CORPORATE TAX INCENTIVES AND CAPITAL STRUCTURE: EVIDENCE FROM UK TAX RETURN DATA

Transcription

1 CORPORATE TAX INCENTIVES AND CAPITAL STRUCTURE: EVIDENCE FROM UK TAX RETURN DATA Jing Xing, Giorgia Maffini, and Michael Devereux Centre for Business Taxation Saïd Business School University of Oxford

2 Research background Research question: Do corporate tax incentives affect firms capital structure? New data: We use confidential UK tax return data matched with accounting data for analysis; Identification strategies: We use variation in the tax incentives that firms face due to: kinks in the corporate tax rate schedule; asymmetric tax treatment between profit and loss.

3 Contribution of the current study Advantages: We use tax return data instead of financial statements; More accurate measure of MTR especially for firms with positive taxable profits; Data matched with ownership information; Findings: A positive and sizable corporate tax effect on firms capital structure; Large difference in terms of the taxable profits between the tax return data and financial statements; Studies using financial statements are likely to find a smaller tax effect on leverage ratio; Multinational firms capital structure responds to corporate tax incentives mainly via internal debt policies.

4 Structure of the presentation A brief literature review Description of the data Estimation results Estimating tax effects based on a partial adjustment model of leverage; Comparison between results based on tax returns and those based on financial statements; Does ownership matter? Domestic versus multinational firms. Conclusions

5 Literature review Three types of empirical studies: Type 1: Graham (1996a), (1998) Data: firm-level data from a single country; Identification: variation in the simulated marginal tax rate (SMTR) due to the asymmetric tax treatment of loss and profit for identification (Graham, 1996a, 1998); Results: 0.07 in Graham (1998); Comments: potentially large measurement noises in the SMTR. Type 2: Rajan and Zingales (1995); Faccio and Xu (2012) Data: cross-country firm-level data; Identification: variation in the statutory CIT across countries Results: 0.4 in Faccio and Xu, 2012 Comments: the top CIT rates could differ from the true tax incentives; tax rates are highly likely to be related to country-specific business cycles. Type 3: Desai, Foley, and Hines (2004) Data: subsidiary-level data of multinational firms; Identification: variation in the statutory CIT across countries; Results: significant tax effects on debt shifting. Comments: tax effects on debt shifting is different from the tax effects on the overall leverage; domestics firms different from multinationals.

6 Sample construction Tax return data matched with FAME: ; Criteria: Unconsolidated accounts only; fiscal year starting from April for 12 months; non-financial firms; continuous observations for at least 4 years; Two samples: Sample 1: Firms with continuous positive taxable profits only; Sample 2: Firms making both profits and losses; Key variables: Leverage ratio=(total debt)/(total debt+book equity); Corporate tax incentives: After-financing statutory marginal tax rate (based on box37 in tax returns); Before-financing statutory marginal tax rate (box 37 plus interest expenses); perfect-foresight marginal tax rate (sample 2): MTR P i,t = Control variables: size, tangibility, and profitability; MTR i,t, π i,t > 0 MTR i,t+s 1 + r s, π i,t 0, π i,t 1 0, π i,t+s > 0 MTR i,t 1 (1 + r), π i,t 0, π i,t 1 > 0, loss carried backward Standard data cleaning procedure to deal with outliers, gaps, etc: majority of firms do not report leverage ratio in FAME.

7 UK corporate tax rate schedule: an illustration MTR 2000/ /02 MTR 2002/ /06 MTR 2006/ K 300K ,500K Taxable profits

8 Levergae ratio Data description: Average leverage ratio Unbalanced Balanced

9 Partial adjustment model of capital structure We estimate a partial adjustment model as: Lev i,t = αlev i,t 1 + β 1 MTR i,t + β 2 MTR i,t 1 + γ 1 Z i,t + γ 2 Z i,t 1 + μ i + θ t + ε i,t Re-arrange to the Error-correction model (ECM): Lev i,t = α 1 Lev i,t 1 + (β 2 + β 1 )MTR i,t 1 + (γ 2 +γ 1 )Z i,t 1 + β 1 MTR i,t + γ 1 Z i,t + μ i + θ t + ε i,t where the LR tax effect on leverage is measured by

10 Distribution of the statutory marginal tax rate: Sample 1

11 Distribution of the statutory marginal tax rate: Sample 2

12 Table 1:Transitional probability of the taxable profits Year t Taxable Profits LOSS 0-50K K >300K LOSS 69.57% 16.04% 9.62% 4.77% Year t K 19.98% 59.02% 18.30% 2.70% K 11.55% 13.81% 58.96% 15.68% >300K 8.79% 2.89% 16.40% 71.93%

13 Table 2: Number of tax status changes within firms Sample 1: Total number of tax status changes Number of changes Number of firms Percent of total firms Number of Percent of total observations observations % 11, % % 13, % % 12, % % 7, % % 3, % >= % 11, % Total 9, % 51, % Sample 1: Total number of moving in and out of the 300,000 tax bracket 0 6, % 32, % 1 1, % 8, % 2 1, % 6, % >= % % Total 9, % 51, % Sample 2: Total number of tax status changes % 16, % % 17, % % 22, % % 16, % % 10, % >= % 9, % Total % 93, % Sample 2: Total number of moving in and out of losses 0 8, % 44, % 1 3, % 22, % 2 2, % 16, % >= % 10, % Total 16, % 44, %

14 Endogeneity of the marginal tax rate Sources of endogeneity: Debt policy directly affects both the leverage ratio and firms after-financing taxable profits (Graham, 1996); Kinks in the tax rate schedule may give incentives for firms to reduce their marginal tax rate by using more debt; Firms taxable profits (before or after-financing) are highly likely to be correlated with unobserved firm-level shocks; Solutions: Using the before-financing MTR as the proxy for corporate tax incentives. Instrumental variables (IVs) estimations: we use lags of the current marginal tax rate as instruments

15 Table 3:Estimation results using kinks in the tax rate schedule OLS FE IVs GMM VARIABLES (1) (2) (3) (4) Convergence speed *** *** *** *** (0.002) (0.008) (0.015) (0.034) LR coefficient Tax *** *** 0.399** 0.764** (0.091) (0.026) (0.183) (0.379) Profitability *** *** *** (0.103) (0.039) (0.069) (0.828) Size ** 0.074*** 0.089*** 0.507* (0.005) (0.008) (0.012) (0.296) Tangibility 0.034** 0.141*** 0.151*** * (0.017) (0.021) (0.030) (0.586) Hansen test Weak instrument NA F-statistics NA AR(1) AR(2) AR(3) Year dummies Yes Yes Yes Yes Firm FE No Yes Yes Yes Observations 41,612 41,612 32,173 32,173 Number of group 9,439 9,439 9,439 9,439

16 Table 4:Estimation results using Sample 2 Statutory afterfinancing MTR Perfect-foresight MTR (2-3) VARIABLES (1)GMM (2)GMM (3)GMM Convergence speed *** *** *** (0.024) (0.024) (0.026) LR coefficient Tax 1.044*** 0.980*** 1.115** (0.243) (0.279) (0.443) Observations 61,011 61,011 50,906 Number of group 16,124 16,124 15,158 Notes: We use variables dated at t-3 and t-4 as instruments in the GMM estimations. In Column 3, we omit the first and last observations with losses and re-estimate the ECM.

17 Table 5: Correspondence between tax return data and accounting data CT600 FAME Taxable profits Loss 0-50,000 50, ,000 >300K Loss , , , >300,

18 Table 6: Comparing CT600 and FAME, GMM estimations Sample 1 (1&2) Sample 2 (3-6) Statutory MTR Statutory MTR Perfect-foresight MTR (1) (2) (3) (4) (5) (6) Convergence speed *** *** *** *** *** *** (0.034) (0.034) (0.024) (0.024) (0.026) (0.026) LR coefficients TAX (CT600) 0.764** 1.044*** 1.115** (0.379) (0.243) (0.443) TAX (FAME) ** (0.455) (0.291) (0.601) Hansen test AR(1) AR(2) AR(3) Observations 32,173 32,173 61,011 61,011 50,906 54,797 No. of groups 9,439 9,439 16,124 16,124 15,158 15,747 Notes: We report GMM estimation results of the ECM in this table, using MTRs calculated either using tax returns or accounting data. In Columns 5 and 6, we omit the first and last observations for each firm if they are in the loss-making positions. The sample size changed as there is no perfect match between the tax return data and FAME.

19 Does ownership matter? We define a firm to be domestic if: it is independent and has no foreign subsidiary; it is a subsidiary of a UK group and the group has no foreign subsidiary. A firm is defined to be multinational if : it is independent and has foreign subsidiaries; it is a subsidiaries of a UK group and the UK group has foreign subsidiaries; it is a subsidiary of a non-uk group. Caveats: We do not observe the ownership information for two-thirds of the sample; Ownership information is only available in the most recent year reported in FAME and it is assumed to be unchanged over time.

20 Table 7: Domestic versus multinational firms Total leverage External Internal (1) (2) (3) (4) VARIABLES Multinationals Whole sample Multinationals Multinationals Convergence speed *** *** *** *** (0.047) (0.025) (0.052) (0.067) LR coefficient MTR (Domestics) 0.946*** (0.310) MTR (Multinationals) 1.310*** 1.415*** ** (0.434) (0.456) (0.333) (0.371) Test of equal coefficients (p-values) Hansen test AR(1) AR(2) AR(3) Year dummies Yes Yes Yes Yes Firm fixed effects Yes Yes Yes Yes Observations 18,214 61,011 18,214 18,214 Number of groups 4,597 16,124 4,597 4,597

21 Conclusions We use kinks in the UK corporate tax rate schedule to identify the effects of corporate tax incentives on firms leverage ratio; We find a positive link between corporate tax incentives and firms capital structure, once we control for the endogeneity of the marginal tax rate; Simply using the current before-financing marginal tax rate is unlikely to correct the endogeneity of the MTR; Using accounting data for this analysis is likely to bias downward the estimated tax effects on leverage; There is some evidence that the capital structures of multinational firms respond more strongly to the corporate tax incentive than domestic firms.