A Founding Decision in a Finance Law
The National Institute of Statistics and Economic Studies, known by its French abbreviation Insee, was established under Article 32 of Law No. 46-854 of April 27, 1946. It belonged to the Ministry of the National Economy and covered metropolitan France and its then overseas territories. The law required centralization of statistical, economic-research and documentation services, leaving specific powers and coordination procedures to administrative regulations under Article 33.
First director Francis-Louis Closon later recalled that National Economy Minister André Philip and Secretary of State Albert Gazier advanced the plan, seeking a stronger statistical and research institution to support postwar public decisions. Establishment was thus both administrative reorganization and part of rebuilding economic knowledge, not an expansion of a university laboratory.
Building on Existing Personnel and Equipment
Decree No. 46-1432 of June 1946 listed the bodies incorporated: the National Statistical Service, the economic-conjuncture research organization, relevant planning and documentation services in the ministry, and foreign-economy services. Their powers and working resources were transferred together. The decree also defined a framework of central departments, regional offices, ministry-linked units and an applied-training school.
In his 1971 retrospective, Closon noted that many staff in Paris and provincial offices were skilled in administration, records and mechanical tabulation, while professional statisticians and economists were comparatively few. Combining these traditions required training, division of work and continuing coordination. Gabriel Chevry's 1948 paper already recorded a Paris central directorate, a Paris regional office and seventeen provincial offices, showing the practical importance of the local network.
Statistics Production Alongside Economic Research
The founding decree did not confine the institute to totaling figures. Article 1 required compiling and updating population and property statistics, observing economic conditions, undertaking statistical and economic research, disseminating results and training specialists. Sample surveys were explicitly included. Recording conditions, analyzing change and building technical capability formed one institutional design.
The coordination responsibility was especially concrete: departments could retain their own data, but methods, classifications and codes had to connect. Article 2 authorized relations with counterpart institutions abroad and participation in professional international conferences. Both domestic interdepartmental cooperation and international methodological exchange had an institutional basis; “centralization” did not mean every survey could be performed only by one office.
References: [2]
Funding Arrangements and Early Capacity
The institute depended on the state budget. Article 33 of the founding law permitted the economic and finance authorities to transfer or cancel appropriations affected by reorganization through decrees and eliminate unneeded posts. Funding and staffing therefore had to adjust with service reorganization; the law itself supplied no fixed total that could be called a “founding fund”.
Gaël de Peretti and Béatrice Touchelay's institutional history found that early staffing and budget limits affected survey expansion and the audience for results. Closon's economic “green bulletins” primarily served a small group of governmental and economic decision-makers and were strictly controlled in distribution. The dual objectives of research and dissemination did not mean an open statistical service for the entire public existed from the start.
Making an Establishment Register a Cooperative Tool
Chevry's 1948 paper described an early priority: organizing the register of industrial and commercial establishments. The institute checked old records against returns from the March 10, 1946 census, separating confirmed establishments from unverified ones, and sampled blocks in some large cities for building-by-building visits. The census predated Insee's legal creation; the new institution inherited the subsequent processing.
In 1947, social-insurance bodies adopted the institute's establishment identifiers, and about 650,000 employer records supplied another round of checks. Business-change notices and municipal investigations filled further gaps. Cooperation depended on distinguishing the same enterprise from an establishment at the same location so records could connect across departments. Chevry also warned against directly extrapolating omission rates found in large-city samples to all France.
References: [5]
From Administrative Registration to an Employment Sample Survey
The employment survey launched in 1950 demonstrated another research path. Between population censuses, job-seeker registrations and unemployment-benefit records could not fully describe employment: registered job seekers might still have jobs, while unemployed people not claiming benefits might be missed. Sampling households supplied evidence independent of those records to supplement estimates.
Dominique Goux's study notes that designers initially wanted quarterly surveys, but could conduct them only twice yearly; resource shortages reduced frequency to annually in 1954. Early questionnaires and questioning techniques were repeatedly tested. Questions added in 1953, including father's occupation, also supplied material for examining social mobility. Frequency, questionnaire scope and funding jointly shaped results; the later continuous-survey system cannot be projected onto 1950.
References: [6]
Data Cooperation Had to Address Confidentiality and Trust
Centralizing and linking records also prompted concerns. Article 7 of the 1946 decree already required staff to take an oath and observe professional confidentiality. Later debate over population records as possible instruments of oppression showed that technical efficiency did not automatically create public trust. The 1951 law on statistical obligations, coordination and secrecy placed survey organization and information protection in a clearer framework.
The 2023 study emphasizes that the early institute first sought a position among government and specialist departments, only gradually expanding ties with other users. Its founding experience shows that an effective statistical system requires more than mathematics or tabulating machines: it must assign responsibilities among data providers, enumerators, analysts and user departments and explain which data may be used for which purposes.