Selected Issues

Abstract

Selected Issues

What Drives Private Savings in Singapore?1

Singapore has one of the highest national saving rates in the world, which contributes to its large current account surplus. National savings are driven largely by the private sector, with private savings constituting about 80 percent of national savings and 37 percent of GDP. Using data over the last four decades, this paper empirically examines the factors associated with the high private saving rate in Singapore. Based on time series and panel data analysis, the findings indicate that demographics and per capita income are the most important drivers of private savings in Singapore. While there is no strong evidence of private-public saving substitutability for the full sample period, private savings may have been partially offsetting changes in public savings since the 2000s.

A. Background

1. Singapore’s current account balance has been in a persistent surplus since the mid-1980s, which peaked at about 26 percent of GDP in 2007 (Figure 1). While the current account surplus almost halved in the immediate aftermath of the global financial crisis, it rebounded sharply in 2010–11 to 23 percent of GDP. Since then, it has averaged about 18 percent of GDP, largely because of a decline in the services and income balance.

Figure 1.
Figure 1.

Singapore: Current Account Balance and National Savings

Citation: IMF Staff Country Reports 2017, 241; 10.5089/9781484312858.002.A004

2. From a savings and investment perspective, Singapore’s large current account surplus since the early 2000s can be attributed to a rise in national savings, combined with lackluster investment that was sluggish to recover after the Asian financial crisis. National savings fell during the global financial crisis, but recovered in 2011, and have averaged about 46 percent of GDP since then. This ratio is one of the highest among Asian countries—most of which have also been running current account surpluses over the last decade, though of a smaller magnitude than Singapore (Figure 1). As in other Asian countries, however, the main contributor to national savings is private savings—with its share in national savings rising steadily from 68 percent in the 1990s to 75 percent in 2000s, and further to about 80 percent in 2012–16 (Figure 2).2

Figure 2.
Figure 2.

Singapore: Public and Private Saving

Citation: IMF Staff Country Reports 2017, 241; 10.5089/9781484312858.002.A004

3. Presently at around 37 percent of GDP, Singapore’s private saving rate is the highest among other advanced countries. Private savings averaged about 27 percent of GDP in the 1980s, before increasing to 33 percent in the 1990s, and further to 37 percent in 2010-16. The 10 percent of GDP increase in private savings over the last four decades is the sharpest among advanced countries (Figure 3). While some other major global financial centers such as Ireland and Switzerland also have a high private saving rate, for most advanced countries with similar per capita incomes, the average private savings rate is almost one-half of that.

Figure 3.
Figure 3.

Singapore: Labor Income and Corporate Profit

Citation: IMF Staff Country Reports 2017, 241; 10.5089/9781484312858.002.A004

4. Why is the private saving rate so high in Singapore? While several explanations have often been put forward to explain Singapore’s high private saving rate (such as demographics, high per capita income, global financial center status, a fully funded pension system), a comprehensive empirical analysis on the determinants of private saving has been lacking. In this paper, we address this gap by systematically examining the drivers of private saving in Singapore—taking into account its specific economic and structural features—using data covering the most recent time period (1980–2016). Since disaggregated data of private saving into household and corporate saving is unavailable, we analyze the behavior of total private saving (in percent of GDP).

A04ufig1

Singapore: Public and Private Saving

(In Percent of GDP)

Citation: IMF Staff Country Reports 2017, 241; 10.5089/9781484312858.002.A004

Sources: IMF, World Economic Outlook; and IMF staff

5. The analysis focuses on three key questions. First, how important are the macroeconomic and structural drivers of savings (such as demographics, income level, economic growth, public savings) in explaining the private saving behavior in Singapore. Second, has the importance of the various factors changed over time as Singapore has evolved into a rich advanced economy. And third, how does the behavior of private savings in Singapore compare with that of other advanced countries. To examine these questions, we conduct a country-specific time-series analysis of private savings in Singapore, as well as panel data analysis of private saving behavior in a broad sample of advanced economies over 1980–2016.

6. Our findings show that demographics and income per capita are important determinants of private savings in Singapore. Consistent with the life-cycle consumption model, a higher elderly dependency ratio is associated with significantly lower savings, while an increase in the projected age dependency ratio raises savings. Private savings in Singapore tend to rise with the income level and real interest rates, as well as with higher income growth. There is no evidence of substitutability between private and public savings over the full sample, though it appears that over the last decade, private savings may have been partially offsetting changes in public savings.

7. Results based on the panel data analysis suggest that in addition to demographic and macroeconomic variables, structural factors may also be important in explaining private saving behavior. Global financial centers like Singapore tend to have a higher private saving rate. While adequate and comparable data on pension systems across countries is unavailable, there is some evidence that countries with a mandatory defined contribution system have, on average, a higher private saving rate. Controlling for the relevant demographic, macroeconomic, and structural characteristics, Singapore’s average private saving rate turns out to be higher (by about 3-5 percent) than other advanced countries, but the “excess” saving seems to have fallen in recent years.

8. The rest of the paper is organized as follows. Section B outlines the analytical framework and estimation model for examining the determinants of private savings in Singapore. Section C presents the country-specific time series estimation results for Singapore, as well as the results obtained from panel data analysis. Section D discusses the policy implications of our findings and concludes.

B. Analytical Framework

9. Given the significant role that savings can play in promoting economic growth by stimulating investment, an extensive body of theoretical and empirical literature investigates the determinants of savings. Earlier literature was dominated by the life-cycle model, which emphasizes the role played by the age structure of the population in determining savings, while later studies have explored the importance of other possible factors (such as public savings, pension systems, income level, financial development, inflation, terms of trade, etc.) in stimulating savings.

Life-Cycle Hypothesis

10. The life-cycle model has been the standard theory for explaining the behavior of consumption (and savings) over time and across countries. The model is built around the saving behavior of a representative agent maximizing the present value of lifetime utility, subject to a budget constraint (which restricts lifetime consumption to be equal to the current net worth and the present value of expected labor income over the working life of the agent). Assuming perfect capital markets and perfect foresight, the model predicts that consumption depends on expected lifetime income (and not on current income) as individuals smooth consumption over their lifetimes. Given the fluctuations in income, saving thus depends on the stage in the life cycle—with individuals being net savers during their working years and dis-savers during retirement (Modigliani, 1986).

11. At the macro level, the obvious implication of the life-cycle model is that a high share of working age population would translate into higher savings as workers provide for their retirement. Conversely, the private (and aggregate) saving rate would decline as the share of elderly population that has reached retirement age increases. Given the rising longevity, however, not only the current but also future age dependency ratio may matter for the saving behavior—while a higher current share of elderly population implies lower savings, a higher share of elderly population in the future would imply higher savings as the population prepares for retirement.3

12. The impact of demographics on savings has been well established empirically. Both country-specific and cross-country studies generally find a strong negative effect of the share of elderly population (in relation to working age population) on private savings (e.g., Modigliani and Sterling, 1983; Graham, 1987; Masson and Tryon, 1990; Masson and others., 1998; Athukorala and Sen, 2004). For Singapore, demographics appear to have been an important factor in explaining the saving behavior. Husain (1995), for example, finds that the sharp increase in the share of working age population in total population raised the national saving rate by about 20 percentage points over 1970-92. Similarly, IMF (2006) finds that the rising share of elderly population in total population over the 1990s and 2000s contributed negatively to private savings in Singapore.

13. Based on the life-cycle model, income growth also has implications for savings. With an unchanged savings rate by age group, higher income growth would increase the aggregate income of the working age population (relative to the older age population not earning labor income), thereby raising aggregate savings (Modigliani, 1966). Tobin (1967), however, notes that unchanged saving rates are only consistent with myopic expectations of future income—if workers expect income growth to be permanent, then based on the life-cycle model, such wealth effects may induce more consumption today (for example, if young workers borrow to finance current consumption in view of the high lifetime wealth). Thus, individual saving rates could fall sufficiently to offset the aggregate saving effects of higher income growth. Empirical literature—including for Singapore (e.g., Lahiri, 1989; Husain, 1995)—has generally documented a significant positive effect of income growth on private (and aggregate) savings.

14. Real interest rate on bank deposits and total wealth may also influence private savings under the life-cycle model. The net effect of interest rates on savings is theoretically ambiguous because of potentially offsetting substitution and income effects. A higher real interest rate increases the present price of consumption relative to the future price, providing an incentive to increase saving. But if the individual is a net creditor in financial assets, a higher real interest rate increases lifetime income, and may increase consumption (reduce saving) through the income effect.4 Paralleling the theoretical ambiguity, empirical research documents mixed results on the impact of real interest rates on savings (Bosworth, 1993; Schmidt-Hebbel and others, 1992). For wealth, while the theoretical prediction is straightforward—that is, by reducing dependence on current income, higher wealth allows higher consumption (and lower saving)—the empirical evidence is less clear. While several studies find a statistically significant effect of changes in wealth (proxied by changes in house prices given that housing wealth tends to be the dominant form of household wealth) on private saving in advanced countries (e.g., Muellbauer and Murphy, 1997; Benjamin and others, 2004; Case and others, 2005), the association between private residential property prices and the private saving rate is not documented to be statistically strong for Singapore (Phang, 2002; IMF, 2006).5

Public-Private Saving Offset

15. Under the Ricardian equivalence hypothesis, public and private savings are perfect substitutes. Any change in public savings is offset by private savings as it will be accompanied by a change in future taxation; public saving will thus not affect national saving (Barro, 1974). Empirical evidence on the private saving offset to government saving is decidedly mixed with studies generally finding limited evidence of a full offset (e.g., Bernheim, 1987; Corbo and Schmidt-Hebbel, 1991; Haque and Montiel, 1989; Mason and others, 1998). This may be because of liquidity constraints faced by the private sector (Hayashi, 1985; Campbell and Mankiw, 1989), or because of the differential effects of different types of government expenditure and revenue on private savings (Callen and Thimann, 1997). In particular, if government investment is viewed as productive, and not expected to require tax increases in the future, it may not generate a response by the private sector. By contrast, government consumption spending, or investment that is not expected to generate future revenues, might induce a larger private saving offset.

16. The role played by the public sector in influencing private savings in Singapore is unclear. While some studies report high substitutability between public and private savings, others finding no strong offset by private savings to public savings (Husain, 1995; IMF, 2006). Looking at Figure 2, it appears that part of the problem in detecting a relationship between private and public savings may be the lack of a stable relationship between the two series. Through the 1980s up to the mid-1990s, for example, public and private savings are positively correlated. But this trend appears to change after the Asian financial crisis, and then more recently in the aftermath of the global financial crisis, when public savings declined (as a result of government efforts to stimulate the economy), while private savings remained high.

Pension System

17. Given the life-cycle hypothesis, and the growth in retirement programs, the impact of pension plans on private saving behavior has garnered much academic interest. Several studies argue that the existence of private pension plans imply a net increase in private savings. Economic theory, however, suggests that it may simply reflect a shift in the form of savings as employees covered by pension plans reduce their own direct savings in response to expected retirement benefits. While comparable cross-country data on pension systems is lacking, there is some evidence that individuals reduce their direct saving in anticipation of promised pension benefits, though the offset may not be perfect (Munnell and Yohn, 1991; Bosworth and Burtless, 2004).6

18. In Singapore, a key aspect of private savings is the mandatory government-managed Central Provident Fund (CPF). Initiated in 1955, the CPF requires both employees and employers to contribute a certain portion of labor wages to the fund. The contribution rates were originally set at levels (5 percent for both employees and employers) to help individuals finance a basic retirement income. The rates have however increased substantially over time (17 percent for the employer and 20 percent for the employee) as the CPF has evolved into a saving vehicle to help individuals finance housing, health, investment, and other outlays. The sizable mandatory contribution rates have attracted much attention over the years, with critics arguing that they represent a form of “forced saving,” which increases the aggregate saving rate of Singapore. The evidence for the forced saving hypothesis is however weak: Husain (1995), for example, finds that changes in the ratio of CPF savings to private disposable income do not impact private savings significantly—suggesting that CPF savings are offset by individuals’ voluntary savings. Looking at the data for recent years (2010–16), the correlation between CPF contributions (net of withdrawals) to GDP ratio and the overall private saving rate is close to zero, supporting the earlier finding of an offset by voluntary saving to the mandatory CPF saving.

Other Factors

19. Several other macroeconomic and structural factors could also potentially affect the private saving behavior:

  • Differences in per capita income level could be an important factor influencing savings. At subsistence levels, the potential for savings is limited, a rise in per capita income therefore may lead to higher savings. The size of this effect, however, may decline as per capita income rises and may even become negative for rich countries where investment opportunities and growth are relatively lower. Existing studies generally find strong evidence that rising income levels that expand the savings base are associated with private savings. For Singapore also, the rise in income per capita over the years is considered to have significantly affected the private saving rate (Husain, 1995; IMF, 2006).7

  • Financial liberalization and development could lower savings by increasing private sector’s access to credit. At the same time, they could also raise the private saving rate by providing more opportunities for saving (McKinnon, 1973; Shaw, 1973). Empirical evidence generally supports a negative relationship between financial liberalization (access to foreign savings) or financial development (proxied by domestic credit to GDP ratio) and private (or national) savings (e.g., Fry, 1980; Giovannini, 1985; Schmidt-Hebbel and others, 1992; Loayza and others, 2000).

  • Inflation is another factor that may affect saving through several channels. In the life-cycle model, the impact of inflation on saving stems through its effect on the real interest rate (i.e., real returns to saving), but inflation could also impact real wealth (if consumers want to maintain a certain level of wealth relative to income, saving will rise with inflation). Moreover, by increasing uncertainty about future income stream, inflation could lead to higher saving on precautionary grounds as well. Cross-country studies tend to find a positive effect of inflation on the private saving rate in advanced countries (e.g., Mason and others, 1998), but the impact of inflation has been found to be statistically unimportant for Singapore, perhaps because of its historically low inflation rate (Lahiri, 1989; Husain, 1995).

  • Terms of trade improvement could increase savings (through the Harberger-Laursen-Metzler effect); though recent literature argues that the impact may depend on whether the terms of trade change is transitory or permanent.8 A transitory change in income should increase savings, while permanent shocks to the terms of trade would have ambiguous effects that should be small in magnitude. The cross-country evidence generally supports the positive association between terms of trade changes and the saving rate; however, existing studies find no strong statistical association between terms of trade changes and private savings in Singapore (Lahiri, 1989; Husain, 1995; IMF, 2006).

  • Status as a global financial center could influence the private saving rate, as the private sector self-insures against external financial shocks. Countries with large financial centers in general report a higher private saving rate—in 2016, for example, the average private saving rate in Hong Kong SAR, Ireland, Luxembourg, Singapore, and Switzerland was about 10 percentage points higher than in other advanced countries.9

  • Corporate savings is another channel through which private savings may be boosted. From a theoretical perspective, high corporate profits should not necessarily affect savings since, in the absence of tax distortions and liquidity constraints, household saving would decline to offset any change in corporate saving. Empirical studies, however, generally do not find support of a full offset by households to corporate savings (Poterba, 1991; Auerbach and Hassett, 1991). For Singapore, a breakdown of private saving into corporate and household saving is unavailable, which prevents us from rigorously analyzing their dynamics, but estimates based on the high share of corporate income in total economic output suggest that corporate savings likely play a non-negligible role in contributing to private savings (Figure 4). In fact, if proxied by pre-tax corporate profits (less dividends), corporate profits account for over half of private savings in 2010–14 (Figure 4).10 While the high corporate saving rate is by no means unique to Singapore—corporate savings have risen globally over the years—a comparison of publicly listed companies reveals that Singapore’s corporate profitability (measured as earnings before interest, taxes, and return on assets) is at par with the OECD countries, yet the corporate retained earnings are higher (Figure 5). This could be attributed to a significant presence of foreign multinational corporations (MNCs) in

Figure 4.
Figure 4.

Singapore: Corporate Profits and Retained Earnings

Citation: IMF Staff Country Reports 2017, 241; 10.5089/9781484312858.002.A004

Singapore, which are more prone than domestic firms to retain a large share of their earnings (Akhtar, 2017).11

Empirical Model

20. Drawing on the literature outlined above, we empirically examine the drivers of private savings in Singapore by estimating the following equation:

St=α0+α1St1+Σj=1JβjDjt+Σk=1KγkXkt+ɛt(1)

where St is private saving in percent of GDP in year t; St−1 is lagged private saving in percent of GDP to capture any persistence in saving behavior; D includes demographic variables such as the share of elderly and working age population in total population, population growth, and projected elderly age dependency ratio; X indicates different macroeconomic variables that may influence private saving such as real income per capita, income growth, public saving (in percent of GDP), real returns to saving (proxied by the real deposit rate), unemployment rate, change in terms of trade, the inflation rate, and financial sector development (proxied by domestic credit in percent of GDP); and ε is the random error term. We estimate eq. (1) for the period 1980–2016 using the ordinary least squares (OLS) methodology with robust standard errors, as well as the Prais-Winsten generalized least squares method to allow for possible first-order serial correlation in the error term.12

21. In addition to a Singapore-specific private saving function, we also estimate the private saving function for a panel of advanced countries for comparative purposes. While eq. (1) allows us to determine the factors associated with private savings in Singapore, we assess the importance of the different factors relative to other advanced countries by estimating the following equation for a panel of 27 advanced countries:

Sit=γ0+γ1Sit1+Ditθ+Xitυ+Zitδ+λt+ηit(2)

where Sit is private saving in percent of GDP in country i in year t; Sit-1 is lagged private saving to GDP; D and X are matrices including the demographic and macroeconomic variables outlined above, respectively; Z includes variables capturing the structural characteristics of the economy that may influence aggregate savings (such as status as a global financial center; the type of pension system); λ reflects time effects to capture the impact of any common global factors on the private saving rate; and η is the random error term. We estimate eq. (2) for the period 1980–2016 using the OLS and Prais-Winsten methods (without and with country-fixed effects), and calculate panel-corrected standard errors to address potential contemporaneous correlation in the error term.

C. Estimation Results

Private Savings in Singapore

22. The estimation results of equation (1) suggest that demographics play a key role in supporting private savings in Singapore. Controlling for other relevant factors, an increase in the share of elderly people in the population by one percentage point lowers the private saving rate by about 4 percentage points, while an increase in the future elderly dependency ratio by one percentage point increases the private saving rate by about 1 percentage point (Table 1, col. [1]). There does not appear to be a statistically significant association between the share of working-age population and private savings in Singapore, but a higher rate of population growth strongly raises private savings—perhaps because it better captures the increase in labor force associated with foreign workers (col. [2]).

Table 1.

Singapore: Private Saving Function, 1980–2016

article image
Notes: Dependent variable is private saving (in pct. of GDP). d1980, d1990, and d2000 are dummy variables equal to one for the years 1980-89, 1990-99, and 2000-09, respectively, and zero otherwise (the base period is 2010-16). All specifications are estimated with the OLS method, and include a constant. DW-statistic is the Durbin-Watson statistic for serial correlation in the residuals. Robust standard errors are reported in parentheses. ***, **, and * indicate statistical signficance at the 1, 5, and 10 percent levels, respectively.

23. Higher real deposit rates and higher per capita income induce more savings. A one percentage point increase in the real deposit rate implies about a one percentage point increase in the private saving rate, while quantifying the effect of income per capita suggests that—holding everything else constant—about 15 percent of the rise in the private saving rate in Singapore over 1980 to 2016 could be attributed to the increase in its per capita income. The positive and large coefficient on real income per capita conforms with economic intuition that the propensity to save rises with the income level, but could also be capturing the rise in savings as a result of Singapore’s evolution into a global financial center and multinational corporation hub. In addition, higher real GDP growth is also associated with higher savings, suggesting that much of the income gains are perceived to be temporary by the private sector.13

24. For the full sample period, there does not appear to be a private-public savings tradeoff. The coefficient on the public saving rate is negative but wholly statistically insignificant, suggesting that there is no private sector off-set from higher public savings over the sample period. This could be because the private sector is unsure about whether (or when) the fiscal surpluses of today will be available to them in the future, or because credit constraints prevent the private sector from borrowing against future distributions of fiscal surpluses (IMF, 2006). It could also be a cultural preference for thriftiness and prudence, with an overly cautious attitude of individuals toward risk and uncertainty.

25. Among other factors, improved business confidence is associated with higher private savings. While this finding may seem counterintuitive as improved economic expectations and lower uncertainty should reduce the need for precautionary saving, it likely reflects the impact of higher corporate income/profits on savings that is positively correlated with business prospects. Higher unemployment rates imply lower savings, presumably because of lower disposable incomes, but the effect is not statistically significant for the full sample period. The effect of other macroeconomic variables such as terms of trade change, credit to GDP ratio, change in credit to GDP ratio, and change in the real house price index (to capture the wealth effect) is also statistically insignificant. There is strong evidence, however, of persistence in private saving behavior—the coefficient on the lagged private saving rate is highly statistically significant in all specifications, implying a half-life of deviation of about one year. These findings are largely robust to the inclusion of a trend term in the model, or addressing potential serial correlation in the error term (Appendix Table 2).

26. Importantly, the effect of several demographic and macroeconomic variables appears to have changed over time. Including interaction terms between the demographic variables and binary variables equal to one for the decades of the 1980s, 1990s, and 2000s (with the post-global financial crisis years, 2010–16, as the reference category), we find that the dissaving effect of a rising share of elderly population is more pronounced in the recent period than it was in the 1980s and 1990s (Table 1, col. [4]). At the same time, the impact of an increase in population growth and the projected future elderly dependency ratio is smaller in recent years than it was in the earlier decades (cols. [5]-[6]).14 The stronger impact of population aging in earlier years perhaps reflects tighter credit constraints or a less expansive social safety net then as compared to now.15 Adding interaction terms between the public saving rate and the binary variables for the various decades, the results show a negative and statistically significant coefficient on the public savings variable (col. [7]). When combined with the coefficients on the interaction terms, however, formal tests find no support for Ricardian behavior in the 1980s and 1990s, but evidence of an offset since the 2000s.

28. Considering the aggregate savings function confirms these results. Both the current and projected elderly dependency ratios play a key role in driving national savings; the impact of these factors has, however, declined since the 1980s (Appendix Table 3). Rising incomes over the years have also contributed to a higher national savings rate. Public savings contribute significantly to national savings with the coefficient indicating a very modest offset by private savings over the full sample period (formal tests fail to reject the hypothesis that the Ricardian equivalence does not hold; col. [1]). Much of this effect, however, stems from the earlier years, and a rise in public savings since the 2000s does not appear to have been statistically significantly associated with a rise in national savings (col. [7]).16

Cross-Country Analysis

29. Estimating the private savings function for a panel of advanced countries reinforces the important role played by demographics and the income level. The results obtained from estimating eq. (2) suggest that, consistent with the life-cycle hypothesis, a higher share of elderly population implies a lower private saving rate, while a higher projected age dependency ratio implies greater savings (Table 2). The impact of these variables on the private saving rate is, however, smaller in the estimated panel than it is for Singapore: on average, a one percentage point increase in the share of elderly population in total population lowers the saving rate by about 0.1 to 1 percentage point (depending on whether country-fixed effects are considered), while a one percentage point increase in the projected future age dependency ratio raises the saving rate by about 0.1-0.2 percentage points. Similar to the results reported in Table 1, both higher real per capita income and income growth are strongly associated with higher private savings in the panel estimations (though the effect of per capita income is much larger for Singapore than for other advanced countries). The coefficient on the lagged dependent variable is also significantly positive, providing strong evidence on persistence in private saving behavior.

Table 2.

Singapore: Private Saving Function in Other Advanced Countries, 1980–2016

article image
Notes: Dependent variable is private saving (in pct. of GDP). Cols. [1]- [4] are estimated with the OLS method, and cols. [5]-[8] are estimated with the Prais-Winsten method to account for possible first-order serial correlation in the error term. All specifications include a constant. Panel-corrected standard errors are reported in parentheses. ***, **, and * indicate statistical signficance at the 1, 5, and 10 percent levels, respectively.

30. The results suggest a partial but imperfect private sector offset to public savings. The coefficient on the public saving rate is negative and strongly statistically significant, but formal tests reject the Ricardian Equivalence hypothesis. Across specifications, a one percentage point rise in the public saving rate implies a reduction in private savings by about 0.2-0.6 percent of GDP. Among other factors, an improvement in terms of trade, a lower unemployment rate, and a lower credit to GDP ratio are associated with significantly higher private savings, while the coefficient on real deposit rate is statistically insignificant.

31. Structural characteristics also appear to be important drivers of private savings. On average, global financial centers have a significantly higher private saving rate (by about 2 percentage points; Table 2, cols. [6]-[7]), while countries with a defined contribution pension system also have a higher private saving rate by about one percent of GDP. Considering also the size of the economy by including a dummy variable equal to one if the population size is less than 5 million (and zero otherwise), we find the coefficient to be negative and statistically significant at the 5 percent level suggesting that controlling for other factors, smaller states have on average a lower private savings rate.

32. While the considered demographic, macroeconomic, and structural factors explain Singapore’s private saving behavior, they appear to be only part of the story. Including Singapore in the sample of advanced countries and estimating eq. (2), the obtained results are quite similar to those reported above (Table 3). Adding a country-specific (binary) variable for Singapore in the model, however, we find its coefficient to be significantly positive—indicating that private savings in Singapore are on average about 3-5 percent of GDP higher than in other advanced countries (Table 3, cols. [3], [7]). This result does not change if we include quadratic terms for the age dependency ratio or per capita income to capture any non-linear effects of these variables, or consider additional variables such as the size of the economy, the inflation rate, and public health spending to GDP ratio (to capture the variation in the social safety net across countries). This suggests that beyond the traditional variables considered in the literature, some other factors (such as cultural norms, or aspects of the tax system) that encourage thriftiness may be at play in driving Singapore’s high private saving rate. It could also be that the existing variables do not adequately capture Singapore’s structural features such as its status as a non-reserve issuing currency global financial hub, its special pension system, and the extent of its social safety net.

Table 3.

Singapore: Private Saving Function in Advanced Countries Including Singapore, 1980–2016

article image
Notes: Dependent variable is private saving (in pct. of GDP). Cols. [1]- [4] are estimated with the OLS method, and cols. [5]-[8] are estimated with the Prais- Winsten method to account for possible first- order serial correlation in the error term. All specifications include a constant. Panel- corrected standard errors are reported in parentheses. ***, **, and * indicate statistical signficance at the 1, 5, and 10 percent levels, respectively.

33. The unexplained component of Singapore’s private saving behavior, however, has been declining over the years. Estimating the specification in Table 3 (cols. [3], [7]) for the different subperiods (1980-89, 1990-99, 20009-09, and 2010-16), the coefficient on the dummy variable for Singapore is the largest for the 1990s and 2000s, but has more than halved since the global financial crisis and is only weakly statistically significant (Table 4, cols. [4], [8]).17 This suggests that, given Singapore’s economic and structural characteristics, its private saving rate may have been less excessive compared to other advanced countries in the recent years than during the pre-global financial crisis years.

Table 4.

Singapore: Private Saving Function in Advanced Countries: Sub-samples

article image
Notes: Dependent variable is private saving (in pct. of GDP). Cols. [1]- [4] are estimated with the OLS method, and cols. [5]- [8] are estimated with the Prais- Winsten method to account for possible first- order serial correlation in the error term. All specifications include a constant. Panel- corrected standard errors are reported in parentheses. ***, **, and * indicate statistical signficance at the 1, 5, and 10 percent levels, respectively.

D. Conclusion

34. Private saving in Singapore are amongst the highest in advanced countries. Empirical analysis conducted here shows that consistent with the life-cycle consumption model, Singapore’s demographic structure plays a key role in influencing its private savings. A growing share of elderly population is strongly associated with lower savings, while a rise in expected future old age dependency ratio increases savings. In addition, private savings in Singapore are also positively affected by macroeconomic characteristics such as rising real per capita income, real interest rates, and income growth, while higher unemployment rate is associated with dissaving. Over the long run, there is no evidence of public savings being offset by private savings, though it seems that the public-private saving relationship may have changed since the 2000s.

35. Cross-country empirical analysis suggests that Singapore’s average private saving rate is about 3-5 percent higher than other advanced countries, even after controlling for relevant factors. Much of this average difference in the private saving rate can, however, be attributed to the private saving behavior in the 1990s and 2000s; the unexplained component of Singapore’s private saving rate has almost halved in recent years.

36. Looking ahead, aging is expected to significantly lower private savings. Given the projected rise in the age dependency ratio, however, households may save more in the preceding years for precautionary purposes, thereby contributing to the current account surplus (as discussed in Appendix II of IMF, 2017). Higher incomes through productivity growth are also likely to raise private savings. In addition, given the rising offset in private savings due to changes in public savings, there is a possibility of greater fiscal spending (smaller fiscal surpluses) to contribute to even higher private savings in the medium-term. It would thus be important to anchor expectations of the private sector regarding the future course of fiscal policy.

References

  • Abeysinghe, T., and K. Choy, 2004, “The Aggregate Consumption Puzzle in Singapore,Journal of Asian Economics, 15 (3): 563578.

  • Akhtar, S., 2017, “Dividend Policies Across Multinational and Domestic Corporations–An International Study,Account and Finance, forthcoming.

    • Search Google Scholar
    • Export Citation
  • Athukorala, P., and K. Sen, 2004, “The Determinants of Private Savings in India,World Development, 32 (3): 491503.

  • Auerbach, A., and K. Hassett, 1991, “Corporate Savings and Shareholder Consumption,” in D. Bernheim and J. Shoven, eds., National Saving and Economic Performance, Chicago: University of Chicago Press, 7598.

    • Search Google Scholar
    • Export Citation
  • Barro, R., 1974, “Are Government Bonds Net Wealth?Journal of Political Economy, 82 (6): 10951117.

  • Benjamin, J., P. Chinloy, and G. Jud, 2004, “Real Estate versus Financial Wealth in Consumption,The Journal of Real Estate Finance and Economics, 29 (3): 341354.

    • Search Google Scholar
    • Export Citation
  • Bernheim, B., 1987, Ricardian Equivalence: An Evaluation of Theory and Evidence, NBER Working Paper 2330, Cambridge, MA: National Bureau of Economic Research.

    • Search Google Scholar
    • Export Citation
  • Bosworth, B., 1993, Saving and Investment in a Global Economy, Washington DC: Brookings Institution.

  • Bosworth, B., and G. Burtless, 2004, Pension Reform and Saving, Washington DC: Brookings Institution.

  • Callen, T., and C. Thimann, 1997, “Empirical Determinants of Household Saving: Evidence from OECD Countries,IMF Working Paper WP/97/181, Washington DC: International Monetary Fund.

    • Search Google Scholar
    • Export Citation
  • Campbell, J., and N. Mankiw, 1989, “Consumption, Income, and Interest Rates: Reinterpreting the Time-Series Evidence,NBER Working Paper 2924.

    • Search Google Scholar
    • Export Citation
  • Case, K., J. Quigley, and R. Shiller, 2005, “Comparing Wealth Effects: The Stock Market versus the Housing Market,The B.E. Journal of Macroeconomics, 5 (1): 134.

    • Search Google Scholar
    • Export Citation
  • Corbo, V., and K. Schmidt-Hebbel, 1991, “Public Policies and Saving in Developing Countries,Journal of Development Economics, 36 (July): 89115.

    • Search Google Scholar
    • Export Citation
  • Edelstein, R., and S. Lum, 2004, “House Prices, Wealth Effects, and the Singapore Macroeconomy,Journal of Housing Economics, 13(4): 342367.

    • Search Google Scholar
    • Export Citation
  • Fitzgerald, J., 2013, “The Effect of Redomiciled Plcs on GNP and the Irish Balance of Payments,ESRI Research Note 2013/1/2, Dublin, The Economic and Social Research Institute.

    • Search Google Scholar
    • Export Citation
  • Fry, M., 1978, “Money and Capital or Financial Deepening in Economic Development?Journal of Money, Credit, and Banking, 10 (November): 464475.

    • Search Google Scholar
    • Export Citation
  • Giovannini, A., 1985, “Saving and the Real Interest Rate in LDCs,Journal of Development Economics, 18 (August): 197281.

  • Graham, J., 1987, “International Differences in Saving Rates and the Life Cycle Hypothesis,European Economic Review, 31 (8): 1509-1529.

    • Search Google Scholar
    • Export Citation
  • Haque, N., and P. Montiel, 1989, “Consumption in Developing Countries: Tests for Liquidity Constraints and Finite Horizons,Review of Economics and Statistics, 71 (3): 408-415.

    • Search Google Scholar
    • Export Citation
  • Hayashi, F., 1985, “The Permanent Income Hypothesis and Consumption Durability: Analysis Based on Japanese Data,Quarterly Journal of Economics, 100 (November): 10831113.

    • Search Google Scholar
    • Export Citation
  • Husain, A., 1995, “Determinants of Private Savings in Singapore,” in Singapore: A Case Study in Rapid Development, ed., K. Bercuson, IMF Occasional Paper 119, Washington DC: IMF

    • Search Google Scholar
    • Export Citation
  • IMF, 2006, Singapore: Selected Issues, IMF Country Report No. 06/151, International Monetary Fund, Washington DC.

  • IMF, 2017, Staff Report for Singapore 2017 Article IV Consultation, International Monetary Fund, Washington DC.

  • Lahiri, A., 1989, “Dynamics of Asian Savings: The Role of Growth and Age Structure,IMF Staff Papers, 36 (1): 228261.

  • Lee, J., G. Milesi-Ferretti, A. Prati, J. Ostry, and L. Ricci,Exchange Rate Assessments: CGER Methodologies,IMF Occasional Paper 261, Washington DC: International Monetary Fund.

    • Search Google Scholar
    • Export Citation
  • Loayza, N., K. Schmidt-Hebbel, and L. Serven, 2000, “What Drives Private Saving Across the World?Review of Economics and Statistics, 82 (2): 165181.

    • Search Google Scholar
    • Export Citation
  • Mancini-Griffoli, T., and N Stoffels, 2012, “Adjusting the Current Account to Better Capture Wealth Accumulation,mimeo, Swiss National Bank.

    • Search Google Scholar
    • Export Citation
  • Masson, P., and R. Tryon, 1990, “Macroeconomic Effects of Projected Population Aging in Industrial Countries,IMF Staff Papers, 37 (3): 453485.

    • Search Google Scholar
    • Export Citation
  • Masson, P., T. Bayoumi, and H. Samiei, 1998, “International Evidence on the Determinants of Savings,World Bank Economic Review, 12 (3): 483501.

    • Search Google Scholar
    • Export Citation
  • McKinnon, R., 1973, Money and Capital in Economic Development, Washington DC: Brookings Institution.

  • Modigliani, F., 1966, “The Life-Cycle Hypothesis of Saving, the Demand for Wealth, and the Supply of Capital,Social Research, 33 (summer): 160217.

    • Search Google Scholar
    • Export Citation
  • Modigliani, F., 1986, “Life Cycle, Individual Thrift, and the Wealth of Nations,American EconomicReview, 76 (3): 297-313.

  • Modigliani, F., and A. Sterling, 1983, “Determinants of Private Saving with Special Reference to the Role of Social Security: Cross Country Tests,” in, eds., F. Modigliani and R. Hemming, The Determinants of National Saving and Wealth, Macmillan, London.

    • Search Google Scholar
    • Export Citation
  • Muellbauer, J., and A. Murphy, 1997, “Boom and Busts in the UK Housing Market,The Economic Journal, 107: 17011727.

  • Munnell, A., and F. Yohn, 1991, “What is the Impact of Pensions on Saving?Federal Reserve Bank of Boston Working Paper 915.

  • Phang, S., 2002, “The Impact of Housing Prices on Aggregate Consumption: Evidence from an East Asian City-State,Singapore Management University Economics and Statistics Working Paper 7-2002.

    • Search Google Scholar
    • Export Citation
  • Phillips, S., L. Catão, L. Ricci, R. Bems, M. Das, J. Giovanni, D. Unsal, M. Castillo, J. Lee, J. Rodriguez, and M. Vargas, 2013, “The External Balance Assessment (EBA) Methodology,IMF Working Paper WP/13/272, Washington DC: International Monetary Fund.

    • Search Google Scholar
    • Export Citation
  • Poterba, J., 1991, “Dividends, Capital Gains, and the Corporate Veil: Evidence from Britain, Canada, and the United States,” in D. Bernheim and J. Shoven, eds., National Saving and Economic Performance, Chicago: University of Chicago Press, 4971.

    • Search Google Scholar
    • Export Citation
  • Schmidt-Hebbel, K., S. Webb, and G. Corsetti, 1992, “Household Saving in Developing Countries: First Cross-Country Evidence,World Bank Economic Review, 6 (3): 529547.

    • Search Google Scholar
    • Export Citation
  • Shaw, E., 1973, Financial Deepening in Economic Development, New York: Oxford University Press.

  • Tobin, J., 1967, “Life-Cycle Saving and Balanced Growth,” in William Fellner, ed., Ten Economic Studies in the Tradition of Irving Fisher, pp. 23156. New York: John Wiley and Sons.

    • Search Google Scholar
    • Export Citation

Appendix I. Data and Additional Results

Table 1.

Singapore: Data Description and Sources

article image
Table 2.

Singapore: List of Other Advanced Countries in the Sample

article image
Table 3.

Singapore: Private Saving Function with Prais-Winsten Method, 1980-2016

article image
Notes: Dependent variable is private saving (in pct. of GDP). All specifications are estimated using the generalized least-squares method assuming AR(1)errors, and include a constant. Robust standard errors are reported in parentheses. ***, **, and * indicate statistical signficance at the 1, 5, and 10 percent levels, respectively.
Table 4.

Singapore: National Saving Function, 1980-2016

article image
Notes: Dependent variable is national saving (in pct. of GDP). d1980, d1990, and d2000 are dummy variables equal to one for the years 1980- 89, 1990- 99, and 2000- 09, respectively, and zero otherwise (the base period is 2010- 16). All specifications are estimated with the OLS method, and include a constant. Robust standard errors are reported in parentheses. ***, **, and * indicate statistical signficance at the 1, 5, and 10 percent levels, respectively.