## 1. Determinants of Private Saving in Previous Studies on LAC

## Source details

**Canonical URL:** [1. Determinants of Private Saving in Previous Studies on LAC](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15108.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15108.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15108.pdf.json)

---

### I. Introduction — research question, motivations, contributions
- Research question: What policies are most effective in raising saving, especially in LAC?
- Motivations:
  - LAC exhibits much lower private saving rates than the average for the world, with heterogeneity: particularly low saving in smaller countries and high volatility in countries exposed to exogenous shocks (natural disasters, commodity price fluctuations).
  - Importance of terms of trade (TOT) and commodity-price shocks for LAC saving (cited Adler and Magud, 2013; Gruss, 2014).
  - Institutional and macroeconomic reforms in the last three decades (adoption of fiscal rules, inflation targeting, floating exchange rates) — unclear extent of effect on private saving.
- Contributions:
  - Extensive review of previous empirical research on private saving in LAC, summarizing results and identifying contradictions and omissions.
  - Empirical search of determinants of private and public saving rates adding variables: different measures of TOT and commodity-price shocks; macroeconomic policy regimes including fiscal rules, monetary regimes, exchange-rate regimes; linear and interaction specifications.
  - Analysis of statistical differences in saving determinants between LAC and the rest of the world in a nested econometric framework, and differences within three LAC subgroups.

### II. Stylized findings (high level summary)
- Baseline private saving estimation:
  - Generally robust across regional groups; coefficients for the LAC region do not exhibit much difference from the rest of the world.
  - Private saving fundamentally explained by: its own lag, real GPDI per capita, the TOT, share of urban population, and old-age dependency ratio.
- Public saving:
  - Increased income growth reduces public saving (possible procyclical fiscal policy).
  - Increased private sector credit raises public saving (suggesting crowding out).
  - Results differ across LAC subgroups: larger countries show signs of countercyclical fiscal policy and less sensitivity to old-age dependency.
- Commodity-price shocks:
  - TOT and commodity price volatility and accelerations, and their temporary components, raise private saving in both LAC and the rest of the world.
- Policy regimes:
  - Adoption of fiscal rules and floating exchange rate regimes do not affect private saving.
  - Adoption of inflation targeting reduces private saving irrespective of the negative effects of inflation on saving.

### III. Previous empirical literature on LAC — stylized findings and gaps
- Literature characteristics:
  - Study types: macroaggregate models (18 papers), micro survey studies (11 papers), single-equation estimations, simulation exercises, some general-equilibrium or multi-equation sector models.
  - Sample sizes vary widely: macro single-country 20 to 624 observations; panel studies 99 to 128 observations; micro studies 1,762 to 392,263 observations.
- Common determinant categories and empirical patterns (across 26 LAC studies):
  - Income level and income growth: empirical studies more often report positive effects on private saving but with variation.
  - TOT: some studies report positive effects.
  - Demographics: age dependency ratios tend to reduce private saving in aggregate studies.
  - Financial access and credit: mixed results; micro evidence often finds greater access → lower precautionary saving.
  - Public saving/fiscal policy: 16 macro studies show higher public saving reduces private saving; offset coefficients range from 0 to 0.75, most estimates significantly different from both zero and one.
  - Pension reforms (transition to fully-funded systems): five country studies find increases in national saving since reform start by: 2.3-2.7 percentage points in Chile, 1.5 pp in Colombia, 1.0 pp in Mexico, and 0.3 pp in Peru.
- Gaps in prior LAC macro studies:
  - Often absent: terms-of-trade and commodity-price shocks; policy regimes (fiscal rules, monetary regimes, exchange-rate regimes); explicit comparisons between LAC and other regions and within LAC subgroups.
  - Most studies include only a small number of potential determinants, rely on mostly linear specifications, and rarely include interaction effects.

### IV. Data and stylized facts (dataset, measures, regional splits)
- Dataset:
  - Based on Grigoli et al. (2014) dataset with 4,137 observations spanning 32 years for 165 countries (slight revisions cause loss of 12 observations vs Grigoli et al. (2014)).
  - Private saving rate = ratio of gross private saving to gross private disposable income (GPDI).
  - Public saving rate = ratio of gross public saving to GPDI.
- Commodity price measures:
  - Country-specific net commodity price index (NCPI) from Gruss (2014); monthly data converted to annual averages.
  - Permanent and temporary components via HP filter on log index with lambda = 6.25.
- Policy regime dummies:
  - Fiscal rules from IMF Fiscal Rules Dataset; inflation targeting from IMF Finance & Development; exchange rate regime from IMF Annual Report on Exchange Arrangements and Exchange Restrictions.
  - Dummy = 1 if country has fiscal rule (expenditure, budget balance, debt, or any), inflation targeting, or de jure flexible exchange rate regime; 0 otherwise.
- Regional/subgroup classification:
  - Regions per IMF regional departments. LAC split into: LAC7 (seven largest LAC economies), Middle LAC (other South American countries, Central America, Dominican Republic), and Small LAC (small Caribbean economies, Belize, Guyana).
- Stylized statistics (unbalanced panel 164 countries, 1981–2012) — private saving means (percent of GPDI):
  - World: 20.4
  - Africa: 13.4
  - Asia and Pacific: 26.7
  - Europe: 24.4
  - Middle East and Central Asia: 26.1
  - Western Hemisphere: 16.4
  - LAC (regional aggregate): 15.6
  - LAC7: 20.0
  - Middle LAC: 15.1
  - Small LAC: 13.1
- Volatility and persistence:
  - LAC standard deviation and coefficient of variation are below world averages, but Small LAC has volatility comparable to Africa.
  - Autocorrelation (AC1) is high across regions; example AC1 for LAC7 = 0.71.
- Notable trends and country patterns (1981–2012):
  - Global private saving ~20 percent of GPDI stable; LAC ~4.5 pp lower on average.
  - LAC7 saving remained strong post-Global Financial Crisis, exceeding global average by about 2.4 percent since 2006.
  - Example: Small LAC private saving fell from 8.1 to 3.3 percent of GPDI in 2008, rebounding to 9.5 in 2009.
  - Public saving ranged roughly 3.5-7 percent of GPDI; surged mid-2000s (4.4 percent in 2002 → 10.3 percent in 2006), then fell during the Global Financial Crisis and rose by 2012.
  - Country extremes: Venezuela public saving nearly 44 percent of GPDI in 2005, ~15 percent in 2012; St. Vincent and the Grenadines private saving hit -19.3 percent of GPDI in 2008 and remained below -10 percent of GPDI by 2012.
- Household vs corporate saving:
  - Household saving ~5–8 percent of GPDI in Chile, Colombia, Dominican Republic; corporate saving generally higher and more volatile (exceptions Bolivia and Mexico).

### V. Empirical methodology
- Dynamic model estimated with two-step system GMM (S-GMM) following Arellano and Bond (1997), Blundell and Bond (1998), with Windmeijer (2005) finite-sample correction to two-step covariance matrix.
- Baseline private saving specification includes:
  - Endogenous/predetermined covariates: log real per capita GPDI (PPP), real growth rate of per capita GPDI (PPP), public saving/GPDI, inflation, real deposit rate, flow of private sector credit/GPDI; lagged private saving.
  - Exogenous covariates: log TOT, old-age dependency ratio, share of urban population.
- Nested specifications: interactions between regressors and region/subgroup dummies to test differential effects.
- Diagnostics: AR(1), AR(2) tests for serial correlation; Hansen J-test for instrument validity; instruments collapsed; Windmeijer correction applied.

### VI. Private saving — core econometric findings (baseline coefficients, world sample Table 3, Column 1)
- Statistically significant baseline determinants (coefficient (std. err.)):
  - Lagged private saving: 0.593*** (0.066)
  - Ln real per capita GPDI (PPP): 0.171*** (0.032)
  - Real growth rate of per capita GPDI (PPP): 0.255*** (0.040)
  - Ln TOT: 0.052*** (0.015)
  - Public saving/GPDI: -0.253*** (0.064)
  - Inflation: 0.388*** (0.124)
  - Flow of private sector credit/GPDI: -0.082** (0.039)
  - Old-age dependency ratio: -1.129*** (0.242)
  - Share of urban population: -0.392*** (0.099)
- Regional differential findings (nested interactions):
  - Private saving less persistent in Africa (negative interaction on lagged dependent variable).
  - Sensitivity to income level lower in Europe and Middle East and Central Asia: a one pp increase in income level raises private saving by about 0.06 pp, ~half the effect in other economies.
  - Reduced sensitivity of private saving to income growth in poorer countries (Africa); Asia and Pacific shows larger sensitivity to income growth.
  - Old-age dependency ratio particularly strong negative in Europe.
  - Negative impact of urban population share on private saving considerably reduced in Europe, Middle East and Central Asia, and LAC; in LAC a one pp increase in share of urban population associated with reduction in saving of only 0.01 pp (interaction described qualitatively).
- LAC subgroup analysis:
  - No significant differential effects between LAC subgroups and rest of world for baseline regressors, except inflation in Middle LAC (interaction reduces private saving by 0.09 pp).
  - Conclusion: LAC subgroups share sensitivity to fundamental private saving determinants comparable to the rest of the world.

### VII. Contributions to fitted private saving (selected quantitative contributions, percent of GPDI)
- Lag private saving contribution:
  - LAC = 9.3; LAC7 = 11.8; Middle LAC = 9.0; Small LAC = 7.1.
- Ln real GPDI per capita (PPP) contribution:
  - Europe = 43.2; LAC = 29.3; Small LAC = 35.8; LAC7 = 31.5; Middle LAC = 22.9.
- Ln TOT contribution:
  - Range ~23.8 (Middle LAC) to 24.4 (LAC7).
- Share of urban population (negative contributions):
  - Europe = -27.6; LAC = -23.6; LAC7 = -31.6; Small LAC = -16.7.
- Old-age dependency ratio (negative contributions):
  - Europe = -24.0; Asia and Pacific = -12.1; LAC = -11.2; Middle LAC = -10.5; Small LAC = -12.5.
- Other determinants with smaller contributions: growth rate of real GPDI (PPP), inflation, public saving, flow of credit to private sector.

### VIII. Public saving — core econometric findings (Table 5)
- Baseline public saving restricted specification (Table 5, Column 2) — selected coefficients (coefficient (std. err.)):
  - Lagged public saving: 0.295*** (0.110)
  - Ln real per capita GPDI (PPP): 0.222** (0.089)
  - Real growth rate of per capita GPDI (PPP): -0.201*** (0.053)
  - Ln TOT: 0.077*** (0.030)
  - Private saving/GPDI: -0.477*** (0.147)
  - Flow of private sector credit/GPDI: 0.150*** (0.057)
  - Old-age dependency ratio: -2.699*** (0.988)
- Key interpretations:
  - Growth in real GPDI per capita reduces public saving by 0.14 pp in full specification (Column 1) → possible procyclical fiscal policy.
  - Increased private sector credit associated with higher public saving (0.06 pp in full specification) → possible crowding out.
  - Private-public saving interaction: one pp increase in private saving associated with 0.24 pp fall in public saving (Column 1) → offset behavior.
- LAC subgroup differences:
  - LAC7: persistence of public saving > 1; income growth positively impacts public saving (one pp increase in growth raises public saving by 0.41 pp) → countercyclical fiscal policy in larger LAC economies.
  - Middle LAC: one pp increase in real GPDI per capita (PPP) has virtually no effect on public saving → fiscal acyclicality in this subgroup.

### IX. Terms-of-Trade and commodity-price shocks — empirical results (Table 6; projections)
- Tests:
  - Role of level, growth rate, volatility (3-year rolling SD), and temporary/permanent components of TOT and NCPI on private saving; interactions with LAC dummy.
- Main findings:
  - Growth/accelerations:
    - Growth rate of TOT: one pp increase raises private saving by 0.058*** (0.018).
    - Growth rate of NCPI: one pp increase raises private saving by 0.029** (0.013).
    - No differential LAC effect found.
  - Volatility:
    - SD of ln TOT (3-year rolling SD): coefficient 0.113** (0.053).
    - SD of ln NCPI: 0.143** (0.061).
    - No differential LAC effects detected.
  - Temporary vs permanent components (HP filter):
    - Temporary component of TOT: 0.098*** (0.024).
    - Temporary component of NCPI: 0.056*** (0.020).
    - Permanent components not significant.
    - No differential LAC effects detected.
- Policy-relevant projection exercise (WEO January 19, 2015 TOT projections to 2019 using Table 3 Col 1 and Table 5 Col 2 coefficients):
  - On average TOT projected to improve by 0.9 percent by 2019 in LAC (heterogeneity: Grenada +16.8 percent; Venezuela -20.8 percent).
  - Estimated change in private and public saving rates (product of first difference of ln TOT between 2012 and 2019 and TOT coefficients):
    - LAC region average: mild increase in both private and public saving roughly 0.1 percent of GPDI.
    - By subgroup:
      - LAC7: private saving projected to fall by 0.4 percent of GPDI; public saving projected to fall by 0.5 percent of GPDI.
      - Middle LAC: private saving projected to increase by 0.1 percent of GPDI; public saving by 0.2 percent of GPDI.
      - Small LAC: private saving projected to increase by 0.3 percent of GPDI; public saving by 0.5 percent of GPDI.
    - Grenada: largest projected increase (private saving +0.9 percent of GPDI; public saving +1.3 percent of GPDI).
    - LAC oil exporters expected largest fall — 0.6 percent of GPDI in private saving.
  - If projected TOT changes were temporary, final effect on private saving would be larger, "ranging from -2.04 percent for Venezuela and 1.65 percent for Grenada."

### X. Policy frameworks — prevalence and empirical tests (Table 7)
- Prevalence as of end-2012:
  - "75 countries have some category of fiscal rule in place (14 in LAC)"
  - "25 have adopted an inflation targeting regime (6 in LAC)"
  - "61 run flexible exchange rate regimes (7 in LAC)"
- Empirical tests: linear effects of policy regimes and interactions with selected variables, controlling for baseline specification.
- Main empirical findings on private saving:
  - Fiscal rules:
    - Tested expenditure rules, budget rules, debt rules, general fiscal rule dummy.
    - "None of the results are significant at five percent" → adoption of fiscal rules not reflected in changes of private saving.
  - Inflation targeting:
    - "Inflation targeters have lower private saving by 2.66 pp — a worldwide result shared by LAC."
    - Table coefficient excerpt: "Inflation targeting........-2.662**-1.834.."
    - Interaction: "Inflation targeting * Inflation.........-0.231*.."
  - Floating exchange rate regimes:
    - Aggregate effect ambiguous; empirical coefficients not significant: "Floating ER regime..........-1.303-1.118"
- Estimation diagnostics (selected reported values from Table 7):
  - AR(1) p-val.: "0.000" (all columns)
  - AR(2) p-val.: values reported include "0.3430.3380.3470.2940.3360.3060.3390.2890.3440.3420.3910.385"
  - Hansen J-test p-val.: values include "0.9180.9450.8860.4760.9160.2630.9080.4180.9430.9630.5160.359"
  - Observations reported: "3,2433,2433,2433,2433,2433,2433,2433,2433,2433,2432,7912,791"
  - Number of countries: "153153153153153153153153153153153153"

### XI. Conclusions and policy implications
- Robust empirical patterns:
  - Income level, income growth, and TOT raise private saving.
  - Age dependency ratios tend to reduce private saving rates.
  - Household structure variables: formal head and number of income-earning members increase private saving; number of under-age members decrease it.
  - Risk aversion and the discount factor raise saving.
  - Financial access: micro evidence consistent with negative sign (greater access reduces precautionary saving).
  - Pension reforms toward fully-funded regimes raise national saving; mandatory pension saving is partly offset by lower voluntary household saving.
- Policy-variable patterns and implications for LAC:
  - Fiscal consolidation reduces private saving; government pension payments and transfers reduce household/private saving.
  - Taxes on firms reduce corporate saving; personal income tax lowers private saving.
  - Interest rate ambiguous; inflation positively affects private saving.
  - Demographic trends (aging, urbanization) exert substantial negative contributions to private saving; policies addressing long-term pension burdens and demographic adaptation are central.
  - Commodity price shocks: temporary and volatile increases in commodity prices are saved — precautionary and temporary-income saving behavior observed; projected commodity price shifts imply heterogeneous impacts across LAC countries and subgroups (oil exporters vulnerable to declines in saving ratios if TOT deteriorate).
  - Fiscal behavior matters for public saving dynamics: growth episodes associated with lower public saving in some contexts; larger LAC economies show countercyclical policy features suggesting scope for fiscal frameworks that support saving stabilization.
  - Financial deepening and private credit flows: mixed effects on private saving but associated with higher public saving (crowding-out interpretation), highlighting complex interactions between domestic financial development and fiscal balances.
- Research gaps:
  - Need for further exploration of interaction effects between policy regimes and economic outcomes on saving.
  - Continued inclusion of non-standard variables (temporary/permanent income components, detailed commodity price measures, pension-system variables) in macro saving studies for LAC.

*Source: _wp15108, IMF Working Paper content provided.*

### 1. Determinants of Private Saving in Previous Studies on LAC ...............................................5

### 1. Determinants of Private Saving in Previous Studies on LAC ...............................................5

### Major sections (as organized in the content unit)
- 1. Determinants of Private Saving in Previous Studies on LAC ...............................................5
- 2. Descriptive Statistics of Private Saving ...............................................................................11
- 3. Regional Differences ...........................................................................................................20
- 4. Differences Across LAC Subgroups ....................................................................................22
- 5. Public Saving Determinants in LAC Subgroups .................................................................27
- 6. Terms of Trade and Commodity Price Shocks ....................................................................29
- 7. Policy Frameworks’ Impact on Private Saving ...................................................................33

### Figures listed in the content unit
- Figure 1. Average Saving Rates ..........................................................................................................12
- Figure 2. Private Saving Rates in LAC Subgroups .............................................................................14
- Figure 3. Household and Corporate Saving Rates in LAC .................................................................15
- Figure 4. Public Saving Rates in LAC Subgroups ..............................................................................16
- Figure 5. World and Regional Estimates.............................................................................................21
- Figure 6. Contributions........................................................................................................................25
- Figure 7. Change in Saving Rates Based on TOT Projections for 2019 .............................................30
- Figure 8. Policy Frameworks ..............................................................................................................32

### Content structure for ingestion and indexing
- The unit is organized into seven numbered analytical sections (1–7) focusing on:
  - prior empirical literature on private saving in LAC,
  - descriptive statistics,
  - spatial/regional heterogeneity,
  - heterogeneity across LAC subgroups,
  - public saving determinants within subgroups,
  - the role of terms of trade and commodity price shocks,
  - the influence of policy frameworks on private saving.
- Eight figures (Figures 1–8) accompany the text, mapping average and subgroup saving rates, household vs. corporate saving, public saving, world/regional estimates, contributions, projections based on terms-of-trade (TOT) for 2019, and policy frameworks.

*Source: _wp15108 - 1. Determinants of Private Saving in Previous Studies on LAC ...............................................5*

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

### _wp15108 - References .............................................................................................................

### I. Introduction
- Research question: What policies are most effective in raising saving, especially in LAC?
- Motivations:
  - LAC exhibits much lower private saving rates than the average for the world, with heterogeneity: particularly low saving in smaller countries and high volatility in countries exposed to exogenous shocks (natural disasters, commodity price fluctuations).
  - Importance of terms of trade (TOT) and commodity-price shocks for LAC saving (cites Adler and Magud, 2013; Gruss, 2014).
  - Institutional and macroeconomic reforms in the last three decades (adoption of fiscal rules, inflation targeting, floating exchange rates) — unclear extent of effect on private saving.
- Contributions of the paper:
  - Extensive review of previous empirical research on private saving in LAC, summarizing results and identifying contradictions and omissions.
  - Empirical results from an extensive search of determinants of private and public saving rates, adding previously neglected variables (different measures of TOT and commodity-price shocks; macroeconomic policy regimes including fiscal rules, monetary regimes, exchange-rate regimes), including linear and interaction specifications.
  - Analysis of statistical differences in saving determinants between LAC and the rest of the world in a nested econometric framework, and differences within three LAC subgroups.

- Key summary results (high level):
  - Baseline private saving estimation is generally robust across regional groups; coefficients for the LAC region do not exhibit much difference from the rest of the world.
  - LAC private saving fundamentally explained by: its own lag, real GPDI per capita, the TOT, share of urban population, and old-age dependency ratio.
  - Public saving: increased income growth reduces public saving (possible procyclical fiscal policy); increased private sector credit raises public saving (suggesting crowding out). Results differ across LAC subgroups: larger countries show signs of countercyclical fiscal policy and less sensitivity to old-age dependency.
  - TOT and commodity price volatility and accelerations, and their temporary components, raise private saving in both LAC and the rest of the world.
  - Adoption of fiscal rules and floating exchange rate regimes do not affect private saving; adoption of inflation targeting reduces private saving irrespective of the negative effects of inflation on saving.

### II. Previous empirical literature for LAC — stylized findings and gaps
- Literature characteristics:
  - Wide range of behavioral assumptions, model specifications, econometric techniques, and data samples leading to contradictory results and large dispersion in parameter estimates.
  - Studies include macroeconomic aggregate models (18 papers), microeconomic survey-based studies (11 papers), single-equation estimations, simulation exercises, and some general equilibrium or multi-equation sector models.
  - Sample sizes vary widely: macro single-country 20 to 624 observations; panel studies 99 to 128 observations; micro studies 1,762 to 392,263 observations.
- Commonly investigated determinant categories (and empirical patterns across 26 LAC studies):
  - Income level and income growth: theoretically ambiguous; empirical studies more often report positive effects on private saving but with variation (examples from table: Income level current: + (1,2,3,8,10,17,26); 0 (21); − (16)).
  - Terms of trade (TOT): some studies report positive effects.
  - Demographics: age dependency ratios tend to reduce private saving in aggregate studies; micro studies sometimes show younger cohorts save more.
  - Financial access and credit: mixed results; micro evidence often finds expected negative sign (greater access → lower precautionary saving).
  - Public saving/fiscal policy: 16 macro studies show higher public saving (or larger fiscal balance) reduces private saving; offset coefficients reported range from 0 to 0.75, with most estimates significantly different from both zero and one (full Ricardian equivalence rejected).
  - Pension reforms (transition to fully-funded systems): five country studies find increases in national saving since reform start by: 2.3-2.7 percentage points in Chile, 1.5 pp in Colombia, 1.0 pp in Mexico, and 0.3 pp in Peru (point estimates based on ranges in corresponding studies).
- Identified omissions and limitations in prior LAC macro studies:
  - Absent: private saving determinants including terms-of-trade and commodity-price shocks; policy regimes (fiscal rules, monetary regimes, exchange-rate regimes); explicit identification of differences between LAC and other world regions and within LAC subgroups.
  - Most studies include only a small number of potential determinants, rely on mostly linear specifications, and rarely include interaction effects between policy and economic outcome variables.

### III. Data and stylized facts
- Dataset:
  - Based on Grigoli et al. (2014) dataset with 4,137 observations spanning 32 years for 165 countries (note: slight revisions cause loss of 12 observations vs Grigoli et al. (2014)).
  - Private saving rate = ratio of gross private saving to gross private disposable income (GPDI).
  - Public saving rate = ratio of gross public saving to GPDI.
- Commodity price measures:
  - Country-specific net commodity price index (NCPI) from Gruss (2014): monthly data converted to annual averages. Permanent and temporary components constructed by applying HP filter to log of index with lambda set to 6.25.
- Policy regime dummies:
  - Fiscal rules from IMF Fiscal Rules Dataset; inflation targeting framework from IMF Finance & Development; exchange rate regime from IMF Annual Report on Exchange Arrangements and Exchange Restrictions (monthly, collapsed to annual modes).
  - Dummy variables: 1 if country has fiscal rule (expenditure, budget balance, debt, or any of these), inflation targeting framework, or de jure flexible exchange rate regime; 0 otherwise.
- Regional and subgroup classification:
  - Regions per IMF regional departments. LAC split into: LAC7 (seven largest LAC economies), Middle LAC (other South American countries, Central America, Dominican Republic), and Small LAC (small Caribbean economies, Belize, Guyana). (Appendix I lists countries.)
- Stylized facts (private and public saving, unbalanced panel 164 countries, 1981–2012):
  - Private saving means (percent of GPDI):
    - World: 20.4
    - Africa: 13.4
    - Asia and Pacific: 26.7
    - Europe: 24.4
    - Middle East and Central Asia: 26.1
    - Western Hemisphere: 16.4
    - LAC (regional aggregate): 15.6
    - LAC7: 20.0
    - Middle LAC: 15.1
    - Small LAC: 13.1
  - Volatility and persistence:
    - LAC standard deviation and coefficient of variation are below world averages, but Small LAC has volatility comparable to Africa.
    - Autocorrelation (AC1) is high across regions; private saving in LAC7, Asia and Pacific, and Europe exhibit higher inertia (e.g., AC1 for LAC7 = 0.71).
  - Trends and notable country patterns (1981–2012):
    - Global private saving ~20 percent of GPDI stable; LAC ~4.5 pp lower on average.
    - LAC7 saving remained strong post-Global Financial Crisis, exceeding global average by about 2.4 percent since 2006.
    - Small LAC experienced sharp crisis effects: example private saving fell from 8.1 to 3.3 percent of GPDI in 2008, rebounding to 9.5 in 2009.
    - Public saving more volatile, ranged roughly 3.5-7 percent of GPDI; surged mid-2000s (4.4 percent in 2002 → 10.3 percent in 2006), then fell during the Global Financial Crisis and rose by 2012.
    - Country-specific extremes: Venezuela public saving nearly 44 percent of GPDI in 2005, later ~15 percent in 2012; St. Vincent and the Grenadines private saving hit -19.3 percent of GPDI in 2008 and remained below -10 percent of GPDI by 2012.
  - Household vs corporate saving:
    - Household saving ~5–8 percent of GPDI in Chile, Colombia, Dominican Republic; large fluctuations elsewhere; corporate saving generally higher and more volatile (except Bolivia and Mexico).

### IV. Is LAC different? — methodology and empirical findings
H3: Empirical methodology
- Estimation framework:
  - Dynamic model estimated with two-step system GMM (S-GMM) following Arellano and Bond (1997), Blundell and Bond (1998), with Windmeijer (2005) finite-sample correction to two-step covariance matrix.
  - Baseline private saving specification includes endogenous (and predetermined) covariates: log real per capita GPDI (PPP), real growth rate of per capita GPDI (PPP), public saving/GPDI, inflation, real deposit rate, flow of private sector credit/GPDI; exogenous covariates: log TOT, old-age dependency ratio, share of urban population.
  - Use of nested specifications: interactions between regressors and region or subgroup dummies to test differential effects (equations (2)-(4) in source).
  - Diagnostics: AR(1), AR(2) tests for serial correlation; Hansen J-test for instrument validity. Instruments collapsed; Windmeijer correction applied.

H3: Private saving — core econometric findings
- Baseline results (Table 3, Column 1; world sample) — statistically significant determinants include:
  - Lagged private saving: coefficient 0.593*** (std. err. 0.066)
  - Ln real per capita GPDI (PPP): 0.171*** (0.032)
  - Real growth rate of per capita GPDI (PPP): 0.255*** (0.040)
  - Ln TOT: 0.052*** (0.015)
  - Public saving/GPDI: -0.253*** (0.064)
  - Inflation: 0.388*** (0.124)
  - Flow of private sector credit/GPDI: -0.082** (0.039)
  - Old-age dependency ratio: -1.129*** (0.242)
  - Share of urban population: -0.392*** (0.099)
- Regional differential findings (nested interactions):
  - Private saving less persistent in Africa (interaction on lagged dependent variable significant and negative in Column 2).
  - Sensitivity to income level lower in Europe and Middle East and Central Asia: a one pp increase in income level raises private saving by about 0.06 pp, ~half the effect in other economies.
  - Reduced sensitivity of private saving to income growth in poorer countries (Africa); Asia and Pacific shows larger sensitivity to income growth.
  - High-growth Asian economies show stronger responses to inflation and real deposit rates.
  - Old-age dependency ratio has particularly strong negative effect in Europe.
  - Negative impact of urban population share on private saving is considerably reduced in Europe, Middle East and Central Asia, and LAC; in LAC a one pp increase in share of urban population is associated with a reduction in saving of only 0.01 pp (interaction result described qualitatively).
- LAC subgroup analysis (Table 4):
  - No significant differential effects between LAC subgroups and rest of world for baseline regressors, except inflation in Middle LAC (interaction reduces private saving by 0.09 pp).
  - Conclusion: LAC subgroups share sensitivity to fundamental private saving determinants comparable to the rest of the world.
- Contributions to fitted private saving (Figure 6, Table 3, Column 1 coefficients):
  - Largest positive contributions: lagged private saving, real GPDI per capita (PPP), TOT.
  - Largest negative contributions: demographic variables (old-age dependency, share of urban population).
  - Quantitative contributions (selected regional examples, percent of GPDI):
    - Lag private saving contribution: LAC = 9.3; LAC7 = 11.8; Middle LAC = 9.0; Small LAC = 7.1.
    - Ln real GPDI per capita (PPP) contribution: Europe = 43.2; LAC = 29.3; Small LAC = 35.8; LAC7 = 31.5; Middle LAC = 22.9.
    - Ln TOT contribution: range ~23.8 (Middle LAC) to 24.4 (LAC7).
    - Share of urban population (negative contributions): Europe = -27.6; LAC = -23.6; LAC7 = -31.6; Small LAC = -16.7.
    - Old-age dependency ratio (negative contributions): Europe = -24.0; Asia and Pacific = -12.1; LAC = -11.2; Middle LAC = -10.5; Small LAC = -12.5.
  - Other significant determinants with small contributions across regions: growth rate of real GPDI (PPP), inflation, public saving, flow of credit to private sector.

H3: Public saving — core econometric findings (Table 5)
- Baseline public saving specification (Table 5, Column 2, restricted specification) — significant determinants and magnitudes:
  - Lagged public saving: 0.295*** (0.110) in Column 2 baseline restricted.
  - Ln real per capita GPDI (PPP): 0.222** (0.089) in Column 2 baseline restricted.
  - Real growth rate of per capita GPDI (PPP): -0.201*** (0.053)
  - Ln TOT: 0.077*** (0.030)
  - Private saving/GPDI: -0.477*** (0.147)
  - Flow of private sector credit/GPDI: 0.150*** (0.057)
  - Old-age dependency ratio: -2.699*** (0.988)
- Key interpretations:
  - Growth in real GPDI per capita reduces public saving by 0.14 pp in full specification (Column 1), interpreted as possible procyclical fiscal policy.
  - Increased private sector credit is associated with higher public saving (0.06 pp in full specification), interpreted as possible crowding out.
  - Private-public saving interaction: a one pp increase in private saving associated with 0.24 pp fall in public saving (Column 1), consistent with offset behavior.
- LAC subgroup differences (Columns 3–6):
  - LAC7: persistence of public saving > 1; income growth has positive impact on public saving (one pp increase in growth raises public saving by 0.41 pp) — evidence suggesting countercyclical fiscal policy in larger LAC economies.
  - Middle LAC: increase of one pp in real GPDI per capita (PPP) has virtually no effect on public saving → fiscal acyclicality in this subgroup.

H3: Terms-of-Trade and Commodity Price Shocks (Table 6; Figures 7)
- Tests performed:
  - Role of level, growth rate, volatility (3-year rolling SD), and temporary/permanent components of TOT and NCPI on private saving.
  - Interactions with LAC dummy to test differential effects.
  - Controls: baseline specification variables and their interactions with LAC dummy kept in estimations.
- Main empirical findings (Table 6):
  - Growth/accelerations:
    - Growth rate of TOT: a one pp increase in growth rate of TOT raises private saving by 0.058*** (0.018) (Column 1).
    - Growth rate of NCPI: a one pp increase in growth rate of NCPI raises private saving by 0.029** (0.013) (Column 2).
    - No differential effect for LAC found in interactions.
  - Volatility (income uncertainty proxies):
    - SD of ln TOT (3-year rolling SD): coefficient 0.113** (0.053) (Column 3) → a one pp increase in SD of TOT associated with 0.113** increase in private saving.
    - SD of ln NCPI: 0.143** (0.061) (Column 4).
    - No differential LAC effects detected.
  - Permanent vs temporary components (HP filter decomposition):
    - Temporary component of TOT: 0.098*** (0.024) (Column 5) → a one pp increase in temporary component raises private saving by 0.098***.
    - Temporary component of NCPI: 0.056*** (0.020) (Column 6).
    - Permanent components not significant.
    - No differential LAC effects detected.
- Policy-relevant projection exercise (using WEO January 19, 2015 TOT projections to 2019, and coefficients from Table 3 Col 1 and Table 5 Col 2):
  - On average TOT projected to improve by 0.9 percent by 2019 in LAC (heterogeneity: Grenada +16.8 percent; Venezuela -20.8 percent).
  - Estimated change in private and public saving rates (product of first difference of ln TOT between 2012 and 2019 and the TOT coefficients):
    - LAC region average: mild increase in both private and public saving roughly 0.1 percent of GPDI.
    - By subgroup:
      - LAC7: private saving projected to fall by 0.4 percent of GPDI; public saving projected to fall by 0.5 percent of GPDI.
      - Middle LAC: private saving projected to increase by 0.1 percent of GPDI; public saving by 0.2 percent of GPDI.
      - Small LAC: private saving projected to increase by 0.3 percent of GPDI; public saving by 0.5 percent of GPDI.
    - Grenada: largest projected increase (private saving +0.9 percent of GPDI; public saving +1.3 percent of GPDI).
    - LAC oil exporters expected to observe largest fall—0.6 percent of GPDI in private saving (text truncated at end of provided content).

### V. Conclusions and policy implications (drawn from findings)
- Determinants and robustness:
  - Private saving behavior in LAC largely mirrors global patterns: persistence, income level, TOT, urbanization, and demographics (old-age dependency) are the primary drivers.
  - Interactions and additional tested variables (policy regime dummies for fiscal rules, inflation targeting, flexible exchange rates) show:
    - No detectable effect of fiscal rules or de jure floating exchange rate regimes on private saving.
    - Adoption of inflation targeting is associated with reduced private saving (result reported qualitatively in introduction).
- Implications for policymakers in LAC:
  - Demographic trends (aging, urbanization) exert substantial negative contributions to private saving; policies addressing long-term pension burdens and demographic adaptation are central.
  - Commodity price shocks:
    - Temporary and volatile increases in commodity prices (TOT and NCPI temporary components and volatility) are saved — precautionary and temporary-income saving behavior observed.
    - Projected commodity price shifts imply heterogeneous impacts across LAC countries and subgroups; oil exporters may face declines in saving ratios if TOT deteriorate.
  - Fiscal policy behavior matters for public saving dynamics:
    - Growth episodes are associated with lower public saving (procyclical fiscal policy) in some contexts; larger LAC economies show signs of countercyclical policy, suggesting scope for fiscal frameworks that support saving stabilization.
  - Financial deepening and private credit flows: mixed effects on private saving but associated with higher public saving (crowding-out interpretation), highlighting complex interactions between domestic financial development and fiscal balances.
- Research gaps identified:
  - Need for further exploration of interaction effects between policy regimes and economic outcomes on saving.
  - Continued inclusion of non-standard variables (temporary/permanent income components, detailed commodity price measures, pension-system variables) in macro saving studies for LAC.

*Source: _wp15108 - References, IMF working paper content provided.*

### 0.9 percent of GPDI in public saving—with Venezuela observing the largest fall (1.1 percent of

### _wp15108 - 0.9 percent of GPDI in public saving—with Venezuela observing the largest fall (1.1 percent of

### Terms-of-Trade (TOT) shocks and saving
- "0.9 percent of GPDI in public saving—with Venezuela observing the largest fall (1.1 percent of GPDI in private saving and 1.6 percent of GPDI in public saving)."
- If projected changes in TOT were temporary, the final effect on private saving would be larger, "ranging from -2.04 percent for Venezuela and 1.65 percent for Grenada," as the magnitude of the coefficient is almost twice as large as the one for actual TOT.
- WEO subgroup of oil importers includes some relatively minor oil exporters and, as a result, the final effect on saving is close to zero.

### Fiscal rules, inflation targeting, and exchange rate flexibility — prevalence and empirical tests
- Prevalence as of end-2012:
  - "75 countries have some category of fiscal rule in place (14 in LAC)"
  - "25 have adopted an inflation targeting regime (6 in LAC)"
  - "61 run flexible exchange rate regimes (7 in LAC)"
- Empirical strategy: tested linear effects of policy regimes and interactions between policy regimes and selected relevant variables, for the world and differential effects in LAC, while maintaining controls from the baseline specification.

### Empirical findings from Table 7 (impact on private saving / GPDI)
- Adoption of fiscal rules:
  - Tested expenditure rules, budget rules, debt rules, and a general fiscal rule dummy.
  - "None of the results are significant at five percent, suggesting that the adoption of fiscal rules is not reflected in changes of private saving."
- Inflation targeting:
  - "The results suggest that the inflation targeters have lower private saving by 2.66 pp—a worldwide result shared by LAC."
  - Table coefficient: "Inflation targeting........-2.662**-1.834.."
  - Interaction: "Inflation targeting * Inflation.........-0.231*.."
- Floating exchange rate regimes:
  - Aggregate effect ambiguous a priori (buffers vs. higher volatility).
  - Empirical coefficients are not significant: "The results in Columns 11 and 12 confirm this ambiguity as the coefficients are not significant."
  - Table coefficient excerpt: "Floating ER regime..........-1.303-1.118"
- Estimation and diagnostics (as reported in Table 7):
  - Time-fixed effects: "Yes" across specifications.
  - Lags/Instruments: values listed include "1/671/711/671/711/671/711/671/711/671/711/591/63"
  - AR(1) p-val.: "0.000" (all columns)
  - AR(2) p-val.: values reported include "0.3430.3380.3470.2940.3360.3060.3390.2890.3440.3420.3910.385"
  - Hansen J-test p-val.: values reported include "0.9180.9450.8860.4760.9160.2630.9080.4180.9430.9630.5160.359"
  - Observations reported: "3,2433,2433,2433,2433,2433,2433,2433,2433,2433,2432,7912,791"
  - Number of countries: "153153153153153153153153153153153153"

### Synthesis of conclusions and policy implications
- Robust common findings in literature and this paper:
  - Income level, income growth, and TOT raise private saving.
  - Age dependency ratios tend to reduce private saving rates.
  - Household structure variables: formal head and number of income-earning members increase private saving; number of under-age members decrease it.
  - Risk aversion and the discount factor raise saving.
- Policy-variable patterns:
  - Fiscal consolidation reduces private saving; government pension payments reduce private saving; government transfers reduce household saving.
  - Taxes on firms tend to reduce corporate saving; personal income tax rate lowers private saving.
  - Interest rate has ambiguous impact on private saving; the level of inflation positively affects private saving.
  - Financial access measures: micro evidence consistent with negative sign (greater access reduces precautionary saving).
  - Foreign saving inflows tend to partially offset aggregate private saving; restrictions on capital flows raise aggregate private saving.
- Pension reforms:
  - Transition toward a fully-funded pension regime raises household (and private) saving due to fiscal deficit effects.
  - Mandatory pension saving by active workers is partly (at most totally) offset by lower voluntary household saving.
- Regional and subgroup findings:
  - Nested baseline private saving estimation generally robust across regional groups; "LAC private saving behaves in line with the rest of the world."
  - LAC is much less responsive to urbanization.
  - From contribution analysis: LAC saving fundamentally explained by its own lag, real GPDI per capita, the TOT, share of urban population, and old-age dependency ratio.
  - Saving persistence contributes more to saving in LAC7 than in Small LAC; urbanization contributes negatively twice as much to saving in LAC7 than in Small LAC.
- Additional findings on volatility and temporary components:
  - TOT and NCPI volatility and accelerations as well as the temporary components of TOT and NCPI raise private saving in both LAC and the rest of the world.
- Public saving behavior:
  - Many private saving determinants affect public saving similarly.
  - Increased income growth reduces public saving (possible procyclical fiscal policy).
  - Increased private sector credit raises public saving (suggesting some crowding out).
  - For LAC and Small LAC, public saving behaves no differently than the rest of the world.
  - In LAC7, public saving shows more persistence, is positively affected by income growth (implying countercyclical fiscal policy), and is less sensitive to old-age dependency.
  - In Middle LAC countries, income level reduces public saving (suggesting higher procyclical fiscal policy) and public saving is less responsive to old-age dependency.

*Source: Authors’ calculations, IMF Working Paper content as provided.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15108.pdf_