Comments on IMF Working Paper on Estimating the Size of Shadow Economy in Europe

This post is a continuation of a series of the difficulties of modelling illicit financial activity and the need to understand that the results of such models are estimates not facts when we use those results.

Earlier posts here. Transparency International’s Corruption Perceptions Index, Difficulties in Modelling Illicit Financial Flows, The Origin of the Estimate of Global Money Laundering.

On 13 Dec 2019 the IMF published Working Paper 19/278 Explaining the Shadow Economy in Europe: Size, Causes and Policy Options written by B. Kelmanson, K. Kirabaeva, L. Medina, B. Mircheva and J. Weiss.

Before discussing this publication, just a note that IMF WPs do not represent the IMF’s official position, but report on research in progress and are designed to elicit comment and feedback.

While the WP is “rich” in content, I’d like to highlight a few points for comment. Because as usual AA has an axe to grind.

Defining and Measuring the “Shadow Economy”  (Pages 5-6)

There are several key take-ways here.

1. Different Definitions - There are different definitions of the “shadow economy”.  And so, it’s important to know which definition is being used in an estimate.  It’s like Transparency International’s Corruption Perception Index.  Are we measuring all corruption in a country or just a subset?
2. Estimates Not Measurement - What is actually going on is not measurement in a formal sense.  But an attempt to estimate the size. A critical difference.  One can directly measure AA’s weight.  Or the distance from Bayt Meri to Beirut. But one can’t directly measure this or that person’s intelligence.  One has to estimate it.
3. There is No Single Infallible Method for Such Estimations – in the authors’ words “There is no ideal or leading method to measure the shadow economy, each of them have some conceptual or practical strengths and weaknesses. The choice of the methodology can be governed by data availability, or the research objectives. Multiple methods can be employed to improve accuracy of the estimations.”
4. Each Model has its Limitations - In discussing the model they used (MIMIC), the authors state: “The shortcomings of this method include sensitivity to changes in data and specifications, the sample used, calibration procedures, and starting values (Breusch 2005).”

Results – False Precision

1. The tables of results are on page 25 and 26.
2. The results appear very precise.
3. The authors have come up with results to the tenth of one percent.  Or in non percentage terms .001.  That is some pretty fine parsing for estimates of the size of something that is unknown.
4. Not only that but they have been able to order the sizes of the shadow economies among various countries.
5. In the first table (page 25), France is 15.3% while in Germany is at  15.9%.  In the second table, the scores are 15.0% and 16.7% respectively.
6. Is this precision really possible?
7. What about the caveats about models and estimates in the previous section?
8. Shouldn’t we expect to see less precision?
9. Use of ranges?  Grouping of countries into similar baskets.  For example, Germany and France have roughly the same size shadow economies.
10. AA has his axe out now. I made the argument in an earlier post on Transparency International’s CPI that when valuing the credit worthiness of an issuer rating agencies place similar firms in broad categories, AAA, AA, A etc.  They don’t parse creditworthiness of individual firms within a category. And since I have a habit of repeating myself made the same arguments in another post about modelling illicit financial flows.
11. Similarly in valuing firms, stock analysts come up with a range for the value of stock, not a single point estimate.
12. The point I want to make yet again is that we need to treat results from such modelling efforts as directional not locational. They are not precise but only give an indication of the real state of affairs. And when we use them, we need to keep that in mind.

Implications: Likely Underreporting of “Total” GDP

Ignoring the shadow economy means that we are likely underreporting the total or “true” GDP of a country – that is the GDP from its formal sector and that from its informal or shadow sector.  

Where the shadow economy is relatively small, this probably doesn't make much of a difference. 

But if the authors’ estimates that the shadow economy in the CIS countries is around 40% of formal GDP (page 7) are correct, then typical characterizations of CIS “dismal” economic performance based on only formal GDP are probably less true than they appear.

This does not of course “excuse” the fact that a good portion of this activity is taking place “off the books”.