+1 Added to my documents.
Please be aware your selection is temporary depending on your cookies policy.
Remove this selection here

Classifying the emerging economies: Amundi's methodology

Using a fundamental and statistical approach, we establish a classification of the emerging economies that groups similar profiles together. We focus on 20 emerging economies and collect 14 macroeconomic variables on a quarterly basis, which we group into four areas1: (i) macroeconomic outlook, (ii) quality of public balance sheets, (iii) investment and savings, and (iv) vulnerability.

At mid-2015, the classification shown on the right in a dendrogram presents us with six different groups of emerging economies: (i) commodity exporters, (ii) structural surplus economies, (iii) structural deficit economies, (iv) economies with declining current accounts, (v) vulnerable economies, and finally (vi) dissaving economies. It is striking to note at what point the regional approaches show through this classification process.

Indeed, the group of structural surplus economies is composed solely of Emerging Asian economies, specifically China and its main trading partners. The group of economies with declining current accounts is also composed solely of Emerging Asian economies. India and Indonesia have a current account deficit, while Malaysia is seeing its current account deteriorate rapidly. As for the group of structural deficit economies, it is composed solely of Central and Eastern European economies, which are traditionally more in debt than their counterparts in Emerging Asia for example. This regional approach also applies to the group of dissaving economies, composed solely of Latin American commodity-exporting countries. However, we consider it separate from the group of commodity-exporting countries (Chile and Russia) where the economic outlook has worsened rapidly but where investment and savings have remained strong. Finally, the group of vulnerable economies stands out for its relative singularity. Indeed, Turkey, Mexico and South Africa have poor economic outlooks, declining current accounts, and a low rate of savings, as well as little room to manoeuvre in terms of their foreign exchange reserves.

In 2016, the divergence in monetary policies within the developed economies – leading to an appreciation of the dollar – and the weakness of commodity prices could add to uncertainties and exacerbate the regional approaches seen at work in 2015. However, concerns over the Chinese slowdown could fade, and the renminbi’s integration into the SDR slated for the fourth quarter of 2016 could be a necessary, but not sufficient, condition for this hiatus. 2016 is expected to be a pivotal year for the commodity-exporting countries and, more generally, for vulnerable economies.


Amundi’s methodology

We propose a system of scoring based on 14 quarterly macroeconomic variables2 grouped into four categories across a sample of 20 countries considered to be emerging3. Our system of scoring is based on four categories:

  1. Macroeconomic outlook: growth of real GDP and inflation (YoY % change);
  2. Quality of public balance sheets: government debt, government defi cit, gross external debt (% of GDP) and short-term gross external debt (% of gross external debt);
  3. Investment and savings: current balance, change in current balance, gross domestic savings and change in gross domestic savings (% of GDP);
  4. Vulnerability: currency reserves (months of imports), basic balance, trend changes in domestic credit (% of GDP) and M2 money supply (% of FX reserves).

As it happens, the superimposition of all these scores using an average, as is often done, would have very little meaning. To classify emerging countries based on the scores achieved, we have decided to use the Hierarchical Ascendant Classification (HAC) method. This method of automated classification is frequently used in data analysis and offers two advantages: (i) we start off with proximity measures (here, the scores) between the elements (in this instance, emerging countries) that we wish to merge; (ii) one of the results is the dendogram, which is used to graphically depict the iterative merging of data. We can then get a good idea of the appropriate number of classes into which emerging countries can be classified.

The HAC method is simple. We start by calculating the proximity between 20 emerging countries, then we merge two emerging countries by minimising the Euclidean distance4 between the scores, thereby creating a class out of these two emerging countries. We then calculate the proximity between this class and the other 18 emerging countries and again merge the emerging countries by minimising the Euclidean distance. We repeat these steps until all the emerging countries are merged into clusters. These successive mergers yield a binary classification tree: the dendogram.


1 For more details on the methodology applied, see Cross Asset Investment Strategy Monthly June 2014, "An emerging economies typology following last year’s stress episode”.

2 To mitigate the volatility of the data, we use weighted averages. The weighting systemstrengthens the most recent data. The weight of one quarter’s data counts for twice that of the previous quarter. 

3 China, South Korea, India, Indonesia, Malaysia, Philippines, Taiwan, Thailand, Russia,Turkey, Poland, Hungary, Czech Republic, Romania, South Africa, Brazil, Mexico, Chile, Colombia and Peru. 


4 Euclidean distance is defined as follows: 



BEN ABDALLAH Marc-Ali , Senior Analyst, Investment Solutions
MOUSSAVI Julien , Strategy and Economic Research at Amundi
Send by e-mail
Classifying the emerging economies: Amundi's methodology
Was this article helpful?YES
Thank you for your participation.
0 user(s) have answered Yes.