## _wp1003

## Source details

**Canonical URL:** [_wp1003](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1003.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1003.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1003.pdf.json)

---

### I. Introduction and overview of investment trends
- Aggregate investment in Argentina rose sharply after the crisis earlier this decade and helped spur the recovery in economic growth since 2002.
- Research focus: impact of the changing economic environment on investment decisions of large firms using a panel of balance sheet data for just under 90 listed firms.
- Main empirical links examined:
  - Investment and a measure of the productivity of capital.
  - Investment and a proxy for financing constraints.
  - Investment and macro variables such as the real exchange rate and the cost of borrowing.
- Key empirical summary:
  - Investment is positively correlated with a more appreciated real exchange rate for the large, listed firms in the sample.
  - The effect on investment is halved for firms in the traded goods sector.
  - A firm-specific effective interest rate measure is developed and found to be a significant determinant of investment.
  - Evidence consistent with increased impact of financing constraints on investment decisions after the 2001 crisis.

### A. Macroeconomic developments and aggregate investment
- Historical context and policy regime change:
  - Crisis produced a cumulative fall in real GDP of 18 percent between 1998 and 2002.
  - Fixed exchange rate system gave way to a floating exchange rate, sovereign debt default, and forcible conversion of previously dollar-denominated financial contracts to pesos.
- Post-crisis recovery drivers:
  - Favorable external environment, supportive policy, reversal of fiscal and external deficits into surpluses.
  - Real depreciation of the peso, real wage erosion, declines in real energy costs, short-term real interest rates became negative.
- Gross fixed capital formation (GFCF):
  - After 2001, GFCF rose steadily as a proportion of GDP; ratio remained within range of values in Latin America and low compared with other emerging markets.
  - Investment growth led by construction, especially residential.
  - Investment in vehicles, equipment and machinery recovered more slowly; capital stock per worker declined substantially from pre-2002 peaks.
- Capital stock construction note:
  - Maia and Nicholson (2005) use 17 percent depreciation for durable goods versus 3 percent for construction.

### Exposure of sectors to relative price changes (1997, Table 1)
- Trade and energy intensity (percent of gross production):
  - Agriculture and fishery: Exports 14.5, Imports 2.1, Electricity use 0.6
  - Mining: Exports 25.5, Imports 2.1, Electricity use 1.4
  - Extraction of metals: Exports 36.9, Imports 15.3, Electricity use 23.6
  - Manufacturing: Exports 14.4, Imports 10.6, Electricity use 0.9
  - Oils and their subproducts: Exports 82.2, Imports 5.1, Electricity use 0.6
  - Motor vehicles: Exports 36.2, Imports 38.3, Electricity use 0.3
  - Other secondary: Exports 0.0, Imports 2.8, Electricity use 0.8
  - Services: Exports 1.2, Imports 1.1, Electricity use 0.8
  - Air transport: Exports 4.1, Imports 17.4, Electricity use 4.5
  - Total: Exports 6.3, Imports 4.1, Electricity use 0.9
- Interpretation:
  - Most trade-intensive sectors: primary extractive industries and manufacturing.
  - Many manufacturing industries, mining and some services are energy-intensive (electricity inputs as proportion of gross production).

### B. Investment and the real exchange rate
- Channels through which real depreciation affects firm-level profitability and investment:
  - Share of exports in sales; reliance on imported inputs; impact on foreign-currency debt service.
- Sectoral outcomes:
  - Devaluation led to sharp rise in value added for tradable activities; prices in mining, agriculture and manufacturing increased far more rapidly than for services.
  - Value added per hour worked in US dollars barely dipped in mining and agriculture; agriculture rose well above the economy average.
  - Rising commodity prices aided these trends.
- Large-firm evidence:
  - In 2005, 75 percent of value added generated by the 500 largest firms was in mining and manufacturing, compared to 26 percent in total gross value added.
  - These firms’ production, value added and profits jumped after the devaluation; firms in non-tradable sectors (communications, electricity and gas) experienced losses.
  - The 500 largest firms accounted for about 20 percent of gross value added.
- Capital goods and imports:
  - Prior to the crisis, imports of capital goods were equivalent to half the value of durable goods investment; on devaluation, fall in imported capital goods accounted for more than three-quarters of the fall in durable goods investment.
  - As recovery progressed, capital goods imports grew as proportion of total durable goods investment to exceed previous peaks.
- Debt dollarization and asymmetric pesification:
  - Dollarization of domestic financial institutions’ credit to the private sector had risen to 72 percent by 2001.
  - For the 500 largest firms, burden of servicing this debt jumped by 2½ percent of GDP on devaluation.
  - Financial expenses declined rapidly thereafter, reflecting benefits to firms of “asymmetric pesification.”

### C. Financial conditions
- Despite policy-led fall in short-term real interest rates, financial conditions facing firms worsened after the crisis:
  - Sovereign default and asymmetric pesification provoked a banking crisis and curtailed access to foreign financing.
  - Bank lending was weak for several years; restructuring of dollarized corporate debt often long-drawn-out.
  - Restrictions on dollar lending or indexation of debt to prices or exchange rate contributed to reduction in long-term credit; dollar lending did not recover.
  - Decline in aggregate credit despite recovery suggests financing constraints became worse and credit rationing more prevalent until 2006.

### II. Investment behavior at the firm level — empirical approach and data
- Empirical approach:
  - Unbalanced panel of 87 non-financial listed firms for 1993 to 2006.
  - Construct firm-specific cost of capital: effective interest rate derived from balance sheet debt and interest payments.
  - Infer financing constraints through sensitivity of investment to balance sheet measures of net worth or liquidity.
- Firm size and sector classification:
  - Firms categorized into “small” and “large” using two definitions: (i) annual highest/lowest quartiles by assets (composition changes by year); (ii) constant-composition small firms (lowest quartile in at least 9 of 13 years plus three short-lived firms).
  - Traded sectors in sample: mining, agriculture or manufacturing.
  - Through the sample: 67 percent of firms are in the traded goods sector; 71 percent of observations are in the traded goods sector; 58 percent of firms are in manufacturing; 61 percent of observations are in manufacturing.
- Measurement:
  - Investment measured using capital expenditures from cash flow statement.
  - Capital stock derived via permanent inventory approach; depreciation rate assumed 8 percent.
  - Investment and capital stock deflated by INDEC WPI machinery and electrical equipment price index; sales and cash flow deflated by overall WPI.
  - Real exchange rate: log of trade-weighted, CPI-based multilateral index.
  - Cost of capital: firm-specific effective interest rate from debt and interest payments.

### Empirical model specifications
- Baseline accelerator model and augmentations:
  - Eqn. 1: Iit/Kit-1 = β1 ∆Sit/Kit-1 + β2 Cit/Kit-1 + νi + εit
  - Eqn. 2: adds interaction C * DUM_Small
  - Eqn. 3: adds log(REER) and log(REER)*DUM_Tradable
  - Eqn. 4: adds firm-specific cost of debt (r)
- Expectations:
  - ∆Sales coefficient positive.
  - Cash stock coefficient positive (may reflect financing constraints or correlation with expected profitability).
  - Cash × small-firm positive implies stronger financing constraints for small firms.
  - Cost of borrowing coefficient expected negative.
  - Real exchange rate coefficient indeterminate; depreciation expected to favor tradable firms.

### Specification problems and endogeneity handling
- Concerns:
  - Cash flow/stock may correlate with marginal productivity of capital (endogeneity).
  - Simultaneous determination of investment, cash, and sales.
  - Dynamic investment adjustment causes serial correlation; lags violate strict exogeneity.
- Remedies/trials:
  - Use of cash stock (vs flow) to reduce correlation with marginal productivity.
  - Instrumental variables (IV): GDP growth, commodity price indices, corporate tax and VAT receipts.
  - System GMM estimators with forward orthogonal deviations and Windmeijer correction.
  - Recursive estimation to assess parameter stability.

### Main empirical findings (selected quantitative results preserved exactly)
- General coefficient interpretations:
  - Investment-capital ratio rises by between 0.02 and 0.03 percentage points for each percentage point increase in the ratio of the change in sales to capital.
  - Investment-capital ratio rises by between 0.05 and 0.08 percentage points for each point increase in the cash stock to capital ratio.
- Selected estimated coefficients (Table 2; FE and IV specifications; p-values in parentheses):
  - ∆S: 0.030*** (0.001) in Eqn 1 (FE); other reported ∆S coefficients include 0.029** (0.030), 0.022** (0.035), 0.027*** (0.000), 0.028*** (0.001), 0.025*** (0.001), 0.025** (0.047), 0.025** (0.041), 0.033*** (0.000), 0.028*** (0.000).
  - C: 0.057* (0.056) in Eqn 1 (FE); other reported C coefficients include 0.033 (0.499), 0.869*** (0.001), 0.0775** (0.041), 0.084** (0.011), 0.472** (0.045), 0.050 (0.214), 0.053 (0.195), 0.057* (0.075), 0.091*** (0.002).
  - RER (log real exchange rate): 0.142*** (0.000) and 0.121*** (0.001) in relevant FE specifications.
  - Trade-RER interaction: -0.076* (0.060), -0.075* (0.057), -0.087** (0.034).
  - r (firm-specific cost of borrowing): -0.217** (0.023), -0.427*** (0.005).
  - Post-crisis interactions: ‘02‘06_r = 0.365** (0.033); ‘02’06_C_S = 0.261*** (0.000); ‘02’06_C shows 0.025 (0.258) and 0.097* (0.010) in reported variants.
- Real exchange rate effects:
  - A one percent increase in the real exchange rate (i.e., an appreciation) is correlated with an increase in the investment-capital ratio of about 0.15 percentage points.
  - For traded goods sector firms, the beneficial impact of an appreciation is reduced by about half (negative and significant Trade-RER interaction).
- Effective interest rate effects:
  - Investment-capital ratio falls by about 0.2 of a percentage point for each percentage point increase in the effective interest rate (r).
  - Post-2002 interaction indicates weaker total interest-rate sensitivity for the post-crisis period (interaction ‘02‘06_r = 0.365** (0.033)).
- Financing constraints evidence:
  - Cash × small-firm interactive term positive but generally insignificant across baseline specifications.
  - Post-crisis: cash holdings for small firms become positive and highly significant (holds for both definitions of small firms).
  - Interactive term including the real exchange rate yields a positive and significant coefficient for cash holdings in the post-crisis period.
  - Interpretation: results are not inconsistent with a significant worsening of financing constraints in 2002–2006, though baseline evidence is mixed.

### Robustness, GMM, and estimation diagnostics
- FE and random effects often similar; Hausman test cannot reject null of non-systematic differences, but firm fixed effects used.
- IV estimations (instruments: non-fuel and agricultural commodity prices; GDP growth; corporate tax and VAT receipts) yield qualitatively similar results; instrumented cash coefficient can be several times larger than FE estimate.
- System GMM approach (Annex Table 2) — key reported estimates and diagnostics:
  - Investment (L1): Eqn 1: 0.241*** (0.000); Eqn 2: 0.260*** (0.000); Eqn 3: 0.260*** (0.000).
  - I (L2): Eqn 1: 0.058* (0.073); Eqn 2: 0.025 (0.417); Eqn 3: 0.027 (0.400).
  - Sales: Eqn 1: 0.039** (0.050); Eqn 2: 0.025** (0.034); Eqn 3: 0.024* (0.056).
  - Cash stock: Eqn 1: 0.128 (0.448); Eqn 2: -0.090 (0.122); Eqn 3: -0.101* (0.078).
  - No. of observations: 656 for Eqn 1–3.
  - Instrument count: Eqn 1: 47; Eqn 2: 59; Eqn 3: 60.
  - Hansen p-value: Eqn 1: 0.630; Eqn 2: 0.201; Eqn 3: 0.158.
- GMM caveats:
  - Structural break from the crisis violates steady-state/orthogonality assumptions underpinning internal instruments; forward orthogonal deviations used; Windmeijer correction applied.
  - GMM results sensitive to instrument count and specification; coefficients for cash, cost of debt capital, or real exchange rate are volatile across specifications.
  - Arellano-Bond tests: reject no first-order autocorrelation; cannot reject absence of second-order autocorrelation.
- Recursive estimation:
  - Recursive estimates starting with 1993–2001 and progressively adding years show coefficient estimates generally stable and signed as expected; stock-of-cash insignificant in shortest subsample but otherwise stable.

### Descriptive statistics (Annex Table 1 medians and percentiles)
- I/K (847 Obs): 25th pcile 0.02; Median 0.06; 75th pcile 0.13
  - Pre-crisis (499 Obs) Median 0.08; Post-crisis (348 Obs) Median 0.04
  - Small (208 Obs) 25th 0.01; Median 0.04; 75th 0.08; Pre-crisis (123 Obs) Median 0.05; Post-crisis (85 Obs) Median 0.02
  - Large (212 Obs) 25th 0.04; Median 0.09; 75th 0.15; Pre-crisis (126 Obs) Median 0.11; Post-crisis (86 Obs) Median 0.04
  - Traded (608 Obs) 25th 0.02; Median 0.06; 75th 0.14; Pre-crisis (380 Obs) Median 0.07; Post-crisis (228 Obs) Median 0.04
  - Non-traded (239 Obs) 25th 0.03; Median 0.06; 75th 0.12; Pre-crisis (119 Obs) Median 0.10; Post-crisis (120 Obs) Median 0.04
- ∆Sales/K (832 Obs): 25th pcile -0.10; Median 0.02; 75th pcile 0.13
  - Pre-crisis (491 Obs) Median 0.01; Post-crisis (341 Obs) Median 0.02
  - Small (196 Obs) 25th -0.26; Median 0.01; 75th 0.21; Pre-crisis (116 Obs) Median -0.04; Post-crisis (80 Obs) Median 0.05
  - Large (212 Obs) 25th -0.06; Median 0.01; 75th 0.09; Pre-crisis (126 Obs) Median 0.02; Post-crisis (86 Obs) Median 0.01
  - Traded (593 Obs) 25th -0.12; Median 0.02; 75th 0.15; Pre-crisis (372 Obs) Median 0.00; Post-crisis (221 Obs) Median 0.07
  - Non-traded (239 Obs) 25th -0.07; Median 0.01; 75th 0.09; Pre-crisis (119 Obs) Median 0.02; Post-crisis (120 Obs) Median 0.00
- Cash/K (847 Obs): 25th pcile 0.02; Median 0.05; 75th pcile 0.17
  - Pre-crisis (499 Obs) Median 0.04; Post-crisis (348 Obs) Median 0.07
  - Small (208 Obs) 25th 0.02; Median 0.07; 75th 0.22; Pre-crisis (123 Obs) Median 0.07; Post-crisis (85 Obs) Median 0.07
  - Large (212 Obs) 25th 0.02; Median 0.05; 75th 0.13; Pre-crisis (126 Obs) Median 0.03; Post-crisis (86 Obs) Median 0.08
  - Traded (608 Obs) 25th 0.01; Median 0.06; 75th 0.19; Pre-crisis (380 Obs) Median 0.05; Post-crisis (228 Obs) Median 0.07
  - Non-traded (239 Obs) 25th 0.02; Median 0.05; 75th 0.13; Pre-crisis (119 Obs) Median 0.04; Post-crisis (120 Obs) Median 0.07
- Cost of K (601 Obs): 25th pcile 0.07; Median 0.09; 75th pcile 0.12
  - Pre-crisis (378 Obs) Median 0.10; Post-crisis (223 Obs) Median 0.08
  - Small (113 Obs) 25th 0.08; Median 0.11; 75th 0.14; Pre-crisis (74 Obs) Median 0.12; Post-crisis (39 Obs) Median 0.10
  - Large (188 Obs) 25th 0.07; Median 0.08; 75th 0.10; Pre-crisis (111 Obs) Median 0.09; Post-crisis (77 Obs) Median 0.08
  - Traded (410 Obs) 25th 0.07; Median 0.10; 75th 0.13; Pre-crisis (273 Obs) Median 0.10; Post-crisis (137 Obs) Median 0.08
  - Non-traded (191 Obs) 25th 0.07; Median 0.08; 75th 0.11; Pre-crisis (105 Obs) Median 0.09; Post-crisis (86 Obs) Median 0.08

### Conclusions and policy-relevant implications
- Aggregate and sectoral changes post-crisis:
  - Important relative price changes (real exchange rate, real interest rates, energy costs, terms of trade, real wages) affected investment incentives and favored tradable and energy-intensive firms (primary production and manufacturing).
  - Aggregate investment rose to levels exceeding those seen during the 1990s as the economy rapidly recovered.
  - Despite much lower real interest rates, aggregate credit fell substantially, indicating financial constraints facing firms may have increased.
  - Capital per worker did not return to previous peaks in the sample period.
- Firm-level conclusions (87 listed firms, 1993–2006):
  - For large listed firms, real exchange rate appreciation is correlated with higher investment; beneficial impact roughly half as large for firms with tradable outputs.
  - Increases in the effective cost of borrowing reduce investment; estimated effect about -0.2 percentage points in I/K per percentage point increase in the effective interest rate.
  - Strong evidence of financing constraints is not found for the full sample period, but post-crisis interactions suggest financing constraints worsened during 2002–2006.
  - Given rise in aggregate investment, firms not captured in the sample—particularly small and medium enterprises—must collectively have greatly increased investment.
- Methodological caveats:
  - Dataset includes only listed, relatively large firms and over-represents tradable sectors; caution advised in generalizing to the wider economy.
  - System GMM and IV approaches provide cross-checks but are sensitive to instrument choices and the structural break in the sample.

*Source: _wp1003 - Annexes 1–2 and main text of the supplied content.*

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

### References

### I. Introduction and overview of investment trends
- Aggregate investment in Argentina rose sharply after the crisis earlier this decade and helped spur the recovery in economic growth since 2002.
- Key changes in the investment environment:
  - Significant real exchange rate depreciation.
  - Negative real interest rates.
  - Increased international demand for Argentine products.
  - Financial conditions affected by sovereign default and banking crisis.
- Research focus: impact of the changing economic environment on investment decisions of large firms using a panel of balance sheet data for just under 90 listed firms.
- Main empirical links examined:
  - Investment and a measure of the productivity of capital.
  - Investment and a proxy for financing constraints.
  - Investment and macro variables such as the real exchange rate and the cost of borrowing.
- Key empirical findings summarized:
  - Investment is positively correlated with a more appreciated real exchange rate for the large, listed firms in the sample.
  - The effect on investment is halved for firms in the traded goods sector.
  - A firm-specific effective interest rate measure is developed and found to be a significant determinant of investment.
  - Evidence consistent with increased impact of financing constraints on investment decisions after the 2001 crisis.
- Structure of the paper:
  - Descriptive overview of aggregate investment and macro variables.
  - Model and panel estimation results.
  - Conclusions.

### A. Macroeconomic developments and aggregate investment
- Historical context:
  - The 1990s featured relative stability, with recession in 1995 and sharper decline starting in 1998 culminating in the 2001/2002 crisis.
  - Crisis produced a cumulative fall in real GDP of 18 percent between 1998 and 2002.
  - Policy regime change: fixed exchange rate system gave way to a floating exchange rate, sovereign debt default, and forcible conversion of previously dollar-denominated financial contracts to pesos.
- Post-crisis recovery:
  - Recovery aided by favorable external environment and supportive policy.
  - Fiscal and external deficits reversed into surpluses.
  - New investment opportunities as global demand for Argentine products rose, particularly commodities.
  - Key relative price changes: real depreciation of the peso, real wage erosion, declines in real energy costs, short-term real interest rates became negative (Figure 1).
- Gross fixed capital formation (GFCF):
  - After 2001, GFCF rose steadily as a proportion of GDP.
  - Ratio remained within the range of values in Latin America and is low compared with other emerging markets.
  - Investment growth led by construction, especially residential.
  - Investment in vehicles, equipment and machinery recovered more slowly; capital stock per worker declined substantially from pre-2002 peaks.
- Capital stock construction note:
  - Maia and Nicholson (2005) use 17 percent depreciation for durable goods versus 3 percent for construction.

### Exposure of sectors to relative price changes (Table 1, 1997)
- Table 1. Exposure to Relative Price Changes by Sector, 1997 (In percent of gross production)
  - Agriculture and fishery: Exports 14.5, Imports 2.1, Electricity use 0.6
  - Mining: Exports 25.5, Imports 2.1, Electricity use 1.4
  - Extraction of metals: Exports 36.9, Imports 15.3, Electricity use 23.6
  - Manufacturing: Exports 14.4, Imports 10.6, Electricity use 0.9
  - Oils and their subproducts: Exports 82.2, Imports 5.1, Electricity use 0.6
  - Motor vehicles: Exports 36.2, Imports 38.3, Electricity use 0.3
  - Other secondary: Exports 0.0, Imports 2.8, Electricity use 0.8
  - Services: Exports 1.2, Imports 1.1, Electricity use 0.8
  - Air transport: Exports 4.1, Imports 17.4, Electricity use 4.5
  - Total: Exports 6.3, Imports 4.1, Electricity use 0.9
- Interpretation:
  - Most trade-intensive sectors: primary extractive industries and manufacturing.
  - Many manufacturing industries, mining and some services are energy-intensive (measured as electricity inputs as a proportion of gross production).

### B. Investment and the real exchange rate
- Channels through which real depreciation affects firm-level profitability and investment:
  - Share of exports in sales.
  - Reliance on imported products in production.
  - Impact on foreign-currency debt service.
- Sectoral outcomes:
  - Devaluation led to a sharp rise in value added for tradable activities (Figure 4).
  - Prices in mining, agriculture and manufacturing increased far more rapidly than for services and other secondary activities.
  - Value added per hour worked in US dollars barely dipped in mining and agriculture; agriculture rose well above the economy average.
  - Rising commodity prices aided these trends.
- Large-firm evidence:
  - Data on the 500 largest firms show concentration in tradable sectors: in 2005, 75 percent of the value added they generated was in mining and manufacturing, compared to 26 percent in total gross value added.
  - These firms’ production, value added and profits jumped after the devaluation, with mining as an outlier.
  - Firms in non-tradable sectors (communications, electricity and gas) experienced losses in the post-crisis period.
  - The 500 largest firms accounted for about 20 percent of gross value added.
- Capital goods and imports:
  - Prior to the crisis, imports of capital goods were equivalent to half the value of durable goods investment (Figure 2).
  - On devaluation, the fall in imported capital goods accounted for more than three-quarters of the fall in durable goods investment.
  - As recovery progressed, capital goods imports grew as a proportion of total durable goods investment to exceed previous peaks.
- Debt dollarization and asymmetric pesification:
  - Dollarization of domestic financial institutions’ credit to the private sector had risen to 72 percent by 2001.
  - For the 500 largest firms, the burden of servicing this debt jumped by 2½ percent of GDP on devaluation (Figure 5).
  - Financial expenses declined rapidly thereafter, reflecting benefits to firms of “asymmetric pesification” that short-circuited balance-sheet channels.

### C. Financial conditions
- Despite policy-led fall in short-term real interest rates, financial conditions facing firms worsened after the crisis:
  - Sovereign default and asymmetric pesification provoked a banking crisis and curtailed access to foreign financing.
  - Domestic capital markets were affected by increased uncertainty.
  - Bank lending (the dominant source of firm financing) was weak for several years; restructuring of dollarized corporate debt was often long-drawn-out.
  - Restrictions on dollar lending or indexation of debt to prices or the exchange rate contributed to a reduction in long-term credit; dollar lending did not recover.
  - Decline in aggregate credit despite recovery suggests financing constraints became worse and credit rationing more prevalent until 2006.

### II. Investment behavior at the firm level
- Empirical approach:
  - Use firm-level panel data to exploit cross-sectional differences and avoid time-series limitations and structural breaks caused by the crisis.
  - Assess impact of relative price changes—specifically the real exchange rate and firms’ costs of external borrowing—on investment.
  - Construct a firm-specific cost of capital variable: an effective interest rate derived from balance sheet data on debt and interest payments (a novel approach in recent literature).
  - Infer presence of financing constraints through the sensitivity of investment to balance sheet measures of net worth or liquidity.
- Theoretical considerations:
  - Under perfect capital markets (Modigliani and Miller (1958)), financial structure should not matter for investment.
  - With financing constraints, firms with higher net worth are more creditworthy; investment depends on internal funds as well as expected marginal productivity of capital.
  - Evidence of financing constraints typically inferred from a strong correlation of investment with measures of internal funds (cash flow) or stock of cash.
  - Potential endogeneity: cash flow may reflect future profitability of investment and thus be correlated with investment for fundamental reasons, complicating interpretation as a pure constraint signal.
- Empirical strategy to address confounding:
  - Sort firms into groups expected a priori to face differing access to market finance (e.g., firm size).
  - Use interaction terms (e.g., cash stock × small firms) to test whether sensitivity of investment to cash is higher for firms likely to be financially constrained.
  - Use stock of cash for a subset of small firms as an alternative measure of internal funds.
- Data:
  - Unbalanced panel of 87 non-financial firms for the period 1993 to 2006.

*References*

### Annex 1 for details). The dataset includes only listed firms, introducing important caveats in

### Dataset coverage and firm classification

### Dataset coverage and caveats
- The dataset includes only listed firms.
- These firms by definition have some access to external finance, and are all “large” by Argentine standards.
- Many of the companies included are part of international groups, which may give them greater access to finance through intra-firm or international capital market operations.
- A much greater proportion are involved in tradable sectors than in the economy as a whole.
- The dataset therefore introduces important caveats in drawing conclusions for the wider economy.

### Firm size definitions and grouping
- Firms are categorized into “small” and “large” groups.
- Two criteria were used to define firm size in the sample:
  - First definition: firms were ranked by assets and year, with those in the highest and lowest quartiles defined as large and small respectively, for each year (i.e., the group composition changes by year).
  - Second definition (constant-composition small firms): selects firms ranked in the lowest quartile by assets in at least 9 of the 13 years of the sample, plus three firms with shorter lives that were ranked as small in all or all-but-one year of their lives.
- Empirical results did not vary significantly for the two definitions (overlap is high).

### Sectoral classification
- Firms are also classified by whether they are in the traded or non-traded sectors.
- Traded sectors defined in the sample include mining, agriculture or manufacturing.

### Key statistics
- Sample span: 13 years (composition reference).
- Constant-composition small-firm criterion: at least 9 of the 13 years.
- Through the sample:
  - 67 percent of firms are in the traded goods sector.
  - 71 percent of observations are in the traded goods sector.
  - 58 percent of firms are in manufacturing.
  - 61 percent of observations are in manufacturing.

*Source: _wp1003 - Annex 1 for details). The dataset includes only listed firms, introducing important caveats in drawing conclusions for the wider economy.*

### Annex Table 1 presents median values for the variables derived from the balance sheet. An

### _wp1003 - Annex Table 1 presents median values for the variables derived from the balance sheet. An

### Data overview and descriptive patterns
- Median balance-sheet based variables indicate:
  - Smaller firms do not in fact seem to face worse financing constraints in this sample.
  - Tradable sector firms have done much better in the post-crisis period.
  - Investment behavior shows a clear change from the pre- to the post-crisis period, while other variables are far more stable and may have relatively low overall explanatory power.
- Caveat: balance sheet data for small firms was not available to the author for some analyses.

### Investment and sales dynamics
- Investment-capital ratio:
  - The median investment-capital ratio declined steadily from the commencement of the recession in 1998, and remained well below its values in the 1990s even at the end of the sample.
  - The gap in investment between small and large firms narrowed as investment of the largest firms fell substantially more than that of the smaller ones.
  - Firms with both traded and non-traded outputs invested less after the crisis, with the fall being twice as large for firms in the non-traded sector.
- Sales relative to capital:
  - Pre-crisis: sales of the median small firm were shrinking.
  - Post-crisis: sales growth in relation to capital for small firms exceeds that of the larger firms fivefold (although the median for small firms is much more volatile).
  - Firms in the traded goods sector suffered considerably more than firms with non-traded outputs in the 1998-2001 recession but have had considerably more dynamic sales in the post-crisis period.

### Cash holdings and debt
- Reserves of cash and near-cash:
  - Overall increase in holdings after 2002 is consistent with more constrained access to external financial resources and the collapse of the banking system.
  - Debt fell modestly compared to the pre-crisis period: median debt-asset ratio fell from 24.4 to 22.6 percent.
  - Large firms exhibited a striking increase in holdings of cash; smaller firms’ cash holdings remain roughly constant and fall below those of the large firms.
  - No substantial change in cash holdings across tradable/non-tradable sectors.
  - Medians are volatile for both large and small firms.

### Effective interest rates
- Effective interest rates paid by firms fell in the post-crisis period; the decline is modest compared to the decline in benchmark rates.
- Evidence suggests access to external credit may have been limited by rationing rather than price.
- Smaller firms faced higher effective interest rates than larger firms in both periods.
- Relative cost of debt for producers of traded vs non-traded outputs:
  - Prior to 2002: traded producers paid more than non-traded.
  - Subsequently: traded producers paid less than non-traded (difference is small).

### Empirical strategy (model specification)
- Baseline accelerator model (investment explained by change in sales, cash stock captures financing constraints):
  - Equation 1: Iit/Kit-1 = β1 ∆Sit/Kit-1 + β2 Cit/Kit-1 + νi + εit  (Eqn. 1)
  - Equation 2 adds interaction of cash with small-firm dummy: includes C * DUM_Small  (Eqn. 2)
  - Equation 3 adds log(REER) and interaction with tradable dummy: includes log(REER) + log(REER)*DUM_Tradable  (Eqn. 3)
  - Equation 4 adds firm-specific cost of debt (r)  (Eqn. 4)
- Measurement notes:
  - Investment measured using capital expenditures from cash flow statement.
  - Capital stock derived using a permanent inventory approach (accounts for depreciation).
  - Real exchange rate is CPI-based, effective rate; cost of borrowing is firm-specific effective interest rate constructed from debt and interest payments.
- Expectations on coefficients:
  - Coefficient on sales expected positive.
  - Coefficient on stock of cash expected positive (could reflect financing constraints or correlation with expected profitability).
  - Positive coefficient on cash × small-firm interaction required to conclude financing constraints operate more strongly for small firms.
  - Coefficient on cost of borrowing expected negative.
  - Coefficient on real exchange rate indeterminate; depreciation expected to favor tradable firms.

### Specification problems and endogeneity handling
- Key concerns:
  - Correlation of cash flow/stock with marginal productivity of capital complicates interpretation as financing constraint measure.
  - Simultaneous determination of investment, cash, and sales creates endogeneity (would bias standard panel estimators).
  - Dynamic determination of investment (adjustment costs, serial correlation) suggests potential need for dynamic estimators; inclusion of lags violates strict exogeneity for fixed effects.
- Remedies/trials employed:
  - Use of stock of cash (rather than flow) to reduce correlation with marginal productivity.
  - Instrumental variables (IV) using exogenous macro variables (GDP growth, commodity price indices, corporate tax and VAT receipts) as cross-checks.
  - System GMM estimators to address dynamics using internal instruments (noted caveat: approach assumes absence of structural breaks, an assumption violated in this dataset).
  - Recursive estimation to assess parameter stability.

### Empirical results — main findings
- General:
  - Change in sales (∆S) and cash stock (C) variables are signed as expected, significant, and of similar magnitude across most specifications.
  - Exception: cash variable becomes insignificant when time dummies and the firm-specific cost of capital variable are included (Equations 4 and 5 in Table 3).
  - R-squared statistics are low, but variables are always highly jointly significant.
- Magnitudes (interpretation from estimated coefficients):
  - Investment-capital ratio rises by between 0.02 and 0.03 percentage points for each percentage point increase in the ratio of the change in sales to capital.
  - Investment-capital ratio rises by between 0.05 and 0.08 percentage points for each point increase in the cash stock to capital ratio.
  - In comparable studies, cash-flow coefficients can be much larger; cash-flow vs cash-stock differences noted.
- Selected estimated coefficients from Table 2 (FE and IV specifications; p-values in parentheses):
  - ∆S: 0.030*** (0.001) in Eqn 1 (FE); other reported ∆S coefficients include 0.029** (0.030), 0.022** (0.035), 0.027*** (0.000), 0.028*** (0.001), 0.025*** (0.001), 0.025** (0.047), 0.025** (0.041), 0.033*** (0.000), 0.028*** (0.000).
  - C: 0.057* (0.056) in Eqn 1 (FE); other reported C coefficients include 0.033 (0.499), 0.869*** (0.001), 0.0775** (0.041), 0.084** (0.011), 0.472** (0.045), 0.050 (0.214), 0.053 (0.195), 0.057* (0.075), 0.091*** (0.002).
  - RER (log real exchange rate): 0.142*** (0.000) and 0.121*** (0.001) in relevant FE specifications.
  - Trade-RER interaction: -0.076* (0.060), -0.075* (0.057), -0.087** (0.034).
  - r (firm-specific cost of borrowing): -0.217** (0.023), -0.427*** (0.005).
  - Post-crisis interactions: ‘02‘06_r = 0.365** (0.033); ‘02’06_C_S = 0.261*** (0.000); ‘02’06_C shows 0.025 (0.258) and 0.097* (0.010) in reported variants.
- Real exchange rate effects:
  - A one percent increase in the real exchange rate (i.e., an appreciation) is correlated with an increase in the investment-capital ratio of about 0.15 percentage points.
  - For traded goods sector firms, the beneficial impact of an appreciation is reduced by about half (negative and significant Trade-RER interaction), consistent with cheaper imported inputs and easier servicing of dollar-denominated debt dominating for non-traded firms.
- Effective interest rate effects:
  - Increases in the effective cost of borrowing have a negative effect on investment: investment-capital ratio falls by about 0.2 of a percentage point for each percentage point increase in the effective interest rate.
  - Inclusion of an interactive post-2002 term for cost of capital increases magnitude and significance for the whole period; interaction term is positive and significant, yielding a weaker total effect for the post-crisis period. Interpretation: interest rate sensitivity of investment declined after 2002, possibly reflecting greater rationing of external finance.
- Financing constraints evidence:
  - Across specifications, the cash × small-firm interactive term is positive but insignificant (holds using both definitions of “small” reported).
  - Dummy for firms in the largest quartile yields similar insignificance, suggesting either lenders treat firms in the sample similarly or that most acute constraints may be faced by medium-sized firms not well-captured here.
  - However, including post-crisis interaction terms provides supportive evidence:
    - Cash holdings for small firms in the post-crisis period: positive and highly significant (holds for both definitions of small firms).
    - Including the real exchange rate yields a positive and significant coefficient for the interactive term on cash holdings in the post-crisis period.
  - Interpretation: evidence is not inconsistent with a significant worsening of financing constraints in the post-crisis period (2002–2006), even though baseline small-firm interaction insignificance tempers a definitive conclusion.

### Robustness and estimation notes
- Fixed effects (FE) and random effects produce similar results in many cases; Hausman test cannot reject null of non-systematic differences, but firm-level fixed effects are believed important.
- Instrumental variables (IV) estimation (instruments: non-fuel commodity prices and agricultural commodity prices; GDP growth, corporate tax and VAT receipts, other commodity indices tried) yields qualitatively similar results; instrumented cash coefficient can be several times larger than FE estimate.
- System GMM attempted to capture dynamics; caution advised due to structural break in middle of sample and unbalanced panel.
- Time dummies (year fixed effects) are jointly significant; time dummies included in all estimations except those with REER (colinearity).

*Source: _wp1003 - Annex Table 1 presents median values for the variables derived from the balance sheet. An*

### Annex 2 for a fuller discussion and a table of results). With this caveat, the results suggest

### _wp1003 - Annex 2 for a fuller discussion and a table of results). With this caveat, the results suggest

### GMM estimation approach and methodological caveats
- Designed for panels with relatively short time periods compared to cross-sectional units, “system” GMM handles endogeneity from inclusion of lags by transforming the data and adds information on levels to improve efficiency.
- Assumptions and transformations used:
  - Idiosyncratic disturbances are assumed uncorrelated across individuals; year dummies included.
  - The lagged, transformed series must be uncorrelated with individual fixed effects—an assumption analogous to cross-sectional units being in a steady-state; this appears invalid given the sample includes a major crisis generating a structural break.
  - Forward orthogonal deviations were used instead of differencing to avoid magnifying data loss from sample gaps.
  - Two-step estimation was used, with the Windmeijer correction applied to avoid small sample bias in standard errors.
  - Given high instrument count, additional results reported for the “collapsed” instrument set.
- Overidentification and instrument concerns:
  - Overidentification arises because the number of instruments rises exponentially with available lags.
  - Sargan-Hansen tests lose power as instrument count rises; Roodman (2006) suggests Hansen test p-values should lie between 0.1 and 0.25.
  - Hansen tests were preferred here because the structural break implies errors are likely heteroskedastic.

### GMM estimation results and robustness
- Dynamic specification evidence:
  - Results suggest a dynamic specification is appropriate, with positive and significant coefficients on the first and second lag of the dependent variable. (Elosegui et. al. (2006) do not find conclusive evidence.)
- Coefficients and significance:
  - Coefficient on the sales variable is significant (the lags are not), similarly signed and only slightly larger than in the FE estimation.
  - Coefficient on the cash variable is not significant, although signed as expected and larger than in the FE estimations.
  - The first lag of the dependent variable and the sales variable are always signed as expected and significant across specifications.
- Robustness to inclusion of other variables:
  - GMM results were generally not robust to inclusion of other variables, hinting at effects of the instrument overidentification problem.
  - Hansen statistics of overidentifying restrictions were usually close to (some passing, some failing depending on the exact specification) the upper bound suggested by Roodman (2006).
  - Coefficients for cash, cost of debt capital, or real exchange rate were volatile, mostly insignificant and sometimes changed signs depending on the exact specification.

### Recursive estimation and parameter stability
- Recursive estimation procedure:
  - Basic equation estimated recursively starting with subsample 1993–2001 and progressively adding one year.
- Recursive estimation findings:
  - For the shortest sample period the coefficient estimate on the stock of cash is not significant.
  - Coefficient estimates are otherwise always significant, with the appropriate signs, and the variables remain highly jointly significant as sample expands.
- Graphical results (described):
  - Recursive estimation plots shown for coefficient on change in sales and coefficient on cash stock with point estimates and 95% confidence intervals across sample years included.

### Conclusions from empirical and qualitative analysis
- Aggregate and sectoral changes post-crisis:
  - Investment process in Argentina changed in the post-crisis period.
  - Important relative price changes—in the real exchange rate, real interest rates, energy costs, the terms of trade and real wages—affected investment incentives and favored tradable and energy-intensive firms (essentially primary production and manufacturing) relative to their peers.
  - Aggregate investment rose to levels exceeding those seen during the 1990s as the economy rapidly recovered.
  - Data on the 500 largest firms shows traded goods sectors were more profitable than non-traded sectors after 2002.
  - Despite much lower real interest rates, aggregate credit fell substantially, indicating financial constraints facing firms may have increased.
  - Lower real wages favored labor-intensive activity; capital per worker did not return to previous peaks in the sample period.
- Firm-level empirical results (87 listed firms, 1993–2006):
  - Firms in sample are large by Argentine standards; paper does not draw conclusions about small and medium firms.
  - Investment model linked investment with changes in sales and the stock of cash, augmented with real exchange rate and cost of capital (firm-specific effective interest rate).
  - Interaction terms used to assess differing impacts across traded vs non-traded sectors and for smaller firms in the sample.
  - Basic findings are generally robust across different specifications and estimation techniques, subject to methodological caveats.
- Key substantive findings:
  - For large firms, real exchange rate appreciation is correlated with higher investment, suggesting benefits of cheaper imported inputs and greater ease of servicing dollar-denominated debt could outweigh export-return benefits.
    - This effect is only about half as large for firms with tradable outputs.
    - Consistent with observed lower overall investment displayed by tradable firms, but tradable sector firms invested relatively more.
  - Given the rise in aggregate investment, firms not captured in this sample—particularly small and medium size enterprises—must collectively have greatly increased investment.
  - Financing conditions:
    - Investment declines in the effective interest rate paid on firms’ debt.
    - Overall debt has fallen modestly in the post-crisis period.
    - Some evidence that interest rate sensitivity of investment decreased after the crisis, perhaps reflecting greater rationing of credit.
    - Strong evidence of financing constraints is not found for the full sample period, possibly reflecting insufficient variation in firm size across the sample.
    - Results are not inconsistent with financing constraints having worsened during 2002–2006.

### Data, sample construction, and descriptive statistics
- Sample coverage and construction:
  - Data covers a sample of 87 non-financial firms listed on the Argentine stock exchange, from 1993 to 2006, drawing from financial statements.
  - Sample is unbalanced; relatively few firms exist at both beginning and end of sample.
  - Gaps exist in the data for some firms around 2002.
  - Investment, sales and stock of cash and short term investments derived from balance sheet information.
  - Capital stock derived via permanent inventory approach, taking initial capital stock as property plant and equipment at earliest available date and assuming a depreciation rate of 8 percent.
  - Investment and capital stock variables deflated by an index of machinery and electrical equipment prices from INDEC WPI; sales and cash flow deflated by overall WPI.
  - Real exchange rate variable is the log of a trade-weighted, CPI-based multilateral index calculated by Fund staff.
  - Cost of capital for each firm is the effective rate derived from debt and interest payment data from financial statements.
- Data quality and outlier handling:
  - Debt and interest payment data are the weakest, with many “illogical” or extreme values.
  - After eliminating values less than zero or in excess of twice the contemporaneous prime lending rate, the number of observations is reduced by almost thirty percent (i.e. a substantively different sample).
  - Problems arise from 2002 crisis and accounting for restructurings, provisions, reversals, and contract redenomination/renegotiation varying widely across firms.
  - Six observations dropped due to very low and rapidly rising entries for property, plant and equipment in the first year of firms’ lives.
  - Three firms dropped entirely due to insufficient or inconsistent data.
- Annex Table 1: Descriptive statistics (medians and percentiles shown for full sample and subperiods)
  - Full sample (1994-2006), Pre-crisis (1994-2001), Post-crisis (2002-06)
  - I/K (847 Obs): 25th pcile 0.02; Median 0.06; 75th pcile 0.13
    - Pre-crisis (499 Obs) Median 0.08; Post-crisis (348 Obs) Median 0.04
    - Small (208 Obs) 25th 0.01; Median 0.04; 75th 0.08; Pre-crisis (123 Obs) Median 0.05; Post-crisis (85 Obs) Median 0.02
    - Large (212 Obs) 25th 0.04; Median 0.09; 75th 0.15; Pre-crisis (126 Obs) Median 0.11; Post-crisis (86 Obs) Median 0.04
    - Traded (608 Obs) 25th 0.02; Median 0.06; 75th 0.14; Pre-crisis (380 Obs) Median 0.07; Post-crisis (228 Obs) Median 0.04
    - Non-traded (239 Obs) 25th 0.03; Median 0.06; 75th 0.12; Pre-crisis (119 Obs) Median 0.10; Post-crisis (120 Obs) Median 0.04
  - ∆Sales/K (832 Obs): 25th pcile -0.10; Median 0.02; 75th pcile 0.13
    - Pre-crisis (491 Obs) Median 0.01; Post-crisis (341 Obs) Median 0.02
    - Small (196 Obs) 25th -0.26; Median 0.01; 75th 0.21; Pre-crisis (116 Obs) Median -0.04; Post-crisis (80 Obs) Median 0.05
    - Large (212 Obs) 25th -0.06; Median 0.01; 75th 0.09; Pre-crisis (126 Obs) Median 0.02; Post-crisis (86 Obs) Median 0.01
    - Traded (593 Obs) 25th -0.12; Median 0.02; 75th 0.15; Pre-crisis (372 Obs) Median 0.00; Post-crisis (221 Obs) Median 0.07
    - Non-traded (239 Obs) 25th -0.07; Median 0.01; 75th 0.09; Pre-crisis (119 Obs) Median 0.02; Post-crisis (120 Obs) Median 0.00
  - Cash/K (847 Obs): 25th pcile 0.02; Median 0.05; 75th pcile 0.17
    - Pre-crisis (499 Obs) Median 0.04; Post-crisis (348 Obs) Median 0.07
    - Small (208 Obs) 25th 0.02; Median 0.07; 75th 0.22; Pre-crisis (123 Obs) Median 0.07; Post-crisis (85 Obs) Median 0.07
    - Large (212 Obs) 25th 0.02; Median 0.05; 75th 0.13; Pre-crisis (126 Obs) Median 0.03; Post-crisis (86 Obs) Median 0.08
    - Traded (608 Obs) 25th 0.01; Median 0.06; 75th 0.19; Pre-crisis (380 Obs) Median 0.05; Post-crisis (228 Obs) Median 0.07
    - Non-traded (239 Obs) 25th 0.02; Median 0.05; 75th 0.13; Pre-crisis (119 Obs) Median 0.04; Post-crisis (120 Obs) Median 0.07
  - Cost of K (601 Obs): 25th pcile 0.07; Median 0.09; 75th pcile 0.12
    - Pre-crisis (378 Obs) Median 0.10; Post-crisis (223 Obs) Median 0.08
    - Small (113 Obs) 25th 0.08; Median 0.11; 75th 0.14; Pre-crisis (74 Obs) Median 0.12; Post-crisis (39 Obs) Median 0.10
    - Large (188 Obs) 25th 0.07; Median 0.08; 75th 0.10; Pre-crisis (111 Obs) Median 0.09; Post-crisis (77 Obs) Median 0.08
    - Traded (410 Obs) 25th 0.07; Median 0.10; 75th 0.13; Pre-crisis (273 Obs) Median 0.10; Post-crisis (137 Obs) Median 0.08
    - Non-traded (191 Obs) 25th 0.07; Median 0.08; 75th 0.11; Pre-crisis (105 Obs) Median 0.09; Post-crisis (86 Obs) Median 0.08

*Source: Annexes 1–2 and main text of the supplied content.*

### Annex Table 2. System GMM Estimations

### Annex Table 2. System GMM Estimations

### Main estimation results (coefficients and p-values)
- Investment (L1)
  - Eqn 1: 0.241*** (0.000)
  - Eqn 2: 0.260*** (0.000)
  - Eqn 3: 0.260*** (0.000)
- I (L2)
  - Eqn 1: 0.058* (0.073)
  - Eqn 2: 0.025 (0.417)
  - Eqn 3: 0.027 (0.400)
- Sales
  - Eqn 1: 0.039** (0.050)
  - Eqn 2: 0.025** (0.034)
  - Eqn 3: 0.024* (0.056)
- ∆S (L1)
  - Eqn 1: 0.006 (0.605)
  - Eqn 2: 0.010 (0.351)
  - Eqn 3: 0.009 (0.364)
- ∆S (L2)
  - Eqn 1: 0.0005 (0.958)
  - Eqn 2: 0.0008 (0.935)
  - Eqn 3: -0.001 (0.880)
- Cash stock
  - Eqn 1: 0.128 (0.448)
  - Eqn 2: -0.090 (0.122)
  - Eqn 3: -0.101* (0.078)
- C (L1)
  - Eqn 1: 0.000 (0.999)
  - Eqn 2: 0.062 (0.268)
  - Eqn 3: 0.061 (0.293)
- C (L2)
  - Eqn 1: 0.025 (0.417)
  - Eqn 2: 0.055 (0.101)
  - Eqn 3: 0.061 (0.110)
- Cash stock / small firms
  - Eqn 1: 0.070 (0.364)
  - Eqn 2: 0.076 (0.309)
- C_S (L1)
  - Eqn 1: -0.027 (0.568)
  - Eqn 2: -0.025 (0.611)
- C_S (L2)
  - Eqn 2: -0.089** (0.035)
  - Eqn 3: -0.087** (0.033)
- Log real exchange rate
  - Eqn 2: -0.707 (0.566)
- RER / tradable firms
  - Eqn 3: -0.019 (0.242)

### Model and sample diagnostics
- No. of observations: 656 (for Eqn 1, Eqn 2, and Eqn 3)
- Instrument count:
  - Eqn 1: 47
  - Eqn 2: 59
  - Eqn 3: 60
- Hansen p-value:
  - Eqn 1: 0.630
  - Eqn 2: 0.201
  - Eqn 3: 0.158
- Estimation method: Equations system-GMM results with the collapsed instrument matrix.
- Significance notation: *, **, and *** indicate significance at 10, 5 and 1 percent. P-values reported in parentheses.

### Specification, instruments, and robustness notes
- Variables added to the level equation: two lags of the dependent variable, sales, and cash stock.
- Estimations including the cost of external borrowing were weaker and are not reported.
- Instrument strategy:
  - Lags of the real exchange rate were included as exogenous instruments.
  - Lags of two and deeper for investment, sales and cash stock were included as “GMM-style” instruments.
  - Rationale: twice-lagged instruments are appropriate for endogenous variables, while a single lag will suffice for pre-determined variables (including the first lag of sales on the assumption it is predetermined rather than endogenous did not affect the results).
- Arellano-Bond serial correlation tests:
  - Tests reject a hypothesis of no first-order autocorrelation in the residuals (a consequence of the transformation of the data).
  - The absence of second-order autocorrelation cannot be rejected, suggesting no residual serial correlation is present in the errors.

*Source: _wp1003 - Annex Table 2. System GMM Estimations.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1003.pdf_