Advanced digitalisation and resilience during the COVID-19 pandemic: firm-level evidence from developing and emerging economies

ABSTRACT

Advanced digital production technologies – often clustered under the label of ‘Industry 4.0’– are reshaping industrial production. Using novel firm-level data collected by UNIDO this paper investigates the extent to which these technologies shielded developing countries’ manufacturing firms in times of COVID-19. The results of the analysis show that the adoption of the latest vintage of digital technologies strengthened firms’ robustness against the shock and, at the same time, facilitated their readiness to respond and adapt to the new context. These findings pose relevant implications for the design of post-COVID recovery strategies to strengthen future industrial resilience in developing and emerging economies.

KEYWORDS:

• Industrial development
• digital technologies
• resilience
• fourth industrial revolution
• firm-level analysis
• COVID-19

JEL CLASSIFICATION:

• D22 , D24, F61, L 20, O12, O14, O33

Acknowledgements

The authors are grateful to the feedback received on an earlier version of this work during the 8^thEuropean Conference on Corporate R&D and Innovation CONCODRI (22-25 November 2021) and to comments provided by participants of the 18th International Schumpeter Society Conference (8-10 July 2021) and the 12th MEIDE Conference (29-30 November 2021). The authors thank all UNIDO staff members and colleagues, national and local government agencies and institutions, industry associations, business chambers, universities and non-governmental organizations that supported the implementation of the UNIDO COVID-19 Firm level surveys, whose data has been used in this paper. The authors also thank the two anonymous reviewers for their constructive comments and suggestion.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available from UNIDO. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from the authors with the permission of UNIDO.

DISCLAIMER

This paper is an extension of the background paper “Advanced digital technologies and industrial resilience during the COVID-19 pandemic: A firm-level perspective” produced for the UNIDO Industrial Development Report 2022. The views expressed in this paper are those of the authors and do not necessarily reflect the views of the Secretariat of UNIDO.

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Notes

¹ It is anyway worth to mention the work of Gal et al. (2019) on digitalisation and productivity in EU countries.

² It has to be noticed, however, that more firm-level data on new digital technologies have been collected in recent years in advanced economies, such as the European Manufacturing Survey (EMS), the Eurostat ICT Usage and E-commerce in Enterprises survey, the Investment Survey by the European Investment Bank (EIBIS). Some firm-level information on digitalisation can also be found in the executive surveys carried out by international consulting firms (McKinsey & Company 2020; PwC 2018). Besides being are rarely accessible, these executive surveys are designed to capture the opinions of business leaders, often using Likert-scale questions and not collecting information on firm’s characteristics (i.e. revenues, investments, number of employees). Thus, the data of these executive surveys poses challenges to the implementation of empirical analyses, although they can serve to complement the information collected by international organisations and national statistical offices through conventional firm-level surveys.

³ See Conz and Magnani (2020) for a more complete review of the in business and management literature on resilience.

⁴ For more information on the UNIDO COVID-19 survey, see UNIDO (2021) and the survey dedicated webpage (https://www.unido.org/covid19_surveys).

⁵ This is the same approach employed by other data collection exercises on digitalisation in a developing and emerging context (see Ferraz et al. (2019) and Kupfer, Ferraz, and Torracca (2019) in Brazil; Albrieu et al. (2019) in Argentina; UNIDO (2019) in Ghana, Thailand and Viet Nam; Avenyo, Bell, and Nyamwena (2022) in South Africa).

⁶ When needed, the original English questionnaire was translated into local language(s).

⁷ We acknowledge that heterogeneity may still surge, even if the same questionnaire is delivered in each country. The following section explains how we use country fixed-effects to control for that in the proposed empirical analysis.

⁸ The UNIDO COVID-19 survey collected information for a total of 3,880 respondents. We checked whether selection into the subsample used may have been driven by the other considered variables, but we did not find systematic differences between the subsample of 2,700 with information on production technologies and the remaining 1,180 firms without information on production technologies. See Appendix A for more information about data collection and sample composition.

⁹ Obtaining a statistically representative sample from a firm survey is per se a difficult exercise in developing context, where complete and updated business registries are often missing. This operation became even more challenging during the COVID-19 pandemic, due to the short time available to gather relevant information about the consequences of the COVID-19 crisis in a large number of countries and within a certain span of time. In some countries (Argentina, Cote d’Ivoire, Mauritius, Pakistan, Zambia) the support of local partners facilitated the access to national registries and databases, making it possible to obtain a more representative sample (taking into account the distribution of firm size by sector); unfortunately, this was not the case for most of the countries. In addition, social distancing and containment measures made it impossible to collect data through in-person interviews and the questionnaire had to be delivered via telephone or online through the interface of a survey manager platform. The latter clearly increased the risk of selection bias towards firms with Internet access and already familiar with email services.

¹⁰ See UNIDO (2019) and Delera et al. (2022) for a more detailed explanation of the ‘technological generations’ approach and of the variables related to the level of digitalisation of a firm.

¹¹ See in Appendix D for a summary of how the categories of the variable Production technologies (PT_i) were obtained.

¹² See Table B1 and Figure B1 in Appendix B for the results of PCA.

13 We thank an anonymous reviewer for this suggestion.

¹⁴ Specifically, a Stringency Index above 67.5. The period considered for the world average is March 2020 to June 2021. See Appendix C for more details on this variable.

¹⁵ Following Brancati, Brancati, and Maresca (2017) and Delera et al. (2022), we define GVC_i as a binary variable taking the value of 1 when a firm is either: an active exporter of intermediate products; a two-way trader (that is, a firm that exports and imports); or outsourced from abroad. We consider as two-way traders firms whose exports and imports represent over 10 percent of sales and expenditure on inputs, respectively (Del Prete, Giovannetti, and Marvasi (2017) and Montalbano, Nenci, and Pietrobelli (2018) apply this same threshold).

¹⁶ We thank an anonymous reviewer for this suggestion.

¹⁷ The results of separate OLS regressions for the models in Equation 1(1) $Robusteness\equiv \hspace{0.17em}\mathrm{\Delta }{S}_i\mathrm{o}\mathrm{r}\mathrm{\Delta }{P}_i={\beta }_0+{\beta }_1{x}_i+{\beta }_{2}ADP{T}_i+{\beta }_3{c}_i+{\beta }_4{s}_i+{\beta }_{5}Pandemi{c}_i+{\epsilon }_i$(1) and Equation 2(2) $Readiness\equiv TChang{e}_i={\beta }_0+{\beta }_1{x}_i+{\beta }_{2}ADP{T}_i+{\beta }_3{c}_i+{\beta }_4{s}_i+{\beta }_{5}Pandemi{c}_i+{\epsilon }_i$(2) are reported as reference in Table D1 in Appendix D. To improve the transparency of results, in Table D2 in Appendix D we provide a correlation matrix including the variables used in the presented models.

¹⁸ To complete the scope of our analysis, we also conducted a preliminary exploration of the factors possibly affecting the adoption of advanced technology and estimate a probability model on the binary variable ADPT_i. Results are available upon request.