Firm-level innovation and efficiency in the food sector: Insights from a literature-based innovation output indicator

ABSTRACT

Innovation activity is commonly regarded as one prerequisite for the competitiveness of food manufacturers. However, new product introductions typically face a high failure rate, which frequently deprives food manufacturers of returns on new product developments. Accordingly, existing empirical evidence on the effect of innovation activity on firm performance is not unanimous in the food sector. In this study, we explore contemporaneous and lagged effects of product and process innovation on production efficiency of dairy processors in the German market. Endogeneity of innovation variables and production inputs is accounted for by instrumentation using a System GMM approach. While we find no positive effect of patents or the number of new product launches on efficiency, we find a consistent positive effect of successful new products on the dairy processors’ technical efficiency. On the methodological side, our results confirm the relevance of the literature-based innovation output indicator as a differentiated measure of product innovation suitable for measuring the innovation effect on firm performance. Regarding innovation management in the food sector, the results highlight the challenge related to high failure rates of new products in the food sector and emphasize the importance of new product quality for business performance.

KEYWORDS:

• Productivity
• dairy processors
• literature-based innovation output indicator
• innovation

JEL CLASSIFICATION:

• C23
• D24
• L25
• L66
• O31

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Although innovation activity also has other possible effects on the production process of the firm as, for example, the input composition, we focus purely on its effect on technical efficiency.

² Although a more flexible functional form is in general regarded as advantageous, we prefer a Cobb-Douglas form for its parsimony in parameters and consequently the number of required instruments. Translog specifications of the models did not yield different or clearer results.

³ This approach resembles a corrected ordinary least squares approach with additional efficiency determinants (the innovation variables) incorporated in the frontier estimation.

⁴ We reduce the instrument count by restricting lag lengths of ‘GMM style’ instruments, as well as ‘collapsed’ instrument matrices, both options that can be implemented with the user-written command ‘xtabond2’ (Roodman 2009b) in STATA.

⁵ We decide on a separate examination of contemporaneous and lagged effects for the greater flexibility in the choice of instrumental variables.

⁶ In 2017, the production value of the dairy processing sector amounted to approximately 17% of the food sector of the EU 28 (Eurostat 2019).

⁷ For example, the German Federal Office for Agriculture and Food indicates that the number of dairy companies in Germany has decreased from 194 in 2009 to 124 dairies in 2015 (BLE 2017).

⁸ Quota volumes were gradually increased 2007 through 2014 with the intention of a soft-landing approach. Until 2007, cows’ milk collection stagnated, and has increased since then at a yearly rate of 1.4% until 2018 (European Commission 2019).

⁹ According to information from the journals’ editorial office, dairies generally notify the journal about new product introductions. Because dairies have an interest in making new product introductions public, it is to be expected that almost all new products in the market are covered by the journal’s review. Some new product introductions that are not reported to the journal are additionally researched by the journal itself. Therefore, especially for German dairies, which constitute the majority of our sample, we can expect almost complete coverage of all new products introduced into the market.

¹⁰ On general strengths and weaknesses of the indicator see also Coombs, Narandren, and Richards (1996).

¹¹ When the excluded interaction terms were included, their coefficients did not show statistical significance.

Additional information

Funding

This work was supported by the Bavarian Ministry of Food, Agriculture and Forestry.