Singapore: Selected Issues
Author:
International Monetary Fund
Search for other papers by International Monetary Fund in
Current site
Google Scholar Close
Publication Date:
26 Apr 2004
eISBN:
9781451834208
Language:
English
Keywords:
ISCR; CR; Singapore; surplus; external shock; investment; Singapore's economy; growth experience; factor share; production function approach; growth record; Output gap; Capital accumulation
Download PDF (236.8 KB)

The paper first uses the production function to analyze the sources of past growth in Singapore and compares it with the experience of other Asian and industrialized economies. This study also provides some thoughts on how to boost medium-term growth prospects in Singapore, and assesses the growth slowdown of the past few years in Singapore reflecting cyclical versus structural factors. The assessment given in this paper suggests that there are returns to be had from investment in education and structural reforms.

Abstract

The paper first uses the production function to analyze the sources of past growth in Singapore and compares it with the experience of other Asian and industrialized economies. This study also provides some thoughts on how to boost medium-term growth prospects in Singapore, and assesses the growth slowdown of the past few years in Singapore reflecting cyclical versus structural factors. The assessment given in this paper suggests that there are returns to be had from investment in education and structural reforms.

II. Medium-Term Growth Prospects1

A. Introduction

1. Despite the downturn of recent years, Singapore’s growth record over the past four decades has been impressive. Per capita GDP growth from 1960–2003 averaged about 8 percent. In the past few years, however, growth has slowed down considerably. It was negative in 2001 and only slightly positive in 2002 and 2003. This chapter addresses the following questions: does this slowdown reflect a permanent decrease in the growth potential of the Singapore’s economy, and what is the medium-term growth potential? In other words, how much of the growth slowdown is permanent, and how much is cyclical?

2. Estimating a projected path for potential growth is sensitive to the theory of growth that is applied. The standard view is the so-called “neo-classical” growth theory in which aggregate production is subject to decreasing returns on capital and labor. Assuming competitive factor markets, this implies that a country with a low capital stock relative to advanced economies will experience temporarily higher growth rates as physical and human capital levels catch up to those observed in advanced economies. As the economy matures, however, there are decreasing returns to scale, meaning that each additional unit of physical and human capital yields lower returns, and hence the growth rate of the economy converges to the growth rate of the advanced economies.

3. This chapter reviews Singapore’s past growth experience, and assesses its future growth potential by applying a neoclassical production function. The chapter first uses the production function to analyze the sources of past growth in Singapore and compares it to the experience of other Asian and industrialized economies. The results indicate that a substantial part of Singapore’s past growth can be explained by capital accumulation. These results are then used to gain insights into medium-term growth potential which is projected to be in the range of 4–5 percent per year. The chapter also discusses alternative estimates of the output gap, indicating that there is currently a considerable output gap in Singapore, on the order of 2.5–4.5 percent. This suggests that in the short run the economy may overshoot its medium-term growth potential as the economy recovers.

B. The Neoclassical Growth Theory and Growth Accounting

4. This section reviews Singapore’s growth experience and contrasts it to that of other Asian and industrialized countries.2 (See Box II.1, where the results are put in the context of a large and controversial literature on Singapore’s growth.) It is assumed that production is characterized by a Cobb-Douglas production function:

The TFP Debate

The debate began with Young (1992) who argued, contrary to popular perception at the time, that economic growth in Singapore was driven by factor accumulation rather than by productivity growth. Of the average GDP growth rate of 8 percent, he estimated that growth in total factor productivity (TFP) accounts for less than 0.5 percentage points from the mid-1960s to 1990, with the remainder attributable to the accumulation of factor inputs. A study by Wong and Seng (1997) largely confirmed these results for 1975–1985 but showed a marked improvement in TFP growth in 1985–1995 (the latter half of that period is not covered in Young’s estimate). Young’s conclusions were widely discussed, for example by Krugman (1994), who interpreted them as indicating that Singapore could face a “Soviet-style” growth collapse.

TFP growth in Singapore has been broadly in line with that in advanced economies, and even slightly higher according to the estimate presented in this paper. The difference in these results, relative to Young’s, is mainly due to the differing assumptions about the factor shares in the production function (1). The table shows how the results presented in this chapter would change if TFP were calculated by the average value of the factor shares assumed by Young. If Young’s factor shares are assumed, the results would be much closer to what he obtained. As has been emphasized by Gollin (2002), however, there are important measurement problems with factor shares. He finds that once these problems are addressed, factor shares are nearly constant across countries and time in the sample he studies. His result, therefore, provides some justification for the assumption maintained in this exercise. An estimate by Sarel (1997) gives further justification for the assumed value of the factor shares for Singapore.

TFP in Singapore

(In percent)

article image

Other recent work points to similar conclusions as those in this chapter, namely that TFP growth in Singapore has not been substantially lower than in other advanced and emerging economies. Toh and Low (1996) identify possible imperfections in the labor market in Singapore, due to “latent factors.” If these factors are ignored they find that a researcher estimating factor shares from national accounts data would tend to overestimate the share of capital in the production function. In another study Wu and Thia (20003) estimates TFP but takes explicitly into account distortions caused by the housing market in Singapore. These adjustments have implications for factor shares, the estimate of output and the capital stock. The net effect of these adjustments has similar effect as assuming the low (relative to Young’s) constant factor shares assumed in this chapter. The reported TFP growth by Wu and Thia for 1990-2000 is 1.6 percent, which is the same as reported here. Hsieh (2002 calculates TFP in Singapore using a dual approach that builds on using factor prices to estimate the capital stock, rather than investment data from the national accounts. The result of that exercise is that TFP growth is dramatically higher than Young’s, and somewhat higher than in the results reported here. In sum, the results reviewed suggest that TFP growth in Singapore has not been as dramatically different from the industrial economies as was first suggested by Young (1992).

y t = A t ⁢ K t α ⁢ ( H t ⁢ L t ) 1 − a ( 1 )

where K is physical capital, H is human capital, L is the labor force and A is total factor productivity (TFP). Human capital is measured by the function Ht=(1.07) s where s is years of schooling. This reflects the assumption that one year of additional schooling of the labor force increases aggregate production by 7 percent.3 The parameter α, where 0<α<1, is the capital share and measures the relative importance of capital and labor in production. It is assumed to be 0.35.4 With production function (1) growth can be decomposed into four components due to increases in the labor force, the capital stock, human capital and TFP. Importantly, the production function implies decreasing return to scale for each factor.

5. The decomposition reveals that Singapore’s growth since 1960 has been driven largely by capital accumulation (Table II.1).5 The columns labeled “weighted” in the table shows the contribution of each factor to growth. In the period 1960–70, for example, capital’s contribution to growth was 5.6 percent. It is noteworthy that the contribution of physical capital to growth has been decreasing over time. In the period 1960–1970, the table indicates that about 60 percent of annual growth was attributed to an increase in the capital stock, while in the period 1990–2003, about 40 percent of growth was accounted for by capital accumulation. The contribution of TFP has remained relatively stable, i.e., between 0.9–1.6 percent from 1960–2003. In total, over the past 43 years, capital accumulation has accounted for more than a half of the growth in Singapore and TFP less than one fifth.

Table II.1.

Singapore: Annual Growth

(In percent)

article image

6. Similar results hold for East Asia as a whole (excluding China).6 Roughly half of the GDP growth over the period 1960–2000 can be explained by an increase in the capital stock, and again, the contribution of capital to growth shows a downward trend. As in the case of Singapore, TFP accounts for less than one-fifth of growth. It is instructive to compare the growth experience of East Asia, and Singapore in particular, to the U.S. and the rest of the industrial countries. Table II.3 illustrates that in the U.S. physical capital accumulation has accounted for less than one-third of GDP growth, and that TFP has accounted for about one-fourth. Turning to the industrialized countries, the contribution of capital is about 40 percent (Table II.4).7 As would be expected from the neoclassical theory of growth, the contribution of the capital stock to growth is slightly higher than in the U.S., since this group includes countries that were relatively less economically advanced.

Table II.2.

East Asia Less China: Annual Growth

(In percent)

article image
Table II.3.

United States: Annual Growth

(In percent)

article image
Table II.4.

Industrial Countries: Annual Growth

(In percent)

article image

7. The pattern of capital accumulation and growth in Singapore is consistent with the prediction of the neoclassical growth model. In particular, the large increase in the capital/output ratio, from 0.8 in 1960 to 3.1 today, gives some evidence for decreasing returns to capital. This has important implications for predicting medium-term growth in Singapore. Decreasing returns to capital, coupled with the fact that Singapore’s capital/output ratio has reached those observed in advanced economies, provide some analytical support for the view that growth in coming years may be well below the average observed in the last 40 years, as our projection in Section D indicates. The contribution of capital to growth is likely to continue to decline. The contribution of labor is also likely to decline, as labor force participation rates have moved from relatively low levels in 1960 to levels comparable to those observed in industrialized economies today.

C. Future Sources of Growth

8. This section provides some thoughts on how to boost medium-term growth prospects in Singapore. In the context of the growth accounting framework outlined above, two sources of growth can be identified: further factor accumulation and faster TFP growth. Given the high labor force participation rates in Singapore and the relatively high capital/output ratio, the highest returns are likely to be derived from TFP growth and investment in education.

9. Further investment in human capital may yield substantial gains. Using average years of schooling as a measure of human capital (Barro and Lee, 2000), Singapore is seen to rank below the G7 countries (Table II.5). Based on the growth accounting exercise of the previous section, the return to education is high. For example, if educational achievement were to reach the same level as in the U.S. (12.12 years) Singapore’s GDP would increase by 17.5 percent according to the production function (1).

Table II.5.

Average Years of Education of Population

(15 years and older)

article image

10. Other possible growth opportunities include policy measures to enhance efficiency. One example is the Free Trade Agreements reached with the U.S., Japan, Australia, New Zealand, and others. These agreements are likely to have two impacts in the framework outlined above. Lower tariffs in trading partners increase the profitability of exporting companies for given inputs, leading to higher TFP. While free trade agreement could in principle have negative effects through trade diversion, this is unlikely to arise in Singapore’s case given almost no import tariffs. In addition, further deregulation and structural reforms may also enhance efficiency by increasing the productivity of the existing capital stock and labor force. Measures to increase competition and enhance entrepreneurship would also work in this direction.

D. Estimating Medium-Term Growth Potential

11. This section assesses Singapore’s medium-term growth potential. The production function (1) and the associated growth accounting exercise outlined in Section B provide a useful framework to this end. Based on the assumption that the contribution of capital and labor continues to decline as Singapore moves into the league of highly-developed nations, the growth potential of the economy is estimated at 4–5 percent over the medium term.

12. Three possible growth scenarios, ranging from 4.5 to 7.7 percent, can be considered based on differing TFP growth levels (Table II.6). In these examples it is assumed that capital accumulation will continue to trend downward.8 Similarly, it is assumed that labor input contribution to growth will continue to decline based on demographic projections.9 Education levels are assumed to continue to rise.10 In the baseline scenario TFP growth is assumed to be equal to Singapore’s average level from 1960 to 2003. Case A shows how this result would change if the contribution of TFP were equal to the level in the industrial countries in the period 1960– 2000. Finally, case B shows the level of TFP growth that would be needed to support GDP growth at the 1960–2003 average level of 7.7 percent. This scenario is unlikely, as TFP growth would need to be 4.6 percent, an extremely high level by historical standards.

Table II.6.

Growth Projections: Varying TFP

(In percent)

article image

13 The baseline projection is fairly robust to different assumptions about capital accumulation. Table II.7 shows a sensitivity analysis of the baseline projection based on assuming different rates of investment. In case D the capital stock is projected by assuming that investment, as a fraction of GDP, stays constant at the average investment rate observed over the past 10 years. This results in higher capital accumulation than in the baseline scenario where the future capital stock was projected using a time series model. This scenario gives a reasonable upper bound on the growth projection for capital since a decline in investment is to be expected for an economy that is moving from the status of a developing economy to a highly developed one. Case C shows how the result changes assuming the same rate of investment that was observed in 2003 (which is the lowest rate of investment as a fraction of GDP that has been observed in more that 30 years). This number is a bit below what is to be expected over the medium run, since some part of the decline in investment in 2003 relative to past years is likely to be due to cyclical factors.

Table II.7.

Growth Projections: Varying Capital Growth Rates

(In percent)

article image

14. The baseline projection is relatively more dependent on differing assumptions of factor shares (i.e., the value of α in the production function). Case E considers a case in which the value for the capital share is assumed to be 0.5 in line with Young’s (1992) estimate (see Box II.1). This changes the projected TFP growth (which is assumed to be the same as the historical average of TFP growth in Singapore) and the contribution of each factor to growth. The net result is that the projected medium-term growth rate declines to 3.6 percent. Although this is a percentage point below the baseline projection it does not indicate the growth “collapse” that has sometimes been suggested (see discussion of Krugman (1994) in Box II.1).

Table II.8.

Growth Projections: Varying the Capital Share

(In percent)

article image

E. The Growth Slowdown of the Past Few Years

15. This section assesses the extent to which the growth slowdown of the past few years in Singapore reflects cyclical versus structural factors. Growth in Singapore was slightly negative in 2001 and it has only been modestly positive since. The result from the previous exercise indicates that potential growth is above the recent growth experience. In order to determine how much of the recent slowdown has been due to temporary shocks (e.g., to demand) and how much is due to reduction in long-term growth potential (e.g., due to the neoclassical convergence) the output gap is estimated. The output gap can be estimated using a wide range of approaches, from simple detrending techniques, to more structural approaches, such as the production function approach. However, no single method is generally accepted. Two examples are given here, one based on the production function, and the other based on the Hodrick-Prescott (HP) filter.

16. The production function approach (PF) is appealing for its close link to the growth exercises used in the previous sections. Assuming the production function (1), potential output can be written as:

Y t * = A t * ⁢ K t * α ⁢ ( H t * ⁢ L t * ) 1 − α ( 2 )

where the asterisk denotes potential. It is assumed here that potential refers to the output that can be produced at full employment of the existing physical and human capital stock. Potential output can then deviate from actual output because labor is under utilized (due to temporary shocks) and/or because productivity is below potential (due to temporary shocks). To calculate the series for potential At* and Lt* the HP filter was applied to the series for Lt and the estimated series for At from Section B. This filter extracts short-term fluctuations from long-term trends in these two series. This interpretation of potential output is thus the output that would be produced absent short-term fluctuations in productivity and employment. Figure II.1 shows the output gap calculated by this method (PF), measured as the difference between actual and potential output. Another common estimate of the output gap is to approximate potential output by applying the HP filter directly to the output series.11

Figure II.1.
Figure II.1.

Output Gap

(In percent)

Citation: IMF Staff Country Reports 2004, 103; 10.5089/9781451834208.002.A002

17. The estimated output gap for 2003 is in the range of 2.5–4.5 percent. According to these estimates, a quite large part of the slowdown can be explained by short-term fluctuations, rather than movements in long-term trends. The negative output gap indicates that in the short run—as the output gap is closed—growth may be temporarily above the economy’s medium-term growth potential as the economy recovers.

F. Conclusion

18. Singapore has recorded a remarkable record of economic development and growth over the last four decades. In recent years, however, there has been an abrupt slowdown in the growth rate of the economy partly due to temporary shocks. More broadly, however, an overall slowdown in growth potential can be expected, compared to the high growth rates of the past few decades. A lower growth potential in the coming years is a reflection of Singapore’s success in converging to per capita income levels of the world’s wealthiest economies. The assessment given in this paper suggests that, looking forward, there are returns to be had from investment in education and from structural reforms to enhance TFP.

References

  • Barro, Robert and Johg-Wah Lee (2000), “International Data on Educational Attainment, Updates and Implications,” NBER Working Paper 7911 (Cambridge, Massachusetts), September.

    • Search Google Scholar
    • Export Citation

  • Bosworth, Barry P. and Susan M. Collins (2003), “The Empirics of Growth: An Update,” Brookings Institute, working paper.

  • Gollin, Douglas (2002), “Getting Income Shares Right,” Journal of Political Economy 110(2): pp. 458–74.

  • Hsieh, Chang-Tai (2002), “What Explains the Industrial Revolution in East Asia? Evidence from the Factor Markets,” American Economic Review, June.

    • Search Google Scholar
    • Export Citation

  • Krugman, Paul (1994), “The Myth of the Asian Miracle,” Foreign Affairs, November.

  • Low, A., Ouliaris, S., Robinson, E. and Mei, W. (2004), “Education for Growth: The premium on Education and Work Experience in Singapore,” Staff Paper No. 26, Monetary Authority of Singapore,

    • Search Google Scholar
    • Export Citation

  • Sarel, Michael (1997), “Growth and Productivity in ASEAN Countries,” WP/97/97 (International Monetary Fund, Washington, D.C.).

  • Toh, Mun-Heng and Linda Low (1996), “Differential Total Factor Productivity in the Four Dragons: The Singaporean Case,” The Journal of International Trade & Economic Development, Vol. 5:2, pp. 161–181.

    • Search Google Scholar
    • Export Citation

  • Wu, Friedrich, and Thia Jang Ping (2003), “Total Factor Productivity with Singaporean Characteristics: Adjusting for Impact of Housing Investment and Foreign Workers.”

    • Search Google Scholar
    • Export Citation

  • Wong, Soon Teck and Bentson Sim Song Seng (1997), “Total Factor Productivity Growth in Singapore: Methodology and Trends.”

  • Young, Alwyn (1992), “Tale of Two Cities: Factor Accumulation and Technical Change in Hong Kong and Singapore,” NBER Macroeconomic Annual 1992, pp. 13–54.

    • Search Google Scholar
    • Export Citation
1

Prepared by Gauti Eggertsson (ext. 34918).

2

This section is based on data collected by Bosworth and Collins (2003). For Singapore the data was extended to 2003.

3

This is a fairly conservative estimate for the returns on education. A new study by the Monetary Authority of Singapore (MAS), for example, estimates that an additional year of education increases earning on average by 13.2 percent (see Low, Ouliaris, Robinson, and Mei (2004)).

4

The return on education and a are based on estimates by Bosworth and Collins (2003). The estimate of a is also consistent with Sarel’s (1997) estimate of factor shares in Singapore. Sensitivity of the results to varying the factor shares are discussed in Box II.1.

5

In this exercise the capital stock is estimated by investment data from the national accounts using the perpetual inventory model, the labor force is measured as total hours worked and the measure of human capital is obtained from Barro and Lee (2000). For 1960–2000 Bosworth and Collins (2003) estimate of the capital stock and the labor force is used, and the national account data are used to extent the series to 2003.

6

The countries in this sample are: Indonesia, Korea, Malaysia, the Philippines, Singapore, Taiwan Province of China, and Thailand. The data is obtained from Bosworth and Collins (2003) and each country is equally weighted. China is excluded due to data problems.

7

The countries in this sample are: Australia, Austria, Belgium, Canada, Denmark, Finland France, Germany, Greece, Iceland, Ireland, Italy, Japan, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, and the United States. The data is obtained from Bosworth and Collins (2003) and each country is equally weighted.

8

The assumption on the contribution to growth is obtained from a projection of the capital stock using a simple auto regression model with 3 lags. The number of lags was determined by estimating a model with an arbitrary large number of lags and then eliminating those (recursively) that were not statistically significant. The model was also estimated with a trend, but it was not statistically significant. This number reflects an average over 15 years horizon to reflect the “medium-term.” The model was estimated in logs.

9

Based on World Bank Population projections to 2020.

10

The assumed education’s contribution to growth is equal to its contribution during 1990– 2000 but higher than in the period 1960–1990. This projection reflects that the data exhibits a structural break in 1990.

11

The largest part of the difference between the two estimates is that they are obtained by applying filters to quarterly versus annual data.

Same Series

Other IMF Content

Other Publishers

  • Collapse
  • Expand
Cover IMF Staff Country Reports
Singapore: Selected Issues
Author:
International Monetary Fund
Volume/Issue:
Volume 2004: Issue 103
Publisher:
International Monetary Fund
ISBN:
9781451834208
ISSN:
1934-7685
Pages:
47
DOI:
https://doi.org/10.5089/9781451834208.002
Figure II.1.

Output Gap

(In percent)