As with the case of urban and rural households, the comparison of household saving
between the north and the south of the country' is undertaken using both the per capita
saving and the analysis of saving determinants in the saving level model. In
correspondence with the higher per capita income level of households in the south
compared to those in the north (mean values of per capita income of households in the
south are 1975 thousand dong compared to 1297 thousand dong in the north, or a factor of
around 1.5 times), per capita saving of southern households is also higher than that of
northern households: 496 and 344 thousand dong, respectively (Table 6.5). However, the
gap of per capita saving of the two regions is not as high as in the urban/rural case, but
about the same as the gap of per capita income o f south/north: 1.4 times.
The econometric results from the 2SLS regressions for the two geographic regions are
reported in Table 6.6. The results show that for the group o f households in the north the
variables of income, education, number of children, marital status and gender of household
head are found to be important. For the southern households, only income, assets and
than that of those in the south, and the difference is considerable: 0.52 compared to 0.62.
Taking the fact that per capita income of the households in the south is more than that in
the north in Table 6.5, this result shows that the regional group with a higher income level
has higher mps.
Table 6.5: Per capita income and saving of households in the north and in the south (thousand dong)
Northern households Southern households Total sample
Per capita income 1297 1975 1613
Per capita saving 344 496 415
The effect o f the number of children on household saving is found to be almost the same
for households in the two regions. The impact is negative and, as in the case of the whole
sample, the marginal effect of an extra child is a fall in saving o f approximately 0.2 million
dong. The number of old people in the households seems to have no effect on saving in
either region. The signs and the magnitudes o f the coefficients as well as the significance
levels and the signs of the intercept terms were consistent with those for the whole sample.
In particular, the coefficient of assets (W) for the south is -0.12, close to the corresponding
Similarly, while the marginal effects of education, marital status and gender of household
heads differ across the north and south, they are close to those found for the whole sample.
Table 6.6: Saving of households by geographic regions
Variables Northern Southern
Coefficients t-ratios Coefficients t-ratios
Y 0.52** 4.34 0.62** 7.12 W -0.14 -1.39 -0.12** -2.47 AGE 0.004 0.27 0.0001 0.003 EDUY 0.14** 2.30 0.09 1.15 NCHILD -0.21** -2.63 -0.22* -1.90 NOLD -0.30 -1.00 0.08 0.19 MAR 0.95** 2.20 0.56 0.91 GEN -0.85** -2.12 -0.41 -0.79 Constant -1475.67** -2.05 -2160.16** -2.00 R-squared 0.73 0.72 Observations 2446 2142
Note: The dependent variable is the saving level, in million dong. The asterisks * and ** denote the significance levels at the 10 per cent and 5 per cent, respectively. Income (Y) and assets (W) are in million dong.
Except for the mps and the marginal effect of the number of children, it is hard to compare
the marginal effects of other factors on household saving across these two regions because
the coefficients of the other variables were significant in one regression but not in the
other. However, it is possible to draw a conclusion about the impact of the saving
determinants in a particular group as well as for the whole sample. For example, with the
whole sample, education and marriage were found to be important in inducing households
to save more, while the coefficient for gender implies that households with female heads
save more. The results also suggest that the effect o f these factors found in the case of the
whole sample may be mainly accounted for by households in the northern region.
Conversely, assets were found to have a negative effect on household saving in the south,
and to have a more dominant role than those of northern households in contributing to the
effect of this factor for the whole sample. This result and that of the previous sub-section
suggest that the assets factor has a more pronounced effect on saving in households that are
relatively well off. In particular, assets were found to have an impact on saving in urban
areas and in the south which has higher living standards than rural households and
households in the north.
Overall, this sub-section shows that if the sample o f households is divided by geographical
regions, the southern region households, that have higher mean per capita income than the
northern households, also have a higher mps. While the effect o f assets, education, marital
the same across regions and is likely to induce households to dissave, as was found for the
whole sample.