ASSESSING INCOME DISTRIBUTION FOR INDIA AT S ON O N A A THE - - PowerPoint PPT Presentation

ASSESSING INCOME S ON O N A A DISTRIBUTION FOR INDIA AT THE DISTRICT LEVEL USING THE DISTRICT LEVEL USING NIGHTTIME SATELLITE IMAGERY

Mayuri Chaturvedi, Tilottama Ghosh, Laveesh Bhandari Indicus Analytics Private Limited New Delhi India New Delhi, India

32nd APAN Meeting

Overview Overview

I t d ti Wh I Di t ib ti d Ni ht Li ht

Introduction: Why Income Distribution and Night Lights Literature review: Potential of Nighttime Lights Research Objective Research Objective Method Data Used State-level graphical analysis Developing the Model

R l

Results Model estimates and diagnostics Error maps Error maps Discussion Conclusion

Why Income Distribution and Nightlights? Why Income Distribution and Nightlights?

I l i th f th j li th t i

Inclusive growth one of the major policy thrust areas in

the current as well as next Five-Year Plan

Income distribution data not easy to come by Income distribution data not easy to come by Limitations include:

Under-reporting, Over-reporting, Misreporting Under reporting, Over reporting, Misreporting Inappropriate sampling and/or weighting Lack of standardized methodology across sampling

i i

• rganizations

Huge time lags between collection and publication, and low

frequency of data collection q y

Coarse spatial resolution, Modifiable Areal Unit Problem

Nightlights (NL) can help circumvent these problems

Potential of Nighttime Lights Potential of Nighttime Lights

• Imhoff et al 1997
• Imhoff et al 1997
• C. Small 2001
• Milesi et al 2003

Mapping urban extent and its ecological impact

• P.C Sutton 1997

Estimating urban populations and

• Sutton et al 2001
• Sutton et al 2003

Estimating urban populations and intra-urban population density

• Elvidge et al 2004

Estimating impervious surface

g

area

• Doll et al 2000
• Oda and Maksyutov 2010
• Ghosh et al 2010

Mapping Green-House Gas emissions

Ghosh et al 2010

• Cova et al 2004

Mapping fire and fire-prone areas

• Ebener et al 2005
• Doll et al 2006
• Sutton et al 2007

Estimating and mapping GDP at the national and sub- national levels

El id l 2009

• Elvidge et al 2009
• Ghosh et al 2010

Creating global grids of GDP and poverty

Research Objective Research Objective

In this paper, we take a look at the relationship between night lights and Income distribution, as captured by the b f h h ld i diff t number of households in different income brackets Use multinomial regression Use multinomial regression techniques to study the statistical relationship Map the prediction errors to Map the prediction errors to identify regions of maximum estimation errors Use socio-economic insights to Use socio economic insights to understand probable reasons behind the errors Radiance-calibrated nighttime image of India, 2004

Method I: Data used… Method I: Data used…

Radiance-calibrated Nighttime Image of India, 2004 (NOAA NGDC) 2004 (NOAA, NGDC) Households Income Data, 2004 (Indicus Analytics) 2004 (Indicus Analytics) States and District Shapefiles for India Shapefiles for India (Indicus Analytics)

Method I: Data used Method I: Data used

Th t i f h h ld

Three categories of households

defined on the basis of annual household income

U i h h ld ( i

Upper Income Households

Upper income households (earning

more than Rs 10 lakh per annum, or US $22,500)

Middle income households (earning Rs

Middle Income Households

Middle income households (earning Rs

3-10 lakh per annum, or US $6,700 – US $22,500)

Lower income households (earning less

Households Lower Income

Lower income households (earning less

than Rs 3 lakh per annum, or US $6,700)

Sum of lights extracted for the States

Households

Sum of lights extracted for the States

and the Districts

Sum of lights and number of

households in each income category households in each income category graphed at the State level

Method II: State-level graphical analysis g p y

A look at the Sum of Lights and Households’ Income data

Method II: State-level graphical analysis

Inferences

Nightlights inadequately capture household income

data across Indian states

Some examples that highlight the need to conduct a

finer spatial resolution analysis p y

Maharashtra and Andhra Pradesh (similar lights,

dissimilar incomes) d ss a co es)

Madhya Pradesh and Rajasthan (similar incomes,

dissimilar lights) dissimilar lights)

Uttar Pradesh in the graph and in the NL Image

(variegated lighting pattern) (variegated lighting pattern)

Complex role of population highlighted

Method III: Developing the Model

The relationship between Night Lights and Households data suggested a The relationship between Night Lights and Households data suggested a logarithmic model…

Scatter plot for 585 districts

• X- Sum of lights
• Y- No. of households in Upper

Scatter plot for 585 districts

• X- Natural log of sum of lights
• Y- Natural log of no. of households

income group in Upper income group

Method III: Developing the Model

…and the use of dummy variables for commercially and administratively …and the use of dummy variables for commercially and administratively important districts that are also high population zones

2.Suburbs of Metros 1.Metropolitan Districts Delhi M b i 4 State 3.Large Industrialized Towns Agra Noida Thane Gurgaon Faridabad Mumbai Kolkata Chennai Pune Other 4.State Capitals Patna Khordha Mumbai Jamnagar Kanpur Rangareddi Cuttack Nagpur Hugli Surat… Bangalore Ahmadabad Hyderabad Other Districts All the remaining… Mumbai Jaipur… Krishna…

Method III: Developing the Model

Hypotheses Proposed…

N f N

Upper

Upper Income

Number of households Contribution to Nightlights

pp Income

Middle Income

Upper Income

Middle Income Income

Lower Income

Income

Lower Income

• While we can have data on households in different income brackets, we can obtain

information only on total sum of lights in a region

• Hypothesis One: NL should be more closely associated with the richer in any given region

than with the poorer

• Hypothesis Two: NL will most likely tend to under-estimate the number of poor households

and over estimate the rich households and over-estimate the rich households

• Logarithmic multivariate regression model used for all three income categories using the

same predictor variables

Results I: Model Estimates and Diagnostics

Model Co-efficients

Ln Y = α + β1 (Ln X1) + β2 X2 + β3X3 + β4X4 + β5X5

Predictor Variables

Natural log of number of households in the 'lower' income group Natural log of number of households in the 'middle' income group Natural log of number of households in the 'upper' income group

Natural log of Sum of Lights, β1

0.45* 0.51* 0.55*

Dummy for Metro cities, β2

1.01* 1.60* 2.11*

Dummy for Suburbs of Metros, β3

0.37$ 1.35* 1.67*

Dummy for Suburbs of Metros, β3 Dummy for Large Towns, β4

0.18# 0.55* 0.72*

Dummy for Capital cities, β5

‐0.30$ 0.46* 0.76*

I t t

8 08* 4 26* 1 86*

* Significant at the 99% Confidence Interval, $ Significant at the 95% Confidence Interval, #

Intercept, α

8.08* 4.26* 1.86*

Adjusted R2

0.61 0.75 0.76

Significant at the 90% Confidence Interval

Results I: Model Estimates and Diagnostics

Inference

Ti ht i f l ti hi b t NL d h h ld ’

Tightening of relationship between NL and households’

categories as the income goes up as seen in higher adjusted R2 values for middle and upper income category models

Magnitude of the co-efficient for NL (β1) increases for the

more affluent segments

Most of the predictor variables significant at the 99% level of Most of the predictor variables significant at the 99% level of

significance

Co-efficients of all dummy variables also go up monotonically

for higher income group

Lights are better able to estimate households in more affluent

categories (Hypothesis One) categories (Hypothesis One)

Β’s consistently highest for the Metropolitan dummy followed

by dummy for Suburbs of Metros for all three models

Results I: Model Estimates and Diagnostics

Scatter graphs for the three income category households with Nightlights

Results II: Error Maps

From a predominant pattern of under-estimation to an emerging pattern of From a predominant pattern of under estimation to an emerging pattern of

• ver-estimation

Results II: Error Maps Results II: Error Maps

L I H h ld

Lower Income Households

Under-estimation among states: Bihar, parts of UP

, Jharkhand, WB, Orissa, Assam, AP and Kerala

Over-estimation among states: Punjab, Haryana, HP

, parts of Gujarat, Rajasthan, some N-E states’ districts

Middle Income Households

Under-estimation among states: UP

, Bihar, Jharkhand, WB, Kerala and some NE districts

Over-estimation among states: Changlang (Arunachal), Udhampur (J&K),

Senapati (Manipur), Yanam (Pondicherry)

Upper Income Households

Under-estimation among states: Kerala, parts of UP

, Punjab, HP , Bihar, U de es a o a o g s a es: e a a, pa s o U , u jab, , a , Jharkhand, WB, Assam, Nagaland

Over-estimation among states: Gujarat, Rajasthan, Karnataka

Discussion Discussion

A d l ti hi i t b t i htti li ht d i

A good relationship exists between nighttime lights and income

distribution at the district level, with the relationship being stronger for households in the higher income brackets U d f b f h h ld l

Under-estimation of number of households in lower income category

for highly populated states with over 70% rural population

Over-estimation of lower income households in sparsely populated

states

This under-estimation much less for middle income households, and

an emerging pattern of variegated over-estimation appears g g p g pp

This pattern of over-estimation spreads further to more districts for

upper income households

Under estimation of upper income households by NL observed in Under-estimation of upper income households by NL observed in

high population density states of UP , Bihar and Kerala

Thus, it can be said that Hypothesis Two also holds true for Indian

di t i t districts

Conclusion Conclusion

F l l l f h l h ff

Finer spatial resolution analysis of nightlights more effective

in applying this technology as a proxy for economic activity

Examination of the relationship between nighttime lights and Examination of the relationship between nighttime lights and

income distribution at the district level for India shows promising results

The relationship is non-linear at the district level, and shows

strong correlations with each income category

Highest co-efficient of determination (R2) exhibited for the

model with the highest income cut ff ff f

Nightlights co-efficient, along with co-efficients of all dummy

variables also highest for upper income category

Thus nightlights shown to be more closely associated with Thus, nightlights shown to be more closely associated with

the richer in a region than with the poor

Conclusion Conclusion

Ni htli ht d ti t th b f

Nightlights under-estimate the number of poorer

households in most Indian districts

The error maps demonstrate a progressive movement The error maps demonstrate a progressive movement

towards over-estimation by nightlights with a mixed pattern of over- and under-estimation for middle and i h h ld upper income households

High population density, lack of govt. provision of

public amenities high human development affluent public amenities, high human development, affluent farmers, big share of rural population, presence of military base some of the characteristics noticed of f districts with anomalous estimates of economic activity by nightlights