We use here different databases in three major fields: economy, demography and policy. The main sources are mentionned in the table. In this part, we briefly describe the main databases used in the analysis of flows across the world.

1. Economic flows

Trade flows

The CHELEM database was developed by the CEPII and provides a 'country vs. country vs. merchandises' matrix for each year since 1967. Unfortunately, some countries are grouped together in this database resulting in 80 blocks of countries before 1990 and 94 after the communist fall. For each block, we have the destination of these flows and their structure in term of merchandises up to 147 different products according to the CITI classification. We have chosen to work on 5 different periods since 1967 at regular intervals. At each time, we have taken into account the average between three years to avoid conjuncture effects (1967-69, 1977-79, 1987-89, 1994-1996 and 2004-2006). The three first periods are before the fall of USSR and we have chosen to analyse one period just and after this major change in the geopolitical situation (1987-1989 and 1994-1996). The period 2004-2006 signs the most recent data available on the CHELEM database.

Because the CHELEM database does not consider all countries, we also use the IMF database on trade. The IMF database is an annual 'country vs. country' matrix which starts from 1950 and includes nearly all countries (160 countries before 1990, 179 after). We analyze the IMF data on the same dates than the CHELEM database. However, before 1990, data for communist countries had to be completed using CHELEM database. IMF database has been completed in two steps for some missing communist countries before 1980. First, we don’t have data for some communist countries but we do have statistics for the rest of the world, including their trade with the missing communist countries. It means that most of the missing links can be completed easily. However, some intra-communist countries links are still missing. Second, for those exceptional missing links, we consider the CHELEM database and apply the share of I to J in the world trade according to CHELEM to the IMF database.

FDI flows

According to FDI, we have built a nearly complete matrix of flows between countries. This matrix is based on UNCTAD FDI flows, completed by OECD, as well as national sources. All the data from these three last sources have been homogenized by using the total amount of FDI of UNCTAD. All the FDI flows have been converted in million of US dollars using the annual average rate of exchange between national currency and US dollar. This work has been realized on two periods of five (1998-2002) and three years (2006-2008) in order to avoid the conjuncture effects and incompleteness. The flow between two countries is the average per year of all the flows during the entire considered period. The result is two rectangular matrixes (150 countries vs. 100 countries in 1998-2002, 180 vs 80 in 2006-2008) of FDI flows for both periods. The matrixes are not square because data are often more complete on outward flows than inward flows. Concerning inward flows, the information is limited to 100 countries in 1998-2002 and 80 in 2006-2008 while for nearly all countries we have information about outward flows. Even if only 15% of all possible links are filled in our matrix, the flows content in the matrixes are equal to about 98% of all the FDI flows in 1998-2002 and 90 per cent in 2006-2008. Of course, some areas such as Africa or Middle East are subject to more uncertainty in terms of origin or destination of flows but they represent very limited parts of world FDI. Also, we have collected some data about the structure of FDI inwards by sector of activity. This matrix has been built with data from the same institutes than the previous one: UNCTAD, OECD and national sources. Only three sectors have finally been taken into account (primary, secondary and tertiary) considering the lack of more precise data for most of the countries. This database contains the distribution per sector for about 100 countries for both periods: 1998-2002 and 2006-2008. As for the origin/destination, sectoral data are the most incomplete for Africa and Middle East.

Financial flows

The financial dataset is a matrix of cross-listing flows giving the number of companies from a country i that are listed on a financial market to a country j. As far as the origin of the firm is concerned, the operational address has been chosen, which enables to get round the tax havens bias and the dummy legal addresses located in Bermuda, British Virgin Island, and so on. The data have been collected from the official listings of 123 stock markets, and more precisely from the cash markets of shares and related products. The data-gathering occurred in late 2007.

2. Political flows: Embassies and Consulates

The consulates matrix is a directed square matrix 206x206 spatial units (mainly states plus entities like Greenland, Cook Islands or Netherlands Antilles) which includes all embassies and consulates sent by a country I to a country j. Information was gathered in 2009 in the framework of the EuroBroadMap project. The full dataset also provides the presence of embassies and accreditations.

3. Migratory flows

Migratory flows are based on the OECD database on the stock of migrants according to their nationalities or their place of birth, depending on the countries. It gathers the migrants from all countries in the OECD countries in 1995 and 2005. In some cases, data have been completed by national sources.