Land use and agricultural productivity in Ghana CSAE Conference Denis Cogneau, Lara Tobin & Liam Wren-Lewis 24 March 2015
Introduction “In developing countries, 6 million ha of additional land will be brought into production each year to 2030. Two-thirds of this expansion will be in Sub-Saharan Africa and Latin America, where potential farmland is most plentiful.” Deininger et al. (2011) “Africa has very little empty, good, cheaply cultivable cropland, despite some misinterpreted recent claims.” Lipton (2012) D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Uncertain prospects Sub-Saharan Africa is experiencing strong demographic growth → Will agricultural production keep up? Uncertainties regarding land suitable for agriculture ◮ Heady and Jayne (2013); Lipton (2012); Deininger et al. (2011). Limited increases in yields ◮ Hunt & Lipton (2011); Duflo et al. (2008); Jayne & Raschid (2013). Reactions of households/institutions to land scarcity ◮ Boserup (1965); Pande & Udry (2006); Demont et al. (2007) D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Aims of this project 1 Methodological: ◮ Can we use household surveys to measure agricultural land expansion? ◮ Can we use satellite data to measure land availability? 2 Empircial ◮ How have households responded to changes in land availability? ◮ When will Ghana run out of land? D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Answers so far.... 1 Methodological: ◮ Can we use household surveys to measure agricultural land expansion? Yes (probably) ◮ Can we use satellite data to measure land availability? Maybe (we’re working on it) 2 Empircial ◮ How have households responded to changes in land availability? A number of possibilities (incl. migration, input use, livestock) ◮ When will Ghana run out of land? Between 2020 and 2100 (i.e. we don’t know) D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Potential data sources on land use We compare several sources of data on land use in Ghana: The Food and Africulture Organisation (FAO) / Ministry of Agriculture ◮ Problems: Methological issues, political economy of data ◮ Young (1999); Carletto et al. (2013); Jerven (2012); Sandefur (2013); Devarajan (2013). The Ghana Living Standards Survey (GLSS) ◮ Problems: Conversion factors, self reported data ◮ Carletto et al. (2011); Capeau and Dercon (2011). Satellite data ◮ Problems: Mismeasurement, consistency of processing, clouds ◮ Codjoe (2007), Van der Geest et al. (2010); Yiran et al. (2012) D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Land under cultivation 80 Land under cultivation (1000 sq km) 70 60 50 40 1990 1995 2000 2005 2010 Years GLSS FAO US Geo Survey D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Land cultivated per capita .4 Land under cultivation (Ha/capita) .35 .3 .25 .2 1990 1995 2000 2005 2010 Years GLSS FAO D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Where has this cultivated land come from? Cultivated land increased by 24,000 km 2 between 1991 and 2005: Equal expansion in forest and savannah zones (GLSS) 2/3 from reduction in non-agricultural land (FAO) 1/3 from reduction of long fallows/pastures (FAO) ◮ Large reduction in HHs with livestock, mainly in forest zone (GLSS) ◮ Fallow data inconsistent over time (GLSS/satellite) D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Until when is this sustainable? The sustainability of this expansion unclear: If Ha/capita stays constant & all land is suitable: - Ghana will run out of land in 2100 If Ha/capita continues to expand & only agricultural land suitable: - Ghana will run out of land in 2020/25 Moreover, land productivity may fall if: ◮ New land is less suitable for agriculture ◮ Reducing fallows decreases soil fertility ◮ Deforestation has environmental consequences D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Do we see evidence of lower productivity? Productivity and yields: Both GLSS and FAO show (small) increase in yields across crops This is despite decrease in population working in agriculture (GLSS/Census) Large increases in use of inputs (GLSS) At aggregate level, we cannot distinguish between the following stories: 1 No negative productivity effect of land expansion 2 Lower land productivity ⇒ stimulating compensatory input use 3 Lower land productivity hidden by unconnected increasing input use D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Can satellite data tell us anything? In theory: Satellite data can give us information on: ◮ Land use ◮ Fallows ◮ Yields Data sufficiently fine to match with HH survey cluster D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Can satellite data tell us anything? In practice: We have three land use maps from satellite data: 1990: Produced by CERGIS, Ghana 1 2000: Produced by CERGIS, Ghana 2 2005: Produced by Globcover, ESA 3 Translating satellite images to land use requires many assumptions... ... and no consistent practice ⇒ Impossible to measure trends from existing datasets D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
CERGIS Land Use map, 1990 D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
CERGIS Land Use map, 2000 D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Use satellite data in cross-sections Future work: Working with a team of remote sensing specialists Will generate land use measures consistently over time Also attempt to measure fallows / yields For now: Is there evidence satellite data can measure anything relevant? We construct measures of land availability at the cluster level Correlate with economic variables from HH surveys D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Cross-sectional evidence on land availability Regressions at cluster level (GLSS): Share of Share of Average Share of Land farmed HHs in HHs using no. of HHs who per capita agriculture inputs livestock sold land Share of 10k radius 34.409*** 0.193 -25.131*** 2.245* 6.859 available for agriculture (7.632) (0.169) (6.839) (1.286) (12.069) Nb. Obs 580 580 462 579 461 R 2 0.53 0.29 0.21 0.45 0.24 Sources: Dependent variables from GLSS 5, independent variable from 2000 land use and 2005 land cover satellite images. Controls include: distance from road, ecological region dummies, elevation, suitability for agriculture index D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Cross-sectional evidence on fires Agricultural growth rates at district level (Census): 1 Agricultural population growth .5 0 −.5 −1 6 8 10 12 14 16 Log of number of fires D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Conclusions What we have found: FAO data appears consistent with Ghanain household surveys Cultivated land roughly doubled between 1991 and 2005 This was a result of: ◮ Conversion of non-agricultural land (e.g. forests) ◮ A reduction in pastures ◮ Possibly a reduction in fallows Existing satellite data is inconsistent over time D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Conclusions What we hope to find out soon: What has happened between 2005 and 2010? Are trends from satellite data consistent with surveys? Insights on how land availability influences household behavior A better guess as to when Ghana will run out of land D. Cogneau, L. Tobin & L. Wren-Lewis Land use and agricultural productivity in Ghana
Recommend
More recommend
