr website is the Institute s dissemination medium of choice, with more than 30 million visits annually. A wide variety of data are available: key figures, chronological series, detailed results, local data, etc. and also studies published by INSEE. To optimise access to all these products, the Institute carried out a major overhaul of the website. A new version has been online since 21 November 2016 (see below, Testimonial from Lucile Olier, Head of the Dissemination Division). Since 2016, the publication calendar for all Ministerial Statistical Departments has been available for consultation on the insee.fr website. It therefore showcases the entire range of production by the official statistical system. To meet the requests of its users, INSEE provides continuous telephone or assistance for all those who are unable to find the information they need or who need further explanations. And for users who would like additional tables to those available on the website, INSEE even offers tailor-made products to meet their specific needs. In 2016, 455 such requests were dealt with. Preparing to make SIRENE freely available As well as producing and disseminating statistics and studies, INSEE also manages files and registers required for administrative purposes, including the Computerised System for the Register of Companies and their Establishments (Répertoire national des entreprises et de leurs établissements, SIRENE) (see section Gérer des grands répertoires ). With new legislation paving the way for a Digital Republic, the Institute worked throughout 2016 to go even further in data dissemination by making the SIRENE register freely available, as from January In just a few months, dedicated teams completed all the necessary changes to improve our offer even further: the possibility of creating free files containing up to 100,000 establishments from the sirene.fr website; making the database available in the context of Etalab s public service data mission; providing support for the network of SIRENE s major customers on the legal and technical consequences of opening up the database. In addition to this considerable amount of work, INSEE and Etalab organised the #OpenSirene Hackathon on 15 November in Paris, as part of the Week of public innovation. The 80 participants were able to use datasets that INSEE made available to all from January Anticipating the needs of the general public and especially younger generations It is INSEE s goal to give the general public, and especially younger generations, the opportunity to understand statistics and to be aware of the key figures in the French economy and society. Every year the Institute is very involved in two events that provide an understanding of the economy for a non- 26 INSEE 2016 Annual Report

28 Meeting such a wide variety of needs is a real challenge! Testimonial from Lucile Olier, Head of the Dissemination Division Hackathon #OpenSirene, November , Sensespace Paris specialist public, made up essentially of university and high school students: Printemps de l Économie organised in Paris and the Journées de l Économie in Lyon. During Printemps de l Économie 2016, held in mid- April, INSEE organised a round table on Quality of life in the French territories, how can we measure it? What are the issues? Sylvie Marchand, Head of the Regional Action Division, highlighted INSEE s contribution: by providing objective indicators to look at different aspects of quality of life, health, education, housing, employment, income, social relations rather than giving a list of the best regions to live in, the Institute showed everyone how to evaluate quality of life according to their own criteria. At the Journées de l Économie in Lyon, INSEE also focused on the educational aspects of their presentations. Some of the subjects discussed: How do we measure poverty? What do key notions like purchasing power mean? What does GDP measure? Among the new communication products for the general public disseminated in 2016, the video A day just like any other (Un jour comme les autres) provides figures about the French population collected by INSEE. Every day, more than 85,000 people visit the INSEE website and their profiles are very varied: students, teachers, researchers, consultants, R&D, journalists, decision-makers from the public and private sectors, etc. They may be looking for the latest value of an index to revise a contract, or they want to download a database, consult the information available on their municipality, or read one of the Institute s recent studies. Meeting such a wide variety of needs is a real challenge! In the satisfaction surveys, our online visitors complained that they could not find what they were looking for easily. It is true that we disseminate more than 16,000 statistical products. To clarify the architecture of information on our website and to make searching easier, we carried out a number of qualitative and quantitative tests to help us understand how these different categories of users assess and classify INSEE s statistical products and to determine the most common types of search. We were then able define relevant search criteria and their modalities. Tests on the beta versions of the future site were conducted on users every three months throughout the development process. The site offers new features. From now on, all data and publications are gathered under a single heading, Statistics. Everyone can select the products that interest them via a multicriteria search, with options by topics, geographical levels, types of product or collections. Another advantage is the search engine allows a faceted search, optimising results with the option to sort according to the desired criteria. Finally, everyone can access the entire contents of the site, whatever their mobile device, and the new layout allows for more comfortable navigation. 27

29 Improving its methods and innovating INSEE has to meet a growing social demand to cover subjects of a varied and complex nature. At the same time, statisticians have more and more information available to them, thanks particularly to their use of administrative records but also to the increasing amounts of Big Data produced by our digital society. Data processing techniques are improving, but in order to use these new sources effectively some significant methodological work will be required. Innovation is taking place in all the Directorates at INSEE, and especially in the Statistical Methods Division of the Methodology, Statistical Coordination and International Relations Directorate (DMCSI). In close collaboration with the other Directorates, and also with the Ministerial Statistical Departments, this Division conducts methodological research and experiments in the following areas: Innovative methods in sample selection and postcollection processing for household surveys and business surveys; Development of online data collection: virtually all business surveys are now conducted online, and the process is ongoing for household surveys in the framework of our multimode programme, combining face-to-face interviews, telephone interviews, and online questionnaires; Incorporation of geographical information into the statistical information system, to disseminate statistical data at a very detailed geographical level, but still respecting the rules of confidentiality; Development of indicators to measure the level of social segregation in lower secondary schools, to establish a baseline for the degree of social mix, with support from the Department of Evaluation, Prospects and Performance at the French Ministry of Education Big Data: another area of opportunity for official statistics The Statistical Methods Division is also putting a great deal of work into the possible uses of Big Data some of them being available on the internet. To improve or complete some of the official statistics datasets, it is participating in European initiatives on this issue. However, the use of these new data is a technical and statistical challenge and domain statistician need to have a good understanding of them in order to make judicious methodological choices. Teams at INSEE are therefore specialising in mastering these new technologies, such as web scraping, which consists in harvesting data from websites, machine learning methods, or using the Semantic Web which enables applications to converse with each other. Not forgetting the need to develop among all our statisticians some specific data science skills: datamining, data-architecture, data visualisation, etc. These methods can also be applied to the more traditional areas of statistical production. At the end of November 2016, the Institute organised a seminar entitled, Big Data and official statistics: a question of methods to present the state of the art at the European level and report on experimental work being carried out, especially at INSEE. Among work produced by INSEE was a study on using data from Google Trends to analyse the economic outlook, published in Conjoncture in France in March To date, experts at the Institute have concluded that, for their analyses of the economic outlook, internet data cannot entirely replace the survey data or administrative records that are normally used. However, they do provide some interesting additional insights. An enhanced future price index thanks to scanner data Regarding Big Data, the datasets held by private companies are of great interest. This is especially the case for scanner data held by the major retail distribution enterprises which could be used to calculate the Consumer Price Index (CPI). Their volume is particularly impressive, as it is measured at several gigabytes per day. Indeed, a hypermarket records hundreds of thousands of sales every day, and there are more than 1,000 hypermarkets in France! There are many advantages: it would improve the quality of the CPI by using many more data, it would 28 INSEE 2016 Annual Report

30 reduce production costs by limiting the amount of price collection required in the field, and new indicators could be produced, such as price changes in certain market segments like organic products or quality labels, etc. INSEE has been looking into the use of scanner data since After contacting some major retail outlets, INSEE has been testing the use of scanner data to calculate the CPI since This experimental phase has been useful for consolidating all the data processing and familiarising ourselves with Big Data technology: how to receive these data on a daily basis? How to store them securely? How to process them quickly, as the CPI is calculated every month? The statistical methodology was put to the test as various problems emerged, such as replacing a product in the price index basket during the year. These calculations did not concern all products. In the first instance, INSEE chose to select industrial food products (yogurts, coffee, etc.), household cleaning products (detergents, etc.) and hygiene and beauty products (shampoo, etc.). These represent 17% of household consumption and they are articles that are stable over time, unlike clothing, which changes with each collection. Of course, INSEE will check the quality of the data transmitted. Automatic checking procedures are planned and checks will also be made in the field with surveys. The Law for a Digital Republic, adopted on 7 October 2016, will enable the Institute to continue with the acquisition and processing of these data: it will become possible for official statistics to access data which until now has been held by private companies. INSEE will incorporate scanner data into the regular production of the price index in Mobile phone data to produce statistical indicators? Testimonial from Zbigniew Smoreda, researcher at Orange Labs /SENSE We have worked with INSEE in the framework of a research partnership supported by Eurostat. Our aim is to estimate the potential of mobile phone data to produce statistics. To do this, we compared indicators constructed from mobile phone data with INSEE data. We were very quickly able to work together. Our partners at INSEE are experts in using innovative methods to analyse large databases and our social science department is used to handling data from INSEE. Mobile phone data, providing information on the date, type and place of a communication, are generated automatically by the mobile phone network and we are able to transform them into indicators at low cost. This has tremendous potential, which we hope to demonstrate! From Orange s point of view, the indicators could be marketed, and we could strengthen our analyses with more accurate information on the quality of our data. In the developing countries, where the statistical system is less efficient, they could also improve their knowledge of their populations using these telephone data. Concerning official statistics, collecting our data in real time could mean that indicators could be supplied at more detailed geographical and time levels. For example, our first experiments focused on measuring the definition of territories (major centres or urban centres), according to profiles of the present population. There are also questions on measuring mobility. 29

31 Contributing actively, with the French Official Statistical System, to European and International statistics INSEE and the Ministerial Statistical Departments (SSM) make up the French Official Statistical System (SSP). INSEE is responsible for coordinating the SSP and must therefore interact with the SSMs. Together, they help to consolidate the European Statistical System and participate in consultations required at the international level. The challenge? Sharing methods and know-how to produce high quality statistics, which are comparable at the European and international levels. Sharing quality requirements across the Official Statistical System INSEE guarantees the quality of what it does in all its activities, but it also supports the Ministerial Statistical Departments to ensure that quality procedures are applied systematically across the entire Official Statistical System, based on the European Statistics Code of Practice. Particular emphasis is placed on the coherence and comparability of all the statistics produced, whether in France or at the European level. In 2016, INSEE and the Ministerial Statistical Departments set out quality guidelines to formalise the strengthening of the coordinating role played by the Directors of National Statistical Institutes, as stipulated by Regulation (EC) 223/2009 amended, on European statistics. At the European conference on Quality in Official Statistics, organised by Eurostat, the Statistical Office of the European Union, and the National Statistical Institute of Spain (INE), from 31 May to 3 June 2016, in Madrid, INSEE presented its work on company profiling with a view to improving the relevance of observations on the globalisation of the economy. 1 On behalf of several European partners, INSEE also raised a crucial methodological issue, that of seasonal adjustments, and presenting a new way of assessing the quality of these adjustments. The SOeS, observation and statistical service of the French Ministry of the Environment, Energy and the Sea, focused on its work on improving the building permit database to improve the reliability of short-term analyses of the construction sector. Participating actively and in a coordinated fashion in European groups and projects The Statistical and International Coordination Division organises and coordinates the participation of INSEE and the Ministerial Statistical Departments in the European Statistical System (ESS), chaired by Eurostat. It also contributes to official statistical documents submitted to the Council of the European Union, co-legislator with the European Parliament. As representative of the French Official Statistical System (SSP), it raises any concerns it may have and promotes its methods to our European partners. Indeed, the majority of the production of the SSP falls under European regulations. These define the characteristics of the statistics to be produced in order to ensure their quality and comparability. In addition, certain standards have to be applied: all Relevant and reliable statistics but not always easy to access INSEE undertakes to respect the European Statistics Code of Practice which aims to provide statistical results of high quality. Several criteria are used to assess the quality of these results: relevance, accuracy and reliability, timeliness, coherence and comparability between regions and countries, and accessibility. At the beginning of 2016, INSEE surveyed its users to measure their perception of the quality of its statistical production. Main findings: for 93% of respondents, INSEE s statistics are relevant, i.e. they meet their needs, and for 90%, they are reliable. 79% agree or agree somewhat that these statistics are an accurate reflection of the economic situation in France or that they are easy to understand. Respondents are more critical of their timeliness and punctuality, with 21% thinking that the statistics are not available in time for their needs. Lastly, 25% consider that the statistics are difficult to access. INSEE s new website was launched in November and should ensure that the Institute scores some more points in this area! 1. See section Proviiding details about the productive system 2. See section Disseminating its statistics and studies 30 INSEE 2016 Annual Report

32 types of classifications, defining the statistical units to be surveyed, transmission deadlines, format of data, etc. INSEE is involved in several Eurostat projects on clearly defined topics such as the construction of a European system of interoperable business registers, and making available products and services that are constantly better and more responsive to users needs. European project has support from the National Statistical Institutes of Italy and Lithuania. The aim is to modernise the Tunisian statistical system. On the programme: the legislative and institutional framework of the statistical system, revision of the national accounting system, the development of regional statistics and the dissemination and communication policy. In 2016, two new themed networks involving INSEE emerged. The first concerns Sharing common functionalities in the ESS. Steered by INSEE, this network was launched in January with the aim of developing the architectures of the information systems of Statistical Institutes so that they share a common core of services. In February, a network was launched devoted to exploratory work on using Big Data. The planned research topics include using job offer data available online to produce unemployment statistics, using smart meters to produce statistics on household electricity consumption, or using mobile phone data to measure the population present in a given territory. In addition to INSEE, DARES (statistical department of the Ministry of Labour) is participating in these studies, which are planned to continue until Contributing to international statistics: multilateral relations in cooperation actions At the international level, INSEE collaborates in statistics projects with the United Nations (UN), the International Monetary Fund (IMF), the Organisation for Economic Cooperation and Development (OECD) and the World Bank. It is a member of the UN Statistical Commission, the UN Economic Commission for Europe, located in Geneva, and the OECD Committee on Statistics and Statistical Policy. The Institute also manages an international cooperation programme, mainly with countries in North Africa and Sub-Saharan Africa, and with some non-eu-member countries in Europe. In 2016, INSEE and Statistics Tunisia were twinned. This 2-year The need for comparable statistics INSEE promotes the need to produce statistics that are harmonised at the European level. The challenge is to be able to compare figures from the different countries and inform decision-makers and political and economic players. Comparisons are complex, however, and have to be interpreted with caution. This is what emerged at the conference organised in May 2016 by the National Council for Statistical Information (Conseil national de l information statistique, CNIS) on international comparisons. Also, when INSEE proposes studies focusing on comparisons between European countries, such as Tourism in Europe in Spain and France leading in overnight stays published in July 2016, it sets out the methodological limitations of the exercise by highlighting differences, especially regarding the scope of the surveys. 31

33 The profession of systems integrator The challenge of ensuring that applications are working as they should About one in ten INSEE agents work in the field of computers. They are analysts, developers, systems engineers, networks or security, responsible for providing assistance to users... These skills are essentially distributed across our six national computing centres or services, located in Aix-en-Provence, Lille, Metz, Nantes, Orléans and Paris. The Metz statistical centre was opened in 2011 and the teams there are responsible for IT production. The projects they are working on include management of all of INSEE s applications. Here we focus on the profession of systems integrator, with testimonial from Olivier Kremer. I am part of the unit dealing with the integration and management of applications and software (CIGAL). In the chain of deployment of IT tools and services, my colleagues and I come in at the end. We put in place all of INSEE s computer applications, many of which have been developed internally. The scope of these applications is very wide indeed. They range from everything to do with simple working tools like intranet, the internal directory, to all the systems used to gather, analyse and disseminate data. It s our responsibility to get the applications into production, which means launching and running the survey portals, like the population census portal, or the COLTRANE portal which gradually assembles all the business surveys, the Institute s website, the ESANE system for preparing the annual business statistics, etc. In practical terms, there are four stages. The first, which is often the longest, is to configure the servers on which the applications will run. On average, an application deploys on about ten servers. The second is carrying out tests to identify any possible technical blockages. If these are problems specific to this phase of the process, such as insufficient memory or incompatibility of the equipment, we make the necessary adjustments. If there are technical problems like a programming error, then we inform the IT development teams of our diagnosis, who then take over. When all the tests show that the application is ready to be launched and the decision has been taken to go into production, then we launch the application. Finally, when the application is operational, we monitor it, and make all the updates. We are involved in the project on using scanner data from major retail outlets to calculate the price index. The challenge for us is that we must consolidate the technical infrastructure needed to collect and analyse these Big Data. How do we best distribute the calculation capacities of the servers? What type of technology should we use? So many questions that make this a very interesting project! What do I like about my job? The variety of technical problems, the challenge of ensuring that applications are working as they should and a certain pride in delivering the right tool to each member of the INSEE staff so that they can do their job under the best possible conditions! 32 INSEE 2016 Annual Report

34 INSEE in facts and figures 34 Key figures 36 Head Office organisation chart 38 INSEE: a nationwide presence 39 Workforce by region and by category 40 The men and women at INSEE 41 Training and budget 42 Surveys carried out in

35 Europe and international Key figures INSEE Contact Service 78,000 calls annually to the voic server 40,000 telephone responses 32,000 requests of which 97% receive a response within two business days Price index 200 survey workers 90 staff employed 200,000 price listings collected 29,000 sales outlets 180,000 prices collected every month (SNCF, electricity, etc.) Unemployment 108,000 people surveyed for the Labour Force Survey 113 groups of European experts and 29 groups of international experts involving INSEE 163 international cooperation actions of which 40 % in national accounting and economic outlook INSEE online 32 million visits to insee.fr annually 50,000 Twitter 40,000 subscribers to insee.net actualités Economic outlook 340 Informations Rapides a year 20,000 businesses surveyed every month 1,800 households interviewed every month 34 INSEE 2016 Annual Report

36 Major registers 106,800,739 people registered in the National Directory for the Identification of Natural Persons 45.7 millions voters on electoral lists in France 1.5 million new voters 10 million active enterprises and 11 million active establishments in SIRENE 390 staff working on this Population census 9 million people surveyed 24,500 census agents 450 supervisors 5 million homes 8,128 municipalities of which 7,140 have fewer than 10,000 inhabitants Regions and territories 743 regional studies of which 1/3 conducted in partnership with regional public bodies INSEE in the media 72,000 quotes in print and online 7, 20 0 subjects or reports in the audiovisual press Knowing about businesses 258,000 businesses surveyed 28 surveys fully online Observing society 500,000 households interviewed 865 survey interviewers 16 household surveys 35

37 Head Office organisation chart Executive Committee Information Technology Services Olivier Lefebvre IT Production and Infrastructure Software Development and Projects Alain Bayet General Secretariat Financial Affairs and Planning of Work resources Human Resources Living and Working Conditions Legal Affairs and Litigation Statistical Centre of Metz Coordinator of the executive secretaries of regional offices Coordinator of the managers of regional offices Étienne Traynard Jean-Luc Tavernier Director General Pascal Rivière Internal Audit Unit Sylvie Lagarde Methodology, Statistical Coordination and International Relations Directorate Christel Colin Business Statistics Directorate Statistical Methods Registers, Infrastructures and Structural Statistics Quality Short-Term Statistics Statistical and International Coordination Sectoral Economic Outlooks Regional Office Antilles-Guyane Auvergne-Rhône-Alpes Bourgogne-Franche-Comté Bretagne Centre-Val-de-Loire Corse Grand-Est Hauts-de-France Île-de-France La Réunion-Mayotte Normandie Nouvelle-Aquitaine Occitanie Pays de la Loire Provence-Alpes-Côte d Azur Manager Didier Blaizeau Pascal Oger Patrick Pétour Olivier Biau Yvonne Perrot Alain Tempier Joël Creusat Daniel Huart Marie-Christine Parent Valérie Roux Daniel Brondel Fabienne Le Hellaye Jean-Philippe Grouthier Pascal Seguin Patrick Redor Didier Blanchet Economic Studies and National Accounts Directorate Short-Term Economic Analysis General Economic Studies National Accounts Valérie Albouy* Director General's Office Chantal Cases Demographic and Social Statistics Directorate Demography Employment and Earnings Household Resources and Living Conditions Consumer Price Index and Household Surveys Demographic and Social Studies Françoise Maurel Dissemination and Regional Action Directorate INSEE Info-Service Dissemination Regional Action Library Resources and Archives National Council for Statistical Information (CNIS) Secretariat * replaced in December by Benoît Ourliac 36 INSEE 2016 Annual Report 37