Transcript Income diversification patterns in rural Sub-Saharan

Income diversification patterns in
rural Sub-Saharan Africa:
Reassessing the evidence
Benjamin Davis, Alberto Zezza and Stefania di Giuseppe
Annual Bank Conference on Africa
Paris, June 23, 2014
AG RICULTURE
IN AFRICA
T E L L I N G FA C T S
FROM MYTHS
“East Asian countries grew rapidly by replicating, in
a much shorter time frame, what today’s advanced
countries did following the Industrial Revolution.
They turned their farmers into manufacturing
workers, diversified their economies, and exported
a range of increasingly sophisticated goods. Little of
this process is taking place in Africa. … Optimists say
that the good news about African structural
transformation has not yet shown up in
macroeconomic data.” (Rodrik, 2013)
Page 2
Is Africa different when it comes to
rural income diversification?
• Revisit the facts: Are rural households in Africa
diversifying less out of agriculture than elsewhere?
• Spatial aspects of income diversification in Africa
– Agricultural potential
– Distance from urban centers
– Small vs large cities
• Implications:
– Structural change
– Welfare
– Approach to rural development
Page 3
Diversification and RNF literature:
Not so many myths after all
• Large rural non-farm (or off-farm) sector (though
estimates vary)
• Positively related to household income and GDP
• Role of assets (edu, land, infrastructure)
• Barriers to entry, dualism (high/low skills/returns)
• Likely good for poverty reduction; mixed evidence on
inequality
• But despite efforts:
– Data issues remain (comparability, measurement issues)
– Is there an African specificity?
– Not much on spatial analysis
Page 4
Countries included in the study
•
•
•
•
•
•
•
•
•
Ethiopia (2011)
Ghana (1992, 1998 and 2005)
Kenya (2005)
Madagascar (1993)
Malawi (2004 and 2011)
Niger (2010-11)
Nigeria (2004 and 2011)
Tanzania (2009)
Uganda (2005-06 and 2009-10)
•
•
•
•
•
•
•
•
•
•
•
•
•
Nepal (1996 and 2003)
Bangladesh (2000 and 2005)
Tajikistan (2003 and 2007)
Pakistan (1991 and 2001)
Nicaragua (1998, 2001 and 2005)
Indonesia (1993 and 2000)
Bolivia (2005)
Guatemala (2000 and 2006)
Albania (2002 and 2005)
Ecuador (1995 and 1998)
Bulgaria (1995 and 2001)
Panama (1997 and 2003)
Vietnam (1992, 1998 and 2002)
Page 5
We use the following
income categories
7 income categories:
•
1. Crop production
2. Livestock production
3. Agricultural wage
employment
4. Non-agricultural wage
employment
5. Non-agricultural selfemployment
6. Transfer
7. Other
•
•
•
•
Agricultural income
– crop + livestock + agricultural
wage
Non agricultural income
– non-agricultural wage + nonagricultural self + transfer + other
On farm
– crop + livestock
Non farm
– non-agricultural wage + nonagricultural self
Off farm
– agricultural wage + nonagricultural wage + nonagricultural self + transfers +
other
Page 6
Rural households in most countries
have an on farm activity
50
60
70
(%)
80
90
100
Participation in on farm activities
6
7
8
9
GDP (log)
Participation Africa
Overall Trend
Africa without NGA
Log of 2005 PC GDP
Participation Non-Africa
African trend
Page 7
And a large share a non farm activity
(non agricultural wage and self emp)
60
40
20
(%)
80
100
Participation in non farm activities
6
7
8
9
GDP (log)
Participation Africa
Overall Trend
Africa without NGA
Participation Non-Africa
African trend
Page 8
Increasing share of non agricultural income
with GDP: Is Africa different?
Page 9
Or just still at lower levels of GDP?
40
20
0
(%)
60
80
Share of non agricultural income
6
7
8
9
GDP (log)
Africa
Overall Trend
Africa without NGA
Non-Africa
African trend
Page 10
Similar story for non agricultural
wage income—increasing with GDP
20
10
0
(%)
30
40
Share of non agricultural wage income
6
7
8
9
GDP (log)
Africa
Overall Trend
Africa without NGA
Non-Africa
African trend
Page 11
Do rural households in African have a tendency
towards more on farm specialization?...
Household defined as specialized if receives more
than 75 percent of income from single source and
diversified if no single source is greater than 75
percent
Page 12
… Possibly! Decreasing, but still high,
specialization in on farm activities in Africa
40
20
0
(%)
60
80
Share specializing on farm
6
7
8
9
GDP (log)
Africa
Overall Trend
Africa without NGA
Non-Africa
African trend
Page 13
Increasing specialization in non
agricultural wage income with GDP
0
5
10
(%)
15
20
25
Share specializing non agricultural wage
6
7
8
9
GDP (log)
Africa
Overall Trend
Africa without NGA
Non-Africa
African trend
Page 14
Most African countries specialize in on farm
activities, while most non African are diversified
Page 15
Implications for welfare: Stochastic
dominance analysis
Tanzania
Malawi
Page 16
The role of geography: Theory and
literature
• Von Thünen (1842)
• New economic geography: Institutions vs
geography (Mostly macro, x-country)
• Agglomeration, dispersion forces
• Micro studies: Fafchamps et al. (Nepal); Foster
and Rosenzweig (India); Diechmann et al
(Bangladesh); Yamano and Kijima (Uganda)
• Interaction of location, ag potential, mediated
by infrastructure, tradability, wages, etc.
Page 17
Basic hypotheses on diversification
and location (theory and literature)
Specialization outside of farming
Distance to cities
Agricultural
potential
Low
High
Low
++
(?)
High
+(?)
-
Nonlinearities, interactions complicate the
picture
Page 18
The role of geography: Our estimates
•
•
•
•
•
Multinomial logit of specialization categories
On-farm specialization the base
Quadratic terms for distance, ag. potential
Interaction term b/wen distance and ag potential
Non-linearities not included unless jointly
significant
• Separately for different city sizes (20K to 1 mln.)
Page 19
The role of geography: Results
• “It depends…”: Non-linearities matter, the role
of distance changes with potential and city
size
• Role of distance more muted where ag
potential is high
• Smaller towns linked to diversification; larger
towns to non-ag
Page 20
Malawi: Non ag wage specialization, ag
potential, and distance from cities
0
Low Distance
-0.5
 High potential flatter: Ag
driving
High Distance
-1
Dependent variable
-1.5
 Low potential, large cities: Non-ag
higher declines with distance
-2
-2.5
Low Potential
High Potential
-3
-3.5
-4
0
Low Distance
-0.5
-1
Small town
 Low potential, small towns: Non-ag
higher with distance
Dependent variable
-4.5
-5
High Distance
-1.5
-2
Low Potential
-2.5
High Potential
-3
-3.5
-4
-4.5
Large city
Page 21
Tanzania: Non ag wage specialization,
ag potential, and distance from cities
0
Low Distance
 High potential flatter: Ag
driving (with nuances)
High Distance
-2
 Low potential, large cities: Non-ag
higher declines with distance
Low Potential
-3
High Potential
-4
0
Low Distance
-0.5
-5
High Distance
-1
-1.5
-6
Dependent variable
Dependent variable
-1
-2
Low Potential
-2.5
Mid-size town
High Potential
-3
-3.5
 Low potential, mid-size towns: Nonag lower with distance
-4
-4.5
-5
Large city
Page 22
Conclusions
• Diversification patterns in Africa do not seem to differ
markedly, for given GDP pc
• Non-farm and household welfare
• We know about barriers to entry: Role of education
• Need to consider spatially explicit policies:
– Ag potential
– City size
– Land abundance
• Data now allow for more in-depth, spatially explicit analyses
• Need (and opportunity) for revitalizing ‘rural development’
discourse in Africa?
Page 23
Income diversification patterns in
rural Sub-Saharan Africa:
Reassessing the evidence
Benjamin Davis, Alberto Zezza and Stefania di Giuseppe
Annual Bank Conference on Africa
Paris, June 23, 2014
AG RICULTURE
IN AFRICA
T E L L I N G FA C T S
FROM MYTHS