## _wp13132

## Source details

**Canonical URL:** [_wp13132](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13132.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13132.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13132.pdf.json)

---

### I. Introduction: scope and contributions
- Focus: experiences of nearly 70 LICs using data from the early 1950s.
- Three contributions:
  - Exclusive focus on LICs distinct from emerging markets.
  - Exhaustive classification of each LIC’s growth experience into episodes of growth takeoff versus other periods.
  - Complement statistical analysis with historical case studies to assess policies that sustained or derailed takeoffs.

### Key definitions and methodology
- Advanced economies: member economies of the OECD as of 1990, with the exception of Turkey.
- LIC definition:
  - Economy whose average real output per capita over the previous five years is lower than a time-varying low-income threshold.
  - Low-income threshold in 1990: bottom 45th percentile of average EMDE output per capita (about $2,600 in purchasing-power-parity-adjusted constant 2005 U.S. dollars).
  - Threshold extrapolated backward and forward using average growth rate of global output per capita during 1950–2011 (about 2.3 percent per year).
- Sample exclusions: economies with average 1950–2011 population less than 1 million; China and India are included in EMDEs but not LICs.
- Turning points detection: Harding and Pagan (2002) algorithm to pick local peaks and troughs in annual PPP-adjusted real GDP per capita.
- Cycle constraint: cycle (one upswing and one downswing) must be at least five years long.
- Takeoff definition: an upswing in LIC output per capita that lasts at least five years, with average annual growth of at least 3.5 percent.
- Postconflict takeoffs: some episodes following serious internal or external conflicts were excluded (see Appendix 1 in source).

### Empirical identification and counts
- Identified takeoffs:
  - 29 growth takeoffs during 1990–2011.
  - 41 takeoffs prior to 1990.
- Some episodes ongoing as of 2011 use duration and average growth as of 2011.

### Major empirical findings and stylized facts
- Frequency and timing:
  - Two waves of LIC growth takeoffs: the 1960s and early 1970s, and again since the 1990s.
  - Frequency of growth takeoffs declined after 2008, partly due to data censoring and a drop in share of LICs that sustained takeoffs.
  - One-third of LICs sustained their takeoffs as of 2011 compared with an average of 20 percent during the 1980s.
- Duration and growth outcomes:
  - Recent takeoffs have lasted longer than those prior to the 1990s.
  - Median duration for recent takeoffs:
    - 9 years for growth episodes already completed.
    - 12 years for episodes still ongoing as of 2011.
  - Median growth in output per capita:
    - 6¼ percent for recent takeoffs that have already ended.
    - 5¼ percent for recent takeoffs that were still ongoing as of 2011.
  - For takeoffs before 1990:
    - Median duration about 7 years.
    - Median growth about 5 percent.
- Payoff of takeoffs:
  - Takeoffs typically lead to 50–60 percent increases in real per capita GDP within 10 years after takeoff.
  - Resource-rich LICs in the current generation typically saw GDP per capita rise by 80 percent in 10 years (many resource-rich peers did not take off).
  - Manufacturing-oriented dynamic LICs in both generations saw a 50 percent rise in GDP per capita after 10 years.
- Global conditions and timing:
  - Recent takeoffs started under weaker global growth and higher global interest rates compared with pre-1990 takeoffs; conditions improved after takeoff for the current generation but deteriorated for the previous generation.
  - Pre-takeoff terms-of-trade (TOT) growth was more favorable for recent takeoffs than for previous-generation takeoffs; TOT growth rose for both generations after takeoff.
  - LICs that failed to take off faced similar global conditions as those that did, indicating domestic conditions and policies matter.
- Macroeconomic and structural correlates (recent vs predecessors):
  - Recent takeoffs associated with stronger post-takeoff macroeconomic outcomes: lower debt and inflation, more competitive exchange rates, higher investment levels.
  - Structural reforms and institutions in recent takeoffs: better infrastructure, lower regulatory burden, more political stability.
  - Financing patterns: greater reliance on FDI-financed investment instead of debt-financed investment in recent takeoffs.
- Case-study evidence:
  - Sustaining strong macroeconomic and structural policies is key to sustaining growth takeoffs; sustained improved policies increase likelihood of avoiding reversals experienced by earlier generations.

### Crises and historical caution
- Crises incidence:
  - Close to one-third of previous takeoffs ended with a currency, debt, or banking crisis.
  - Less than 15 percent of recent takeoffs have ended with crises thus far.
  - Over 1970–89, 32 percent of growth takeoffs were associated with a financial crisis near their end; over 1990–2011, the corresponding incidence was 14 percent.
- Examples of crisis-linked takeoffs (preserved entries):
  - Indonesia: Takeoff Start 1964, Takeoff End 1981, 1979 (currency)
  - Thailand: Takeoff Start 1973, Takeoff End 1982, 1983 (banking)
  - Mali: Takeoff Start 1975, Takeoff End 1986, 1987 (banking)
  - Vietnam: Takeoff Start 1975, Takeoff End 1979, 1981 (currency)
  - Republic of Congo: Takeoff Start 1977, Takeoff End 1984, 1986 (debt)
  - Zimbabwe: Takeoff Start 1977, Takeoff End 1983, 1983 (currency)
  - Lao P.D.R.: Takeoff Start 1980, Takeoff End 1986, 1997 (currency)
  - Sierra Leone: Takeoff Start 1981, Takeoff End 1987, 1989 (currency)
  - Indonesia (later): Takeoff Start 1983, Takeoff End 1997, 1997 (banking), 1998 (currency), 1999 (debt)
  - Tanzania: Takeoff Start 1985, Takeoff End 1991, 1990 (currency)
  - Uganda: Takeoff Start 1988, Takeoff End 1994, 1994 (banking)
  - Lao P.D.R. (later): Takeoff Start 1989, Takeoff End 1997, 1986 (currency)
  - Yemen: Takeoff Start 1992, Takeoff End 1998, 1996 (banking)
  - Nigeria: Takeoff Start 1996, Takeoff End 2008, 2009 (banking)
  - Zambia: Takeoff Start 2000, Takeoff End 2008, 2009 (currency)
  - Mongolia: Takeoff Start 2002, Takeoff End 2008, 2008 (banking)

### Investment, saving, and financing patterns
- Takeoffs in both generations correlated with higher investment and national saving rates than in LICs that did not take off.
- Current account deficits broadly similar across generations.
- A larger share of current account deficits was financed by FDI flows for the current generation of takeoffs.
- FDI flows rise sharply after takeoff for the current generation of dynamic LICs (especially resource rich dynamic LICs); for manufacturing-oriented LICs, current-generation FDI levels exceed previous generation levels but did not increase sharply following takeoff.

### Public and external debt, inflation, and financial indicators
- Among current-generation dynamic LICs, within 10 years after takeoff:
  - Public debt decreased from more than 90 percent of GDP to 44 percent of GDP.
  - External debt fell from more than 70 percent of GDP to about 44 percent of GDP.
- Excluding economies that received debt relief, typical patterns:
  - Public debt declines from 50 percent of GDP before takeoff to 40 percent of GDP in 10 years after takeoff.
  - External debt declines from 60 percent of GDP to about 45 percent of GDP in the same period.
- Previous generation patterns:
  - Public and external debt stood at 40 and 33 percent of GDP, respectively, before takeoff, but more than doubled within 10 to 20 years after takeoff, with inflation tending to increase.
- Financial imbalances:
  - No compelling evidence that recent takeoffs are accompanied by rising financial imbalances; credit-to-GDP ratio tends to increase gradually in recent takeoffs.
  - Credit-to-GDP ratios in dynamic LICs were lower than in LICs with weaker growth and in LICs that took off in the previous generation.

### Competitiveness, export structure, and real exchange rate dynamics
- Export performance:
  - Both generations: LICs with takeoffs experienced stronger export growth than LICs with weaker growth.
  - Recent takeoffs tend to have more geographically diversified exports but diversification reversed in the 10 years after takeoff for the current generation (partly due to resource discoveries and specialization).
  - Greater trade exposure to other EMDEs implies greater exposure to risks in those EMDEs and commodity-price risks.
- Real exchange rate dynamics:
  - Recent LIC takeoffs: real exchange rates versus the U.S. dollar typically depreciated during the five year periods before and at the start of a takeoff, with appreciation pressure during years 6 to 10 after takeoff.
  - Real exchange rate deviation: dynamic LICs are typically some 25 to 30 percent weaker than their productivity-adjusted long-term level (Figure 8, panels 9 and 10).
  - For takeoffs before 1990, the real exchange rate was as much as 10 to 15 percent stronger than its productivity-adjusted long-term level until five years after takeoff.
- Foreign reserves: greater accumulation in dynamic LICs, which may have helped maintain price competitiveness.

### Structural reforms, human capital, and institutions
- Dynamic LICs (both generations) tend to have:
  - Smaller governments (Size of Government index, 0 to 10).
  - Lower regulatory barriers (index, 0 to 10).
  - Better infrastructure (Telephones per thousand people).
  - Higher human capital (Educational Attainment in years of schooling).
- Recent takeoffs (post-1990) show stronger improvements after takeoff:
  - Size of government and regulatory barriers continued to decline after takeoff.
  - Infrastructure and education continued to improve.
  - Income inequality (Gini): recent takeoffs display more income equality versus higher inequality in the previous generation.
  - Political institutions (Constraints on the Executive index 0 to 1): stronger in the current generation of takeoffs.

### Interpretation and policy-relevant associations
- Strong investment growth associated with takeoffs, but sustainability matters: investment financed by FDI and accompanied by macroeconomic discipline featured in recent successful takeoffs.
- Risks remain: previous-generation takeoffs frequently ended in crises; reversals occurred within 20 years for some.
- Structural and policy complementarities important:
  - Policies that avoid inducing macroeconomic vulnerability while supporting productive investment are necessary.
  - Continued economic and export diversification essential to improve resilience, especially where post-takeoff export concentration increases due to resource specialization.
- Stylized facts are correlations and should be interpreted as associations, not proven causal drivers.

### Case-study lessons (selected)
- Brazil and Korea, 1960–80:
  - Brazil: inward-oriented, import-substitution, large domestic-market focus; high public debt growth and eventual debt crisis; slow educational advancement and persistently high income inequality.
  - Korea: export-orientation, promotion of labor-intensive then higher-value-added industries; strong investment in human capital and targeted policies to sustain competitiveness; fiscal discipline and monetary targeting kept inflation under control.
- Indonesia:
  - Early energy-led growth; oil windfalls directed to public investment in rural infrastructure, agriculture, health, and education in the 1960s–1970s.
  - Later shift toward manufacturing via deregulation and liberalization; experienced 1998 crisis and subsequent rebound.
  - Annual growth in real GDP output per capita averaged 3¾ percent in the 2000s.
- Mozambique:
  - Post-1992 peace and stability, 5¾ percent average annual growth in per capita real GDP over 16 years.
  - Growth capital-intensive and resource-focused with limited employment generation and modest poverty reductions.
  - Policy priorities: transport and energy infrastructure, human capital, broader access to financing, agricultural expansion.
- Cambodia:
  - Output per capita grew at average annual rate of 6 percent over the past decade.
  - Growth supported by export-oriented textile industry (three-quarters of goods exports), tourism, agriculture, and FDI.
  - Credit-to-GDP ratio quadrupled to 35 percent in less than 10 years.
  - Policy recommendations: remove infrastructure bottlenecks, deepen financial intermediation with prudential supervision, improve public debt management, mobilize fiscal revenue for human capital.

### Takeaways from the case studies
- Growth takeoffs feasible under various strategies; sustaining strong growth requires continued effort to reduce external and internal imbalances.
- Avoid financing investment by excessive debt—today’s dynamic LICs, now only 9 to 12 years into their takeoffs, should avoid excessive debt financing.
- Structural reforms raise productivity and help broad-based growth; human-capital and productivity-enhancing policies are central.
- Policies must adjust to changing global conditions; avoid procyclical policies despite ultralow global interest rates.

### Conclusions — stylized facts and policy implications
- Scope: examined episodes of growth takeoffs in nearly 70 developing economies or LICs over the past six decades.
- Waves: first wave in the 1960s and early 1970s; rebound since the 1990s.
- Share sustained: more than half of recent takeoffs, or one-third of all LICs, continued to expand through 2011.
- Duration:
  - Recent takeoffs: 9 to 12 years on average.
  - Prior to 1990: typically 7 years.
- Payoff: 50 to 60 percent rise in per capita output in 10 years after start of takeoff, versus only 5 to 15 percent for LICs that did not take off.
- Common patterns: higher investment rates and greater trade integration in takeoffs across generations; export growth higher in recent takeoffs.
- Lower vulnerabilities in current generation: declines in debt and inflation after takeoff; greater reliance on FDI vs debt.
- Determinants of takeoff probability (controlling for global conditions):
  - Probability rises with higher global growth, initial size, years of schooling, and initial level and increases in investment.
  - Probability falls with initial income per capita, appreciation/overvaluation of the real exchange rate, and increases in public debt.
  - Overall chances of starting a takeoff tripled in the 2000s compared to pre-1990, with domestic conditions and policies accounting for a majority of the rise.
- Caveats: many LICs still far from advanced-economy income levels; sectoral concentration and reliance on FDI can pose Dutch disease risks; ensuring broad-based living-standard improvements remains a challenge.
- Policy priorities: continue reforms, avoid major macroeconomic imbalances, maintain external competitiveness.

### Logit regressions — approach, selected results, and robustness
- Model: annual-frequency logit where g_it = 1 if economy i starts a strong growth takeoff at time t.
- Structural break in relationships in 1990 investigated; relationships are associational, not causal.
- Selected logit estimation results (preserved as in source — Logit Coef. ; Aver. Marg. Eff.):
  - Contemporaneous World Real GDP Growth:
    - Full Sample: 0.800** ; 2.250**
    - Before 1990: 0.859** ; 2.450**
    - 1990–2011: 1.866*** ; 4.200***
  - Contemporaneous U.S. Three-Month Treasury Bill Real Rate:
    - Full Sample: 0.032 ; 0.091
    - Before 1990: 0.110 ; 0.313
    - 1990–2011: 0.433 ; 0.973
  - Initial Log Real GDP per Capita:
    - Full Sample: –2.439*** ; –6.880***
    - Before 1990: –1.543 ; –4.400
    - 1990–2011: –7.095*** ; –16.000***
  - Initial Educational Attainment:
    - Full Sample: 0.301* ; 0.848*
    - Before 1990: 0.048 ; 0.137
    - 1990–2011: 0.903** ; 2.030*
  - Change in Real Investment Divided by GDP:
    - Full Sample: 0.149*** ; 0.420***
    - Before 1990: 0.234*** ; 0.668***
    - 1990–2011: 0.177*** ; 0.397***
  - Change in Public Debt Divided by GDP:
    - Full Sample: –0.003 ; –0.009
    - Before 1990: –0.019 ; –0.055
    - 1990–2011: –0.014*** ; –0.031**
- Baseline predicted annual probability of a new takeoff:
  - less than 1 percent before 1990 (0.8 percent predicted probability reported for the subsample before 1990).
  - close to 3 percent during the 2000s (2.8 per cent predicted probability reported for the 2000–11 subsample).
- Contributions to the percent change in the odds ratio (2000s vs before 1990):
  - Structural conditions (notably more years of schooling) contributed most to the increase.
  - Macro-economic conditions (higher investment growth, lower debt) contributed materially.
  - Global conditions, initial size, and openness/integration also contributed positively; higher initial income in the 2000s lowered the chance (convergence effect).

### Robustness checks and alternative specifications
- Alternative LIC samples and takeoff definitions assessed, including:
  - HPR (Hausmann, Pritchett, and Rodrik, 2005) definition.
  - Time-invariant income threshold of $2,600 (2005 PPP).
  - Higher growth threshold of 5 percent.
  - Merging episodes separated by ≤ 5 years.
- Effects on episode counts:
  - HPR algorithm yields 55 growth accelerations in LIC sample: 31 during 1990–2011, and 24 prior to 1990.
  - Excluding temporary delays reduces episodes from 29 to 24 during 1990–2011 and from 41 to 31 prior to 1990s.
  - Raising cutoff to 5 percent reduces takeoffs to 17 from 29 during 1990–2011 and to 20 from 41 during 1950–89.
- Stylized facts robust across alternative definitions; some indicators (income inequality, political institutions) vary depending on definition used.
- Alternative estimation methods for rare events and small-sample bias (Firth, King and Zeng, complementary log-log, random effects) yield broadly similar signs and magnitudes.
- AUC values (full sample across methods) around 0.814–0.818.

### Data definitions and key sources (Appendix 1 highlights)
- Primary data sources: IMF’s World Economic Outlook (WEO); Penn World Table 7.1 (PWT); World Bank WDI.
- Crisis and conflict data: Laeven and Valencia (2012); The New COW War Data, 1816–2007 v4.0.
- Sectoral and structural indicators: Papageorgiou and Spatafora (2012); Barro and Lee (2010); Lane and Milesi-Ferretti (2007) updated to 2011; IMF databases; UN Comtrade.
- Transformations for logistic regression:
  - initial — once-lagged, backward-looking five-year average;
  - contemporaneous — current year, forward-looking five-year average;
  - change — difference between contemporaneous and initial values.
- Country grouping and classification rules preserved (thresholds, percentile cutoffs, population limits, resource/manufacturing/agricultural classification rules).

*Source: IMF staff calculations and text in the provided content unit.*

### References .............................................................................................................

### _wp13132 - References

### I. Introduction: scope and contributions
- Focus: experiences of nearly 70 LICs using data from the early 1950s.
- Three contributions:
  - Exclusive focus on LICs distinct from emerging markets.
  - Exhaustive classification of each LIC’s growth experience into episodes of growth takeoff versus other periods.
  - Complement statistical analysis with historical case studies to assess policies that sustained or derailed takeoffs.

### Key definitions and methodology
- Advanced economies: member economies of the OECD as of 1990, with the exception of Turkey.
- LIC definition: economy whose average real output per capita over the previous five years is lower than a time-varying low-income threshold.
  - Low-income threshold in 1990: bottom 45th percentile of average EMDE output per capita (about $2,600 in purchasing-power-parity-adjusted constant 2005 U.S. dollars).
  - Threshold extrapolated backward and forward using average growth rate of global output per capita during 1950–2011 (about 2.3 percent per year).
- Sample exclusions: economies with average 1950–2011 population less than 1 million; China and India are included in EMDEs but not LICs.
- Turning points detection: Harding and Pagan (2002) algorithm used to pick local peaks and troughs in annual PPP-adjusted real GDP per capita.
- Cycle constraint: cycle (one upswing and one downswing) must be at least five years long.
- Takeoff definition: an upswing in LIC output per capita that lasts at least five years, with average annual growth of at least 3.5 percent.
  - The 3.5 percent threshold corresponds to the 60th percentile of growth in output per capita in all EMDEs in the past two decades and aligns with prior studies.
- Postconflict takeoffs: some episodes following serious internal or external conflicts were excluded from analysis (see Appendix 1 in source).

### Empirical identification and counts
- Identified takeoffs:
  - 29 growth takeoffs during 1990–2011.
  - 41 takeoffs prior to 1990.
- Some episodes ongoing as of 2011 use duration and average growth as of 2011.

### Major empirical findings and stylized facts
- Frequency and timing:
  - LIC growth takeoffs occurred in two waves: the 1960s and early 1970s, and again in the last two decades (since the 1990s).
  - Frequency of growth takeoffs declined after 2008, partly due to data censoring and a drop in share of LICs that sustained takeoffs.
  - One-third of LICs sustained their takeoffs as of 2011 compared with an average of 20 percent during the 1980s.
- Duration and growth outcomes:
  - Recent takeoffs have lasted longer than those prior to the 1990s.
  - Median duration for recent takeoffs:
    - 9 years for growth episodes already completed.
    - 12 years for episodes still ongoing as of 2011.
  - Median growth in output per capita:
    - 6¼ percent for recent takeoffs that have already ended.
    - 5¼ percent for recent takeoffs that were still ongoing as of 2011.
  - For takeoffs before 1990:
    - Median duration about 7 years.
    - Median growth about 5 percent.
- Payoff of takeoffs:
  - Takeoffs typically lead to 50-60 percent increases in real per capita GDP within 10 years after takeoff (though some previous-generation takeoffs ended in crises or reversed sharply).
  - Resource-rich LICs in the current generation typically saw GDP per capita rise by 80 percent in 10 years, although many resource-rich peers did not take off.
  - Manufacturing-oriented dynamic LICs in both generations saw a 50 percent rise in GDP per capita after 10 years, while many manufacturing-oriented peers failed to take off.
- Global conditions and timing:
  - Recent takeoffs started under weaker global growth and higher global interest rates compared with pre-1990 takeoffs; conditions improved after takeoff for the current generation but deteriorated for the previous generation.
  - Pre-takeoff terms-of-trade (TOT) growth was more favorable for recent takeoffs than for previous-generation takeoffs; TOT growth rose for both generations after takeoff.
  - LICs that failed to take off faced similar global conditions as those that did, indicating domestic conditions and policies matter.
- Macroeconomic and structural correlates (current-generation vs predecessors):
  - Recent takeoffs associated with stronger post-takeoff macroeconomic outcomes:
    - Lower debt and inflation.
    - More competitive exchange rates.
    - Higher investment levels.
  - Structural reforms and institutions in recent takeoffs:
    - Better infrastructure.
    - Lower regulatory burden.
    - More political stability.
  - Financing patterns: greater reliance on FDI-financed investment instead of debt-financed investment in recent takeoffs.
- Case-study evidence:
  - Case studies emphasize that sustaining strong macroeconomic and structural policies is key to sustaining growth takeoffs.
  - Sustained improved policies increase likelihood of avoiding reversals experienced by earlier generations.

### Paper organization (as given)
- Section 2: identifies LIC growth takeoffs and compares them in historical perspective.
- Section 3: empirical analysis — documents behavior of key macroeconomic and structural policy variables underlying LIC takeoffs in the two generations (1990–2011 and prior to the 1990s); presents multivariate regression analysis and robustness checks.
- Section 4: case studies.
- Section 5: conclusion.

*Source: IMF staff calculations and text in the provided content unit.*

### 1. LICs excl ude countr i es exper i enci ng or  r ecover i ng fr om a ser i ous exter nal  or  i nter nal

### _wp13132 - 1. LICs excl ude countr i es exper i enci ng or  r ecover i ng fr om a ser i ous exter nal  or  i nter nal

### Key empirical findings on takeoffs and output per capita
- Output per capita typically rose by 60 percent after 10 years following takeoff for takeoff LICs, compared with about 15 percent for LICs that did not take off (Figure 5, panel 1).
- For the previous generation, output per capita increased by 50 percent in 10 years after takeoff and doubled within 25 years (Figure 5, panels 2 and 3).
- In the previous-generation takeoffs, the pace of growth slowed after 10 years; those in the bottom quartile reversed their gains in output per capita within 20 years after takeoff (Figure 5, panel 3).

### Crises incidence and historical caution
- Close to one-third of previous takeoffs ended with a currency, debt, or banking crisis (Table 3).
- Less than 15 percent of recent takeoffs have ended with crises thus far.
- Table 3: growth takeoffs are shown with start and end years and crises (1970–2011); over 1970–89, 32 percent of growth takeoffs were associated with a financial crisis near their end; over 1990–2011, the corresponding incidence was 14 percent.
- Countries listed in Table 3 experienced specific crises (examples from table entries preserved as text): 
  - Indonesia: Takeoff Start 1964, Takeoff End 1981, 1979 (currency)
  - Thailand: Takeoff Start 1973, Takeoff End 1982, 1983 (banking)
  - Mali: Takeoff Start 1975, Takeoff End 1986, 1987 (banking)
  - Vietnam: Takeoff Start 1975, Takeoff End 1979, 1981 (currency)
  - Republic of Congo: Takeoff Start 1977, Takeoff End 1984, 1986 (debt)
  - Zimbabwe: Takeoff Start 1977, Takeoff End 1983, 1983 (currency)
  - Lao P.D.R.: Takeoff Start 1980, Takeoff End 1986, 1997 (currency)
  - Sierra Leone: Takeoff Start 1981, Takeoff End 1987, 1989 (currency)
  - Indonesia (later): Takeoff Start 1983, Takeoff End 1997, 1997 (banking), 1998 (currency), 1999 (debt)
  - Tanzania: Takeoff Start 1985, Takeoff End 1991, 1990 (currency)
  - Uganda: Takeoff Start 1988, Takeoff End 1994, 1994 (banking)
  - Lao P.D.R. (later): Takeoff Start 1989, Takeoff End 1997, 1986 (currency)
  - Yemen: Takeoff Start 1992, Takeoff End 1998, 1996 (banking)
  - Nigeria: Takeoff Start 1996, Takeoff End 2008, 2009 (banking)
  - Zambia: Takeoff Start 2000, Takeoff End 2008, 2009 (currency)
  - Mongolia: Takeoff Start 2002, Takeoff End 2008, 2008 (banking)

### Investment, saving, and financing patterns
- Takeoffs in both generations were correlated with higher levels of investment and national saving rates compared with LICs that did not take off (Figure 6, panels 1–4).
- Current account deficits were broadly similar in both generations (Figure 6, panels 5–6).
- A larger share of current account deficits was financed by foreign direct investment (FDI) flows for the current generation of takeoffs compared with the previous generation.
- FDI flows rise sharply after takeoff for the current generation of dynamic LICs compared with both LICs with weak growth and the previous generation of dynamic LICs (Figure 6, panels 7–8).
- Note from footnote: the sharp increase post-takeoff is particularly observed for the resource rich dynamic LICs; for manufacturing-oriented LICs, FDI levels for the current generation exceed those in previous generations but did not increase sharply following takeoff.

### Public and external debt, inflation, and financial indicators
- Among current-generation dynamic LICs, within 10 years after takeoff:
  - Public debt decreased from more than 90 percent of GDP to 44 percent of GDP.
  - External debt fell from more than 70 percent of GDP to about 44 percent of GDP.
- Excluding economies that received debt relief, typical patterns are:
  - Public debt declines from 50 percent of GDP before takeoff to 40 percent of GDP in 10 years after takeoff.
  - External debt declines from 60 percent of GDP to about 45 percent of GDP in the same period.
- The current generation’s reliance on FDI and greater macroeconomic policy discipline appears associated with lower inflation after takeoff compared with the previous generation (Figure 7, panels 5–6).
- For the previous generation, public and external debt stood at 40 and 33 percent of GDP, respectively, before takeoff, but more than doubled within 10 to 20 years after takeoff, with inflation tending to increase.
- There is no compelling evidence that recent takeoffs are accompanied by rising financial imbalances; the ratio of credit to GDP tends to increase gradually in recent takeoffs (Figure 7, panels 7–8).
- Credit-to-GDP ratios in dynamic LICs were lower than in LICs with weaker growth and in LICs that took off in the previous generation.

### Competitiveness, export structure, and real exchange rate dynamics
- Both generations: LICs with takeoffs experienced stronger export growth than LICs with weaker growth (Figure 8, panels 1–2).
- Today’s LIC takeoffs tend to have more geographically diversified exports, which may help sustain export growth despite weak advanced-economy demand (Figure 8, panels 3–4).
- Greater trade exposure to other EMDEs implies greater exposure to risks to growth in those EMDEs and related commodity-price risks.
- Export structures were more diversified in dynamic LICs of both generations than in those with weak growth, but diversification reversed in the 10 years after takeoff for the current generation (Figure 8, panels 5–6).
- The greater concentration of exports after takeoff partly relates to increased specialization in commodity-related activity in LICs that discover natural resources.
- Real exchange rate behavior:
  - Recent LIC takeoffs: real exchange rates versus the U.S. dollar typically depreciated during the five year periods before and at the start of a takeoff, with appreciation pressure during years 6 to 10 after takeoff (Figure 8, panels 7–8).
  - The analysis shows statistically significant movements in some periods (notations such as t [–4,0]*, t [1,5], t [6,10], and associated significance markers appear in figure panels).

### Interpretation and policy-relevant associations
- Strong investment growth is associated with takeoffs, but sustainability matters: investment financed by FDI and accompanied by macroeconomic discipline has featured in recent successful takeoffs.
- Risks remain: previous-generation takeoffs frequently ended in crises; a nontrivial share of dynamic LICs in the past have seen reversals in output per capita within 20 years of takeoff.
- Structural and policy complementarities appear important for sustaining takeoffs:
  - Policies that avoid inducing macroeconomic vulnerability while supporting productive investment are necessary.
  - Continued economic and export diversification will be needed to improve resilience, particularly where post-takeoff export concentration increases due to resource specialization.
- Stylized facts are correlations and should be interpreted as associations with takeoffs rather than proven causal drivers.

*Source: IMF staff calculations and analyses as presented in the chapter.*

### Appendix 1 for the definition of conflict and the source of the conflict data. *, **, and *** denote statistically signi

### _wp13132 - Appendix 1 for the definition of conflict and the source of the conflict data. *, **, and *** denote statistically signi

### Definitions and data notes
- *, **, and *** denote statistically significant difference in distributions (based on the Kolmogorov-Smirnov test) at the 10 per cent, 5 per cent, and 1 per cent levels, respectively.
- Significance tests on the x-axis are for the difference in the distributions between groups of strong and weak growth. Significance tests on the blue bars are for the difference in the distributions across 1990–2011 and before 1990 (not shown for red bars).
- A constant composition sample underlies each panel to ensure comparability within the group of strong and weak growth episodes across time for that panel.
- Appendix 1 defines conflict and provides the source of the conflict data (not reproduced here).

### External competitiveness, exports, diversification, and reserves (Figure 8 highlights)
- Real exchange rate change (percent change): displayed panels compare LICs with strong growth vs LICs with weak growth across Before 1990 and 1990–2011 for t [–4,0], t [1,5], t [6,10].
- Real exchange rate deviation (percent difference from fitted value): panels 9 and 10 indicate dynamic LICs are typically some 25 to 30 percent weaker than their productivity-adjusted long-term level (text reference to Figure 8, panels 9 and 10).
- For takeoffs before 1990, the real exchange rate was as much as 10 to 15 percent stronger than its productivity-adjusted long-term level until five years after takeoff.
- Foreign reserves (percent of GDP) show greater accumulation in dynamic LICs (Figure 8, panels 11 and 12), which may have helped maintain price competitiveness.

### Structural reforms, infrastructure, human capital, and institutions (Figure 9 highlights)
- Dynamic LICs (both generations) tend to have:
  - Smaller governments (Size of Government index, 0 to 10).
  - Lower regulatory barriers (index, 0 to 10).
  - Better infrastructure (Telephones per thousand people).
  - Higher human capital levels (Educational Attainment in years of schooling).
- Recent takeoffs (post-1990) show stronger improvements after takeoff:
  - Size of government and regulatory barriers continued to decline after takeoff.
  - Infrastructure and education continued to improve.
  - Earlier generation (pre-1990) generally did not show these sustained improvements except for education.
- Income inequality (Gini coefficient):
  - Recent takeoffs display more income equality, whereas income inequality was typically high in the previous generation of takeoffs.
- Political institutions (Constraints on the Executive index, 0 to 1 with unlimited authority = 0 and executive parity = 1):
  - Political institutions are stronger in the current generation of takeoffs—possibly reflecting the end of conflicts or greater democratization.

### Stylized conclusions from descriptive evidence
- Recent LIC growth takeoffs are generally underpinned by stronger fundamentals compared with takeoffs prior to 1990.
- Strong investment- and externally-oriented growth in recent takeoffs relied less on foreign borrowing, giving dynamic LICs more room to maneuver policies.
- Growth was helped by productivity-enhancing structural reforms and more inclusive institutions.
- Further export diversification will be essential to improve economic resilience.
- Many of these indicators are regarded as key determinants of sustained growth and bode well for today’s dynamic LICs, particularly if they maintain policy momentum.

### Logit regressions—approach and caveats
- A conditional probability of an LIC new growth takeoff is estimated at an annual frequency using a logit model. The binary dependent variable g_it equals 1 if economy i starts a strong growth takeoff at time t, and 0 otherwise.
- The model form (probability of g_it = 1) follows the logit functional form with K potential covariates x_j and coefficients β_j; estimated by maximum likelihood.
- The possibility of a structural break in relationships in 1990 is investigated.
- All estimated relationships are associational, not causal.
- Limited availability of data and the relative rarity of takeoffs imply results should be interpreted with caution.
- ROC curve and area under the ROC curve (AUC) are used to assess model performance. The Youden index and its cutoff π* are used to choose an operating threshold.

### Selected logit estimation results (Table 4 — reported values preserved as in source)
- Global Conditions
  - Contemporaneous World Real GDP Growth:
    - Full Sample: Logit Coef. 0.800** ; Aver. Marg. Eff. 2.250**
    - Before 1990: Logit Coef. 0.859** ; Aver. Marg. Eff. 2.450**
    - 1990–2011: Logit Coef. 1.866*** ; Aver. Marg. Eff. 4.200***
  - Contemporaneous U.S. Three-Month Treasury Bill Real Rate:
    - Full Sample: 0.032 ; 0.091
    - Before 1990: 0.110 ; 0.313
    - 1990–2011: 0.433 ; 0.973
  - Contemporaneous Terms-of-Trade Growth:
    - Full Sample: 0.008 ; 0.024
    - Before 1990: 0.031 ; 0.088
    - 1990–2011: 0.002 ; 0.005

- Income per Capita and Size
  - Initial Log Real GDP per Capita:
    - Full Sample: –2.439*** ; –6.880***
    - Before 1990: –1.543 ; –4.400
    - 1990–2011: –7.095*** ; –16.000***
  - Initial Log Real GDP Level:
    - Full Sample: 0.538* ; 1.520*
    - Before 1990: 0.363 ; 1.030
    - 1990–2011: 1.707*** ; 3.840***

- Openness and Integration
  - Initial Real Exchange Rate vs. U.S. Deviation:
    - Full Sample: –0.013* ; –0.038*
    - Before 1990: 0.005 ; 0.015
    - 1990–2011: –0.069*** ; –0.154***
  - Change in Real Exchange Rate vs. U.S.:
    - Full Sample: –0.021* ; –0.058*
    - Before 1990: –0.004 ; –0.010
    - 1990–2011: –0.087*** ; –0.195***
  - Initial Trade Openness:
    - Full Sample: 0.001 ; 0.003
    - Before 1990: –0.005 ; –0.015
    - 1990–2011: 0.036 ; 0.080
  - Initial Exports to EMDEs Divided by GDP:
    - Full Sample: 0.027 ; 0.075
    - Before 1990: –0.298** ; –0.851*
    - 1990–2011: 0.012 ; 0.026

- Structural Conditions
  - Initial Indicator for Constraint on Executive:
    - Full Sample: 0.063 ; 0.176
    - Before 1990: 1.470 ; 4.190
    - 1990–2011: –2.472 ; –5.560
  - Initial Life Expectancy:
    - Full Sample: 0.012 ; 0.033
    - Before 1990: 0.059 ; 0.170
    - 1990–2011: 0.044 ; 0.099
  - Initial Educational Attainment:
    - Full Sample: 0.301* ; 0.848*
    - Before 1990: 0.048 ; 0.137
    - 1990–2011: 0.903** ; 2.030*
  - Initial Real Investment Divided by GDP:
    - Full Sample: 0.066 ; 0.186
    - Before 1990: 0.160*** ; 0.456***
    - 1990–2011: 0.010 ; 0.023

- Macroeconomic Conditions
  - Change in Real Investment Divided by GDP:
    - Full Sample: 0.149*** ; 0.420***
    - Before 1990: 0.234*** ; 0.668***
    - 1990–2011: 0.177*** ; 0.397***
  - Change in Inflation:
    - Full Sample: –0.002 ; –0.006
    - Before 1990: –0.004 ; –0.012
    - 1990–2011: 0.019 ; 0.043
  - Change in Public Debt Divided by GDP:
    - Full Sample: –0.003 ; –0.009
    - Before 1990: –0.019 ; –0.055
    - 1990–2011: –0.014*** ; –0.031**

- Additional model performance and metadata (as reported)
  - Observations, Pseudo R Squared, Number of Cases, Log Likelihood, AUC, 90% Lower Bound for AUC, 90% Upper Bound for AUC, Optimal Youden Cutoff, True Positive Rate (%), False Positive Rate (%) — reported in Table 4 (values not reproduced here verbatim in the excerpt).

### Interpretation and robustness
- Univariate correlations cannot identify key causal drivers; recent takeoffs likely result from a combination of several factors interacting with global conditions.
- Some variables that differed significantly for recent takeoffs (e.g., net FDI flows, external debt, foreign reserves) could not be included in the logit due to limited data coverage.
- Robustness checks, including consideration of the rare-events problem, were performed and findings broadly hold.

*Source: IMF staff calculations (excerpt from _wp13132 Appendix 1 and related figures and Table 4).*

### Appendix 4.1. Heteroscedasticity and autocorrelation within country robust standard errors are in parentheses under the 

### _wp13132 - Appendix 4.1. Heteroscedasticity and autocorrelation within country robust standard errors are in parentheses under the

### Key regression findings on determinants of growth takeoffs
- The probability of a new growth takeoff is related to global economic environment, initial levels and changes in domestic macroeconomic conditions, and structural characteristics.
- Global factors:
  - Takeoffs in both generations were supported by higher global growth.
  - Contemporaneous World Real GDP Growth coefficients (selected estimates): 0.640*, 0.561, 1.392*, 0.788**, 1.896***, 0.509*, 0.403, 2.191* (standard errors shown in parentheses in the source tables).
- Income and size:
  - A low initial level of per capita income raises the probability of a growth takeoff, consistent with conditional convergence.
  - Initial Log Real GDP per Capita coefficients (selected estimates): –2.691***, –1.642, –7.016***, –0.010, –0.382, –1.551**, –1.445, –9.854***.
  - Initial Log Real GDP Level coefficients (selected estimates): 0.582**, 0.391, 1.687***, 0.301, 0.612*, 0.128, –0.005, 1.966**.
- Openness and external competitiveness:
  - The initial deviation of the real exchange rate from its productivity-adjusted long-run level (measure of overvaluation) and appreciation of the real exchange rate are negatively related to the probability of a takeoff.
  - Initial Real Exchange Rate vs. U.S. Deviation (selected): –0.017**, 0.006, –0.072***, –0.014*, –0.033***, –0.012*, –0.003, –0.088***.
  - Change in Real Exchange Rate vs. U.S. (selected): –0.027**, –0.004, –0.091***, –0.022**, –0.046***, –0.016, –0.017, –0.099**.
  - Initial Exports to EMDEs Divided by GDP: mixed signs; some significant negative and positive estimates (examples: 0.025, –0.321**, 0.014, –0.027, –0.031, 0.040**, –0.030, –0.054).
- Structural and human capital:
  - Takeoffs are positively correlated with the level of educational attainment, significant primarily in the last two decades.
  - Initial Educational Attainment coefficients (selected): 0.417***, 0.017, 0.882**, 0.212, 0.330, 0.144, –0.335, 0.975***.
- Investment and macroeconomic conditions:
  - Both the initial level of investment and the increase in investment during takeoff are positively correlated with the probability of a takeoff in both generations.
  - Initial Real Investment Divided by GDP (selected): 0.044, 0.170***, 0.016, 0.001, 0.050, 0.096***, 0.128***, –0.131.
  - Change in Real Investment Divided by GDP (selected): 0.145***, 0.241***, 0.181***, 0.054, 0.151**, 0.152***, 0.190***, 0.217***.
  - Change in Public Debt Divided by GDP (selected): –0.006, –0.018, –0.013**, –0.006**, –0.008**, –0.001, –0.017, –0.019*** (negative coefficients imply reductions in public debt increase takeoff probability).

### Magnitude and decomposition of the change in takeoff probabilities (2000s vs before 1990)
- Baseline predicted annual probability of a new takeoff:
  - less than 1 percent before 1990 (0.8 percent predicted probability reported for the subsample before 1990).
  - close to 3 percent during the 2000s (2.8 per cent predicted probability reported for the 2000–11 subsample).
- Overall: The chances of takeoff more than tripled during the 2000s compared with the period before 1990.
- Contributions to the percent change in the odds ratio (2000s versus before 1990) by variable group (percent change in odds ratio; variable groups correspond to Table 4 in the source):
  - Global conditions: contributed positively to the change.
  - Initial income per capita: higher initial income in the 2000s lowered the chance of takeoff (convergence effect).
  - Initial economic size: larger economic size raised the chance of takeoff (economies of scale).
  - Openness and integration (including more exports to EMDEs and a more competitive real exchange rate): contributed positively.
  - Structural conditions (notably more years of schooling): contributed most to the increase.
  - Macro-economic conditions (higher investment growth, lower debt): next most important contributors.
- Methodology notes:
  - The odds ratio is the probability of starting a takeoff divided by the probability of not starting one.
  - The estimated contribution of variables to the percent change in the predicted odds ratio is based on logistic regression coefficient estimates in Table 4 for the full sample.
  - The associated predicted probabilities at average values are 0.8 percent (before 1990) and 2.8 per cent (2000–11). To calculate the overall change, the product of contributions is used.

### Robustness checks: alternative samples, takeoff definitions, and model specifications
- Alternative LIC samples considered:
  - Baseline time-varying income threshold (LIC if average real output per capita during previous five years is below threshold), excluding conflict cases at start of takeoffs (baseline).
  - Baseline sample including LICs experiencing or recovering from conflict.
  - Time-invariant income threshold: LIC if average real output per capita over previous five years is below $2,600 in purchasing-power-parity-adjusted constant 2005 U.S. dollars (45th percentile of EMDE per capita real GDP as of 1990); this sample excludes conflict cases.
- Alternative takeoff definitions:
  - Hausmann, Pritchett, and Rodrik (2005; HPR): episode at least eight years, GDP per capita growth averages at least 3.5 percent, average growth at least 2 percentage points higher than the prior eight years, and output at end exceeds pre-takeoff peak.
  - Allowing temporary delays or breaks: two takeoffs separated by ≤ 5 years are merged into a continuation.
  - Higher growth threshold: baseline definition but growth threshold raised from 3.5 percent to 5 percent.
- Effects on episode counts:
  - HPR algorithm yields 55 growth accelerations in LIC sample: 31 during 1990–2011, and 24 prior to 1990.
  - Excluding temporary delays reduces episodes from 29 to 24 during 1990–2011 and from 41 to 31 prior to 1990s.
  - Raising cutoff to 5 percent reduces takeoffs to 17 from 29 during 1990–2011 and to 20 from 41 during 1950–89.
- Stylized facts robust across alternative definitions:
  - Dynamic LICs (both generations) experienced high investment and national saving rates relative to LICs that did not take off.
  - Current account deficits broadly similar across generations, but a larger share financed by FDI for current-generation dynamic LICs.
  - Recent LIC takeoffs: sharp decreases in inflation and public and external debt (contrasts with increases in previous generation).
  - Both generations: stronger export growth; current-generation takeoffs show more geographically diversified exports and more competitive exchange rates.
  - Dynamic LICs, especially current generation: smaller governments, better infrastructure, higher human capital than non-taking-off LICs.
- Differences under alternative criteria:
  - Using HPR methodology:
    - Income inequality is still lower in dynamic LICs than non-dynamic, but current-generation dynamic LICs do not have lower income inequality than pre-1990 dynamic LICs.
    - Current-generation dynamic LICs do not have stronger political institutions (constraints on the executive) than previous generation or non-dynamic LICs.
  - Using 5 percent threshold:
    - Recent takeoffs have lower income inequality and stronger political institutions than takeoffs prior to the 1990s, but not relative to LICs that did not take off.
    - All other stylized facts broadly similar to baseline.

### Robustness of logit estimates to alternative specifications and estimation methods
- Alternative model specifications:
  - Adding decadal dummies to baseline; results similar to baseline.
  - HPR-derived definition: similar general pattern of coefficient signs but structural conditions often statistically insignificant (possibly due to fewer HPR accelerations and concentration since 1990). HPR model not estimable for pre-1990 LIC takeoffs due to paucity of events and data.
  - Fixed income-per-capita threshold specification also shows robustness of baseline findings.
- Alternative estimation methods to account for rare events and small-sample biases:
  - Firth’s (1993) bias-reducing transformation of the log likelihood.
  - King and Zeng’s (2001) approximately unbiased coefficients procedure.
  - Complementary log-log transformation to account for skew in the dependent variable distribution.
  - Random effects logit model.
- Findings across estimators:
  - Signs and magnitudes of key coefficients broadly similar across estimation methods.
  - Examples of estimates from alternate estimators (full sample):
    - Contemporaneous World Real GDP Growth: 0.800**, 0.760**, 0.765**, 0.754**, 0.927**.
    - Initial Log Real GDP per Capita: –2.439***, –2.252***, –2.258***, –2.441***, –2.989***.
    - Initial Educational Attainment: 0.301*, 0.291**, 0.293**, 0.295*, 0.255.
    - Change in Real Investment Divided by GDP: 0.149***, 0.138***, 0.138***, 0.139***, 0.171***.
  - AUC values (full sample across methods) around 0.814–0.818, with 90% bounds reported in source tables.
- Additional checks:
  - Including conflict cases and merging takeoff episodes within five years retains baseline results, though merging reduced number of takeoffs in the logit sample from 29 to 17 (interpret with caution).

### Historical case lessons (selected): Brazil and Korea, 1960–80
- Commonalities and contrasts:
  - Both pursued industrialization and experienced strong growth between 1960 and 1980; post-1980 trajectories diverged sharply.
  - Brazil: output per capita stagnated for more than two decades after a debt crisis in the early 1980s.
  - Korea: after a recession in 1980, regained momentum and continued growth.
- Growth strategies:
  - Brazil:
    - Inward-oriented growth model focused on large domestic market.
    - Import substitution central to strategy (discouraged imports, subsidized domestic producers).
    - Growth driven mainly by domestic demand; slow export growth.
  - Korea:
    - Shift away from import-substitution beginning in the 1960s toward export orientation.
    - Initially promoted labor-intensive exports; later promoted higher-value-added industries.
    - Large-scale investment in shipbuilding, steel, and petrochemicals led to global leadership in these sectors.
- Financing and macroeconomic management:
  - Brazil:
    - High national saving rate did not keep pace with investment.
    - Rising current account deficits increasingly financed by external borrowing.
    - Public debt rose beginning in the 1970s.
    - Overheating pressures intensified when stimulus policies were not adjusted after first oil shock; second oil shock and higher world interest rates contributed to debt crisis.
  - Korea:
    - Large current account deficit until early 1970s financed by foreign aid and external borrowing.
    - Rapidly growing saving rates over time; budget deficit stayed relatively low.
    - Policies encouraged personal saving (mandatory long-term saving for civil servants and other employees) and corporate saving (mandated low dividends).
    - Narrowed current account deficit in the 1970s; rose after second oil shock but fell soon after due to strong export growth.
    - Fiscal discipline and strict monetary targeting helped keep inflation under control.
- Competitiveness and productivity policies:
  - Korea maintained a relatively depreciated real exchange rate (step devaluations within an implicit crawling peg) to preserve external competitiveness.
  - Exporters received incentives; labor skills in key sectors were upgraded.
  - These policies supported sustained investment productivity and improved external balances.

*Source: IMF staff calculations.*

### 2. Real Invest ment

### 2. Real Investment

### Overview
- Figures compare long-run experiences of economies using indicators defined in Appendix 1 and sourced to Abbas and others (2010); Barro and Lee (2010); IMF databases; Lane and Milesi-Ferretti (2007) updated to 2011; Penn World Table 7.1; World Bank WDI (2012); and IMF staff calculations.
- Note: Public debt data for Brazil are missing from 1962 to 1969, and for Korea for 1970.

### Brazil and Korea: divergent human-capital and inequality dynamics during 1960–90
- Korea prioritized vocational and in-plant training early, later strengthening engineering education and establishing specialized research institutes, maintaining relatively low income inequality after takeoff.
- Brazil emphasized general education during early labor-intensive industrial policies but experienced slow educational advancement and persistently high income inequality.
- Figure reference: panel 8 shows educational attainment trends across the period.

### Indonesia, mid-1960s to present — growth with shared prosperity
- Structural shift: growth initially led by the energy sector until the early 1980s and increasingly by manufacturing afterward (Figure 12, panels 1 and 2).
- Use of commodity windfalls:
  - Large share of government revenue from commodity windfall gains directed to public investment in rural infrastructure, agriculture, health, and education in the 1960s–1970s.
  - Oil boom windfalls helped fund infrastructure and human-capital investments that supported inclusive growth.
- Policy and reforms:
  - After the oil boom ended in the early 1980s, government policies supported a shift toward manufacturing through industrial deregulation and trade, capital account, and financial liberalization.
  - Agricultural productivity improvements included adoption of high-yield seeds, increased use of fertilizers and irrigation (Green Revolution technologies).
- Macroeconomic outcomes and risks:
  - Monetary targets reduced inflation from triple digits in the mid-1960s to less than 15 percent by the end of that decade.
  - Fiscal targets in the late 1970s kept public debt relatively low.
  - Financial deregulation without adequate prudential regulation triggered a credit boom centered in the property sector beginning in the 1980s, financed by short-term capital flows with a pegged exchange rate regime.
  - Crisis and rebound: economy hit by contagion in 1998, experienced banking and balance of payments crisis, and rebounded in 2000 on stronger macroeconomic policies and structural reforms.
- Growth performance:
  - Annual growth in real GDP output per capita averaged 3¾ percent in the 2000s.
  - Growth was accompanied by sharp declines in poverty levels and relatively low income inequality (Figure 12, panel 4).

### Mozambique, 1990s to present — resource-driven investment and unfinished agenda
- Growth drivers and financing:
  - Peace and political stability after 1992 supported rebound; annual growth in per capita real GDP averaged 5¾ percent over the subsequent 16 years.
  - Investment surge supported by aid-financed reconstruction initially, then by public-private infrastructure to develop the resource sector; government attracted FDI to finance public-private projects.
  - Mozambique qualified for debt relief under the Heavily Indebted Poor Country Initiative and Multilateral Debt Relief Initiative, freeing fiscal space for government contributions to infrastructure projects.
- Policy reforms to attract investment:
  - Measures included establishing a one-stop investment center, improving investor property rights and contract enforcement, providing generous tax incentives, and creating “development corridors” to form industrial clusters linked to ports.
- Outcomes and challenges:
  - Growth has been capital intensive and resource-focused, generating limited employment and limited fiscal gains due to tax exemptions for projects.
  - Only modest declines in poverty and income inequality and slow improvement in health and education despite donor support; Mozambique ranks among the poorest performers in UNDP’s Human Development Report.
  - Risks: reduced vulnerabilities from external borrowing but elevated risk of Dutch disease effects.
- Policy priorities recommended:
  - Develop transport and energy infrastructure.
  - Continue to enhance human capital.
  - Ensure broader access to financing to attract domestic private investment.
  - Expand use of agricultural land to enhance agricultural productivity.

### Cambodia, 1990s to present — rapid growth with further structural work needed
- Growth and investment patterns:
  - Real GDP per capita gained momentum in the mid-1990s; output per capita grew at an average annual rate of 6 percent over the past decade (Figure 14, panel 1).
  - Growth supported by rising investment related to the export-oriented textile industry and more recently in infrastructure.
  - Textile sector accounts for three-quarters of total exports of goods; tourism and agricultural products follow.
- External financing and FDI:
  - Cambodia has relied heavily on FDI to finance its saving-investment gap; recent FDI harnessed into public-private initiatives to improve power generation.
  - Preferential market access historically under the Multi-Fiber Arrangement (MFA) catalyzed takeoff; MFA ended in 2005 but preferential access to the European Union continued.
- Structural constraints and financial trends:
  - Investment growth decelerated in the early 2000s due to regulatory burden concerns; recent government efforts improved the business climate (Doing Business rank moved up by eight places in 2012 to 133rd out of 185).
  - Credit intermediation deepening: the credit-to-GDP ratio has quadrupled to 35 percent in less than 10 years and continues to rise.
- Policy recommendations to sustain growth:
  - Remove infrastructure bottlenecks and continue improving the business climate to attract private investment and diversification.
  - Deepen financial intermediation while maintaining financial stability through strong prudential supervision and regulation.
  - Improve public debt management to lower risks from contingent fiscal liabilities related to public-private initiatives.
  - Mobilize fiscal revenue to build buffers for development needs, including human capital development through improved health and education (Figure 14, panels 5 and 6).

*Source: _wp13132 - 2. Real Invest ment (IMF staff calculations and cited datasets).*

### 6. Educat ion and Healt h

### 6. Education and Health

### E. Takeaways from the Case Studies
- Growth takeoffs are feasible under a variety of development strategies; growth was strong in all five case-study economies despite different economic structures and strategies.
- Country strategies observed:
  - Cambodia, Indonesia, Korea, and Mozambique: promoted growth through investment and exports.
  - Brazil: investment geared toward the domestic market.
- Degree of government involvement varied:
  - Mozambique and Cambodia in the 1990s: government focused on maintaining political stability in the postwar era and developing an investment-friendly environment.
  - Brazil and Korea in the 1960s: heavier public sector involvement with varying macroeconomic effects.
- Key lessons:
  - Sustaining strong growth requires continued effort to reduce external and internal imbalances.
    - For all five economies, the growth takeoff was accompanied by some narrowing of fiscal and external current account deficits, but not all sustained this momentum.
    - Where imbalances grew or growth relied excessively on foreign borrowing, takeoffs ended disruptively or were interrupted (Brazil in 1982, Indonesia in 1997).
    - Recommendation: today’s dynamic LICs, now only 9 to 12 years into their takeoffs, should avoid financing investment by excessive debt.
  - Structural reforms are instrumental in raising productivity and ensuring broad-based growth.
    - Korea: labor training in export-oriented sectors helped move manufacturing up the value chain.
    - Korea and Indonesia in the 1960s: measures to upgrade agricultural productivity, infrastructure, and human capital raised living standards broadly.
    - Brazil in the 1960s: growth from infrastructure projects and import substitution did not alleviate income inequality.
    - Mozambique: capital-intensive growth with limited employment generation may increase social vulnerabilities unless emphasis continues on improving productivity, education, and health.
    - Mozambique’s FDI-financed growth strategy produces less debt but could produce Dutch disease challenges as the economy broadens its growth strategy.
  - Policies must adjust to changing global conditions.
    - Strong global growth, low interest rates, and terms-of-trade gains (or preferential access to larger markets) benefited all five economies at different times.
    - Indonesia’s timely shift from natural resources helped maintain strong growth after the end of the oil price boom in the 1980s, underscoring the significance of further economic diversification for many of today’s dynamic LICs.
    - Brazil’s struggle to adjust domestic demand to the oil price shocks of the 1970s exacerbated its external imbalances.
    - Important lesson for today’s LICs: avoid procyclical policies despite the prevalence of ultralow global interest rates.

### V. Conclusions — Stylized Facts and Policy Implications
- Scope of analysis:
  - Examined episodes of growth takeoffs in nearly 70 developing economies or low income countries (LICs) over the past six decades.
  - Identified a second wave of LIC takeoffs since the 1990s.
- Historical context:
  - First wave of takeoffs in the 1960s and early 1970s.
  - Takeoffs fell in the late 1970s and 1980s as global conditions turned worse, but have rebounded in the past two decades.
  - The share of LICs that sustained their takeoffs increased sharply in the past two decades, with more than half of recent takeoffs, or one-third of all LICs, continuing to expand through 2011.
- Duration of takeoffs:
  - Recent takeoffs have lasted 9 to 12 years on average.
  - Takeoffs prior to the 1990s typically lasted only 7 years.
- Payoff from igniting a takeoff:
  - Defined as an expansion in per capita output that lasts at least five years, with average growth during the period of at least 3½ percent.
  - Even just igniting a takeoff pays off over the long run: a 50 to 60 percent rise in LICs per capita output in the 10 years after the start of a takeoff, compared to only 5 to 15 percent for LICs that did not take off.
- Common patterns across takeoffs:
  - Both recent takeoffs and takeoffs before 1990 were based on higher investment rates and greater trade integration.
  - Export growth rose faster in dynamic LICs than in LICs that were unable to take off, and it was higher in recent takeoffs than in earlier ones.
- Lower vulnerabilities in current-generation takeoffs:
  - Current-generation takeoffs associated with declines in debt and inflation after takeoff, compared to increases in these imbalances in previous-generation takeoffs.
  - Partly related to greater reliance on FDI-financed investment rather than debt-financed investment for the current generation.
- Other distinguishing economic conditions for recent takeoffs:
  - More competitive exchange rates.
  - Deeper links with emerging market and developing economies in trade.
  - Faster-paced implementation of structural reforms and institution building: lower regulatory burdens, better infrastructure, higher education levels, and greater political stability.
- Determinants of takeoff probability (controlling for global conditions):
  - Probability of takeoffs rises with higher global growth, initial size, years of schooling, and the initial level and increases in investment.
  - Probability falls with initial income per capita, an appreciation and an overvaluation of the real exchange rate, and an increase in the public debt.
  - Overall, the chances of starting a takeoff have tripled in the 2000s compared to the period before the 1990s, with domestic conditions and policies accounting for a majority of the rise.
- Caveats and remaining challenges:
  - Income per capita in LICs typically still a fraction of that in advanced economies; long journey before convergence.
  - Many LICs experience growth in a few sectors only and remain vulnerable unless they further diversify.
  - Greater reliance on FDI flows—while reducing vulnerabilities from debt-financed investment—could raise Dutch disease problems in the future.
  - Ensuring that growth results in broad-based improvements in living standards remains a challenge.
  - Case studies remind that takeoffs can fall apart even after many years of strong growth.
- Policy priorities for today’s dynamic developing economies:
  - Continue with reforms.
  - Avoid major macroeconomic imbalances.
  - Maintain external competitiveness.

*Source: _wp13132 - 6. Education and Health*

### APPENDIX 1. DATA DEFINITIONS AND SOURCES AND COUNTRY GROUPINGS

### APPENDIX 1. DATA DEFINITIONS AND SOURCES AND COUNTRY GROUPINGS

### Data definitions and primary sources
- Primary data sources: IMF’s World Economic Outlook (WEO); Penn World Table version 7.1 (PWT; Heston, Summers, and Aten, 2012); World Bank’s World Development Indicators (WDI).
- For indicators with multiple sources, splicing order is followed (extend level of primary series using growth rate of secondary series).
- Example: aggregate real GDP and real GDP per capita in constant 2005 purchasing-power-parity U.S. dollars are from the PWT, and where missing, are extended with data from the WEO and WDI.

### Domestic shocks
- Bank, currency, and debt crises: Laeven and Valencia (2012).
- Conflict: indicates involvement in a serious internal or external conflict in a given year in which output per capita falls by more than 3 percent; derived from The New COW War Data, 1816–2007 v4.0 and real output per capita series.
- Low-income country (LIC) episodes of strong or weak growth are excluded if they occur in the year after a conflict to avoid confounding a growth takeoff with a simple bounce back from a war.

### Economic structure and sectoral measures
- Export concentration: Papageorgiou and Spatafora (2012); corresponds to Theil index on updated UN-NBER harmonized COMTRADE data at four-digit SITC (Rev. 1).
- Exports to emerging and developing economies: IMF Direction of Trade Statistics; expressed as a percent of nominal GDP in U.S. dollars from the WDI, extended with the WEO.
- National saving to GDP: share of real gross national product in real GDP from the WDI minus share of private and public consumption in real GDP from the PWT.
- Real exports to GDP: real exports of goods and services as percent of GDP, from the WDI, extended with the WEO.
- Real investment (percent of GDP): PWT.
- Real share of manufacturing and real share of resources in value added: WDI. Resources = contribution of industry in value added minus contribution of manufacturing in value added. Total value added = agriculture + industry + services.
- Textile exports (percent of goods exports): United Nations Comtrade Statistics database.

### Key indicator data sources (Table 7 — selected entries)
- Global Growth (percent): IMF, World Economic Outlook Database (2012); Penn World Table 7.1 (2012).
- U.S. Real Interest Rate (three-month treasury-bill rate minus realized inflation rate; annualized percent): Haver Analytics.
- Aid Flows (millions of current U.S. dollars): World Bank, World Development Indicators Database (2012).
- Bank Crises / Currency Crises / Debt Crises: Laeven and Valencia (2012).
- Conflict: The New COW War Data, 1816–2007 v. 4.0 (2011).
- Current Account Balance (percent of GDP): WDI (2012); WEO (2012).
- Credit (percent of GDP): IMF, International Financial Statistics.
- Educational Attainment (years of schooling): Barro and Lee (2010).
- Export Concentration: Papageorgiou and Spatafora (2012).
- External Debt (percent of GDP): Lane and Milesi-Ferretti (2007) updated to 2011.
- Inflation (percent): WDI (2012); WEO (2012).
- Life Expectancy (years): WDI (2012).
- Net FDI Flows (percent of GDP): IMF, Balance of Payments Statistics Database; WEO (2012).
- Real GDP and Real GDP per Capita (PPP, 2005 U.S. dollars): PWT 7.1 (2012); WEO (2012); WDI (2012).
- Real Investment (percent of GDP): PWT 7.1 (2012).
- Trade Openness: WDI (2012); WEO (2012).
- Trade-Weighted Terms-of-Trade Growth (percent): WDI (2012); WEO (2012).
(Note: table lists many additional indicators and their data sources.)

### External policies
- Aid flows: WDI; real aid flows obtained by deflating by the U.S. consumer price index.
- Current account balance (percent of GDP): WDI, extended with WEO.
- Foreign reserves to GDP: External Wealth of Nations Mark II Database (Lane and Milesi-Ferretti, 2007).
- Net FDI Flows (percent of GDP): IMF Balance of Payments Statistics Database (line 4500).
- Trade openness: (imports + exports of goods and services) / GDP; components from WDI, extended with WEO.

### Global environment
- Global growth: world GDP growth aggregate from the WEO, weighted by PPP GDP, extended by growth of aggregate GDP PPP levels from the PWT.
- U.S. real interest rate: U.S. three-month Treasury bill rate (secondary market, annual average) minus realized U.S. inflation rate; both series from Haver Analytics; expressed in annualized percent.

### International relative prices
- Real exchange rate: PWT; defined as price level of GDP versus that of the United States.
- Real exchange rate deviation: residual from linear regression of log real exchange rate on productivity differential proxied by difference in log real GDP per capita with the United States.
- Real exchange rate change: percent change over a five year period in the five-year average of the real exchange rate.
- Trade-weighted terms of trade: percent change of terms-of-trade index constructed using deflators of exports and imports and nominal series for GDP, exports, imports — all from WDI and WEO. Terms-of-trade index = export price deflator^(share of exports in GDP) / import price deflator^(share of imports in GDP).

### Monetary and fiscal policies
- Credit (percent of GDP): IMF International Financial Statistics; refers to bank credit to the private sector (line 22D).
- External debt to GDP: External Wealth of Nations Mark II Database (Lane and Milesi-Ferretti, 2007).
- Inflation: log difference of the consumer price index (CPI); CPI from WDI, extended with WEO.
- Public debt (percent of GDP): Abbas and others (2010); Lane and Milesi-Ferretti (2007) updated to 2011; public-debt-to-GDP ratio taken as ratio to GDP (GDP from WDI, extended with WEO) and extended using change in external debt to GDP.

### Structural and political conditions
- Constraints on the executive: Political Regime Characteristics and Transitions Database (2011), rescaled to zero to 1 (from 1 to 7): unlimited authority = zero; executive parity = 1.
- Educational attainment: years of schooling from Barro and Lee (2010).
- Income inequality: Gini coefficient of household disposable income from Solt (2009).
- Life expectancy at birth (years): WDI.
- Poverty headcount: percent of population living on $2 a day in PPP terms (WDI).
- Regulatory barriers and size of government: Economic Freedom of the World 2012 Annual Report (Gwartney, Lawson, and Hall, 2012); original indices 0 to 10 with 10 indicating most freedom; transformed (10 minus original values) so higher scores indicate more restraints and larger size, respectively.
- Missing data for poverty headcount, regulatory restraints, and size of government: intervening years linearly interpolated.
- Telephones per capita: Banks and Wilson Cross-National Time-Series Data Archive (2012); units = telephones per thousand people.

### Transformations for the logistic regression
- Three variable forms used:
  - initial — once-lagged, backward-looking five-year average (average behavior in five years before potential takeoff);
  - contemporaneous — current year, forward-looking five-year average (average behavior in first five years of a potential takeoff);
  - change — difference between contemporaneous and initial values (captures average trajectory during first years of potential takeoff).
- Moving average calculated only if there are at least 2 nonmissing observations for the indicated variable during the window.

### Country groups and sample restrictions
- Advanced economies (AEs): member economies of the Organization for Economic Cooperation and Development before 1990, with the exception of Turkey.
- Other economies classified as emerging market and developing economies (EMDEs).
- Low-income countries (LICs): economies in which output per capita, averaged over the previous five years, is lower than the time-varying low-income threshold.
  - Low-income output per capita threshold represents bottom 45th percentile of EMDEs’ output per capita in 1990 ($2,600 in 2005 U.S. dollar PPP terms).
  - Threshold spliced back for pre-1990 and forward for post-1990 using average growth rate of global output per capita during 1950–2011 (about 2.3 percent per year).
- Sample restricted to economies with average population at least 1 million but no more than 500 million (excludes China and India from LICs).
- For bar charts comparing cases and referents from Figure 3 onward, a constant composition sample underlies each panel to ensure comparability within group of cases or referents across time.

### Classification of country episodes by economic structure
- Sector shares constructed using WDI sectoral value added in local currency at constant prices.
- Nonmanufacturing industry treated as proxy for resources (includes mining and quarrying, construction, utilities).
- Procedure:
  - Compute shares of agriculture, manufacturing, resources, other in total value added.
  - Calculate 10-year average of these shares from start of growth episode (or first year for valid referent).
  - Predominantly agricultural if 10-year average agriculture share is in 70th percentile for whole sample of country episodes between 1960 and 2011.
  - Manufacturing oriented if 10-year average manufacturing share is in 70th percentile for 1960–2011.
  - Resource rich if 10-year average resources share is in 70th percentile for 1960–2011.
  - “Other” includes episodes not classified as agricultural, manufacturing oriented, or resource rich.
- For episodes with insufficient data, classification complemented with WDI data on resource rents (resource rents = value of production at world prices minus total costs of production for oil, natural gas, coal, minerals, forestry; calculated at current prices).
  - An episode classified as resource rich if 10-year average resource rents as percent of GDP are in 70th percentile (1960–2011).
  - An episode classified as manufacturing oriented if 10-year average industry sector value-added share is in 70th percentile and 10-year average resource rents as percent of GDP are not in 70th percentile (1960–2011).

*Source: APPENDIX 1. DATA DEFINITIONS AND SOURCES AND COUNTRY GROUPINGS*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13132.pdf_