Impact of research and development tax credits on the innovation and operational efficiencies of Internet of things companies in Taiwan

Abstract

Following the emergence of the Internet, the Internet of things (IOT) brought about another wave of technological and economic revolutions. Through the lens of the production process, this study utilises the dynamic network slack-based measure model in data envelopment analysis to evaluate 32 IOT companies in Taiwan in terms of their innovation efficiency, operational efficiency and overall efficiency for the period of 2007–2017. Empirical results reveal that the average operational and overall efficiencies of IOT companies in Taiwan have been decreasing considerably since 2008. However, their average innovation efficiency remains stable over the sample period owing to government reductions in enterprise research and development (R&D) tax credit incentives. Through the impulse response function method, this study further confirms that the Statute for Industrial Innovation, which was implemented in 2010 and revised and reimplemented in 2016, specifically, policies concerning enterprise R&D tax credits, affect the efficiencies of IOT companies in Taiwan. Overall, this study reveals the performance evaluation process of IOT companies by showing that their innovation capability affects their operational efficiency. Thus, the government is advised to incorporate innovation measures into relevant industrial policies to achieve policy effectiveness.

Appendices

Appendix 1: Statute for Industrial Innovation in Taiwan

Innovation is the main source of profits for a company (Grant 1991). Considering that innovation capabilities are the key to successful innovations, corporations must continue innovating to maintain a competitive advantage (Wang et al. 2008). Therefore, numerous governments emphasise the investment and efficiency of enterprise R&D as crucial means for economic development and propose tax credit policies as an incentive for R&D expenditure. By implementing the Statute for Upgrading Industry in 1991, the Taiwanese government stipulated R&D tax credits for companies and lowered eligibility requirements to increase the utilisation rate of investment tax credits, thereby encouraging enterprise R&D investment. To help industries succeed against international competition and apply the concept of light taxes and simple administration as a direction for tax reform, the Statute for Industrial Innovation was implemented in 2010 as an extension of the Statute for Upgrading Industry. The new regulation aims to promote industrial innovation, improve the industrial environment and enhance industrial competitiveness.

The core implementation measures of the Statute for Industrial Innovation are as follows: provide counselling to strengthen industry competition, promote sustainable industry development, use policies as a flexible tool and establish various development funds, levy industrial park lands and leverage national resources to assist industries with product innovations and technological upgrades. One of the reforms states that employees who have worked 2 years or more will receive stock-based employee compensation. When a stock is transferred, a company can be taxed on either the acquisition price or the actual transfer price of the stock depending on which price is lower. Companies can pay dividends through stocks, thereby retaining talented employees. In addition, industry-based tax credits are cancelled and used for counselling and formulating other policies to replace R&D tax credits for companies.

An analysis of the differences between the Statute for Upgrading Industry and Statute for Industrial Innovation reveals three major changes in terms of company R&D tax credits. Firstly, the R&D tax credit rate is reduced. The Statute for Upgrading Industry specifies the rate of annual R&D expenditure as 30%, whereas the Statute for Industrial Innovation reduces this rate to 15%. Secondly, incremental tax credits are cancelled. The Statute for Upgrading Industry stipulates that if a company’s annual R&D expenditure exceeds the average amount of that for the two previous years, then the excess will be offset by 50%. That is, the Statute for Upgrading Industry increases the incremental tax credit rate by up to 50%, whereas the Statute for Industrial Innovation does not have such an incentive.

Finally, R&D tax credits have considerable limitations in the Statute for Industrial Innovation. The Statute for Upgrading Industry stipulates that a corporation may be exempt from up to 50% of the taxable income of profit-seeking enterprises, whereas the Statute for Industrial Innovation reduces this credit rate to 30%. The Statute for Upgrading Industry regulates unused credits, which may be carried forward to deduct the taxable income of profit-seeking enterprises in each of the following 5 years, and the amount deducted in the final year is not subject to the 50% limit of the current year. The Statute for Industrial Innovation stipulates that the use of enterprise R&D tax credits is limited to the current year, and unused credits may not be carried forward to use in subsequent years. The government loosened the limitations of this article on December 30, 2015, thereby allowing enterprises to select either the original deduction rate of 15% or the deduction rate of 10% with an extended time limit of 3 years. This amendment came into effect in 2016.

To maintain fairness in taxation as well as a balance between industrial development and government revenue and expenditure, tax credit incentives vary considerably between the Statute for Industrial Innovation and Statute for Upgrading Industry. These changes reduce R&D tax credits for companies substantially, thereby making it the first time that such a large-scale reduction in tax incentives occurred in Taiwan. Its effect on the innovation efficiency of domestic enterprises requires the government’s attention. By using the DNSBM model and IRF targeting the IOT industry, this study explores the effects of R&D tax credits from a new perspective and analyses whether statute implementation has other effects on the efficiencies of companies in the IOT industry. The results may serve as a reference for policy formulation.

Appendix 2: Non-positive output data in the slack-based measure

In the case of the slack-based measure, dealing with negative outputs in the evaluation of efficiency is crucial, as negative data should play an important role in measuring efficiency. If neglected, a large deficit (loss) is worse than a small one, especially for profit (loss) in this case. Therefore, we employ a new scheme (Düzakın and Düzakın 2007; Tone 2017) to solve this problem.

Suppose $y_{ro} \le 0$. We define $\overline{y}_{r}^{ + }$ and $\underline{y}_{r}^{ + }$ as

$$ \begin{aligned} \overline{y}_{r}^{ + } = Max_{j = 1, \ldots ,n} \left\{ {y_{rj} \left| {y_{rj} > 0} \right.} \right\} \hfill \\ \underline{y}_{r}^{ + } = Min_{j = 1, \ldots ,n} \left\{ {y_{rj} \left| {y_{rj} > 0} \right.} \right\} \hfill \\ \end{aligned} . $$

(A.1)

We replace the term ${{s_{r}^{ + } } \mathord{\left/ {\vphantom {{s_{r}^{ + } } {y_{ro} }}} \right. \kern-0pt} {y_{ro} }}$ in the objective function with the following (notice that we never change the value $y_{ro}$ in the constraints): if $\overline{y}_{r}^{ + } > \, \underline{y}_{r}^{ + }$, then the term is replaced with

$$ s_{r}^{ + } /\frac{{\underline{y}_{i}^{ + } \left( {\overline{y}_{r}^{ + } - \underline{y}_{r}^{ + } } \right)}}{{\overline{y}_{r}^{ + } - y_{ro} }}. $$

(A.2)

If $\overline{y}_{r}^{ + } = \, \underline{y}_{r}^{ + }$, then the term is replaced with

$$ {{s_{r}^{ + } } \mathord{\left/ {\vphantom {{s_{r}^{ + } } {\frac{{(\underline{y}_{r}^{ + } )^{2} }}{{B\left( {\overline{y}_{r}^{ + } - y_{ro} } \right)}}}}} \right. \kern-0pt} {\frac{{(\underline{y}_{r}^{ + } )^{2} }}{{B\left( {\overline{y}_{r}^{ + } - y_{ro} } \right)}}}}, $$

(A.3)

where $B$ is a large positive number. In any case, the denominator is positive and strictly less than $\underline{y}_{r}^{ + }$. Furthermore, it is inversely proportional to the distance $\overline{y}_{r}^{ + } - y_{ro}$. This scheme positively considers the magnitude of the non-positive output. The score obtained is also unit invariant, that is, it is independent of the units of measurement used (Bowlin 1998).