Accounting for the Diversity of Rural Income Sources in Developing Countries: The Experience of the RIGA Project Katia Covarrubias, Ana Paula de la O & Alberto Zezza ESA Wye City Group Meeting on Statistics on Rural Development and Agriculture Household Income Rome, June 11-12, 2009 The Rural Income Generating Activities Project Database of 34 living standards surveys Outputs: Income Aggregates Household Level Indicators Access to capital Demographic indicators Additional analysis-specific indicators Methodological Goal: Consistency and Comparability RIGA Data: 34 Survey Countries Africa Ghana GLSS (1992, 1998*) Kenya KIHBS (2005) Madagascar EPM (1993, 2001) Malawi IHS (2004*) Nigeria (2004*) Asia Bangladesh IHS (2000*, 2005) Cambodia SES (2004) Indonesia FLS (1992, 2000*) Nepal LSS (1996, 2003*) Pakistan HIES (1991, 2001) Vietnam LSS (1992, 1998*, 2002*) Eastern Europe/Central Asia Albania LSMS (2002, 2005*) Bulgaria IHS (1995, 2001*) Tajikistan LSMS (2003*, 2007) Latin America Bolivia EH (2005) Ecuador ECV (1995*, 1998) Guatemala ENCOVI (2000*, 2006) Nicaragua EMNV (1998*,2001*) Panama ENV (1997, 2003*) * Labor Data also Available at the Individual and Job Levels Income Aggregates: Defining Income Income must: Occur regularly Contribute to current economic well-being (available for current consumption) Income must not: Arise from a reduction in current net-worth Arise from an increase in household liabilities Source: ILO, Resolution I “Resolution concerning household income and expenditure statistics” Available from: http://www.ilo.org/public/english/bureau/stat/download/res/hiestat.pdf Income Aggregates: Basic Characteristics Household-level Annual Labor data also available at the Job and Individual levels Wage income data: also for daily and monthly time frames Net of costs Purchases and sales of durables, investments and windfall gains excluded Local currency units Rural (and urban) Outlier checks Issues and Lessons Learned Income Estimation RIGA Components of Total Household Income Dependent Wage Income agricultural non-agricultural Independent Crop Livestock Self Employment Transfers public private Other Sources Total Household Income Classifications Total Income: Yi Agwgei Nonagwgei Cropi Livestocki Selfempi Transferi Otheri Yi Agriculturali Nonagriculturali Agricultural: Non-agricultural: Agwge + Crop + Livestock Nonagwge + Selfemp + Transfers + Other Yi Onfarmi Offarmi On-farm: Crop + Livestock Off-farm: Agwage + Nonagwge + Selfemp + Transfers + Other Non-farm: Nonagwge + Selfemp Total Household Income Agricultural On-farm Livestock Crop Agwage Off-farm Non-Agricultural Transfer Non-farm Nonagwage Selfemp Other Dealing with Costs Issue: Dealing with investment/durables expenditures Misclassification: bias total income Example: raw materials purchases (Albania; Vietnam) Recommendations: Clear classification of costs in survey instrument Appropriate choice of reference periods and frequencies Gross versus Net Issue: Inconsistent reporting & estimation of gross/net income Recommendations: In Qx: deductions and taxes should be asked about and reported In income estimation: Net: agricultural, self-employment and wage income Gross: rental income and transfer income Issues and Lessons Learned Questionnaire Design RIGA Reference Periods Issue: Defining appropriate reference periods Choice of Short v. Long seasonal fluctuations relevance to recall error link to survey timing phrasing of questions Recommendations: Reference periods should reflect frequency of Inc/Exp Short: Regular or frequent sources (food exp, wages, etc.) Long: Infrequent sources (business costs; ag inputs, etc.) Units & Coding Issue: Comparability and Standardization of Units and Coding Variability of unit reporting Lack of equivalence scales in data and documentation Inconsistency in units and codification of items across survey modules Agricultural Production and Food Expenditure modules Recommendations: YES to local unit reporting but: Inclusion of equivalence scales Consistency in codification within/across survey modules Lessons Learned From Key RIGA Results RIGA RIGA Results: Main Components of Rural Household Income On farm Activities Agricultural Wages Transfers and Other Non-Labour Sources Non-farm Activities Note: 1. Surveys sorted by increasing per capita GDP 2. Expenditure quintiles move from poorer to richer 03 PA N 01 BU L 95 U 05 EC B AL G U A 00 00 IN D 01 N IC 01 PA K 04 IG N VN M 98 03 TA J A 98 03 G H EP N G 00 93 BN AD M M AL 04 0 20 40 60 80 100 Share of total income from main income generating activities by expenditure quintiles On-farm income falls and Non-farm rises... Share of On-farm Income by Per Capita GDP Share of Non-farm Income by Per Capita GDP 70 60 MAD 93 MAL 04 NIG 04 ECU 95 40 TAJ 03 30 NEP 03 20 PAK 01 GUA 00 NIC 01 50 VNM 98 IND 00 GUA 00 BNG 00 GHA 98 ALB 05 ALB 05 VNM 98 NEP 03 NIG 04 BUL 01 30 NIC 01 IND 00 BNG 00 PAK 01 40 GHA 98 Share of Income from Non-farm Sources 60 50 PAN 03 10 ECU 95 BUL 01 PAN 03 20 6.5 7 7.5 8 Log Per Capita GDP (PPP, Constant 2005 $) Note: On-farm income is comprised of income earned from crop and livestock activities. 2. Fitted curve fits the quadratic prediction of the income shares on per capita GDP. 8.5 TAJ 03 MAD 93 MAL 04 9 6.5 7 7.5 8 Log Per Capita GDP (PPP, Constant 2005 $) 8.5 Note: Non-farm income is comprised of income earned from non-agricultural wages and self employment. ...with increasing per capita GDP levels. 9 RIGA Results: Diversification of Rural Household Income Defining Specialization and Diversification: Specialization >= 75% Diversification <75% Influenced by survey timing and reference period: seasonal diversification individuals member diversification Rural income diversification is the trend On-farm specialization falls with PCGDP Specialization in Non-Agricultural Wage Labour by Per Capita GDP 40 50 On-Farm Specialization by Per Capita GDP MAD 93 PAN 03 NIG 04 20 TAJ 03 10 NEP 03 ECU 95 30 NIC 01 PAK 01 GUA 00 BUL 01 IND 00 ALB 05 20 VNM 98 BNG 00 NEP 03 TAJ 03 GHA 98 PAK 01 10 30 GHA 98 Share of Households Specialized in Non-Agricultural Wages (%) 40 MAL 04 NIC 01 MAL 04 ALB 05 IND 00 NIG 04 MAD 93 VNM 98 GUA 00 ECU 95 0 PAN 03 BUL 01 6.5 7 7.5 8 Log Per Capita GDP (PPP, Constant 2005 $) 8.5 9 0 BNG 00 6.5 7 7.5 8 Log Per Capita GDP (PPP, Constant 2005 $) 8.5 9 ...but Non-agricultural wage specialization rises. RIGA Results: Defining the Agricultural Household “Rural” as “Agricultural” Thresholds of income lack of data to create comparable rural definition urban agriculture dwelling versus job location diversity of rural economy Non-zero (basic participation) Higher cut-offs Occupation of the household head RIGA Results: Sensitivity and Criteria in Agricultural Households Definition Survey Percent Rural Shares of Households with Positive Agricultural Income Shares of Households with Agricultural Income > 30% Share of Agricultural Households by Occupation Total 20.3 19.9 26.7 28.7 52.2 58.4 31.4 71.7 52.3 37.3 62.3 63.5 26.7 59.8 77.5 Total 36.4 26.4 33.0 29.8 48.4 68.0 38.7 n.a. 49.7 31.5 60.8 62.3 48.2 49.5 66.3 45.92 46.4 Peru (2003) Ecuador (1995) Bolivia (2002) Nicaragua (2001) Zambia (1998) Ethiopia (2000) Guatemala (2000) Cambodia (1999) Ghana (1998) Pakistan (2001) Vietnam (1998) Madagascar (2001) Bangladesh (2000) Nepal (2003) Malawi (2004) 35.9 37.4 42 43.9 47.8 50.7 56.7 60 63.3 71 71.2 75.8 79.7 87.4 88.1 Total 37.5 43.9 96 80.5 70.1 71.3 70.2 86.1 65.5 55.9 92.7 71.2 61.3 90.6 89 Unweighted average 60.72 72.12 Source: Aksoy, et al. (2009) Summary and Conclusions Estimation of Income Questionnaire Design: Various approaches for characterizing household income Costs classification Reporting of deductions/taxes relevant Reference periods should reflect frequency of income and expenditures Need for equivalence scales/conversion factors Unit and coding consistency within surveys. Analysis: Different definitions of agricultural household exist; generate differing characterization of results Thank You! Questions?