TMF Money Trails: Methodology

Nicolas Kayser-Bril edited this page Aug 14, 2015 · 2 revisions
Clone this wiki locally

Just after publishing The Migrants Files in March 2014, we knew we had to look in detail at the financial cost of the Fortress Europe policies. Later that year, a grant from gave us the excuse we needed to get started.

Finding the amount paid by refugees to facilitators

From a methodological perspective, finding the amounts paid by refugees and migrants to facilitators was the easiest task. We simply had to find an average price for a journey, find the number of people who did this journey, and multiply the two numbers. Things weren't that easy.

First, the number of people who did a specific journey is not known. Frontex does aggregate data by "migration route", which is in effect the number of persons who were detected by local border guards. The journey is only defined by its ending point. Someone traveling from Tunisia to Italy is considered to be on the Central Mediterranean route, but someone traveling from Turkey to Italy, a much longer voyage, is on the same route. More importantly, some figures are impossible to obtain. An irregular border crossing to an island is almost always detected. But refugees and migrants coming by air with forged documents or traveling within Europe are rarely detected and no estimate could be found or made regarding their annual numbers. It's a shame, because that's where the biggest share of the facilitation market probably lies.

We had to focus only on crossings by land and sea. Frontex aggregates data from 2008 on, and Clandestino, an academic project, reconstituted comparable data for previous years.

Now, we had to find the average price of a journey. We pored through hundreds of sources, from news articles to direct testimonies to court documents and private archives. We had approximately 2,500 prices from 2000 to 2015. 1,700 of these came from a dataset of the Spanish Red Cross, who surveyed refugees and migrants who traveled the West African route in 2008. Only 2,000 data points could be associated precisely to a migration route. Over 500 data points were related to longer journeys (e.g Syria to Denmark) and could not be used. Each data point was converted in euros using historical exchange rates and adjusted for inflation using the European CPI, to make sure that an increase we might notice wasn't due to the general price variation.

We had too few data points to carry out an analysis by route and year. We nevertheless computed five-year averages for each route. What was most interesting in the data was what it did not show, namely an increase in the prices paid to facilitators. Some commentators argue that the increase in refugees pushed prices up, but the data does not confirm that: it still costs around 1,000€ to cross from Libya to Italy, a price similar to what was paid in the early 2000's.

The Turkish police ran interviews with refugees and migrants in Turkey and found that 15% were traveling for free, either because of their age or of special deals made with facilitators. The Spanish Red Cross dataset showed that less than 3% of refugees did not pay. We decided to use the largest deflator and multiplied the results of our estimate by .85 to account for non-paying travelers.

Our global estimate of 16 billion paid by refugees and migrants to facilitators is extremely conservative, because it does not include travels by air, travels within Europe or travels from the place of departure to the European border (e.g the crossing from Niger to Libya, which costs around €500). A journey from Syria to Germany can cost 10,000€, but our estimate only focuses on the approximately €2,000 paid to cross into Europe (from Libya to Italy or from Turkey to Greece).

Finding the amount paid by European citizens to prevent refugees and migrants from coming or staying

We first set out to list all the amounts we could get our hands on related to migration in Europe and structure them. We soon realized that the task would be too large and decided to focus exclusively on amounts spent (i.e not the amounts predicted or budgeted) to prevent refugees and migrants from coming or staying in Europe. We did not take into accounts actions related to asylum or to the sheltering of migrants upon arrival.

R&D projects

European research and development projects dedicated to keeping refugees and migrants at bay, even if they represent modest amounts (less than a euro per citizen over the past 15 years), played a key role in framing the issue. We decided to analyze them separately. We listed 39 projects that were representative of the Fortress Europe policies (border patrolling robots, sensors that detect - or smell - migrants, satellites that spot refugees in boats etc.) Using the European Commission financial transparency system and the Cordis database, it was straightforward to match a beneficiary of EU funds to a project. We then had to find the main shareholder of the beneficiaries. We could not do this systematically on the 500 beneficiaries of EU research funds, we had to focus on the companies that were part of the Group of Personalities that set up the framework for these research projects (mostly arms manufacturers). We only focused on amounts disbursed by the European Commission, not on the total cost of projects, because we know that the amounts said to be invested by private companies in European projects are more often than not simple accounting tricks (i.e a company says it co-finances a project by 50% but really works half of the time it budgeted).


As we gathered amounts paid for by European governments, we realized that deportations were the biggest expenditure incurred because of the Fortress Europe policies. We also noticed that governments did not keep track of this type of expenditure. Some of the Interior ministries kept tabs on the expenditures, but most of the data was limited in scope (most often, only transport costs were available) and in time. Interestingly, when members of parliament in France and Italy looked into the costs of deportations, their estimates were two to four times higher than those from the police. This is due mostly to the fact that the police never counts the salaries of the officers as part of the cost of deportations. However, the personal dedicated to manning detention centers (these jails are rarely operated by the ministries of Justice) or accompanying deportees to Pakistan or Nigeria (a two-day trip, at least) does weight heavily on the public purse.

No institution had tried to estimate the costs of deportations at the European level. We decided to give it a shot. Just like we did for the amount paid to facilitators, we decided to find the average cost of a single deportation based on the elements some police forces had published or that we obtained through freedom of information requests. In total we gathered close to 100 data points (a expense for a country and a year), from 10 countries.

Unfortunately for us, this was the easy part. Not only do our governments not know how much money they spent to deport men and women, they don't even know how many of them are deported. The laws restricting the right for someone to simply be on European territories are so vast that you have 10 different categories of "returns":

  1. Deportation of asylum seekers who have been turned down
  2. Deportation of third-country nationals (not asylum seekers)
  3. Deportation of EU nationals who are "abusing the social system" (e.g Roma)
  4. Deportation of foreign nationals under Dublin regulation
  5. Deportation of foreign nationals who are legally residing in a member state but don't have the right to be in another member state, e.g an asylum seeker in Frankfurt / Oder who crosses the bridge to Slubice in neighboring Poland to buy cigarettes and is arrested by the police.
  6. Voluntary return of third-country nationals to another member state under the Dublin regulation
  7. Voluntary return of asylum seekers who are still in the asylum process with assistance from the state
  8. Voluntary return of asylum seekers who are still in the asylum process without assistance from the state
  9. Voluntary return of asylum seekers who have been turned down, without assistance from the state
  10. Voluntary return without an order to leave (e.g leaving a country with an expired visa)

All of these are in use by one member state or another. Governments tend to put these figures together if they want to look tough on foreigners, others will focus on a much narrower subset. Sometimes, the police holds data for one category and the asylum authorities the data for another one. More interestingly, when we asked for details about a data set we received, not a single administration could tell us with certainty what categories of returns were counted. We unfortunately did not have the resources to ask every single European administration for the detailed data and methodology around "returns".

To work with a data set that had comparable data in time and space, we used Eurostat's count of "people returned following an order to leave", which includes categories 1, 2 and 9 (or should include - Eurostat was unable to answer our numerous requests for clarification), starting in 2008. Luckily, a group of academics did reconstitute time series starting in 2000 using the same methodology for the most populous countries part to the Dublin regulation.

We then had a fairly coherent number for most years and most countries. We could derive a cost per "return following an order to leave". Even if this amount was virtual (it did not correspond to any reality, as a person returned following an order to leave could be a deportee for which a government pays 20,000€ or a person leaving voluntarily), is let us derive an total cost of deportations per year and per country. The enormous assumption we made here is that the ratio between categories 1, 2 and 9 was constant across time and space. Based on the numbers of actual deportations that we obtained, this is not always true. Southern countries used to have more voluntary returns in the early 2000's, compared to their northern neighbors. This means that our global estimate is under-representing the amounts spent by northern governments.

Despite the problems with this methodology, the amounts that we subsequently collected from governments matched our estimates. We do not know the precise amount European citizens pay to deport non-European citizens, but we know for sure that is it, at a minimum, around 1 billion euros a year.

Other costs

The Fortress Europe policies imply lots of other expenditures, such as databases for fingerprints of asylum seekers, coordination centers, satellites, speedboats, walls etc. Here, the financial transparency system of the European Commission was almost useless. Most of the amounts spent are co-financed by member states, but member states very rarely mention the precise amount paid for by the European Union. To avoid double counting, we always considered amounts from member states only (unless the expense was a European project, such as Frontex). We are lacking a large chunk of the European Return Fund, for instance, set up to assist member states in deporting refugees and migrants.

We did not include any amount that could have been considered as "normal" running costs for border guards. We focused only on expenditures designed to prevent migrants and refugees from entering (as opposed to measures against drug traffickers, for instance).