## 1. Overview of (Partial) ACE Systems Around the World

## Source details

**Canonical URL:** [1. Overview of (Partial) ACE Systems Around the World](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06259.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06259.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06259.pdf.json)

---

### Introduction
- ACE concept developed in the 1980s (Boadway and Bruce, 1984; Wenger, 1983); proposed in reform terms by IFS (1991). Recent proponents include Fehr and Wiegard (2003).
- Early adopters and reversal:
  - Croatia implemented an ACE in 1994; announced return to standard corporate taxation in 2001.
  - Italy and Austria introduced partial ACE systems in 1997 and 2000, respectively; these systems were dismantled thereafter.
- Brazil has applied a variant of an ACE since 1996; this Brazilian variant has been understudied.
- Primary aim: overview of ACE systems in practice, review literature, and supplement with empirical analysis for Brazil.

### ACE tax concept and mechanics
- Definition (following Devereux and Freeman, 1991):
  - The ACE (notional return) = end-of-last-year equity stock, Et-1, multiplied by a notional interest rate, iˆ.
  - The notional interest rate should be the risk free nominal interest rate; it can be approximated by the rate on government bonds.
- Practical implementation note:
  - Not identical to Boadway and Bruce (1984) theoretical tax (which suggested disallowing interest deductibility and granting allowance based on interest cost of total assets); however, practical implementation is equivalent if interest rate on debt ≈ notional interest rate.
- Key desirable properties (assuming iˆ is correct):
  - Neutrality for financing choices: indifference between debt and equity for corporate tax purposes.
  - Neutrality to investment: no tax on marginal projects; notional return equals pre-tax profits for marginal projects.
  - Method of tax depreciation is irrelevant under ACE.
  - Unaffected by inflation: inflation-induced monetary profit increases are offset by higher notional return; indexation unnecessary.
- Main drawbacks:
  - Narrower tax base requires a higher tax rate to collect same revenue, potentially harmful under international tax competition for mobile economic rents.
  - Doubts whether other countries will accept ACE-based corporate tax payments as a basis for double tax relief.

### Interaction with the personal income tax system
- Precondition for ACE neutrality between debt and equity:
  - Personal income tax system must treat capital income consistently (either complete exemption as an expenditure tax or complete taxation as a comprehensive income tax).
- Dual income tax systems can be compatible if the rate on different types of capital income is the same.
- Relevance of personal taxes may be limited:
  - Shareholders may avoid direct capital income taxation by saving through pension funds.
  - In small open economies, marginal providers of funds can be foreign investors; taxation of resident shareholders may be less relevant for cost of capital.

### ACE systems in practice — variants and distinguishing features (country highlights)
- Closest practical applications to theoretical ACE: Croatian and Belgian systems.
- Most other practical applications are partial ACE systems deviating from the theoretical model.
- Italian and Austrian systems:
  - Did not exempt normal economic profits but applied a lower tax rate to them than to economic rents.
  - Only considered new (post-reform) equity when calculating the normal return (declining importance over time).
- Brazilian system special feature:
  - Allows deduction of notional interest only when actually paid out to shareholders as “interest on equity”.
  - Functionally: a special type of dividend that is tax deductible, capped by equity stock × notional interest rate.
  - Closed companies may credit owners' accounts with notional interest without cash payout, making their system closer to a standard ACE.
- Naming conventions vary: “notional/protective interest”, “remuneration of equity”, “Dual Income Tax” (Italy), “risk capital deduction / notional interest deduction” (Belgium).
- Israel’s system (since 1982) deducts an allowance = inflation rate × equity stock to neutralize inflation-driven debt preference; targets only inflation neutrality.

### Table 1 — Selected country implementations (preserved figures)
- Austria (2000–04)
  - Base/Rate: Book value of new (post-reform) equity / Average return of government bonds in secondary markets plus 0.8 pp.
  - Details: Notional return taxed at a reduced rate of 25 percent instead of 34 percent.
- Belgium (Since 2006)
  - Base/Rate: Book value of equity / Average monthly government bond rate of year preceding fiscal year by two years. Rate capped at 6.5 percent and cannot change by more than 1 pp from year to year. Special SME rate is 0.5 pp higher.
  - Details: The notional return is deductible.
- Brazil (Since 1996)
  - Base/Rate: Book value of equity / Rate applicable to long-term loans.
  - Details: Up to the level of the notional return, dividends can be paid as “interest on equity”. This is deductible for all corporate income taxes and subject to the usual withholding tax on interest.
- Croatia (1994–2000)
  - Base/Rate: Book value of equity / 5 percent plus inflation rate of industrial goods if positive.
  - Details: The notional return is deductible.
- Italy (1997–2003)
  - Base/Rate: Book value of new (post-reform) equity. From 2000: 120 percent of new equity. In 2001: 140 percent of new equity, then again 100 percent of new equity / 7 percent 1997–2000, 6 percent 2001.
  - Details: Notional return taxed at a reduced rate of 19 percent. Other profits taxed at 37 percent (34 percent in 2003). Before 2001, the average tax must be at least 27 percent.
  - Additional features: notional interest qualified for a reduced corporate income tax rate rather than being tax deductible; only post-reform equity counted initially; changes to post-reform equity percentages in 2000 and 2001; minimum tax rate provision abolished in 2001.

### Previous analyses and literature coverage
- Italy: most intensively studied due to economy size and data availability; Italian variant restricted and frequently changed; key works include Bordignon, Giannini, and Panteghini (1999 and 2001) and overview references such as Keen (2003).
- Croatia: some work exists; advantage of full ACE system, disadvantage of limited data.
- Austria and Brazil: no detailed studies appear to exist yet at time of document (Austrian reform briefly described in some papers including Genser (2002)).
- Belgium: too early for empirical assessment as of document; some theoretical analysis exists (Gerard 2006a, 2006b).

### Italian ACE variant — description and empirical evidence
- Reform design:
  - System mixes an ACE variant with a CBIT variant and additionally taxes wages; interest deductibility is removed.
  - Partial system and frequent changes complicated analysis.
- Theoretical/simulation findings (Bordignon et al.):
  - Reform reduced tax preference for debt finance.
  - Simulations estimate reduced cost of capital for most cases, except where debt-equity ratios are high.
  - Effective tax rates confirm reduction in equity discrimination.
- Firm-level findings:
  - Staderini (2001): leverage decreased after introduction; firms issuing new equity were more profitable and had high investment rates; many firms not induced to issue equity (data up to 1998).
  - Santoro (2005): large and profitable firms more likely to issue new equity; small and southern firms less likely.
  - Oropallo and Parisi (2005): abolition raised tax rates for most firms (except those benefiting from a new tax consolidation provision); did not investigate debt-equity tradeoff.

### Croatian ACE reform — description and empirical evidence
- Croatia introduced a full ACE system (Rose and Wiswesser, 1998).
- Keen and King (2003) findings:
  - Croatian ACE raised similar amounts of revenue relative to GDP as EU and Central and Eastern European countries on average.
  - Protective interest reduced tax revenues by roughly one third in 1998, but other tax features compensated.
  - FDI higher than in all neighboring countries except the Czech Republic.
  - Conclusion: “the ACE passed its first practical test.”
  - Caveat: transition effects make ACE-specific identification difficult.

### The Brazilian ACE variant — 1996 reform summary (Box 1 preservation)
- Pre-1996 system:
  - Interest deductibility for debt only; system defined in Fiscal Reference Units because of high inflation.
  - Pre-reform statutory tax rate: 47.7 percent, composed of a main tax rate (25 percent), a surcharge (18 percent) and a corporate social security charge (10 percent).
  - Withholding taxes pre-1996: dividends 15 percent; interest 25 percent.
- 1996 reform:
  - Abolished Fiscal Reference Units.
  - Introduced deductibility of notional interest on equity stock, but only when paid out to shareholders (or credited to owners in closed companies).
  - Reduced statutory tax rate to 30.7 percent, composed of a main rate (15 percent), a surcharge (10 percent) and a social security charge (10 percent).
  - Final withholding taxes reduced to 0 percent on dividends and 15 percent on interest, including notional interest.
  - Depreciation allowances unchanged: 4 percent for industrial buildings and 10 percent for plant and machinery.
- Footnote on overall rate calculation: (1-.18-.1)*(.1+.18+.25).

### Theoretical effects of the Brazilian variant
- Key distinction from standard ACE:
  - Notional interest deductible only if paid out to shareholders; retained notional interest is not deductible.
  - In absence of transactions costs, firms could distribute notional interest and then issue new shares to restore capital — making effects similar to standard ACE.
  - With transactions costs, many firms may not issue new equity; such firms cannot fully deduct the notional return and face a system effectively giving deductibility of dividends for tax purposes (as “interest to shareholders”).
- Classification of firms:
  - Type L (low dividend payments): dividends typically less than their notional return (1ˆ − < tt EiD). For L firms, issuing equity to restore distributions may be too costly; they effectively have deductibility of dividends if paid as “interest to shareholders.”
  - Type H (high dividend paying): dividends exceed the notional return (1ˆ − > tt EiD). For H firms, the system operates as a standard ACE since they can deduct the full notional return.
  - Investment projects may switch firms between H and L types, changing applicable tax treatment and effective cost of capital.

### Assumptions and baseline parameters (preserved)
- Nominal rate of interest: 10 percent.
- Rate of notional return: 10 percent.
- Rate of inflation (hypothetical scenario): 4 percent.
- For EATR calculations, a financial return of 30 percent is assumed.
- Table 3 actual observed rates:
  - Rate of inflation: 1995: 66.0 percent, 1996: 15.8 percent.
  - Real rate of interest: 1995: -7.6 percent, 1996: 10.1 percent.

### Effective tax rate findings (selected EMTR and EATR figures reproduced exactly)
- Personal Taxes Excluded
  - 1995:
    - EMTR: Ret. earnings 45.1; New equity 45.1; Debt -5.0
    - EATR: Ret. earnings 46.7; New equity 46.7; Debt 37.2
  - 1996 No ACE:
    - EMTR: Ret. earnings 37.6; New equity 37.6; Debt -4.9
    - EATR: Ret. earnings 32.8; New equity 32.8; Debt 23.3
  - 1996 ACE, L-type:
    - EMTR: Ret. earnings 37.6; New equity -4.9; Debt -4.9
    - EATR: Ret. earnings 12.7; New equity -1.0; Debt -1.0
  - 1996 ACE, H-type:
    - EMTR: Ret. earnings 0.0; New equity 0.0; Debt 0.0
    - EATR: Ret. earnings 24.0; New equity 24.0; Debt 24.0
- Personal Taxes Included
  - 1995:
    - EMTR: Ret. earnings 26.8; New equity 37.8; Debt -5.0
    - EATR: Ret. earnings 46.8; New equity 49.1; Debt 42.9
  - 1996 No ACE:
    - EMTR: Ret. earnings 22.0; New equity 22.0; Debt -4.9
    - EATR: Ret. earnings 27.4; New equity 27.4; Debt 22.5
  - 1996 ACE, L-type:
    - EMTR: Ret. earnings 22.0; New equity -4.9; Debt -4.9
    - EATR: Ret. earnings 16.0; New equity 10.1; Debt 10.1
  - 1996 ACE, H-type:
    - EMTR: Ret. earnings -2.4; New equity -2.4; Debt -2.4
    - EATR: Ret. earnings 22.9; New equity 22.9; Debt 22.9

### Interpretation of effective tax rate results
- The ACE element materially alters effective tax rates across financing sources and firm types.
- L-type firms can face substantially different (often lower or negative) EMTRs and EATRs for certain financing sources compared to H-type firms and pre-reform values.
- With personal taxes included, most tax rates are lower because the shareholders discount rate is reduced; dividend taxation changed from 1995 to 1996 (15 percent to 0 percent), affecting comparisons.
- In theory the reform removed tax-driven debt preference for all firms; in practice H-type firms may have continued to prefer debt because the notional interest rate was set much lower than the average bank interest rate.
- With personal taxes included, the system can subsidize marginal investment in practice:
  - Subsidization observed for debt-financed investment and new equity financed investment of L-type firms.

### Table 3 selected ETR/EMTR figures using actual 1995/1996 rates (preserved)
- Personal taxes ignored, 1995:
  - EMTR (Ret. earnings): 45.1
  - EMTR (New equity): 45.1
  - EMTR (Debt): -5.0
  - EATR (Ret. earnings): 48.9
  - EATR (New equity): 48.9
  - EATR (Debt): 60.4
- Personal taxes ignored, 1996:
  - No / ACE (aggregate rows):
    - EMTR: 47.9 (No/ACE); EMTR Debt: -7.7
    - EATR: 41.7 (No/ACE); EATR Debt: 18.5
  - L-type:
    - EMTR (Ret. earnings): 47.9; EMTR (New equity): -7.7; EMTR (Debt): -7.7
    - EATR (Ret. earnings): 31.0; EATR (New equity): -2.4; EATR (Debt): -2.4
  - H-type:
    - EMTR (Ret. earnings): 25.4; EMTR (New equity): 25.4; EMTR (Debt): -2.7
    - EATR (Ret. earnings): 28.1; EATR (New equity): 28.1; EATR (Debt): 19.5
- Personal taxes included, 1995:
  - EMTR (Ret. earnings): 26.8
  - EMTR (New equity): 37.8
  - EMTR (Debt): -5.0
  - EATR (Ret. earnings): 66.7
  - EATR (New equity): 64.1
  - EATR (Debt): 71.3
- Personal taxes included, 1996:
  - Aggregate (No/ACE):
    - EMTR: 30.1 (No/ACE); EMTR Debt: -7.7
    - EATR: 29.0 (No/ACE); EATR Debt: 16.9
  - L-type:
    - EMTR (Ret. earnings): 30.1; EMTR (New equity): -7.7; EMTR (Debt): -7.7
    - EATR (Ret. earnings): 20.8; EATR (New equity): 6.0; EATR (Debt): 6.0
  - H-type:
    - EMTR (Ret. earnings): 12.1; EMTR (New equity): 12.1; EMTR (Debt): -5.1
    - EATR (Ret. earnings): 21.9; EATR (New equity): 21.9; EATR (Debt): 10.5

### Empirical analysis: data and sample
- Data source: Thomson Financial.
- Sample construction:
  - Over 300 nonfinancial firms for years since 1999; earlier years scarcer.
  - Selection: firms with at least three observations, including at least one pre-reform and one post-reform observation; years restricted to 1990–2000.
  - Final sample: 98 firms and 835 observations (sample size reduced for some regressions).
- Reporting: due to high inflation, ratios used rather than levels.

### Descriptive statistics (Table 4; preserved means and s.d.)
- Observations:
  - Debt/Total Assets: 835 observations.
  - Equity/Total Assets: 835 observations.
  - Profits/Sales: 827 observations.
  - Real Sales Growth: 710 observations.
  - Dividends/Sales: 828 observations.
- Means and dispersion:
  - Mean Debt/Total Assets: 13.3
  - Mean Equity/Total Assets: 41.1
  - Mean Profits/Sales: -6.5
  - Mean Real Sales Growth: 15.4
  - Mean Dividends/Sales: 2.5
  - S.d. Debt/Total Assets: 11.6
  - S.d. Equity/Total Assets: 44.7
  - S.d. Profits/Sales: 124.1
  - S.d. Real Sales Growth: 94.4
  - S.d. Dividends/Sales: 4.9

### Dividend payout distribution and behavior
- Notional return estimation:
  - Estimated by multiplying the legally prescribed interest rate by previous year’s book value of equity.
  - For years prior to 1995, rate predicted via regression of the money market rate on prescribed interest rates; prescribed rate tends to increase by 44 percent with the money market rate.
- Distribution of L-type and H-type firms (selected table entries preserved):
  - 1992: L-Type Number 52 (proportion 19 percent), H-Type Number 100 (36.5 percent).
  - 1995: L-Type 69 (25); H-Type 2 (2.8) with H-Type proportion 97.2 (35.2) in table formatting.
  - 2000: L-Type 73 (24); H-Type 13 (15.1) with H-Type proportion 84.9 (27.9).
- Main empirical findings:
  - Great majority of firms pay dividends lower than the notional return.
  - Prior to 1995, no firm in the sample paid dividends in excess of the notional return.
  - For most Brazilian firms the system functioned effectively as dividend deductibility rather than a standard ACE.
- Proximity to notional return:
  - Very few firms pay out dividends reaching at least 90 percent of the notional return: peak year 1998 share among low-payout firms is 5.4 percent; other years < 5 percent.
  - More firms pay just above the notional return (up to 110 percent); among high pay-out firms, share reaches 38 percent in 2000 (but is 0 percent in some years).

### Average share of notional interest paid out (Table 6; preserved yearly means, s.d., medians)
- 1992 Mean: 3.1; S.d.: 4.7; Median: 0.7 (Obs. 52)
- 1993 Mean: 4.5; S.d.: 5.9; Median: 2.4 (Obs. 57)
- 1994 Mean: 1.2; S.d.: 1.2; Median: 1.0 (Obs. 58)
- 1995 Mean: 14.0; S.d.: 23.5; Median: 7.3 (Obs. 71)
- 1996 Mean: 18.7; S.d.: 29.7; Median: 10.3 (Obs. 96)
- 1997 Mean: 45.5; S.d.: 66.0; Median: 28.5 (Obs. 86)
- 1998 Mean: 35.9; S.d.: 54.6; Median: 17.4 (Obs. 84)
- 1999 Mean: 30.0; S.d.: 51.8; Median: 15.1 (Obs. 83)
- 2000 Mean: 45.8; S.d.: 75.0; Median: 23.9 (Obs. 91)
- Interpretation:
  - Important increase in dividends relative to notional returns after the 1996 reform, with most increase starting in 1997.
  - Distribution skewed; median always below mean.

### Regression analysis of dividend payouts (Table 7; key coefficients preserved)
- Dependent variable: Cash dividends.
- Selected regression results:
  - (1) OLS: Profits coefficient 0.159 (0.048)***.
  - (2) OLS with Post dummy and interaction: Profits 0.059 (0.042); Post * Profits 0.106 (0.070).
  - (3) Firm fixed effects: Profits 0.017 (0.017); Post * Profits 0.109 (0.046)**.
  - (4) Tobit with firm fixed effects: Profits 0.067 (0.042); Post * Profits 0.116 (0.041)***.
  - Post dummy coefficients (levels) significantly positive in (2)-(4), reflecting higher nominal dividends after reform.
- Interpretation:
  - Controlling for firm fixed effects and left-censoring, a significantly higher share of profits is paid out after the reform.
  - Pre-reform dividends may not have been strongly linked to profits; consistent with firms minimizing dividends.

### Capital structure: aggregate trends (Table 8 preserved)
- Equity/Total Assets and LT Debt/Total Assets (in percent):
  - 1992: Equity/Total Assets 50.6; LT Debt/Total Assets 10.0
  - 1995: Equity/Total Assets 50.2; LT Debt/Total Assets 09.9
  - 1996: Equity/Total Assets 42.2; LT Debt/Total Assets 13.5
  - 1997: Equity/Total Assets 38.4; LT Debt/Total Assets 16.4
  - 1998: Equity/Total Assets 32.0; LT Debt/Total Assets 17.1
  - 1999: Equity/Total Assets 28.0; LT Debt/Total Assets 17.5
  - 2000: Equity/Total Assets 23.7; LT Debt/Total Assets 16.4
- Aggregate implication:
  - Post-reform equity ratios fell substantially, reaching about half pre-reform level by 2000.
  - Long-term debt ratios increased; residual financing share increased steadily.

### Capital structure regressions (Table 9; key coefficients preserved)
- Dependent variable: Debt/Total Assets (specifications 1–3) and Equity/Total Assets (4).
- Post dummy coefficients:
  - (1) Post 0.061 (0.008)*** — OLS
  - (2) Post 0.062 (0.009)*** — OLS with controls
  - (3) Post 0.057 (0.007)*** — Firm fixed effects
  - (4) Post -0.125 (0.012)*** — Firm fixed effects (dependent = Equity/Total Assets)
- Profitability effects:
  - (2) coefficient -0.010 (0.004)** (more profitable firms issue less debt).
  - (3) firm fixed effects: -0.016 (0.004)*** for debt; (4) mirror image for equity 0.020 (0.010)**.
- Real sales growth:
  - (3) firm fixed effects: coefficient 0.011 (0.005)** positive for debt.
- Interpretation:
  - Increase in debt and decrease in equity after the reform cannot be fully explained by observable firm-level changes.
  - Majority of firms appear to face dividend-deductibility incentives rather than ACE, encouraging external finance; if external finance is partly debt, debt/equity increases.

### Investment effects (Table 10 and analysis; key regression coefficients preserved)
- Investment regressions — dependent variable Investment/Capital:
  - (1) OLS: Post 0.039 (0.013)***.
  - (2) OLS with controls: Post 0.045 (0.014)***.
  - (3) Firm fixed effects: Post 0.026 (0.012)**.
  - (4) Firm fixed effects with lagged profitability and real sales growth:
    - Post 0.023 (0.013)*.
    - Profitability t-1 0.089 (0.037)**.
    - Real sales growth and its lag positive; real sales growth t-1 coefficient significant at (0.004)*.
  - (5) Firm fixed effects with Tobin’s Q: Post -0.001 (0.016) (insignificant); Tobin’s Q 0.041 (0.022)*.
- Interpretation:
  - Investment rates increased following the reform (post dummy positive and significant in most specifications).
  - Lagged profitability positive and significant suggests firms rely on retained earnings to finance investment.
  - Including Tobin’s Q renders the post dummy insignificant, consistent with the reform affecting valuation (Q).

### Conclusions from the Brazilian Reform
- Design and immediate effects:
  - 1996 reform reduced corporate income taxes and allowed dividends to be deductible up to a notional return on equity.
  - The effect depends on dividend payout ratios: for most firms the system operated as dividend deductibility, not as a standard ACE.
- Firm-level incidence:
  - L-type firms (low payout): effectively faced dividend deductibility.
  - H-type firms (high payout): faced ACE-like treatment.
  - Very few firms had dividends high enough to ensure ACE treatment; most firms experienced increased dividends and a slow increase in H-type firms.
- Effects on financing and capital structure:
  - Theoretical neutrality: dividend deductibility should neutralize debt/equity choice.
  - Empirical outcome: equity ratios fell and debt ratios rose post-reform; no move toward more equity finance observed.
  - Dividend deductibility incentivized raising external funds and increasing dividends; transactions costs and legal restrictions limited full ACE turnover.
- Investment and macro outcomes:
  - Investment increased following the reform.
  - Ambiguity remains whether increase due to lower statutory tax rates or the dividend-deductibility/ACE structure.
- Longer-run dynamics and research implications:
  - Over time, system could resemble a true ACE if firms increase dividend payments and adjust capital structure.
  - Further research with longer panels and cross-firm analysis needed to assess longer-term transitions.
- Key preserved parameter figures:
  - Year of reform: 1996.
  - Personal tax rate on remuneration to equity (1996 example): 15 percent.
  - Corporate income tax rate (1996 example): 30.4 percent.
  - γ (tax discrimination factor under the 1996 parameterization): 1.22.

### Appendix I — Methodological highlights (effective tax rates and ACE properties)
- Method: builds on Devereux and Griffith (EATR and EMTR) adapted for dividend deductibility and Brazilian variant.
- Formal properties of an ACE (when iˆ = i):
  - Neutrality between debt and equity finance.
  - Irrelevance of depreciation schedules (provided personal taxes excluded).
  - EMTR of 0 if personal taxes excluded: ACE neutral to marginal investment.

*Source: IMF working paper _wp06259 — https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06259.pdf*

### 1. Overview of (Partial) ACE Systems Around the World........................................................8

### 1. Overview of (Partial) ACE Systems Around the World

### Introduction
- The allowance for corporate equity (ACE) concept was developed in the 1980s (Boadway and Bruce, 1984; Wenger, 1983) and proposed in practical reform terms by IFS (1991). Recent proponents include Fehr and Wiegard (2003).
- Early adopters: Croatia implemented an ACE in 1994. Italy and Austria introduced partial ACE systems in 1997 and 2000, respectively.
- By 2000–2001 these European experiments largely reversed: Croatia announced a return to standard corporate taxation in 2001; Italian and Austrian systems were dismantled thereafter.
- Brazil has applied a variant of an ACE since 1996; this Brazilian variant has been surprisingly understudied.
- Primary aim of the paper: provide an overview of all ACE systems in practice, review literature assessing their applications, and supplement with empirical analysis for Brazil.

### ACE tax concept and mechanics
- Definition (following Devereux and Freeman, 1991):
  - The ACE (notional return) = end-of-last-year equity stock, Et-1, multiplied by a notional interest rate, iˆ.
  - The notional interest rate should be the risk free nominal interest rate; it can be approximated by the rate on government bonds.
- Practical implementation note:
  - Not identical to Boadway and Bruce (1984) theoretical tax (which suggested disallowing interest deductibility and granting allowance based on interest cost of total assets); however, since total assets ≈ total debt + equity, practical implementation is equivalent if interest rate on debt ≈ notional interest rate.
- Key desirable properties of ACE (assuming correct notional interest rate):
  - Neutrality for financing choices: indifference between debt and equity for corporate tax purposes.
  - Neutrality to investment: no tax on marginal projects; notional return equals pre-tax profits for marginal projects.
  - Method of tax depreciation is irrelevant under ACE.
  - Unaffected by inflation: inflation-induced monetary profit increases are offset by higher notional return; indexation unnecessary.
- Main drawbacks:
  - Narrower tax base requires a higher tax rate to collect same revenue, which can be harmful under international tax competition for mobile economic rents.
  - Doubts whether other countries will accept ACE-based corporate tax payments as a basis for double tax relief.

### Interaction with the personal income tax system
- Precondition for ACE neutrality between debt and equity: personal income tax system must treat capital income consistently (either complete exemption as an expenditure tax or complete taxation as a comprehensive income tax).
- Dual income tax systems can be compatible if the rate on different types of capital income is the same.
- Relevance of personal taxes may be limited:
  - Shareholders may avoid direct capital income taxation by saving through pension funds (where the tax rate faced by the pension fund may be the relevant rate).
  - In small open economies, marginal providers of funds can be foreign investors; thus taxation of resident shareholders may be less relevant for cost of capital (references: Brennan (1970); Bond, Devereux, and Klemm (2006)).

### ACE systems in practice — variants and distinguishing features
- Closest practical applications to theoretical ACE: Croatian and Belgian systems.
- Most other practical applications are partial ACE systems deviating from the theoretical model.
- Italian and Austrian systems:
  - Did not exempt normal economic profits but applied a lower tax rate to them than to economic rents.
  - Only considered new (post-reform) equity when calculating the normal return (declining importance over time).
- Brazilian system special feature:
  - Only allows deduction of notional interest when actually paid out to shareholders as “interest on equity”.
  - Functionally: a special type of dividend that is tax deductible, capped by equity stock × notional interest rate.
  - Closed companies may credit owners' accounts with notional interest on equity without cash payout, making their system closer to a standard ACE.
- Naming conventions vary across countries: “notional/protective interest”, “remuneration of equity”, “Dual Income Tax” (Italy), “risk capital deduction / notional interest deduction” (Belgium).
- Israel’s system (since 1982) deducts an allowance = inflation rate × equity stock to neutralize inflation-driven debt preference; it is not listed as a partial ACE in Table 1 because it targets only inflation neutrality and does not affect non-inflation-induced debt preference.

### Table 1 — Overview of (Partial) ACE Systems Around the World (country-by-country details)
- Austria (2000–04)
  - Name: Notional interest 1/
  - Base/Rate: Book value of new (post-reform) equity / Average return of government bonds in secondary markets plus 0.8 pp
  - Details: The notional return is taxed at a reduced rate of 25 percent instead of 34 percent.
- Belgium (Since 2006)
  - Name: Risk capital deduction / notional interest deduction
  - Base/Rate: Book value of equity / Average monthly government bond rate of year preceding fiscal year by two years. Rate capped at 6.5 percent and cannot change by more than 1 pp from year to year. Special SME rate is 0.5 pp higher.
  - Details: The notional return is deductible.
- Brazil (Since 1996)
  - Name: Remuneration of equity
  - Base/Rate: Book value of equity / Rate applicable to long-term loans
  - Details: Up to the level of the notional return, dividends can be paid as “interest on equity”. This is deductible for all corporate income taxes and subject to the usual withholding tax on interest.
- Croatia (1994–2000)
  - Name: Protective interest
  - Base/Rate: Book value of equity / 5 percent plus inflation rate of industrial goods if positive.
  - Details: The notional return is deductible.
- Italy (1997–2003)
  - Name: Dual income tax
  - Base/Rate: Book value of new (post-reform) equity. From 2000: 120 percent of new equity. In 2001: 140 percent of new equity, then again 100 percent of new equity / 7 percent 1997–2000, 6 percent 2001.
  - Details: The notional return is taxed at a reduced rate of 19 percent. Other profits are taxed at 37 percent (34 percent in 2003). Before 2001, the average tax must be at least 27 percent.
  - Additional Italian features and changes:
    - Notional interest qualified for a reduced corporate income tax rate rather than being tax deductible.
    - Only post-reform equity was counted initially; policy changes: 120 percent of post-reform equity in 2000; temporarily 140 percent in 2001; then back to 100 percent.
    - In 2001 the minimum tax rate provision (minimum average tax of 27 percent) was abolished.
- Footnote in Table: 1/ “Kalkulatorische Zinsen auf den Eigenkapitalzuwachs.”

### Previous analyses and literature coverage
- Italy: most intensively studied ACE variant due to economy size and data availability. Italian variant was restricted, frequently changed, and interacted with other taxes in complicated ways. Key descriptive/assessing work: Bordignon, Giannini, and Panteghini (1999 and 2001); overview references include Keen (2003).
- Croatia: some work exists; advantage of being a full ACE system, disadvantage of limited data.
- Austria and Brazil: no detailed studies appear to exist yet (Austrian reform is briefly described in some papers including Genser (2002)).
- Belgium: too early for empirical assessment (as of the document); some theoretical analysis already exists (Gerard 2006a, 2006b).

*Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06259.pdf*

### introduction of a tax on source-based value added. This also addresses differences in

### _wp06259 - introduction of a tax on source-based value added. This also addresses differences in

### Italian ACE variant — description and empirical evidence
- Reform design:
  - System mixes an ACE variant with a CBIT variant and additionally taxes wages; interest deductibility is removed.
  - The reform removed tax preference for debt finance by restricting interest deductibility.
  - The system was partial and subject to frequent changes, complicating theoretical and empirical analysis.
- Theoretical / simulation findings (Bordignon and others (1999); Bordignon, Giannini, and Panteghini (2001)):
  - Reform achieved a reduction in the tax preference for debt finance.
  - Simulations estimate reduced cost of capital for most cases, except where debt-equity ratios are high (because interest is taxable under the local tax).
  - Effective tax rates confirm a reduction in equity discrimination.
  - Reform process balances improved internal efficiency (addressed by ACE feature) with the need for a low tax to prevent profit shifting and capital displacement; restrictions reduced need for compensating tax rate increases.
- Firm-level empirical findings:
  - Staderini (2001) using panel data:
    - Leverage decreased following introduction of the partial ACE.
    - Firms that issued new equity were mainly more profitable firms and those with high investment rates.
    - For many firms, improved tax status of equity was not enough to induce equity issuance during the sample period (data up to 1998).
  - Santoro (2005):
    - Large and profitable firms were more likely to issue new equity and benefit from the system.
    - Small firms and southern firms were less likely to issue equity despite potentially higher cost of debt.
    - Cross-sectional results do not allow a judgment against the pre-reform period.
  - Oropallo and Parisi (2005):
    - Abolition of the Italian ACE variant raised tax rates for most firms (except those benefiting from a new tax consolidation provision).
    - Study did not investigate effects on debt-equity tradeoff.

### Croatian ACE reform — description and empirical evidence
- Reform design and immediate observations (Rose and Wiswesser, 1998):
  - Croatia introduced a full ACE system.
  - Paper explains reform aims and theoretical effects; notes tax revenues increased sharply following the reform but does not attribute increase specifically to the ACE.
- Empirical assessment (Keen and King (2003)):
  - Croatian ACE raised similar amounts of revenue relative to GDP as the EU and Central and Eastern European countries on average.
  - Protective interest reduced tax revenues by roughly one third in 1998, but other tax features compensated.
  - FDI was higher than in all neighboring countries except the Czech Republic.
  - Conclusion: “the ACE passed its first practical test.”
  - Caveat: macroeconomic and transition effects make identification of ACE-specific impacts difficult; more detailed data needed.

### Conclusions from previous research (Italy and Croatia)
- Overall assessment:
  - Results of studies on both reforms are rather weak.
  - Italian results weakened by partial ACE design and frequent system changes.
  - Croatian reform was more radical but analysis hampered by lack of data and confounding effects from transition to a market economy.

### The Brazilian ACE variant — overview (1996 reform)
- Key features of the 1996 reform (Box 1):
  - Until 1995: standard tax system with interest deductibility for debt only; system defined in Fiscal Reference Units to deal with very high inflation.
  - Pre-reform statutory tax rate: 47.7 percent, composed of a main tax rate (25 percent), a surcharge (18 percent) and a corporate social security charge (10 percent).
  - Withholding taxes pre-1996: dividends 15 percent; interest 25 percent.
  - 1996 reform:
    - Abolished use of Fiscal Reference Units.
    - Introduced deductibility of notional interest on equity stock, but only when paid out to shareholders (or credited to owners in closed companies).
    - Reduced statutory tax rate to 30.7 percent, composed of a main rate (15 percent), a surcharge (10 percent) and a social security charge (10 percent).
    - Final withholding taxes reduced to 0 percent on dividends and 15 percent on interest, including notional interest.
    - Depreciation allowances unchanged: 4 percent for industrial buildings and 10 percent for plant and machinery.
  - Footnote on overall rate calculation: (1-.18-.1)*(.1+.18+.25).

### Theoretical effects of the Brazilian variant
- Key distinction from standard ACE:
  - Notional interest deductible only if paid out to shareholders; retained notional interest is not deductible.
  - In absence of transactions costs, firms could distribute notional interest and then issue new shares to restore capital — making effects similar to standard ACE.
  - With transactions costs, many firms may not issue new equity; such firms cannot fully deduct the notional return and face a system effectively giving deductibility of dividends for tax purposes (as “interest to shareholders”).
- Classification of firms:
  - Type L (low dividend payments): firms where dividends are typically less than their notional return (1ˆ − < tt EiD). For L firms, issuing equity to restore distributions may be too costly; they effectively have deductibility of dividends if paid as “interest to shareholders.”
  - Type H (high dividend paying): firms where dividends exceed the notional return (1ˆ − > tt EiD). For H firms, the system operates as a standard ACE since they can deduct the full notional return.
  - Investment projects may switch firms between H and L types, changing applicable tax treatment and effective cost of capital for the project.

### Findings from effective tax rate calculations (EMTR and EATR)
- Methodology:
  - Use of effective average tax rate (EATR) as in Devereux and Griffith (2003); EMTR is obtained as a special case when expected post-tax rent is exactly zero.
  - For the post-reform system, methodology adapted:
    - L-type firms: deductibility of dividends equivalent to a tax credit on dividends (case discussed in Devereux and Griffith).
    - H-type firms: standard ACE treatment; equity stock depends on investment, financing choices, and depreciation rates.
  - Calculations are based on hypothetical values of inflation and interest rates to isolate tax reform effects between 1995 and 1996.
- Selected effective tax rate results (Table 2; EMTR and EATR presented for three sources of finance and both firm types; personal taxes excluded/included; numbers reproduced exactly):
  - Personal Taxes Excluded
    - 1995:
      - EMTR: Ret. earnings 45.1; New equity 45.1; Debt -5.0
      - EATR: Ret. earnings 46.7; New equity 46.7; Debt 37.2
    - 1996 No ACE:
      - EMTR: Ret. earnings 37.6; New equity 37.6; Debt -4.9
      - EATR: Ret. earnings 32.8; New equity 32.8; Debt 23.3
    - 1996 ACE, L-type:
      - EMTR: Ret. earnings 37.6; New equity -4.9; Debt -4.9
      - EATR: Ret. earnings 12.7; New equity -1.0; Debt -1.0
    - 1996 ACE, H-type:
      - EMTR: Ret. earnings 0.0; New equity 0.0; Debt 0.0
      - EATR: Ret. earnings 24.0; New equity 24.0; Debt 24.0
  - Personal Taxes Included
    - 1995:
      - EMTR: Ret. earnings 26.8; New equity 37.8; Debt -5.0
      - EATR: Ret. earnings 46.8; New equity 49.1; Debt 42.9
    - 1996 No ACE:
      - EMTR: Ret. earnings 22.0; New equity 22.0; Debt -4.9
      - EATR: Ret. earnings 27.4; New equity 27.4; Debt 22.5
    - 1996 ACE, L-type:
      - EMTR: Ret. earnings 22.0; New equity -4.9; Debt -4.9
      - EATR: Ret. earnings 16.0; New equity 10.1; Debt 10.1
    - 1996 ACE, H-type:
      - EMTR: Ret. earnings -2.4; New equity -2.4; Debt -2.4
      - EATR: Ret. earnings 22.9; New equity 22.9; Debt 22.9
- Interpretation:
  - The ACE element materially alters effective tax rates across financing sources and firm types.
  - L-type firms can face substantially different (often lower or negative) EMTRs and EATRs for certain financing sources compared to H-type firms and pre-reform values.
  - Table 2 calculations assume an investment in plant and machinery, with a true economic rate of depreciation (table context continued in source).

*Source: IMF working paper _wp06259 - introduction of a tax on source-based value added. This also addresses differences in*

### 12.25 percent. The nominal rate of interest and the rate of the notional return are assumed to be 10 percent. The

### _wp06259 - 12.25 percent. The nominal rate of interest and the rate of the notional return are assumed to be 10 percent. The

### Assumptions and baseline parameters
- Nominal rate of interest: 10 percent.
- Rate of notional return: 10 percent.
- Rate of inflation (hypothetical scenario): 4 percent.
- For EATR calculations, a financial return of 30 percent is assumed.
- Table 3 uses actual observed rates in 1995 and 1996 instead of the hypothetical rates:
  - Rate of inflation: 1995: 66.0 percent, 1996: 15.8 percent.
  - Real rate of interest: 1995: -7.6 percent, 1996: 10.1 percent.

### Effective tax rate (ETR) findings (corporate-level focus)
- Reform outcome:
  - The reform reduced tax rates in the great majority of cases.
  - Exception: effective marginal tax rates for debt-financed investment (some cases not reduced).
- Pre-reform characteristics:
  - Debt-financed investment faced much lower effective tax rates; for investments that just break even, the tax rate was even negative due to depreciation allowances and interest deductibility.
  - Negative tax rates require either other profitable investments or expected future profitability to be exploitable; otherwise the lowest possible rate is 0 percent.
- Post-reform (1996) by firm type:
  - L-type firms:
    - New equity and debt finance are equally attractive from a tax point of view, and both imply an investment subsidy.
    - Finance by retained earnings faces a huge tax disadvantage.
  - H-type firms (effective ACE tax system):
    - Typical ACE outcomes: tax rates independent of source of finance; tax rates for marginal projects are 0.

### Effects when personal income taxes are included
- General effects:
  - Most tax rates are lower when personal taxes are included because the shareholders discount rate is reduced (defined as the post-tax interest rate).
  - In 1995 dividends were taxed; in 1996 dividends were tax free — this change affects comparisons across years.
- L-type firms:
  - Additional dividends are treated as remuneration of equity and are therefore taxed, increasing effective average tax rates.
- H-type firms:
  - Effective marginal tax rates can become negative, implying subsidization of marginal investment.
- Practical neutrality:
  - In theory the reform removed tax-driven debt preference for all firms.
  - In practice, H-type firms may have continued to prefer debt because the notional interest rate was set much lower than the average bank interest rate.

### Table 3 (select extracted ETR/EMTR figures using actual 1995/1996 rates)
- Personal taxes ignored, 1995:
  - EMTR (Ret. earnings): 45.1
  - EMTR (New equity): 45.1
  - EMTR (Debt): -5.0
  - EATR (Ret. earnings): 48.9
  - EATR (New equity): 48.9
  - EATR (Debt): 60.4
- Personal taxes ignored, 1996 (types reported: No, ACE, L, H with values):
  - No / ACE (aggregate rows):
    - EMTR: 47.9 (No/ACE), EMTR Debt: -7.7
    - EATR: 41.7 (No/ACE), EATR Debt: 18.5
  - L-type:
    - EMTR (Ret. earnings): 47.9; EMTR (New equity): -7.7; EMTR (Debt): -7.7
    - EATR (Ret. earnings): 31.0; EATR (New equity): -2.4; EATR (Debt): -2.4
  - H-type:
    - EMTR (Ret. earnings): 25.4; EMTR (New equity): 25.4; EMTR (Debt): -2.7
    - EATR (Ret. earnings): 28.1; EATR (New equity): 28.1; EATR (Debt): 19.5
- Personal taxes included, 1995:
  - EMTR (Ret. earnings): 26.8
  - EMTR (New equity): 37.8
  - EMTR (Debt): -5.0
  - EATR (Ret. earnings): 66.7
  - EATR (New equity): 64.1
  - EATR (Debt): 71.3
- Personal taxes included, 1996:
  - Aggregate (No/ACE):
    - EMTR: 30.1 (No/ACE); EMTR Debt: -7.7
    - EATR: 29.0 (No/ACE); EATR Debt: 16.9
  - L-type:
    - EMTR (Ret. earnings): 30.1; EMTR (New equity): -7.7; EMTR (Debt): -7.7
    - EATR (Ret. earnings): 20.8; EATR (New equity): 6.0; EATR (Debt): 6.0
  - H-type:
    - EMTR (Ret. earnings): 12.1; EMTR (New equity): 12.1; EMTR (Debt): -5.1
    - EATR (Ret. earnings): 21.9; EATR (New equity): 21.9; EATR (Debt): 10.5

### Overall conclusions on effective tax rates
- The reform reduced taxes in most cases; exception: debt-financed projects with low profitability.
- In theory removed tax-driven debt preference for all firms; in practice H-type firms may have retained debt preference due to notional interest rate set too low.
- With personal income taxes included, the system is not neutral and can subsidize investment at the margin:
  - In theory: for all firms (L-type only if using external funds).
  - In practice: subsidization for debt-financed investment and new equity financed investment of L-type firms.

### Empirical analysis: Data and sample construction
- Data source: Thomson Financial.
- Panel characteristics:
  - Over 300 nonfinancial firms for years since 1999; earlier years scarcer.
  - Sample selection: firms with at least three observations, including at least one pre-reform and one post-reform observation; observations restricted to years 1990–2000.
  - Final sample: 98 firms and 835 observations (sample size reduced for some regressions).
- Reporting practice given high inflation: only ratios of variables are shown.

### Descriptive statistics (Table 4; in percent except counts)
- Observations:
  - Debt/Total Assets: 835 observations.
  - Equity/Total Assets: 835 observations.
  - Profits/Sales: 827 observations.
  - Real Sales Growth: 710 observations.
  - Dividends/Sales: 828 observations.
- Means and dispersion:
  - Mean Debt/Total Assets: 13.3
  - Mean Equity/Total Assets: 41.1
  - Mean Profits/Sales: -6.5
  - Mean Real Sales Growth: 15.4
  - Mean Dividends/Sales: 2.5
  - S.d. Debt/Total Assets: 11.6
  - S.d. Equity/Total Assets: 44.7
  - S.d. Profits/Sales: 124.1
  - S.d. Real Sales Growth: 94.4
  - S.d. Dividends/Sales: 4.9
- Distribution percentiles provided (Lower decile, Lower quartile, Median, Upper quartile, Upper decile) for each ratio.

### Dividend payout distribution and behavior
- Notional return estimation:
  - Estimated by multiplying the legally prescribed interest rate by previous year’s book value of equity.
  - For years prior to 1995 (no legal rate), the rate is predicted via regression of the money market rate on prescribed interest rates; prescribed rate tends to increase by 44 percent with the money market rate.
- Table 5: Distribution of low (L-type) and high (H-type) payout firms, selected entries:
  - 1992: L-Type Number 52 (proportion 19 percent), H-Type Number 100 (36.5 percent), firms omitting dividends 0.
  - 1995: L-Type 69 (25) ; H-Type 2 (2.8) ; proportion for H-Type 97.2 (35.2) reported in table formatting.
  - 2000: L-Type 73 (24) ; H-Type 13 (15.1) ; proportion for H-Type 84.9 (27.9) reported.
- Main empirical findings:
  - Great majority of firms pay dividends lower than the notional return.
  - Prior to 1995, no firm in the sample paid dividends in excess of the notional return.
  - For most Brazilian firms the system functioned effectively as dividend deductibility rather than a standard ACE.
- Proximity to notional return:
  - Very few firms pay out dividends reaching at least 90 percent of the notional return: peak year 1998 share among low-payout firms is 5.4 percent; other years < 5 percent.
  - More firms pay just above the notional return (up to 110 percent); among high pay-out firms, share reaches 38 percent in 2000 (but is 0 percent in some years).

### Average share of notional interest paid out (Table 6; in percent)
- Yearly mean share of notional interest paid out by firms:
  - 1992 Mean: 3.1; S.d.: 4.7; Median: 0.7 (Obs. 52)
  - 1993 Mean: 4.5; S.d.: 5.9; Median: 2.4 (Obs. 57)
  - 1994 Mean: 1.2; S.d.: 1.2; Median: 1.0 (Obs. 58)
  - 1995 Mean: 14.0; S.d.: 23.5; Median: 7.3 (Obs. 71)
  - 1996 Mean: 18.7; S.d.: 29.7; Median: 10.3 (Obs. 96)
  - 1997 Mean: 45.5; S.d.: 66.0; Median: 28.5 (Obs. 86)
  - 1998 Mean: 35.9; S.d.: 54.6; Median: 17.4 (Obs. 84)
  - 1999 Mean: 30.0; S.d.: 51.8; Median: 15.1 (Obs. 83)
  - 2000 Mean: 45.8; S.d.: 75.0; Median: 23.9 (Obs. 91)
- Interpretation:
  - Important increase in dividends relative to notional returns after the 1996 reform, with most of the increase starting in 1997.
  - Distribution is skewed; median always below mean.

### Regression analysis of dividend payouts (Table 7)
- Dependent variable: Cash dividends (various specifications).
- Key regression results:
  - Regression (1) OLS: Profits coefficient 0.159 (0.048)***.
  - Regression (2) OLS with Post dummy and interaction: Profits 0.059 (0.042); Post * Profits 0.106 (0.070).
  - Regression (3) Firm fixed effects: Profits 0.017 (0.017); Post * Profits 0.109 (0.046)**.
  - Regression (4) Tobit with firm fixed effects: Profits 0.067 (0.042); Post * Profits 0.116 (0.041)***.
  - Post dummy coefficients (levels) are significantly positive in (2)-(4) reflecting higher nominal dividends after reform (inflation effects).
- Interpretation:
  - After controlling for firm fixed effects and left-censoring (Tobit), a significantly higher share of profits is paid out after the reform.
  - Pre-reform dividends may not have been strongly linked to profits; consistent with firms minimizing dividends (new view).

### Capital structure: aggregate trends (Table 8)
- Equity/Total Assets and LT Debt/Total Assets over time (in percent):
  - 1992: Equity/Total Assets 50.6; LT Debt/Total Assets 10.0; Residual 39.4
  - 1995: Equity/Total Assets 50.2; LT Debt/Total Assets 09.9; Residual 40.0
  - 1996: Equity/Total Assets 42.2; LT Debt/Total Assets 13.5; Residual 44.3
  - 1997: Equity/Total Assets 38.4; LT Debt/Total Assets 16.4; Residual 45.2
  - 1998: Equity/Total Assets 32.0; LT Debt/Total Assets 17.1; Residual 50.9
  - 1999: Equity/Total Assets 28.0; LT Debt/Total Assets 17.5; Residual 54.5
  - 2000: Equity/Total Assets 23.7; LT Debt/Total Assets 16.4; Residual 59.9
- Aggregate implication:
  - Post-reform equity ratios fell substantially, reaching about half pre-reform level by 2000.
  - Long-term debt ratios increased; residual financing share also increased steadily.

### Capital structure regressions (Table 9)
- Dependent variable: Debt/Total Assets (specifications 1–3) and Equity/Total Assets (4).
- Key coefficients:
  - Post dummy positive and significant for Debt/Total Assets across specifications:
    - (1) Post 0.061 (0.008)*** — OLS
    - (2) Post 0.062 (0.009)*** — OLS with controls
    - (3) Post 0.057 (0.007)*** — Firm fixed effects
    - (4) Post -0.125 (0.012)*** — Firm fixed effects (dependent = Equity/Total Assets)
  - Profitability:
    - Regression (2): coefficient -0.010 (0.004)** (more profitable firms issue less debt).
    - Regression (3) firm fixed effects: coefficient -0.016 (0.004)*** for debt; regression (4) mirror image for equity 0.020 (0.010)**.
  - Real sales growth:
    - Regression (3) firm fixed effects: coefficient 0.011 (0.005)** positive for debt.
- Interpretation:
  - Increase in debt and decrease in equity after the reform cannot be fully explained by observable firm-level changes in profitability or sales growth.
  - Majority of firms appear to face dividend-deductibility incentives rather than ACE, encouraging external finance and increased dividends; if external finance is partly debt, debt/equity increases.

### Investment effects (Table 10 and analysis)
- Motivation: ACE variant intended to reduce discrimination of equity finance; ultimate goal to increase investment.
- Investment regressions — dependent variable Investment/Capital:
  - (1) OLS: Post 0.039 (0.013)***.
  - (2) OLS with controls: Post 0.045 (0.014)***.
  - (3) Firm fixed effects: Post 0.026 (0.012)**.
  - (4) Firm fixed effects with lagged profitability and real sales growth:
    - Post 0.023 (0.013)*.
    - Profitability t-1 0.089 (0.037)** positive and significant.
    - Real sales growth and its lag positive; real sales growth t-1 (0.004)* reported.
  - (5) Firm fixed effects with Tobin’s Q: Post -0.001 (0.016) (insignificant); Tobin’s Q coefficient 0.041 (0.022)*.
- Interpretation:
  - Investment rates increased following the reform (post dummy positive and significant in most specifications).
  - Lagged profitability being positive and significant suggests firms rely on retained earnings to finance investment.
  - When Tobin’s Q is included (a sufficient statistic under certain conditions), the post-reform dummy becomes insignificant, consistent with the reform affecting valuation (Q) rather than immediate observable investment regressors.
- Overall conclusion on investment:
  - Investment rates rose after the reform; however, causality cannot be definitively attributed solely to the reform given possible other explanations.

*Source: Author’s calculations and tabulations as presented in the supplied content.*

### Conclusions from the Brazilian Reform

### Conclusions from the Brazilian Reform

### Design and immediate effects of the 1996 reform
- In 1996, Brazil introduced a tax reform that reduced corporate income taxes and allowed dividends to be deductible up to a maximum defined by notional return on equity.
- The reform’s mechanism: dividend deductibility up to a notional return makes the system a variant of an ACE (Allowance for Corporate Equity), but the precise incidence depends on dividend payout ratios.
- Empirical finding: the combined effect of the cut in statutory tax rates and changes in the method of taxation reduced effective tax rates for most investment projects, except those with very low levels of profitability financed by debt.

### Firm-level incidence: L-type versus H-type firms
- Distinction:
  - L-type firms: low dividend payout ratios; they face full dividend deductibility (not ACE in practice).
  - H-type firms: high dividend payout ratios; they face an ACE-like system (deduction up to notional return).
- Empirical observations:
  - Very few firms have dividends high enough to ensure they face an ACE system; for most firms the new system operates as dividend deductibility.
  - Firms with high payout ratios face an ACE system, while those with low ratios have full dividend deductibility.
  - In practice, the new system produced an increase in dividends and a slow increase in the share of H-type firms.

### Effects on corporate financing and capital structure
- Theoretical neutrality: dividend deductibility achieves neutrality between debt and equity as sources of finance for new investment.
- Empirical outcome: despite the expectation of a shift from debt to equity if the previous system favored debt, the data did not find a move toward more equity finance.
- Behavioral incentive: dividend deductibility creates an incentive to raise external funds and increase dividends, in principle up to the point where dividends exceed the notional return. Transactions costs and legal restrictions limiting dividends to profits limit this turnover.
- Observed financing outcomes:
  - Data reveal an increase in dividend payments.
  - Data show an increase rather than a fall in debt-equity ratios.

### Investment and macroeconomic outcomes
- Investment increased following the reform.
- Ambiguity remains: it is not possible to disentangle whether the increase in investment was due to lower statutory tax rates or the specific structure of the new system (dividend deductibility / ACE variant).

### Longer-run dynamics and research implications
- In the longer run, the Brazilian system may increasingly resemble a true ACE system if firms increase their debt-equity ratios and dividend payments, causing more firms to switch from dividend deductibility to classical ACE treatment.
- The paper calls for further research using longer panels and analyses of differential effects across firm types to evaluate longer-term transitions.
- Caution for policy transfer: given the special features of the Brazilian ACE variant and limited empirical evidence from other partial ACE systems (Croatia being the only other example noted), it is difficult to draw firm lessons for countries (e.g., Belgium) planning to introduce an ACE. The paper hopes that future implementations (e.g., Belgium) will last long enough to permit detailed analysis.

### Appendix I — Effective tax rates and properties of ACE (methodological highlights)
- Methodology: builds on Devereux and Griffith effective tax rate framework (EATR and EMTR) adapted to incorporate dividend deductibility and the Brazilian ACE variant.
- Numeric parameters highlighted in the Brazilian context (1996):
  - Personal tax rate on remuneration to equity: 15 percent.
  - Corporate income tax rate: 30.4 percent.
  - No tax on unrealized capital gains, yielding γ = 1.22 for L-type firms under the specified parameterization.
- Key formal properties of an ACE (when notional interest is set at the correct level, î = i):
  - (i) Neutrality between debt and equity finance: financing effects are zero under the ACE adaptation.
  - (ii) Irrelevance of depreciation schedules: depreciation allowances do not affect effective tax rates under the ACE adaptation (provided personal taxes are excluded).
  - (iii) Effective marginal tax rates (EMTR) of 0 if personal taxes are excluded: the ACE system is neutral to investment in that case, with the cost of capital equalling the real rate of interest and EMTR = 0.

### Key empirical and parameter figures (preserved)
- Year of reform: 1996.
- Personal tax rate on remuneration to equity (1996 example): 15 percent.
- Corporate income tax rate (1996 example): 30.4 percent.
- γ (tax discrimination factor under the 1996 parameterization): 1.22.

*Conclusions from the Brazilian Reform — IMF working paper _wp06259*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06259.pdf_