Tax Incentives Policy, Firm Investment, Firm exports, and Gross Output: Panel Econometric Modelling

A panel econometric model consisting of 118,380 firms, spanning 2014 to 2019 was used to determine the impact of tax incentive policy on firm investment, firm gross output, and exports. A two-stage modelling approach was used, first the decision to invest or export was modelled using a binary logit model. In the second phase, the impact of the tax incentives policy was estimated. The decisions to export and invest are marginally driven by tax incentive policy. A shilling given as tax expenditure increases the probability of investing and exporting by 0.018% and 0.48% respectively. The results from the study imply that export and investment-related tax incentives are either redundant or have a negligible impact on their respective target variables. – were applied to data generated from returns for 2019.


Figure 1. Tax Expenditure in Ksh Million and % of GDP
Source: KRA, 2020 A detailed breakdown shows that VAT and income tax exemptions account for most of the revenue lost. Tax expenditure under VAT is the largest. This represents revenue foregone due to exemptions and zero-rating of certain goods and services as well as exemptions from payment by certain bodies or persons (Table 2). It is estimated at Ksh. 370.4 billion (or 4.2% of GDP) in 2018. Tax expenditure under personal income tax (PIT) is minimal but largely benefits higher-income households. This contains tax forgone due to personal relief, insurance relief, relief related to persons with disability (PWD), and mortgage relief among others (Table 3). Tax expenditure under custom taxes is in the form of exemptions and zero-rated items under import duty, excise, and VAT on ordinary imports and petroleum imports. It also includes Road development levy (RDL) and Import Declaration Fee (IDF) exemptions.

Research Problem
Kenya has tax expenditure programs predominantly aimed at inducing investment and promoting exports.
The revenue foregone annually due to these incentives has grown five times from Ksh 100 billion in 2012 to Ksh. 536 billion in 2018. The benefits derived by the country are however not comprehensively quantified. Often, the fiscal cost of tax incentive policy may outweigh the envisaged benefits; ultimately undermining the much-needed revenue for public spending on infrastructure, public services, or/and social safety nets. As a rule of thumb, only those tax incentive programs that can have net benefits, in terms of, the economy and revenue should be granted. This study sought to determine the impact and net revenue effect of investment and export tax incentives in Kenya focusing on corporate income tax. The general objective of the study was to determine the impact and efficiency of investment-oriented tax incentives in Kenya. The specific objectives of the study were; 1.
To determine the impact of tax incentives on investment 2.
To determine the impact of tax incentives on export values 3.
To determine the impact of tax incentives on economic growth 4. Determine the net effect of tax incentives policy

Literature Review
Several theoretical frameworks link tax incentives to investment. These include the capital arbitrage theory, the neoclassical investment theory, and the neoclassical ownership, location, and Internalization (OLI) theory as summarised in (Munongo, Akanbi, & Robinson, 2017). The capital arbitrage theory argues that capital movement responds to the differentials in rates of return predominantly linking foreign firm investments in domestic markets to tax incentives. The theory established that capital will move from capital-rich countries to capital-scarce countries in search of higher returns and the process will continue until the returns on capital are equalized between jurisdictions. This theory explains the location of Multinational Corporations (MNCs) in developing countries where capital is scarce.
The neoclassical investment theory postulates that firms accumulate capital as long as the costs of doing so are less than the benefits. Since firms' investments are subject to decreasing returns, the optimal investment is at the point where the present value of returns from capital equals the present value of costs.
lower tax rates reduce the cost of capital and increase the investment in more capital stock (Van Parys & James, 2010). The neoclassical investment theory thus suggests that tax incentives encourage the growth of established firms through reinvestments and also lures new investments since it reduces the cost of capital (Munongo, Akanbi, & Robinson, 2017).
In analyzing investment behavior at the firm level, the accelerator theory and the Euler model have yielded empirically testable investment equations (Bigsten, Collier, Dercon, Gauthier, Gunning, Isaksson, & Sylvain, 1999). The basic assumption is that firms seek to maximize profit. The accelerator equation is based on Clark's (1923) accelerator theory which asserts that investment levels can fluctuate with consumer demand. Accelerator models emphasize the role of expectations and convention where there is a link between the expectation of profits in the next period, given output growth in the current and earlier periods. The accelerator model assumes a fixed capital to output ratio, which implies that prices, wages, tax rates, and interest rates have an indirect impact on investments in capital stock. Koyck (1954) introduced the flexible accelerator model to allow for capital stock adjustment in several periods other than instantaneously. On the other hand, the Euler equation model seeks to address uncertainty by explicitly including dynamic elements and expectations in the optimization problem.
Firm ability to export is predominantly captured in the old international trade theory of comparative advantage (see Heckscher-Ohlin comparative advantage), Bernard, Jensen, Redding, and Schott (2007).
However, recent trade patterns lean more towards firm heterogeneity models. Firm heterogeneity models point out that there are significant differences between international trading and non-trading firms. In deciding to export or the magnitude of export propensity (exports to total sales ratio), firm heterogeneity matters (Roberts & Tybout, 1997;Bernard & Jensen, 1997;Niringiye & Tuyiragize, 2010;Kahia, 2017).
Several facts emerge: First, tax incentives work for certain kinds of investments, in specific situations, and specific sectors, such as export-oriented investments. They may also be used to effectively target public goods in sectors that have high returns. In countries where the level of public goods is very low, the marginal benefit from an additional amount of public good is more than the marginal cost justifying the use of incentives. Secondly, tax incentives are very useful when targeting investment programs with positive externalities. They include investments in Research and Development or High-tech industries that upgrade worker skills, infrastructure projects that encourage business growth, among others.
Nonetheless, due to scarcity in the capital, some countries and economic blocs are caught up in incentives competition against each other to offer more generous incentives. There is evidence that tax competition is occurring between developing countries and is successful in attracting mobile investments. Tax competition leads to a race to the bottom phenomena-a scenario where effective tax rates fall drastically as countries incentivize foreign direct investments (Abbas & Klemm, 2013). Consequently, tax revenue declines as firms employ advanced tax planning, taking advantage of the complicated interactions of international tax systems. The effect is more notable in Africa, to the extent of creating effectively a parallel tax system where rates have fallen to almost zero.
Political economy exerts a powerful influence on incentives too. Governments' behavior is not always driven by economic rationality, and, political rather than economic considerations often tip the balance in favor of incentives (James, 2009). Several factors have been highlighted as political drivers of tax incentives. First, due to political interests, and the need for the government to reward voters, elites can influence and direct policymaking or even control the tax administration and therefore increase tax incentives. This is known as elite influence. Second, the business sector is habitually organized into formal organizations that lobby for the interests of their members. Such groups sustain a certain level of tax incentives by effectively exerting pressure on the government. Thirdly, lack of transparency in the tax system and discretion of social planners in issuing tax incentives facilitate exploitations by organized groups and make lobbying easier. Fourth, for political survival reasons, the government may find it attractive to offer tax incentives (Santos de Souza, 2013).
Empirical evidence at both macro-level (aggregated variables) and micro-level (firm-specific level) point at select cases of successful use of tax incentives to attract foreign direct investment and to crowd in private sector participation in economic and social programs (see Kosonen & Harju, 2018  gross private fixed capital formation or growth. Also, Nallareddy, Rouen, and Serrato (2018) show that corporate tax cuts increase both income inequality as well as a real investment. The implication is that corporate tax cuts increase investment but the gains from this investment are concentrated on top earners.
Employing panel data consisting of 51 countries including Kenya, Stausholm (2017) (2015), UN (2015) and World Bank (2015) studies find that tax incentives are lowly ranked in investment climate surveys, mostly redundant and most investments would have taken place without them (see James, 2009;and James, 2013). Andersen et al. (2017) reveal that the impact of incentives on FDI depends on the nature of the investment in the first place. Tax incentives are more effective in attracting efficiency-seeking FDI focusing on lowering production cost but not for those investments attracted by domestic markets and natural resources.
Turning to the cost-benefit analysis of tax incentives, various methods have been applied in the literature. Andersen et al. (2017) suggest the use of a survey to know the motivation behind a certain investment.
Based on the survey data, the proportion of investment that would have occurred without the incentive is interpreted as an incentive redundancy rate across firms and sectors. Together with tax expenditure, investment, and employment rates, cost-benefit ratios are generated for comparison. This method requires samples large enough to disaggregate the resultant redundancy rates by sector, which is costly.
The second and more reliable method is the user cost of capital (UCC) method. UCC is the pre-tax minimum rate of return required for an investment to be considered profitable. Comparing UCC with and without tax incentives permits an estimation of the change in fixed assets that is due to existing tax incentives. This methodology has produced rigorous measures of the net fiscal costs per job created, or unit of investment, for different sectors and incentive instruments in the Dominican Republic, Malaysia, and South Africa. But its heavy data needs make this approach difficult to replicate in many lower-middle-income countries.
A social accounting method (SAM) has also been used to analyze the cost-benefits of tax incentive policy (see Calitz, Wallace, & Burrows, 2013). Similarly, the United Nations (2018) guidelines on assessment and design of tax incentives show input-output models and computable general equilibrium (CGE) models as very effective in assessing tax incentive cost-effectiveness. However, such models are hardly available in developing countries due to resource constraints. Alternately, United nations (2018) has developed a prototype model for assessing the cost and benefit of any given tax incentive program. The data requirement here is a combination of firm-based financial and tax data, which are assumed accessible by the revenue authorities.
A study by World Bank Group, James (2009)

Empirical Methodology
A regression technique was used to determine the impact of tax incentive instruments on outcome variables. The estimation equation (1) is specified as follows: where K t is capital stock, i t is investment (motor vehicle, purchase of plant and equipment), b is debt, c is profit, dt is a time dummy, μi is an unobserved firm-specific effect, and ν it is an error term. i t is a vector of tax incentive instruments including if a firm is benefiting from the tax holiday, location (EPZ and SEZ), concessional tax rates, investment deduction, or capital allowance. The vector also includes firm-specific attributes like business subtype, sector, and firm size. Explicitly, (I/k)t-1 is lagged investment in plant and equipment to the capital stock, c/k profit rate, b/k is indebtedness (defined as past formal borrowing ) to capital ratio while s/k is sale to capital ratio. λ 1 to λ 6 parameters to be estimated. Ideally, λ 1 should be positive and greater than 1, λ 2 is negative and greater than one in absolute, λ 3 is negative while λ 5 is positive under imperfect competition and is zero under perfect competition. Parameter λ 4 is controlled for non-separability between borrowing and investment decisions and is zero if financing and investment decisions are independent (Kirui, 2018).
To determine the impact of tax incentives on exports, the study employed a decision to export model borrowed from Bernard and Jensen (1997 where y is a binary variable on whether a firm exports or not, X it is a vector of tax incentives while Z it is a vector of firm-specific attributes including size, wage expenditure, and capital intensity. The tax incentives under consideration are; investment deduction, industrial building deduction, wear and tear allowance, and location in SEZ or EPZ. Lastly, to determine the impact of incentives on output, the study linearized a simple Cobb-Douglas production function where the proxy for output is value addition (total sales minus purchases).

Data and Data Sources
The raw data set consisted of 264,810 firms however only 118,380 firms' data were reliable after data cleaning. Data was obtained from KRA. Corporate income tax returns from 2015-2018, which cover 151 EPZs and 8 SEZs, were used to retrieve information on the nature of investments and firm turnover.

Empirical Results
This section covers descriptive statistics and the results of the analysis.

Descriptive Results
The descriptive statistics (mean and standard deviation) are in Table 6. The average investment rate for the period of the analysis is 0.1473 while the sales to capital ratio is high on average at 11.5408.

The Decision to Invest and Export
In this section, we used a logit model to determine if investment incentives determine the decision of a firm to invest and export. The dependent variable is binary with 1 for firms that have an investment rate greater than zero and zero otherwise for any given return period. Capital expenditure (a sum of the respective investment-related deductions) is used as a control variable. The results show that a shilling deduction towards capital expenditure positively and significantly increases odd ratios in favour of investment. Controlling for both time and sector-specific fixed effects reveals consistent results regarding the odd ratios. A marginal analysis at means reveals that a shilling given as tax expenditure increases the probability of investing by 0.018%. Detailed results are in Table 7 below. The total expenditure per employee is used proxy for quality of labour force, assuming that the higher the labour cost per head, the higher the wage that corresponds largely, educated, and skilled workforce. The age of the firm is computed as the difference between the registration date and the date of analysis. A log of short-term liabilities is used as a proxy for financial availability through borrowing. According to theory and empirical findings, the factors are expected to influence exports of a firm positively (Papadogonas, Voulgaris, & Agiomirgianakis, 2007).
The analysis reveals that an exporter is likely to be large, old, accessing credit, labor-intensive, seeking cheap labour, and enjoying some capital expenditure deductions.

Impact of Tax Incentive on the Intensity of Investment and Firm's Value-added
The coefficient of sales to capital ratio and debt to capital squire are positive and significant after controlling for time, business subtype, and sector fixed effects. The positive and significant debt to capital squire implies that financing and investment decisions are dependent. In other words, firms tend to borrow for investment. The coefficient of lagged profit to the capital ratio used as a proxy for cash flow is negative and significant as theoretically expected in a non-financially constrained market. It further implies that a firm can raise as many finances as it desires at a given cost, (Hall, 1991). Detailed results are presented in Table 9. The study sought to determine the effect of tax incentives on the firm's value-added. Since firm-level data was used, value addition suffices as a proxy for economic growth. Value addition is computed as the difference between firm output at market price and intermediate consumption. In estimation, the study mimics a simple but log linearized Cobb-Douglas production function where output is a function of capital and labor. The total number of employees and total fixed assets are used as labor and capital proxies respectively. The results are shown in Table 10. Labour and capital elasticity coefficients are 0.56% and 0.19% respectively. The analysis shows that a 1 percent increase in capital expenditure is associated with an increase in the firm's value-added by between 0.15% and 0.36%.  Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.

Cost-benefit Analysis
In this section, the study aggregates the benefits and costs associated with tax incentives and matches them to determine the net benefits. The profound rationale for offering tax incentives is to rejuvenate the economy, increase investment, and export values. The study computed revenue cost as the value of revenue forgone due to capital expenditure-related deductions. The induced gross value addition, investment, and exports were used as benefits. The respective multipliers were obtained from the econometric model estimates from the study.
Based on the analysis in Table 11, the net revenue effect on tax incentives for the period (2014 to 2019) of the analysis is a loss of Kshs. 206,658 million. The revenue foregone through tax incentives out weights the sum of induced gross value added, induced exports, and investments across all the tax return periods. Annually, the average foregone CIT revenue due to capital expenditure deductions is Kshs.
60,367 million. Similarly, the cost outweighs the benefits of kshs. 34,443 million on average.

Summary of Findings and Policy Recommendations
This study sought to determine the impact of tax incentives on investment, export values, and economic growth. It also sought to determine the net effect of investment and export tax incentives on the economy and give policy recommendations and guidelines on tax incentives in Kenya. To achieve the objectives of this study, we employed a two-stage approach. In the first stage, we model the effect of tax incentives on the firm's decisions to invest and export using a logistic model. In stage two, we modelled the effects of tax incentives on the rate of investment and export values for those firms who invested. Investment Euler equation was used to analyze the effect of tax incentives on investment levels while log-linearized Cobb-Douglas production function was used to analyse the effects of tax incentives on the firm's outputproxied by gross value added. These methods were applied to 118,380 firm-level panel data generated from tax returns for the period 2014 to 2019.
Regarding the impact of tax incentives affects the decision to invest and to export, the results show that the probability of capital deduction expenditure influencing the decision to invest is 0.018%. This implies that a hundred shilling increase in investment deduction increases the propensity of investing by 0.018. The estimates of Euler coefficients show that financing and investment decisions are dependent. In other words, firms tend to borrow for investment. Based on the sample studied, there is no evidence for financial constraints in the market implying that a firm can raise as much finances as it desires at a given cost (Hall, 1991). Concerning exports, doubling capital expenditure deduction (100% increase) only increases the propensity to export by 0.5. The impact of tax incentives on investment rate and firms' gross value added is small. For each Ksh 10 of total capital deduction expenditure, about Ksh 5 is realized as an investment. Based on the value-added production function, a 10% increase in capital expenditure deduction increases growth in the firm's gross value added by between 0.15% and 0.36%, depending on the model assumptions. We used the sum of investment, export, and gross value added values induced by tax incentives policy as benefits and compared this to the average foregone tax revenue (cost) due to capital deductions. The cost on average outweighs the benefits of Ksh.34.4 billion every year.

Policy Recommendations
These results imply that export and investment-related tax incentives are either redundant or have a trivial impact on their respective targets. The study, therefore, recommends a review of the second and the third schedules of the Income Tax Act to: 1. Discontinuing further issuance of redundant tax incentives to newly registered firms.

2.
Gradually phasing-out out the existing redundant tax incentives.

3.
Develop national guidelines on provisions of tax incentives to guide evaluation and enactments so as to ensure that only beneficial incentives are provided for in law