Two-way relationship between innovation and market structure: evidence from Indian high and medium technology firms

ABSTRACT

This study analyses a two-way relationship between innovation and market structure in Indian high and medium technology firms. We perform an empirical analysis based on a simultaneous equation model on data for 991 high and medium technology firms during 2000–2015. Patent applications and patent grants capture innovation activity of the firms. We find that the market structure has an insignificant impact on innovation. A separate analysis of neck-and-neck firms also suggests that market structure has no impact on innovation. On the other hand, patenting has a positive and significant impact on market structure in both high and medium technology firms. The study concludes that in-house technology creation is a vital source of market dominance in Indian high and medium technology firms.

KEYWORDS:

• Market structure
• patenting
• high and medium technology firms

JEL Classification:

• L10
• L11
• L12
• L60
• O30

Acknowledgment

We would like to acknowledge the support given by Indian Institute of Technology Indore to conduct this research. We are thankful to the Editor, Prof. Cristiano Antonelli and anonymous referees for their helpful comments. We thank Prof. N.S. Siddharthan, Prof. B. N. Goldar, Prof. K. Narayanan, Prof. Lakhwinder Singh and Dr. Shomik Dasgupta for their helpful comments.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 These three amendments are: Patent (Amendment) Act 1999, Patent (Amendment) Act 2002 and Patent (Amendment) Act 2004.

2 Creative destruction pattern means those innovations which were never innovated before. These innovations are also known as Schumpeter Mark I pattern. On the other side, creative accumulation pattern means those innovations which were introduced before also. These innovations are called as Schumpeter Mark II pattern.

3 See for instance, Lunn (1986), Albert (1995), Koeller (1995, 2005), Vossen (1999) and Sridhar, Narayanan, and Srinivasan (2014).

4 Large and dominant firms may conduct technological innovations due to strategic reasons like to gain anticipated monopoly power, protection against infringement, strengthening competitive advantage, creation of entry barriers and protection from litigation (Setterstrom et al. 2013; Collinson and Narula 2014).

5 Koetter, Kolari, and Spierdijk (2012), and Clerides, Delis, and Kokas (2015) define adjusted Lerner index as: [(∏_i + tc_i − mc_i q_i)/(∏_i + tc_i)], where ∏_i is profit, tc_i is total cost, mc_i is marginal cost and q_i is the output. If marginal cost is assumed constant then adjusted Lerner can be defined as: [(∏_i / p_i q_i)].

6 Using previous studies (Tingvall and Poldahl 2006) $\mathrm{T}\mathrm{G}\mathrm{A}\mathrm{P}=\mathrm{T}\mathrm{F}{\mathrm{P}}_{\mathrm{m}\mathrm{a}\mathrm{x},jt}-\mathrm{T}\mathrm{F}{\mathrm{P}}_{ijt}/\mathrm{T}\mathrm{F}{\mathrm{P}}_{\mathrm{m}\mathrm{a}\mathrm{x},jt}$, where TFP_max,j,t is the total factor productivity (TFP) of the firm with maximum TFP in industry j in time period t, whereas TFP_ijt is the TFP of firm i in time period t.

7 In the presence of potential correlation between inputs levels and unobserved productivity shock, ordinary least square (OLS) estimation of production estimation yields biased results (Levinsohn and Petrin 2003). Levinsohn and Petrin (2003) attempt to solve this problem by utilizing intermediate input like materials, electricity, power and fuels as proxy variables.

8 Following Tukey (1977), we correct firm-level productivity outliers. We consider an observation as an outlier if it lies below Q_L − 1.5 (Q_U − Q_L) and/or above Q_U + 1.5 (Q_U − Q_L) where Q_U and Q_L are upper and lower quartile, respectively.

9 Indian Patent Advanced Search System (InPASS) is introduced in 2015 which is updated version of Indian Patent Information Retrieval System (IPAIRS).

10 We have used STATA version 14.2 for estimation. When SLM test statistics is insignificant, U-test in STATA does not reject null hypothesis (H₀ = Monotone or inverse U shape). However, it also does not show the value of test statistics. Hence, we have not reported the value of SLM test statistics when it produces insignificant relationships.

11 Mean value of TGAP for the full sample is 0.58 for the high technology firms it is 0.59 and for the medium technology firms the mean value is 0.58. In the full sample around 42.60% observations lies below the mean value of TGAP. Similarly, for the high technology sample 42.13% and for the medium technology firm 44.85% observations lies below the mean value of TGAP.

12 HHI is defined as follow: $\mathrm{H}\mathrm{H}{\mathrm{I}}_{jt}={\sum }_{i=1}^n{s}_{it}^2$. HHI_jt is the Hirschman–Herfindahl index of industry j in time period t. S_it is the market share of ith firm in t time period. Market share (S) is calculated as sales of a firm divided by total sale of the industry.

13 We use HHI as a measure of market structure for robustness check only. According Davies and Geroski (1997), and Tingvall and Poldahl (2006), industry specific measures of market structure like HHI are more stable in the case of high cross sectional variation and heavy competition which may underestimate the actual competitive pressure in the market. This stability leads to low variability in time series observation and large standard errors.

14 It would be interesting to theoretically explore one-way relationship between market structure and innovation for developing countries on lines of Aghion et al. (2005) in another work as it is beyond the scope of this text to formalize such a relationship.

15 In a similar vein, Hashmi (2013) finds mildly negative relationship due to higher technology gap among U.S. manufacturing industries.

16 The significant inverted U-shaped relationship is found for the full sample (where granted patents are utilized to measure innovation) and medium technology firms (patent applications are a proxy for innovation). These findings are also supported by SLM test.