Information about Asia and the Pacific Asia y el Pacífico
Chapter

II Accounting for Growth in Singapore

Author(s):
Kenneth Bercuson
Published Date:
December 1995
Share
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details
Author(s)
Rachel van Elkan

Since achieving independence, Singapore has experienced rapid growth. In the period 1960-91, real GDP rose at an average annual rate of 8 1/4 percent, or by 6 1/4 percent a year in per capita terms. In fact, over the same period, growth of per capita GDP was more rapid in Singapore than in any other Asian country with the exception of Korea. Singapore’s performance was even more noteworthy when compared with developing countries in Latin America and Africa. In addition, per capita growth exceeded that of the industrial countries, and of most by a wide margin. For example, since the early 1960s real per capita GDP has grown at average annual rates of 2 1/2 percent in Germany and 1 3/4 percent in the United States.1

The purpose of this section is to attempt to quantify the factors that have been responsible for Singapore’s remarkable performance. In this way, insight can be gained into answering the question of whether comparable growth rates can be sustained in the future. The approach taken here is to use a production function to decompose growth into the contributions of primary factors including land, labor, and capital. Often, however, growth in output cannot be fully explained by increases in these factor inputs. The residual—the part of growth not explained by the factors of production—usually reflects more efficient use of resources or the adoption of new production technologies, in other words, improvements in total factor productivity (TFP).2

The importance of TFP to the growth process is easily demonstrated. In the short term, which in this context may be a period of several decades, high levels of growth can be achieved through rapid accumulation of factors of production. However, in the long run, or steady state, constraints imposed by population growth, together with diminishing returns that set in as capital intensity is increased, form natural limits to growth. The only way to secure growth rates beyond these limits is to secure ongoing increases in TFP.

One of the main conclusions of the study is that, of the 8 1/4 percent annual increase in real GDP that occurred between 1960 and 1991, 6 1/2 percentage points are estimated to have been attributable to factors of production (unadjusted for quality changes) and the remaining 1 3/4 percentage points to TFP growth. The relatively large contribution of factors of production was due to unprecedented capital accumulation, since labor grew only moderately and land reclamation was limited. As a result of the imbalance in factor growth, a marked capital deepening together with a decline in the productivity of capital have occurred—developments that are likely to limit future growth opportunities through this channel. The implication, therefore, is that sustaining high levels of growth will require boosting growth of TFP.

Performance of Output and Productive Factors in Singapore, 1960-92

Output Performance

Since 1960, GDP in Singapore has increased 12-fold. There has been considerable variation, however, in sectoral development, with manufacturing and financial services registering the most rapid growth. From 1960 to 1992, real value added rose 21-fold in manufacturing and 23-fold in business and financial services. As a consequence, the share of business and financial services in real GDP rose from 14 percent in 1960 to 26 percent in 1992 while, over the same period, the share of manufacturing in GDP increased from 17 percent to 28 percent (Chart 2-1). Further, within manufacturing, there has been an equally dramatic shift of resources, with, for example, the share of electronics and electrical equipment rising from 11 percent of the total in 1970 to 40 percent in 1992. The increasing importance of electronics is a reflection of the underlying change in technological sophistication of Singaporean industry that occurred during this period.

Chart 2-1Structure of GDP

(At 1985 market prices)

Source: Singapore, Ministry of Trade and Industry. Economic Survey of Singapore.

Physical Capital

As mentioned previously, factors of production made the greatest contribution to growth in Singapore and, of these factors, capital accumulation was by far the most important. The scale of investment has been impressive. Since the early 1970s, the ratio of gross fixed investment to GDP has exceeded, on average, 35 percent and has been the highest in the world, far exceeding, for example, that of Japan and the other dynamic Asian economies (Chart 2-2). On the basis of cumulative net investment data, Singapore’s capital stock has increased 33-fold since 1960, doubling on average once every six years. This exceptionally rapid pace of investment resulted in a tenfold increase in the capital-labor ratio between 1960 and 1992.

Chart 2-2Gross Fixed Capital Formation as a Share of GDP1

(In percent)

Source: IMF, international Financial Statistics.

1 In current prices.

The sharp expansion in capital accumulation was accounted for by the private sector, as public sector investment fell in relation to total investment from 29 percent in 1960 to 19 percent in 1992. Still, at 8 percent of GDP in 1992, public sector investment—which is concentrated in residential construction and infrastructure—was high by international standards. As for private sector investment spending, there has been a marked change in its composition. Most notably, in relation to total investment, residential construction declined from 23 percent in 1960 to 11 percent in 1992 while, over the same period, nonresidential construction rose from 6 percent to 19 percent of the total. Expenditure on machinery and equipment has also increased in importance, rising from 29 percent of total investment in 1960 to 35 percent in 1992.

Foreign direct investment has played a major role in Singapore’s development, particularly in the manufacturing and financial services sectors. Between 1980 and 1991, foreign investment represented 25 percent of fixed capital formation and over 60 percent of investment in manufacturing. Within manufacturing, foreign investment was most pronounced in the electronics subsector, accounting for well over 90 percent of total investment.

The contribution of capital accumulation to output growth is due both to increases in the size of the capital stock and to improvements in the quality of capital goods as reflected in their technological sophistication. As to the latter, Singapore has benefited from relatively advanced foreign technologies, both via the importation of capital equipment that embodies such technologies (imports of capital goods have accounted for about a quarter of total imports since 1960) and directly through the adoption of production techniques developed abroad by parent companies. In addition, the technological sophistication of foreign direct investment, which may be proxied by per capita income levels in the source countries, has increased over time, providing a basis for the continual upgrading of Singapore’s capital stock.3 The tabulation below provides an index (1980 = 100) of the technology level of Singapore’s FDI inflows from 1980 to 1989.

1980100.01985206.2
1981119.21986228.3
1982143.91987252.1
1983160.11988285.0
1984178.71989318.0

This index measures the volume of real FDI inflows in any year weighted by relative per capita GDP of the source country. It indicates that the technology level of Singapore’s FDI inflows indeed has increased monotonically during this period.

Labor

Labor’s contribution to growth has been much smaller than that of capital: between 1960 and 1992, employment only trebled. Nevertheless, this increase exceeded the growth of the population as tight labor market conditions throughout the period led to both rising participation rates and an influx of foreign workers.

Between 1960 and 1992, the average productivity of labor, measured by real output per worker, increased by some 300 percent, for an average annual growth in labor productivity of 4 1/2 percent. This rapid growth has been due to capital deepening in the economy as well as improvements in the quality of the capital stock and the labor force. Significant improvements in the quality of the work force, as measured by educational attainment, have taken place since 1970, partly in response to the increasing scarcity of labor. Table 2-1, which sets out a cross-country comparison of the educational attainment of the labor force, indicates a sharp increase in the proportion of Singaporean workers who have received at least a secondary education, from 17 percent in 1970 to 52 percent in 1991. In addition, the share of the Singaporean labor force who have received some form of tertiary education has risen from 4 percent in 1970 to 22 percent in 1991. Nevertheless, by comparison with the industrial countries and developing countries within the East Asian region, the average level of education of the Singaporean work force remains quite low.

Table 2-1Intercountry Comparison of Educational Attainment of the Work Force

(Percent of country’s total work force)1

UniversityPost-SecondarySecondaryPrimary and Below
Singapore
1970421383
1980392860
19917153047
Hong Kong, 19906342337
Japan, 19891961203
Korea, Republic of, 198913372030
Malaysia, 1989524847
Taiwan Province of China, 19876342040
United Kingdom, 199010341930
United States, 198926204014
Source: Singapore. Ministry of Trade and Industry, Economic Survey of Singapore. 1992.

Except for the United States, data refer to employed persons aged 15 years and over; U.S. data refer to employed persons aged 25 to 64 years.

includes both university and post-secondary.

includes secondary, primary, and below.

Source: Singapore. Ministry of Trade and Industry, Economic Survey of Singapore. 1992.

Except for the United States, data refer to employed persons aged 15 years and over; U.S. data refer to employed persons aged 25 to 64 years.

includes both university and post-secondary.

includes secondary, primary, and below.

Land

Land, by far, has been the slowest-growing factor. Since 1960, land available for agricultural, commercial, industrial, and residential purposes has risen by some 5 percent, an amount that reflects both reclamation and release of state-owned land for development. As would be expected with rapid development, land used for agricultural purposes has declined markedly since 1960, while nonagrarian land usage has nearly doubled.

Description of Methodology

This study employs a growth-accounting methodology to decompose output growth into contributions attributable to factor accumulation on the one hand and TFP growth on the other. In this methodology, a weighting scheme is developed that allows the contribution of land, labor, and capital to be aggregated. As mentioned previously, the difference between output growth and the portion accounted for by the factors of production is a measure of TFP growth.

Total factor productivity growth is a residual, but the question is, what does it measure? The basic answer is that it measures the contribution to output growth of the improvements in the efficiency and technology with which resources are employed. The level of TFP can be raised by adopting more advanced production techniques and by reducing distortions in the economy. For example, lowering tariff barriers, allowing flexible wage determination, and phasing out selective interventions that target certain sectors will lead to a reallocation of resources to more productive activities.4 Dynamic gains in efficiency, generating a permanent increase in TFP growth, may result from the elimination of (or reduction in the level of) a distortion if that distortion discouraged investment in a high-productivity (potential) sector. In addition, TFP includes among other things the effects, if any, of scale economies (or diseconomies) in the aggregate, and cyclical and stochastic factors that induce a greater response in output than in factor inputs.

Note that the estimate of TFP that will be derived should be interpreted with caution, since the methodology used here does not adjust factor inputs for quality changes. For example, a unit of capital input is assumed to be the same in 1990 as it was in 1960. The implication is that the incremental effect on growth of embodied technological advancement is not attributed to capital but rather is measured as a higher level of TFP. The same measurement problem can also arise in the case of labor. As education and on-the-job training act to improve the quality of labor, measured TFP will be enhanced. This “mis-measurement” of TFP may well be significant in the case of Singapore, a country that has experienced rapid improvements in embodied technology in the past three decades.

The methodology used here, which follows the pioneering work of Gollop and Jorgenson (1980) and Jorgenson, Gollop, and Fraumeni (1987), assumes a translogarithmic form of the aggregate value-added production function (Y), using inputs capital (K), labor (N), land (L), and time (t):5, 6

The assumption of constant return to scale implies that the parameters satisfy the following restrictions:

and

In equilibrium, the share of the value of output that a producer is willing to pay a factor of production is equal to the elasticity of output with respect to that input. Hence,

where θi is equal to the value share of factor i in total output, and where Py and Pi denote, respectively, the current price of output and input i. Thus, both the elasticities and income shares sum to unity.

First differencing the natural logarithm of the production function in equation (1) gives an expression for the growth of value added in terms of the growth of the individual factor inputs between period τ—1 and τ.

where

and where θi denotes the average income share of factor i in total factor payments. The translog index of TFP growth between periods τ—1 and τ (denoted TFPτ-1,τ) provides a measure of the amount by which (the log of) output would have increased had all inputs remained constant between the two time periods.

Measuring inputs, Factor Shares, and Results

Inputs and Factor Shares

The estimate of output used in this study is real GDP (measured in 1985 prices) adjusted for indirect taxes.7 As to factor inputs, labor services are measured by annual person-hours, derived from aggregate employment and average weekly hours worked. The real capital stock series was computed from cumulative net investment data adjusted by the gross investment deflator. The initial value of the capital stock (in 1960) used to generate the series was taken from estimates of the real capital stock provided by the Monetary Authority of Singapore (MAS). The calculations are based on the perpetual inventory method, assuming an annual depreciation rate of 5 percent.8 The flow of capital services to current production is proxied by the available stock since, once capital has been produced, it has no alternative use and, therefore, it yields its services inelastic ally. For this reason, no adjustment to capital inputs is made for changes in the intensity of capital utilization over the phase of the business cycle. The third primary input, land’s services, is assumed to arise from the area of agricultural land and built-up land.

As to factor shares, data on labor’s share in GDP at factor cost was provided by the MAS. The income allocated to land and capital is defined to be the difference between total factor payments and labor’s share. Since information on land rents is not available, the nonlabor share of income was distributed between land and capital by assuming that net rates of return to capital and land are equal, and that the ratio of the value of the capital stock to the value of land is the same in Singapore as in the United States.9

The assumption that income shares reflect output elasticities is likely to be violated for capital goods, especially in the case of foreign direct investment, where monopoly suppliers of capital goods may earn a return in excess of the marginal product of capital. Accordingly, in what follows, income accruing to capital is reduced by the amount of the monopoly rent, assumed to be equivalent to 1 percent of the current value of the capital stock (see MAS (1993)). This amount is then redistributed to the other factors of production in proportion to their initial income shares.

Results

Annual growth rates for output and for inputs of capital, labor, and land for 1961-91 are set out in Table 2-2. The income shares of capital and labor in nominal value added are also shown. The value share of land is equal to one minus the shares accruing to the other two factors. The contribution of factor inputs to growth in output, measured by factor growth rates weighted by their income shares, is also presented in Table 2-2, as is the rate of growth of total factor productivity. Averages for the entire period (1961-91) appear at bottom.

Table 2-2Contributions to Growth in Aggregate Output1
Growth of:Average Share of Capital InputAverage Share of Labor InputContribution to Growth in Value Added of:
OutputCapitalLaborLand(In percent)(In percent)CaptialLaborLandTotal factor productivity growth
19610.0830.0650.02941.248.10.0270.0140.042
19620.0690.0790.0290.03841.448.10.0330.0140.0040.018
19630.1040.0970.02841.448.20.0400.0130.050
1964–0.0510.1110.0270.00441.348.30.0460.013–0.110
19650.0570.1110.026–0.02641.248.40.0460.013–0.0030.001
19660.1000.1000.0260.00541.248.50.0410.0120.045
19670.1260.1020.026–0.01241.348.50.0420.012–0.0010.072
19680.1350.1300.0250.00641.348.50.0540.0120.0010.069
19690.1230.1490.024–0.00441.348.50.0620.0120.050
19700.1240.1740.0240.05241.248.60.0720.0120.0050.036
19710.1210.1810.0690.00341.048.80.0740.0330.012
19720.1240.1750.069–0.00540.949.00.0720.034–0.0010.020
19730.1090.1550.066–0.01540.949.10.0630.033–0.0020.015
19740.0770.1480.0170.04240.549.50.0600.0090.0040.004
19750.0390.1190.0130.00640.150.20.0480.0070.001–0.016
19760.0710.1100.0430.03839.750.90.0440.0220.0040.002
19770.0730.0970.0420.05639.251.60.0380.0220.0050.007
19780.0800.1010.06639.051.90.0390.0340.006
19790.0850.1040.061–0.02638.652.40.0400.032–0.0020.016
19800.0960.1160.056–0.00437.953.20.0440.0300.023
19810.0900.1200.0620.02137.054.10.0440.0340.0020.010
19820.0640.1300.053–0.01636.055.30.0470.029–0.001–0.010
19830.0720.1260.029–0.00434.956.90.0440.0160.012
19840.0740.1200.014–0.00633.858.20.0410.0080.025
1985–0.0140.085–0.050–0.02333.358.70.028–0.029–0.002–0.011
19860.0380.061–0.005–0.01733.158.90.020–0.003–0.0010.022
19870.0890.0580.051–0.01333.059.00.0190.030–0.0010.041
19880.1060.0600.052–0.03333.258.70.0200.031–0.0030.058
19890.0770.0680.068–0.01633.558.20.0230.039–0.0010.016
19900.0780.0750.0950.00933.758.00.0250.0550.001–0.003
19910.0650.083–0.004–0.01033.758.00.028–0.002–0.0010.041
Average0.0800.1100.0360.00238.252.40.0430.0190.018

‘Growth in variable x is defined by dln(x). where ln denotes the natural logarithm.

‘Growth in variable x is defined by dln(x). where ln denotes the natural logarithm.

Table 2-2 shows that aggregate real output increased in all but two years—1964 and 1985—during 1961-91. Output growth during the period averaged 8.0 percent a year. Growth in the capital stock was positive throughout the period. Growth of the labor input was positive in all periods except 1985, 1986, and 1991. Available land area fluctuated throughout but, on average, remained unchanged. The data also reveal that the income share of capital declined over the period, while that of labor increased. As mentioned previously, the declining share of capital may reflect the capital deepening that took place during this period. For the period as a whole, the average factor share accruing to labor was 38.2 percent, while the average factor share accruing to capital was 52.4 percent.

Capital generated a positive contribution to output growth in each year of the period, accounting on average for 4.3 percentage points, or more than half, of GDP growth. Capital’s contribution was greatest from the late 1960s through the early 1970s—coinciding with the period of fastest output growth. During this period, capital contributed about 7 percentage points to the average annual growth rate. Labor’s contribution to growth exhibited considerable variability, ranging from minus 3 percentage points to nearly 6 percentage points. On average, however, growth in the supply of labor added 1.9 percentage points to annual growth in output. The effect of increased land availability on GDP growth was insignificant.

Total factor productivity—which here reflects both the level of efficiency with which factors are used as well as the quality of the factors themselves—contributed 1.8 percentage points, or more than one fifth, to Singapore’s average annual growth rate during the period.10 Productivity growth displayed much greater variability than growth of factor inputs, being negative for 5 of the 31 years and ranging from minus 11 percentage points to more than (plus) 7 percentage points. In 1964 and 1985, corresponding to the years of real GDP decline, total factor productivity growth was negative. Indeed, in 1964, with the capital stock and employment increasing, the decline in total factor productivity was larger than the decline in output.

Capital was the largest contributor to GDP growth in 20 of the 31 years and was consistently the most important factor in growth during 1969-85. Labor made the greatest contribution in 1989 and 1990. Growth in total factor productivity accounted for the greatest share of output growth in 9 of the 31 years. Abstracting from cyclical factors, total factor productivity growth was particularly important during 1967-74 and again from 1986 onward. The contribution of TFP growth through time can be seen in Chart 2-3, which shows a five-year backward moving average of TFP growth. Following a peak of 5.7 percentage points in 1970, the average growth rate of total factor productivity declined until 1978, when it reached a minimum of 0.1 percentage point per year. Since 1980, however, the rate of TFP growth has increased and has shown significant strengthening, especially since 1986.

Chart 2-3Total Factor Productivity Growth

(Translogarithmic index)

Source: IMF staff estimates.

Conclusion

This section has used a simple growth-accounting methodology to analyze the main determinants of Singapore’s growth performance over the period 1960-91. This methodology enables one to decompose output growth into two main sources: increases in productive factors (land, labor, and capital) and increases in total factor productivity.

Although factor accumulation has been dominant in accounting for Singapore’s superior growth performance, the role of TFP over the past three decades should not be minimized. In fact, total factor productivity as measured here (which includes improvements in both the quality of productive factors and economic efficiency) raised the level of per capita GDP in Singapore by some 76 percentage points between 1960 and 1991. Furthermore, despite substantial capital deepening, relatively high growth has persisted because of the increasing contribution of TFP.

This study suggests that sustaining the high rates of growth Singapore has traditionally experienced will require that TFP play an increasing role. However, as Singapore approaches the world technology frontier, thereby exhausting its opportunities for technology catch-up and its potential for further TFP growth from this source, alternative channels for improving TFP will need to be sought. Public policy will need to play a role in this endeavor. One area may be education. While there has been heavy investment in human capital in Singapore through formal education and on-the-job training (the benefits of which are likely to be experienced, at least in part, in future years), a significant gap in educational attainment remains to be closed between Singapore and the industrial world. In addition, experience in a number of other countries suggests that reducing distortions in the economy—which may arise because tax, trade, or credit policies alter relative prices faced by producers and consumers in comparison with those that would prevail in a free market—may serve to increase a country’s TFP growth rate permanently by generating dynamic gains in efficiency.

All GDP data reported in this chapter are from country sources, as reported in IMF, International Financial Statistics Yearbook. If Summers and Heston (1988) data are used, the rankings are slightly different but the overall picture is unchanged. As reported by the World Bank (1993), Singapore’s 1960–85 GDP growth (based on Summers-Heston data) ranks fifth in the world, behind only Botswana, Taiwan Province of China, Indonesia, and Hong Kong.

The residual is defined formally as the growth in output that occurs with unchanged levels of the factor inputs. As such, the interpretation of this residual depends on the definitions of factor inputs employed in the analysis. Factors of production possess numerous dimensions or characteristics that affect their usefulness in production. Labor’s productivity is affected, among other things, by educational attainment, work experience, and gender. The productivity of capital is affected, among other things, by the age of the equipment, the level of technology embodied within it, and whether the capital good is publicly or privately owned.

The index of the technology level (T) of FDI is defined to be a weighted average of real FDI inflows (Ii), where i denotes the source country, and where the weights (γi) are defined to be the ratio of per capita GDP in country i to the sum of per capita GDPs in all countries contributing FDI to Singapore:

Thus, for a given volume of FDI inflows, improvements in the technology level occur when the source of FDI shifts from a low to a high per capita GDP country. This measure assumes, however, that differences in the technology of FDI across source countries reflect differences in relative per capita GDP.

In the absence of policy distortions, the reallocation of resources will have no effect on aggregate output, since each factor, which receives an identical return in each activity, will generate at the margin the same contribution to value added in each activity.

Time is the proxy for the level of total factor productivity.

Young (1992) notes that, in effect, the translog production function may be interpreted as a second-order approximation to any given production function.

The data used here are from IMF, International Financial Statistics Yearbook, from official Singapore publications, and from the on-line Singapore Department of Statistics PATS service.

The series generated under these assumptions was very similar to the MAS series of the real capital stock.

These are the assumptions adopted by the MAS.

Since TFP growth is calculated as the residual in the growth accounting methodology, it will also reflect errors in the measurement of factor inputs and factor shares. There is no reason to believe, however, that the net effect of errors in measurement will bias calculated TFP growth.

    Other Resources Citing This Publication