Les indices de concentration géographique à l’épreuve de l’agrégation des données

To characterise the spatial concentration of economic activities, reliable statistical measures are needed. This allows assessment of existing disparities and comparison of concentration levels by sector in time and space. Space is continuous but its discretisation due to spatial grouping of observations at different geographical scales (municipalities, d?partements, regions) can induce a measurement error (Briant et alii, 2010), thus affecting the representation of the concentration. Since it is not always possible to utilise the exact position of the entities, this work proposes to study, from simulated data, the extent to which the most commonly used indices of geographic concentration of activities can be biased by geographical aggregation. We show that index values are sensitive to the geographical scale on which they are calculated and that some indices are more robust than others to geographic aggregation.