ictd government revenue dataset

18
The ICTD Government Revenue Dataset Wilson Prichard Interna1onal Center for Tax and Development

Upload: international-centre-for-tax-and-development-ictd

Post on 02-Jun-2015

153 views

Category:

Economy & Finance


0 download

DESCRIPTION

Presentation by Wilson Prichard (U. Toronto and Research Director, ICTD) Introduction Over the last decade, the issue of taxation has benefited from a growing interest from researchers and policy makers, especially regarding its impact on key development components such as economic growth, governance and poverty reduction. However, for a long period, the low quality of available data significantly hindered the quality of the research on the topic and the robustness of the results. This led to the proliferation of small-scale datasets and impeded proper comparison and replication of studies. It is with the aim of coping with this observation that the ICTD developed the GRD. The GRD has been built by compiling and harmonizing numerous existing datasets from various sources, yielding a homogeneous set of data covering a large range of countries over a long period. In addition to expanding the coverage and the quality of the data, the GRD also includes a clear separation between resource and non-resource government revenue, allowing for precise analyses of non-resource tax collection. Although the GRD still suffers from imperfection - particularly because of the successive merging of databases realized from different collection methods - the fact remains that it represents a significant enhancement, which will enable for deeper and more accurate research and improve our understanding of taxation and its effect on economies. Why this dataset is needed: The limitation of existing data. The ICTD GRD is based on the understanding that the quality of international revenue data is not only poor, but also insufficient to sustain analysis, thereby leading to misleading or insufficiently robust findings on tax and development. The ICTD’s goal therefore was to create a single composite dataset that is more complete and more accurate than alternatives, in which one could look up for every country year, any available source of data and compare them, thereby getting the best available source for that country year. Existing international sources (IMF GFS – Pre and post 1990, OECD, CEPALstat, OECDLatAm, OECD AEO, World Bank, Keen and Mansour) all suffer from substantial limitations – reflected in researchers relying increasingly on composite and ad hoc datasets, which are subject to errors, lack transparency and difficulties of comparability. This also makes them hard to replicate, and with huge scope for errors. Indeed, most of the existing databases exhibit missing data stemming from incomplete range of revenue categories, and failure at consistently distinguishing natural resource wealth. Moreover, non-tax revenues are often not included in databases, thus giving an incomplete picture of government finances. Finally, in many countries, GDP may be vastly under-estimated, leading to sizable overestimation of key variables as shares of GDP, hence a need to rebase. Equally, irregular rebasing exercises can lead to major breaks in time series

TRANSCRIPT

Page 1: Ictd government revenue dataset

The  ICTD  Government  Revenue  Dataset

Wilson  Prichard  Interna1onal  Center  for  Tax  and  Development  

     

Page 2: Ictd government revenue dataset

Overview •  The  ICTD  GRD  responds  to  major  limits  of  exis1ng  sources  for  conduc1ng  

cross-­‐country  tax  research,  with  major  improvements  in  data  coverage  and  accuracy  by  combining  data  from  mul1ple  sources  –  including  a  more  consistent  approach  to  natural  resource  revenues  

•  This  is  a  cri1cal  complement  to  work  at  interna1onal  organiza1on  to  improve  data  over  the  long-­‐term,  as  it  offers  a  much  improved  founda1on  for  immediate  research.  

•  However,  it  is  a  very  par1al  solu1on:    There  are  inescapable  limita1ons,  which  reflect  the  limits  of  any  available  sources,  and  the  imperfec1ons  of  merging  data  from  mul1ple  sources  

•  There  is  a  need  for  coopera1on  and  consensus  to  maintain  the  dataset  as  a  resource  for  researchers  while  new  efforts  at  the  IMF,  OECD  and  elsewhere  begin  to  bear  fruit.  

Page 3: Ictd government revenue dataset

Outline

1.  Mo1va1on  

2.  Construc1on  of  the  Dataset  

3.  Limita1ons  

4.  Lessons  and  Next  Steps  

Page 4: Ictd government revenue dataset

Motivation •  Weaknesses  of  exis,ng  data  raise  serious  concerns  

about  the  robustness  of  tax  and  development  research,  and  reduces  value  of  data  for  broader  descrip1ve  and  compara1ve  exercises  

•  Exis1ng  interna1onal  sources  all  suffer  from  substan1al  limita1ons  –  reflected  in  researchers  relying  increasingly  on  composite  and  ad  hoc  datasets  

•  However,  ad  hoc  datasets  subject  to  errors,  lack  of  transparency  and  difficul1es  of  comparability  

   

Page 5: Ictd government revenue dataset

Weaknesses  of  Existing  Sources •  Missing  data  in  sources  with  full  country  coverage  •  Limited  coverage  and  comparability  of  regional  

sources  •  Non-­‐tax  revenue  o>en  not  included,  thus  giving  

incomplete  picture  of  government  finances    •  Failure  to  consistently  dis,nguish  natural  resource  

revenues  in  most  exis1ng  databases    •  Incomplete  range  of  revenue  categories  in  many  

researcher  databases    •  Simple  errors,  most  notably  in  researcher  databases  

–  and  oYen  driven  by  merging  of  sources  •  Problems  with  inconsistencies  in  many  GDP  series    

Page 6: Ictd government revenue dataset

Potential  for  Complementarity

 

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

1990 1993 1996 1999 2002 2005 2008 Year

Ghana: Total tax as % GDP

K & M

GFS

IMF CR

WB WDI

AEO

Page 7: Ictd government revenue dataset

Construction  of  the  ICTD  GRD 1.  A  Standard  Revenue  Classifica1on  2.  Compiling  Available  Interna1onal  Sources  3.  Compiling  and  Adding  Ar1cle  IV  data  4.  Dealing  with  Natural  Resources  5.  Addi1onal  Issues  6.  A  Common  GDP  Series  7.  Manual  Data  Cleaning  

 

Page 8: Ictd government revenue dataset

Construction  of  the  ICTD  GRD:    A  Standard  Revenue  Classification

•  Tax  and  Non-­‐tax  

•  Natural  Resources  

•  Social  Contribu1ons  

 !!

Total!Gov’t!

Revenue!

Total!Gov’t!Revenue!Excluding!Grants!

Grants!

Tax!Revenue!

Non7Tax!Revenue!!

!

Social!Contributions!

Non7Resource!

Direct!Taxes!

Indirect!Taxes!

Non_Resource!Taxes!on!

Incomes,!Profits!and!Capital!Gains!

Property!Taxes!

Taxes!on!Individuals!

Non_Resource!Taxes!on!

Corporations!

Taxes!on!International!

Trade!

Taxes!on!Goods!and!Services!

Other!Taxes!

Sales!Taxes/VAT!

Excises!

Imports!

Exports!

Resource!Non7Tax!

!

Non7Resource!Non7Tax!Revenue!

Non7Resource!Tax!

Revenue!

Resource!Tax!

Revenue!

Resource!Direct!Taxes!

Resource!Taxes!on!Incomes,!Profits!and!Capital!Gains!

Resource!Taxes!on!Corporations!

Page 9: Ictd government revenue dataset

Construction  of  the  ICTD  GRD:  Compiling  Alternative  Sources

•  IMF  GFS  (pre  and  post-­‐1990)  •  OECD  •  CEPALSTAT  •  OECD  LatAm  •  OECD  AEO  •  World  Bank  •  Keen  and  Mansour    

Page 10: Ictd government revenue dataset

Construction  of  the  ICTD  GRD;  Article  IV  Data

•  Ar1cle  IV  data  oYen  available  where  other  sources  missing  –  though  is  less  rigorously  reviewed,  so  should  be  used  when  it  matches  surrounding  sources  

•  Requires  careful  categoriza1on,  as  revenue  categories  vary  across  countries  and  over  1me    

0%

5%

10%

15%

20%

25%

30%

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

Rat

io (%

)

Year

Albania: tax/GDP ratio (%)

Art. IV (GG) GFS (GG) GFS (CG) GFS (CG+SS) Michigan Ross WTD WB

Page 11: Ictd government revenue dataset

Construction  of  the  ICTD  GRD:  Natural  Resources

Angola  1996  

 •  Interna1onal  sources  are  inconsistent  in  classifying  resource  revenue  between  taxes  and  non-­‐tax  revenue  

•  Non-­‐resource  tax  revenue  is  the  analy1cally  interes1ng  category,  which  requires  excluding  natural  resource  component  of  tax  

•  Some1mes  possible  using  OECD,  most  oYen  rely  on  IMF  Ar1cle  IV  

Total  Revenue  

Total  Tax   Taxes  on  Income  

Total  Non-­‐Tax  Rev  

Resource  Revenue  

Non-­‐Resource  Non-­‐Tax  

Pre-­‐Adjustment  

48.9%   48.6%   32%   0.3%   -­‐   0.3%  

Post-­‐Adjustment  

48.9%   4.8%   0.9%   44.1%   43.8%   0.3%  

Page 12: Ictd government revenue dataset

Construction  of  the  ICTD  GRD:  Other  Issues

•  Consistent  approach  to  social  contribu,ons:    Varia1on  across  sources  in  the  inclusion  of  social  contribu1ons  can  lead  to  incompa1bility.    We  report  all  figures  inclusive  and  exclusive  of  social.  

•  Dealing  with  federal  states:    Focusing  exclusively  on  central  government  can  vastly  understate  tax  collec1on  in  federal  states.    We  adopt  general  government  data  where  it  is  significantly  different  from  central  data,  and  consistent  over  1me.  

•  Direct  and  Indirect  Taxes:  Owing  to  differences  across  sources  in  sub-­‐categories  of  taxa1on,  we  calculate  direct  and  indirect  taxes  for  all  country-­‐years.  

 

Page 13: Ictd government revenue dataset

Construction  of  the  ICTD  GRD:  Common  GDP  Series

•  There  are  simple  differences  across  sources  in  GDP  figures,  making  transparency  and  consistency  about  GDP  figures  as  important  as  the  tax  data    

•  Growing  recogni1on  that  underes1ma1on  of  GDP  can  lead  to  vast  overes1ma1on  of  key  variables  as  shares  of  GDP  

•  Equally,  irregular  rebasing  exercises  can  lead  to  major  breaks  in  1me  series  data  unless  applied  retroac1ve  to  earlier  periods  –  which  is  frequently  not  the  case  

Page 14: Ictd government revenue dataset

Construction  of  the  ICTD  GRD:  Common  GDP  Series

 

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

1980 1984 1988 1992 1996 2000 2004 2008 Year

Ghana: Total tax as % source-specific GDP

K & M GFS IMF CR WB WDI

0.0%

2.0%

4.0%

6.0%

8.0%

10.0%

12.0%

14.0%

16.0%

1980 1983 1986 1989 1992 1995 1998 2001 2004 2007 2010 Year

Ghana: Total tax as % common GDP series

K & M GFS IMF CR WB WDI

Page 15: Ictd government revenue dataset

Construction  of  the  ICTD  GRD:  Manual  Data  Cleaning

•  Data  merging  would,  ideally,  be  automated,  filling  gaps  in  a  “baseline”  source  with  pre-­‐determined  “second  best”  data  

•  However,  inconsistencies  across  sources  –  some  reflec1ng  differing  methods,  others  reflec1ng  simple  data  discrepancies  –  imply  that  automated  processes  result  in  incompa1ble  and  inconsistent  1me  series  

•   As  such,  it  is  necessary  to  manually  clean  the  data  to  ensure  consistency  within  countries  between  data  sources  

 

Page 16: Ictd government revenue dataset

Developing  Country  Data  Coverage

Data  coverage  is  drama1cally  more  complete  than  for  other  sources,  including  the  most  widely  used  composite  dataset,  from  the  IMF  FAD.     ICTD  GRD   IMF  FAD   IMF  Art  IV   IMF  GFS   WDI  

Total  Revenue   2317   1913   1484   1391   1060  

Total  Tax   2348   1976   1895   1396   1060  

Taxes  on  Income,  Profits  and  Capital  Gains  

1900   1909   1341   1395   1060  

Taxes  on  Goods  and  Services  

1952   1856   1092   1395   1060  

Page 17: Ictd government revenue dataset

Continued  Limitations 1.   S,ll  significant  missing  data  

2.   Challenges  in  dealing  with  resource  revenues  1.  Some1mes  data  is  not  available,  so  countries  excluded  

from  analysis  2.  OYen  impossible  to  dis1nguish  resource  revenue  from  

other  non-­‐tax  revenue  3.  Defini1onal  issues  in  deciding  what  classifies  as  resource  

revenue    3.   Varia,on  across  sources  o>en  inexplicable,  data  

inherently  imperfect  –  and  merging  choices  inevitably  subjec1ve  

 

Page 18: Ictd government revenue dataset

Lessons  and  Next  Steps 1.  Dealing  with  resource  revenues  is  cri1cal,  needs  to  be  integrated  with  

interna1onal  databases  and  requires  a  common  framework  

2.  Ajen1on  to  GDP  figures  equallly  cri1cal,  and  any  dataset  should  separately  provide  LCU  figures,  %  of  GDP  figures  and  clearly  documented  GDP  series  

3.  Any  dataset  should  deal  with  both  tax  and  non-­‐tax  revenues  in  order  to  be  analy1cally  useful  for  research,  while  also  adop1ng  a  consistent  approach  to  social  contribu1ons  

4.  There  remain  opportuni1es  for  much  improved  interna1onal  coopera1on,  as  there  is  currently  major  overlap  and  duplica1on  –  some1mes  even  within  organiza1ons  –  and  new  ini1a1ves  have  tended  to  address  some,  but  not  all,  of  the  challenges  noted  here.  

5.  Merging  data  from  mul1ple  sources  for  research  is  fraught  with  risks  –  and  is  extremely  1me  intensive  –  thus  placing  a  premium  on  establishing  a  single  accepted  source,  transparency  and  providing  resources  for  long-­‐term  maintenance  of  the  dataset