3. Is the Growth Momentum in Latin America Sustainable?
- International Monetary Fund. Western Hemisphere Dept.
- Published Date:
- May 2013
Latin America has enjoyed strong growth momentum during the last decade. While factor accumulation remains the main driver of GDP growth, the recent acceleration is mainly explained by higher total factor productivity (TFP). However, moving forward, this growth momentum might not be sustainable given some natural constraints on labor, despite recent capital and TFP trends.
Highly favorable external conditions—interrupted only temporarily during the 2008–09 global financial crisis—coupled with prudent macroeconomic policies bolstered GDP growth in most of Latin America during the last decade. The Latin America and the Caribbean (LAC) region has grown by an average of 4 percent per year since 2003, compared with less than 2½ percent annually in 1980–2002 (Figure 3.1). But, what explains this remarkable growth performance from a supply-side perspective, and will this momentum be sustainable in the years ahead?
Figure 3.1.Latin America and the Caribbean: Real GDP Growth Rate
Sources: IMF, World Economic Outlook; and IMF staff calculations.
1 Weighted average of countries in Latin America and the Caribbean.
This chapter addresses these questions by identifying the proximate causes of the recent strong growth performance and estimating potential growth rates for the period ahead based on standard (Solow-style) growth accounting methodologies.1 Our analysis is based on a group of 19 LAC countries starting in 1980.2 First, we decompose the sources of output growth into accumulation of factors of production and total factor productivity. The results are compared with the region’s performance in the past as well as with other regional benchmarks. Then, we analyze the sustainability of the recent strong growth momentum by estimating the potential growth rate ranges using the production function approach. To this end, we use a battery of common filtering techniques to measure the trend of the subcomponents of output (namely, capital, labor, and TFP), smoothing out cyclical fluctuations. We then use these trend series obtained with alternative methods to compute potential growth rate ranges for each country rather than a specific point estimate. To investigate the sustainability of recent high growth rates, we explore possible constraints on factor accumulation for the region’s growth performance.
This study is, to the best of our knowledge, the first to examine growth decomposition while looking at potential GDP growth rates in LAC countries from a cross-country perspective with actual data extended to 2012. Existing research usually focuses on only one country or a small group of countries, typically analyzing long-term developments over a 30–40 year time-span; and importantly, the analysis is usually concentrated in the period prior to the global financial crisis.3 To incorporate the latest available data, we create a new database for the subcomponents of output using data from Penn World Table 7.1 and the latest IMF World Economic Outlook database.
What Factors Drove the Recent Strong Growth Performance?
Although there is consensus that the robust growth performance in recent years has been to a great extent due to favorable external conditions (namely strong global growth, high commodity prices, and easy external financing conditions) that fueled external and domestic demand, it is less clear what the main drivers were from a supply perspective. To examine the latter, we use a simple accounting framework that decomposes output growth into the contributions from the accumulation of capital and (quality-adjusted) labor, and changes in TFP (see Annex 3.1 for details on the methodology and data).
Our key findings can be summarized as follows (Figure 3.2):
Factor accumulation (especially labor), rather than TFP growth, remains the main driver of output growth. In Latin America, total factor accumulation explained 3¾ percentage points of annual GDP growth in 2003–12, compared with ¾ percentage points by TFP. Interestingly, similar patterns are observed throughout Latin America irrespective of the country’s financial integration, export base/orientation, or market structure. Factor accumulation was also the main driver of growth in the Caribbean, but growth performance in this region during the recent period has been weaker than in the previous decade.
The recent growth pickup in Latin America is mainly explained by higher TFP. During the recent period, TFP has increased in most countries, in contrast to the lukewarm performance of the 1990s. Our estimates suggest that TFP explains about 1–1½ percentage points of the higher growth performance since 2003 compared with the 1990–2002 period. The contribution of physical capital also increased, though to a lesser extent, partly reflecting favorable external financial conditions and high investment (including foreign direct investment) in the primary sector associated with the commodity price boom.
Growth in the LAC region remains below that of emerging Asia, with most of the growth differential being explained by differences in TFP performance. On the positive side, Latin America’s growth gap vis-à-vis emerging Asia has narrowed compared with the 1990s, on account of a reduction in differences in capital contributions. However, large TFP growth differentials remain, accounting for most of the GDP growth gap in 2003–12. The labor contribution, in turn, has historically been larger in Latin America (especially in Central America) than in emerging Asia.
Declining unemployment is behind the strong labor contribution to growth in recent years. Much like in the 1990s, labor continues to be the main contributor to growth during 2003–12. However, the factors explaining this high contribution to growth have changed significantly. While increases in the working-age population and higher participation rates were the main factors in 1990–2002, their contribution (while still positive) has been smaller in 2003–12. Instead, increases in the rate of employment—a factor hindering growth in the previous period—played a key role more recently, consistent with near-record low unemployment levels in many countries. The contribution of improvements in human capital to output growth has typically been positive and broadly stable over time, accounting for about ½ percentage point of GDP growth.
TFP performance generally improved in 2003–12, although important differences across countries remain. After exhibiting declines in most of the region in previous decades, TFP growth mostly turned positive (particularly strong growth is recorded in Panama, Peru, and Uruguay), with a few exceptions (Figure 3.3).4 This partly reflects the expansionary phase of the economic cycle in most of these economies in 2003–12, as well as idiosyncratic factors in some cases (such as the canal expansion in Panama).5 In Chile—one of the few countries with positive TFP growth in Latin America during the 1980s and 1990s, TFP growth has turned negative in the last decade, partly reflecting declining productivity in the mining sector. This is in line with the experience in commodity-exporting advanced economies (such as Australia, Canada, and Norway) in the recent past, and is to a significant extent related to the expansion of energy and mining production to areas (fields or mines) of lower marginal productivity—where production has become profitable due to the commodity price boom. A few caveats about the estimation of TFP are worth mentioning, which imply that the results should be interpreted with caution.6 The TFP measure is by definition a residual—the difference between output growth and growth in the quantity (and quality) of inputs. Thus, any measurement errors in the labor and capital series are automatically imputed to TFP. For instance, (i) changes in the quality of the capital and labor stocks that we fail to account for, (ii) changes in the level of capital utilization, and/or (iii) changes in the use of land (a factor our methodology does not account for) would be reflected in the TFP component.
Figure 3.2.Decomposition of Real GDP Growth
Sources: IMF, World Economic Outlook; International Labor Organization; Penn World Table 7.1; World Bank, World Development Indicators; and IMF staff calculations.
1 Simple average of countries within each group. Latin America includes all Latin American countries in our sample. LA6 includes Brazil, Chile, Colombia, Mexico, Peru, and Uruguay. Other South America (SA) includes Bolivia, Ecuador, Paraguay, and Venezuela. Central America (CA) includes Costa Rica, Dominican Republic, El Salvador, Honduras, Nicaragua, and Panama. The Caribbean includes Barbados, Jamaica, and Trinidad and Tobago. Emerging Asia includes China, Indonesia, Malaysia, Philippines, and Thailand. Advanced commodity exporters includes Australia, Canada, New Zealand, and Norway.
2 For Central America: 1992–2002.
3 Excludes Paraguay and Nicaragua, owing to data limitations.
Figure 3.3.Latin America and the Caribbean: TFP Growth
Sources: Barro-Lee (2010); IMF, World Economic Outlook; Penn World Table 7.1; and IMF staff calculations.
1 For Central America: 1992–2002.
Is the Recent Strong Performance Sustainable?
To address this question, we estimate potential growth rate ranges for 2013–17 in LAC countries using a simple accounting framework that decomposes trend GDP growth into the contribution of changes in capital and labor inputs and TFP.
We find that, if recent historical trends for capital and TFP continue, and given some natural constraints on labor, then the current strong growth momentum is unlikely to be sustainable. While the region has, on average, grown by 4 percent during 2003–12, our estimates suggest that the average potential GDP growth rate in 2013–17 is closer to 3¼ percent.7 Indeed, the strong GDP growth rates observed in recent years are higher than (or close to the upper bound of) the potential output growth ranges for 2013–17 in most countries (Figure 3.4).8
Figure 3.4.Latin America and the Caribbean: Potential Output Growth Rate Ranges, 2013–171
This envisaged growth deceleration (from the recent high growth to projected potential growth rates) reflects lower contributions from all sources in the coming years:
Growth of physical capital is expected to moderate somewhat, reflecting a normalization of the easy external financing conditions and the stabilization of commodity prices—both key factors driving the recent strong domestic and foreign direct investment in the region.
The contribution of labor to output growth in the future will likely be limited by some natural constraints (Figure 3.5), including: (i) population ageing (the dependency ratio is expected to reach its minimum over the next years in several countries); (ii) limited scope to further increase labor force participation rates (including for females), which are relatively high already by international standards;9 and (iii) record low unemployment rates (which declined significantly, now representing a key driver of the labor contribution to output growth). Stronger contributions from human capital will require important improvements in the quality of schooling.10
TFP growth would also slow down, in line with the normalization of the business cycle. Therefore, TFP performance, which remains a concern despite its recent improvement, will be pivotal to sustain high growth rates in the region.
Figure 3.5.Labor Constraints to Future GDP Growth
Sources: Penn World Table 7.1; World Bank, World Development Indicators; and IMF staff calculations.
1 Emerging markets (EM) include China, Czech Republic, Estonia, Hungary, India, Indonesia, Poland, Russia, Slovakia, Slovenia, and Turkey.
Potential output ranges vary significantly across countries (see Figure 3.4). In this chapter, we do not attempt to explain cross-country differences in growth potential, although these often reflect differences in economic institutions, natural resource endowments, income inequality, financial sector deepening, and trade openness.
In light of the expected moderation in capital accumulation and the existing natural constraints on labor, the strong growth momentum in the region is unlikely to be sustainable unless TFP performance improves significantly.11 Although TFP contributed positively to GDP growth in recent years in most countries, its contribution was modest, especially considering cyclical issues and compared with other regions. Thus, fostering TFP growth remains a key challenge and priority for the LAC region.
The causes of low TFP growth in LAC countries are many and varied, and designing a policy agenda to unleash productivity is a difficult task. Policymakers should aim at policies that help reduce distortions in the allocation of resources, and this typically entails country-specific measures. Policies to be considered include: improving the business climate and enhancing competition; strengthening entry and exit regulation to facilitate the reallocation of resources to new and high-productivity sectors; improving infrastructure; promoting deeper and more efficient financial markets; enhancing research and development and innovation; and strengthening institutions to secure property rights and stamp out corruption. In the Caribbean, efforts are needed to tackle high debt levels and weak competitiveness, which have held back growth.
This annex describes the data and methodology used in the growth decomposition and in estimating potential growth rate ranges.12
The growth accounting exercise is based on the following standard Cobb-Douglas production function:
where Yt represents domestic output in period t, Kt the physical capital stock, Lt the employed labor force, ht human capital per worker, and At total factor productivity (TFP).13
We use annual data from Penn World Table 7.1 (PWT) for the period 1980 until 2010 and other sources—mainly the IMF’s World Economic Outlook (WEO) database for the subsequent years. Specifically, data on output, measured by real GDP, are obtained from PWT until 2010 and extended using WEO for 2011–12. The capital stock series is constructed with investment data from the PWT using the perpetual inventory method until 2010, and investment data from WEO for 2011–12. Our labor input series (measured by employment) refers to inputs effectively used in the production process. The employment series is obtained using the labor force series from PWT and the employment rate (one minus unemployment rate) from WEO. For 2011–12, we assume that the labor force rises in line with United Nation’s (U.N.) Population Projections (constant fertility scenario) for individuals aged 15 and over. To get quality-adjusted labor, we follow Bils and Klenow (2000) and Ferreira, Pessoa, and Veloso (2013) to model human capital as a function of the average years of schooling, using data from Barro and Lee (2010).
Using equation (1), we can decompose GDP growth as follows (denoting by
To estimate potential growth rates, we first estimate TFP using equation (1). We then obtain trend series for capital, labor, human capital, and TFP (KT, LT, hT, AT) for the period 1980–2017 using the Hodrick-Prescott (for both λ = 6.25 and λ = 100), Baxter and King, and Christiano and Fitzgerald filters.14 The following assumptions about the behavior of K, L, h, and A in 2013–17 are made: (i) we assume that both capital and TFP will grow by the average annual rate observed in 2000–12; and (ii) to project the labor input, we use projected unemployment rates (from WEO) and we assume that labor force grows in line with working-age population from U.N.’s Population Projections database and labor force participation rates remain constant at their latest observation. Finally, our measure of human capital increases at the 2005–10 average annual growth rate. Potential output growth
The potential growth rate is a good theoretical proxy for the long-term sustainable growth rate.
Our sample includes the following countries: Barbados, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Honduras, Jamaica, Mexico, Nicaragua, Panama, Paraguay, Peru, Trinidad and Tobago, Uruguay, and Venezuela. Argentina, Guatemala, and most of the small Caribbean islands are excluded owing to data limitations.
Our growth decomposition estimates (in particular the contribution of TFP growth) for the 1980s, 1990s, and early 2000s are in line with the findings in the literature (see Ferreira et al., 2013; Inter-American Development Bank, 2010; and Loayza et al., 2005).
Labor productivity has been particularly strong in the services sector in 2003–12, with declining trends in the mining sector. See Sosa and Tsounta (2013).
TFP measures the efficiency with which factors of production are used in the production process, and includes technology as well as the efficiency of markets.
These estimates are based on the assumption that both capital and TFP will grow at the average annual rate observed in 2000–12. This period covers a full economic cycle in most countries, whereas 2003–12 includes mainly the expansionary phase of the cycle. See Annex 3.1 for more details on our assumptions to project capital, labor, and TFP.
Mexico (strongly affected by the 2008–09 global financial crisis given its tight linkages with the U.S. economy) and Paraguay (owing to some idiosyncratic shocks) are exceptions.
In fact, the contributions to output growth of both changes in working-age population and the labor force participation rate have already narrowed significantly in 2003–12 compared with those of the 1990s. It is worth noting that these constraints on labor are less binding in countries with a large informal sector (e.g., Colombia, Mexico, Peru, and several Central American countries).
Although LAC region’s performance in terms of average years of schooling is relatively good compared with countries with similar income per capita, the quality of education has ample room for improvement (the region generally underperforms in terms of standard international tests).
Mobilizing higher domestic saving (which is low in the LAC region by international standards) could enhance the contribution of capital to long-term growth. Improving the quality of education would also help increasing potential output growth.
Our assumptions for the capital share of output, α, are country-specific and based on World Bank (2005) and are typically around 0.40 (in line with Gollin, 2002). Our main findings, however, are robust to a range of reasonable values for this parameter.
We include projections through 2017 to avoid the end-of-sample bias.