Draft:The Walker Gravity Model

DRAFT

The Walker Gravity Model of Money Laundering was developed by (now) Professor John Walker after joining @The Australian Institute of Criminology in 1979.

Walker began to focus on the economics of crime which, at the time, were rarely studied owing to the lack of hard data. Law enforcement agencies rarely compiled data on the proceeds of crime or the costs of crime to victims or to governments – such data were not relevant to their task of prosecuting offenders. While economists studied “the shadow or black economy” [1], which involves economic activity which is illegal or untaxed, their focus was never on the criminal economy. There were, therefore, no data, other than the costs of policing, courts and corrections systems themselves, upon which governments could prioritise activities in the criminal justice system. Thus, while governments use cost-benefit analyses to prioritise investments in health, education, defence, science etc there was no way to prioritise the activities of the criminal justice system other than the strident calls of community groups.

While on secondment to the British Home Office Research and Planning Unit, Walker was invited to join discussions which eventually led to the establishment of the International Crime Victims Survey which, in its first round, covered the U.K., the Netherlands, Switzerland, Australia, Canada and the USA [2]. A similar survey was later conducted to compile data on crimes against businesses [3]. At Walker’s suggestion questions were included such as the extent of property losses or costs incurred due to injury or hospitalisation. The responses in the Australian surveys were later used by Walker in his 1992 publication on the cost of crime in Australia [4].

A project in the early 1990s by The Financial Action Task Force to estimate the amount of proceeds of crime in the global economy concentrated almost entirely on the proceeds of illegal drugs. Walker was invited by The Australian Transactions Reports and Analysis Centre (AUSTRAC) to follow up his work on the costs of crime to see if it were possible to quantify the extent of money laundering in and through Australia.

Studying the few research reports that addressed the issue, Walker found that they all attempted to quantify money laundering on the basis of the suspicious transactions reports reported by financial institutions to their country’s counter-money laundering agency. Recognising the inherent problems of such data – including the high probability of false positives and false negatives – he looked at the data obtained from his crime victimisation surveys.

The stepwise model Walker eventually developed for Australia was:

   1. How much crime, of each type, is committed?

   2. How much profit comes from the proceeds of those crimes?

   3. How much of the profit would be laundered, as opposed to simply spent or hidden? and

   4. Where was it likely to be laundered – e.g in Australia or overseas?

Later, a fifth step was added, noting the damage that laundered money can do to an economy:

   5. How much harm does it do there? 

MORE FRAUD THAN DRUGS

Walker quickly established that estimates of the proceeds of fraud in Australia significantly exceeded those of illicit drugs – a result supported by discussions with criminal intelligence agencies. In the absence of hard data to address step 3, Walker conducted a small survey of professionals and practitioners, including senior police, criminologists and finance professionals, noting that – logically – when the proceeds of crime are obtained in relatively small amounts, as in, for example, burglary and theft, the proportion laundered would be vanishingly small, while for “big-money” crimes such as illicit drug trading and major frauds, the proportion would be much higher. Consensus figures were reached from the survey upon which estimates of the quantity of laundered money could be based. No-one was under any illusion that these estimates were “accurate”, but the important thing was that they had credibility among professionals and politicians.

Most importantly, it had credibility amongst economists, as demonstrated by the enthusiastic reception to Walker’s appearance at the 1995 Cambridge International Symposium on Economic Crime, with some attendees having obtained copies of the Austrac report [5]. An updated version of the conclusions was published by the Australian Institute of Criminology in 2004 using the same methodology. [6]

On 10th March 1998, a Note by the FATF Secretariat proposed to hold an International Experts Meeting on Estimating the Magnitude of Money Laundering, “At the meeting, the experts would be expected to review prior survey approaches undertaken by Australia, the United Kingdom, and the United States in order to develop a survey methodology for measuring criminal proceeds for money laundering” and “recommended that the FATF agree upon a method of estimating (through surveys) criminal proceeds and the amount of such proceeds available for money laundering in as many countries as possible, starting with FATF member countries”. Walker was invited to present his proposals at the meeting. Considerable criticism was made on the basis that “experts” still considered that the illicit drugs trade far outweighed any other crime type, and that the inclusion of fraud would result in inflated estimates of money laundering.

In 1999, Walker published his seminal article in the Journal of Money Laundering Control – “How Big is Global Money Laundering?” [7]. The summary Abstract states:

The purpose of this paper is to describe a comparatively simple crime‐economic model, constructed from readily available international databases, that closely ‘predicts’ a range of such expert assessments, and appears to offer a framework for determining and monitoring the size of money‐laundering flows around the world. Further research is required, but the exercise of constructing the model has identified a number of gaps in existing knowledge which could readily be addressed by well‐targeted research. Initial output from the model suggests a global money‐laundering total of S2.85bn per year, heavily concentrated in Europe and North America.

As of October 2024 it had been downloaded almost 3000 times.

The Attractiveness to Money Launderers

The article described a naïve form of index of “Attractiveness to Money Launders”, using the algebraic form:

Attractiveness to Money Launderers = [GNP/capita] x [3 x BankSecrecy + GovAttitude + SWIFTmember — 3 x Conflict — Corruption + 15]

where GNP per capita is measured in US$, BankSecrecy is a scale from 0 (no secrecy laws) to 5 (bank secrecy laws enforced), GovAttitude is a scale from 0 (govern­ment anti-laundering) to 4 (tolerant to laundering), SWlFTmember is 0 for non-member countries and 1 for members of the SWIFT international fund transfer network, Conflict is a scale from 0 (no conflict situation) to 4 (conflict situation exists), Corruption is the modified Transparency International Index (1 = low, 5 = high corruption), and the constant '15' is included to ensure that all scores are greater than zero.

While this version of the Attractiveness Index has been heavily criticised, the concept has become mainstream in recent years. The problem for critics was that it produced rankings that were by and large, extremely credible, for example: 1. Luxembourg, score 686; 2. USA, 634; 3. Switzerland 617; 4. Cayman Islands 600; 5. Austria 497; 6. Netherlands 476; 7. Liechtenstein 466; 8. Vatican City 449; 9. UK 439; 10. Singapore 429.

The inclusion of Vatican City high in the list raised eyebrows, but Walker’s presentation of the model to a crime symposium in Rome coincided with the apparent suicide of a Vatican Banker in London. The inclusion of the USA and the UK at the top of the list generated indignation in those financial centres, but nothing that would refute the findings – money laundering is much easier to hide in a busy financial market. The inclusion of the Netherlands in the list brought initial denial in the Dutch Government, who asked leading economist, Prof Brigitte Unger, to conduct research in that country.

Critics continued to focus on the extremely high figure placed on global money laundering, and particularly the fraud component, but a 2005 book by eminent authority on financial crime, Raymond Baker, entitled “Capitalism’s Achilles Heel” [8] came to very similar quantitative conclusions based on an entirely different methodology, and again emphasised the extent of fraud and corruption around the world.

In 2006, in collaboration with the Australian National University, Unger was introduced, by co-author Dr. Greg Rawlings, to the Walker Gravity Model, devoting a whole chapter to examining the model in detail, and then using it to estimate money laundering in the Netherlands. “Measuring Money Laundering for Australia and the Netherlands - The Scale and Impacts of Money Laundering” [9]. While she concludes that Walker’s results overestimated money laundering in the Netherlands, she found that the model itself had sufficient validity in economic theory to deserve serious attention and improvement.

She wrote: Walker (1995 and 1998) was a pioneer who attempted to measure money laundering worldwide, using an ad hoc equation. His model still looks like the most promising one. Though heavily criticized as ‘ad hoc’, as lacking a solid theoretical or methodological background, and as overestimating the amount of money laundered, it still provides a relevant point of departure for further improving the measurement of money laundering. The Walker Model examines two different aspects of the money laundering process. First, it scrutinizes money generated for laundering per country. Second, it examines flows of generated money from one country to another. Money can be laundered in the country in which it was generated or sent to another country for laundering. An important point within this model is that as soon as money has travelled (flowed) at least once, it is ‘white washed’, or laundered.

This last point answers the question of whether money laundering must be measured at the various steps in the process – that approach would, of course, result in double – or multiple-counting.

In November 2007, she invited Walker and 65 experts including Baker, both critical and supportive of the Model, to Utrecht for the “Tackling Money Laundering” Conference organized by the Tjalling C. Koopmans Institute. While some remained to be convinced, Unger and others resolved to work particularly on the data deficiencies and the structure of the Attractiveness Index. Unger recognised it as an application of the Leontief model commonly used in international trade analysis.

In 2009 Unger and Walker co-wrote “Measuring Global Money Laundering: ”The Walker Gravity Model” [10], in which they concluded:

Using triangulation, we show that the original Walker model estimates are compatible with recent findings on money laundering. Once the scale of money laundering is known, also its macroeconomic effects and the impact of regulation and law enforcement effects on money laundering and transnational crime can be measured. Furthermore, given the great success of the micro foundation of the gravity model in international trade theory, we are optimistic that an economics of crime micro-foundation of the “Walker model” will follow soon.

The use of a Walker-derived model in the EU.

In 2008, the University of Utrecht was commissioned by the European Commission’s Directorate for Justice, Freedom and Security to develop a methodology for assessing the economic and legal effectiveness of anti-money laundering and combating terrorist financing policy (AML/CTF policy) in the 27 EU Member States. Initial steps included developing a methodology on how to do a threat analysis, how to evaluate legal effectiveness and how to do a cost-benefit analysis. The project, named “ECOLEF” [11] was led by Unger and assisted by Walker, and reported in February 2013. The brief required the researchers firstly to estimate the amount of money that might, in the absence of AML activities, be laundered in the EU. The Report states that “A modified Walker-style gravity-type model can be used to express those distance, trade, language and other factors that make some countries more attractive to launderers than other countries, and produce estimates of the proportion of total money available for laundering that MIGHT be laundered in each EU country. In a perfect world, the results of such analysis could be compared with actual data, but such data are not available, and probably never will be. The approach we are forced to take is to assess the credibility of the results, by reference to known facts.”

The Model begins by computing the Intrinsic Attractiveness for each country, which is comprised of measures of GDP per capita, Finance Services as a proportion of Export Trade, and a “Combined Money laundering Capacity Index”. This index reflects both the “capacity” and the “willingness” of a country’s finance sector to launder money, and is compiled from data on numbers of Transnational Affiliates per population, the Financial Secrecy Index, and an FATF Compliance Index, which itself is compiled from the Mutual Evaluations provided by the FATF, and the list of “Major Money Laundering Countries” published by the United States Department of State. The algebra of the model to determine each country’s attractiveness to launderers was:

Attractiveness of Country i = GDP/Capitai * Finance Services % of GDPim * Combined ML Capacity Indexin

where m and n are empirically derived constants.

Those countries with high GDPs sophisticated banking systems and high levels of international trade head the list of countries according to their intrinsic attractiveness to money launderers. The contrast between the scores based on the FATF Compliance Index and those based on the other indices suggests that mere “book” compliance with the FATF’s recommendations – as assessed by the Mutual Evaluations teams – is a very poor indicator of actual outcomes.

The model then computes an index of “Affinity of Country i to Country j”, which is a highly modified version of the Distance component of standard gravity models. Affinity of Country i to Country j = Attractiveness of Countryjp * Tradelinksq * (1+Language Loading) * (1 + Cultural Loading) (Distance between Country I and Country j)r where, for i=j, Tradelinks=1. Language Loading, Cultural Loading, p q and r are empirically determined constants.

The Model then proceeds to the final step, in which the estimates of money being sent for laundering from each country to every other country, are generated. Money Laundered from Country i to Country j = Total Money generated in Country i for Laundering * Affinity of Country i to Country j Σk(Affinity of Country i to Country k)

The Model was used in two quite distinct ways. Firstly, it was used to estimate the total Threat posed to each EU country by money launderers around the world. Secondly, it was used to estimate the amounts of money actually laundered in each EU country. The ratio of the two is suggestive of a measure of the effectiveness of AML/CTF in each country.

In 2011, the Walker Model was employed in a project conducted by the United Nations Office on Drugs and Crime – “Estimating illicit financial flows resulting from drug trafficking and other transnational organized crimes” [12]. The Preface notes that: UNODC’s research report, “Estimating illicit financial flows resulting from drug trafficking and other transnational organized crimes”, attempts to shed light on the total amounts likely to be laundered across the globe, as well as the potential attractiveness of various locations to those who launder money. As with all such reports, however, the final monetary estimates are to be treated with caution. Further research and more systematic collection of data on this topic are clearly required.

Prior to this report, perhaps the most widely quoted figure for the extent of money-laundering was the IMF’s ‘consensus range’ of between 2-5 per cent of global GDP, made public in 1998. A study-of-studies, or meta-analysis, conducted for this report, suggests that all criminal proceeds are likely to have amounted to some 3.6 per cent of GDP (2.3 - 5.5 per cent) or around US$2.1 trillion in 2009. The resulting best estimate of the amounts available for money-laundering would be within the IMF’s original ‘consensus range’, equivalent to some 2.7 per cent of global GDP (2.1 – 4 per cent) or US$1.6 trillion in 2009. From this figure, money flows related to transnational organized crime activities represent the equivalent of some 1.5 per cent of global GDP, 70 per cent of which would have been available for laundering through the financial system. The largest income for transnational organized crime seems to come from illicit drugs, accounting for a fifth of all crime proceeds.

The analysis in the report closely followed the Walker-Unger models. However, the study depended heavily on the UNODC’s holdings of data, which are far more detailed in the area of drug trafficking than any other form of transnational organised crime, reflecting its overwhelming focus and investment in research on trafficking of illicit drugs.

Furthermore, the focus on “organised” crime largely ignored crimes by entrepreneurs, business or politicians.

It is likely, therefore, that money generated and laundered from fraud and corruption is greatly underestimated in that study.

In the UNODC-UNCTAD Expert Consultation on the SDG Indicator on Illicit financial flows in Vienna in 2017, Walker advocated greater use of UNCTAD and other data on trade in financial services to address this issue.

The versatility of the Walker Gravity Model has been shown in various types of analyses in many countries, for example to quantify outflows of laundered money from the Chinese economy [13] and to measure the impacts of organised crime on legitimate business in Europe [14].

In this paper, the term "billion" is taken as 1,000 million and "trillion" as one million million.