Diese Präsentation wurde erfolgreich gemeldet.
Die SlideShare-Präsentation wird heruntergeladen. ×

CFAR-m Presentation English

Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige

Hier ansehen

1 von 52 Anzeige

Weitere Verwandte Inhalte

Andere mochten auch (20)

Ähnlich wie CFAR-m Presentation English (20)

Anzeige

Aktuellste (20)

Anzeige

CFAR-m Presentation English

  1. 1. Cliquez pour modifier le style des sous-titres du masque
  2. 2. Summary 1. C o m po s ite indic a to rs a nd ra nk ing 2. T ra ditio na l m etho ds fo r c o ns truc ting c o m po s ite indic a tors a nd their w ea k nes s 3. T he C FA R -m a lg o rithm 4. S om e ex a m ples of im plem enting C FA R -m 2
  3. 3. 1. Composite indicators and ranking What is it ?… A C o m po s ite I ndic a tor is a n a g reg a te index tha t s um m a rizes a la rg e a m o unt o f info rm a tion g iven by s ing le indic a tors . C om pos ite I ndic a to rs a re inc rea s ing ly being us ed to m ea s ure m ultidim ens iona l perfo rm a nc e a nd to ra nk c ountries , firm s , c lients , ins titutions , etc ., in m a ny fields , s uc h a s :  Competitivity (Global Competitivity Index - FEM)  Country risk (ICRG-PRS group) Cliquez pour modifier le style des sous-titres du  masque Well-being (Health System Achievement Index-WHO)  Environment (Environmental Sustainability Index- WEF)  Governance (The Corruption Perceptions Index - Transparency International)  Innovation (Technology Achievement Index- UN) 3
  4. 4. 1. Composite indicators and ranking A real interest … D em a nd for, a nd produc tion o f C om pos ite I ndic a to rs a re ra pidly g ro w ing . 2 rea s o ns , ba s ic a lly : Google search results for "composite indicators" 1- C om plex ity o f m o dern ec ono m y : jus t one, o r a s et of s ing le indic a tors is not enoug h a ny m o re. 2- D evelopm ent o f I C T s : it m ea ns tha t a hug e m a s s o f inform a tio n ha s to be pro c es s ed 4
  5. 5. 2. Traditional methods for constructing composite indicators and their weakness A great number … Most used weighting schema in aggregation methods: W eig hts ba s ed o n s ta tis tic a l m o dels  E qua l w eig hts  D a ta E nvelopm ent A na lys is (D E A )  P rinc ipa l C o m po nent A na lys is (P C A )  U no bs erved C o m po nents M o dels (U C M ) Cliquez pour modifier le style des sous-titres du W eig hts ba s ed o n ex perts ’ o pinio ns masque  B udg et a llo c a tio n W eig hts ba s ed o n the s ta tis tic a l qua lity o f da ta  S ta nda rd devia tion 5
  6. 6. 2. Traditional methods for constructing composite indicators and their weakness Many problems… Drawbacks of traditional methods : They are exogenous They are linear They lose information They offer a no posle style des sous-titres du Cliquez pour modifier itive capability to assist masque decision-making processes 6
  7. 7. 3. The CFAR-m algorithm Our solution … A n orig ina l m ethod ba s ed on a rtific ia l intellig enc e for the c o ns truc tion of c om pos ite indic a to rs tha t a llow s to perform releva nt ra nk ing . Innovation T he w eig hting s c hem a o f s ing le indic a tors is g enera ted thro ug h a lea rning proc es s , from inform a tiona l c ontent of the va ria bles them s elves a nd their interna l dyna m ic s . 7
  8. 8. 3. The CFAR-m algorithm Our solution … C -FA R m w ork s in three s ta g es tha t a re s truc tura lly c o m bined : S ta g e 1 : Firstly, it carries out a c la s s ific a tio n (self-organization) of objects (records, points, cases, samples, entities, or instances) through a lea rning pro c es s that takes into account interactions between the attributes (variables, fields, characteristics, or features) in ho m o g eneo us c lus ters . Preliminary stage : Stage 1 : Preparing the data base Classification 8
  9. 9. 3. The CFAR-m algorithm Our solution … Stage 1 : Stage 2 : Classification Generating weights: one vector is defined for each object S ta g e 2 : S ec ondly, a n a ppropria te w eig hts vec tor is g enera ted for ea c h o bjec t. 9
  10. 10. 3. The CFAR-m algorithm Our solution … Stage 2 : Stage 3 : Generating weights: one vector is defined Computing the composite indicators and for each object rankingthe objects S ta g e 3 : T hirdly, w eig hts vec tors a re a pplied to the orig ina l da ta to c om pute C FA R -m c om pos ite indic a to rs a nd fina lly to c a rry out the overa ll ra nk ing o f objec ts . 10
  11. 11. 3. The CFAR-m algorithm Our solution … Preliminary stage : Stage 1 : Preparing the data base Classification Stage 3 : Stage 2 : Computing the composite indicators Generating weights: one vector is and rankingthe objects defined for each object 11
  12. 12. 3. The CFAR-m algorithm O ur s olution is ba s ed on a n orig ina l tec hnique tha t us es neura l netw ork s a nd, unlik e ex is ting m etho ds , pres ents the follo w ing c ha ra c teris tic s : Objectivity T here is no m a nipula tion of w eig hts . The Weights used to aggregate single indicators are generated automatically from the database through a learning process. Our model provides a fundamental s olution to the main aggregation problem. S pecificity E a c h objec t ha s a s pec ific equa tion to compute its composite indicator. Decis ion s upport I t a llow s perform ing of s im ula tio ns and therefore, can help to decide on appropriate actions and corrections. 12
  13. 13. 4. Some examples of implementing C-FARm Case study 1 : Computing a CFAR-m Human Development Index (comparison with the UNDP aggregation methodology based on equal weights) Case study 2 : Computing a CFAR-m indicator Governance Index (comparison with the MINEFI-France aggregation methodology using weights based on statistical quality of data ) Case study 3 : Computing a CFAR-m Country Risk Index (comparison with the PRS Group aggregation methodology based on expert opinion) 13
  14. 14. 4. Some examples of implementing C-FARm Case study 1 : Computing a CFAR-m Human Development Index (comparison with the UNDP aggregation methodology based on equal weights) Cliquez pour modifier le style des sous-titres du masque 14
  15. 15. Case study 1 : Computing a CFAR-m Human Development Index I n its firs t Human Development R eport (1990), the U nited N a tio ns D evelo pm ent P ro g ra m (U N D P ) intro duc e d a new index : H um a n D evelo pm ent I ndex (H D I ). H D I is intended to s um m a rize in o ne m ea s ure three dim ens io ns o f the develo pm ent pro c es s : lo ng evity, educ a tio na l a tta inm ent, a nd s ta nda rd o f living . D im ens io ns V a ria ble s (ba s ic indic a to rs ) H ea lth L ife ex pec ta nc y a t birth Cliquez pour modifier le style des sous-titres du masque A dult litera c y ra te E duc a tio n HDI P rim a ry, s ec o nda ry a nd tertia ry s c ho o ling enro lm ent ra tio s S ta nda rd of L iving G D P per c a pita 15
  16. 16. Case study 1 : Computing a CFAR-m Human Development Index T o c o m pute the H D I , U N D P c o ns ider the s im ple a vera g e (equa lly w eig hted s um ) o f the tree dim ens io ns . T he three dim ens io ns ha ve the s a m e w eig ht Cliquez pour modifier le style des sous-titres du masque L ife ex pec ta nc y E duc a tio n GDP index inde x inde x 16
  17. 17. Case study 1 : Computing a CFAR-m Human Development Index … … , thus , c o m pa ris o ns a m o ng different c o untries /reg io ns a re c a rried o ut. Cliquez pour modifier le style des sous-titres du masque 17
  18. 18. Case study 1 : Computing a CFAR-m Human Development Index The main arguments against HDI: I m po rta nt dim ens io ns a re no t c o ns idered (freedo m , hum a n rig hts , g o verna nc e, etc .) H D I is hig hly c o rrela ted to the G D P (0,89 a c c o rding to M a c G illivra y, 1991). T he three dim ens io ns a ls o a re hig hly c o rrela ted to the G D P W eig hting o f the three dim ens io ns is to o s ubjec tive 18
  19. 19. Case study 1 : Computing a CFAR-m Human Development Index The main critics made to the HDI : “The best known macro-indicator in the world is probably the Human D evelopment Index (HD I) developed by the United Nations D evelopment P rogram. It has been severely criticized for combining together indicators of income, health and education to create a composite index, both on the grounds that the weights are arbitrary and unjus tified and on the grounds that the three components of the index are highly correlated and hence give redundant results ” Literature Review of Frameworks for Macro- indicators Andrew Sharpe (2004) 19
  20. 20. Case study 1 : Computing a CFAR-m Human Development Index S ta g e 1 : C o untry c la s s ific a tio n 20
  21. 21. Case study 1 : Computing a CFAR-m Human Development Index S ta g e 2 : G enera ting w eig hts : one vec to r is defined fo r ea c h c o untry W eig hting s c hem e differs fro m o ne c o untry to a nother : C FA R -m is non- linea r L ife ex pec ta nc y E duc a tio n GDP index index index 21
  22. 22. Case study 1 : Computing a CFAR-m Human Development Index S ta g e 2 : G enera ting s pec ific w eig hting s w ith C FA R -m Weights are generated automatically through a learning process from the database : O bjec tivity Each country has a specific equation to compute its development index : S pec ific ity CFAR-m allows the identification, for each country, of the dimension that most influenced the calculation of its index, and therefore its ranking : I ntens ity a nd S ig n T he ra nk ing of C FA R -m w ill be both objec tive a nd releva nt. 22
  23. 23. Case study 1 : Computing a CFAR-m Human Development Index S ta g e 3 : C om puting a C FA R -m H um a n D evelo pm ent I ndex a nd ra nk ing c o untries HDI dimensions - Year 2005   CFAR-m's results for year 2005 Countries Life Education GDP   Country CFAR-m rank topping the list expectancy index index CFAR-m index minus rankt UNDP rank               I c ela nd 0.941 0.978 0.985   IS L 1 0 N o rw a y 0.913 0.991 1.000   NOR 2 0 A us tra lia 0.931 0.993 0.962   AUS 3 0 C a na da 0.921 0.991 0.970   CAN 4 0 I re la nd 0.890 0.993 0.994   IR L 5 0 S w eden 0.925 0.978 0.965   S WE 6 0 U nite d S ta te s 0.881 0.971 1.000   US A 7 5 S w itzerla nd 0.938 0.946 0.981   CHE 8 1 J a pa n 0.954 0.946 0.959   JPN 9 1 23
  24. 24. Case study 1 : Computing a CFAR-m Human Development Index S ta g e 3 : C om puting a C FA R -m H um a n D evelo pm ent I ndex a nd ra nk ing c ountries HDI dimensions - Year 2005   CFAR-m's results for year 2005 Countries closing Life Education GDP   Country CFAR-m CFAR-m the list expectancy index index rank rank index minus UNDP rank               Burundi 0.391 0.522 0.325 BDI 169 2 Central Afr. Rep. 0.311 0.423 0.418 CAF 170 1 Mozambique 0.296 0.435 0.421 MOZ 171 1 Guinea-Bissau 0.347 0.421 0.353 GNB 172 3 Chad 0.423 0.296 0.444 TCD 173 3 Mali 0.469 0.282 0.390 MLI 174 1 Sierra Leone 0.280 0.381 0.348 SLE 175 2 Burkina Faso 0.440 0.255 0.417 BFA 176 0 Niger 0.513 0.267 0.343 NER 177 3 24
  25. 25. Case study 1 : Computing a CFAR-m Human Development Index C FA R -m a s a dec is io n s uppo rt s o lutio n A s w eig ht s c hem a s a re s pec ific , it a llow s to perfo rm s im ula tions Impact on the rank of an improvement 0.1 in one dimension (here, for North African countries) Life expectancy index Number of ranks gained in overall ranking 25
  26. 26. Case study 1 : Computing a CFAR-m Human Development Index C FA R -m a s a dec is io n s uppo rt s o lutio n A s w eig ht s c hem a s a re s pec ific , it a llow s to perfo rm s im ula tions Impact on the rank of an improvement 0.1 in one dimension (here, for North African countries) Education index Life expectancy index Number of ranks gained in overall ranking 26
  27. 27. Case study 1 : Computing a CFAR-m Human Development Index C FA R -m a s a dec is io n s uppo rt s o lutio n A s w eig ht s c hem a s a re s pec ific , it a llow s to perfo rm s im ula tions Impact on the rank of an improvement 0.1 in one dimension (here, for North African countries) GDP index Education index Life expectancy index Number of ranks gained in overall ranking 27
  28. 28. Case study 2 : Computing a CFAR-m Governance Index (comparison with the MINEFI-France aggregation methodology using weights based on statistical quality of data) Cliquez pour modifier le style des sous-titres du masque 28
  29. 29. Case study 2 : Computing a CFAR-m Governance Index T he " I ns titutio na l pro files " da ta ba s e It gathers a whole set of indicators characterizing the institutions of 85 developed and emerging countries 132 I nfo rm a tio n 9 variables a g g reg a tio n governance pro c es s indicators Each variable is weighted according to its standard deviation 29
  30. 30. Case study 2 : Computing a CFAR-m Governance Index T he " I ns titutio na l pro files " da ta ba s e Gathers a whole set of indicators characterizing the institutions of 85 developed and emerging countries 9 g o verna nc e 132 variables indic a to rs 1 : Political institutions 2 : Public order 85 countries 3 :Perfomance of Administration I nfo rm a tio n 4 :Efficiency of free markets a g g reg a tio n proc es s 5 :Prospective and planning 6 : Security of transactions 7 : Regulation 8 : Foreign openness 9 : Social cohesion 30
  31. 31. Case study 2 : Computing a CFAR-m Governance Index S ta g e 1 : C ountry ra nk ing 1st dimension's case : ″political institutions″ 31
  32. 32. Case study 2 : Computing a CFAR-m Governance Index S ta g e 2 : G enera ting s pec ific w eig hts w ith C FA R -m R em inder : in the M I N E FI 's m ethod, the w eig ht o f o ne va ria ble c o m es fro m its s ta nda rd devia tion The component The component that weighs the that weighs the most in the least in the computation computation Components of the 1st dimension re. political institutions  H ow leg itim a te a re tho s e w eig hting s ?  A nd w ha t a bo ut the fa c t tha t they a pply to a ll c o untries ? 32
  33. 33. Case study 2 : Computing a CFAR-m Governance Index S ta g e 2 : G enera ting s pec ific w eig hting s w ith C FA R -m T here is no em piric a l m a nipula tio n. W eig hting s a re pro c es s ed us ing the s o le inform a tio n em bedded in the va ria bles . Kuwait Components of the 1st dimension re. political institutions 33
  34. 34. Case study 2 : Computing a CFAR-m Governance Index S ta g e 2 : G enera ting s pec ific w eig hting s w ith C FA R -m T here is no em piric a l m a nipula tio n. W eig hting s a re pro c es s ed us ing the s o le inform a tio n em bedded in the va ria bles . Kuwait Components of the 1st dimension re. political institutions 34
  35. 35. Case study 2 : Computing a CFAR-m Governance Index S ta g e 2 : G enera ting s pec ific w eig hting s w ith C FA R -m 1st Dimension: "Political Institutions" Countries CFAR-m MINEFI Ranking topping the list ranking ranking spread         Sweden 1 1 0 France 2 3 -1 New Zeland 3 2 1 Spain 4 6 -2 Canada 5 4 1 Germany 6 5 1 Norway 7 7 0 USA 8 15 -7 Italy 9 12 -3 India 10 9 1 Czech Rep. 11 8 3 Ireland 12 11 1 Senegal 13 16 -3 Brazil 14 18 -4 Israel 15 21 -6 Hong Kong 16 26 -10 Greece 17 10 7 Hungary 18 14 4 Argentine 19 19 0 35
  36. 36. Case study 2 : Computing a CFAR-m Governance Index S ta g e 3 : C om puting a C FA R -m G o verna nc e I ndex a nd ra nk ing c o untries O nc e a ll dim e ns io ns o f the ins titutio na l pro file ha ve bee n c o m puted w e ha ve pro c e s s ed w ith the fina l a g g re g a tio n, pro duc ed a C FA R -m indic a to r fo r ea c h c o untry a nd the n a g lo ba l ra nk ing , w hic h the M I N E FI c o uld no t c o m ple te ! Ranking according to CFAR-m Governance Index Countries Countries topping the list closing the list         1 Sweden 76 Nigeria 2 Ierland 77 Cameroon 3 Israel 78 Yemen 4 Spain 79 Ouzbekistan 5 Canada 80 Mauritanie 6 Norway 81 Egypt 7 Italy 82 Syria 8 Germany 83 Iran 9 Portugal 84 Ivory Coast 10 Hungary 85 Chad 36
  37. 37. Case study 2 : Computing a CFAR-m Governance Index C FA R -m is a va lua ble dec is ion s upport T his is the dim ens io n tha t a llow s to pro g res s the quic k er in the ra nk ing Ranks gained in the world ranking Security of openness Prospective & planning transact. Foreign Public order cohesion Perf. of free Perf. of Admin. Social Regulation institutions markets Political Dimensions of the "institutional profile" when affected with a 10% increase 37
  38. 38. Case study 3 Computing a CFAR-m Country Risk Index (comparison with the PRS Group aggregation methodology based on expert opinion) 38
  39. 39. Case study 3 : Computing a CFAR-m Country Risk Index P ro c es s : E c ono m ic I nfo rm a tio n R is k a g g reg a tio n va ria bles pro c es s indic a to r "B la c k bo x " Generally, there is no indication about the computation method 39
  40. 40. Case study 3 : Computing a CFAR-m Country Risk Index A pplic a tion to thehe P R S G ro up's International C ountry Risk Guide T he I C R G brea k s the c ountry ris k into 3 s ub-c la s s es : C om pos ite indic a to r : C ountry-ris k indic a to r S ub-indic a to r #1 : S ub-indic a to r #2 : S ub-indic a to r #3 : P o litic a l ris k E c o nom ic ris k Fina nc ia l ris k E a c h s ub-indic a to r is c o m po s ed w ith s evera l fa c tors to o : 40
  41. 41. Case study 3 : Computing a CFAR-m Country Risk Index A pplic a tion to thehe P R S G ro up's International C ountry Risk Guide E very s ub-indic a tor is a c om pos ite its elf : 12 factors Score (max) S ub-indic a to r #1 : P o litic a l R is k A Government's stability 12 B Social and Economic environment 12 C Investment environment 12 D Internal conflicts 12 E External conflicts 12 F Corruption 6 G Military's influence on policy 6 H Influence of religions on policy 6 I Law and regulation 6 J Ethnic lobbying 6 K Democratic responsibility 6 M Administration and stability of the 4 institutions Total 100 41
  42. 42. Case study 3 : Computing a CFAR-m Country Risk Index A pplic a tion to thehe P R S G ro up's International C ountry Risk Guide S ub-indic a to r #2 : 5 factors Score (max) E c o nom ic ris k A GDP per capita 5 B GDP growth 10 C Inflation rate 10 D Balance of payments (% of GDP) 10 E Current account (% of GDP) 15 Total 50 42
  43. 43. Case study 3 : Computing a CFAR-m Country Risk Index A pplic a tio n to thehe P R S G ro up's International C ountry Risk Guide S ub-indic a to r #3 : 5 factors Score (max) Fina nc ia l ris k A External debt (% of GDP) 10 B Cost of external debt (% of GDP) 10 C Current account (% of goods and 15 services exports) D International net liquidity (months 5 of import funding) E Exchange rate stability 10 Total 50 43
  44. 44. Case study 3 : Computing a CFAR-m Country Risk Index C o untry-ris k indic a to r P o litic a l ris k E c onom ic ris k Fina nc ia l ris k M ea s uring the po litic a l-ris k fa c tor for yea r 2006 Country Govern Social Invest Internal External Corrup Military's Influence Law and Ethnic Democrat Adminis ment's and ment conflicts conflicts tion influence of regula lobbying ic tration stability Econo environ on policy religions tion responsib and mic ment on policy ility stability environ of the ment institu tions Albania 8.5 5.5 8.0 10.0 11.0 1.0 5.0 5.0 2.5 4.5 5.0 2.0 Algeria 9.6 5.8 9.1 8.9 10.0 1.5 3.0 2.5 3.0 3.5 4.5 2.0 Angola 9.6 2.0 7.9 9.3 11.0 2.0 2.0 4.0 3.0 3.0 2.0 1.0 Argentina 10.2 5.2 6.6 10.0 10.0 2.5 4.5 6.0 2.5 6.0 4.5 3.0 Armenia 8.4 4.0 8.0 8.6 7.6 1.5 3.5 5.0 3.0 5.5 3.0 1.0 Australia 10.3 9.7 12.0 9.3 9.6 4.6 6.0 6.0 5.5 4.0 6.0 4.0 ………. 44
  45. 45. Case study 3 : Computing a CFAR-m Country Risk Index S ta g e 1 : C ountry ra nk ing 45
  46. 46. Case study 3 : Computing a CFAR-m Country Risk Index R em inder : P R S G ro up m etho do log y  W eig hting o f ea c h va ria ble defined by ex perts  W eig hting s a re the s a m e w ha tever the c o untry 12 factors V1 Government's stability V7 Military's influence on policy V2 Social and Economic V8 Influence of religions on policy environment V3 Investment environment V9 Law and regulation V4 Internal conflicts V10 Ethnic lobbying V5 External conflicts V11 Democratic responsibility V6 Corruption V12 Administration and stability of the institutions 1st dimension factors re. political institutions 46
  47. 47. Case study 3 : Computing a CFAR-m Country Risk Index S ta g e 2 : defining s pec ific w eig hting s w ith C FA R -m N ot a ll c o untries ha ve the s a m e w eig hting s : it s how s tha t C FA R -m is a no n-linea r proc es s 1st dimension factors re. political institutions 47
  48. 48. Case study 3 : Computing a CFAR-m Country Risk Index S ta g e 2 : defining s pec ific w eig hting s w ith C FA R -m N o t a ll c ountries ha ve the s a m e w eig hting s : it s how s tha t C FA R -m is a no n-linea r proc es s 1st dimension factors re. political institutions 48
  49. 49. Case study 3 : Computing a CFAR-m Country Risk Index M ea s uring the c o untry-ris k fa c to r fo r yea r 2006   CFAR-m results   PRS Group results Countries   Ranking Country   Ranking Country Spread topping the list         1 Finland   1 Finland 0 2 Iceland   2 Luxembourg 1 3 Luxembourg   3 Iceland -1 4 Sweden   4 Ireland 1 5 Ireland   5 Sweden -1 49
  50. 50. Case study 3 : Computing a CFAR-m Country Risk Index M ea s uring the c o untry-ris k fa c tor for yea r 2006   CFAR-m results   PRS Group results Countries   Ranking Country   Ranking Country Spread in the middle of the list         ……… ……… ……… ……… 68 Saudi Arabia 68 Saudi Arabia 0 69 El Salvador 74 El Salvador -5 70 Guatemala 80 Guatemala -10 71 Ghana 67 Ghana 4 72 Brazil 76 Brazil -4 ……… ……… ……… ……… 50
  51. 51. Case study 3 : Computing a CFAR-m Country Risk Index M ea s uring the c o untry-ris k fa c tor fo r yea r 2006   CFAR-m results   PRS Group results Countries   Ranking Country   Ranking Country Spread closing the list         135 Haiti 135 Ivory Coast 1 136 Ivory Coast 136 Haiti -1 137 Serbia 137 Congo, RD 3 138 Montenegro 138 Iraq 1 139 Iraq 139 Serbia -2 140 Congo, RD 140 Montenegro -2 141 Somalia 141 Somalia 0 51
  52. 52. Cliquez pourQuestions & Answers modifier le style des sous-titres du masque Going forward...

×