1

Research question and its motivation

Clearly state your research question

Then explain why this question is important and original

e.g.

Has democratization in Africa reduced infant mortality?

(NOT "Does democracy improve people's welfare?")

e.g.

Infant mortality in Africa is high: 100 per 1,000 live births

There is no convincing evidence of the impact of democracy on living standards

1

Research question and its motivation

Clearly state your research question

Then explain why this question is important and original

e.g.

e.g.

It's a big mystery why non-democratic China manages to sustain economic growth over several decades

There is very little evidence on political selection in non-democratic countries

Who becomes a top politician in China?

(NOT "What's the political factor for growth in China?")

2

Conceptual framework

For empirics

e.g.

The median voter theorem

probabilistic voting model, and

political agency model

all predict democratization induces politicians

to pick policies favored by the poor

In Africa, one such policy is to reduce infant mortality

2

Conceptual framework

For empirics

e.g.

Build a career concern model without moral hazard

to derive a prediction:

Top politicians promote the best-performing officials

among those connected to themselves

(I do not observe Chinese politicians' behavior)

3

Data

For empirics

Data source

Sample construction

Which areas: 28 African countries

What period: 1960-2004

Sampling method: Babies born to a nationally representative sample of women in each country around 2000

Variables to be used in the analysis

Indicator of death within 12 months after birth

Demographic and Health Surveys

3

Data

For empirics

Data source

Sample construction

Which areas: 31 Chinese provinces

What period: 1993-2009

Sampling method: Provincial party secretaries and governors in office with data from China Vitae (check sample selection bias in terms of provincial economic growth rate)

Variables to be used in the analysis

Indicator of promotion to higher positions in Beijing

Indicator of being a former colleague of top politicians

China Vitae (Online database of Chinese politicians' CV)

4

For empirics

Equations to be estimated

Empirical method

Estimation methods

OLS (mother fixed effects estimation)

Standard errors clustered at the country level

Also report p-value from wild cluster bootstrap-t of Cameron et al (2008)

Statistics to be reported in the table

4

For empirics

Equations to be estimated

Empirical method

Estimation methods

OLS (Probit and Logit in the appendix) 

Standard errors clustered at the province level

Also report p-value from wild cluster bootstrap-t of Cameron et al (2008)

Statistics to be reported in the table

n.a.