## WHY HAS BUSINESS INVESTMENT SLOWED DOWN IN AUSTRALIA?

## Source details

**Canonical URL:** [WHY HAS BUSINESS INVESTMENT SLOWED DOWN IN AUSTRALIA?](https://www.imf.org/-/media/files/publications/cr/2020/english/1ausea2020002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2020/english/1ausea2020002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2020/english/1ausea2020002.pdf.json)

---

### Introduction: scope and headline findings
- Non-mining business investment in Australia started declining around the time of the global financial crisis (GFC) and has remained subdued since, contributing to weaker productivity growth.
- The slowdown is broad-based across sectors, with a larger decline in machinery and equipment than in non-residential structures.
- Policy recommendations highlighted in the chapter: reduce domestic policy uncertainty (for example, in energy policy), ease credit constraints for small- and medium-sized enterprises (SMEs), incentivize R&D, and continue product market and tax reforms.
- Near-term risk flagged: a renewed escalation in U.S.-China trade tensions could spill over to investment and growth in Australia.

### Quantified contribution to productivity and investment trends
- Labor productivity growth (excluding mining):
  - Average 1995-96 to 2007-08: 2.3
  - Average 2012-13 to 2017-18: 1.1
  - Difference: -1.2
  - Contributions from:
    - Capital deepening: 1.2 → 0.5 (Difference: -0.8)
    - Multifactor productivity: 0.9 → 0.4 (Difference: -0.5)
    - Other: 0.2 → 0.3 (Difference: 0.0)
- Labor productivity growth (including mining):
  - Average 1995-96 to 2007-08: 2.3
  - Average 2012-13 to 2017-18: 2.0
  - Difference: -0.2
  - Contributions from:
    - Capital deepening: 1.3 → 1.2 (Difference: -0.1)
    - Multifactor productivity: 0.8 → 0.6 (Difference: -0.2)
    - Other: 0.2 → 0.3 (Difference: 0.1)
- The decline in capital accumulation (“capital shallowing”) accounts for almost two-thirds of the observed decline of 1.2 percentage points in labor productivity growth between the pre- and post-GFC periods.
- Real R&D investment has declined in recent years, causing the stock of R&D to depreciate and likely contributing to lower multifactor productivity growth.

### Long-term drivers — aggregate (accelerator) model findings
- Model approach: benchmark accelerator model (investment driven by past economic activity) augmented with:
  - global and Australia-specific economic policy uncertainty indices,
  - terms of trade,
  - real effective exchange rate (REER).
- Key determinants identified:
  - Economic activity (positive lagged effect).
  - Global policy uncertainty (negative effect on non-mining investment).
  - Australia-specific policy uncertainty (negative effect).
  - REER appreciation (negative effect).
  - Terms of trade (positive effect).
- Model performance:
  - Benchmark accelerator model: statistically significant positive impact of past activity but relatively poor fit to observed investment.
  - Augmented accelerator model: dramatically improved fit; the fitted investment-to-capital ratio tracks actual investment with a secular decline during 2012-16, a brief pickup around 2017, and a decline since.
- Selected coefficients and statistics (sample period 1998Q4–2019Q3; Newey-West standard errors):
  - ΔY(-1)/K(-1): Benchmark Model 0.1907 (0.3338); Augmented Model 0.2263 (0.228)
  - ΔY(-5)/K(-1): Benchmark Model 0.5453** (0.2571); Augmented Model 0.5677*** (0.1936)
  - ΔY(-6)/K(-1): Benchmark Model 0.4602* (0.2474); Augmented Model 0.4944*** (0.1697)
  - 1/K(-1): Benchmark Model -8449.604* (4349.279); Augmented Model -5139.88 (7387.49)
  - Term-of-Trade (Augmented Model): 0.000175*** (0.000041)
  - REER (-1) (Augmented Model): -0.000076* (0.000038)
  - Global Uncertainty (Augmented Model): -0.000023*** (0.000008)
  - Australia Uncertainty (Augmented Model): -0.000019** (0.000008)
  - Constant: Benchmark Model 0.0213*** (0.0018); Augmented Model 0.018** (0.0071)
  - Adjusted R-squared: Benchmark Model 0.051747; Augmented Model 0.571325
  - Number of Observations: 84 (both models)

- Decomposition (contributions to cumulative change from 2007Q2):
  - Rising global policy uncertainty has been a drag on non-mining investment since the mid-2010s.
  - Domestic policy uncertainty has also been a drag.
  - In the early 2010s, positive spillovers from the commodity boom (terms of trade) to non-mining investment were partly offset by REER appreciation.

### Firm-level analysis — financial constraints and uncertainty
- Data and methodology:
  - Panel regressions on publicly-listed non-mining corporations, 1994–2018.
  - Dependent variable: investment-to-lagged-capital ratio.
  - Explanatory variables (lagged): cost of debt, leverage (debt-to-asset), liquidity (current assets/current liabilities), Tobin’s Q (market/replacement value), firm-level uncertainty (std. dev. of weekly stock prices), expected profits (or sales).
- Main regression results (full sample):
  - Cost of Debt (-1): -0.2908*** (standard error 0.0934)
  - Leverage (-1): -0.0575*** (standard error 0.0128)
  - Liquidity (-1): 0.0054*** (standard error 0.0018)
  - Uncertainty (-1): -0.0343*** (standard error 0.0102)
  - Expected Profits (-1): 0.0064** (standard error 0.0027)
  - Sales (-1): 0.0103* (standard error 0.006)
  - R-squared: 0.0361
  - Sample Period: 1994-2018
  - Number of Observations: 8,019
- Interpretation:
  - Higher cost of debt, higher leverage, and greater firm-level uncertainty are associated with lower investment-to-capital ratios.
  - Greater liquidity and higher expected profits (or sales) are associated with higher investment-to-capital ratios.
  - Findings indicate that financial constraints matter for investment; firms with high debt-to-asset ratios and/or low liquidity invest less.
- Heterogeneity by firm size and age:
  - Smaller firms (below-median asset size): liquidity is a more important determinant of investment than in the full sample.
  - Younger firms (5 years old or younger): coefficient sizes for liquidity and leverage are larger than in the full sample, implying stronger dependence of investment on internal finance for younger firms.
- Caveat:
  - Survivorship bias: firms that ceased operating due to severe financial constraints are excluded, suggesting the regression estimates may understate the true effect of financial constraints.

### The role of financial constraints — intangible investment
- Coefficients on leverage and liquidity in regressions for intangible asset investment are "much larger" than for overall investment, indicating financial constraints are more binding for intangible investment.
- The coefficient on uncertainty is "also larger for intangible assets," suggesting wait-and-see behavior under uncertainty is more pronounced for investment in new technologies.
- Firm-level coefficients for intangible assets (selected):
  - Cost of Debt (-1), Intangible Assets: -0.3954** (0.2005)
  - Leverage (-1), Intangible Assets: -0.1509*** (0.0421)
  - Liquidity (-1), Intangible Assets: 0.011*** (0.0035)
  - Uncertainty (-1), Intangible Assets: -0.1195*** (0.0291)
  - Expected Profits (-1), Intangible Assets: 0.0153** (0.007)
- Model fit and sample size for intangible assets:
  - R-squared: 0.0518
  - Sample Period: 1994-2018
  - Number of Observations: 6,550

### Role of Structural Reforms: Cross-Country Analysis
- Product market reform can:
  - increase ease of entry of firms, increasing competitiveness and incentivizing investment;
  - reduce firms’ costs of adjusting capital stock.
- Panel regression results suggest that, on average, an episode of significant product market reforms could lead to an increase in the investment-to-GDP ratio of about 4 percentage point after three years.
- Australia-specific scope for reform (areas for improvement):
  - barriers to trade and investment;
  - public ownership;
  - administrative burden for start-ups;
  - simplification and evaluation of regulations.
- Further reforms could target online and digital businesses to ensure free entry and reduce or prevent oligopolistic behavior.

### Short-Term Drivers: Trade Policy Uncertainty (TPU) and Investment
- TPU is a major international driver of global uncertainty, quantified using Trade Policy Uncertainty indices based on news chatter in a database of over 650 million news articles.
- Methodology: local projection model (Jordà, 2005) estimates impact of a sustained rise in TPU (back to peak levels observed over 2018/19) on investment and corporate interest rate spreads across Asia and the Pacific, with ANZIMF (a DSGE model) used to simulate full impacts including spillovers.
- Estimated impacts for Australia (numeric findings):
  - Investment falls overall by over 0.7 percent on impact.
    - 0.4 percentage point of the investment fall is the direct impact from reduced business confidence.
    - The remainder is from higher corporate spreads pushing down investment.
  - Real GDP would reach a trough at almost 0.2 percent below the baseline when considering spillovers from trading partners.

### Key findings and policy implications
- Summary of drivers of the investment slowdown:
  - global policy uncertainty (often focused on international trade);
  - domestic policy uncertainty (including around energy policy, taxation, and R&D treatment);
  - the commodity price cycle;
  - the real exchange rate;
  - financial constraints, especially for smaller and younger firms;
  - across countries, the pace of product market reforms.
- Policy recommendations (priorities and specific measures):
  - Reduce domestic policy uncertainty:
    - set clear policy directions;
    - provide predictability in energy policy;
    - set out clear paths for tax policy and R&D incentives.
  - Continue government actions to encourage multilateralism to help address global trade policy uncertainty over time.
  - Further product market reform:
    - broaden the mandate of the Deregulation Taskforce over time to reduce barriers to trade and investment and the administrative burden for start-ups.
    - pursue competition legislation for intellectual property and reforms in online/digital sectors.
  - Ease financial constraints, especially for SMEs and younger firms:
    - ongoing measures include the Australian Business Securitization Fund (ABSF) and the Australian Business Growth Fund (ABGF), noted as "relatively small" with up to A$2 billion for the ABSF and A$1 billion for the ABGF; these could be expanded in size after a trial period and evaluation.
    - introduce policies and incentives to encourage banks to lend more to businesses (possibly by reducing banks’ concentration in mortgage lending).
    - promote venture capital investments (use of venture capital in Australia is less than half of the OECD average) to stimulate private sector R&D.
  - Promote R&D:
    - evaluate existing programs (R&D tax credit reform legislation; increased STEM funding through the 2019 budget and NISA starting in 2016) and expand effective programs.
    - faster implementation of the 2018 review Australia 2030: Prosperity Through Innovation could help promote R&D and innovation.
    - refine government R&D support to more effectively target younger firms.
  - Complete tax reform agenda:
    - conclude corporate income tax reform by reducing SME tax rates fully to 25 percent and extending tax rate cuts to all firms to maintain international competitiveness.
    - consider further support for new investment through tax measures, possibly including targeted investment allowances.
  - Monitor geopolitical risks, notably U.S.-China trade tensions, as an important near-term downside risk to investment.

### Annex: accelerator and firm-level model methodology (summary)
- Accelerator models:
  - Benchmark model dependent variable: private non-mining investment-to-lagged capital; regressor: private sector non-mining GVA.
  - Augmented model adds news-based global and Australia uncertainty indices, terms of trade, and REER (-1).
  - All variables are four-quarter backward-looking rolling averages.
  - Sample period: 1998Q4 to 2019Q3.
  - Newey-West heteroskedastic-and-autocorrelation-consistent standard errors used.
- Firm-level regressions:
  - Dependent variable: firm i’s capital expenditure (change) with key regressors including cost of debt, leverage, liquidity, firm-level uncertainty, expected profits (Tobin’s Q); one-year lags used.
  - Includes firm and time fixed effects; firm-clustered robust standard errors.
  - Data source: IMF Corporate Vulnerability Unit Database (Thomson Reuters Worldscope).
  - Sample period: 1994 to 2018; mining sector and public administration removed; outliers excluded (investment-to-lagged capital > 100 percent; cost of debt > 15 percent).
  - Subsamples: smaller firms (below-median asset size); younger firms (≤ five years old).

*Source: International Monetary Fund, "WHY HAS BUSINESS INVESTMENT SLOWED DOWN IN AUSTRALIA?", chapter prepared by Yosuke Kido, Dirk Muir, Masahiro Nozaki, Yu Ching Wong, Yong Sarah Zhou, and Sandile Hlatshwayo; February 6, 2020.*

### References __________________________________________________________________________ 15

### WHY HAS BUSINESS INVESTMENT SLOWED DOWN IN AUSTRALIA?

### Introduction: scope and headline findings
- Non-mining business investment in Australia started declining around the time of the global financial crisis (GFC) and has remained subdued since, contributing to weaker productivity growth.
- The slowdown is broad-based across sectors, with a larger decline in machinery and equipment than in non-residential structures.
- Policy recommendations highlighted in the chapter: reduce domestic policy uncertainty (for example, in energy policy), ease credit constraints for small- and medium-sized enterprises (SMEs), incentivize R&D, and continue product market and tax reforms.
- Near-term risk flagged: a renewed escalation in U.S.-China trade tensions could spill over to investment and growth in Australia.

### Quantified contribution to productivity and investment trends
- Labor productivity growth (excluding mining):
  - Average 1995-96 to 2007-08: 2.3
  - Average 2012-13 to 2017-18: 1.1
  - Difference: -1.2
  - Contributions from:
    - Capital deepening: 1.2 → 0.5 (Difference: -0.8)
    - Multifactor productivity: 0.9 → 0.4 (Difference: -0.5)
    - Other: 0.2 → 0.3 (Difference: 0.0)
- Labor productivity growth (including mining):
  - Average 1995-96 to 2007-08: 2.3
  - Average 2012-13 to 2017-18: 2.0
  - Difference: -0.2
  - Contributions from:
    - Capital deepening: 1.3 → 1.2 (Difference: -0.1)
    - Multifactor productivity: 0.8 → 0.6 (Difference: -0.2)
    - Other: 0.2 → 0.3 (Difference: 0.1)
- The decline in capital accumulation (“capital shallowing”) accounts for almost two-thirds of the observed decline of 1.2 percentage points in labor productivity growth between the pre- and post-GFC periods.
- Real R&D investment has declined in recent years, causing the stock of R&D to depreciate and likely contributing to lower multifactor productivity growth.

### Long-term drivers — aggregate (accelerator) model findings
- Model approach: benchmark accelerator model (investment driven by past economic activity) augmented with:
  - global and Australia-specific economic policy uncertainty indices,
  - terms of trade,
  - real effective exchange rate (REER).
- Key determinants identified:
  - Economic activity (positive lagged effect).
  - Global policy uncertainty (negative effect on non-mining investment).
  - Australia-specific policy uncertainty (negative effect).
  - REER appreciation (negative effect).
  - Terms of trade (positive effect).
- Model performance:
  - Benchmark accelerator model: statistically significant positive impact of past activity but relatively poor fit to observed investment.
  - Augmented accelerator model: dramatically improved fit; the fitted investment-to-capital ratio tracks actual investment with a secular decline during 2012-16, a brief pickup around 2017, and a decline since.
- Decomposition (contributions to cumulative change from 2007Q2):
  - Rising global policy uncertainty has been a drag on non-mining investment since the mid-2010s.
  - Domestic policy uncertainty has also been a drag.
  - In the early 2010s, positive spillovers from the commodity boom (terms of trade) to non-mining investment were partly offset by REER appreciation.

### Firm-level analysis — financial constraints and uncertainty
- Data and methodology:
  - Panel regressions on publicly-listed non-mining corporations, 1994–2018.
  - Dependent variable: investment-to-lagged-capital ratio.
  - Explanatory variables (lagged): cost of debt, leverage (debt-to-asset), liquidity (current assets/current liabilities), Tobin’s Q (market/replacement value), firm-level uncertainty (std. dev. of weekly stock prices), expected profits (or sales).
- Main regression results (full sample):
  - Cost of Debt (-1): -0.2908*** (standard error 0.0934)
  - Leverage (-1): -0.0575*** (standard error 0.0128)
  - Liquidity (-1): 0.0054*** (standard error 0.0018)
  - Uncertainty (-1): -0.0343*** (standard error 0.0102)
  - Expected Profits (-1): 0.0064** (standard error 0.0027)
  - Sales (-1): 0.0103* (standard error 0.006)
  - R-squared: 0.0361
  - Sample Period: 1994-2018
  - Number of Observations: 8,019
- Interpretation:
  - Higher cost of debt, higher leverage, and greater firm-level uncertainty are associated with lower investment-to-capital ratios.
  - Greater liquidity and higher expected profits (or sales) are associated with higher investment-to-capital ratios.
  - Findings indicate that financial constraints matter for investment; firms with high debt-to-asset ratios and/or low liquidity invest less.
- Heterogeneity by firm size and age:
  - Smaller firms (below-median asset size): liquidity is a more important determinant of investment than in the full sample, consistent with tighter financial constraints for smaller firms.
  - Younger firms (5 years old or younger): coefficient sizes for liquidity and leverage are larger than in the full sample, implying stronger dependence of investment on internal finance for younger firms.
- Caveat:
  - Survivorship bias: firms that ceased operating due to severe financial constraints are excluded, suggesting the regression estimates may understate the true effect of financial constraints.

### Policy implications distilled from the analysis
- Address global and domestic policy uncertainty to reduce wait-and-see behavior by firms (examples include clearer energy policy).
- Ease credit constraints for SMEs and younger firms to facilitate investment, especially in intangible assets where external financing frictions are acute.
- Incentivize R&D to rebuild the stock of intangible capital and support multifactor productivity growth.
- Continue product market and tax reforms to improve the investment environment and competitiveness.
- Monitor geopolitical risks, notably U.S.-China trade tensions, as an important near-term downside risk to investment.

*Source: International Monetary Fund, "WHY HAS BUSINESS INVESTMENT SLOWED DOWN IN AUSTRALIA?", chapter prepared by Yosuke Kido, Dirk Muir, Masahiro Nozaki, Yu Ching Wong, Yong Sarah Zhou, and Sandile Hlatshwayo; February 6, 2020. https://www.imf.org/-/media/files/publications/cr/2020/english/1ausea2020002.pdf*

### 13.      The role of financial constraints

### 13.      The role of financial constraints

### Financial constraints and intangible investment
- Coefficients on leverage and liquidity in firm-level panel regressions for intangible asset investment are "much larger" than in the analysis for overall investment, suggesting financial constraints may be more binding for intangible investment.
- The coefficient on uncertainty is "also larger for intangible assets," suggesting firms’ wait-and-see behavior under uncertainty is more pronounced for investment in new technologies.
- Implication: More restricted access to credit would lead affected firms to cut back investment particularly in intangible assets.

### Firm-level conclusions
- For all non-mining firms, in addition to expected profits and financial conditions (cost of debt), financial frictions (as evidenced by the significance of leverage and liquidity) and uncertainty play key roles in investment decisions.
- Both smaller and younger firms face more difficulties in accessing finance; these firms tend to have less capacity to self-finance than established, larger firms.
- Investment in intangible assets is more sensitive to financial conditions and uncertainty.

### Policy relevance (financial side)
- Easing financial constraints is important to support intangible and overall business investment, with special attention to SMEs and younger firms.

---

### D.   Role of Structural Reforms: Cross-Country Analysis

### Product market reform and investment
- Product market reform can: (i) increase ease of entry of firms thereby increasing market competitiveness and encouraging firms to invest more to remain viable; and (ii) reduce firms’ costs of adjusting their capital stock in response to changes in fundamentals (Alesina and others, 2005).
- Panel regression results suggest that, on average, an episode of significant product market reforms could lead to an increase in the investment-to-GDP ratio of about 4 percentage point after three years.

### Australia-specific scope for reform
- OECD data show Australia’s product market regulations are less restrictive than the average level for its OECD peers, but scope remains for improvement in:
  - barriers to trade and investment;
  - public ownership;
  - administrative burden for start-ups;
  - simplification and evaluation of regulations.
- Further reforms could target online and digital businesses to ensure free entry and reduce or prevent oligopolistic behavior.

---

### E.   Short-Term Drivers: Trade Policy Uncertainty and Investment

### Trade Policy Uncertainty (TPU) as a near-term risk
- Major international driver of global uncertainty: trade tensions between China and the United States, quantified using Trade Policy Uncertainty (TPU) indices (Hlatshwayo and others, 2019).
- TPU and other policy uncertainty indices are based on news chatter in a database of over 650 million news articles, with individual country indices.

### Quantifying TPU impacts (methodology summary)
- A local projection model (Jordà, 2005) is used to estimate the impact of a sustained rise in TPU (back to peak levels observed over 2018/19) on investment and corporate interest rate spreads across Asia and the Pacific, controlling for other factors including fiscal and monetary policy uncertainty.
- First-round impacts: investment (two years) and corporate spreads (one year) are fed into ANZIMF (a DSGE model) to simulate full impacts, including spillovers.

### Estimated impacts for Australia (numeric findings)
- Investment falls overall by over 0.7 percent on impact.
  - 0.4 percentage point of the investment fall is the direct impact from reduced business confidence.
  - The remainder is from higher corporate spreads pushing down investment.
- Real GDP would reach a trough at almost 0.2 percent below the baseline when considering spillovers from trading partners.

---

### F.   Key Findings and Policy Implications

### Summary of drivers of the investment slowdown
- Key drivers:
  - global policy uncertainty (often focused on international trade);
  - domestic policy uncertainty (including around energy policy, taxation, and R&D treatment);
  - the commodity price cycle;
  - the real exchange rate;
  - financial constraints, especially for smaller and younger firms;
  - across countries, the pace of product market reforms.

### Policy recommendations (priorities and specific measures)
- Reduce domestic policy uncertainty:
  - set clear policy directions;
  - provide predictability in energy policy;
  - set out clear paths for tax policy and R&D incentives.
- Continue government actions to encourage multilateralism to help address global trade policy uncertainty over time.
- Further product market reform:
  - broaden the mandate of the Deregulation Taskforce over time to reduce barriers to trade and investment and the administrative burden for start-ups.
  - pursue competition legislation for intellectual property and reforms in online/digital sectors.
- Ease financial constraints, especially for SMEs and younger firms:
  - ongoing measures include the Australian Business Securitization Fund (ABSF) and the Australian Business Growth Fund (ABGF), noted as "relatively small" with up to A$2 billion for the ABSF and A$1 billion for the ABGF; these could be expanded in size after a trial period and evaluation.
  - introduce policies and incentives to encourage banks to lend more to businesses (possibly by reducing banks’ concentration in mortgage lending).
  - promote venture capital investments (use of venture capital in Australia is less than half of the OECD average) to stimulate private sector R&D.
- Promote R&D:
  - evaluate existing programs (R&D tax credit reform legislation; increased STEM funding through the 2019 budget and NISA starting in 2016) and expand effective programs.
  - faster implementation of the 2018 review Australia 2030: Prosperity Through Innovation could help promote R&D and innovation.
  - refine government R&D support to more effectively target younger firms.
- Complete tax reform agenda:
  - conclude corporate income tax reform by reducing SME tax rates fully to 25 percent and extending tax rate cuts to all firms to maintain international competitiveness.
  - consider further support for new investment through tax measures, possibly including targeted investment allowances.

---

### Annex: Accelerator Models – Methodology and Selected Results

### Model setup and data
- Benchmark and augmented accelerator models estimated following IMF (2015) and Barkbu (2015).
- Benchmark model explains investment dynamics based on output: dependent variable is private non-mining investment-to-lagged capital; private sector non-mining GVA is used as regressor.
- Augmented model adds non-output drivers X_t: news-based global and Australia uncertainty indices (Baker and others, 2016); the terms of trade; and the REER (REER included with a one-quarter lag).
- All variables are four-quarter backward-looking rolling averages.
- Sample period: 1998Q4 to 2019Q3.
- Newey-West heteroskedastic-and-autocorrelation-consistent standard errors are used.

### Selected coefficients and statistics (Annex Table I.1)
- ΔY(-1)/K(-1):
  - Benchmark Model: 0.1907 (0.3338)
  - Augmented Model: 0.2263 (0.228)
- ΔY(-2)/K(-1):
  - Benchmark Model: 0.3409 (0.2973)
  - Augmented Model: 0.3421 (0.229)
- ΔY(-3)/K(-1):
  - Benchmark Model: 0.3537 (0.2676)
  - Augmented Model: 0.3101 (0.1883)
- ΔY(-4)/K(-1):
  - Benchmark Model: 0.2175 (0.261)
  - Augmented Model: 0.2635 (0.1802)
- ΔY(-5)/K(-1):
  - Benchmark Model: 0.5453** (0.2571)
  - Augmented Model: 0.5677*** (0.1936)
- ΔY(-6)/K(-1):
  - Benchmark Model: 0.4602* (0.2474)
  - Augmented Model: 0.4944*** (0.1697)
- 1/K(-1):
  - Benchmark Model: -8449.604* (4349.279)
  - Augmented Model: -5139.88 (7387.49)
- Term-of-Trade (Augmented Model): 0.000175*** (0.000041)
- REER (-1) (Augmented Model): -0.000076* (0.000038)
- Global Uncertainty (Augmented Model): -0.000023*** (0.000008)
- Australia Uncertainty (Augmented Model): -0.000019** (0.000008)
- Constant:
  - Benchmark Model: 0.0213*** (0.0018)
  - Augmented Model: 0.018** (0.0071)
- Adjusted R-squared:
  - Benchmark Model: 0.051747
  - Augmented Model: 0.571325
- Number of Observations:
  - Both models: 84
- Sample period reiterated: 1998Q4–2019Q3

*Italicized source attribution below provided per overlay requirements.*

*Source: IMF staff calculations (chapter "The role of financial constraints" from the provided PDF).*

### Annex II. Firm-Level Regressions – Methodology and Detailed

### Annex II. Firm-Level Regressions – Methodology and Detailed Results

### Methodology
- Panel regression model employed to analyze drivers of non-mining business investment:
  - 퐼퐼_{푖푖,푡푡+푘푘} − 퐼퐼_{푖푖,푡푡−1} = 훼훼_{푖푖} + 훾훾_{푡푡} + 훽훽_{푘푘} 푅푅_{푖푖,푡푡} + 휃휃_{푘푘} 푋푋_{푖푖,푡푡} + 휀휀_{i,푡푡}
- Dependent variable: firm i’s capital expenditure at time t (퐼퐼_{푖푖,푡푡}).
- Key regressors:
  - 퐾퐾_{푖푖,푡푡}: firm i’s capital stock.
  - 푋푋_{푖푖,푡푡}: set of firm-level variables (all explanatory variables included with a one-year lag).
    - cost of debt (interest rate expenditure-to-debt)
    - leverage (debt-to-asset ratio)
    - liquidity (current asset-to-current liability)
    - firm-level uncertainty (firm-level stock volatility)
    - expected profits (Tobin’s Q: sum of market value of equity and book value debt liability divided by book value of assets)
- Regression specification details:
  - Includes firm-level and time fixed effects.
  - Firm-clustered robust standard errors are estimated.
  - Robustness check: specification that includes sales (sales-to-asset ratio) instead of expected profits also estimated.
- For intangible asset investment analysis:
  - 퐼퐼_{푖푖,푡푡} denotes the change in intangible assets.
  - 퐾퐾_{푖푖,푡푡} denotes the stock of intangible assets.

### Data and Sample
- Data source: IMF Corporate Vulnerability Unit Database (based on Thomson Reuters Worldscope database).
- Sample period: 1994 to 2018.
- Exclusions:
  - Mining sector and public administration removed.
  - Outliers removed:
    - samples with investment-to-lagged capital of more than 100 percent.
    - observations with cost of debt of more than 15 percent.
- Data frequency: annual.
- Subsample definitions:
  - Smaller firms: firms with below-median asset size.
  - Younger firms: firms that are five years of age or younger.

### Firm-Level Regression Results (selected figures from Annex Table II.3: Total Investment and Intangible Assets)
- Dependent Variable: Investment-to-Lagged Capital
- Coefficients (standard errors in parentheses)
  - Cost of Debt (-1):
    - Total Investment: -0.2908*** (0.0934)
    - Intangible Assets: -0.3954** (0.2005)
  - Leverage (-1):
    - Total Investment: -0.0575*** (0.0128)
    - Intangible Assets: -0.1509*** (0.0421)
  - Liquidity (-1):
    - Total Investment: 0.0054*** (0.0018)
    - Intangible Assets: 0.011*** (0.0035)
  - Uncertainty (-1):
    - Total Investment: -0.0343*** (0.0102)
    - Intangible Assets: -0.1195*** (0.0291)
  - Expected Profits (-1):
    - Total Investment: 0.0064** (0.0027)
    - Intangible Assets: 0.0153** (0.007)
- Model fit and sample size:
  - R-squared:
    - Total Investment: 0.0361
    - Intangible Assets: 0.0518
  - Sample Period: 1994-2018 (both columns)
  - Number of Observations:
    - Total Investment: 8,019
    - Intangible Assets: 6,550
- Notes:
  - Firm-clustered robust standard errors are reported.
  - In the table, *, ** and *** indicate significant at the 1, 5, or 10 percent level, respectively.

### Annex III — Cross-Country Regression Methodology (brief)
- Panel local projection model focused on advanced economies used to estimate impact of product market reforms on business investment:
  - Dependent variable: non-mining business investment in country i at time t (in logarithms), 퐼퐼_{푖푖,푡푡}.
  - 푅푅_{푡푡}: dummy variable for product market reform (1 in year(s) when a reform takes place, 0 otherwise).
  - 푋푋_{푡푡}: set of control variables including contemporaneous and lagged variables of other structural reforms (such as reforms of unemployment benefits and employment protections), lagged product market reform dummies, and crisis event dummies.
  - Definition of product market reform events follows IMF (2016), identifying reform events based on OECD Economic Surveys and country-specific sources.
- Data coverage: sample of 18 advanced economies from 1990 to 2016.
- Estimation details:
  - Both country fixed effects and year fixed effects included.
  - Clustered robust standard errors are estimated.

*Source: IMF staff calculations.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2020/english/1ausea2020002.pdf_