Traveling Agents: Political Change and Bureaucratic Turnover in India

1
Traveling Agents Political Change and
Bureaucratic Turnover in India

• Lakshmi Iyer (Harvard Business School)
• Anandi Mani (Warwick University)
• IPES Conference
• November 2007

2
Politicians and bureaucrats in India

• We use data from the Indian Administrative
  Service to document the relationship between
  politicians and bureaucrats.
• India has a career civil service with strong
  constitutional provisions against political
  influence on bureaucrats. In particular,
  politicians cannot hire or fire bureaucrats.
• However, politicians can influence the assignment
  of bureaucrats across positions.
• We document a significant rise in bureaucrat
  reassignments (transfers) when a new politician
  assumes office. This is a politician effect,
  not a party effect.
• We present some evidence that politicians trade
  off the benefits of control over policy
  implementation with the cost of losing
  job-specific expertise, when making transfer
  decisions.

3
Data on bureaucrat assignments

• Complete career histories of all IAS officers in
  October 2005 (Ministry of Personnel) 4149
  officers in 19 major states.
• 74 of officers are recruited via competitive
  examinations they are assigned to specific
  states, where they spend most of their careers.
  Only 1/3rd of examination recruits are assigned
  to serve in their home state.
• Position-level data on district administrators,
  Chief Secretaries and Finance Secretaries over
  time.
• Data on bureaucrats ranking during recruitment
  as a measure of ability.

4
Data on political events

• Annual data on 19 states 1980-2004.
• State and national elections change of Chief
  Ministers or Prime Minister change in the party
  in power in the state.
• Elections are scheduled every 5 years in Indias
  parliamentary system.
• The Chief Minister (CM) of a state can also
  change due to non-election related events such as
  within-party political infighting or coalition
  changes. In our data, a state has a new CM every
  3 years on average.

5
Bureaucratic transfers vary over time
6
Political change and bureaucratic transfers
7
Regression specification

• State-level Regressions for mean transfers in
  state k and year t with state and year fixed
  effects
• MeanTransferkt ak bt cPolChangekt ekt
• District-level Regressions for transfer dummy in
  district d of state k and year t with district
  and year fixed effects, and interactions with
  district characteristics
• Transferdkt ad bt cPolChangekt
  dPolChangekt Xdt edkt
• Individual level Regressions for officer i in
  state k and year t, with officer fixed effects
  and interactions with officer/post
  characteristics
• Transferikt ai bt cPolChangekt
  cPolChangekt Xit uikt

8
(No Transcript)
9
(No Transcript)
10
(No Transcript)
11
Conclusions

• Bureaucrat turnover is significantly increased by
  changes in politician identity. Politicians
  appear to take costs and benefits into account
  when making bureaucratic reassignments.
• We do not find significant negative impacts of
  such politically-induced bureaucrat transfers
  on broad district level indicators of policy
  implementation and development outcomes (progress
  of immunization programs and road projects,
  poverty reduction).
• Ongoing/future work classify posts into
  good/prestigious and not so good and document
  transfers across these two types. Document
  personal links between politicians and
  bureaucrats and see whether that matters for
  bureaucrat assignments.