CAPITULO 4 CONSTRUCCIÓN DE LA SOLUCIÓN PROPUESTA
4.2 D ISEÑO DE LA BASE DE DATOS
Vietnam’s General Statistical Office (GSO) has undertaken five high quality household surveys in the transition period (GSO, 2008). Initially supported through the World Bank’s LSMS programme, the Vietnam Household Living Standards Survey (VHLSS) is now undertaken every two years. As the data in Table 5.2 and the chart in Figure 5.7 illustrate, the level of poverty reduction achieved has been exemplary, with the Headcount measure falling from some 58 per cent in 1993 to under 16 per cent of the population in 2006 (columns 2 and 6). This represents an overall reduction of 72 per cent reduction on the base. It is also noteworthy that the reductions in the secondary measures have been stronger, with the Depth measure, for example, falling by 83 per cent over the full 13 year period. It is also apparent that the reductions have been maintained throughout the period, though the level of change has declined. However, it is important to note that although the source is reliable and fully accessible, the Severity measure had to be estimated, as this is not published by the GSO (estimated results are shown in italic text in the table)107. In addition, for domestic policy
107 Estimated using POVCAL, based on distributional data given in GSO (2008)
purposes the national or a non-standard, more
Table 5.2: Poverty Aggregates from the VHLSS 1993 to 2006
1 imputed values. The data in the table is
109 The Vietnam Living Standards Survey (VLSS) supported by
national authorities concentrate on poverty aggregates based on standard, more a tightly drawn, definition of consumption
: Poverty Aggregates from the VHLSS 1993 to 2006109
2 3 4
Sources: GSO (1994 and 2008), World Bank (1999 and 2004)
: Poverty aggregates 1993 to 2006
Source: GSO (1994 and 2009) and World Bank (1999 and 2004
In line with the pattern of post-transition development elsewhere incidence of poverty are substantial. Table 5.3, which provides H
2004 and 2006 surveys, makes clear that the regions surrounding the major cen
oi (the Southeast and the Red River Delta) enjoy considerably lower
a measure of consumption – “Consumption required for living data in the table is compliant with the standard international he Vietnam Living Standards Survey (VLSS) supported by
the national Vietnam Household Living Standards Survey (VHLSS) in 2002.
VLSS93 VLSS98 VHLSS02 VHLSS04 VHLS06
elsewhere, regional variations in the Headcount data for the 2002, the regions surrounding the major centres of Ho Chi oi (the Southeast and the Red River Delta) enjoy considerably lower
Consumption required for living”, which excludes international definitions.
incidences of poverty: their respective 2006 headcount ratios are 5.8 per cent and 8.8 per cent (column 4); versus a national rate of 16.0 per cent. Moreover, urban areas, although suffering an increase in the latest survey (column 4) enjoyed dramatically lower levels of poverty - at 3.9 per cent versus 20.4 per cent for rural areas in 2006. The Headcount remains highest in the most geographically peripheral areas: the North Central Coastland region (29.1 per cent in 2006), the Central Highlands (28.6 per cent), and especially the Northwest (49.0 per cent).
These areas also have the highest proportions of Vietnam’s minority populations (excluding the comparatively wealthy Chinese minority). The disadvantaged groups include the Tay, Tai, Muong, Khmer, Nung and Hmong, and collectively account for around 12 per cent of the population (Kang, 2008). Several research studies on the spatial pattern of poverty in Vietnam confirm that remoteness and ethnicity are covariant drivers of poverty (Kang, 2008;
VASS, 2006).
Table 5.3: Poverty Incidence by Region 2002 to 2006
1 2 3 4
Category 2002 Headcount 2004 Headcount
(% change)
2006 Headcount (% change)
Urban 6.6 3.6 (-83.3) 3.9 (+7.7)
Rural 35.6 25.0 (-42.4) 20.4 (-22.6)
Regions:
Red River Delta 22.4 12.1 (-85.1) 8.8 (-37.5)
Northeast 38.4 29.4 (-30.6) 25.0 (-17.6)
Northwest 68.0 58.6 (-16.0) 49.0 (-19.6)
North Central Coast 43.9 31.9 (-37.6) 29.1 (-9.6)
South Central Coast 25.2 19.0 (-32.6) 12.6 (-50.8)
Central Highlands 51.8 33.1 (-58.6) 28.6 (-15.7)
Southeast 10.6 5.4 (-96.3) 5.8 (+6.9)
Mekong Delta 23.4 19.5 (-20.0) 10.3 (-89.3)
Source: GSO (2008)
However, as Figure 5.8, which disaggregates poverty by region, makes clear, incidence does not correlate with the overall numbers in poverty. Both the Central Highlands and the Northwest rank below the more populous Northeast and two River Deltas, in terms of their contribution to the total. Following the standard policy targeting regime followed within most
PRSs, which focuses on the Headcount, the anticipated policy response would be to concentrate on these core regions. Yet as noted in Section 5.3, this is precisely what the authorities have not done. Instead, anti-poverty interventions continue to target remote regions and ethnic minority populations through a variety of area-based programmes. This reflects a deeper commitment to equity and social cohesion. The data in Table 5.3, however, also shows that these efforts are meeting with mixed success.
Figure 5.8: Rank ordered regional contributions to overall poverty
Source: Author’s calculations based on GSO (2008)
Analysis was undertaken to test the sensitivity of the poverty level to changes in the threshold value. Using the poverty simulation program, POVCAL, the threshold value was flexed up and down by 5 and 10 per cent (column 2), and the poverty aggregates were then re-estimated (columns 3, 4 and 5)110. The results, provided in Table 5.4, show strong, but asymmetric, sensitivity, with proportionate increases in the poverty line resulting in disproportionate rises in the aggregates. For example, a reduction in the 2006 threshold value by 5.0 per cent is
110See Chen, Datt and Ravallion (1991) for a guide to POVCAL and the discussion in 4.4.2.
0%
Table 5.4 : Sensitivity of poverty aggregates to changes in the poverty line
Data disaggregated by demographic characteristics are hard to obtain from official sources.
However, analysis undertaken by the World Bank and the Vietnamese Academy of Social Sciences (VASS), suggest Vietnam’s poverty profile has fairly familiar characteristics to other LDCs. Poorer households are generally larger, have weaker educational backgrounds and are more likely to have no formal sector employment (World Bank, 2004; VASS, 2006). The gender dimensions are more complex, with the data showing a surprising degree of balance for households headed by men and women; however, participatory studies suggest there is substantial inequality in intra-household allocations (World Bank, 2004).
A more specific facet of poverty is the extent of deprivation within the ethnic minority populations. Moreover, as Figure 5.9, based on data within the VASS 2006 Poverty Assessment, shows, the variation in the Headcount and Depth measures between the Kinh majority (plus the Chinese minority) and the ethnic minorities appears to have widened over time. In the diagram, the trend lines for the majority (the solid line) and the disadvantaged minorities (the broken line) are cleanly divergent. Although, as noted, a considerable part of this variation is also driven by location, which co-varies with ethnic origin, marginal analysis provided by the World Bank finds that ethnic groups are on average 14 per cent poorer after allowing for other characteristics (World Bank, 2004). This further suggests that official policy responses, which seek to target these groups, are having limited success. However, Kang (2008) finds that the situation is more nuanced, and has changed markedly in the latest survey data. He notes that the dynamics of poverty reduction vary within groups, according to
their location and household characteristics, with some minorities closing the gap with the Kinh and Chinese groups in recent years.
Figure 5.9: Majority (solid line) versus minority (broken line) poverty rates
Note: the broken line indicates the minority population trends and the solid line the majority trends.
Source: VASS (2006) page 25
5.4.3 Inequality
As has been emphasized elsewhere in this thesis, inequality is not merely important in its own right, but is central to determining the level of poverty reduction in an economy, both directly, and by moderating the impact of economic growth at different points in the distribution. As the discussion in Section 2.4 makes clear, the policy consensus has coalesced around a position that places importance on both inequality and growth.
Successful management of inequality has been a defining feature of the Vietnamese transition.
As Table 5.5 illustrates, each of the aggregate measures records a limited deterioration in the 2002 and 2004 surveys (columns 4 and 5), and an improvement in 2006 (column 5). The Gini coefficient, for example, remains virtually static between 1993 and 1998, worsens by 5.7 per cent in 2002 and by a further 4.3 per cent in 2004, but improves by 3.6 per cent in 2006. For the full 13 year period, the Gini is only some 5.9 per cent higher. This pattern of change is remarkable given that theoretical and empirical accounts would have predicted the reverse trajectory, on the basis of the growth record and the dispersion of asset ownership (particularly land tenure) in Vietnam111. Indeed, the improvement in distribution seen in the
111 See the discussion in Milanovič (1998) 54
2006 survey somewhat confounds past donor concerns over the trajectory given by the 2002 and 2004 data112.
Table 5.5: Summary comparison of inequality measures 1993 to 2006
1 2 3 4 5 6 Source: Author’s calculations based on Povcalnet database (accessed on line)
A further important dimension, which is evident from the Lorenz curves for the 1993 and 2006 surveys shown in Figure 5.10, is that the deterioration has predominantly affected the upper part of the distribution. Indeed, the post-PRS (i.e. post-2002) surveys show a similar pattern, albeit with a worsening in 2004, offset by a compensating improvement in 2006.
Figure 5.10: Lorenz Curves for 1993 and 2006
Source: GSO (1994 and 2008) and author’s calculations