By Nanditha Mathew, Sant’Anna School of Advanced Studies, Italy (firstname.lastname@example.org)
For Citation: SITE4Society Brief No. 2-2018
Related to SDG Goals and Indian National Programmes: #SDG8 (Decent Work and Economic Growth) #SDG9 (Industry, Innovation and Infrastructure) #IIGP (India Innovation Growth Programme) #AIM (Atal Innovation Mission)
SITE focus: Innovation in Indian industries; Country Focus: India;
Sub-disciplines: Industrial organization, Development economics, Economics of innovation.
Based on: Giovanni Dosi, Marco Grazzi and Nanditha Mathew, The cost-quantity relations and the diverse patterns of “learning by doing”: Evidence from India, Research Policy, 46, 377–393 (2017).
Caveat: Please note that our academic works are based on a detailed study of a niche context constrained by the limitations of time and resources. Thus, our policy inferences, which have been extrapolated from such studies should be considered as indicators to be re-confirmed in every new context considered for application.
Context: Theoretical and empirical studies in economics consider “learning-by-doing” as a process in which an increase in experience in a particular type of production (‘doing’) yields an improvement in efficiency (‘learning’). The simplified version of “learning by doing”, henceforth LBD, presents significant drawbacks.
First, LBD, as shown in the innovation literature is only one of several, often complementary forms of knowledge accumulation.
Second, even when strictly applied, it is often considered that learning is a costless activity and an automatic by-product of continued production activities.
Third, it is generally assumed that all organizations have the same capacity to learn and there are no differences in absorptive capacities that might lead to differences in the intensity of learning across different organizations.
Fourth, product characteristics are unlikely to remain invariant once firms invest in learning. Rather, the object of ‘learning’ often improves product performance, but at the same time, triggers changes in its production costs.
Are the above notions true for Indian firms? Needless to say, this query is pertinent for policy design.
Research Questions: What do learning curves look like at the product-level? How do learning curves behave? Are the slope of the learning curves affected by R&D and fixed investment activities by firms?
Motivation for Research Questions: At the outset, it must be noted that there is no standard way to measure learning by firms. However, if a firm’s per unit cost or price falls over time as its production increases, then it could be due to learning. With this intuition, then it is possible to study learning curves of firms by tracing and analysing their cost or price and production portfolios over time.
Applying this method over the entire manufacturing sector, if one finds Indian organizations to have widespread and relatively uniform learning curves, it would support the view that at any point of time, learning is simply the unintentional side effect of production. It would also indicate that over time, the larger the scale of production, the greater the learning experience. Indeed a wide ensemble of growth models are based on such notions.
On the other hand, if one were one to find a lot of inter-product/inter technological/inter-firm diversity in learning rates, then it would mean that firm learning depends on both the specificities of the different technologies and the characteristics and strategies of different firms.
The policy implications for fostering greater productivity and innovation generation would be very different depending on which form of firm learning corresponds to reality. More generally, the latter perspective calls for a greater examination of the complementarity between production-related learning on the one hand, along with other drivers of knowledge accumulation, such as investment in R&D.
Methodology Used: The data we use comes from the Prowess database, provided by the CMIE (Centre For Monitoring Indian Economy Pvt. Ltd.). The learning curves are derived by plotting the cost-quantity relationships, in other words, how cost or price change over time with increased production (or experience) at the product-level.
The relationship is estimated through different functional forms (linear, exponential, power and logarithmic functions) by performing a firm-level fixed effects regression. The differences between the learning parameters across sectors is further tested using Fligner–Policello location test. The relationship between the observed learning parameters and innovative activities of firms (product and process innovation) are investigated separately for each group of sectors through standard regression techniques.
Main findings: The findings indicate that the evolution of cost-quantity relationships over time differ a lot across products belonging to sectors with different “technological intensities”.
Puzzlingly, in quite a few cases, the relation price/cumulative quantities is increasing, in other words, the price (or cost) increase in time with increase in production. In technology-based sectors (science-based and specialized supplier sectors, like pharmaceuticals and electronics) and for technology-based products (for example, Monitors, Led Displays, Cellular phones etc), the relation price/cumulative quantities is increasing. In other words, here the price (or cost) increase in time with increase in production. For some products, we are able to visually identify a sudden increase in the cost or price, for instance in the case of CDs or DVDs around the year 2000. We conjecture that this is in fact due to quality improvement of the product. The general idea is that there are two types of learning, namely one associated with an increased efficiency in the production of a given product and another one linked to the ability of producing new/improved products. The case is illustrated in Figure 1 with unit costs/prices on the y-axis and, for convenience, time on the x-axis. On the other hand, in low-tech industries, like the traditional manufacturing sectors (like food, textiles etc), we find that the cost or price trend tends to decline in time.
The results also indicate that differential learning patterns are affected by firm spending on R&D and capital investments – with learning increasing with in-house efforts.
Hence, the evidence suggests that “learning”, or performance improvement over time is not just a by-product of the mere repetition of the same production activities, but rather it seems to be shaped by deliberate firm learning efforts (which include investments on capital and R&D). This would imply that subsidies or other policy intervention should not simply be on production, but also on R&D investment of firms.
Policy Relevance: In academic debates about industrial policy, much of the attention is placed in the mere “automatic” experience in production, with much less attention on the firm-level learning strategies or inter-sectoral differences. Academics have not paid enough attention to the so-called “non-doing” mechanisms of learning. Our findings show that we need to go beyond measuring “learning-by-doing” to looking at “learning by experience”. The latter will capture also those processes of capability accumulation, technological adoption, imitation, and finally cumulative innovation, at the level of firms and sectors.
Worldwide, including in India, in high-tech sectors, for firms in technology based sectors (like pharmaceuticals and electronics) which are continuously involved in product upgradation, special attention is given to facilitating R&D investment. Our findings confirm that this is in the right direction. Furthermore, we suggest that promotion of “learning by firms” should be based on measurements of the deliberate efforts undertaken by firms to improve their efficiencies. These might take the forms of investment in tangible assets (e.g. equipment embedding latest technology) or R&D spending. Finally, more than providing subsidies for production activities, governmental policy and aid should be targeted at R&D investment.
*For references and more details, please check – Dosi et al (2017), The cost-quantity relations and the diverse patterns of “learning by doing”: Evidence from India, Research Policy, 46, 377–393.
source of CMIE image: https://www.google.fr/search?q=CMIE+India&source=lnms&tbm=isch&sa=X&ved=0ahUKEwjbu6CBo6XZAhWRaFAKHfatCjAQ_AUICygC&biw=1366&bih=637#imgrc=sHg9Pkg01U8RPM: