The Chinese household saving rate is high and has been rising sharply. Between 1983 and 2011, the average urban household saving rate rose by about 20 percentage points – from 10.4% to a staggeringly high level of 30.5%. This stands in sharp contrast with the low household savings rate in developed countries (about 5% in OECD economies). A fast-growing economy should in principle be borrowing against future income to bring forward consumption. Why the Chinese saved so much over a period of rapid income growth and at an increasing rate has led to something of a conundrum – a ‘Chinese saving puzzle’.
Economists have put forward a handful of plausible explanations. Blanchard and Giavazzi (2005) believe that Chinese households save for precautionary reasons. Wei and Zhang (2011) argue that saving is a response to the ferocious marriage-market competition induced by gender imbalance. Changes in income profiles (Song and Yang 2010) and fast growth in the face of tight household credit constraints (Ceourdacier et al. 2013) are also among the contenders.
The saving puzzle coincided with another unique phenomenon in China – the enforcement of the one-child policy in the early 1980s aimed at curbing China’s population spiral. The consequence of fertility control policies was a drastic reduction in the average urban fertility rate from a bit above 3 children per family in 1970 to about 1.2 by 1982.
Can a drastic reduction in fertility raise household saving rates?
We explore this question in our recent paper, “The One-Child Policy and Household Savings’’ (Choukhmane et al. 2013). Part of the answer can be disarmingly obvious: fewer children beget lower spending – hence a higher saving rate (the ‘expenditure effect’). Part of the answer relates to a cultural phenomenon: in a society in which the notion of intergenerational support is not only a norm but also stipulated by law, a reduction in the number of children supporting parents in old age lowers expected income (deriving from children) and raises the need for saving (the ‘transfer effect’). Indeed, as Figure 1 shows, intergenerational transfers (family support) are the main source of revenues for more than half of elderly people. Moreover, the total amount of transfers parents receive is increasing in the number of offspring they have. This includes not only financial transfers but also in-kind benefits, such as cohabitation – the likelihood of which increases significantly with the number of offspring one has.
Figure 1. Main source of livelihood (60+ year olds)
Source: Census 2005.
These ‘expenditure’ and ‘transfer’ effects, which induce households to save more when fertility drops, capture the behavioural changes at the household level. Of course, at the aggregate level, the shifting demographic compositions also exert an impact on saving. This is the channel linking demographics and saving that is conventionally emphasised. The reduction in the ratio of the young to working-age population increases the share of the main savers of the economy, and yet the share of the elderly dissavers relative to the working-age population does not increase until one generation after (when the only-child generation reaches prime working age). Thus, during this demographic transition period, the rise in the ratio of the middle-aged pushes up the saving rate. We estimate that the micro-level and macro-level channels can explain about 35–45% of the total 20 percentage-point rise in China’s household saving rate.
The birth of twins as an exogenous source of variation in fertility
To understand the relationship between fertility and saving, one cannot naively compare the saving rates of households with different numbers of children. Many factors that affect fertility, such as income and education, can also be systematically related to saving behaviour. One also cannot simply compare the household saving rate before and after the one-child policy, since many economic factors were changing during that period. A natural experiment we exploit is the birth of twins under the one-child policy – an arguably exogenous deviation in fertility.
Do households with twins save less than only-child households? And by how much? Even casual observation suggests that there are significant differences in their saving behaviour – a pattern that holds across all income groups (Table 1). More systematically, across all urban households over the period 1992–2009, regression analysis shows that an additional child (twin) reduces the saving rate by about 6-7 percentage points. The lower saving rate stems from higher expenditure – on food, and particularly on education. One additional child increases food expenditures (as a share of total household expenditure) by about 2.5 percentage points and increases education expenditures by about 7 percentage points. That is to say, an only-child household will on average spend 10.6% of their household income on education, whereas those with twins will need to spend an average of 17.3% of household income.
Table 1. Household saving rate: Difference in twin and only-child households by income group
Source: Urban Household Survey 2002-09.
The ‘transfer channel’ is harder to tease out given the limitations of the data, and can only be indirectly inferred. We find that parents with twins at a later stage of their life will tend to consume significantly more (eight percentage points more as a share of household income) on average than parents with an only child – on expenditures that are non-education/child related. Our explanation is that these parents consume more because they are expecting more transfers coming from their two offspring.
The savvy reader may wonder if in the face of a restriction in the quantity of children, the parents may substitute for ‘quality’ – an effort to maximise expected transfers through heavier investment in education. Data strongly supports this view. As Figure 2 shows, while there is little discernible difference between mandatory education expenditures incurred for a twin versus an only child before the age of 15, discretionary education expenditures for children of ages 15–21 exhibit vast differences. At age 20, the education investment per capita for twins is almost half of that of an only child. Discretionary education expenditures are also large – constituting 15–25% of total household expenditures.1 The consequence is a lower human capital attainment on average for a twin compared to an only child – twins are 40% less likely to pursue higher education than their only-child peers, and for secondary education, 40% more likely to attend a technical high school (instead of an academic high school) compared to an only child.
Figure 2. Education expenditures (twins vs. only children)
Implications for a relaxation of fertility restrictions
The third plenum of the Chinese Communist party just announced a loosening of the one-child policy to a two-child policy, so long as either of the parents is an only child. This would apply to most couples of child-bearing age. Fertility is bound to rise, and a reversal of the saving trend may be expected. Our estimates suggest that as much as an 8% decline in the household saving rate over the next few decades wouldn’t come as a surprise.
Blanchard, Olivier J and Francesco Giavazzi (2005), “Rebalancing Growth in China: a Three-Handed Approach”, mimeo.
Choukhmane, Taha, Coeurdacier, Nicolas, and Keyu Jin (2013), “The One-Child Policy and Household Savings”, CEPR Discussion Paper 9688.
Coeurdacier, Nicolas, Stephane Guibaud, and Keyu Jin (2013), “Credit Constraints and Growth in a Global Economy”, CEPR Discussion Paper 9109.
Song, Zheng and Dennis T Yang (2010), “Life cycle earnings and the household saving puzzle in a fast growing economy”, mimeo.
Wei, Shang-Jin and Xiaobo Zhang (2011), “The Competitive Saving Motive: Evidence from Rising Sex Ratios and Savings Rates in China”, Journal of Political Economy, 119(3): 511–564.
1 These may take the form of extracurricular activities, tutorials, academic versus vocational high schools, university education etc.