TNAG-0531-FCO40-626-Application-of-International-Labour-Convention-to-Hong-Kong-1975 — Page 133

111

(capital and labour). Within this framework, it has been suggested that social security contributes to economic growth (1) by increasing capital formation through increasing savings and (2) by improving the quality of labour inputs, This section considers the first point, deferring the second to the next section.

There are two relevant questions under the heading of savings. One is whether social security increases the propensity of the covered persons to save. The other question is whether social security channels savings more effectively to apital formation than other ways in which people save, provided social security at least keeps society's savings constant.

on

a

A priori, it would be difficult to answer the first question one way or another. Much depends upon how the social security benefits are financed. If all social security benefits are financed by allocations from general tax revenues pay-as-you-go basis, savings in the name of social security are only a part of the fiscal surplus. But whether this increase in the fiscal surplus represents a net increase in society's aggregate savings is questionable. Contributory social insurances are often considered conducive to greater savings.

But, even here,

savings as differences between receipts of contributions and payments of benefits may result nore from actuarial prudence by overestimating the extent of contingencies (as may be possible in cases of sickness and unemployment insurances) or from the transitional phase of reserve accumulation before the maturity of a programme (as in the cases of pensions and survivors' insurances) than from an increase in society's propensity to save. Granted that participation in a social insurance is a form of saving for individuals, it is not clear whether it represents a net increase in savings of the participants rather than a diversion of savings from other forms to the social insurance. For all these and many other reasons, the usual reference to the surplus ΟΙ deficit on the social security accounts hardly proves anything about whether or not the propensity of the general public to save has increased because of social security progrannes.'

Be that as it may, it should be mentioned that none other than W. Arthur Lewis feels that social security stimulates savings among the working classes which ordinarily (as assumed in conventional growth economics) do not save or save far less than the middle classes. Lewis flatly declares with no careful argument Cr evidence: "The most powerful stimulus to working class savings is a social security system, which provides income for the rainy day. Working class savings then begin to build up, the working classes begin to have the same ambitions as the middle classes: to save for education, housing, furniture, and so

on. "2 This statement of Lewis's clearly implies that thanks to social security, the propensity of the working classes to save rises. Although logically and empirically I remain unconvinced about it, I bow to Lewis's authority and suggest that the hypothesis be put to test by appropriate data. Among developed countries, Aaron's study showed a negative cross-country correlation between household savings and social security developemnt. Among the Asian and Oceanic countries considered in this paper, the correlation co-efficient is too low (0.28) to be statistically significant. 3 In relation to Japan's unusually high

savings ratio, many Japanese economists have attributed it in part to the relative backwardness of Japan's social security system compared with developed countries, almost standing Lewis on his head.*

relationship

between

At best, one can conclude that the debate about the social security and household savings is inconclusive. But there is one

1 The most helpful discussion of how social security reserves are held is found in Paul Fisher's article mentioned above. However, the question asked in the text is not raised by Fisher.

2 W. Arthur Lewis, "Development economics: An outline", p. 33.

3 The correlation here is between the GNP proportion devoted to social security and the ratio of net savings to national income. On the other hand, when the former indicator is correlated with social security benefits as per cent of aggregate consumption, the correlation coefficient is almost perfect (0.9953). Compare this with the correlation between the indicator in question and per capita national income (0.9303). Both coefficients are very good, but one wonders what the much tighter correlation between a social security development indicator and the consumption ratio implies.

Taira and Kilby, "Differences

#*

• · p. 11.

E-1195-2F:5