Estimer la somme d'argent (document en anglais)

Probably the most cited and most controversial estimations of the amount of money

laundering are based on the Walker-model. Walker (1995 and 1999) is a pioneer in the

field estimating the amount of money laundering per country using a type of gravity

model. His estimation procedure basically consists of four main steps:

1. Identify all the relevant crime data for all countries in the world

2. Estimate how much money is made with these crimes

3. Estimate which percentage of these proceeds are laundered

4. Estimate how this money is allocated over the world

Crime data is never a perfect representation of the actual amount of crime in a

country, but its use (especially when combined with crime victim surveys, as done by

Walker) seems to be generally accepted. Because crime data is not available for all

countries in the world, Walker (1999) uses regional averages to fill the gaps. Once crime

data for all countries in the world is available, we can in step 2 multiply them with the

average proceeds of these crimes to come to the overall estimate of money made by

crime in all countries in the world. To come to the total amount of money that needs of

laundering in step 3, Walker estimates how many of these proceeds are in need of

laundering by using a percentage for each type of crime.9 This procedure basically gives

the amount of money that is in need of laundering in the world with the following

formula:

Total amount of money laundering = number of crimes * proceeds per crime * %

laundered

Step 4 of the Walker-model is a type of gravity equation to estimate how this money

ready for laundering is spread over the world. For over decades the gravity model has

been successfully applied to flows of the most widely varying types, such as migration,

9 Note that step 2 and 3 are basically merged in Walker (1999) by using earlier obtained estimates (in

Walker, 1995) on the average amount of money laundered per recorded crime. This also means that he

circumvents the issue in step 1 that recorded crimes do not represent the amount of actual crimes.

[13]

buyers distributed across shopping centres, recreational traffic, commuting, patient flows

to hospitals and interregional as well as international trade. (see Chapter 5 of this

dissertation) The assumption of Walker is that money laundering flows can be explained

with the following formulas:

2

, , (( )* ) (Distance) ij i i j j j j i j

F  F  GNP Population Attractiveness

3 3 j j j j j Attractiveness  BS GA  Swift  CF 15 j CR 

where GNPj is GNP per capita, Fij is the amount of money laundering flowing from

country i to country j, BS is banking secrecy and GA government attitude towards money

laundering, Swift indicates countries with financial institutions that are member of Swift,

CF refers to conflict and CR stands for corruption. 15 is added to the attractiveness to

avoid negative values.

Walker (1995 and 1999) assumes which country characteristics attract money

launderers and the importance of these characteristics; there is no empirical

underpinning why for instance Bank Secrecy is three times as important as Corruption.

Apart from choosing the relevant characteristics and their weights, Walker also makes

important assumptions on the form of the formula. Walker’s form deviates from a

standard gravity equation in the sense that instead of a complete multiplicative model,

Walker chooses a combination of multiplication and addition (see Chapter 5 of this

dissertation for more details).

Note that step 4 of the Walker model is needed to estimate the amount of money

laundering per country, but that for the worldwide estimate step 4 is unnecessary.

Unger (2006 and 2007) adapts the Walker-model and applies it to the Netherlands.

She adapts the original Walker-model by a) replacing the distance simply measured in

kilometres by a cultural distance, which includes three ‘cultural’ factors: sharing a

common language, having colonial ties or being major trading partners; b) including two

new factors to the attractiveness indicator, being the ‘membership of the Egmont group’,

a cooperation of national Financial Intelligence Units fighting money laundering, and

[14]

‘financial sector size’ (measured as deposits), as a proxy for the fact that extended

financial markets make it easier to launder criminal proceeds; c) using the distance

between the countries in the attractiveness factor, instead of its square, because

empirical findings show most gravity equations for trade come up with an estimate for

the coefficient of distance of around -0.9 (Helliwell, 2000), d) replacing GNP per capita by

GDP

...