Company Size and Effective Corporate Tax Rate

Company Size and Effective Corporate Tax Rate Study on Indian Private Manufacturing Companies

Economic and Political WeeklyMay 19, 20071870From equations (2) and (3):jptaxit = τt πg – τt Σδjit – Σcrpitj=1 p=1jpor, taxit /πg = τt – τt /πg Σδjit –1/πgΣcrpit...(4)j=1 p=1Then they argue that there is nothing in the data to enable themto distinguish between cr and δ. To identify the model they haveused proxies for cr and δ. But the problem is that (4) is not anequation but an identity. Proxies for cr and δ do not alter that.So regression estimation here is problematic.The regression model formulated by Gupta and Newberry islogical but it can be improved upon by introducing scheduledtaxrate asdone by Harris and Fenny. Here, we have introduced thescheduled tax rate into Gupta and Newberry’s model. This modelis different from the Gupta and Newberry model for other reasonstoo. These differences have arisen because of our attempt tocapture thespecificities of the Indian corporate tax system. Thevariable forprofitability of companies is dropped from the equationas Indiancompanies irrespective of their profitability faced asingle scheduled tax rate in the 1990s. We have incorporatedthe export intensityas a tax exemption that was available for theprofit earned by export activity. A dummy for minimum alter-native tax is also introduced. Hence, it is an improvement overboth the models and also has taken into consideration the speci-ficities of the Indian tax system.In Indian corporate tax literature, there is very little on therelationship between effective corporate tax rate and companysize. Goyal (1996) shows that between 1975-76 and 1994-95,companies with paid up capital of more than Rs 25 crore as agroup had lower effective tax rate than the group of companiesconsisting of those having paid up capital less than Rs 1 crore.This study does not provide any evidence to establish a causalrelationship between the company size and corporate effectivetax rate. Also, it was not set in a multivariate framework. Anotherlimitation of this study is that it is silent about the company groupwith a paid up capital that is between Rs 1 crore to Rs 25 crore.The most probable reasons for this negative relationship betweeneffective tax rate and company size may be that larger companiesare use more capital-intensive technology and they have a greaterproportion of borrowed funds as a source of their funds. Thispaper could not eliminate the possibility that these two companycharacteristics are the reasons behind this negative relationship.Our paper by using a multivariate framework provides evidenceregarding the negative causal relationship between ETR and thesize of the companies, after eliminating the effect of companies’use of more capital-intensive technology and greater use ofborrowed funds as a source of funds.Section I briefly discusses the trends in scheduled corporatetax rates and effective corporate tax rates. Section II consists ofa statistical exercise in a bivariate framework and an econometricmodel through which the relationships between ETRs and varioustax exemptions, scheduled corporate tax rates and the size of thecompany are examined. Section III presents the results and fewconcluding remarks.IOverview of Tax Rates for Indian Corporate SectorCorporate tax in India is a fixed percentage of the net profitafter allowing for admissible expenses. It has to be self-assessedin nature and the tax provisions for payment of taxes are madein the company balance sheet. During the last two decades, thecorporate tax rates underwent repeated changes. The trend hasbeen to lower the rates (Table 1). Graph A clearly shows thedownward trend in the last two decades.At the beginning of the 1980s, the scheduled tax rate wasaround 60 per cent. By the end of that decade it came down tothe range of 50 to 54 per cent. In the year 1991-92, the scheduledtax rate came down to 46 per cent for private and public limitedcompanies.4 In the year 1992-93, the scheduled tax rate increasedto 52.5 per cent. From 1992-93 to 1999-2000, the scheduled taxrates were reduced or remained the same in almost every year.But by 1999-2000, the scheduled rate was reduced to 35 per cent,for both types of companies. Following the introduction ofsurcharge, the scheduled rate again climbed up in 2000-01 and2001-02. Reduction in the surcharge once again brought downthe scheduled tax rate close to 35 per cent by the assessmentyear 2002-03. To prevent companies from reducing their taxliability to zero, initially, the budget for 1996-97 introduced theMAT. But ultimately in the 2000-01 budget proposal, it waslevied uniformly at 7.5 per cent of the book profits as determinedunder the companies act.5In the Indian corporate tax system, a variety of concessionsin the form of tax exemptions are provided to corporations, whichresult in substantial reduction in their overall tax liabilities. Itmay be the cause of lower tax provision by the Indian corporatesector, as a percentage of the profits before tax than the scheduledrates [Gupta 1981, 1984 and Goyal 1996]. Table 2 and GraphAshow, that the ETR has decreased over the last two decades andit is much lower than the scheduled tax rate.Table 1: Income Tax and Surcharge for Private and PublicLimited Companies Assessment Years: 1990-91 to 2002-03(Per cent)Assessment YearDomestic Company (Private and Public)Income Tax RateSurchargeScheduled1980-81557.559.1251981-82557.559.1251982-83552.556.3751983-84552.556.3751984-8555557.751985-8655557.751986-8750552.51987-88500501988-8950552.51989-9050552.51990-91508541991-924015461992-93451551.751993-94451551.751994-95451551.751995-964015461996-974015461997-98407.5431998-9935Nil351999-0035Nil352000-01351038.502001-02351339.552002-0335235.70Notes:(1)For 1990-91 surcharge rate was 8 per cent of corporate tax of adomestic company, if total income exceeds Rs 50,000.(2)1991-92 to 1996-97 surcharge at the rate of 15 per cent of corporatetax of a domestic company, if total income exceeds Rs 75,000.(3)1997-98 surcharge at the rate of 7.5 per cent of corporate tax of adomestic company, if total income exceeds Rs 75,000.(4)2000-01 surcharge at the rate of 10 per cent on corporate tax ispayableby the domestic company on the whole amount of income tax.(5)2001-02 surcharge at the rate of 13 per cent on corporate tax ispayable by the domestic company on the whole amount ofincometax.(6)2002-03 surcharge at the rate of 2 per cent on income tax is payableby the domestic company on the whole amount of income tax.Source:Budget documents of various years, government of India.

Economic and Political WeeklyMay 19, 20071872are major company characteristics, which can affect ETR. InIndia, tax exemption is permitted for companies on the profitthey earn by foreign export. Export intensity of a company isanother characteristic that can affect ETR. The above discussionsuggests the following hypotheses to be examined:H1: Indian private manufacturing companies’ average effectivetax rates are related to their sizes.H2: Indian private manufacturing companies’ average effectivetaxrates are related to their financing and investment decisions.H3: Indian private manufacturing companies’ average effectivetaxrates are positively related to the scheduled tax rate, which theyface.There are many other activities on which tax exemptions7 areavailable. At the firm level, the required data of these are notavailable to the public. Our empirical model does not includethem because of this reason. An advantage of the panel dataanalysis used in our study is that it likely accounts for the effectsof these variables if they do not vary greatly over the sampleperiod. In India there was no great change in the exemption itemsin the sample period. Also, total tax exemptions available to thecompanies for these items are not high enough. At the most, itis around 15 per cent. The reason behind it is that the proportionof interest payment and depreciation in the total of tax exemptionsand tax payments is more than 84 per cent. This 84 per cent willincrease further, if the amount of tax payment is not includedin total of tax exemptions and tax payments.Sample Selection and Data SourceThe sample consists of panels of Indian private manufacturingcompanies from the Prowess corporate database of 2002, pub-lished by the Centre for Monitoring Indian Economy for theperiod 1992-2001. The Indian government does not publish thedetailed corporate tax payment at the company level. The onlysource available is the tax provisions shown by different com-panies in their annual reports. That does not provide the detailedbreak up of corporate tax paid by the companies on differentaccounts. So we do not have all the data at the company levelregarding tax exemptions, which companies get through differentprovisions. To address this problem we have identified the majorcompany characteristics that may help them to get more taxconcessions. The required data are collected from the Prowesscorporate data.This 10-year period is chosen because it is the first 10 yearsafter the new economic policies was taken by the governmentfollowing the Structural Adjustment Programme formulated bythe International Monetary Fund and World Bank. As a result,this period has witnessed a rapid reduction in the corporatescheduled tax rate. We have selected all the Indian privatemanu-facturing companies whose financial variables are available forthe whole sample period. Following this criteria, the number ofcompanies in the sample is 901. Among these, the sales dataforsome of the companies is not available for some years. So our panelis an unbalanced one. Total number of observations is 8,900.Empirical Evidence in Bivariate FrameworkTo examine the relationship between ETR and company size,the correlation coefficient between ETR and real total assets (realtotal assets is used as the indicator of size of the company) ofeach company is calculated. We found that the correlationcoefficient is negative for the period 1992-2001. Further, thecorrelation coefficients between these two variables for each yearare calculated. With the exception of 1992 and 2001, for otheryears it is negative (Table 3).Empirical Model in Multivariate FrameworkOur regression equation isETRit = α + ß1sizeit + ß2levit + ß3capinvintit + ß4tot_expsaleit+ ß5strt +ß6matwhere ETRit is the average effective tax rate for company i inyear t and the independent variables (with subscripts omitted)include proxies for company size, capital structure (lev), aasetmix (capinvint), export intensity (tot_expsale), scheduled tax rate(tr) and a dummy variable (mat) for the effect of minimumalternative tax. ETR is generally measured as tax liability divided by income.But there is considerable amount of controversy regarding theappropriate definition of both the numerator and denominator.We are not elaborating on this issue as it is sufficiently discussedelsewhere [Gupta and Newberry 1997; Omer et al 1991; p 59-62; Shevlin and Poter 1992, p 63-64; Wilkie and Limberg 1993;p 83-85]; also it is beyond the scope of this small paper. WithTable 6: Fixed Effects Regression Results of Effective TaxRates on various Firm CharacteristicsModel: ETRit =α + ß1sizeit + ß2levit + ß3capinvintit + ß4tot_expsaleit + ß5strt+ß6matwhere: size = natural logarithm of real total asset (lrtot_asset)VariablePredicted SignValue of CoefficientStd Err zP>|z|capinvint--.0658302.0063433-10.380.000lev--.0674635.0045904-14.700.000Lrtot_asset--.0164674.0019862-8.290.000tot_expsale--.0247781.0074558-3.320.001mat+.0101049.00318253.180.002str+.0017404.00025556.810.000Cons+.0559584.01299754.310.000Table 3: Correlation Coefficient betweenCompany Size and Effective Tax Rate1992-20011992199319941995199619971998199920002001Correlationcoefficient-0.0230.036-0.022-0.046-0.033-0.008-0.02-0.03-0.0110.007-0.01Table 4: Random Effects Regression Results of Effective TaxRates on Various Firm CharacteristicsModel: ETRit =α + ß1sizeit + ß2levit + ß3capinvintit + ß4tot_expsaleit + ß5strt+ß6matwhere: size = natural logarithm of real total asset (lrtot_asset)VariablePredicted SignValue of CoefficientStd Err zP>|z|capinvint--.0638153.0054182-11.780.000lev--.0599431.0038337-15.640.000Lrtot_asset--.0078946.001315-6.000.000tot_expsale--.0385549.0065829-5.860.000Mat+.0094078.00317972.960.003str+.0019392.00025277.670.000cons+.0499675.01306393.820.000Table 5: Specification Tests for Appropriate Panel Data ModelSpecification Tests ×2 ProbabilityBreusch and Pagan Lagrangian multipliertest for random effects×2 (1) = 10576.95Prob > χ2 = 0.000Hausman specification test×2 (6) = 51.91Prob > χ2 = 0.0000

Economic and Political WeeklyMay 19, 20071873regard to the numerator, we are using the current portion of eachcompany’s income tax expense in India. This is because thereare very few Indian transnational corporations and our sampleofcompanies excludes the foreign companies. For the denominator,profit before tax depreciation and interest payment (PBDIT) isused. Since we are capturing the impact of tax exemptions onETR, the taxable income (profit before tax) is not being used.If both the numerator and the denominator are partially or fullyaltered with tax exemptions, then any systematic variation inETRs due to tax exemptions may not be captured. For thosecompanies with negative PBDIT (ie, suffered loss), the cor-responding ETRs are replaced by zero and for those companieswhose ETRs are greater than one, the corresponding values ofETRs are replaced by one, when PBDIT is greater than zero.Company size is measured as the natural logarithm of real totalasset. The deflator is the corresponding year’s wholesale priceof industries. Financial leverage (lev) is measured as the ratiobetween total debt and total assets. Capintinv is a proxy for assetmix. It is defined as the ratio of net property, plant and equipmentto total assets. Export intensity (tot_expsale) of a company ismeasured by the ratio between export and sale. The scheduledtax rate (str) is calculated with the following formula: scheduledtax rate = corporate tax rate + corporate tax rate × percentageof surcharge. “mat” is a dummy variable, which captures theeffect of MAT on effective tax rate of companies.A panel data fixed effects regression model has several ad-vantages over those estimated with cross-section or time seriesdata or a simple pooled cross section time series model. Thesemodels will not provide unbiased and consistent parameterestimates if the unobserved firm specific characteristics arecorrelated with the included explanatory variables. A fixed effectsmodel overcomes this problem by accounting for individual firmheterogeneity via firm specific constant or intercepts in the model,which capture the effects of unobserved or unmeasurable com-pany characteristics that vary by companies but are relativelystable over time for a given company. However, the limitationof a fixed effects model is that it gives parameter estimates, whichare conditional or sample specific, so inferences can not begeneralised outside the sample. A random effects model over-comes this limitation by viewing the individual specific char-acteristic as a normally-distributed random variable but it alsoassumes that individual-specific effects are uncorrelated with theregressors. In this case, it is hard to justify. The scheduled taxrate is one item, which should have an influence on the ETRof companies. This does not vary by companies but over time.Following Harris and Fenny (1999), we have used scheduled taxrate as a proxy for time specific effects in this fixed effect model.We have taken the help of specification tests to justify our modelselections. Those tests are the Breusch-Pagan Lagrangian mul-tiplier test for random effects and Hausman specification test.Table 6 summarises the fixed effects regression results. Table4gives us the random effects regression results and Table 5 providesus the result of specification tests. The results in Table 5 suggestthat the fixed effects model is most appropriate model amongthe fixed effects, random effects and simple-pooled cross sectiontime-series model. The Breusch-Pagan Lagrangian multiplier testresult suggests that the company specific effects models arepreferred over the simple-pooled cross section time-series model.The Hausman specification test tells us that the fixed effectsmodel is more appropriate than the random effects model. If wecompare the value of the coefficients of both the fixed and randomeffects models, not much of a difference is found either in thesigns or values of the coefficients. Further, to evaluate therobustness of the whole sample, the multicolinearity test wasperformed. As far as outliers are concerned, those observationswere removed from the sample and the regressions were run.The result was similar those of Tables 4, 5 and 6. Rather, it wasslightly better: in the sense, that negative coefficients becamemore negative and positive coefficients became more positive.IIIResults and ConclusionsThe proxy for company size (lrtot_asset) provides evidencerelated to hypothesis 1. Company size has a negative and signifi-cant association with average ETR. Also, other company char-acteristics such as capital structure, asset mix and export intensityhave negative associations with the average ETR. Introductionof these company characteristics, separately, in the regressionequation should reduce the negative association between com-pany size and average ETR. It is because of these characteristicsthat companies capture the major tax exemptions. Despite sepa-rating out the impact of these company characteristics, theassociation between the company size and average ETR is negative.It gives scope for an unknown factor, which is built into thepolitical-administrative system of our country, through whichlarger companies are able to reduce their ETR. It may be thatthey have greater power to lobby for obtaining favourable taxexemptions, which are otherwise insignificant in amount inrelation to total tax exemptions, have the capacity to hire moreefficient tax law experts who can reduce their tax burden, byutilising these insignificant tax exemptions or they can buy taxofficials who can reduce their tax burden by making lowerassessments. (This is possible, because government does notcome out with details of tax returns filed by individual companies.It is true that companies are coming out with their annual balancesheet but it does not give details of tax returns filed by them.Also, these balance sheets are prepared following the companyact and not the income tax act.) To know more precisely the routesthrough which big companies can reduce their tax payments, weneed to have more transparency on the part of the tax departmentof government of India, in revealing information regarding thetax returns of the companies.Generally, removal or reduction of these exemption provisions(the major ones like, exemptions on interest payment, deprecia-tion costs, profits earned from exports) may reduce the scopefor lobbying or manipulative tax management for lower effectivetax rate of individual companies. The present government istaking steps in a contradictory manner. They have reduced thetax exemption on the profit earnings from exports. Also, themaximum amount of depreciation is reduced from 25 to 15 percent of stock plant and machinery. These steps have actuallyincreased the ETR in last couple of years. But the recent devel-opments in the special economic zone (SEZ) policy will haveserious impacts on ETR in the opposite direction. This SEZ policyallows private developers to develop SEZ. SEZs are required tobe of a minimum size of 1,000 hectares of land for multi-productand 10 hectares of land for specific products. Primarily bigcompanies or big business houses are taking up these projectsas they have the required resources to do so. For instance, theReliance group has initiated construction of big SEZs near Delhiand Mumbai. It can be expected that many of their factories will

[li