Docstoc

Household Responses to Pensioner out-migration and mortality in

Document Sample
Household Responses to Pensioner out-migration and mortality in Powered By Docstoc
					Household Responses to Pensioner out-migration and mortality in South Africa

V. Ranchhod ERSA Conference: Sept. 2006 email: vranchho@umich.edu

Question + Motivation
• How do households cope with the departure of pensioners? • Background:
– Non-contributory, means tested and age eligibility (≥ 60 for women, ≥ 65 for men.) – Most African elderly people receive it (77%)

• Broad coverage, High Value (relatively)
– 780 R/month – 19.3% of all HHs report “pensions and grants” as ‘main income’ in Sept. 2002 LFS – In African headed HHs with an elderly member, the corresponding percent is >70%.

Related Literature
• Pension improves health outcomes (Duflo 2000 & 2003), Case(2001) • ↓ Labor supply of prime aged adults in HH (Bertrand et al, 2003) • ↓ remittances sent to pensioner households (Jensen 2004) • Δ in HH composition : ↑ # of kids & young women, ↓ # of prime working aged women (Edmonds et al, 2005) • ↑ in child schooling enrollment & ↓ in child labor in rural areas (Edmonds 2006)

Theory (Informal)
If a pensioner leaves the household: • Expect the loss of income to ↑ Labor supply ( LF Activity) • Affect household composition, especially in cases where pensioners time was spent in childcare
– Change in composition (possible):
• Increase in number of adults (producers) • Decrease in number of kids (dependents)

• Increase in remittances

Data + Sample Selection
• From Sept 2001 – Mar 2004, 80% of HHs in SA LFS were rotated out, and a new 20% introduced (Waves 4 – 9). • ‘Pension’ only asked for those not currently working, so I use age-eligibility as a proxy • Sample selection: – HHs w/ a common hhid in subsequent waves, – w/ no members’ age missing, – w/ the eldest member an African, – w/ only 1 dwelling on that property, – w/ at least 1 ‘pensioner’ in the `first’ wave (wavet ) – and at least 1 ‘good’ person level match using the StatsSA matched dataset.

Method
• Final Sample: 12,374 HHs, separate into: – `Losers’ - 1,653 HHs – `Keepers’ - 10,275 HHs – `Gainers’ – 446 HHs (these are omitted from analysis) • Analysis: – Calculate mean change in `outcome’ for Losers between wavet and wavet+1 – “Difference” – Compare to similar change for Keepers – “Difference in Difference” – Estimate regression coefficient of ‘difference-in-difference’
• Include # pensioners before; wave, province and urban identifiers as additional covariates

• Selection Bias: – Reweight by inverse of the predicted Pr(Inclusion in Panel) to get the panel to look more like the cross-section (on observables)

Difference in Means: Main Income Source
Table 6: Summary statistics: Main Income Source Distribution of Main Income Source in Household (%) T0 T1 Keeper Loser Keeper Loser Salaries and/or wages 15.6 27.5 15.4 31.9 Remittances 4.5 8.5 4.2 23.2 Pensions and grants 77.8 58.6 77.9 36.7 Sales of farm product 0.4 0.4 0.4 1.2 Other non-farm income 1.5 3.1 1.8 5.3 no income 0.2 1.7 0.2 1.6 Unspecified 0.0 0.2 0.1 0.0 N 6,060 965 4215 688 note: 1. The T0 data relates to observations in Panels 4-5, 6-7 and 8-9 the T1 data relates to observations in Panels 5-6 & 7-8

Difference in means: Composition
HH size # kids young # kids school # youth # young adults # young adult - M # young adult - F # mid-aged adults # mid-aged adult - M # mid-aged adult - F T0 T1 Keeper Loser Keeper Loser Delta_L DD Reg. 5.358 6.294 5.316 5.515 -0.779 -0.736 *** 0.862 0.952 0.839 0.932 -0.020 0.004 1.107 1.285 1.113 1.207 -0.078 -0.084 *** 0.593 0.754 0.582 0.715 -0.039 -0.027 1.017 1.209 0.467 0.578 0.550 0.631 0.464 0.519 0.212 0.218 0.252 0.301 0.992 1.192 0.456 0.566 0.537 0.626 0.467 0.608 0.210 0.227 0.258 0.381 -0.017 -0.012 -0.006 0.088 0.009 0.080 0.007 -0.001 0.008 0.085 *** 0.011 0.074 ***

# older adults 0.169 0.254 0.177 0.613 0.358 0.350 *** # older adult - M 0.079 0.129 0.080 0.290 0.160 0.159 *** # older adult - F 0.090 0.125 0.097 0.323 0.198 0.191 *** # pens age 1.166 1.349 1.166 0.289 -1.060 -1.060 *** N 10275 1653 * denotes statistical significance at the 10% level, ** at the 5% level, and *** at the 1% level

Difference in Means: Activities (a)
T0 T1 Keeper Loser Keeper Loser Delta_L DD 1.071 1.235 1.080 1.174 -0.061 -0.070 ** 0.449 0.551 0.439 0.532 -0.019 -0.009 0.021 0.007 0.014 0.196 0.094 0.102 0.145 0.082 0.063 0.046 0.024 0.023 0.019 0.007 0.012 0.241 0.108 0.133 0.176 0.091 0.085 0.085 0.038 0.047 0.019 0.006 0.013 0.184 0.087 0.097 0.147 0.084 0.063 0.046 0.024 0.022 0.025 0.011 0.014 0.269 0.120 0.150 0.232 0.137 0.095 0.201 0.104 0.097 0.006 0.004 0.003 0.028 0.011 0.017 0.057 0.046 0.011 0.116 0.066 0.050 0.008 0.005 0.003 0.041 0.019 0.022 0.054 0.044 0.011 0.117 0.066 0.051

# of kids school # of youth school # of youth Work # of youth Work - F # of youth Work - M # of young adult Work # of young adult Work - F # of young adult Work - M # of mid-adult Work # of mid-adult Work - F # of mid-adult Work - M # of older adult Work # of older adult Work - F # of older adult Work - M

** * *** *** *** *** ***

Difference in Means: Activities (b)

T0 T1 Keeper Loser Keeper Loser Delta_L DD # of youth in LF (broad) 0.133 0.192 0.133 0.161 -0.031 -0.031 * # of youth in LF (broad) - F 0.069 0.088 0.068 0.077 -0.010 -0.009 # of youth in LF (broad) - M 0.063 0.105 0.065 0.084 -0.021 -0.023 * # of young adult in LF (broad) 0.836 0.989 0.816 0.986 -0.003 0.017 # of young adult in LF (broad) - F 0.447 0.505 0.437 0.506 0.001 0.011 # of young adult in LF (broad) - M 0.389 0.484 0.378 0.479 -0.005 0.006 # of mid-adult in LF (broad) 0.354 0.401 0.364 0.443 0.042 0.032 # of mid-adult in LF (broad) - F 0.193 0.215 0.202 0.262 0.047 0.038 ** # of mid-adult in LF (broad) - M 0.161 0.186 0.163 0.181 -0.005 -0.006 # of older adult in LF (broad) 0.082 0.131 0.084 0.280 0.149 0.146 *** # of older adult in LF (broad) - F 0.040 0.056 0.044 0.139 0.083 0.079 *** # of older adult in LF (broad) - M 0.042 0.075 0.041 0.141 0.066 0.067 ***

Difference in Means: Activities (c)

T0 T1 Keeper Loser Keeper Loser Delta_L DD # of youth in LF (narrow) 0.066 0.083 0.066 0.083 0.001 0.000 # of youth in LF (narrow) - F 0.031 0.035 0.031 0.036 0.001 0.000 # of youth in LF (narrow) - M 0.036 0.048 0.035 0.048 0.000 0.001 # of young adult in LF (narrow) 0.560 0.675 0.539 0.678 0.003 0.025 # of young adult in LF (narrow) - F 0.279 0.321 0.269 0.326 0.004 0.015 # of young adult in LF (narrow) - M 0.281 0.354 0.270 0.352 -0.001 0.010 # of mid-adult in LF (narrow) 0.268 0.303 0.275 0.355 0.052 0.045 # of mid-adult in LF (narrow) - F 0.144 0.156 0.149 0.202 0.045 0.039 # of mid-adult in LF (narrow) - M 0.124 0.147 0.126 0.153 0.007 0.005 # of older adult in LF (narrow) 0.063 0.104 0.062 0.241 0.136 0.137 # of older adult in LF (narrow) - F 0.030 0.043 0.031 0.116 0.073 0.073 # of older adult in LF (narrow) - M 0.032 0.061 0.031 0.124 0.063 0.065

** *** *** *** ***

Caveats
• Not necessarily a causal inference, since no exogenous variation • Simply a reduced form partial correlation coefficient • Use `Death’ for exogenous variation, but sample sizes are very small noisy estimates • Need to qualify results for sample selection and attrition, and acknowledge the potential impact of measurement error, especially attenuation bias. • An offsetting increase in remittances increases the significance of the results.

Conclusion
• Both household composition and the labor supply of household members change when we observe that a pensioner has left the household. • Labor supply of youth seems generally unaffected, as does their schooling enrollment rates. • It seems that females’ residency and labor force activity is more sensitive to the departure of a pensioner. • Strongest effects seem to manifest amongst middle aged women and older adults. • The importance of remittances increased drastically in Loser households. • The estimates are consistent with the ‘loose’ theory presented.


				
DOCUMENT INFO
Shared By:
Stats:
views:39
posted:12/19/2009
language:English
pages:13
Description: Household Responses to Pensioner out-migration and mortality in