## 1ausea2021002 — Reigniting Productivity Growth in Australia

## Source details

**Canonical URL:** [1ausea2021002 — Reigniting Productivity Growth in Australia](https://www.imf.org/-/media/files/publications/cr/2021/english/1ausea2021002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2021/english/1ausea2021002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2021/english/1ausea2021002.pdf.json)

---

### Overview and introduction
- Productivity growth slowed in many advanced economies pre-pandemic due to hysteresis from the global financial crisis and structural headwinds (waning of the ICT boom, slowdown in global trade, demographic changes).
- Australia experienced a marked deterioration in productivity dynamics despite avoiding a GFC recession and initially benefiting from a mining boom.
- Medium-term focus: productivity-enhancing investments and competition.
- Key priorities to reignite productivity growth:
  - Increase productivity-supporting investments in R&D and information and communication technology (ICT).
  - Renewed product market reforms to enhance competitive forces.
  - Reduce uncertainty; step up direct government R&D spending; incentivize university-business collaboration.
  - Use tax incentives to promote R&D investment, especially among smaller firms.
  - Enhance competition through deregulation and reducing entry barriers and financing constraints for SMEs.

### Australia’s aggregate and industry-level productivity performance
- Labor productivity and investment rates:
  - Average labor productivity growth 1990–2015: 1.6 percent.
  - Average labor productivity growth 2016–2019: about 0.4 percent.
  - FY2019/20 preliminary estimate for TFP growth: -0.7 percent.
  - Average investment rate 2010–2015: almost 27 percent of GDP.
  - Investment rate in 2019: about 23 percent of GDP.
- Utilization-adjusted industry-level findings (ABS KLEMS; Basu et al. (2006) approach):
  - Non-Mining Market Sectors (Solow residuals by five-year periods): 0.65; 0.32; 0.04; 0.45; 0.16.
  - Non-Mining Market Sectors (Utilization Adj. TFP by five-year periods): 0.81; 0.33; 0.22; 0.67; 0.36.
  - Mining Sector (Solow residuals by five-year periods): 0.83; -2.58; -2.42; -0.27; 1.11.
  - Mining Sector (Utilization Adj. TFP by five-year periods): 0.82; -2.54; -2.44; -0.28; 1.15.
  - A productivity slowdown in non-mining sectors is observed even when using utilization-adjusted measures; slowdowns are broad based (manufacturing, construction, accommodation and food services, wholesale trade).
  - Mining sector productivity picked up in recent years, but utilization-adjusted estimates for mining are highly uncertain.

### The role of declining R&D and ICT investment
- Trends and levels:
  - R&D investment: from about 3 percent of GDP in 2010 to less than 2.5 percent of GDP in 2019.
  - ICT investment: fell from a peak of 3.5 percent (2001–05 average) to below 2 percent of GDP in recent years.
- Composition of R&D spending:
  - Business R&D: fell from over 1.2 percent of GDP in 2009-10 to less than 1 percent of GDP in 2019-20.
  - Government R&D: declined by almost 40 percent as a share of GDP, from about 0.27 percent of GDP in 2010-11 to 0.17 percent of GDP in 2019-20.
  - University R&D: maintained at about 0.6 percent of GDP over the last decade.
- Non-R&D innovative activities:
  - Spending on non-R&D innovative investments (e.g., organization capital, process efficiency) is almost as large as R&D spending in Australia.
  - The share of firms undertaking innovative activities has increased.
- Caveats:
  - R&D is used broadly to include all investments in intellectual property, including business research expenditure, mineral exploration, artistic originals, etc.
  - Relying solely on R&D as a share of GDP can be misleading; non-R&D innovative investment and GDP developments may affect the ratio.

### Empirical evidence: Impact of R&D and ICT investment on TFP (cross-country industry-level)
- Data and strategy:
  - EU KLEMS data for 40 industries across 21 countries, 1995–2017.
  - Regression of sector-country-year TFP growth on lags (up to 5 years) of investment-to-value-added ratios for R&D, ICT, and other capital investment; controls include lagged productivity growth, country-time, country-sector, and sector-time fixed effects. Standard errors clustered two-way at country and industry level.
- Main estimated coefficients (selected exact values reported):
  - R&D Investment:
    - Lag 1: 0.0598 (insignificant; column 1)
    - Lag 3: 0.0386*** (column 2)
    - Lag 5: 0.0983*** (column 3)
  - ICT Investment:
    - Lag 1: 0.0739 (insignificant; column 4)
    - Lag 3: 0.0305* (column 5)
    - Lag 5: 0.0883** (column 6)
  - Other Investment:
    - Lag 1: 0.0143** (column 7)
    - Lag 3: 0.0054*** (column 8)
    - Lag 5: 0.0237* (column 9)
  - Lagged TFP growth:
    - L.TFP_gwth: values near -0.07 (all columns, significant at 1 percent).
    - L2.TFP_gwth: values near -0.04 to -0.05 (all columns, significant at 1 percent).
  - Observations by column: 9,570; 9,570; 8,640; 9,486; 9,486; 8,556; 9,486; 9,486; 8,556.
  - R-squared values reported: 0.319; 0.319; 0.330; 0.320; 0.320; 0.331; 0.320; 0.321; 0.331.
- Interpretation:
  - R&D and ICT investment are positively associated with TFP growth, with effects materializing over the medium term (lags of 3 to 5 years).
  - Coefficients on R&D and ICT are larger than on other investment, implying a 1 percentage point increase in R&D or ICT investment has a bigger medium-term positive effect on TFP growth than a comparable increase in other investment.
- Robustness (Annex II highlights):
  - Results robust across specifications, samples (excluding G-7, manufacturing-only, services-only), and when summing lags 1–5.
  - Selected robustness coefficients (preserve reported values):
    - Multiple-investment regression (Annex Table II.1): R&D Investment (lag 5) = 0.0668* (0.033); Other Investment (lag 5) = 0.0163 (0.014); Observations = 8,640; R-squared = 0.331.
    - ICT Investment (lag 5) = 0.0465** (0.020); Other Investment (lag 5) = 0.0197** (0.009); Observations = 8,556; R-squared = 0.331.
    - Combined R&D and ICT Investment (lag 5) = 0.0431*** (0.008); Other Investment (lag 5) = 0.0162 (0.012); Observations = 8,556; R-squared = 0.331.
    - Selected II.3 values: R&D Investment (lag 5) = 0.1022***; 0.1337**; 0.1435*** in different restricted samples. ICT Investment (lag 5) = 0.1117***; 0.1267; 0.0779*; 0.0888* in various columns.

### Closing the investment gap: illustrative impacts on TFP growth
- Impact estimates of raising Australian investment to peer benchmarks:
  - Raising R&D investment to the OECD median associated with increase in TFP growth of about 0.1 percentage point.
  - Raising ICT investment to the OECD median associated with increase in TFP growth of about 0.05 percentage point.
  - Increasing R&D investment to the OECD top 5 average can increase TFP growth by about 0.3 percentage point.
  - Increasing ICT investment to the OECD top 5 average can increase TFP growth by about 0.2 percentage point.
- These estimates are illustrative and indicate magnitudes involved.

### Determinants of innovative and intangible investment — cross-country and firm-level evidence
- Cross-country aggregate evidence:
  - Government support (tax credits, government-funded R&D) effective in stimulating R&D expenditure; uncertainty adversely affects R&D.
  - Regression (Equation 2) finding: an increase in R&D tax incentive of 0.1 percentage point of GDP (nearly doubling) associated with a boost in innovative investment of about 11-16 percent, depending on specification.
- Firm-level evidence (IMF CVU database, listed firms, 2001–2018, financial sector removed):
  - Firm-level regression (Equation 3) dependent variable: growth rate of intangible capital.
  - Selected reported coefficients (Annex Table III.1, preserve values):
    - Sales Growth = .2464*** (0.0609) in column (1); .2652*** (0.0519) in column (6).
    - Uncertainty = -0.0251 (0.0516) in column (1).
    - Sales Growth * Uncertainty = - .3318*** (0.1071) in column (1); - .3600** (0.0870) in column (6).
    - Uncertainty * Lagged Dependent Variable = 1.5171*** (0.0655) in column (1); 1.5090*** (0.0577) in column (6).
    - High Ext. Finance Dep. * R&D tax incentives (-1) = .3279*** (0.1094) in column (1); .3697*** (0.0878) in column (6).
    - Manufacturing * R&D tax incentives (-1) = 1.1048* (0.6241) in column (1); 0.9972* (0.5465) in column (6).
    - Small * R&D tax incentives (-1) = 1.0199*** (0.4211) in column (1); 0.9682** (0.3843) in column (6).
    - High Exp. Growth * R&D tax incentives (-1) = 0.2529*** (0.1221) in column (1); 0.3057*** (0.1016) in column (6).
  - Fit and sample:
    - R-squared around 0.7597 to 0.7633.
    - Number of observations: 4,006 in column (1); 5,435 in column (6).
  - Heterogeneity:
    - Increasing tax incentives by 0.1 percentage point of GDP on growth of intangible capital next year is about 10.2 percentage points stronger for SMEs.
    - Manufacturing and firms more dependent on external finance see larger increases in intangible capital in response to aggregate incentives.
    - Firms with higher Tobin’s Q increase intangible investment more in response to tax incentives.

### Effects of uncertainty on intangible investment
- Uncertainty reduces responsiveness of intangible investment to sales growth and increases persistence of intangible investment decisions.
- Quantitative examples:
  - Without uncertainty, 10 percent growth in sales would boost growth rate of intangible capital investment by 2.7 percentage points.
  - A 10 percentage points increase in the uncertainty measure (firm’s stock volatility) would shave the growth of intangible capital by 0.4 percentage point contemporaneously.

### Is declining competition contributing to lower productivity growth?
- Evidence of reduced competition in Australia (consistent with global trends):
  - Concentration (Orbis): median concentration ratio increased from about 74.5 percent in 2013 to 77 percent in 2018; concentration ratios are above the median for advanced economies.
  - Markups: sales-weighted average markups from Worldscope rising in Australia (De Loecker and Eeckhout, 2020).
  - Entry and exit: entry rate lower than pre-GFC levels; exit rate declining and low relative to peers; share of “zombie” firms increasing.
  - Administrative-data studies (BLADE): concentration ratios and markups trended upwards since early 2000s with reduced reallocation toward more productive firms.
- Interpretation caveats:
  - Increases in concentration may coexist with productivity growth if driven by expansion of productive industries.

### Outlook for productivity growth — risks and upside
- Downside risks:
  - Subdued R&D and slow recovery of R&D investment post-pandemic could undermine medium-term productivity.
  - Low exit of firms, partly due to pandemic support, may delay reallocation and harm aggregate productivity.
  - Shortage of skilled workers due to border closures and skill erosion could hinder innovation.
  - Deterioration in education quality or disruptions during lockdowns could hurt long-term productivity.
  - Skill erosion from long-term unemployment could reduce human capital accumulation.
- Upside potential:
  - Acceleration of digitalization could boost ICT-related investment and productivity.
  - Resource reallocation toward more productive sectors following structural pandemic-induced changes could lift medium-term productivity.
  - Faster digital technology adoption by Australian firms could generate positive spillovers from frontier markets.

### Policy considerations and recommendations
- Promote R&D and ICT investment:
  - Use targeted tax incentives (with special focus on smaller firms and young innovators); monitor and reduce administrative burdens.
  - Step up direct government R&D spending, including direct funding of business R&D in priority areas.
  - Incentivize university-business collaboration; encourage managerial focus on innovation and integration into global value chains.
  - Swiftly implement the A$1.2 billion Digital Economy Strategy to build skills and infrastructure for digitalization.
- Macroeconomic stabilization and uncertainty reduction:
  - Reduce policy and economic uncertainty (e.g., quick vaccine rollout, integrated energy and climate policies) to support innovative investment.
- Enhance competition and SME financing:
  - Continue product market deregulation, streamline administrative burdens for start-ups, simplify regulations.
  - Alleviate SME financing constraints: improve matching between businesses and investors, financial literacy, reduce SME risk weights to lower interest spreads.
  - Deepen venture capital markets via expanded government-sponsored or co-investment funds and remove barriers to early-stage investment.
  - Expand national Automatic Mutual Recognition of Occupational Registrations across states/territories to reduce entry barriers and improve labor mobility.
- Labor market, education, and health:
  - Scale up active labor market policies (ALMPs), training in digital skills, and job search assistance; review JobMaker Hiring Credit effectiveness.
  - Improve teacher training and student outcomes to address declining PISA performance and high dispersion in scores.
  - Address mental health to reduce productivity losses; support National Mental Health and Suicide Prevention Plan.
- FDI and infrastructure:
  - Provide clear guidance on the national security test for foreign investment and use it judiciously to preserve FDI benefits.
  - Continue infrastructure spending to close gaps in electricity, telecoms and transportation.

_Italic: Source: IMF staff chapter "Reigniting Productivity Growth in Australia" (prepared by Yosuke Kido and Siddharth Kothari), November 4, 2021._

### 1. Utilization-Adjusted Productivity Growth Rates for Mining and Non-Mining Sectors  4

### 1. Utilization-Adjusted Productivity Growth Rates for Mining and Non-Mining Sectors

### Overview and introduction
- Productivity growth had slowed significantly in many advanced economies prior to the pandemic; causes include hysteresis from the global financial crisis and structural headwinds (waning of the ICT boom, slowdown in global trade, demographic changes).
- Australia experienced a marked deterioration in productivity dynamics in recent years despite avoiding a GFC recession and benefiting initially from a mining boom.
- The paper takes a medium-term view focusing on productivity-enhancing investments and competition as drivers of Australia’s productivity slowdown.
- Key priorities identified to reignite productivity growth:
  - Increase productivity-supporting investments in R&D and information and communication technology (ICT).
  - Renewed product market reforms to enhance competitive forces.
  - Reduce uncertainty, step up direct government R&D spending, incentivize university-business collaboration.
  - Use tax incentives to promote R&D investment, especially among smaller firms.
  - Enhance competition through deregulation and reducing entry barriers and financing constraints for SMEs.

### Australia’s aggregate and industry-level productivity performance
- Labor productivity growth:
  - Average growth 1990–2015: 1.6 percent.
  - Average growth 2016–2019: about 0.4 percent.
  - FY2019/20 preliminary estimate for TFP growth: -0.7 percent (driven by COVID disruptions).
- Investment rates:
  - Average investment rate 2010–2015: almost 27 percent of GDP.
  - Investment rate in 2019: about 23 percent of GDP.
- Findings from utilization-adjusted industry-level analysis (ABS KLEMS data; utilization-adjusted productivity following Basu and others (2006)):
  - Utilization-adjusted measures control for capital utilization and labor hoarding (labor effort proxied by change in hours per worker).
  - Table 1 (utilization-adjusted productivity growth rates, average growth rates, percent) reports:
    - Non-Mining Market Sectors:
      - Solow residuals by five-year periods: 0.65; 0.32; 0.04; 0.45; 0.16
      - Utilization Adj. TFP by five-year periods: 0.81; 0.33; 0.22; 0.67; 0.36
    - Mining Sector:
      - Solow residuals by five-year periods: 0.83; -2.58; -2.42; -0.27; 1.11
      - Utilization Adj. TFP by five-year periods: 0.82; -2.54; -2.44; -0.28; 1.15
  - A productivity slowdown in non-mining sectors is observed even when using utilization-adjusted measures; slowdowns are broad based (manufacturing, construction, accommodation and food services, wholesale trade).
  - Mining sector productivity picked up in recent years, though there is high uncertainty around utilization-adjusted estimates for mining.

### The role of declining R&D and ICT investment
- R&D and ICT investments have declined in Australia and are below OECD medians:
  - R&D investment: from about 3 percent of GDP in 2010 to less than 2.5 percent of GDP in 2019.
  - ICT investment: fell from a peak of 3.5 percent (2001–05 average) to below 2 percent of GDP in recent years.
- Composition of R&D spending:
  - Business R&D: fell from over 1.2 percent of GDP in 2009-10 to less than 1 percent of GDP in 2019-20.
  - Government R&D: declined by almost 40 percent as a share of GDP, from about 0.27 percent of GDP in 2010-11 to 0.17 percent of GDP in 2019-20.
  - University R&D: maintained at about 0.6 percent of GDP over the last decade.
- Non-R&D innovative activities:
  - Spending on non-R&D innovative investments (e.g., organization capital, process efficiency) is almost as large as R&D spending in Australia.
  - The share of firms undertaking innovative activities has increased.
- Noted caveats:
  - R&D is used broadly to include all investments in intellectual property, including business research expenditure, mineral exploration, artistic originals, etc.
  - Relying solely on R&D as a share of GDP can be misleading; non-R&D innovative investment and GDP developments may affect the ratio.

### Empirical evidence: Impact of R&D and ICT investment on TFP (cross-country industry-level)
- Data and empirical strategy:
  - Uses EU KLEMS data for 40 industries across 21 countries, 1995–2017.
  - Regression specification (equation 1) regresses sector-country-year TFP growth on lags of investment-to-value-added ratios for R&D, ICT, and other capital investment, controlling for lagged productivity growth and fixed effects: country-time, country-sector, and sector-time fixed effects.
  - Standard errors clustered two-way at country and industry level.
  - Investment variables are lagged up to 5 years to address timing of effects.
  - Recognized limitations: residual endogeneity concerns may remain; lagging investment reduces but may not eliminate such concerns.
- Main empirical findings (Table 2 summary, coefficients reported exactly as in source):
  - R&D Investment:
    - Lag 1: 0.0598 (insignificant; column 1)
    - Lag 3: 0.0386*** (column 2)
    - Lag 5: 0.0983*** (column 3)
  - ICT Investment:
    - Lag 1: 0.0739 (insignificant; column 4)
    - Lag 3: 0.0305* (column 5)
    - Lag 5: 0.0883** (column 6)
  - Other Investment (non-R&D, non-ICT):
    - Lag 1: 0.0143** (column 7)
    - Lag 3: 0.0054*** (column 8)
    - Lag 5: 0.0237* (column 9)
  - Lagged TFP growth coefficients (reported across columns):
    - L.TFP_gwth: values near -0.07 (all columns, significant at 1 percent)
    - L2.TFP_gwth: values near -0.04 to -0.05 (all columns, significant at 1 percent)
  - Observations and R-squared:
    - Observations reported per column: 9,570; 9,570; 8,640; 9,486; 9,486; 8,556; 9,486; 9,486; 8,556
    - R-squared values reported: 0.319; 0.319; 0.330; 0.320; 0.320; 0.331; 0.320; 0.321; 0.331
  - Interpretation:
    - R&D and ICT investment are positively associated with TFP growth, with effects materializing over the medium term (lags of 3 to 5 years).
    - Coefficients on R&D and ICT are larger than on other investment, indicating that a 1 percentage point increase in R&D or ICT investment has a bigger medium-term positive effect on TFP growth than a comparable increase in other types of investment.
  - Robustness:
    - Results hold under several robustness checks (see Annex II in source).

### Key implications and policy considerations (as highlighted in the chapter)
- To address the productivity slowdown, policy actions include:
  - Increase R&D and ICT investment through a mix of measures:
    - Tax incentives targeted to promote R&D, especially for smaller firms.
    - Reduce uncertainty to encourage private investment.
    - Step up direct government spending on R&D.
    - Incentivize university-business collaboration.
  - Enhance product market competition:
    - Continue deregulation.
    - Promote efficient resource allocation by reducing entry barriers and financing constraints for SMEs.
  - Recognize the timing of investment benefits:
    - R&D and ICT investments have stronger positive associations with TFP growth over medium-term horizons (lags of 3–5 years).

*Source: IMF staff chapter "Reigniting Productivity Growth in Australia" (prepared by Yosuke Kido and Siddharth Kothari), November 4, 2021.*

### 14.      Closing the investment gap in R&D and ICT investment between Australia and the

### 14.      Closing the investment gap in R&D and ICT investment between Australia and the

### Impact of closing R&D and ICT investment gaps on TFP growth
- Raising R&D investment in Australia to the OECD median can be associated with an increase in TFP growth of about 0.1 percentage point.
- Raising ICT investment in Australia to the OECD median can be associated with an increase in TFP growth of about 0.05 percentage point.
- Increasing R&D investment to the OECD top 5 average can increase TFP growth by about 0.3 percentage point.
- Increasing ICT investment to the OECD top 5 average can increase TFP growth by about 0.2 percentage point.
- These estimates are illustrative and provide a sense of magnitudes involved.

### Determinants of innovative investment: cross-country aggregate evidence
- Empirical literature highlights effectiveness of government support (tax credits, government-funded R&D) in stimulating R&D expenditure (Hall and Van Reenen (2000); Becker (2014); Bloom and others (2002); Hussinger (2008); Cerulli and Poti (2012)).
- Adverse effects of uncertainty on R&D investments are documented (Bloom (2007); Aghion and others (2012)).
- Simple regression model (Equation 2) specification:
  - Dependent variable: log-scaled real business R&D (푅푅푅푅푖푖,j,푡푡).
  - Key regressors: output gap (퐺퐺퐺퐺퐺퐺푖푖,푡푡−1) and R&D tax incentives in percent of GDP (퐼퐼퐼퐼퐼퐼퐼퐼퐼퐼퐼퐼퐼퐼푖푖,푡푡−1).
- Estimated effect of tax incentives:
  - An increase in R&D tax incentive of 0.1 percentage point of GDP (nearly doubling) would be associated with a boost in innovative investment of about 11-16 percent, depending on specification.

### Determinants of intangible investments: Australian firm-level evidence
- Firm-level model (Equation 3) specification:
  - Dependent variable: growth rate of intangible capital for firm i at time t (퐼퐼퐼퐼퐺퐺푖푖,푡푡).
  - Key regressors include growth rate of sales (훥훥푆푆푆푆푆푆 퐼퐼푆푆푖푖,푡푡), firm-level uncertainty proxied by volatility in weekly stock returns (휎휎푖푖,푡푡), lagged government tax incentives as a share of GDP (퐼퐼퐼퐼퐼퐼퐼퐼퐼퐼퐼퐼퐼퐼푡푡−1) interacted with firm-characteristic dummies.
  - Firm-characteristic dummies: High External Finance Dependence (above median), Manufacturing, Small (asset size below 25th percentile), High Expected Growth (Tobin’s Q above median).
  - Data: annual Australian firm-level data from IMF Corporate Vulnerability Unit Database (Thomson Reuters Worldscope), 2001 to 2018, financial sector removed; database covers listed firms only.
- Core firm-level regression findings:
  - Sales Growth coefficients: .2464*** to .2510*** across specifications (standard errors reported in source).
  - Uncertainty (un-interacted) coefficients: negative but not consistently significant (examples: -0.0251, -0.0333, -0.0230, -0.0360, -0.0186).
  - Sales Growth * Uncertainty interaction: -.3318*** to -.3358***, indicating higher uncertainty reduces responsiveness of intangible investment to sales growth.
  - Uncertainty * Lagged Dependent Variable: 1.5171*** to 1.5212***, indicating increased persistence of intangible investment under uncertainty.
  - High External Finance Dependence * RD tax incentives (-1): .3279*** and .3267***.
  - Manufacturing * RD tax incentives (-1): 1.1048* and 1.1420*.
  - Small * RD tax incentives (-1): 1.0199*** and 1.1134***.
  - High Expected Growth * RD tax incentives (-1): 0.2529*** and 0.2823***.
  - Lagged dependent variable values effectively reported as -.0000 with negligible Nickell bias given time series length.
  - R-squared values around 0.7597 to 0.7623; sample period 2001-2018; number of observations 4,006.
- Heterogeneity in tax-incentive effects:
  - Increasing tax incentives by 0.1 percentage point of GDP (nearly doubling) on the growth of intangible capital next year is about 10.2 percentage points stronger for SMEs.
  - Manufacturing sector and firms more dependent on external financing see a bigger increase in intangible capital when aggregate incentives increase.
  - Firms with higher expectations for growth (higher Tobin’s Q) increase intangible investment more in response to government tax incentives than less viable firms.

### Effects of uncertainty on intangible investment
- Uncertainty tends to make intangible investment less responsive to changes in business situations and makes firms reluctant to change investment plans (intangible investment becomes more persistent) (Bloom (2007)).
- Quantitative example from the source:
  - Without uncertainty, 10 percent growth in sales would boost the growth rate of intangible capital investment by 2.7 percentage points.
  - A 10 percentage points increase in the uncertainty measure (the firm’s stock volatility) would shave the growth of intangible capital by 0.4 percentage point contemporaneously.

### Is declining competition contributing to lower productivity growth?
- Theory: Relation between productivity and competition can be positive or non-monotonic; examples include inverted-U-shaped relation (Aghion and others (2005)).
- Multiple measures point to reduced competition in Australia, consistent with global trends and recent administrative-data literature (Bakhtiari, 2020; Hambur, 2021):
  - Concentration (Orbis data, operating revenue ratio of top 4 firms relative to top 10 firms across 2-digit industries):
    - Median concentration ratio in Australia increased from about 74.5 percent in 2013 to 77 percent in 2018.
    - Concentration ratios in Australia are above the median for advanced economies.
    - Caution: Orbis data coverage differs across countries and is limited in Australia to large, listed firms.
  - Markups:
    - Sales-weighted average markups from Worldscope dataset have been rising in Australia, suggesting deterioration in competition (De Loecker and Eeckhout, 2020).
    - Increase in markups in Australia is in line with global trends.
  - Firm entry and exit rates:
    - Entry rate of new firms lower than pre-global financial crisis levels though still relatively high internationally.
    - Exit rate of firms has been on a declining trend and is low compared to peer advanced economies.
    - Cross-country study shows the share of “zombie” firms in Australia has been increasing, associated with weak productivity growth (Banerjee and Hoffman, 2020).
- Evidence from high-quality administrative data:
  - Hambur (2021) using BLADE (universe of Australian firms) finds concentration ratios and markups trended upwards since early 2000s, with evidence of reduced reallocation of resources toward more productive firms (Andrews and Hansell, 2021).
  - Bakhtiari (2021a) finds increase in aggregate concentration with heterogeneity across sectors and some concentration driven by technological change.
- Notes on interpretation:
  - Literature highlights difficulty in interpreting aggregate trends in concentration; increases in concentration may coexist with productivity growth if driven by expansion of productive industries (Ganapati (2021); Covarrubias and others (2020); Rossi-Hansberg and others (2021)).

### Outlook for productivity growth
- Outlook is highly uncertain, especially given unprecedented nature of COVID-19 shock (Bannister and others (2020); IMF (2021)).
- Downside risks identified:
  - Subdued R&D and slow recovery of R&D investment after the pandemic could undermine medium-term productivity growth.
  - Low exit of firms, in part due to strong government policy support after the pandemic, may delay reallocation and have unintended adverse effects on aggregate productivity in the medium term (Productivity Commission 2021a).
  - Shortage of skilled workers due to border closures and skill erosion could hinder innovation; Australia relies more on foreign labor in some high-productivity sectors relative to other advanced economies.
  - Deterioration in quality of education or disruptions during lockdowns could hurt education outcomes and long-term productivity (Black and Lynch, 1996; Chevalier and others, 2004; Fernald and Li, 2021).
  - Skill erosion from long-term unemployment could further hinder human capital accumulation and reduce productivity.

_Italic: Source: IMF staff estimates and analysis as presented in the cited content unit._

### 25.      On the upside, acceleration of digitalization and resource reallocation toward more

### 1ausea2021002 - 25.      On the upside, acceleration of digitalization and resource reallocation toward more

### Upside: digitalization and resource reallocation could boost productivity
- The pandemic has propelled investment in digital technologies globally (IMF 2021).
- ICT-related investment is associated with productivity growth; positive spillovers from frontier markets could lift Australia’s medium-term productivity growth.
- Faster take up of digital technologies by Australian firms could also boost productivity.
- Resource reallocation from low productivity sectors toward more productive sectors could occur as a result of structural changes induced by the pandemic (Bannister and others, 2020).
- Productivity Commission (2021a) reports that labor shifts from some services to more productive sectors boosted labor productivity during the pandemic; however, such effects are likely to revert once restrictions are lifted, like they did after the previous lockdowns in 2020.

### Policy considerations — overarching goal
- Strong policy actions to boost productivity growth are essential to raise living standards.
- Many factors that constrain productivity in Australia pre-date the COVID-19 crisis; a strong structural reform push is essential for reigniting productivity growth.

### Promoting productivity-enhancing investments in R&D and ICT
- The R&D tax incentive:
  - Annual cost of about 0.1 to 0.2 percent of GDP to the budget.
  - 2018-19 budget changes aimed to improve targeting by supporting smaller companies while refocusing support for larger companies to higher intensity R&D; implementation delays potentially held back R&D investment.
  - 2020-21 budget amended the incentive, making it more generous, which should support R&D investment going forward, especially for smaller firms.
  - Recommendation: Monitor the impact of these changes and work towards better targeting incentives to young innovative firms, including reducing administrative burden of the tax incentive.
- Other innovation policy:
  - Scope for scaling up government spending in R&D (which is below that of peers), including directly funding business R&D in priority areas.
  - While higher-education R&D spending is relatively high, further incentivizing university-business collaboration is possible.
  - Encourage greater managerial focus on innovation (e.g., government’s Entrepreneurs Program) and greater integration into global value chains, including through foreign direct investment.
- Promoting ICT investment:
  - Swift implementation of the A$1.2 billion Digital Economy Strategy is essential to build skills and infrastructure for digitalization, including opportunities for regions and SMEs.

### Macroeconomic stabilization and uncertainty
- Macroeconomic stabilization promotes innovative investment because innovative investment is susceptible to uncertainty.
- A quick vaccine rollout reduces need for lockdowns and reduces uncertainty, supporting intangible investment.
- R&D investment tends to be more susceptible to long run uncertainty (Barrero and others, 2017); a more integrated approach to energy and climate change policies would help reduce policy uncertainty and catalyze innovative investment.

### Enhancing product market competition and SME financing
- Australia compares favorably on product market efficiency, but further reforms can support greater competition and productivity growth.
- Further product market deregulation:
  - Scope for improvements in some areas, such as streamlining administrative burdens for start-ups and simplifying regulations.
  - Recent initiatives—digitization of regulatory procedures and insolvency reforms for SMEs—are welcome.
- Alleviating financing constraints of SMEs:
  - Small firms tend to be riskier and face larger uncertainties and capital constraints; SMEs in Australia often have difficulties accessing finance and hold relatively more cash.
  - OECD cross-country data suggests loan interest rate spreads between SMEs and large firms in Australia are relatively wide.
  - Measures: improve matching efficiency between business and investor needs, improve financial literacy among SMEs, provide financial training.
  - Proposed reduction in risk weights for SMEs, from relatively high levels, would help reduce interest spread between large firms and SMEs.
  - Government initiatives that facilitate SME lending include the SME Recovery Loan Scheme, the Australian Business Growth Fund, and the Australian Business Securitisation Fund.
- Promoting venture capital (VC):
  - VC in Australia has been smaller than in many advanced economies, especially at early stages.
  - Government programs: Venture Capital Limited Partnerships and Early Stage Venture Capital Limited Partnership; VC funding continued to grow through the COVID shock.
  - Scope to further deepen VC markets: expand government-sponsored funds or co-investment funds and remove potential barriers to investment to improve young firms’ access to finance and promote productivity growth.
- Recognition of occupational licenses:
  - Expand the national Automatic Mutual Recognition of Occupational Registrations scheme (currently in place in New South Wales, Victoria, the Australian Capital Territory and the Northern Territory) to more occupations and across all states and territories to reduce entry barriers and improve labor mobility.

### Additional reforms to further boost productivity
- Education:
  - Australia has generally performed better than peers in PISA, but average PISA scores have declined and Australia now trails the OECD average in Math.
  - Evidence of high dispersion in PISA scores points to inequalities in the education system (OECD, 2021).
  - Reforms aimed at improving teacher training and student outcomes can support long-term productivity growth.
- Labor market policies:
  - Pandemic measures included JobKeeper Payments and wage subsidies for apprentices and trainees; the JobMaker Hiring Credit has had relatively minor role.
  - Scope to scale up active labor market policies (ALMPs), which have been relatively small compared to other advanced economies, to address elevated long-term unemployment and promote human capital accumulation.
  - Training, especially in digital areas, and job search assistance can help reallocation of workers.
  - JobTrainer Fund supports free or low-fee training in areas of high labor demand.
  - New Employment Services Model commences from July 2022 to connect job seekers with employment opportunities.
  - Given very low take-up, effectiveness of JobMaker Hiring Credit should be reviewed; parameters such as scope, length, and benefit level can be recalibrated depending on labor market conditions. Consider including disadvantaged workers such as the long term unemployed.
- Mental health:
  - Poor mental health is associated with weak productivity performance and absenteeism.
  - A considerable share of Australians reported having experienced a mental disorder; economic costs of mental illness are estimated to be sizeable (Productivity Commission 2020a).
  - Addressing mental illness (reduce job stress, workplace bullying, adverse impacts from deterioration in job conditions) would help productivity and wellbeing.
  - Recent government initiatives including National Mental Health and Suicide Prevention Plan are welcome.
- FDI regime:
  - FDI inflows into Australia have traditionally been higher than the OECD average and have supported investment with potential productivity spillovers.
  - Australia has a relatively open FDI regime on equity ownership but more onerous screening and approval restrictions.
  - Recent changes introduced a national security test requiring approval for foreign investments in ‘sensitive national security business’, regardless of investment value.
  - Recommendation: provide clear policy guidance on the national security test and use it judiciously (e.g., excluding risks attenuable through other policies like competition policy) to keep FDI regime simple, transparent, and supportive of productivity growth.
- Infrastructure:
  - Infrastructure gaps remain in electricity, telecoms and transportation.
  - Recent increase in infrastructure spending is welcome and should help close the infrastructure gap.

*International Monetary Fund — Australia country chapter excerpts*

### Chapter 2

### 1ausea2021002 - Chapter 2

### Annex I. Utilization-Adjusted Productivity Measures — methodology and key findings
- Purpose: Estimate utilization-adjusted productivity at industry level to remove cyclical influence of input utilization (capital and labor) on Solow residuals (TFP).
- Main approach (following Basu and others (2006)):
  - Production function for gross output Y_{i,t} specified with capital utilization KUR_{i,t}, capital stock K_{i,t}, labor effort E_{i,t}, hours per worker H_{i,t}, number of workers L_{i,t}, intermediates X_{i,t}, and utilization-adjusted productivity G_{i,t} (Equation G1.1).
  - Growth relationship: ΔS I Y_{i,t} = γ_i (ΔS I I_{i,t} + ΔS I K_{i,t}) + ΔS I G_{i,t} (Equation G1.2).
  - Cost-share weighted input ΔS I I_{i,t} defined in Equation G1.3; unobservable resource utilization ΔS I K_{i,t} defined in Equation G1.4.
  - Estimated regression: ΔS I Y_{i,t} = γ_i ΔS I I_{i,t} + β_i ΔS I H_{i,t} + ΔS I G_{i,t} (Equation G1.5). Residual ΔS I G_{i,t} (including constant) is utilization-adjusted productivity growth.
  - Two-stage-least squares used to address endogeneity; instruments: growth rate of spending in national defense (real), growth rate of imported fuel prices deflated by GDP deflator, growth rate of rural commodity prices deflated by GDP deflator, economic policy uncertainty in the United States and China.
  - Utilization coefficients restricted within two groups (mining, utilities and manufacturing; other industries). Assume constant returns to scale (γ_i = 1) for mining and retail sectors. Hours worked series detrended using HP filter with λ=6.25; use average weekly hours worked by industry.
- Alternative (simpler) approach — capital-utilization-adjusted TFP:
  - Start from standard industry-level production function with Solow residual z_{i,t} (Equation G1.6).
  - Capital service M_{i,t} = KUR_{i,t} * K_{i,t} (Equation G1.7).
  - Re-express output using capital services to obtain G_{i,t} that controls for capital utilization (Equation G1.8).
  - Link between Solow residual and utilization-adjusted productivity: ΔS I G_{i,t} = ΔS I z_{i,t} − α_{i,K} ΔS I KUR_{i,t} (Equation G1.9).
  - Simple empirical proxy: assume growth rate of intermediate goods equals growth rate of capital service, so ΔS I KUR_{i,t} = ΔS I X_{i,t} − ΔS I K_{i,t} (Equation G1.10).
- Comparative findings reported in Annex Tables I.1 and I.2:
  - For overall non-mining market sectors, the slowdown in recent years is larger when using the capital utilization-adjusted productivity measure, underscoring greater severity of productivity slowdown in non-mining market sectors.
  - For the mining sector, the capital-utilization adjusted productivity measure shows a different pattern relative to other measures, indicating uncertainty around productivity measurement in mining.

- Selected average growth rates (percent) from Annex Table I.2 (industry- and period-level figures preserved exactly as reported):
  - Non-Mining Market Sectors (period averages shown in table)
    - Solow residual: 0.65; 0.32; 0.04; 0.45; 0.16
    - Utilization Adj. TFP: 0.81; 0.33; 0.22; 0.67; 0.36
    - Cap. Utilization Adj. TFP: 0.77; 0.45; 0.35; 0.43; 0.08
  - Mining Sector
    - Solow residual: 0.83; -2.58; -2.42; -0.27; 1.11
    - Utilization Adj. TFP: 0.82; -2.54; -2.44; -0.28; 1.15
    - Cap. Utilization Adj. TFP: 1.67; -1.04; 0.59; 3.31; -0.62
  - Selected industry-level Solow residuals (period sequence as in table)
    - Agriculture, Forestry & Fishing: 2.82; 1.25; 0.53; -0.29; -0.60
    - Manufacturing: 0.29; 0.07; -0.06; 0.00; -0.20
    - Retail Trade: 0.96; 0.27; 0.85; 0.63; 0.55
    - Professional, Scientific & Technical Services (Solow residual): 0.17; 0.03; 0.55; -0.33; 1.02
  - Selected industry-level Utilization adj. TFP
    - Professional, Scientific & Technical Services: 3.97; 1.66; 2.77; 1.62; 3.47
    - Administrative & Support Services: 1.68; 2.15; 1.27; 1.06; 2.91
  - Selected industry-level Cap. Utilization adj. TFP
    - Retail Trade: 1.64; 1.19; 1.27; 1.27; 0.31
    - Information Media & Telecommunications: 0.29; 0.08; 1.19; 0.35; 0.92

### Annex II. Impact of R&D and ICT Investment on TFP — empirical results and robustness
- Main message: R&D and ICT investments are positively associated with TFP growth; results robust across multiple specifications and samples.
- Multiple-investment regressions (Annex Table II.1):
  - Column (1): R&D Investment (lag 5) = 0.0668* (standard error 0.033); Other Investment (lag 5) = 0.0163 (0.014); Observations = 8,640; R-squared = 0.331; Country-Time FE = Yes; Country-Ind FE = Yes; Industry-Time FE = Yes.
  - Column (2): ICT Investment (lag 5) = 0.0465** (0.020); Other Investment (lag 5) = 0.0197** (0.009); Observations = 8,556; R-squared = 0.331.
  - Column (3): R&D Investment and ICT Investment included together: coefficients on R&D and ICT larger than other investment but insignificant (potential multicollinearity); Observations = 8,556; R-squared = 0.332.
  - Column (4): Combined R&D and ICT Investment (lag 5) = 0.0431*** (0.008); Other Investment (lag 5) = 0.0162 (0.012); Observations = 8,556; R-squared = 0.331.
- Robustness checks (Annex Table II.2 and II.3):
  - Results robust to excluding the G-7 countries from the sample.
  - Results robust when sample restricted to manufacturing only, services only, and services excluding ICT industries.
  - Results robust when including lags one through five of investment types together: the sum of coefficients across lags is positive and significant for R&D and ICT investment.
- Selected coefficients and panels from additional tables (preserve reported values):
  - Annex Table II.3 (examples):
    - R&D Investment (lag 5) in some restricted samples = 0.1022***; 0.1337**; 0.1435*** (standard errors reported in table).
    - ICT Investment (lag 5) reported as 0.1117***; 0.1267; 0.0779*; 0.0888* in various columns.
    - When including lags 1–5: Lag 1 = 0.0444 (0.122) for R&D; Lag 3 = 0.0331*** for R&D (0.009); Lag 4 = 0.0945** for R&D (0.045); Lag 5 = 0.0488 for R&D (0.040).
  - Sample sizes and fit:
    - Observations vary by specification: 8,640; 8,556; 5,379; 5,295; 3,156; 3,154; 2,422; 2,597; 1,846; 2,516 depending on column.
    - R-squared values reported across specifications: 0.331; 0.333; 0.362; 0.364; 0.417; 0.416; 0.350; 0.348; 0.371.

### Annex III. Firm-level determinants of innovative (intangible) investment — selected findings
- Data: IMF CVU firm database. R&D tax incentives measured in percent of GDP. Interaction terms examine heterogeneity by firm characteristics: High External Finance Dependence (Rajan-Zingales index), Manufacturing dummy, Small firm dummy (sales < 25th percentile), High Expected Growth (Tobin’s Q above median).
- Selected reported coefficients from Annex Table III.1 (dependent variable: Growth Rate of Intangible Capital; sample period 2001-2018):
  - Sales Growth = .2464*** (0.0609) in column (1); .2652*** (0.0519) in column (6).
  - Uncertainty = -0.0251 (0.0516) in column (1).
  - Sales Growth * Uncertainty = - .3318*** (0.1071) in column (1); - .3600** (0.0870) in column (6).
  - Lagged Dependent Variable = -.0000 (0.0000).
  - Uncertainty * Lagged Dependent Variable = 1.5171*** (0.0655) in column (1); 1.5090*** (0.0577) in column (6).
  - High Ext. Finance Dep. * R&D tax incentives (-1) = .3279*** (0.1094) in column (1); .3697*** (0.0878) in column (6).
  - Manufacturing * R&D tax incentives (-1) = 1.1048* (0.6241) in column (1); 0.9972* (0.5465) in column (6).
  - Small * R&D tax incentives (-1) = 1.0199*** (0.4211) in column (1); 0.9682** (0.3843) in column (6).
  - High Exp. Growth * R&D tax incentives (-1) = 0.2529*** (0.1221) in column (1); 0.3057*** (0.1016) in column (6).
- Model features and fit:
  - Firm Fixed Effects = Yes; Year Fixed Effects = Yes.
  - R-squared = 0.7597 in column (1); 0.7633 in column (6).
  - Number of observations: 4,006 in column (1); 5,435 in column (6).

*Source: 1ausea2021002 - Chapter 2 (PDF chapter).

---


_Source: https://www.imf.org/-/media/files/publications/cr/2021/english/1ausea2021002.pdf_