R&D tax credit policy, product development, and welfare gains

Over the past two decades or so, India has had a targeted R&D tax credit policy to incentivise innovation within firms in strategic sectors. This article shows that small and medium firms responded to the policy by expanding product scope, while large firms upgraded product quality. This resulted in a decrease in the aggregate price index and significant welfare gains.

India ranks 38^th among 139 countries and leads the group of lower-middle-income economies in the Global Innovation Index (World Intellectual Property Organization, 2025). As a share of GDP (gross domestic product), Indian R&D (research and development) expenditure¹ rose from 0.64% to approximately 0.86% between 1996 and 2008.However, it has been on the decline ever since (Figure 1); in 2020, the percentage is almost equal to what it was in 1996.

Governments worldwide are now increasingly adopting targeted R&D tax credit policies, in the form of industrial policy, to catalyse in-house innovation within firms and strategic sectors. One such initiative was proposed by the Indian government in the Union Budget of 1997-98. The reform was initially introduced for drugs and pharmaceuticals, chemicals, electronic equipment, computers, telecommunication equipment in 1997-98 and later extended to helicopters and aircrafts in 2001 and automobiles and auto parts in 2004. This R&D tax credit policy partitioned firms based on a clearly defined eligibility criterion: in-house R&D units recognised by the Department of Scientific and Industrial Research (DSIR), the nodal unit in the Ministry of Science and Technology, Government of India. This criterion created a set of eligible firms that are directly impacted by the reform (‘treatment group’) compared to others who are not (‘control group’), enabling us to causally estimate the relative effects of the reform both within and between industries.

Figure 1. India’s R&D expenditure (% of GDP), 1996-2020

Source: World Development Indicators, World Bank.

In ongoing research (Chakraborty et al. 2025), we investigate the impact of this R&D tax weighted deduction on firm R&D investment, product scope, and product quality using a rich firm- and firm-product level data from the PROWESS maintained by the Centre for Monitoring the Indian Economy (CMIE).

Impact on R&D expenses by firms

To assess the first-order effects of the policy change, we exploit a key outcome variable – the R&D expenditure of a firm. Firms report information on R&D expenditure, as an attachment to every balance sheet presented at the Annual General Meeting in a report by the board of directors, as per section 217(1)(e) of the Companies Act,. We start by using a simple ‘difference-in-differences’² approach where our variable of interest is an interaction of two terms: (i) a ‘dummy’ representing the R&D reform. It takes a value of 1 in the post R&D tax credit policy period, that is, for the period 1998-2007 for pharmaceuticals, chemicals, electronic equipment, computers, telecommunication equipment, 2002-2007 for helicopters and aircrafts, and 2004-2007 for automobiles and auto parts, and 0 otherwise; and (ii) a dummy representing the treated industry. This takes a value of 1 for pharmaceuticals, chemicals, electronic equipment, telecommunication equipment, helicopters and aircrafts, and automobiles and auto parts. Overall, there are 21 industries at the 4-digit level of the 2004 National Industrial Classification that serve as our treated group, while the rest serve as the control group.

We find that firms belonging to industries where the policy was implemented registered an increase of 11% in their R&D expenses. One of the key features of the R&D tax credit change was that the policy applied only to firms that had an in-house R&D unit. We exploit this information to understand whether a firm that had in-house R&D was differentially affected compared to firms that did not have such facilities. The DSIR provides the directory of the recognised in-house R&D units of firms for each year. The information provided in these directories includes firm names, address, the year in which the in-house R&D facility is recognised, and the year until this recognition is valid. We extract all this information from the annual reports and match these with PROWESS firms based on the names of the firms, using a combination of fuzzy and manual hand coding.

Using this information, we introduce a triple interaction term in which we interact with a dummy that represents whether or not a firm has an in-house R&D unit or not to our previously defined double interaction term. Our estimates show that the entire effect of the increase in R&D expenses is driven by firms with an in-house R&D unit. Moreover, the effect jumps from 11% to 71%.

Further, the DSIR database on in-house R&D units has three different cut-offs to classify firms in terms of their in-house R&D expenditure: (i) firms with in-house R&D of less than Rs. 10 million; (ii) firms with in-house R&D of Rs. 10-50 million; and (iii) firms with in-house R&D of Rs. 50 million. We find that firms that had more than Rs. 50 million of in-house R&D expenses registered about 650% growth in R&D expenses, followed by firms with expenses on R&D in the range of Rs. 10-50 million (148%) and firms with expenses on R&D below Rs. 10 million (12%). Notably, over this period, R&D expenditure in India as a share of GDP increased by 34%.

The plots show that the differences in R&D expenses before the policy change were economically small and not statistically significant, indicating that there were no differential pre-policy trends in R&D expenses between firms in treated industries or between firms in any of these three categories of the control group.^3,4This, changed after the tax credit was implemented in 1998. The policy led to significant increase in the share of R&D expenses, especially by firms with in-house R&D units and firms with in-house R&D units and expenses of more than Rs. 50 million.

Figure 3 shows that all these effects are further amplified for firms producing differentiated products (Panel A) and firms that export (Panel B).

Figure 2. Effects of R&D tax credit on R&D expenditure

Notes: (i) Figure presents the Callaway & Sant’Anna (2021) coefficient estimates and their 95% confidence intervals. A 95% confidence interval means that, if you were to repeat the experiment with new samples, 95% of the time the calculated confidence interval would contain the true effect. (ii) Our coefficient estimates are controlled for firm fixed effects, year fixed effects, and industry-year trends. Standard errors clustered at the industry (4-digit) level.

Figure 3. Effects of policy on R&D expenditure: Firms with more than Rs. 50 million R&D expenditure and in-house R&D unit

Notes: (i) Figure presents the Callaway & Sant’Anna (2021) coefficient estimates and their 95% confidence intervals. A 95% confidence interval means that, if you were to repeat the experiment with new samples, 95% of the time the calculated confidence interval would contain the true effect. Our coefficient estimates are controlled for firm fixed effects, year fixed effects, and industry-year trends. Standard errors clustered at the industry (4-digit) level.

How are these investments used in terms of product development?

One of the key questions we explore in this study is that how do these new R&D investments translate into product development. A key implication of our theoretical foundation is that in response to R&D incentives, in our case a R&D tax credit, bigger firms with larger market share and belonging to the differentiated products industry will invest more in product quality than others. In addition, product scope will decrease with firm size. The intuition for this strategic reallocation is rooted in the trade-off between demand-side cannibalisation and cost-side diseconomies of scope. In multi-product settings, the addition of new varieties induces a cannibalisation effect, whereby a firm’s own products compete for market share, thus reducing the marginal gains from scope expansion. We now test those hypotheses using unique information from our dataset.

Figure 4 shows a significant increase in product quality for firms with in-house R&D units that have R&D investments above Rs. 50 million in industries where the R&D tax credit was implemented after the policy reform, with no changes before the reform. In particular, the quality of the products increased significantly in the three years following the reform, then declined for a few years, and began to rise again thereafter. Our coefficient plots echo the theoretical underpinnings. In contrast, Figure 5 shows that for firms with R&D investments of less than Rs. 50 million, product scope increased significantly after the R&D tax credit was implemented.

Figure 4. Effects of R&D tax credit on product quality: Firms with 50 million R&D and in-house R&D unit

Figure 5. Effects of R&D tax credit on product scope: Firms with R&D and in-house R&D unit

Are there welfare gains from such an R&D tax credit policy?

Given our findings, a key follow-up question arises: did the R&D tax rebate generate welfare gains through increases in product quality and scope? To address this, we use a simple framework following Sheu (2014) and quantify gains from both quality upgrading and expanded product variety using a two-period framework (pre-and post-tax credit policy). Figure 6 shows a steady decline in the aggregate price index over the policy period, indicating substantial welfare improvements. In particular, taking three years before treatment as the pre-period, and three years after treatment as post period, predict a price factor of 0.067. This indicates that the aggregate price index fell by about 93.3% from the pre-reform to the post-reform period.

Figure 6. Evolution of Aggregate Price Index Factor

Notes: (i) Figure presents Sheu (2014) estimates of the price index factor. (ii) To interpret, a price factor of 0.7 indicates that from pre-time period to post-time period , prices fell by 30%, indicating welfare improvements. (iii) The price factor in the figure is calculated using 1996 as the pre-period and subsequently changing the post-period from 1996 to 2005.

By separately holding variety (or quality) constant, we isolate the welfare contributions of each margin and find that both quality upgrading and variety expansion yield significant consumer gains. If product quality in post-period remained the same as in the pre-period, prices would only decline by 83.8% instead of 93.3%. Similarly, if the varieties remained constant, prices would only decline by 43.7%. These results underscore the effectiveness of R&D tax credits in generating sizeable macroeconomic welfare improvements through firm-level innovation.

Policy recommendations

Industrial policies have long been a subject of debate and interest – either because of their political economy or their global implications (Juha´sz and Lane 2024, Millot and Rawdanowicz 2023). However, a new set of studies have now demonstrated that such policies can play a significant role in fostering economic development (Dugo et al. (2025), Garc´ia Herrero and Schindowski 2024). Although industrial policies in developing economies are often directed towards supporting weaker sectors, targeting sectors with strong links to the broader economy may generate positive spillover effects.

Given the resource constraints faced by developing economies, our study shows that R&D tax credit policies in strategically selected sectors have the potential to deliver substantial welfare gains through product scope and product quality. In addition, firm heterogeneity can add another dimension to this type of policy, leading to different forms of gains from the same policy intervention. Developing countries, like India, should actively promote such policies in order to increase not only the innovation investments of firms or the country in question, but also for the larger objective of welfare gains. (Liu 2019)

Notes:

Gross domestic expenditures on R&D include both capital and current expenditures in the four main sectors: Business enterprise, Government, Higher education, and Private non-profit. R&D covers basic research, applied research, and experimental development.
Difference in differences is an empirical strategy that estimates the causal effect of a treatment (the R&D tax policy in this case) by comparing differences in how treatment (eligible firms) and control group (other firms) outcomes change over time. It is a valid strategy under the assumption that both groups would have trended similarly in the outcome over time in the absence of the treatment.
The absence of pre-trends also negates the possibility that other simultaneous reforms such as drop in tariffs due to the WTO (World Trade Organization) membership in 1994 had any effect on R&D expenses and R&D tax credit policy was simply an amplification effect.
Panel A consists of all firms in the ineligible industries as the control group; Panel B consists of two types of firms as the control group: (i) all firms in the ineligible industries; and (ii) firms without in-house R&D in the eligible industries; Panel C consists of three types of firms as the control group: (i) all firms in the ineligible industries; (ii) firms without in-house R&D in the eligible industries; and (iii) firms with in-house R&D in the eligible industries, but with R&D expenditure of less than Rs. 50 million.