The Incidence of Social Transfers in Vietnam Nordic Conference in - PowerPoint PPT Presentation

The Incidence of Social Transfers in Vietnam Nordic Conference in Development Economics 2009 Henrik Hansen and Le Dang Trung Institute of Food and Resource Economics University of Copenhagen

Background Transfers between the central government and the provinces are pro-poor. But rumor has it that direct government transfers are regressive: We take a second look

Outline 1. A brief description of the transfer system 2. Estimation of the behavioral responses to transfers 3. The incidence of the social transfers in 2004 4. The incidence of war related transfers in 2006 5. Policy recommendations

Direct Household Transfers: Social Insurance and Social Protection • It is natural to split the direct social transfers in Vietnam into Social insurance and Social protection In 2004: • The social insurance system provides labor market related VND 15,139 cash transfers, say, in the form of pension and survivorship billion • The social insurance system started in 1947 • In 1995 it was reformed by the establishment of the Vietnam Social Insurance (VSI) program VND 2,141 • S ocial protection is organized in social guarantee funds billion • The Social Guarantee Fund for War Affected Groups (aka. The Social Guarantee Fund for Veterans, War Invalids and Others who have Contributed to National Liberation Struggle) • The Social Guarantee Fund for Regular Relief • The Contingency Fund for Pre-Harvest Starvation and Disaster Relief

Estimation of the incidence of the transfers • In order to estimate the incidence of the social transfers we need to define a welfare measure. • We follow standard WB practice and use a consumption based welfare measure • Most benefit incidence analyses use a very simple measure of the counterfactual expenditure: the observed expenditure less the transfer: E  E  T i i i • This measure implicitly assumes there is no behavioral response to transfers • If there is a response we should not subtract the full transfer, but only the fraction consumed E  E   T i i 1 i

An econometric model for the marginal propensity to consume out of transfers • We assume the marginal propensity to consume out of transfers can be estimated from a reduced form model for household expenditure E     T  X        it 0 1 it it i t it • Some of the social transfers are clearly endogenous being correlated with consumption fixed effects and the shocks • We address this problem by using the changes from 2002 to 2004,  E    T   X   c    it 1 it it it • and by using instruments for the transfers • We use the number of pension age women (above 55) and pension age men (above 60) in the households in 2002 and 2004

Regression results for Social Transfers Dependent variable: Real household expenditure, 1.000 VND January 2004 prices. Transfer: Social insurance and Social protection Estimator Simple w. controls IV Constant 3,231 (15.67) 3,033 (15.48) 2,920 (13.87) Transfer 0.328 (3.29) 0.197 (2.49) 0.988 (2.14) Size of household 3,388 (6.33) 3,580 (6.72) Size of household squared -78 (1.75) -81 (1.83) Gender of household head 1,080 (1.36) 1,017 (1.29) Household members who are Children 0-5 -1,059 (2.20) -1,229 (2.62) Children 6-10 -782 (2.28) -818 (2.54) Primary-educated -144 (0.54) -208 (0.82) Secondary-educated 389 (1.45) 336 (1.27) High-school-educated 1,453 (3.41) 1,058 (2.43) Vocationally educated 2,825 (4.85) 2,334 (4.09) Professionally educated 3,455 (5.25) 3,217 (4.79) University-educated 3,744 (4.09) 3,270 (3.33) Observations 3935 3935 3935 RESET test (p-value) 0.72 0.43 0.71 First stage F-statistic 50.28 Hansen J-test (p-value) 0.43

The estimated counterfactual consumption measure • Based on the modeling considerations and the regression results we conclude that the marginal propensity to consume out of transfers is (close to) 1 • This means that the counterfactual expenditure (the theoretical expenditure without transfers) is computed as E  E  T i i i • This is in stark contrast to previous studies that used E  E  T 1 i i 2 i • We now look at average transfers by quintile and use a decomposition of the averages to understand the distributional shapes       Nq R B 1 N 1 q  q q  T T  T  , q  1,2, ,5.       q qi N qi       N R B i  1 q i  1 q q q Taget Coverage Average transfer population rate to the recipients

The incidence of Social Insurance in 2004 Quintiles (Poorest) (Richest) Q1 Q2 Q3 Q4 Q5 All Average transfer 349 117 136 128 228 191 Reference population (%) 41.2 32.5 29.8 32.5 31.4 33.5 Coverage rate (%) 39.4 21.5 28.1 24.5 37.8 30.7 Average transfer to recipients 2150 1676 1617 1609 1924 1863 Average share of expenditure (%) 61.9 35.1 29.5 23.1 15. 9 36.4 Mpct = 0.5 Average transfer 153 135 136 205 326 191 Reference population (%) 38.3 32.6 30.1 34.0 32.2 33.5 Coverage rate (%) 23.8 24.0 27.7 32.6 46.4 30.7 Average transfer to recipients 1679 1734 1627 1856 2178 1862 Average share of expenditure 63.0 45.4 35.4 30.9 20.0 36.4 All monetary values are VND 1,000 January 2004 prices

The incidence of Social Protection in 2004 Quintiles (Poorest) (Richest) Q1 Q2 Q3 Q4 Q5 All Average transfer 49 31 20 24 11 27 Coverage rate (%) 16.8 8.6 5.1 5.8 2.9 7.9 Average transfer to recipients 291 363 390 420 369 344 Average share of expenditure (%) 13.9 11.6 9.8 8.0 4.0 11.3 Mpct = 0.5 Average transfer 36 30 26 33 12 27 Coverage rate (%) 15.4 8.2 5.9 6.8 3.1 7.9 Average transfer to recipients 215 292 458 468 484 344 Average share of expenditure (%) 12.7 12.8 11.6 9.0 4.3 11.3 All monetary values are VND 1,000 January 2004 prices

The main part of social protection is War Compensation Disaster relief Poorest Richest All Q1 Q2 Q3 Q4 Q5 Average transfer 3,39 1,00 0,5 0,35 0,03 1,13 Incidence 2,35 1,65 0,97 0,81 0,20 1,24 Average transfer to recipients 144,53 60,63 51,85 43,57 13,74 91,11 Average share of expenditure 6,44 1,53 1,13 0,7 0,15 3,39 War related transfers Average transfer 134,65 59,52 49,79 47,08 39,83 68,39 Incidence 18,93 8,12 6,94 6,57 4,82 9,4 Average transfer to recipients 711,17 732,88 717,29 716,9 826,61 727,38 Average share of expenditure 24,44 17,18 12,84 9,55 6,62 17,8 All monetary values are VND 1,000 January 2006 prices

Policy Recommendations • Social insurance and protection in Vietnam consists of 4 kinds of transfers 1. Labor market pensions: Not poverty oriented — by construction • The government should consider its involvement and the generosity of the system could be reconsidered 2. War related compensations: Not poverty oriented — by construction • The government should make the transfers transparent and not list the transfers as social protection 3. Disaster relief: Poverty oriented • The government should make the transfers transparent and probably increase amount spent 4. Regular relief: Appears to be a state secret • The government should, at least, make the transfers transparent

Thank you