Political Decentralization Study Materials Ch 2

3.3 Empirical Studies of Provincial Protectionism in Russia and China The studies described in the previous section focus on capture and corruption within specific communities. The theoretical literature has also addressed possible adverse spillover impacts of decentralization on other parts of the country, such as raising provincial protectionism. As explained above, the Blanchard---Shleifer--Sonin viewpoint in this regard contrasts sharply with the Qian---Roland--- Weingast market---preserving---federalism hypothesis. What does empirical evidence show? In the Russian context, evidence in favor of the `local---state---corroding--- federalism’ hypothesis is provided by Ponomareva and Zhuravskaya (2004) and Slinko et al (2004). Using a sample of Russian firms from the second half of the 1990s, the former show that federal tax arrears accumulated at a faster rate in provinces whose governors had a larger popular base, were in political opposition to the central government, and had a stronger bargaining position. Lambert Mogiliansky et al (2003) show in a sample of over 8,000 Russian firms that firms located in regions with governors that were politically more powerful, independent from the central government and with more opaque tax collection systems were significantly less likely to be liquidated under the new bankruptcy law of 1998 which enlarged the discretion of regional governors over bankruptcy procedures.  Slinko  et  al   (2004)  show  firms  with  more  preferential  treatments  grew  faster  in   profits,  sales,  employment  and  federal  tax  arrears  after  controlling  for  firm   characteristics.  Firms  without  any  such  treatments  invested  more  and  performed   better  when  they  were  located  in  regions  where  political  power  was  less   concentrated.  Regions  with  more  concentrated  power  exhibited  lower  rates  of   development  of  small  businesses  and  retail  turnover.  While  overall  rates  of  growth   of  economic  activity  in  the  region  were  not  significantly  different,  greater   concentration  of  power  was  associated  with  lower  payment  of  federal  taxes.   China  exhibited  substantially  different  results.  Jian  et  al  (1999)  use  data  from   1982E 92  for  30E odd  provinces  to  show  that  fiscal  decentralization  was   positively  related  to  growth  of  per  capita  regional  GDP,  of  nonE agricultural   employment  and  nonE state  industrial output, after controlling for provincial tax rates, regional labor force growth, province and time fixed effects. Administrative decentralization had a significant positive correlation with local fixed investment, ratio of local to central government investment, and growth of non-agricultural employment and non-state industrial output. They explain these results as resulting from the effects of altered fiscal contracts between local and central governments, wherein local governments were allowed to retain a higher fraction of tax revenues collected during 1982---92, compared to the previous decade. Despite this, inequality in provincial per capita budgetary revenues and expenditures at the provincial level declined during 1982—92, while inequality in central transfers moved in the opposite direction. They infer that high incentives on the margin coexisted with infra-marginal redistribution in the fiscal contracting mechanism. A  contrasting  view  of  the  effects  of  increased  fiscal  decentralization  in  China  is   presented  by  Young  (2000)  and  Poncet  (2003,  2004),  who  argue  that  the  increased   autonomy  and  incentives  offered  to  local  governments  induced  them  to  engage  in   provincial  protectionism,  in  the  form  of  high  barriers  to  interE regional  trade.     Moreover,  just  as  in  the  Russian  case,  local  governments  offered  local  enterprises   exemptions  on  central  taxes,  accompanying  them  with  informal  levies  accruing  to   their  own  coffers.  They  show  that  the  composition  of  output  converged  across   regions  between  1978  and  1997,  while  prices,  labor  productivity  and  labor   allocations  diverged,  and  reliance  on  international  trade  increased.  Despite the evidence for growing trends of regional protectionism in China, Young (2000, p.1129) concedes that fiscal decentralization probably contributed to economic growth overall, owing to its success in dealing with control and incentive problems. Most writers on China and Russia seem to concur that the overall effects of fiscal decentralization was positive in China, and negative in Russia, and that this owed to a strong central government in China. While this is plausible, there is little hard evidence in favor of this hypothesis. Most of the empirical results for China pertain to data on differences in regional growth rates and their relation to the nature of fiscal contracting: there are no estimates available of the effects of regional protectionism on growth in China as a whole which trade off the superior incentive properties of decentralization against the corroding effects on regional protectionism. The  only  supporting  arguments  are  provided  by  Jin  et  al  (1999)  who  point  out  that   fiscal  contracts  with  the  provinces  in  the  two  countries  were  quite  different:   budgetary  constraints  for  provinces  were  much  `harder’  in  China.  Blanchard  and   Shleifer  (2000)  argue  that  the  problem  of  capture  of  local  governments  by  local   firms  was  much  less  severe  in  China,  because  China  did  not  start  its  reforms  with   the  large  industrial  enterprises  that  characterized  the  Russian  economy  of  the  early   1990s.     4. Emerging  Issues In  this  section,  I  shall  describe  other  important  questions  that  are  beginning  to   receive  more  attention,  and  need  to  be  addressed  further  in  future  work.    The focus on intra-community allocations and extent to which they reveal mis-targeting, diversion or lack of responsiveness to local need is a natural first step in evaluating the accountability performance of local governments. Much of the existing literature has grappled with the econometric challenge of identifying causal effects of decentralization and of underlying determinants of accountability, using experimental or quasi-experimental methods. A natural concomitant is that effects on a narrow range of outcomes in the short run are evaluated. For instance, patterns of political participation and competition in any given context is taken as given, and viewed as an underlying determinant of the effects of decentralization. The  possibility  that  these  may  in  the  long  term  be  affected  by  decentralization  has   been  largely  ignored.  Moreover,  empirical  identification  of  effects  of   decentralization  policies  often  reveal  very  little  about  the  underlying  processes  at   play.  Less  effort  has  been  devoted  to  understanding  the  political  economy  of  local   governments  in  any  given  context.  Consequently  when  disparate  results  are  found   in  different  contexts,  it  is  difficult  to  draw  any  general  lessons  that  could  be  useful  in   other  areas  where  decentralization  policies  are  yet  to  be  determined.   These  concerns  of  course  apply  more  to  recent  empirical  literature  on  policy   evaluation  more  generally:  identification  of  shortE term  effects  on  a  narrow  range   of  dimensions,  concerns  about  external  validity,  lack  of  understanding  of  underlying   processes,  and  lack  of  a  systemic  (e.g.,  general  equilibrium)  perspective.    In  the   specific  context  of  decentralization,  the  following  sets  of  issues  are  emerging  as   important  new  areas  of  investigation:      widening    the  conceptual  apparatus  of  political  economy  distortions  beyond capture  and  corruption  effects  on  a  wider  range  of  relevant  dimensions,  such  as  inter-­‐community allocations,  and  the  functioning  of  local  democracy  itself  (civic  participation, political  competition,  legitimacy,  leadership  and  learning);  evaluating  decentralization  vis-­à-­vis  other  organizational  alternatives  design  issues  highlighted  by  first  generation  theories  of  fiscal  federalism, such  as  inter-­‐jurisdictional  spillovers,  inter-­‐community  allocations,  and hardness  of  budget  constraints;  the  appropriate  domain  and  extent  of  devolution;  and  the  political  economy  of  implementation. 4.1  Widening  the  Conceptual  Apparatus   While  elite  capture  or  corruption  are  important  political  distortions  associated  with   local  governments,  they  are  not  necessarily  the  only  ones.  Consider  communities   where  literacy  is  high,  local  elections  are  free  and  fair  by  and  large,  involve  high   levels  of  political  turnout  and  awareness  across  all  sections  of  the  population,  and   are  highly  contested.  Should  one  then  assess  the  likely  welfare  benefits  of   decentralization  to  be  high,  on  the  grounds  that  such  a  community  is  not   susceptible  to  much  capture  or  corruption?  Strong  political  competition  could   conceivably  induce  candidates  to  pander  to  voters  in  a  variety  of  ways,  by  offering   shortE term  benefits  (such  a  loan  waivers  or  low  taxes)  at  the  expense  of  longE term  programs.   In  developing  countries,  there  is  considerable  evidence  of  clientelistic  politics  or   voteE buying,  wherein  parties  offer  private  transfers  or  benefits  to  specific    groups   of  swingE voters  conditional  on  the  latter’s  voting  for  them  (e.g.,  Bardhan  et  al   (2009),  Bardhan  and  Mookherjee  (2012),  Khemani  (2014),  Keefer  and  Khemani   (2014),  Kitchelt  and  Wilkinson  (2007),  Robinson  and  Verdier  (2013),  Stokes   (2005),  Wantchekon  (2003)).  Such  clientelistic  relations  tend  to  be  more  common     with  poorer  voters  (as  their  votes  are  `cheaper’  to  purchase).  In  contexts  where   clientelism  is  common,  one  would  therefore  observe  highly  proE poor  targeting.   Viewing  the  outcomes  through  the  singular  lens  of  elite  capture,  one  might  infer  the   outcome  to  be  benign.  Yet  it  could  conceal  the  underlying  distortion  of  clientelism,   associated  with  excess  spending  on  narrowly  targeted  discretionary  private   transfers  of  a  recurring  nature,  at  the  expense  of  more  broadly  targeted   programmatic  transfer  programs  or  public  goods.     In  similar  vein,  mandated  reservations  of  local  government  positions  for   disadvantaged  minorities  defined  by  ethnicity  or  tribe  is  often  associated  with   increased  transfers  to  members  of  such  groups  (as  much  research  in  the  Indian   context  shows).  How  such  outcomes  should  be  evaluated  depend  on  the  relative   importance  of  distortions  related  to  capture  and  clientelism  respectively.  Are  these   increased  transfers  symptoms  of  a  reduction  of  elite  capture,  or  heightened   clientelism?  More  attention  needs  to  be  devoted  to  the  study  of  clientelistic   distortions,  and  whether  they  are  heightened  by  decentralization.  Khemani  (2013)   provides  a  theoretical  model  wherein  central  governments  use  decentralization  as   an  instrument  of  dispensing  targeting  private  transfers,  at  the  expense  of  broad-­‐ based  public  goods.  Khemani  (2014)  subsequently  provides  evidence  from  the   Philippines  of  a  robust  negative  correlation  between  voter  reports  of  having   received  targeted  private  transfers,  and  the  delivery  of  community  health  services   (which  in  turn  was  positively  correlated  with  child  health  outcomes).   In  a  related  vein,  politicians  and  voters  are  often  prone  to  cash  in  on  short-­‐term   benefits  at  the  expense  of  long  term  objectives,  resulting  in  problems  of  credible   commitment  by  politicians  to  policies  that  are  likely  to  achieve  long-­‐term  goals.  For   instance,  there  is  short-­‐term  pressure  to  avoid  levying  or  collection  of  taxes  and   user  fees,  which  undermine  the  fiscal  capacity  of  governments  to  make  long-­‐term   investments  in  infrastructure.    To  gain  popularity,  governments  and  voters  could  be   prone  to  spend  less  on  education,  health  and  sanitation  the  benefits  of  which  are   often  diffuse  and  pushed  into  the  distant  future.  Is  decentralization  more  or  less   likely  to  be  vulnerable  to  such  populist  distortions?  This  issue  is  discussed  in  a   preliminary  way  in  Bardhan  and  Mookherjee  (2013),  but  it  deserves  more  attention,   both  theoretically  and  empirically.   4.2  Wider  Effects  of  Decentralization   Much  attention  has  been  devoted  to  intra-­‐community  allocations  of  government   services:  how  they  are  allocated  between  different  programs,  different  categories  of   beneficiaries,  or  how  efficiently  such  services  have  been  delivered.    In  evaluating  the   relation  between  underlying  determinants  of  elite  capture  such  as  land  or  socio-­‐ economic  equality,  attention  is  often  focused  on  a  narrow  subset  of  programs   administered  by  local  governments  (e.g.,  latrines  in  Ecuador  (Araujo  et  al  (2008)),   schools  in  Argentina  (Galiani  et  al  (2008))  or  allocation  across  a  few  categories  of   public  goods  in  India  (Chattopadhyay  and  Duflo  (2003),  Foster  and  Rosenzweig   (2001)).  Rarely  is  a  more  comprehensive  evaluation  undertaken  which  incorporates   the  larger  range  of  programs.  Moreover,  effects  of  decentralization  on  inter-­‐ community  allocations  are  no  less  important  than  effects  on  intra-­‐community   allocations.  For  instance,  the  1994  decentralization  in  Bolivia  resulted  in  a   remarkable  redistribution  of  central  resources  in  favor  of  poorer  regions  (Faguet   (2004)).  The  Indian  state  of  West  Bengal  (Bardhan  and  Mookherjee  (2006a))   manifested  a  bias  in  the  opposite  direction,  wherein  elite  capture  manifested  itself   in  distorted  inter-­‐community  rather  than  intra-­‐community  allocations.   The  nature  of  civil  society  and  democracy  which  has  been  found  to  be  a  key   determinant  of  effects  of  decentralization  in  a  large  number  of  studies  are  typically   taken  as  given.  Yet,  as  Faguet  (2014)  argues  in  his  introduction  to  a  recent   symposium  on  decentralization,  there  could  be  important  effects  on  the  nature  of   democracy  itself.  Citizens  may  participate  more  in  civic  affairs  when  they  perceive   more  of  a  stake  from  doing  so.  In  Indonesia,  the  2001  fiscal  decentralization  was   followed  by  emergence  of  new  leaders  that  were  elected  rather  than  appointed  by   local  elites  in  about  20%  of  all  villages:  these  villages  manifested  superior  gains   from  decentralization  (Pal  and  Roy  (2014)).  Decentralization  could  provide  room   for  a  new  set  of  political  leaders  to  emerge  and  prove  their  mettle,  conceivably   lowering  entry  barriers  into  national  politics  (Myerson  (2014)).    Even  if  resource   allocations  are  unaffected,  the  fact  of  wider  popular  participation  in  decision-­‐ making  could  generate  greater  satisfaction  among  citizens,  as  Olken  (2010)  finds  in   a  controlled  experiment  in  Indonesia.  This  may  help  enhance  the  legitimacy  of   government  in  the  eyes  of  citizens,  and  limit  threats  of  political  secession.  Beaman  et   al  (2009))  show  that  Indian  voters’  gender  stereotypes  regarding  political  leaders   became  moderated  as  they  become  exposed  to  local  women  leaders.    Women   leaders  may  accumulate  political  and  administrative  experience  over  time,  reducing   and  even  reversing  gender  gaps  in  governance  performance  (Afridi  et  al  (2013)).   Decentralization  provides  the  scope  for  learning  from  policy  experimentation  by   select  jurisdictions,  an  avenue  for  spatial  diffusion  of  successful  policy  innovations,   as  exemplified  by  schooling  reforms  in  Sau  Paulo,  Brazil  (Gemignani  and  Madeira   (2014)).     4.3  Comparisons  with  Other  Organizational  Alternatives   Most  studies  of  political  decentralization  examine  performance  of  local  government     across  different  jurisdictions,  and/or  over  time,  and  relate  these  to  differences  in   political  competition,  inequality  or  social  capital.  Relatively  few  studies  compare  it   with  centralization,  owing  to  lack  of  suitable  data.  Yet  that  is  one  of  the  questions  of   primary  concern.  Comparisons  with  other  alternatives  frequently  arise  in  policy   discussions,  such  as  privatization.  For  instance,  if  the  provision  of  an  infrastructure,   education  or  health  services  by  a  centralized  government  are  considered   unsatisfactory,  would  services  improve  if  responsibility  for  provision  were  devolved   to  local  governments?  Or  would  it  be  better  to  privatize  service  delivery,  while  the   government  limits  its  role  to  financing  via  budgetary  subsidies  to  private  service   providers  or  cash  vouchers  to  households?  The  trade-­‐offs  between  decentralization   and  privatization  has  not  received  much  attention  in  the  literature,  either   theoretical  or  empirical.  An  exception  is  Granados  and  Sanchez  (2014),  who   evaluate  the  comparative  performance  of  decentralization  and  privatization  in   delivery  of  water  services  in  Colombia  in  lowering  infant  mortality  rates.   4.4  Design  Issues  Highlighted  by  First  Generation  Theories    Spillovers     Inter-­‐jurisdictional  spillovers  and  loss  of  scale  economies  are  identified  as  the   principal  costs  of  decentralization  in  the  first  generation  literature.  These  have  to  be   traded  off  against  benefits  from  catering  to  greater  heterogeneity  and  improved   accountability  under  decentralization.  Compared  to  the  latter,  there  is  a  paucity  of   empirical  studies  that  evaluate  the  former.     Nevertheless,  there  has  been  some  interesting  work  on  effects  of  decentralization  on   environmental  quality  across  political  jurisdictions.  Sigman  (2001)  finds  evidence  of   free-­‐riding  across  countries  by  comparing  water  quality  in  national  and   international  rivers  in  49  countries  spanning  1979-­‐90.  Water  quality  was   significantly  poorer  in  monitoring  stations  located  upstream  of  international  rivers   outside  the  European  Union.  Using  data  from  500  river  monitoring  stations  within   the  United  States,  Sigman  (2005)  finds  similar  evidence  of  4%  deterioration  of   water  quality  in  monitoring  stations  located  downstream  from  states  authorized  to   implement  the  Clean  Water  Act.  These  results  obtain  after  controlling  for  state,  river   characteristics  and  monitoring  station  fixed  effects,  with  the  identification  coming   from  time-­‐varying  state  authority.     Lipscomb  and  Mubarak  (2007)  examine  the  effects  of  splitting  county  boundaries  in   Brazil  using  data  from  372  monitoring  stations  located  in  all  eight  major  river  basins.   They  find  a  significant  increase  in  pollution  close  to  a  river’s  downstream  exit  point   out  of  a  newly  created  county,  thereby  revealing  increased  free-­‐riding  incentives.   Yet,  they  find  no  evidence  of  any  effects  on  overall  water  quality  averaging  across  all   locations.  They  explain  these  contrasting  results  by  the  fact  that  splitting  counties   increased  aggregate  budgets  of  county  governments,  resulting  in  offsetting   increases  in  spending  on  sanitation.  Burgess  et  al  (2012)  provide  evidence  of   increased  deforestation  in  Indonesia  resulting  from  splitting  local  governments   between  2000-­‐08.  These  results  inform  the  need  to  correctly  evaluate  costs  of   decentralization,  besides  designing  decentralization  programs  to  control  trans-­‐ boundary  spillovers  via  monitoring  and  corrective  taxes  or  subsidies.   Vertical  Fiscal  Relations  and  Inter--jurisdictional  Allocations   Another  design  issue  stressed  by  the  first  generation  literature  are  arrangements  for   fiscal  transfers  from  national  to  local  governments.  In  many  Latin  American   countries  decentralization  has  been  associated  with  macroeconomic  instability   associated  with  inability  of  central  governments  to  control  spending  of  local   government  owing  to  `soft  budget  constraint’  problems  (Rodden  and  Eskeland   (2003)).  In  most  other  developing  countries  on  the  other  hand,  decentralization  has   transferred  significant  spending  responsibilities  to  local  governments  without   concomitant  increases  in  funds  devolved,  resulting  in  complaints  of  `unfunded   mandates’.     While  the  softness  of  local  government  budget  constraints  is  partly  affected  by   underlying  political  economy  factors,  these  could  also  be  affected  by  the  design  of   vertical  fiscal  transfer  mechanisms.  To  the  extent  there  is  scope  for  such  choices,   what  are  the  associated  trade-­‐offs?  Soft  budget  constraints  can  create  both   problems  of  controlling  fiscal  spending  and  inefficiencies  associated  with  bailouts  of   improvident  localities.  On  the  other  hand,  hardening  budget  constraints  limits   flexibility  in  inter-­‐jurisdictional  allocations  in  response  to  changing  needs  and   circumstances,  and  constrains  the  ability  of  local  governments  to  provide  essential   services.       Fiscal  strain  can  also  induce  increases  in  corruption,  as  shown  in  an  interesting   study  of  Indonesian  local  governments  by  Henderson  and  Kuncoro  (2004).  They   report  the  results  of  a  survey  of  1808  firms  in  2001-­‐2002,  following  a   comprehensive  nation-­‐wide  decentralization.  After  controlling  for  various  firm  and   locality  specific  characteristics,  they  find  that  increasing  the  ratio  of  property  tax   revenues  retained  by  local  governments  to  local  GDP  or  central  government   transfers  to  local  GDP,  had  a  strong  negative  effect  on  the  number  of  licenses  that   local  firms  had  to  obtain.  One  standard  deviation  increase  in  either  of  these   variables  was  associated  with  73%  and  56%  increase  in  the  number  of  licenses,   which  in  turn  was  positively  correlated  with  bribes  (a  doubling  of  number  of   licenses  raised  the  ratio  of  bribes  to  firm  costs  by  about  .8-­‐-­‐.9).  They  interpret  this  as   the  effects  of  fiscal  strain  of  local  governments  relative  to  expenditure   responsibilities  mandated  by  the  decentralization,  which  led  them  to  underpay   government  officials,  thereby  inducing  the  latter  to  rely  on  bribes  to  supplement   incomes.     A  related  issue  is  the  extent  of  discretion  retained  by  upper  level  governments  in   reallocating  transfers  across  jurisdictions.  The  use  of  formula-­‐based  transfers  in   Bolivia  was  instrumental  in  the  redistribution  of  funds  in  favor  of  poorer  regions   observed  (Faguet  (2004)).  In  contrast,  many  Indian  states  allow  upper  level   governments  to  retain  discretion  over  these  transfers,  allowing  scope  for  political   manipulation.  Gupta  and  Mukhopadhyay  (2014)  show  that  the  Rajasthan  state   government  dominated  by  the  Indian  National  Congress  (INC)  allocated  larger  funds   for  the  national  employment  guarantee  program  to  blocks  (a  layer  of  government   between  the  district  and  village  level)  where  the  INC  lost  the  previous  election  by  a   narrow  margin,  and  that  this  helped  the  INC  recover  their  vote  shares.  Retaining   discretion  over  inter-­‐jurisdictional  transfers  enable  parties  dominating  higher  level   governments  to  manipulate  these  to  secure  political  advantages,  in  a  manner  rarely   visible  to  citizens.  More  attention  therefore  needs  to  be  devoted  to  the  design  of   interjursidictional  transfers  and  their  implications.     4.5  Domain  and  Extent  of  Devolution   What  responsibilities  ought  to  be  devolved  to  local  governments?  While  authority   over  allocations  across  different  local  projects  and  their  implementation  is  devolved   in  most  decentralization  contexts,  these  can  be  subject  to  various  mandates  and   regulations  imposed  by  upper  level  governments.  Projects  may  have  to  be  approved   in  advance,  a  time  consuming  procedure  which  restricts  time  available  to  local   governments  to  execute  projects.  There  can  be  restrictions  on  the  use  of  contractors,   or  on  the  ratio  of  costs  incurred  on  material  procurement.  The  intent  of  these   regulations  is  often  to  limit  scope  for  local  corruption,  but  in  practice  can  become   instruments  of  discretionary  control  exercised  by  upper  level  governments  to   secure  their  own  political  advantage  (such  as  restricting  spending  by  local   governments  dominated  by  rival  parties).     Many  service  areas  are  often  partially  decentralized,  in  the  sense  that  responsibility   is  shared  between  central  and  local  governments.  Joanis  (2014)  provides  a   theoretical  analysis  of  the  benefits  (vertical  supply  complementarities  between   different  tiers)  and  costs  (reduced  accountability,  as  voters  cannot  pinpoint   responsibility  for  delivery  failures)  of  such  arrangements.  Despite  the  pervasiveness   of  shared  responsibilities,  very  few  empirical  studies  have  attempted  to  examine   their  consequences.   Another  important  question  is  the  desirability  of  financial  devolution:  should  local   governments  be  empowered  or  authorized  to  levy  taxes  on  local  businesses  or   residents?  Countries  vary  considerably  with  respect  to  the  extent  of  fiscal  autonomy   granted  to  local  governments.  As  reported  in  the  preceding  section,  cross-­‐country   and  country  panel  studies  find  that  expansionary  effects  of  expenditure   decentralization  on  local  services  are  higher,  the  greater  the  extent  of  revenue-­‐ raising  devolution  that  accompanies  the  decentralization.  Bardhan  and  Mookherjee   (2006b)  develop  a  theoretical  model  which  is  consistent  with  this  finding,  but  in   which  devolution  of  revenue  raising  authority  could  nevertheless  reduce  welfare.   The  reason  is  that  with  full  revenue  raising  autonomy,  local  governments  subject  to   substantial  elite  capture  are  motivated  to  impose  regressive  local  taxes  where  the   poor  end  up  paying  for  services  delivered  to  elites.  This  problem  is  avoided  by   restricting  financial  devolution.  On  the  other  hand,  if  the  local  government  is  not   subject  to  elite  capture,  financial  devolution  allows  local  governments  the  flexibility   to  adjust  spending  in  response  to  variations  in  local  need.  Moreover,  it  is  often   argued  that  citizens  would  be  motivated  to  evade  taxes  less  if  they  were  visibly   spent  on  the  jurisdictions  in  which  they  reside.  Hence  there  are  various  pros  and   cons  of  altering  the  extent  of  financial  devolution,  which  do  not  seem  to  have  been   studied  empirically.   The  converse  question  of  effects  of  political  devolution  in  a  setting  with  fiscal   decentralization  have  been  studied  in  the  context  of  China,  where  a  reform  requiring   some  local  government  leaders  to  be  elected  has  been  introduced  in  the  past  two   decades.  Using  a  village  panel  and  staggered  introduction  of  local  government   elections,  Martinez-­‐Bravo  et  al  (2014)  find  that  provision  of  local  public  goods   financed  via  local  taxes  doubled,  and  less  land  was  leased  to  enterprises  resulting  in   reduced  incomes  of  prosperous  residents  and  lower  inequality.  In  similar  vein,   Zhang  et  al  (2004)  find  that  local  elections  increased  share  of  public  investments  in   local  government  spending,  and  a  shift  of  taxes  from  household  to  enterprises.   4.6  Political  Economy  and  Implementation  Issues   Case  studies  of  recent  decentralization  experiences  of  a  number  of  poor  and  middle   income  countries  show  striking  differences  in  design  and  implementation  (Bardhan   and  Mookherjee  (2006c)).  In  some  countries,  local  governments  have  independent   constitutional  authority,  while  in  others  they  exist  at  the  mercy  of  upper  level   governments.  Local  government  leaders  can  be  elected  in  some  countries,  and   appointed  by  upper  level  governments.  Even  when  there  is  some  degree  of  local   democracy,    as  in  China  over  the  past  two  decades,  there  can  be  restrictions  imposed   on  freedom  of  political  parties  to  form  and  contest  these  elections.  Most  countries,   however,  impose  no  such  constraints.  Elected  officials  of  local  governments  may  be   constrained  to  operate  with  bureaucrats  or  service  workers  appointed  by  and   accountable  only  to  upper  level  governments  in  some  countries,  as  in  the  case  of   India  where  most  school  teachers  and  health  workers  cannot  be  hired  or  fired  by   local  governments.  In  others,  local  governments  have  far  greater  control.     Moreover,  the  extent  of  decentralization  de  facto  often  deviates  from  what  is   prescribed  de  jure.  Frequently  responsibilities  are  divided  between  local  and  higher   level  governments,  with  boundaries  that  are  not  clearly  demarcated.  Officials  in   local  governments  often  lack  requisite  administrative  skills  and  training.  The  extent   of  upward  and  downward  accountability  mechanisms  on  local  governments  can  also   differ  greatly,  as  can  the  extent  of  financial  authority  or  range  of  functions  devolved.   As  is  clear  from  the  evidence  described  by  existing  literature,  these  features  have  a   strong  impact  on  the  ultimate  outcomes  of  decentralization.   Are  there  normative  rationales  in  terms  of  underlying  `fundamentals’  for  these   cross-­‐country  or  cross-­‐regional  differences,  or  are  they  instead  the  result  of  the   political  economy  of  the  process  by  which  decentralization  was  implemented?  This   is  a  question  that  requires  more  attention.  To  do  this,  one  has  to  ultimately  confront   the  question  why  central  governments  willingly  devolve  power  to  lower  level   governments,  and  the  circumstances  under  which  they  do  so.  As  country  case   studies  (e.g.,  in  Bardhan  and  Mookherjee  (2006c))  show,  historical  contexts  play  a   critical  role  in  defining  the  motive  for  decentralization,  and  the  shape  that  it   eventually  takes.  Central  authorities  devolve  power  either  because  they  are   compelled  by  threats  to  their  own  authority  (rising  competition  from  competing   parties  or  regional  pressures),  or  seek  to  preempt  such  threats,  or  owing  to  crises,   external  shocks  and  pressures,  or  to  changes  in  political  ideology,  or  some   combination  of  these.  In  some  cases  (such  as  Bolivia,  Brazil,  Indonesia  and  South   Africa)  a  prior  authoritarian  political  regime  was  swept  away  by  a  new  government   wedded  to  an  ideology  more  aligned  with  decentralization,  resulting  in  a  `big  bang’   decentralization  reform.  Such  decentralizations  were  more  comprehensive,   devolving  a  larger  range  of  responsibilities  to  local  governments  with  independent   constitutional  authority  and  in  a  more  transparent  manner.  In  countries  such  as   Uganda  or  China  where  strong  central  authorities  faced  no  challenges,  they  resorted   to  administrative  rather  than  political  devolution  as  a  way  of  improving  the   management  of  government  functions.  In  India  and  Pakistan,  weak  central   governments  tended  to  counter  rising  opposition  from  regional  parties  by   cultivating  local  governments.  Yet  in  India,  the  central  government  was   constitutionally  required  to  seek  approval  from  the  regional  parties,  so  was  forced   to  let  each  state  implement  the  devolution  in  its  own  manner.  This  resulted  in   enormous  heterogeneity  in  the  pattern  of  devolution  across  different  states  in  India,   in  which  many  state  governments  implemented  central  statutes  and  mandates  in   ways  to  hold  on  to  as  much  power  as  possible.  It  is  a  challenging  task  to  develop  a   theory  of  the  institutional  dynamics  of  actual  decentralizations  that  can  provide  a   unified  account  of  these  disparate  experiences.