Microsoft Word EcRefChina Oct06. doc

average 
monthly 
wage for employees in the individual’s geographic area (provided a minimum of 15 years of 
contributions have been paid), the pension level in the second layer can only be guessed at (Ministry of 
Labor and Social Security, 1998, p. 51). For a general discussion of China’s emerging pension system in 
urban areas, see, for instance, Song (2002); and Zhao and Xu (2002). 
124
Broadly speaking, saving by households and enterprises in recent years each accounted for 2/5 of 
aggregate saving (amounting to close to 40-45 percent of GDP), and the government for the remaining 
1/5 (Kuijs, 2005).

---

61
in the future, this argument is not watertight.
125
After all, the pension system chosen 
now should be designed for individuals who will be retirees many decades later.
Moreover, the long-term financial viability of the new pension system may be questioned. 
One reason is the combination of expected future changes in demography (the “graying” 
of the population) and the low pension age (formally 60 years for males and 55 for 
females, with an effective retirement age of only 55 for the former). This problem could, 
in principle, be solved through a higher effective retirement age, provided there is also a 
reasonably well functioning labor market for the elderly (which requires flexible wages 
and working hours).
A more specific Chinese problem is that contributions originally paid into the funded part 
of the system have, in fact, been used to finance deficits in the paygo part of the system – 
resulting in the “problem of empty individual accounts”. As a consequence, the funded 
part of the mandatory pension system so far looks more like a defined-contribution paygo 
system with “notional” rather than real accounts.
126
The Chinese authorities would have 
basically two alternative strategies to deal with the issue of “the empty accounts”. One 
would be to abandon the idea of funded individual accounts altogether and be satisfied 
with a paygo system (possibly with notional individual accounts based on previously paid 
contributions). The other alternative would be to “recapitalize” the funded accounts. This 
could, be done in different ways. One possibility would be tax-financed capital injections 
to the individual accounts. Another possibility, suggested by Pieter Bottelier (2002), could 
be to let the National Council of Social Security Fund take over the shares in a number of 
state firms, and instruct the Fund to sell the shares – at appropriate intervals to avoid 
strong negative effects on share prices. The sales revenues would then be deposited on the 
individual accounts in proportion to what each individual has paid (directly or indirectly 
through his/her employer). Two birds would then be killed in one shot: the funded system 
would be saved and the privatization of state-owned firms and assets would be speeded 
125
Some of the reasons to expect that the household saving rate will fall in the future are changes in 
demography (fewer individuals in working age), slower per capita income growth and the build-up of 
mandatory social insurance – as predicted by the life-cycle saving model; see, for instance, Modigliani 
and Cao (2004). Another reason is that the liquidity constraints on individual households are likely to 
become weaker as domestic credit markets develop. 
126
According to OECD (2005a, p.188), the return to individuals on these accounts has been no more 
than 2.4 percent per year. Some observers have hypothesized that the meager return in this system is an 
important explanation for the low participation as well as the evasion of paying contributions.

---

62
up. A much more modest version of this idea has, in fact, already been implemented: The 
National Social Security Fund is entitled to receive 10 percent of the proceeds from the 
sales of shares in state-owned companies in connection with initial public offerings 
(IPOs) and new share issues. 
A more general issue concerning funded government-run pension systems is whether the 
government should opt for government-operated or privately operated pension funds. 
Decentralized pension funds run by non-government agents are, of course, more 
consistent with the notion of a competitive market economy than either one central 
government-run fund or several separate government-operated funds. Government-
operated funds always run the risk of being controlled (“high-jacked”) by politicians 
insisting that they should decide the portfolio policy of the funds, appoint the members of 
the board of the fund(s), and perhaps also appoint board members in firms where the 
funds have bought shares. In other words, government-operated funds may in a long-term 
perspective result in a basically nationalized and centralized economy. 
The most promising way of significantly reducing the probability of political intervention 
in government-created pension funds, and hence de facto nationalization of the national 
economy, is to opt for a number of decentralized, non-government funds from the very 
beginning. The system could still be mandatory and the contributions collected by the 
government. Considering China’s recent tradition of government ownership and political 
intervention in individual firms, the risk (or “hope” among some observers) that a funded, 
government-created pension system will result in a strongly nationalized economy is 
hardly less in China than in other countries. 
Although China has chosen to build up a mandatory social insurance system, it seems that 
the government will share the responsibility for income security with firms (occupational 
benefits) and individual households (voluntary saving and voluntary insurance); see, for 

---

63
instance, Wong (1998, pp. 198 ff.).
127
Indeed, the Chinese government also encourages 
voluntary organizations to help the poor via donations (“charity”).
128
Let me finally turn to two country-specific problems connected with China’s emerging 
systems of social insurance. One is that risk pooling often only takes place across limited 
geographic areas, such as a city or a province, which creates a high degree of 
fragmentation of the systems across geographical areas and sectors. A second problem is 
the pronounced urban bias of the new emerging social insurance system – a bias which 
may to some extent be seen as an inheritance from the pre-1978 reform period. 
At first glance, the limited geographical domain of risk pooling may look like a trivial 
problem, since these domains are often more populous than many European nations. 
However, the composition of industries often differs strongly across geographical areas, 
so that the rates of payroll taxes also vary considerably across such areas. In particular, 
firms in areas with many unemployed or pensioners are exposed to much higher social 
costs than firms in other areas. For instance, payroll taxes are relatively high in regions 
with old industries, such as mining and steel, whereas they are relatively low in regions 
with new industries, such as banking, electronics and civil aviation. Thus, in spite of 
attempts to unify the social insurance systems across geographical areas and production 
sectors, economic and demographic factors have led to contributions rates that vary 
greatly across localities and sectors of the national economy. As a result, the social 
insurance system in China is highly fragmented. One consequence is that the relative 
competitiveness of firms is influenced in a rather arbitrary way, since local wages may not 
fully adjust to differences in payroll taxes. Another consequence is that variations in 
benefits across firms and regions limit the portability of entitlements and hence impede 
the emergence of a national labor market. 
The most obvious problem inherent in China’s emerging new arrangements for income 
security is, however, the bias in favor of urban insiders. First, within urban areas, 
individuals with only temporary residence permits, or no permits at all, are covered to a 
127
 
By the end of 2003, about 54 million individuals were reported to have signed voluntary old-age 
insurance with individual accounts. The Chinese Government also experiments with economic rewards 
in the form of an additional pension benefit (amounting to 600 yuan per year above the age of 60) for 
households that effectively practice family planning. 
 
128
China’s Social Security White Paper (September 7, 2004) reports that there were 28,000 such social 
donation centers, with aggregate donations amounting to 23 billion RMB during the period 1996-2003. 

---

64
very small extent, if at all.
129
Second, per capita social transfers are about 10 times as 
large in urban as in rural areas (UNDP, 2005, p. 3). Indeed, the dispersion of per capita 
public-sector spending across provinces seems to have widened rather than narrowed in 
recent decades (UNDP, 2005, p. 25).
One rationale for concentrating the arrangements of income insurance to urban areas is 
simply that systems of income support are more difficult to organize and administrate in 
agriculture than in other sectors. Indeed, the income (partly in kind) of individual 
members of farm families is even difficult to assess. Another rationale might be that the 
need for such systems would be less urgent in rural than in urban areas, since informal 
transfers within families, including remittances from relatives in urban areas, might be 
more important in the countryside. A third rationale for confining income insurance to 
urban areas might be that income risk for individuals would be smaller in agriculture than 
in industry, where business cycles are notorious. However, the last two rationales are 
rather weak, since the family is a very narrow unit for risk pooling and since fluctuations 
in factor income are hardly smaller in agriculture than in other production sectors.
130
Moreover, rationales do not constitute explanations. The latter have to be expressed in 
terms of the political process. One explanation why the authorities have favored urban 
insiders might be that these have more political clout than the rural population, possibly 
because the authorities are more concerned about social unrest in the cities than in the 
countryside. Sooner or later, however, strong political demands are likely to emerge for 
comprehensive arrangements of income security also among rural citizens in China. The 
129
According to a study by Gao et al. (2002) concerning five major cities, 74 percent of permanent 
urban residents were covered by pension benefits in 2000, while the corresponding figure for temporary 
residents was 10 percent. The corresponding figures for health insurance were 68 percent and 12 
percent, respectively, for unemployment benefits 33 percent and 1 percent, and for workplace injury 
insurance 25 percent and 14 percent. 
130 
A study by Jalan and Ravallion (1999) suggests, however, that such arrangements are rather limited in rural 
areas as well. Moreover, the need for income insurance is well established also for citizens in rural areas in 
China, for instance in connection with fluctuations in harvests and prices of agriculture products. This seems to 
be a particularly serious problem for the poorest groups of the rural population. For instance, Jalan and 
Ravallion (1999) report that while only 10 percent of the income shocks are passed onto current consumption in 
the case of individual households in the highest wealth decile in rural areas (in their sample), the corresponding 
figure is 40 percent for individual households in the poorest decile. Since many individuals in rural areas are 
tightly liquidity constrained, they may not be able to smooth consumption very much, for instance, in the event 
of fluctuations in agricultural output and prices. In other words, poor people often cannot afford to accumulate 
precautionary savings or take private insurance. Moreover, as an inheritance from the pre-reform period, there 
are probably rather few (formal and informal) moneylenders in the Chinese countryside specialized in loans to 
households as compared, for instance, to India, although the situation has been rapidly changing in recent years.

---

65
increasing number of so-called “social incidents” (unrest) in the countryside (70-80,000 
per year recently) could perhaps be regarded as an indication of a rise in such demands 
(the expropriation of land-lease contracts being another background factor). Indeed, 
experiments with income insurance in rural areas are already underway in a number of 
provinces,
131
and future reforms along these lines have recently been promised by the 
central government, for instance, in connection with the 11
th