## Inflation Anchoring and Industry Growth (wp1836)

## Source details

**Canonical URL:** [Inflation Anchoring and Industry Growth (wp1836)](https://www.imf.org/-/media/files/publications/wp/2018/wp1836.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp1836.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp1836.pdf.json)

---

### I. Introduction and research question
- Central claim: the degree to which inflation expectations are anchored has first-order implications for economic performance; low and stable inflation and well-anchored expectations are argued to reduce uncertainty and foster longer-term investment and growth.
- Empirical challenge: identification of causal effects of inflation anchoring on growth is difficult in aggregate data because of omitted variables and reverse causality; the paper uses industry-level data and a difference-in-difference strategy à la Rajan and Zingales (1998) to mitigate these concerns.
- Measure of anchoring: sensitivity of medium-term (five-year ahead) inflation expectations to inflation surprises — a country-specific, time-invariant parameter that varies only across countries.
- Sample:
  - 22 manufacturing industries
  - 36 advanced and emerging market economies
  - Period: 1990–2014

### II. Identification strategy and empirical framework
- Cross-industry variation enables identification of the interaction effect between country-level inflation anchoring and industry-specific credit-constraint characteristics, avoiding absorption by country fixed effects.
- Main empirical specification (Rajan and Zingales (1998) style):
  - gis,c = αi + αc + δX i Z c + μ log yi,c0 + εi,c
  - Interaction of country anchoring and industry-specific measures of credit constraints is the main independent variable.
- Inflation anchoring estimated from:
  - ∆πi,t+h e = βi h πi,t news + εi,t+h (equation (1)), using Consensus Economics professional forecasters.
  - Baseline: five-year-ahead expectations; quarterly forecast error used as baseline shock.

### III. Credit-constraints channel and proxies
- Conceptual channel: volatility in the economic environment disproportionately harms credit-constrained firms/industries by inducing shorter-term investment and reducing growth potential.
- Industry-level proxies (constructed from U.S. Compustat/firm data):
  - External financial dependence: median across U.S. firms of (total capital expenditures − current cash flow) / total capital expenditures.
  - Asset tangibility: ratio of fixed assets to total assets.
  - R&D intensity: R&D expenditures over total capital expenditure.
- Hypotheses:
  - Industries with higher external financial dependence benefit more from anchoring.
  - Industries with lower asset tangibility (more intangible capital) are more credit constrained and benefit more.
  - R&D-intensive industries face higher credit constraints and gain more from anchoring.

### IV. Main empirical findings (key statistics preserved)
- Primary finding: inflation anchoring fosters growth in industries that are more credit constrained.
- Figure 1 illustration (regression of average industry value added growth (1990–2014) on anchoring and controls, split by external financial dependence):
  - Below-median external financial dependence (left panel): slope coefficient = 0.82; t-statistic = 0.06.
  - Above-median external financial dependence (right panel): slope coefficient = -27.69; t-statistic = -2.14.
  - Interpretation: higher sensitivity (i.e., lower degree of anchoring) is negatively associated with average growth only for industries with above-median external financial dependence.
- Baseline regression results (selected coefficients and statistics):
  - Log of initial share: -0.959*** (I), -0.904*** (II), -0.952*** (III).
  - External financial dependence * Inflation anchoring: -39.860*** (III) (11.911 in parenthesis).
  - Asset tangibility * Inflation anchoring: 66.067** (II) (27.415 in parenthesis).
  - R&D intensity * Inflation anchoring: -26.960*** (III) (8.512 in parenthesis).
  - Magnitude of differential effects: -1.24 (I), 0.61 (II), -1.12 (III).
  - Observations 668; R-squared 0.6 (I), 0.59 (II), 0.59 (III).
- Gross output specifications (selected):
  - Log of initial share: -0.798*** (I), -0.761*** (II), -0.791*** (III).
  - External financial dependence * Inflation anchoring: -35.787** (III) (15.321 in parenthesis).
  - R&D intensity * Inflation anchoring: -23.030*** (III) (7.550 in parenthesis).
  - Magnitude of differential effects: -1.18 (I), 0.34 (II), 0.96 (III).
  - Observations 668; R-squared 0.61 (I), 0.60 (II), 0.60 (III).
- Summary numeric anchors:
  - Average sensitivity coefficient (inverse of anchoring) across final sample: 0.03.
  - Standard deviation of sensitivity coefficients: 0.05.
  - Consensus forecasts initially available for 44 countries (1990–2014); final empirical sample: 36 countries (U.S. excluded).
  - Industry coverage: 22 manufacturing industries (INDSTAT2).

### V. Robustness checks and alternative explanations
- Sorting industries by alternative credit-constraint indicators (asset tangibility, R&D intensity) yields consistent interaction effects.
- Controls and augmentations:
  - Explicitly control for interaction between level of inflation and industry credit-constraint measures; interaction of average level of inflation with credit-constraint measures is not statistically significant.
  - Interact sectoral credit-constraint measures with macro variables Wc: average bank credit to private sector/GDP, average CPI inflation, average government expenditure/GDP, average real GDP growth, countercyclicality of real short-term interest rate.
  - Correlations of anchoring with macro variables: −0.26 (financial development), 0.56 (level of inflation), −0.24 (size of government), −0.13 (overall growth), −0.18 (countercyclicality of real short-term rate).
  - Inclusion of these interactions leaves significant anchoring × credit-constraint effects intact.
- Alternative dependent variable:
  - Gross output growth yields broadly similar signs, sizes, and significance; asset tangibility channel loses statistical significance in some gross output specs.
- Generated regressor uncertainty:
  - WLS re-estimation with weights = inverse of standard deviation of sensitivity coefficients produces parameters similar to OLS.
  - Table 6 WLS highlights:
    - External financial dependence * Inflation anchoring: -48.005*** (III) (11.072).
    - Asset tangibility * Inflation anchoring: 84.018*** (II) (19.455).
    - R&D intensity * Inflation anchoring: -33.032*** (III) (10.206).
    - Magnitude of differential effects: -1.50 (I), 0.78 (II), -1.37 (III).
- Alternative anchoring measures and correlations with baseline:
  - Correlations: 0.58 (short-term expectations), 0.48 (alternative shocks), 0.85 (absolute sensitivity).
  - Results remain consistent across 1-, 2-, 3-, 5-, and 10-year-ahead expectations (results for additional horizons available upon request).

### VI. Instrumental variables and endogeneity
- Instruments (central bank institutional quality from Crowe and Meade (2007)):
  - Central bank governor turnover index, central bank independence index, central bank transparency index.
  - Governor turnover and independence indices also constructed from 1980–1989 values for exogeneity.
  - Instrument availability: three instruments for 25 countries; countercyclicality measure available for 28 countries.
- IV procedure and diagnostics:
  - First stage: anchoring on instruments with industry and country fixed effects.
  - Instruments are strong: Cragg-Donald Wald F-statistics well above Stock and Yogo (2005) critical values.
- IV results (selected statistics):
  - Log of initial share: -0.972*** (I), -0.970*** (II), -0.962*** (III).
  - External financial dependence * Inflation anchoring: -31.289** (I) (13.321).
  - Asset tangibility * Inflation anchoring: 44.842 (II) (83.625).
  - R&D intensity * Inflation anchoring: -23.171* (III) (13.697).
  - Magnitude of differential effects: -0.98 (I), 0.41 (II), -0.96 (III).
  - Cragg-Donald Wald F-statistic: 61.202 (I), 54.172 (II), 57.908 (III).
  - Stock-Yogo 5% critical value: 13.91 (all columns).
  - Hansen J-statistic p-value: 0.118 (I), 0.914 (II), 0.187 (III).
  - Observations 428; R-squared 0.42 (I), 0.39 (II), 0.42 (III).
- IV estimates confirm that inflation anchoring enhances growth more for industries with higher external financial dependence and higher R&D intensity; IV effects are somewhat smaller than OLS but differences are not statistically significant.

### VII. Policy implications and conclusions
- Core conclusion:
  - Inflation anchoring fosters industry growth through the credit-constraint channel, benefiting industries with high external financial dependence, high R&D intensity, and low asset tangibility.
- Quantitative magnitudes and examples:
  - Interaction-effect magnitudes range from 0.6 to 1.2 percentage points across specifications.
  - Example cited: differential growth gains of 1.2 percentage points for improving anchoring from Czech Republic level to Italy level while moving from an industry with low to high external financial dependence.
- Policy-relevant insights:
  - Improving the monetary policy framework to better anchor inflation expectations is likely more growth-friendly in economies with a larger share of credit-constrained industries and during periods when credit constraints are more binding (for example, recessions).
  - The predictability of inflation (anchoring) rather than the level of inflation per se is what matters for industry growth via credit constraints.
- Robustness summary:
  - Findings hold across alternative specifications, growth measures, anchoring definitions, macro controls, and IV approaches.

*Source: wp1836*

### References 21

### References 21

### I. Introduction and research question
- Central claim examined: the degree to which inflation expectations are anchored has first-order implications for economic performance; low and stable inflation and well-anchored expectations are argued to reduce uncertainty and foster longer-term investment and growth.
- Empirical challenge addressed: identification of causal effects of inflation anchoring on growth is difficult in aggregate data because of omitted variables and reverse causality; the paper uses industry-level data and a difference-in-difference strategy à la Rajan and Zingales (1998) to mitigate these concerns.
- Measure of inflation anchoring: the sensitivity of (medium-term, that is, five-year ahead) inflation expectations to inflation surprises — a country-specific, time-invariant parameter that varies only across countries.
- Sample used in analysis:
  - 22 manufacturing industries
  - 36 advanced and emerging market economies
  - Period: 1990-2014

### II. Identification strategy and empirical framework
- Cross-industry dimension motivations:
  - The inflation anchoring parameter is country-specific and would be absorbed by country fixed effects in a pure cross-country analysis; industry-level variation allows identification of the interaction effect.
  - Reverse causality concerns are mitigated because country-level inflation anchoring plausibly affects industry-level outcomes, and the main independent variable is an interaction between country anchoring and industry-specific technological/financial characteristics derived from U.S. firm-level data.
- Main empirical approach: regress industry-level growth on the interaction between country-level inflation anchoring and industry-specific measures of credit constraints, controlling for country and industry fixed effects and initial industry shares.

### III. Credit-constraints channel and proxies
- Theoretical/conceptual channel: volatility in the economic environment (from shocks or policy) disproportionately harms firms/industries that are credit constrained by inducing shorter-term rather than longer-term investment, reducing growth potential.
- Industry-level proxies for credit constraints used in the analysis include:
  - External financial dependence
  - Asset tangibility
  - R&D intensity
  - (These are sourced from U.S. firm-level data and are standard proxies in the literature.)

### IV. Main empirical findings (key statistics preserved)
- Primary finding: inflation anchoring fosters growth in industries that are more credit constrained.
- Illustration from Figure 1 (as described in the text):
  - Regression of average industry value added growth (1990–2014) on the measure of inflation anchoring, industry dummies, and initial industry share, split by external financial dependence:
    - Left panel (industries with below-median external financial dependence): slope coefficient = 0.82; associated t-statistic using robust standard errors = 0.06.
    - Right panel (industries with above-median external financial dependence): slope coefficient = -27.69; associated t-statistic using robust standard errors = -2.14.
  - Interpretation: higher sensitivity (i.e., lower degree of inflation anchoring) is negatively associated with average growth only for industries with above-median external financial dependence.

### V. Robustness checks and alternative explanations
- The results are subjected to a battery of robustness exercises:
  - Sorting industries by alternative indicators of credit constraint (asset tangibility and R&D intensity) in addition to external financial dependence.
  - Controlling explicitly for the interaction between the level of inflation and industry-specific measures of credit constraints to disentangle the effect of inflation anchoring from the effect of the level of inflation; results suggest inflation anchoring — not the level of inflation per se — has the statistical effect on growth through the credit-constraints channel.
  - Controlling for interactions between sectoral credit-constraint measures and a broad set of macroeconomic variables, including financial development, inflation, the size of government, overall economic growth, monetary policy counter-cyclicality, and the level of inflation.
  - Instrumental variable (IV) approach using monetary policy transparency and independence as instruments for inflation anchoring.

### VI. Contribution to the literature
- Adds to two streams:
  - The literature on the effect of inflation on growth (addressing mixed and fragile evidence in prior studies).
  - The literature on how financial frictions amplify the growth effects of uncertainty about the economic environment (shocks or stabilization policies).

### VII. Organization of the paper (structure)
- Section II: outlines the credit-constraint channel and empirical proxies.
- Section III: describes the data and construction of the inflation-anchoring measure.
- Section IV: explains the difference-in-difference methodology.
- Section V: presents main results and robustness exercises.

*Source: wp1836 - References 21*

### Conclusions are in Section VI.

### wp1836 - Conclusions are in Section VI.

### Main research question and methodology
- Objective: Assess how inflation anchoring affects industry growth via credit-constraint channels.
- Empirical strategy:
  - Measures shock-anchoring by estimating ∆πi,t+h e = βi h πi,t news + εi,t+h (equation (1)) using survey-based professional forecasters’ expectations from Consensus Economics.
  - Industry-level analysis follows Rajan and Zingales (1998) style specification: gis,c = αi + αc + δX i Z c + μ log yi,c0 + εi,c (equation (2)), with i for industries and c for countries.
  - Sample and data:
    - Consensus forecasts available initially for 44 countries (1990–2014); final empirical sample: 36 countries (U.S. excluded).
    - Industry-level dependent variables from UNIDO for 22 manufacturing industries (INDSTAT2 2016, ISIC Revision 3) over 1990–2014.
    - Industry characteristics (external financial dependence, asset tangibility, R&D intensity) constructed from U.S. firm-level Compustat data and measures updated from Tong and Wei (2011), Samaniego and Sun (2016), Braun and Larrain (2005).

### Measurement of inflation anchoring and samples
- Baseline inflation anchoring:
  - Uses five-year-ahead inflation expectations as baseline (medium-term), with sensitivity coefficient β interpreted as the degree of anchoring (smaller β = better anchored).
  - The quarterly forecast error is used as baseline measure of inflation shocks.
- Key numeric summaries:
  - Estimates produced for 44 countries where Consensus forecasts are available from 1990 to 2014; final empirical sample used in regression analysis: 36 countries.
  - Industry coverage: 22 manufacturing industries (INDSTAT2).
  - Average sensitivity coefficient (inverse of anchoring) across the final sample: 0.03.
  - Standard deviation of sensitivity coefficients: 0.05 — indicating large cross-country variation.

### Channels and industry characteristics (credit-constraint proxies)
- External financial dependence:
  - Measured as median across U.S. firms in each industry of (total capital expenditures − current cash flow) / total capital expenditures (updated Tong and Wei, 2011).
  - Hypothesized: industries with higher external financial dependence benefit more from inflation anchoring.
- Asset tangibility:
  - Measured as ratio of fixed assets to total assets from U.S. Compustat (Samaniego and Sun, 2016; Braun and Larrain, 2005).
  - Hypothesized: industries with lower asset tangibility (more intangible capital) are more credit constrained and benefit more from anchoring.
- R&D intensity:
  - Measured as R&D expenditures over total capital expenditure (Samaniego and Sun, 2016).
  - Hypothesized: R&D-intensive industries face higher credit constraints and gain more from anchoring.

### Baseline empirical findings
- Catch-up effect:
  - The coefficient on initial sector share (log yi,c0) is negative and statistically significant at the 1 percent level.
- Interaction effects (inflation anchoring × industry characteristics):
  - Inflation anchoring (lower sensitivity of expectations) increases industry growth more for industries with:
    - Higher external financial dependence — statistically significant at the 5 percent level.
    - Lower asset tangibility — statistically significant at the 5 percent level.
    - Higher R&D intensity — statistically significant at the 5 percent level.
- Magnitude of effects (differential growth gains):
  - Interaction-effect magnitudes range from 0.6 to 1.2 percentage points.
  - Example: Differential growth gains of 1.2 percentage points for improving anchoring from Czech Republic level to Italy level while moving from an industry with low to high external financial dependence.
  - Note: moving from the 25th to 75th percentile in sensitivity implies a dramatic change in monetary policy credibility.

### Robustness and sensitivity checks
- Alternative dependent variable:
  - Using gross output growth (instead of value-added) yields broadly similar signs, sizes, and significance for interaction effects; asset tangibility channel loses statistical significance with gross output.
- Uncertainty in anchoring estimates:
  - Re-estimation using Weighted Least Squares (WLS) with weights = inverse of standard deviation of sensitivity coefficients produces parameters similar to OLS — mitigates concerns about generated regressor uncertainty.
- Alternative anchoring measures:
  - Tests with 1-year-ahead expectations, alternative definitions of inflation shocks, and absolute sensitivity of medium-term expectations.
  - Correlations with baseline anchoring measure: 0.58 (short-term expectations), 0.48 (alternative shocks), 0.85 (absolute sensitivity).
  - Results remain consistent: significant anchoring effects via external financial dependence, asset tangibility, and R&D intensity.
  - Additional note: findings robust when replacing 1-year-ahead with 2, 3, and 10 year-ahead expectations (results available upon request).
- Omitted variable and macro controls:
  - Augmented specification interacts industry characteristics with country-level variables Wc to test omitted variable bias (equation (3)):
    - Considered Wc: average bank credit to private sector/GDP (financial development, 1990–2014), average CPI inflation (1990–2014), average government expenditure/GDP (1990–2014), average real GDP growth (1990–2014), and countercyclicality of real short-term interest rate (sensitivity of real short-term rate to real GDP growth, controlling for one-quarter-lagged real short-term rate).
  - Correlations of anchoring with these macro variables: −0.26 (financial development), 0.56 (level of inflation), −0.24 (size of government), −0.13 (overall growth), −0.18 (countercyclicality of real short-term rate).
  - Inclusion of these interactions leaves the significant interaction effects of inflation anchoring and the three credit-constraint measures intact.
  - The interaction of average level of inflation with credit-constraint measures is not statistically significant.

### Instrumental variables and endogeneity
- Instruments for anchoring:
  - Central bank institutional quality indicators from Crowe and Meade (2007): (i) central bank governor turnover index, (ii) central bank independence index, (iii) central bank transparency index.
  - For exogeneity, governor turnover and independence indices also constructed from 1980–1989 values (pre-sample).
  - Availability: three instruments available for 25 countries in sample; countercyclicality measure available for 28 countries.
- IV procedure:
  - First stage: regress anchoring measure on instruments with industry and country fixed effects.
    - Instruments are strong: Cragg-Donald Wald F-statistics well above Stock and Yogo (2005) critical values.
    - Hansen’s J statistic reported for instrument validity.
  - Second stage: re-estimate equation (2) using fitted values of anchoring.
- IV results:
  - Confirm that inflation anchoring enhances growth more for industries with higher external financial dependence and higher R&D intensity, with somewhat smaller effects than OLS (differences not statistically significant).

### Policy implications and conclusions
- Core conclusion:
  - Inflation anchoring fosters industry growth through the credit-constraint channel, benefiting industries characterized by high external financial dependence, high R&D intensity, and low asset tangibility.
- Policy-relevant insights:
  - Improving the monetary policy framework to better anchor inflation expectations is likely more growth-friendly in economies with a larger share of credit-constrained industries and during periods when credit constraints are more binding (for example, recessions).
  - The predictability of inflation (anchoring) rather than the level of inflation per se is what matters for industry growth via credit constraints.
- Final empirical scope:
  - Findings are robust across alternative specifications, growth measures, anchoring definitions, macro controls, and IV approaches.

*Source: wp1836 - Conclusions are in Section VI.*

### References

### wp1836 - References

### References (selected citations)
- Acemoglu, Daron, and Fabrizio Zilibotti. “Was Prometheus unbound by chance? Risk, diversification, and growth.” Journal of Political Economy 105.4 (1997): 709-751.  
- Afonso, Antonio, and Davide Furceri. “Government size, composition, volatility and economic growth.” European Journal of Political Economy 26.4 (2010): 517-532.  
- Aghion, Philippe, George-Marios Angeletos, Abhijit Banerjee, and Kalina Manova. “Volatility and growth: Credit constraints and the composition of investment.” Journal of Monetary Economics 57, no. 3 (2010): 246-265.  
- Bernanke, Ben. “Irreversibility, uncertainty, and cyclical investment.” Quarterly Journal of Economics, Vol. 97, No. 1, (1983), pp. 85-106.  
- Christiano, Lawrence J., Roberto Motto, and Massimo Rostagno. “Risk shocks.” American Economic Review 104.1 (2014): 27-65.  
- Gilchrist, Simon, Jae W. Sim, and Egon Zakrajšek. “Uncertainty, financial frictions, and investment dynamics.” No. w20038. National Bureau of Economic Research, (2014).  
- Pindyck, Robert S. “Irreversible investment, capacity choice, and the value of the firm,” American Economic Review, vol. 78(5), (1988), 969-85, December.  
- Rajan, Raghuram and Luigi Zingales (1998), “Financial Dependence and Growth,” American Economic Review, 88 (3), 559-86  
- Stock, James H., and Motohiro Yogo. “Testing for Weak Instruments in Linear IV Regression.” Identification and Inference for Econometric Models: Essays in Honor of Thomas Rothenberg (2005)  
- Volcker, P., 1979, Statement before the Joint Economic Committee of the U.S. Congress. October 17, 1979. Reprinted in Federal Reserve Bulletin 65 (11): 888–90.  

### Figures (captions and key numeric indicators)
- Figure 1. Inflation anchoring and industry growth: the role of credit constraints  
  - Left panel slope coefficient: 0.82; associated t-statistic using robust standard errors: 0.06.  
  - Right panel slope coefficient: -27.69; associated t-statistic using robust standard errors: -2.14.
- Figure 2. Change in inflation expectations and inflation shocks (percentage points)  
  - Data are quarterly. In panels 3 and 4, blue lines denote the median of inflation shocks, and shaded areas denote interquartile ranges.
- Figure 3. Correlations between the sensitivity of inflation expectations and other factors  
  - Panels plot sensitivity of the medium-term inflation expectations against: Average private credit to GDP; Average CPI inflation; Average general government expenditure to GDP; Average real GDP growth; Real interest rate countercyclicality (axes and scales shown in figure).

### Tables — industry-level characteristics and empirical results (key entries and statistics)
- Table 1. Industry-specific intrinsic characteristics (selected rows)
  - ISIC 15 Food products and beverages: External financial dependence 0.11, Asset tangibility 0.37, R&D intensity 0.06  
  - ISIC 24 Chemicals and chemical products: External financial dependence 0.50, Asset tangibility 0.29, R&D intensity 1.11  
  - ISIC 30 Office, accounting and computing machinery: External financial dependence 0.96, Asset tangibility 0.18, R&D intensity 1.19  
  - ISIC 31 Electrical machinery and apparatus n.e.c.: External financial dependence 0.95, Asset tangibility 0.21, R&D intensity 0.81
- Table 2. Correlation matrix of industry-specific characteristics
  - Correlation between External financial dependence and Asset tangibility: -0.27  
  - Correlation between External financial dependence and R&D intensity: 0.73  
  - Correlation between Asset tangibility and R&D intensity: -0.40
- Table 3. Country coverage and number of industries (selected entries)
  - Australia 11; Brazil 21; Canada 22; China 18; India 21; Japan 20; Korea 22; United Kingdom 20; United States not listed in table excerpt.
- Table 4. The effect of inflation anchoring on industry growth: Value added (coefficients shown across specifications I–III)
  - Log of initial share: -0.959*** (I), -0.904*** (II), -0.952*** (III) (t-statistics in parentheses).  
  - External financial dependence * Inflation anchoring: -39.860*** (III) (11.911 in parenthesis).  
  - Asset tangibility * Inflation anchoring: 66.067** (II) (27.415 in parenthesis).  
  - R&D intensity * Inflation anchoring: -26.960*** (III) (8.512 in parenthesis).  
  - Magnitude of differential effects: -1.24 (I), 0.61 (II), -1.12 (III).  
  - Observations 668; R-squared 0.6 (I), 0.59 (II), 0.59 (III).
- Table 5. The effect of inflation anchoring on industry growth: Gross output
  - Log of initial share: -0.798*** (I), -0.761*** (II), -0.791*** (III).  
  - External financial dependence * Inflation anchoring: -35.787** (III) (15.321 in parenthesis).  
  - Asset tangibility * Inflation anchoring: 36.717 (II) (33.957 in parenthesis).  
  - R&D intensity * Inflation anchoring: -23.030*** (III) (7.550 in parenthesis).  
  - Magnitude of differential effects: -1.18 (I), 0.34 (II), 0.96 (III).  
  - Observations 668; R-squared 0.61 (I), 0.60 (II), 0.60 (III).
- Table 6. The effect of inflation anchoring on industry growth: WLS
  - Log of initial share: -0.927** (I), -0.794* (II), -0.913** (III).  
  - External financial dependence * Inflation anchoring: -48.005*** (III) (11.072 in parenthesis).  
  - Asset tangibility * Inflation anchoring: 84.018*** (II) (19.455 in parenthesis).  
  - R&D intensity * Inflation anchoring: -33.032*** (III) (10.206 in parenthesis).  
  - Magnitude of differential effects: -1.50 (I), 0.78 (II), -1.37 (III).  
  - Observations 668; R-squared 0.60 (all specs).
- Table 7. Alternative measures of the degree of inflation anchoring (selected coefficients)
  - Log of initial share ranges across specs: -0.959** to -0.932*** (specific values and t-statistics reported per column).  
  - External financial dependence * Inflation anchoring: -3.533*** (I), -15.540*** (II), -60.016*** (III) (with parenthesis (1.189), (5.406), (13.118) respectively).  
  - Asset tangibility * Inflation anchoring: 7.435*** (I) (2.389), 13.176 (II) (15.912), 108.864*** (III) (29.334).  
  - R&D intensity * Inflation anchoring: -2.417** (I) (1.128), -13.005*** (II) (4.131), -38.665*** (III) (11.166).  
  - Observations 668; R-squared 0.59–0.60 across columns.
- Table 8. Omitted variable bias and alternative explanation (selected coefficients)
  - Log of initial share: values such as -0.833***, -0.899***, -0.963*** across specifications (standard errors in parentheses).  
  - External financial dependence * Inflation anchoring: -43.443*** (I) (15.663), -44.067** (II) (21.32), -36.761*** (III) (10.911).  
  - Asset tangibility * Inflation anchoring: 72.731** (I) (29.922), 98.231** (II) (41.813), 58.172** (III) (24.849).  
  - R&D intensity * Inflation anchoring: -21.186** (I) (7.985), -16.958* (II) (9.217), -25.677*** (III) (8.969).  
  - Observations 650 (some specs) and 668 (others); R-squared 0.58–0.60.
- Table 9. Omitted variable bias and alternative explanation (continued)
  - Log of initial share: values such as -0.994***, -0.903***, -0.953***, -0.977***, -0.944***, -0.935*** across columns X–XV.  
  - External financial dependence * Inflation anchoring: -41.278*** (X) (11.686), -47.610*** (XI) (12.27).  
  - Asset tangibility * Inflation anchoring: 66.880** (X) (27.773), 83.099* (XII) (46.047).  
  - R&D intensity * Inflation anchoring: -27.224*** (X) (8.398), -32.706*** (XII) (7.651).  
  - Observations 668 for overall growth specs; 415 for interest rate countercyclicality specs; R-squared 0.59 (overall) and 0.56 (interest rate countercyclicality).
- Table 10. The effect of inflation anchoring on industry growth: IV regression (selected statistics)
  - Log of initial share: -0.972*** (I), -0.970*** (II), -0.962*** (III).  
  - External financial dependence * Inflation anchoring: -31.289** (I) (13.321).  
  - Asset tangibility * Inflation anchoring: 44.842 (II) (83.625).  
  - R&D intensity * Inflation anchoring: -23.171* (III) (13.697).  
  - Magnitude of differential effects: -0.98 (I), 0.41 (II), -0.96 (III).  
  - Cragg-Donald Wald F-statistic: 61.202 (I), 54.172 (II), 57.908 (III).  
  - Stock-Yogo weak identification test 5% critical values: 13.91 (all columns).  
  - Hansen J-statistic p-value: 0.118 (I), 0.914 (II), 0.187 (III).  
  - Observations 428; R-squared 0.42 (I), 0.39 (II), 0.42 (III).

### Appendix (industry classification crosswalk)
- Table A.1. Industry classification: INDSTAT2 vs. INDSTAT3 (selected mappings)
  - INDSTAT2 ISIC 15 Food products and beverages → INDSTAT3 ISIC 311 Food  
  - INDSTAT2 ISIC 16 Tobacco products → INDSTAT3 ISIC 313 Beverages  
  - INDSTAT2 ISIC 17 Textiles → INDSTAT3 ISIC 321 Textiles  
  - INDSTAT2 ISIC 24 Chemicals and chemical products → INDSTAT3 ISIC 351 Industrial chemicals; 352 Other chemicals  
  - INDSTAT2 ISIC 29 Machinery and equipment n.e.c. → INDSTAT3 multiple classes including 382 Machinery, except electrical; 383 Machinery, electric

*Source: wp1836 - References (pdf). *

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp1836.pdf_