Book–Tax Differences and Firm Value: Evidence from Firms Listed on the Nairobi Securities Exchange, Kenya

Keywords: Book–Tax Differences, Firm Value, Tax Planning, Nairobi Securities Exchange, Kenya

Paper Type: Research Article

Recommended citation: Achieng, H. F. O. (2026). Book–tax differences and firm value: Evidence from firms listed on the Nairobi Securities Exchange, Kenya. Journal of Economics, Management Sciences and Procurement, 5(1), 286–298.

1. Introduction

Firm value reflects investors’ assessment of a company’s capacity to generate future returns. It matters to shareholders because changes in market valuation affect the value of their investments and influence a company’s ability to attract capital (Bardos, et al., 2020). Tax decisions can enter this assessment through their implications for cash flows and the credibility of reported earnings. A strategy that reduces taxes may preserve resources for investment, but the market may also consider whether the strategy creates uncertainty about financial reporting or future tax obligations (Aprilliasari and Soesetio, 2024). For this reason, the relationship between corporate tax positions and firm value requires empirical examination. One relevant measure is book–tax difference, which captures the divergence between income reported for accounting purposes and income used for taxation (Darmawan & Angelina, 2021).

Book–tax differences arise because accounting income and taxable income are calculated under different rules. Accounting seeks to present financial performance, while tax computation applies the relevant rules for determining taxable profits. Differences may result from the timing of income recognition, depreciation, expenses that receive different treatment, or managerial reporting choices. Temporary differences can reverse in later periods. IAS 12 recognises differences between the carrying amounts of assets or liabilities and their tax bases when determining deferred taxes, subject to the standard’s requirements (IFRS Foundation, n.d.). Consequently, a positive book–tax difference, where accounting income exceeds taxable income, cannot by itself establish unlawful conduct or earnings manipulation. Its interpretation depends on the transactions and reporting choices that created it.

Empirical evidence on book–tax differences and firm value remains mixed. Darmawan and Angelina (2021) reported a negative relationship among Indonesian manufacturing firms. Dyussembina and Park (2024) similarly found a negative relationship in Russia, with dividend payouts weakening that association. In Malaysia, Razali et al. (2018) found a negative but statistically insignificant relationship. These differences show that the valuation of book–tax differences may depend on the institutional setting and the information available to investors. A difference that investors can explain through normal accounting and tax treatment may carry different implications from an unexplained difference. The sign and significance of an estimated relationship therefore need to be considered within the context in which a firm operates.

In Kenya, the NSE provides a setting for examining this issue because listed firms publish audited financial statements that combine accounting information with tax disclosures. Investors can use these reports alongside market information when assessing corporate value. Existing Kenyan work has considered broader tax planning and financial performance. For example, Oeta et al. (2019) examined manufacturing firms listed on the NSE and did not find a statistically significant association between the tax-planning measures considered and financial performance. That study addressed financial performance rather than the specific relationship between book–tax difference and market-based firm value. This distinction matters because accounting profitability and market valuation capture different aspects of firm outcomes. Evidence about one outcome does not automatically establish the relationship for the other.

The research problem is therefore whether differences between accounting and taxable income are associated with the value of firms listed on the NSE after accounting for other tax measures and firm characteristics. Evidence from other countries provides useful comparisons, but it cannot determine the direction of the relationship in Kenya. The present study examined this relationship over the period 2015–2024. Its objective was to establish the relationship between book–tax difference and firm value among firms listed on the Nairobi Securities Exchange, Kenya. The study contributes Kenyan panel evidence to the discussion of how tax-related accounting information is associated with corporate valuation. Its findings can inform the disclosure and oversight of tax positions while recognising that a statistical association does not reveal the motives behind every observed difference.

2. Theoretical Framework

The study was anchored on Agency Theory, developed by Jensen and Meckling (1976). The theory explains the relationship between shareholders, who provide capital, and managers, who make decisions on their behalf. Agency costs arise when their interests differ and when monitoring managerial decisions is costly. In applying this framework to taxation, managers have discretion over transactions, accounting estimates and tax-planning choices that shareholders may find difficult to evaluate. Legitimate tax planning can preserve resources for shareholders. However, opaque reporting may also make it difficult for investors to assess the underlying financial position of a firm. The relevant issue is whether tax and reporting decisions support shareholders’ interests and can be understood from the information disclosed.

Agency Theory provides a basis for expecting an association between book–tax differences and firm value. A substantial unexplained difference may increase uncertainty about reported earnings or the sustainability of a firm’s tax position. Investors may then discount the firm’s valuation. This is an interpretation of the potential information and monitoring costs associated with the difference; it is not evidence that every difference reflects managerial opportunism. Differences arising from ordinary accounting and tax rules may be consistent with shareholder interests. Applied to the present study, the theory supports examining whether higher book–tax differences are associated with lower firm value after accounting for observable firm characteristics and other tax measures. The regression tests this association, while the proposed agency explanation remains a mechanism that the financial data alone cannot establish.

3. Empirical Review (Hypothesis Development)

Dyussembina and Park (2024) examined book–tax differences and firm value using 1,421 firm-year observations from 336 Russian publicly traded firms over 2012–2020. They found a negative relationship, which weakened as dividend payouts increased. The pattern was particularly evident for the component related to discretionary accruals. Their findings connect the valuation of tax-related accounting information with reporting choices and payouts. However, evidence from Russia does not establish whether the same relationship holds in Kenya. The present study examines the Kenyan listed-firm context using a longitudinal panel.

Darmawan and Angelina (2021) examined tax planning among 135 Indonesian manufacturing companies over 2016–2018. They measured tax planning using cash effective tax rate and book–tax difference, and measured firm value using Tobin’s Q. Multiple regression analysis included firm size and leverage as control variables. Book–tax difference was negatively related to firm value, while cash effective tax rate had a positive relationship. Their results show that the direction of the relationship can vary across tax measures. The present study extends the observation period and examines book–tax difference alongside other tax measures and firm characteristics in Kenya.

Abdelrhman and Eissa (2021) examined whether qualified audit opinions affected the relationship between book–tax differences and firm value among Egyptian EGX100 firms. Their sample comprised 335 firm-year observations for 2014–2018, analysed using ordinary least squares regression. They reported a negative relationship and found that the negative association occurred in firms with qualified audit opinions. This result suggests that the information accompanying a book–tax difference may matter to investors. The present study examines the direct relationship in Kenya, without attributing its findings to an audit-opinion effect that was not tested in the focal analysis.

Razali et al. (2018) examined firms listed on Bursa Malaysia using effective tax rate and book–tax differences as tax-planning measures. The study used 387 observations over 2014–2016 and included firm size, leverage, asset tangibility, firm age and dividends as controls. Book–tax difference had a negative but statistically insignificant relationship with firm value. This finding shows that a negative coefficient does not necessarily provide sufficient statistical evidence of a relationship. The result also differs from the significant negative findings reported in Indonesia, Russia and Egypt, strengthening the case for an empirical test in the Kenyan market.

Qolbiyah et al. (2023) studied 12 property and real estate firms listed in Indonesia over 2018–2021. They considered permanent differences, temporary differences and aggregate book–tax differences in relation to profit growth. Permanent and temporary differences were positively and significantly associated with profit growth, while aggregate book–tax difference was not significant. The outcome was profit growth, so the results cannot be treated as a direct test of market-based firm value. Nevertheless, the study shows why separating the sources of book–tax differences can be useful. An aggregate measure may combine components with different economic implications.

Hepfer (2023) examined whether investors misprice book–tax differences. Using time-series asset-pricing tests and cross-sectional return tests, the study found that returns to trading on book–tax differences were explained by the operating cash flows-to-price anomaly. This finding questions whether book–tax differences provide an independent trading signal. It does not directly settle their association with Tobin’s Q, because abnormal returns and firm valuation are different outcomes. It therefore provides a reason to distinguish the outcome and model used in each study rather than treating all results concerning book–tax differences as equivalent.

The reviewed evidence provides grounds for expecting a relationship between book–tax difference and firm value, but its statistical significance and interpretation remain dependent on context. Studies differ in the markets examined, observation periods, valuation measures and control variables. Kenyan research on broader tax planning and financial performance does not resolve this specific question. The present study consequently tested the following hypothesis:

H1: There is a statistically significant relationship between book–tax difference and firm value among firms listed on the Nairobi Securities Exchange, Kenya.

4. Methodology

The study adopted a positivist research philosophy and a correlational longitudinal design. Positivism was appropriate because the analysis used measurable financial information and a predetermined hypothesis that could be tested quantitatively (Park et al., 2020). The longitudinal component involved observations of the same firms over the ten-year period from 2015 to 2024, while the correlational component assessed the relationship between book–tax difference and firm value. Historical financial data allowed the analysis to consider both differences across firms and changes within firms over time. The design establishes statistical association; it does not independently identify a causal effect of tax-related reporting decisions on firm value.

Sampling

The target population comprised 66 firms listed on the Nairobi Securities Exchange as at 31 December 2024. A census approach was used to consider firms meeting the inclusion criteria. Firms were required to have traded continuously during 2015–2024 and to have publicly accessible audited financial statements with information for the study variables. Following application of these criteria, 49 firms were retained. The dataset contained 490 firm-year records, representing 49 firms over ten years. The regression output reported 489 observations from the same 49 firms, with a minimum of nine and a maximum of ten observations per firm. These counts distinguish the descriptive dataset from the regression estimation sample.

Data Collection

Secondary panel data were extracted from audited annual reports and financial statements of the eligible firms. The sources included statements of profit or loss, statements of financial position and accompanying notes, together with company, NSE and Capital Markets Authority publications. A structured secondary data collection sheet guided the extraction of comparable information for each firm and year. The audited reports provided the financial information needed to construct the tax measures and firm characteristics, while market information supported the firm-value measure. Information was cross-checked against the official sources used in the study to improve consistency. The analysis relied on published records and did not include direct interviews about managerial tax-planning decisions.

Measurement of Variables

Firm value was the dependent variable and book–tax difference was the focal explanatory variable. Firm value was operationalised using the Tobin’s Q approximation specified in the study, calculated as market capitalisation plus total debt, divided by total assets. Book–tax difference was computed as pre-tax accounting income minus taxable income, divided by average total assets. It was a signed ratio: a positive value indicated that accounting income exceeded taxable income, while a negative value indicated the reverse. Firm size and firm age were control variables. Effective tax rate, tax retention rate and corporate marginal tax rate were additional covariates in the joint regression model. Table 1 presents their operational measures.

Data Analysis and Model Specification

Data were organised in Microsoft Excel and analysed using STATA version 15. Descriptive statistics summarised the distribution of the variables, and Pearson correlations assessed their bivariate relationships. Fixed-effects and random-effects panel estimators were compared using the Hausman specification test. Diagnostic tests examined residual normality, first-order serial correlation, heteroskedasticity, multicollinearity and panel unit roots. Statistical decisions used a 5% significance level. Fixed-effects estimation accounts for time-invariant firm characteristics that may be correlated with the explanatory variables, subject to the remaining model assumptions (Baltagi, 2021; Wooldridge, 2010). The focal relationship was evaluated using the book–tax difference coefficient in the joint tax-planning model, which also contained the three other tax measures and the two firm controls.

FV_it = β₀ + β₁BTD_it + β₅FS_it + β₆FA_it + α_i + ε_it

In the model, FV denotes firm value; BTD, book–tax difference; ETR, effective tax rate; TRR, tax retention rate; CMTR, corporate marginal tax rate; FS, firm size; and FA, firm age. The subscript i identifies a firm and t identifies the year. The term α_i represents the time-invariant firm effect, while ε_it is the idiosyncratic error. The coefficient β₁ tests the relationship of interest. Estimated coefficients are reported as unstandardised coefficients, denoted by b, so their interpretation follows the units of the operational measures.

5. Results

This section presents descriptive statistics, diagnostic tests, panel unit-root results, correlations and the fixed-effects regression. The descriptive summaries cover 490 firm-year records from 49 firms over 2015–2024. The regression uses 489 observations. Book–tax difference is the focal variable, while the other tax measures and firm controls are included to show the specification underlying the estimated relationship. The analysis first describes the variables, then assesses the reported diagnostics and finally tests the study hypothesis.

Descriptive Statistics

Table 2 presents the pooled descriptive statistics for the 490 firm-year observations included in the study. Firm value ranged from 0.010 to 52.180, with a mean of 3.630 and a standard deviation of 9.410, indicating substantial variation in firm value among the listed firms over the study period. Book–tax difference ranged from −1.468 to 4.953, with a mean of 0.175 and a standard deviation of 0.552. The positive mean suggests that, on average, accounting income exceeded taxable income when scaled by average total assets, while the negative minimum shows that taxable income exceeded accounting income in some firm-year observations. The relatively large standard deviation compared with the mean indicates considerable variation in book–tax differences across firms and over time. Book–tax difference also recorded a skewness of 4.196 and kurtosis of 30.933, indicating a highly positively skewed and leptokurtic distribution, with a small number of observations recording unusually high positive values. Firm size ranged from 0.000 to 9.210, with a mean of 7.253 and a standard deviation of 0.960, suggesting relatively limited variation in firm size across the sampled firms. Its negative skewness of −1.732 indicates that most observations were concentrated toward the higher end of the distribution, while the kurtosis of 15.057 suggests the presence of extreme observations. Overall, the descriptive results show noticeable variation in firm value and book–tax difference among firms listed on the Nairobi Securities Exchange, with the high skewness and kurtosis of book–tax difference indicating the need to consider the influence of extreme values in subsequent regression analysis; however, the descriptive statistics alone do not establish the direction or statistical significance of the relationship between book–tax difference and firm value.

Diagnostic Tests

Table 3 summarises the reported diagnostic results. The joint skewness–kurtosis test (χ² = 4.27, p = .1184) and the Jarque–Bera test (χ² = 4.137, p = .1264) did not reject residual normality. The normality output covered 374 residual observations. The Mardia, Henze–Zirkler and Doornik–Hansen results also did not reject their respective null hypotheses at the 5% level. These findings concern the tested residuals and should be distinguished from the pronounced skewness of the tax variables themselves. The Wooldridge test did not reject the null of no first-order serial correlation, F(1, 48) = 0.774, p = .471. The Breusch–Pagan/Cook–Weisberg test did not reject constant error variance, χ²(1) = 0.07, p = .7847. These tests provide no statistically significant evidence of the specific error problems tested; they do not establish that every possible model problem is absent (Drukker, 2003; Wooldridge, 2010).

The variance inflation factor for book–tax difference was 1.280, with tolerance of 0.779. The VIFs for the other covariates in the direct model ranged from 1.060 to 2.330. These low values indicate that linear dependence among the explanatory variables was unlikely to be a serious concern in the reported diagnostic specification. Taken together, the diagnostic results supported proceeding with the panel analysis, while the distribution of the tax ratios still calls for sensitivity checks when interpreting the magnitude of the coefficients.

Unit Root Test

Table 4 presents the results of the Fisher-type panel unit-root test used to determine whether the study variables were stationary before estimating the panel regression models. The null hypothesis of the test states that all panels contain a unit root. For firm value (FV), all four test statistics were statistically significant: inverse chi-squared (P = 519.705, p < .001), inverse normal (Z = −10.221, p < .001), inverse logit (L* = −18.229, p < .001), and modified inverse chi-squared (Pm = 30.122, p < .001), providing strong evidence against the presence of a unit root. Similarly, book–tax difference (BTD) was significant across all four statistics, with P = 301.762, Z = −5.449, L* = −9.148 and Pm = 14.554, all at p < .001, indicating that BTD was stationary in the panel. Firm age (FA) also recorded significant results across all four tests (P = 768.033, Z = −4.090, L* = −21.311 and Pm = 47.860, all p < .001). For firm size (FS), the inverse chi-squared (P = 204.202, p < .001), inverse logit (L* = −2.248, p = .013), and modified inverse chi-squared (Pm = 7.586, p < .001) rejected the unit-root null, although the inverse normal statistic was not significant (Z = −0.710, p = .239). Since three of the four Fisher-type statistics for firm size rejected the null hypothesis, the overall evidence suggests that firm size was also sufficiently stationary for analysis, although the result should be interpreted with some caution. Overall, the findings provide adequate evidence that the variables were stationary in levels and could therefore be included in the subsequent panel regression analysis without first differencing them; however, rejection of the Fisher-type null indicates stationarity for at least some panels rather than necessarily establishing stationarity for every individual firm.

Correlation Analysis

Table 5 presents the Pearson correlation coefficients among firm value, book–tax difference, firm size, and firm age. Book–tax difference was negatively and significantly correlated with firm value (r = −0.330, p < .01), indicating that higher book–tax differences were associated with lower firm value at the bivariate level. Firm size was positively and significantly correlated with firm value (r = 0.481, p < .01), suggesting that larger firms tended to record higher firm values. In contrast, firm age had a weak negative and statistically insignificant relationship with firm value (r = −0.061, p > .05), indicating that differences in firm age were not meaningfully associated with firm value. Book–tax difference was also negatively and significantly associated with firm size (r = −0.141, p < .01), while its relationship with firm age was very weak and insignificant (r = −0.010, p > .05). In addition, firm size was negatively and significantly correlated with firm age (r = −0.277, p < .01). Overall, the correlation coefficients were below the commonly used threshold of 0.70, suggesting that serious multicollinearity was unlikely among the explanatory variables. However, these results represent bivariate associations only and do not control for other firm-specific factors; therefore, the direction and statistical significance of the relationship between book–tax difference and firm value were further examined using panel regression analysis..

Hypotheses Testing

The Hausman specification test was used to determine the appropriate panel estimator, and the significant test result supported the use of the fixed-effects model. As presented in Table 6, the fixed-effects regression was based on 490 firm-year observations from 49 firms and was statistically significant, F(3, 437) = 22.26, p < .001, indicating that book–tax difference, firm size, and firm age jointly explained significant variation in firm value. The within R² of 0.1325 indicates that the explanatory variables accounted for approximately 13.25% of the variation in firm value within firms over time, while the between R² and overall R² were 0.0445 and 0.0648, respectively. Book–tax difference had a negative and statistically significant effect on firm value (β = −0.337, p < .001, 95% CI [−0.422, −0.252]), indicating that, holding firm size and firm age constant and controlling for unobserved time-invariant firm characteristics, a one-unit increase in book–tax difference was associated with a 0.337-unit decrease in firm value. The confidence interval was entirely below zero, further confirming the negative relationship. Firm size had a negative but statistically insignificant effect on firm value (β = −0.047, p = .406), while firm age also had a negative but statistically insignificant effect at the 5% level (β = −0.165, p = .068). The significant F-test for individual effects, F(48, 437) = 19.92, p < .001, further indicates that firm-specific effects were important and supports the use of a panel estimator that controls for such heterogeneity. Therefore, the null hypothesis that book–tax difference has no statistically significant relationship with firm value among firms listed on the Nairobi Securities Exchange was rejected, supporting the alternative hypothesis that book–tax difference is significantly related to firm value. The negative relationship is consistent with the findings of Dyussembina and Park (2024), Darmawan and Angelina (2021), and Abdelrhman and Eissa (2021), who also reported negative relationships between book–tax differences and firm value, but differs from Razali et al. (2018), who found a negative but statistically insignificant relationship. The result suggests that larger book–tax differences may be associated with lower market valuation, possibly because such differences may increase uncertainty regarding the quality and sustainability of reported earnings and tax positions. However, the finding should be interpreted as an association rather than proof that book–tax differences directly cause lower firm value, since the fixed-effects model controls for time-invariant firm characteristics but does not eliminate all possible time-varying omitted factors.

6. Conclusions

The study concludes that book–tax difference had a negative and statistically significant relationship with firm value among firms listed on the Nairobi Securities Exchange during the 2015–2024 period. The findings indicate that increases in the signed difference between accounting income and taxable income were associated with decreases in firm value, even after controlling for firm size, firm age, and unobserved time-invariant firm characteristics. This suggests that larger book–tax differences may be viewed by the market as a signal of greater uncertainty in firms’ financial and tax reporting positions. The finding is consistent with the agency perspective, which suggests that information asymmetry and uncertainty surrounding managerial reporting choices can influence firm valuation. However, book–tax differences may also arise from legitimate differences between accounting standards and tax rules. Therefore, the study concludes that greater transparency, clear disclosure, and effective oversight of material book–tax differences are important in helping investors understand the nature and implications of such differences for firm value.

7. Recommendations

The recommendations concern disclosure, accounting practice and managerial oversight. They follow from the negative association between signed book–tax difference and firm value, while recognising that legitimate differences can arise from the distinct rules used for financial reporting and tax computation. The aim should be to improve the explanation and monitoring of material differences and to reduce avoidable reporting errors or unsupported adjustments. Reducing every difference to zero would not be an appropriate objective where accounting and tax rules require different treatment.

For policy and regulation, the Capital Markets Authority, NSE and Institute of Certified Public Accountants of Kenya should encourage clear and consistent explanations of material tax-related differences in listed firms’ financial reports. Guidance can help companies distinguish timing differences from items that have a permanent effect on the relationship between accounting and taxable income. Where a tax position involves substantial judgement, the accompanying explanation should make its nature and financial implications understandable. Coordination with the Kenya Revenue Authority can support reporting consistency and appropriate tax administration. The empirical finding does not justify treating the size or sign of a book–tax difference as sufficient evidence of non-compliance.

For accounting practice, finance professionals should maintain regular reconciliations between pre-tax accounting income and taxable income. Each material adjustment should be supported by the relevant transaction records and accounting or tax treatment. Professionals should review the consistency of depreciation, income recognition and expense classification across the financial statements and tax calculations. Changes in deferred tax balances should also be explained clearly where they are relevant to understanding the firm’s reported position. These procedures can reduce unexplained discrepancies and enable auditors, directors and investors to assess the sources of the differences more accurately. Staff development should address the interpretation of the differences as well as their calculation.

For management, tax-planning decisions should be reviewed alongside their reporting implications and the firm’s long-term objectives. Boards and audit committees should require explanations of material changes in book–tax differences and assess whether the supporting assumptions are reasonable. Management should communicate significant tax positions through clear financial reporting and respond to questions about their effects on reported income and future obligations. Legitimate tax savings should be assessed together with the costs of uncertainty and weak disclosure. Investors can use book–tax differences as one input in their assessment, but should also consider cash flows, audit information and the explanations accompanying the accounts.

8. Limitations of the Study

The findings should be interpreted within several limits. First, the analysis used an aggregate signed book–tax difference measure and did not distinguish permanent, temporary and discretionary components. The coefficient therefore cannot identify which component accounts for the relationship. Second, the sample included firms with continuous listing and accessible financial information over 2015–2024; the findings may not generalise to firms excluded by these criteria, unlisted businesses or other countries. Third, published financial records do not directly reveal managerial motives or investors’ interpretation of tax disclosures. Fourth, extreme observations in the tax ratios may affect the magnitude of the estimated coefficients. Fifth, fixed-effects estimation controls for time-invariant firm characteristics but does not resolve all time-varying omitted factors, reverse causality or measurement error. Finally, the correlational longitudinal design establishes association rather than a definitive causal effect. The estimated negative relationship should therefore inform further investigation and reporting practice without being treated as proof of the cause of lower firm value.

9. Suggestions for Future Studies

Future research should separate permanent and temporary book–tax differences and identify the components associated with discretionary accruals. This would help determine whether market valuation depends on the source of the difference. Researchers could compare signed differences with their absolute magnitude and examine whether positive and negative differences have different associations with firm value. Sensitivity checks using alternative valuation measures and careful treatment of extreme ratios would improve assessment of the result. Further studies could consider profitability, leverage, ownership, audit opinions and dividend policy, alongside year effects and standard errors appropriate for repeated firm observations. Comparisons between financial and non-financial sectors, listed and unlisted firms, and East African markets could test the scope of the Kenyan finding. Interviews with investors, auditors and finance managers could explain how tax-related information enters valuation decisions. Designs based on clearly identified tax-policy changes could provide stronger evidence on causality than the present correlational approach.

References