RESEARCH POSTER PRESENTATION DESIGN © 2015 www.PosterPresentations.com 
Definition 
Pro-poor growth is defined as broad-based economic growth that 
increases the poor’s income and welfare proportionately more than the 
non-poor’s. In other words, it is an inclusive growth where the poor do 
not get left behind.  
 
Measurement error refers to the difference between the 
measured/reported value of a variable and its true value. In a 
regression, the measurement error of a predictor variable causes the 
regression analysis to be inaccurate or biased. 
 
Example:   
In the above regression,     indicates the effect of an increase in a unit 
of income on savings. However, if income is measured with error,  
ordinary least square estimates of    will be biased and this will not give 
an accurate estimate of the impact on  savings.  
 
In the case of measurement errors, instrumental variables are used 
to get an unbiased estimate of  . These are variables that are 
correlated with our regressor, income and, uncorrelated with our 
regressand, savings.  
 
Background on Indonesia 
 
Indonesia has experienced an average annual real GDP per capita 
growth rate 5.4% over the last 15 years and is one the world’s 20 
largest economies. 
 
 
 
 
 
 
 
However, there are concerns of the growth being unequal. The Gini 
coefficient, an indicator of inequality, has risen from 0.31 points in 2000 
to 0.43 in 2015. Consumption is also very unevenly distributed, with 
the richest 10% now consuming as much as the poorest 54%, rising 
from 42% in 2000  
 
Introduction 
Research Question 
Data Used: Indonesia Family Life Survey 2007 and 2014 
 
Cross-Section Analysis 
- Compare mean of per capita expenditure of a quintile in 
the 1st year with the mean for households in that same 
quintile in the second year.  
- Shows the distribution of income in a country  
- Reflects the changes in patterns of inequality over time 
 
 
 
 
 
 
 
 
 
 
 
 
Panel-Data Analysis 
- Compare, for a given quintile, the same households’ 
mean per capita expenditure in the first and second year 
- Conveys the degree of mobility for the poor to move up 
into higher quintiles 
 
 
 
 
 
 
 
 
 
 
 
 
Measurement Error 
Simulate a joint distribution of the true values of per capita 
expenditure in 2007 and 2014, by making inferences of the 
mean, variance of the variables and its error terms. 
 
 
 
 
 
 
Methodology 
 Cross-Section Analysis 
 
 
 
 
 
 
 
 
 
 
 
 Panel Data Analysis 
 Panel Data without Correcting for Measurement Error 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Results 
Conclusions 
 Cross-section analysis shows: 
• Overall growth rate of expenditure between 2007 and 
2014 is, on average, 4.63% per year.  
• The growth rates among the 5 quintiles have been equal 
with the third quintile (middle 20%) experiencing the 
highest growth 
• Indonesia’s growth between 2007 and 2014 has, 
arguably, not been pro-poor since the bottom 20% has 
not experienced a higher growth than the other 80%.  
 
 Panel-Data analysis shows:  
 Without correcting for measurement error, the bottom 
20% experienced a high growth, nearly double the 
median of growth rates among the 5 quintiles. 
 Thus, Indonesia’s growth has been pro poor 
 
 However, simulation accounting for measurement error, 
shows that the previous panel-data analysis has 
overestimated the growth rates. 
 Growth for the 1st quintile is as high as suggested and 
growth for 5th quintile is no as low as previously indicated. 
 Growth for the 1st quintile is higher than than the other 4 
quintiles. So, shows that growth has been pro-poor. 
 
 Acknowledgements 
• I would like to thank my mentor Professor Paul Glewwe for his 
tremendous support and constant guidance throughout this research 
project 
• I would like to thank the Undergraduate Research Opportunity 
Project (UROP) for funding this research project.  
 
 
Department of Applied Economics, University of Minnesota 
Ariza Gusti, Prof. Paul Glewwe 
Pro-Poor Growth in Indonesia 
0
1
2
3
4
5
6
7
GDP Growth Rate 
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
1999 2002 2005 2007 2008 2009 2010 2011 2012 2013 2015
Gini Coefficient Index 
Quintile 1 
HH 1 
HH 2 
HH 3 
Quintile 2 
HH 4 
HH 5 
HH 6 
Quintile 3 
HH 7 
HH 8 
HH 9 
Quintile 4 
HH 10 
HH 11 
HH 12 
Quintile 1 
HH 1 
HH 2 
HH 4 
Quintile 2 
HH 3 
HH 5 
HH 6 
Quintile 3 
HH 7 
HH 8 
HH 10 
Quintile 4 
HH 9 
HH 11 
HH 12 
Mean 
Q1 
Mean 
Q2 
Mean 
Q3 
Mean 
Q4 
Mea
n Q1 
Mea
n Q4 
Mea
n Q3 
Mea
n Q2 
Quintile 1 
HH 1 
HH 2 
HH 3 
Quintile 2 
HH 4 
HH 5 
HH 6 
Quintile 3 
HH 7 
HH 8 
HH 9 
Quintile 4 
HH 10 
HH 11 
HH 12 
Quintile 1 
HH 1 
HH 2 
HH 4 
Quintile 2 
HH 3 
HH 5 
HH 6 
Quintile 3 
HH 7 
HH 8 
HH 10 
Quintile 4 
HH 9 
HH 11 
HH 12 
Mean Q1 
HH1 
HH2 
HH3 
Mean Q1 
HH1 
HH2 
HH3 
Mean Q2 
HH4 
HH5 
HH6 
Mean Q3 
HH7 
HH8 
HH9 
Mean Q4 
HH10 
HH11 
HH12 
Mean Q2 
HH4 
HH5 
HH6 
Mean Q3 
HH7 
HH8 
HH9 
Mean Q4 
HH10 
HH11 
HH12 
quantile anngrw 
1 10.67% 
2 7.91% 
3 6.18% 
4 4.30% 
5 0.80% Data 
• Panel Attrition 
 
 
 
  
 
  Households 
# of HH Interviewed in 2007 13,270 
# of HH Interviewed in 2014 15,921 
# of HH Split After 2007 3,950 
# of HH Interviewed Only in 2007 1,852 
# of HH Interviewed Only in 2014 305 
# of HH Interviewed Both in 2007 and 2014 11,666 
quantile Mean PCE07 
Mean 
PCE14 anngrw 
1 247,689 335,571 4.43% 
2 389,618 540,038 4.77% 
3 542,975 756,134 4.84% 
4 777,958 1,073,418 4.71% 
5 1,718,484 2,383,717 4.79% 
Overall 735,276 1,009,687 4.63% 
  All Households Panel Household* 
  2007 2014 2007 2014 
Per-Capita 
Expenditure 735,276 1,009,687 697,377 1,221,980 
Household Size 3.74 3.7 3.87 3.91 
*Panel households include 10,972 households that are found in both 
2007 and 2014 surveys i.e hhid07=hhid14 
quantile Mean PCE07 
Mean 
PCE14 anngrw 
1 245,110 536,909 11.85% 
2 381,829 729,807 9.70% 
3 526,532 868,356 7.41% 
4 746,835 1,022,716 4.59% 
5 1,587,439 1,578,273 -0.08% 
Table 1 
Table 2 
Table 3 
Table 4 
Table 5 
0
200
400
600
800
1000
1200
Real GDP at 2010 US$ (Billions) 
Results (Continued) 
• Simulated Panel Data Correcting for Measurement Error 
 
 
 
 
 
 
 
Methodology (Continued) 
Where 
      is the observed value of expenditure at time 1 
      is the true value of expenditure at time 1  
      is the random measurement error 
Where 
      is the error term of the regression 
      is the coefficient attained from 2SLS (IV) regression 
      is the true value of expenditure at time 2  
Where 
      is the coefficient from OLS regression 
        is the variance of corresponding variable  
       
Acknowledgements 
• Strauss, J., F. Witoelar, & B. Sikoki. “The Fifth Wave of Indonesia Family Life 
Survey(IFLS5): Overview and Field Report”. March 2016. WR-1143/1-NIA/NICHD 
• Strauss, J., F. Witoelar, B. Sikoki & A.M. Wattie. “The Fourth Wave of Indonesia 
Family Life Survey(IFLS5): Overview and Field Report”. April 2009. WR-675/1-
NIA/NICHD 
• Glewwe, P., & Dang, H-A. H. (2011) “Was Vietnam’s Economic Growth in the 
1990s Pro-Poor? An Analysis of Panel Data from Vietnam.” Economic 
Development and Cultural Change, 59, 583-608  
 
 
 
High 
Economic 
Growth 
Pro-Poor 
Growth?