Information about Asia and the Pacific Asia y el Pacífico
Chapter

Chapter 2. How Fast Can Cambodia Grow? Assessing Key Drivers

Author(s):
Olaf Unteroberdoerster
Published Date:
February 2014
Share
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details
Author(s)
Phurichai RungcharoenkitkulThe author was an economist in the IMF Asia and Pacific Department when the initial draft of this book chapter was prepared. He would like to thank Enrique Aldaz-Carroll, Roberto Cardarelli, and Olaf Unteroberdoerster for their insightful comments. He is also grateful to Jaromir Benes for his helpful advice on the IRIS open-source toolbox for macroeconomic modeling and forecasting. This chapter was previously published as IMF Working Paper No. 12/96 (Rungcharoenkitkul, 2012).

Accurate estimates of potential growth have always been in high demand, in both advanced and developing economies. This is hardly surprising, since long-term growth is not only the core objective of development policy but also a key anchor of economic stability. However, applying standard quantitative methods to estimate potential output in developing economies is subject to severe limitations. In particular, poor data coverage and a lack of high-frequency economic indicators make the use of standard filtering techniques more difficult. Moreover, because developing economies are often in the process of important structural changes, any model that relies on stable macroeconomic/behavioral relationships can be a misleading guide to the future. Yet it is far from clear that a judgment-based estimate will necessarily be a better alternative.

This chapter’s primary objective is to introduce a tractable quantitative framework to assess potential growth in a developing economy. The framework is based on a small dynamic macroeconomic model that, by design, economizes on both the data and behavioral assumption requirements and also allows the incorporation of external assumptions and judgment calls. We will show that the model can be used alongside a more qualitative “growth diagnostic” approach (introduced by Hausmann, Rodrik, and Velasco, 2005).

Cambodia’s current macroeconomic backdrop makes it an especially interesting application of the proposed framework. The last decade was, for the most part, a fast-growing period for the country, with its economy expanding by 9.8 percent on average over 1999–2007, doubling its standard of living in the process. The impressive growth performance resulted from a confluence of favorable factors, including the expansion of international trade, sustained investment, and improvements in productivity. However, Cambodia is at an important crossroads. Its productivity growth is constrained by a lack of adequate infrastructure, notably in the electricity and power sectors, and this lack in turn is holding back investment. Overreliance on a narrow export market, notably the low-skilled garment industry, also means the external engine of its growth is approaching its limit, unless the export sector moves up the value chain or diversifies into other products. Cambodia’s growth potential will hinge on whether the country can successfully overcome these challenges in the coming years.

To assess the outlook for potential growth in Cambodia, this chapter proposes (i) a quantitative framework that accounts for the dynamics of the fundamental drivers of growth and (ii) an analytical diagnostic of the constraints to growth. The two parts of this strategy complement each other, with the model informing the analysis of growth impediments and at the same time serving as a tool to assess different scenarios informed by the qualitative analysis.

The model proposed here is an integration of several standard methods, which may be grouped into two main categories: (i) the production function approach and (ii) the state space or statistical filtering approach. The first approach assumes an aggregate production function and estimates each of its components (factor inputs and total factor productivity). The approach works best in more advanced economies, where there are more data on factor inputs (such as capital stock, labor participation, hours worked, and utilization rates), as well as good proxies for total factor productivity. For example, the U.S. Congressional Budget Office and the EU Commission use this method to estimate the potential growth in the United States and the European Union, respectively (see CBO, 2001; and Roeger, 2006).1

One drawback of the production function method, given its bottom-up philosophy, is that it does not exploit other macroeconomic predictions that can be useful for inferring potential output. In particular, macroeconomic theory suggests that, if shocks are transitory, potential output should be close to the smoother trend of realized output. Moreover, according to the Phillips curve, potential output should be relatively low compared with actual output when inflation is accelerating.

The second category of estimates of potential growth attempts to exploit these macroeconomic relationships. Statistical filters that produce smoothed output series (such as the Hodrick-Prescott (HP) and bandpass filters) are the simplest example. A more structural example considers potential output as an unobserved variable within a structural macroeconomic model. An estimation procedure can then be performed to extract the sequence of potential output that is most consistent with the predicted relationship. For an example of this state space/filtering approach, see Benes and others (2010).

Both approaches have advantages and drawbacks. The first method offers an appealing identification of the drivers of potential growth, which is useful for the analysis of growth impediments. However, it requires detailed, granular data, a condition rarely met for developing countries. The second method can economize on the amount of data required by using restrictions from economic theory. Traditionally, the filtering method is most often used to explain the cyclical fluctuations of output around the trend (that is, to assess the output gap), but it does not allow a structural explanation of the sources of long-run growth.

This chapter develops a hybrid model that combines the strengths of both traditional methods. A small-scale structural macro model is proposed; it has a supply-side production function and thus is capable of explaining the dynamics of growth drivers. The model has a state-space structure, where unobserved variables are estimated using a Kalman filter. This eases the burden on data requirements, since not all variables need to be observed. Finally, the model includes the demand-side equations, namely for the processes governing the output gap and the Phillips curve, enabling it to utilize information from macroeconomic theory as in the standard filtering method.

The chapter is organized as follows. First, the small structural macro model is laid out and estimated for Cambodia to project potential growth over the next decade. This serves as a platform to conduct quantitative growth simulations under different scenarios. We then proceed to identify factors that may be hindering growth by examining a number of criteria according to which Cambodia may have fallen behind other lower-income peers. This growth diagnostic is then used in combination with the growth model developed in the first part of the chapter to provide a quantitative assessment of the growth dividends should the identified growth impediments be removed.

Estimating Potential Growth From a Small Macro Model

The Model

The potential output Y¯t is modeled as a simple Cobb-Douglas aggregate production function with a Hicks-neutral productivity term:

where At is the productivity level, Kt is the physical capital stock, Lt is the labor input, and 0 < α < 1. The potential output is therefore determined by the supply side of the economy, with available factor inputs and technology dictating how much output the economy can potentially produce.

The capital stock evolves as a function of an exogenous saving rate, St:

where δ is the depreciation rate, and Yt is the level of actual output. The saving rate is fixed at its realized historical values over the estimation sample, while its future values are pinned by an assumption as part of a forecasting scenario. Since the modeled economy is open, the saving rate St is identical to the rate of total investment, both domestic and foreign. Labor input is determined by the demographics, growing exponentially at a constant rate l:

with a white-noise shock εtL.

The productivity term At is an aggregate of three subcomponents:

where AGt, AMt and AOt represent productivity in the agricultural, manufacturing, and other sectors, respectively. The productivity in each sector is subject to disturbances, but is expected to grow exponentially at a constant rate:

Demand shocks can cause the actual output to deviate from its potential level. For instance, capital and labor may be underutilized during recessions, causing actual output to fall below its potential level. These exogenous shocks are assumed not to have a permanent effect on output, and thus the output gap—gaptlog(Yt/Y¯t)—narrows as the effect of shocks disappears. The demand-side equation can be summarized in a reduced form as

with εtY being a white-noise demand shock.

Inflation dynamics depend on the output gap and are therefore conditional on actual output, revealing information about the potential output. Inflation πt is assumed to follow a canonical Phillips curve with a white-noise disturbance:

In other words, inflation rises with the output gap, moves with inertia, and is subject to supply shocks such as changes in energy and international food prices. In the long run, inflation converges to a constant π¯.

Potential Growth Estimate

The solution to equations (1) through (9) is characterized by a balanced growth path, with the potential output driven by productivity growth and factor accumulation. The model has a state space structure, where a subset of variables is observed, while others can be estimated from Kalman filtering. The observed variables are annual output Yt, inflation πt, labor input Lt, investment rate St, output gap gapt, agricultural productivity AGt, and manufacturing productivity AMt.2 The full estimation period is 1986–2011, and any missing data points are treated as unobserved, to be recovered by the filter algorithm. Appendix Table A2.1 provides details of the data sources and available dates for each series.

The model is log-linearized, solved, and estimated by the Bayesian method.3 In almost all cases, only loose priors about the parameter values are imposed to allow data ample room to speak, by setting large prior standard deviations and wide minimum-maximum bounds. Only in the case of a is the prior maximum value binding after the estimation. Detailed information on the parameter priors and estimation results are given in Appendix Table A2.2.

The estimated model is used to forecast all endogenous variables up to the year 2020. The baseline projection (Figure 2.1) assumes that Cambodia will be able to maintain investment at 20 percent of output in each year (similar to the 2000–10 average of 19.6 percent). The estimate for labor growth is 3 percent, whereas the estimate for productivity growth is 3 percent in the agricultural sector, 15 percent in the manufacturing sector, and 3 percent in other sectors. These growth assumptions are all close to their historical values.

Figure 2.1Potential Growth Estimate, Cambodia, 1989–2020

(Percent)

Sources: Data provided by the Cambodian authorities; and IMF staff estimates and projections.

Findings. The baseline estimate shows that potential growth has been declining gradually from the peak of 8.7 percent in 2004–05 but is expected to stabilize at around 7.5 percent from 2012 onward without further shocks, provided the investment rate remains at 20 percent (Figure 2.1). The estimated output gap is currently in negative territory, but it should gradually close as actual output growth catches up with potential growth.

The breakdown of potential growth into contributions from factor accumulation and productivity gain is shown in Figure 2.2. Growth in productivity contributes about 3 percent to potential growth on average and has been a relatively stable source of growth for Cambodia. Factor accumulation contributes about 4 percent to potential growth on average, with roughly equal contributions from capital and labor accumulation over 1987–2011. However, during 2000–07 a rapid accumulation of capital stock contributed about 3 percent to growth and was a key driver of potential growth during this expansionary period. As factor accumulation slows going forward, growth is projected to moderate from the previous decade, supported for the most part by a continued gain in productivity, followed by sustained capital accumulation and labor growth.

Figure 2.2Contributions to Past and Potential Growth, Cambodia, 1987–2020

(Percent)

Sources: Data provided by the Cambodian authorities; and IMF staff estimates and projections.

Adverse Growth Scenarios

Clearly, if Cambodia fails to sustain its investment or productivity improvement, potential growth can be adversely affected. But by how much? Figure 2.3 compares the baseline potential growth with potential growth under two scenarios: (i) if the productivity in each sector grows at only half the rate as in the baseline, and (ii) if the investment-to-output ratio St is halved, relative to the baseline, to 10 percent.

Figure 2.3Potential Growth Scenarios, Cambodia, 1987–2020

(Percent)

Sources: Data provided by the Cambodian authorities; and IMF staff estimates and projections.

Findings. A 50 percent reduction in productivity growth immediately reduces potential growth by over 1.5 percentage points and lowers potential growth permanently by 1.7 percentage points. On the other hand, a fall in the investment rate by half reduces potential growth by nearly 2 percentage points before sustained productivity gains (assumed to remain unchanged in this scenario) begin to lift potential growth. The boost from productivity only raises potential growth gradually, however, and the growth rate after 10 years still falls short of the baseline by more than 1 percentage point. In both cases, there will be substantial gaps between the levels of potential output compared with the baseline, and the gaps are still diverging after 10 years.

The possibility of highly nonlinear dynamics not captured by the model must be considered when using the model to analyze adverse scenarios. For a small open economy such as Cambodia’s, there can be significant positive feedback between investment and productivity in ways not captured by the model. Higher productivity lowers production costs and raises profit margins, which helps attract foreign direct investment and accelerates capital accumulation. On the other hand, foreign investment brings technical know-how and is an important boost to productivity. Therefore, falling short of sustaining either investment or productivity may risk starting a vicious cycle of anemic growth.

What policy lessons can be drawn from the exercise so far? One clear policy priority is to continue to improve the quality of the factors of production by investing in education and labor skills. Another priority is to speed up the diffusion of technology to improve industrial efficiency. Third, investment in infrastructure to lower energy costs will also provide a significant boost to productivity, especially for the manufacturing sector.

The next step is to promote sectors that have greater room to benefit from productivity growth. In the manufacturing sector, this means moving up the value chain of the dominant garment industry or diversifying into other areas, such as electronics. The low-skilled garment industry has less room for productivity improvement, and its contribution to potential growth may inevitably decline in the longer term.4

Growth Diagnostic via Cross-Country Comparison

What are the major growth-hindering factors in Cambodia? What can be done to alleviate these impediments and what is the likely effect on growth potential? This section aims to identify the constraints to growth drivers using a growth diagnostic exercise that relies on a cross-country comparison.5 Impediments to growth drivers, namely investment and productivity, are then discussed in turn.

Identifying Impediments to Growth

Although investment has contributed significantly to Cambodian growth over the past decade, as a percentage of GDP investment in Cambodia remains among one of the lowest among low-income economies (Figure 2.4). A simple cross-country regression over 2000–09 shows that a 10 percent higher investment rate is associated with a 1.3 percent higher GDP growth rate.6 Meanwhile, a simulation based on the above macro model suggests that a 10 percent increase in investment can raise potential growth immediately by as much as 2 percent and that the positive growth effect can persist even after a decade (see below on Quantifying Growth Dividends). Investment therefore holds significant untapped potential as a key driver of future growth for Cambodia.

Figure 2.4Average Investment-to-GDP, Selected Asian Economies, 2000–09

(Percent)

Sources: IMF World Economic Outlook database.

However, a combination of factors is currently holding investment back. Mitigating these constraints will be an important first step toward unleashing investment. According to the Doing Business 2012 report by the World Bank and the International Finance Corporation, investors have found it harder to start a business, obtain a construction permit, and have contracts enforced in Cambodia than in many other countries. As Figure 2.5 illustrates, Cambodia ranks in 133rd place in overall ease of doing business, trailing most of its lower-income peers, and also scores poorly in the perception of corruption, according to Transparency International (see Chapter 3). Such rent seeking can deter private investment, cause a misallocation of resources that compromises investment efficiency, and ultimately hurt growth.

Figure 2.5Hindrance to Investment, Selected Asian Economies, 2012

(International rankings)

Sources: World Bank, Doing Business; and Transparency International, Corruption Perceptions Index.

A lack of adequate infrastructure is also restraining investment, as well as hampering total factor productivity. As Figure 2.6 shows, Cambodia compares less favorably against its peers in basic indicators such as available telephone lines and Internet users. Meanwhile, prohibitively expensive power has been a binding obstacle for Cambodia, constraining power consumption per capita, which is currently low relative to other developing countries. Newly built hydropower plants will likely help ease some constraints on investment and growth, but they should be supplemented by wider measures to improve other basic infrastructure. There is robust evidence that factors such as the ease of doing business and basic infrastructure are important determinants of private investment in Asia (IMF, 2010).

Figure 2.6Infrastructure, Selected Asian Economies, 2008/09

Source: World Bank, World Development Indicators.

Although productivity gains have been a key driver of Cambodia’s growth, not all sectors have witnessed a surge in productivity. Growth in the agricultural sector, for example, has been helped by an expansion in total harvested area, whereas crop yields remain similar to those in the less productive competitors in the international market (Figure 2.7). As Cambodia’s crop area inevitably stops expanding, and with its export market power still limited, its agricultural sector will need to rely on improvements in yield to continue growing and to increase global market competitiveness. Improving yield through greater use of physical capital such as machinery is one option, and it is another example of the complementary effects of investment and productivity.

Figure 2.7Cereal Yields, Selected Asian Economies, 2012

Source: World Bank, World Development Indicators.

Another important source of growth is labor productivity. The primary school enrollment rate was close to 90 percent in 2007. However, secondary school enrollment was much lower, at 34 percent, while the tertiary enrollment rate was only 5 percent. As the economy climbs up the technology ladder, the shortage of qualified labor will become an even more binding growth constraint. Therefore, an urgent need exists to speed up investment in human resources. However, public expenditure on education as a share of GDP in Cambodia is among the lowest compared to peers even after controlling for country-specific differences in level of development and educational attainment (Figure 2.8).

Figure 2.8Deviations from Predicted Public Spending on Education, Selected Asian Economies, 2011

(In percent of GDP)

Source: IMF staff estimates.

Although the manufacturing and export sectors have probably been major beneficiaries of productivity gains in the past, their exclusive reliance on a garment-led growth model may be more challenging going forward. While Cambodia has made efforts to diversify its export markets, the diversification in the product (as opposed to market) space is limited. As Figure 2.9 illustrates, the Herfindahl index of export diversification (in which a lower number indicates more diversification) has been higher for Cambodia than for many other Asian exporters. Countries such as China and Thailand, which have enjoyed a long period of sustained export-led growth, are highly diversified in the products that they export. Vietnam has also been diversifying. Only Bangladesh and Cambodia seem to have stalled in the progress toward greater diversification. The recent trend points toward less diversification in Cambodia, indicating that the country is following a different development path from a typical developing country.7

Figure 2.9Herfindahl Index of Export Diversification, Selected Asian Economies, 1980–2006

Source: World Bank, Export Diversification Data, PRMED.

Conceptually, there are benefits and costs to both diversification and specialization, but in the case of Cambodia a number of factors currently point toward increasing net benefits from export diversification. First, the uncertain global economic backdrop highlights the need for diversifying risks across both markets and products. Second, following years of rapid growth, Cambodia is now better positioned to start investing more in a wider range of sectors and promoting greater extensive margin trade.8 Third, export diversification is an essential part of export discoveries, which in turn can lead to a renewed wave of productivity growth much needed by Cambodia. Finally, Cambodia’s low labor costs and stable environment have started to attract foreign investors beyond the dominant garment sector, providing an opportunity to diversify its export industry into other areas of light manufacturing, such as car parts and electronics.

Pursuing greater export diversification will depend on the extent to which Cambodia succeeds in sustaining investment. Acemoglu and Zilibotti (1997) suggest that higher incomes make possible the diversification of indivisible risky projects. Consequently, countries should voluntarily choose to diversify more as their incomes grow, leading to a U-shaped pattern in sectoral concentration. The fact that Cambodia is not diversifying more rapidly may therefore be another symptom that investment is being hindered. Removing these constraints and related market failures not only will boost investment, which is an important engine for growth in itself, but will also lead to greater export discoveries and diversification, which will bring the fresh productivity gains that can be the backbone for sustained growth over the next decade.

In sum, despite Cambodia’s strong record of growth in the past, the underlying fundamentals for its sustained growth on many fronts will need to be significantly strengthened before they are comparable to the standards of peers. To ensure robust growth over the longer term, therefore, structural reforms will be necessary to improve infrastructure and to promote investment and total factor productivity.

Quantifying Growth Dividends

How much additional growth could such structural reforms promote? Specifically, if the reforms were to succeed in bridging the gaps in infrastructure quality between Cambodia and its peers, how much would Cambodia’s potential growth increase as a result? Estimates of investment elasticity to various measures of infrastructure quality have been computed by the IMF (2010) and are reproduced in the second column of Table 2.1. Gaps between the corresponding infrastructure quality of Cambodia and the average of its peers are shown in the first column; in all cases, the gaps are positive, indicating that Cambodia lags behind its peers. The calculation shows that if these gaps were to be closed by improved infrastructure, the estimated investment rate, shown in the third column, could be boosted by up to 5 percent of GDP.

Table 2.1Cambodia: Infrastructure Gaps and Investment
Gap with peersInvestment elasticityExtra investment rate (in percent of GDP)
Electric power consumption (kWh per capita)a6182.24.0
Telephone (lines per 100 persons)b81.44.4
Paved roads (square meters per capita)c0.54.04.8
Sources: World Bank. World Development Indicators; and IMF staff estimates.

Based on 2008 data; peers comprise Bangladesh, Bhutan, Mongolia, Nepal, Timor-Leste, and Vietnam.

Based on 2009 data; peers comprise Bangladesh, Bhutan, Lao P.D.R., Mongolia, Nepal, and Vietnam.

Based on 2004 data; peer is Vietnam.

Sources: World Bank. World Development Indicators; and IMF staff estimates.

Based on 2008 data; peers comprise Bangladesh, Bhutan, Mongolia, Nepal, Timor-Leste, and Vietnam.

Based on 2009 data; peers comprise Bangladesh, Bhutan, Lao P.D.R., Mongolia, Nepal, and Vietnam.

Based on 2004 data; peer is Vietnam.

In the event that the infrastructural improvements indeed raise the investment rate permanently by 5 percent of GDP, then the impact on potential growth based on the macro model would be as shown in Figure 2.10. The immediate effect on potential growth is about 1.2 percent, and the effect persists even after a decade. After 10 years, potential growth, despite declining from diminishing returns to physical capital, would still be 0.4 percentage point higher than without improvements in infrastructure. Thus, the growth impact is both substantial and long lasting. The cumulative effect of these growth dividends would lift the aggregate income level by more than 7 percent after 10 years.

Figure 2.10Potential Growth with Investment Boost, Cambodia, 1987–2020

(Percent)

Sources: Data provided by the Cambodian authorities; and IMF staff estimates and projections.

This simulation does not capture the direct effect of infrastructural improvement on total factor productivity, nor the positive feedback between investment and productivity gains, both of which can be significant albeit harder to measure. At the same time, to the extent that structural reforms may have an adverse effect on some sectors (for example, cheaper electricity may render old production technology obsolete and lower the demand for low-skilled labor), the net effect on growth may also be smaller. Notwithstanding these interactional effects, the magnitude of growth dividends from improved infrastructure according to the model estimate is significant and likely to be of first-order importance.

Conclusion

A key advantage of the proposed framework is that the quantitative model complements the investigative approach (i.e., growth diagnostics, as adopted in this chapter) in important ways. The latter identifies key constraints to growth and potential sources of structural breaks, and thus closes the gap left by the model. The model can then be used to simulate the identified potential breaks in order to obtain a quantitative assessment that would otherwise be unavailable. Another advantage is that, since the model has a state space structure, the availability of data is of less concern, since missing or delayed data can be estimated by the filter.

Given these advantages, the framework can also be useful for other developing economies. The details of the model and the investigative research design are flexible and can be suitably adjusted to fit the country application. For example, for countries with more stable macro relationships, the model may be extended to include a Euler equation that endogenizes saving and investment. For countries with fully independent monetary policy, a policy rule can be added. If less is known about the sectoral productivities, modeling aggregate total factor productivity alone will suffice. As the growth diagnostic exercises differ from country to country, the model structure can also be tailored according to the focus of the analysis.

Regarding Cambodia’s potential growth, the main findings are these: (i) Potential growth has indeed moderated from its 2004–05 peak, but the medium-term outlook of 7.5 percent remains relatively robust. (ii) However, the baseline outlook is conditional on Cambodia’s ability to maintain its productivity growth and rate of investment, which are currently constrained by a number of structural impediments. (iii) Mitigating some of these impediments, for example by closing the infrastructural gaps between Cambodia and other developing peers, would translate into a significant boost to investment and a corresponding growth dividend.

Appendix
Appendix Table A2.1Data for Observed Variables and Sources1
VariablesDataPeriod coveredSources
YtReal GDP at 2000 prices1986–2011NIS, IMF
gaptDeviation of Yt from the HP-filtered trend1986–2011NIS, IMF
πtCPI inflation (Phnom Penh)1995–2011NIS, IMF
LtTotal labor force1994–2009WDI
StGross fixed capital formation to GDP1995–2011NIS, IMF
AGtCereal yield (kg per hectare)1994–2009WDI
AMtElectricity production (kWh)1995–2008WDI
Note: CPI = consumer price index; HP = Hodrick-Prescott.

National Institute of Statistics of Cambodia (NIS); and World Bank, World Development Indicators (WDI) 2011.

Note: CPI = consumer price index; HP = Hodrick-Prescott.

National Institute of Statistics of Cambodia (NIS); and World Bank, World Development Indicators (WDI) 2011.

Appendix Table A2.2Priors and Posterior Estimates of Parameters
PriorPosterior
ParametersModeDispersionModeDispersion
a0.20.90.3NA
bG0.050.90.0420.0002
bM0.050.90.0130.0005
cG0.030.90.0330.0004
cM0.030.90.1500.0003
CO0.030.90.0270.0000
δ0.050.90.0780.0006
β0.90.90.4940.0279
γ0.90.90.4240.0346
π¯0.070.50.0610.0003
lg0.030.50.0290.0000
σεAG0.010.90.0991.8437
σεAM0.010.90.0730.9984
σεAO0.010.90.0060.0068
σεL0.010.90.0110.0236
σεY0.010.90.0400.3100
σεπ0.010.90.0530.5314
σεS0.0010.90.0420.3365
Note: NA = not applicable.
Note: NA = not applicable.
References

    AcemogluD. and F.Zilibotti1997Was Prometheus Unbound by Chance? Risk, Diversification and Growth,Journal of Political Economy Vol. 105 No. 4 pp. 70951.

    • Search Google Scholar
    • Export Citation

    BenesJ.K.ClintonR.Garcia-SaltosM.JohnsonD.LaxtonP.Manchev and T.Matheson2010Estimating Potential Output with a Multivariate Filter,IMF Working Paper No. 10/285 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    Congressional Budget Office (CBO)2001CBO’s Method for Estimating Potential Output: An Update,http://www.cbo.gov/ftpdocs/30xx/doc3020/PotentialOutput.pdf.

    • Search Google Scholar
    • Export Citation

    EpsteinN. and C.Macchiarelli2010Estimating Poland’s Potential Output: A Production Function Approach,Working Paper No. 10/15 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    HausmannR.J.Hwang and D.Rodrik2007What You Export Matters,Journal of Economic Growth Vol. 12 No. 1 pp. 125.

    HausmannR.D.Rodrik and A.Velasco2005Growth Diagnostics” (Cambridge, MA: JohnF.Kennedy School of Government Harvard University Center for International Development).

    • Search Google Scholar
    • Export Citation

    HummelsD. and P. J.Klenow2005The Variety and Quality of a Nation’s Exports,American Economic Review Vol. 95 No. 3 pp. 70423.

    • Search Google Scholar
    • Export Citation

    ImbsJ. and R.Wacziarg2003Stages of Diversification,American Economic Review Vol. 93 No. 1 pp. 6386.

    International Monetary Fund (IMF)2010Investment and Rebalancing in Asia,Regional Economic Outlook: Asia and Pacific (October) pp. 5770.

    • Search Google Scholar
    • Export Citation

    RoegerW.2006The Production Function Approach to Calculating Potential Growth and Output Gaps: Estimates for EU Member States and the US,EU Commission DG ECFIN.

    • Search Google Scholar
    • Export Citation

    RungcharoenkitkulPhurichai2012Modeling with Limited Data: Estimating Potential Growth in Cambodia,IMF Working Paper No. 12/96 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    WillmanA.2002Euro Area Production Function and Potential Output: A Supply Side System Approach,Working Paper No. 153 (Frankfurt: European Central Bank).

    • Search Google Scholar
    • Export Citation
1The production function approach can be motivated via an explicit microfoundation, as is done by Willman (2002) in another EU application. For developing economies, a simplified and reduced form of the method may instead be implemented. For example, Epstein and Macchiarelli (2010), using the case of Poland, focus on identifying the trend labor input in the production function (i.e., the natural rate of employment).
2A past estimate of the output gap, strictly speaking, is not required for the estimate of potential output. It is included here to “train” the model to produce a smooth estimate of potential growth. If in the modeler’s view, the latent potential growth can fluctuate meaningfully, then the HP-filtered gap can be dropped from the list of observed variables.
3The procedure is implemented using an open-source Matlab specialized toolbox, IRIS, developed by Jaromir Benes. Available via the internet at http://code.google.com/p/iris-toolbox-project/.
4The link between productivity in the traded goods sector (“export sophistication”) and economic growth is documented by Hausmann, Hwang, and Rodrick (2007). They also highlight the difficulties in transitioning from less to highly sophisticated markets without policy support, owing to the presence of network externalities.
6The estimated cross-sectional equation is y = 1.03 + 0.13x with R2 = 0.38, where y denotes real GDP growth per capita averaged over 2000–2009, and x is the gross fixed capital formation as a percentage of GDP averaged over 2000–2009. The sample countries comprise Bangladesh, Bhutan, Cambodia, China, India, Indonesia, Lao P.D.R., Mongolia, Nepal, Pakistan, the Philippines, Sri Lanka, and Vietnam.
7Imbs and Wacziarg (2003) document robust cross-country evidence that economies typically diversify until a certain income level is reached (typically around US$7,000–9,000 per capita), after which sectoral concentration resumes. Thus, the Herfindahl index typically decreases with income at an early stage of development, before rising later in a U-shaped pattern. Cambodia, on the other hand, has not followed this pattern and has not increased its diversification despite low per capita GDP (US$830 as of 2010).
8Extensive margin trade refers to increased exports of new products, which extends the frontier of exported product space. By contrast, intensive margin trade refers to growth in exports of existing products. Hummels and Klenow (2005) find that extensive margin trade accounts for around 60 percent of the export growth of advanced economies.

    Other Resources Citing This Publication