Growth Redistribution and Inequality Effects on Poverty in Nigeria

Growth, Redistribution and Inequality

Effects on Poverty Changes in Nigeria

Jude Okechukwu Chukwu

PhD Candidate

Department of Economics

University of Nigeria

[email protected]

Visiting Research Fellow

IPC-IG, Brasilia/Brazil

IPC-IG, SEMINAR SERIES, 28th July, 2014

Presentation Outline

1. Introduction: context and motivation

2. Study objectives

3. What does theory tell us?

4. Empirical Evidence 4. Empirical Evidence

5. Estimation Method: Models and data

6. Empirical Results

7. Conclusion and Policy Implications

Introduction/1 � The economy has been growing on average between 6.5% and7.2% for a decade (NPC, 2012).

� GDP growth rate was 3.7% in 2004. By 2010, it had increased to7.8% (World Bank, 2013). Rise of 4.1%, which is abovecontinental Africa’s average growth rate of 4.0%.

� Trickle-down theory: increase in growth rates enables the poor to� Trickle-down theory: increase in growth rates enables the poor toderive maximum benefits.

� Growth is good for the poor (Dollar & Kraay, 2000)

� Yet in rare cases the economic growth might increase inequalityand offset gains of the poor from the economic growth (Esanov,2006).

� The paradox is that the poverty level in Nigeria contradicts thecountry’s immense wealth (Obadan, 2004).

Introduction/2

� Fig. 1: Nigeria GDP Growth Annual (%)

Introduction/3

� Fig. 1: Change in GDP Growth (%) 2004 - 2010

Introduction/4

� Figure 3 show that poverty increased at National levelby 14.6%; in rural sector by 9.9% and urban sectorby 18.6%

� Poverty changes were positive in Rural and Urban� Poverty changes were positive in Rural and Urbansectors, Southern zones but negative in all theNorthern zones

� South-east showed worst evidence of povertyincrease by 29.72%, while South-west showed leastincrease by 3.10% followed by South-south (17.7%)

Introduction/5: Poverty Trends and Changes: 2004 & 2010/Fig 3

50

60

70

80

-20

-10

0

10

20

30

40

National Rural Urban South

South

South

East

South

West

North

Central

North

East

North

West

14.6

9.9

18.6 17.72

29.72

3.1

-10.68

-5.52-3.52

Poverty 2004

Poverty 2010

Poverty Change

Introduction/6

� Inequality rose by 4.1% at national level; increasedby 2.2% and 4.2% at rural and urban sectorsrespectively

� Inequality changes were positive in all Southern� Inequality changes were positive in all Southernzones, North east and North West but negative inNorth central (about -5.4%)

� Highest increase by 18.1% was in the South eastfollowed by South south (oil rich/coastal region) withan increase of 12.8% while South west (Lagos axis)had least increase in inequality by 0.2%

Introduction/7: Inequality Trends and changes: 2004 & 2010/Fig 4

10

15

20

12.8

18.1

8.6

-10

-5

0

5

National Rural Urban South

South

South

East

South

West

North

Central

North

East

North

West

4.1

2.2

4.2

0.2

-5.4

0.7

Inequality 2004

Inequality 2010

Inequality Change

Introduction/8

� Nigeria’s growth paradox is a policy concern since average growth rates have been trending upwards, while poverty and inequality deteriorate

� HNLSS, 2010 indicted the high growth rates between 2004 and 2010 - both poverty & inequality were reported to have and 2010 - both poverty & inequality were reported to have increased: poverty rose by 14.6% and inequality by 4.1%

� Why did the impressive growth in the 2000s not lead to decline in poverty and inequality in Nigeria?

� Has growth been pro-poor and inclusive in Nigeria?

Introduction/9

Departs from Adigun et al (2011); Odozi & Awoyemi (2010); Aigbokhan (2008) in three main respects:

� employed latest 2010 HNLSS for analyses. Previous Nigerian studies used 1996 NLSS and 2004 HNLSS

� considered poverty elasticity with respect to within and between-group inequality using models by Araar (2007)

� evaluates pro-poorness and inclusiveness of growth between 2004 and 2010

Study objectives

i. decomposes poverty changes into growth effect and redistribution effect

ii. estimate marginal FGT impact and FGT elasticity with respect to within-and-between group with respect to within-and-between group inequality

iii. estimate the pro-poorness and inclusiveness of growth

What does theory tell us?/1� Poverty-growth-inequality triangle (Bourguignon, 2004): inequality has direct impact on poverty as well as indirectly on poverty through growth

� Kuznet’s (1955) inverted-U predicts rise in inequality at early periods of high and rising growth rates and fall in inequality during later periodsduring later periods

� Okun’s (1975) treatise on equality and efficiency argue that greater income inequality is an incentive for work and investment

� Trickle down hypothesis states that inequality is good for growth

What does theory tell us?/2� Barro (2000): effect of inequality on growth may be non-linear: it may be bad for growth in poor countries but good for growth in rich countries

Four stylized facts on the relative importance of growth and inequality on poverty exists:

*1. Growth effect is dominant

*2. Growth is less effective in reducing poverty in high inequality economies

*3. Growth is less effective in reducing poverty in LDCs than in mature economies

*4. Growth has larger impact on poverty reduction in rural areas, while distribution has larger impact on poverty reduction in urban areas

Empirical Evidence� Studies that find that poverty change was mainly due to inequality include Esanov (2006) for Kazakhstan, Anwar (2010) for Pakistan, Kirama (2013) for Tanzania etc

� Studies that find that poverty change was mainly due to growth include Baye (2006) for Cameroon, Epoh & Baye (2007) for Cameroon, Odozi and Awoyemi (2010) for Nigeria and so on. Cameroon, Odozi and Awoyemi (2010) for Nigeria and so on.

� Cheema and Sial (2010) for Pakistan found that growth and redistribution effects were negative indicating that both effects reinforced each other to reduce poverty

� Studies that support the prospective viewpoint are Araar (2012) and Klasen & Misselhorn (2008).

Estimation Method/1

Model for objective 1: Shapley Value approach by Shorrocks(1999) and Datt & Ravallion (1992) approach were employed for growth and redistribution decomposition.

Model for objective 2: Araar’s (2007) models for poverty and inequality within-group and between-group were employed. inequality within-group and between-group were employed.

Model for objective 3: Pro-poor indices such as Ravallion and Chen (2003); Ravallion & Chen index – g; Kakwani and Pernia (2000); Kakwani and Pernia (2000) - 1; Kwakani, Khander & Son–PEGR-(2003); PEGR – g; and the Growth Incidence Curve were used

Estimation Method/2

Data and their features: * 2004 NLSS & 2010 HNLSS obtained by NBS were employed

* Both covered 36 states and FCT (37 Strata), and used two-stage multi-stage sampling technique with EAs as Psu-first stage and Households as Usu-second stage Households as Usu-second stage

* 2004 NLSS surveyed 19, 158 households: 4646 in Urban (24.2%) and 14, 512 in rural (75.8%)

•

* 2010 HNLSS surveyed 34, 619 households: 9,348 in Urban (27%) and 25, 271 in rural (73%)

* Both were survey set for sample design prior to estimations

Empirical Results/1

Table 1: Poverty Decomposition into Growth and Redistribution Components, 2004-2010, NIGERIA

FGT Growth Effect Redistribution Effect Residual Difference in Poverty (d2 – d1)

Datt & Ravallion Shapley Datt & Ravallion Shapley Datt & Ravallion (d2 – d1) Datt & Ravallion Shapley Value

Datt & Ravallion Shapley Value

Datt & Ravallion (d2 – d1) 2010 - 2004

P0 -0.3381 (-0.3262)

-0.3321

0.1006 (0.1125)

0.1065 0.0119 (-0.0119)

-0.2256

P1 -0.1624 (-0.1984)

-0.1804 0.0875 (0.0514)

0.0695 -0.0360 (0.0360)

-0.1109

P2 -0.0974 (-0.1343)

-0.1159 0.0657 (0.0288)

0.0473 -0.0369 (0.0369)

-0.0686

Source: Author’s calculation based on NLSS 2004 and HNLSS 2010 Note: Values in brackets are for 2010 as reference period

Empirical Results/2� Table 1 show that P0 fell by 22.56%; the growth = 33.21% in poverty reduction while, redistribution adversely accounted for 10.65%; thus dampened positive impacts of growth on poverty

� It is likely that national poverty would have fallen the more due to growth, iff, inequality had not varied given the realized decrease in poverty of -0.2256 as against the potential of -0.3321 (growth)in poverty of -0.2256 as against the potential of -0.3321 (growth)

� Poverty gap declined by 11.09%, growth accounted for 18.04% in poverty reduction; while redistribution adversely accounted for 6.95%

� Poverty severity fell by 6.86%, the pure growth effect accounted for 11.59% in poverty reduction, while redistribution adversely accounted for about 4.73%

Empirical Results/3� The results suggest that had inequality not increased, the reduction in poverty would have been more

� Tables 2 and 3 show the sectoral decomposition (urban and rural) of poverty changes, and follow same pattern as for Nigeria

� Poverty reduction was more due to growth than due to redistribution

� In urban sector, poverty declined by 35.74%, 16.56% and 10.2% for headcount ratio, poverty gap and poverty severity respectively

� In rural sector, poverty declined by 18.04%, 9.25% and 5.74% for headcount ratio, poverty gap and poverty severity respectively

� Poverty reduction declined faster in urban than rural areas

� Study finds that for all FGT poverty measures, redistribution reduced the positive impacts of growth on poverty

� This finding is consistent with Anwar (2010) for Pakistan

Empirical Results/4

Table 2: Poverty Decomposition into Growth and Redistribution Components, 2004-2010, URBAN Sector/1

FGT Growth Effect Redistribution Effect Residual Difference in Poverty (d2 – d1)

Datt & Ravallion Shapley Value

Datt & Ravallion Shapley Value

Datt & Ravallion (d2 – d1) 2010 - 2004

P0 -0.4719 (-0.4718)

-0.4718 0.1143 (0.1144)

0.1141 0.000011 (-0.000011)

-0.3574

P1 -0.2045 -0.2433 0.1166 0.0778 -0.0776 -0.1656 P1 -0.2045 (-0.2822)

-0.2433 0.1166 (0.0389)

0.0778 -0.0776 (0.0776)

-0.1656

P2 -0.1177 (-0.1888)

-0.1533 0.0868 (0.0157)

0.0512 -0.0712 (0.0712)

-0.1020

Source: Author’s calculation based on NLSS 2004 and HNLSS 2010 Note: Values in brackets are for 2010 as reference period

Empirical Results/5

Table 3: Poverty Decomposition into Growth and Redistribution Components, 2004-2010, RURAL Sector/2

FGT Growth Effect Redistribution Effect Residual Difference in Poverty (d2 – d1)

Datt & Ravallion Shapley Value

Datt & Ravallion Shapley Value

Datt & Ravallion d2 – d1 2010 - 2004 Value Value 2010 - 2004

P0 -0.2853 (-0.2786

-0.2820 0.0982 (0.1049)

0.1016 0.0067 (-0.0067)

-0.1804

P1 -0.1428 (-0.1679)

-0.1553 0.0753 (0.0502)

0.0628 -0.0251 (0.0251)

-0.0925

P2 -0.0871 (-0.1134)

-0.1003 0.0560 (0.0298)

0.0429 -0.0262 (0.0262)

-0.0574

Source: Author’s calculation based on NLSS 2004 and HNLSS 2010 Note: Values in brackets are for 2010 as reference period

Empirical Results/6� Findings in South south, South west, North central and North west are consistent with those of Nigeria, urban and rural sector

� Results for South east and North east differ a little from others: for the south east, poverty reduced by 2.41% for P0, while, P1 and P2 increased marginally by 1.69% and 2.26% respectively

� For North east, change in poverty reduced by 21.7% for P0, the P1 & P2 decreased by 8.45% and 4.13% points respectively. Redistribution impact supplemented the growth effect

� Despite that growth made inequality to deteriorate, the distributional change was targeted towards exiting the marginally poor households out of poverty. This is consistent with the findings of Kang and Imai (2010) for NU minority in Vietnam.

Empirical Results/7

Table 4: Elasticity of poverty with respect to within-group and between-group inequality

Group MII MIP P0

ELS P0

MIP P1

ELS P1

MIP P2

ELS P2

South south 0.00708 0.000328 0.450556 0.000767 2.281928 0.000797 3.998963 South east 0.000688 0.000227 0.321574 0.000729 2.232403 0.000792 4.091086

Source: Author’s calculation using DASP in STATA 13.1 NOTE: MII means marginal impact on inequality; MIP represents marginal impact on poverty; ELS means elasticity; P0 = poverty headcout, P1 = poverty gap; and P2 = poverty severity.

South east 0.000688 0.000227 0.321574 0.000729 2.232403 0.000792 4.091086 South west 0.001113 0.000211 0.184137 0.001186 2.246063 0.001380 4.409713 North central 0.000727 0.000425 0.568770 0.000749 2.169840 0.000745 3.642724 North east 0.000869 0.000457 0.512063 0.000893 2.166069 0.000931 3.808073 North west 0.001121 0.000967 0.839879 0.001159 2.178917 0.000996 3.158547 Within 0.005226 0.002616 0.487000 0.005483 2.210817 0.005641 3.837036 Between 0.000135 0.000076 0.548316 0.000143 2.245673 0.000127 3.364633 Population 0.005421 0.002526 0.453362 0.005704 2.217143 0.005942 3.896517

Empirical Results/8 Elasticity of poverty with respect to

within- and between-group inequality

� Simulated elasticities are all positive

� Results are sensitive to the choice of poverty measure

� Between-group elasticities are numerically larger than within-group elasticities for P0 and P1

� Between-group elasticity is less than within-group in the case of poverty severitypoverty severity

� heterogeneity exists in impact and elasticity estimates across geo-political zonal distributions using all the FGT poverty measures

� Heterogeneity exists despite that same redistributive processes, same “inequality and poverty aversion” parameters and same poverty line are employed.

� Study finds that heterogeneity is due to variations in initial sub-group distributions, not due to disparities in the quality of growth nor differences in redistributive policies

Empirical Results/9

Table 5: Pro-poor indices

NIGERIA 2004 - 2010

Pro-poor indices Poverty Incidence (P0)

Poverty Depth (P1)

Poverty Severity (P2)

Growth Rate of Mean Income (g) 1.1125 1.1125 1.1125 Growth Rate of Mean Income (g) 1.1125 1.1125 1.1125

Ravallion & Chen (2003) index 0.3463 0.3463 0.3463

Ravallion & Chen (2003) - g -0.7662 -0.7662 -0.7762

Kakwani & Pernia (2000) index 0.6672 0.5514 0.5396

Kakwani & Pernia (2000) index -1 -0.3328 -0.4486 -0.4604

Kakwani, Khander & Son (2003) PEGR Index

0.7423 0.6134 0.6003

PEGR - g -0.3702 -0.4990 -0.5121

Empirical Results/10� Growth is pro-poor if the resulting growth rate is greater than the mean income growth rate (Stoterau, 2010; Kabubo-Mariara et al, 2012).

� In table 5, growth in mean income is everywhere greater than all estimated pro-poor indices

� Results show positive growth rate between 2004 and 2010 suggesting that mean income increased suggesting that mean income increased

� Ravallion & Chen (2003) index is 0.3463 which is less than growth rate of the mean income (1.1125): Growth was anti-poor instead of pro-poor

� Growth process b/w 2004 and 2010 could not have been pro-poor since the Kakwani and Pernia (2000) indices are less than unity in all FGT poverty measures

Empirical Results/11

Absolute and Relative Pro-poorness of growth

� Absolute pro-poorness of growth: Ravallion & Chen (2003) index, Kakwani & Pernia (2000), and Kakwani, Khander & Son (2003)

� Relative pro-poorness of growth: Ravallion & Chen (2003) index minus g; Kakwani, Khander & Son (2003) minus g; and the Kakwani & Pernia (2000) minus unity

� Estimates of 3 indices of absolute pro-poorness are statistically greater than zero signifying that the change decreased absolute poverty; while, the estimates of the indices of relative pro-poorness are –ve but not statistically different from zero

� Growth between 2004 and 2010 has not been sufficiently relative pro-poor. Result is consistent with Duclos and Verdier-Chouchane(2010) for Mauritius.

Conclusion and Policy implications� Overall, economic growth does not seem to be accompanied by poverty reduction

� Poverty is worsening in Northern geo-political zones, and they lag behind their southern counterparts in the fight against poverty

� Inequality is also worsening in the Northern zones than in the Southern zones

� Growth has neither been pro-poor nor inclusive

� While growth has been anti-poor, inequality has been rising and inhibiting the growth process

� It is likely that robust inequality reducing policies to complement growth promoting policies are needed

� The low response of growth to poverty could be due to poor targeting of vulnerable groups and lack of quality institutional frameworks

Thank you for your rapt attention

� OBRIGADO!!!!!