A lasting crisis affects R&D decisions of smaller firms: the Greek experience

Giotopoulos, Ioannis; Kritikos, Alexander S.; Tsakanikas, Aggelos Article — Published Version A lasting crisis affects R&D decisions of smaller firms: the Greek experience The Journal of Technology Transfer Provided in Cooperation with: Springer Nature Suggested Citation: Giotopoulos, Ioannis; Kritikos, Alexander S.; Tsakanikas, Aggelos (2022) : A lasting crisis affects R&D decisions of smaller firms: the Greek experience, The Journal of Technology Transfer, ISSN 1573-7047, Springer US, New York, NY, Vol. 48, Iss. 4, pp. 1161-1175, https://doi.org/10.1007/s10961-022-09957-7 This Version is available at: https://hdl.handle.net/10419/308176 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Vol.:(0123456789) The Journal of Technology Transfer (2023) 48:1161–1175 https://doi.org/10.1007/s10961-022-09957-7 1 3 A lasting crisis affects R&D decisions ofsmaller firms: theGreek experience IoannisGiotopoulos1· AlexanderS.Kritikos2,3,4,5· AggelosTsakanikas6 Accepted: 6 July 2022 / Published online: 8 August 2022 © The Author(s) 2022 Abstract We use the prolonged Greek crisis as a case study to understand how a lasting economic shock affects the innovation strategies of firms in economies with moderate innovation activities. Adopting the 3-stage CDM model, we explore the link between R&D, innovation, and productivity for different size groups of Greek manufacturing firms during the prolonged crisis. At the first stage, we find that the continuation of the crisis is harmful for the R&D engagement of smaller firms while it increased the willingness for R&D activities among the larger ones. At the second stage, among smaller firms the knowledge production remains unaffected by R&D investments, while among larger firms the R&D decision is positively correlated with the probability of producing innovation, albeit the relationship is weakened as the crisis continues. At the third stage, innovation output benefits only larger firms in terms of labor productivity, while the innovation-productivity nexus is insignificant for smaller firms during the lasting crisis. Keywords Small firms· Large firms· R&D· Innovation· Productivity· Long-term crisis JEL Classification L25· L60· O31· O33 * Alexander S. Kritikos [email protected] Ioannis Giotopoulos [email protected] Aggelos Tsakanikas [email protected] 1 Department ofEconomics, School ofEconomics andTechnology, University ofPeloponnese, Tripoli Campus, 22100Tripoli, Greece 2 German Institute forEconomic Research (DIW Berlin), Berlin, Germany 3 University ofPotsdam, Potsdam, Germany 4 GLO, Essen, andIAB, Nuremberg, Germany 5 DIW Berlin, Mohrenstr. 58, 10117Berlin, Germany 6 Laboratory ofIndustrial andEnergy Economics, School ofChemical Engineering, National Technical University ofAthens, Zografou Campus, 15780Athens, Greece 1162 I.Giotopoulos et al. 1 3 1 Introduction Pursuing innovative strategies is critical for improving a firm’s output through increased productivity. What is already important in normal times (see inter alia Griffith etal., 2006, Hall etal., 2010; Huergo & Moreno, 2011; Baumann & Kritikos, 2016; Lööf etal., 2017) might become crucial in times of crisis: when firms are confronted with sharp reductions in sales, they must develop a sustainable and dynamic recovery path through an innovation strategy (Archibugi etal., 2013). However, when sales collapse, firms typically put R&D expenditures at the top of the list for cutting. Also public R&D investment often drops simultaneously in economies with moderate innovation activities; examples of this include Greece and other southern European economies (see Pellens etal., 2020). We use the 2008 Geek economic crisis to investigate how firms react in a situation of a deep and prolonged downturn. It is a fact that the 2008 global financial crisis created a turbulent environment for the Greek economy, being particularly harmful for the economic performance and viability of a large number of Greek firms (Giotopoulos etal., 2017; Williams & Vorley, 2015). In this paper, we analyze what kind of innovation strategies the Greek manufacturing sector, separated into small and large firms, practiced during a crisis that turned out not to be just a short shock but rather a long, unabated shock, as well as how the respective strategies influenced the firms’ productivity. The long-term economic crisis in Greece offers a unique case study to explore how strong exogenous shocks to an economy affect small and large firms and their innovative behavior. Although the financial and debt crisis differs from the COVID-19 pandemic with respect to its sources, it shares two significant similarities. As Roper and Turner (2020) point out, both are sharp exogenous shocks rather than business-cycle fluctuations. Furthermore, both affected firms through strongly reduced liquidity, whether through a substantial reduction in the availability of commercial finance (economic crisis) or extensively reduced turnover (COVID-19 crisis) (see Fairlie, 2020, Fairlie & Fossen 2022). As the effects on business performance and the relevant business decisions may share common characteristics, an analysis could offer important insights for fine-tuning policy measures in the post-COVID era, especially in moderate innovation economies. In both cases, financial stringency will force firms to make rapid strategic decisions regarding spending and potential savings. Using a unique Greek data set from a two-wave survey of 524 Greek manufacturing firms during the financial crisis period (2011 and 2013), we employ the well-established model of Crépon etal. (1998) to investigate how firms in different size groups of the manufacturing sector react as the crisis continued to trouble the economy. We find that, for the first stage of our analysis, the continuation of the crisis appears to be harmful for the R&D engagement of smaller firms while it increased the willingness for R&D activities among the larger ones. At the second stage, among smaller firms the knowledge production remains unaffected by R&D investments; among larger firms, the predicted R&D decision is positively correlated with the probability of producing innovation output. At the third stage, we observe that innovation output benefits only larger firms since it significantly improves their labor productivity. With our analysis, we contribute to the literature on the effect of lasting economic shocks on R&D investments. We provide a systematic analysis on potential effect heterogeneities in R&D-innovation-productivity linkages for different firm size groups and we consider—to the best of our knowledge for the first time—how a lasting shock causes large economic imbalances in moderate innovator economies. This adds to the analysis on the 1163 A lasting crisis affects R&D decisions ofsmaller firms: the… 1 3 effects of short-term shocks like the 2008 global financial crisis (as analyzed e.g. by Archibugi etal., 2013) as well as linkages between uncertainty and innovation of firms and the linkages between innovation activities and performance of SMEs for a large number of countries [as analyzed in two studies by Goel and Nelson (2021, 2022)]. In that sense, our research is relevant and novel as it may also allow for designing policy instruments intended to increase the resilience of firms across different size groups, i.e. small vs. large firms, through the R&D-innovation-productivity channel under prolonged turbulent economic conditions. Despite the fact that we analyze one country, we argue that our research has important implications for other countries as well. The unique empirical insights may be valuable for other moderate innovator economies that face such crises and are dominated by SMEs. 2 Theoretical background, data andcrisis measurement 2.1 Theoretical background Investments in R&D and innovation activities are, already in normal times, risky decisions aiming to increase the productivity performance of firms. To analyze this relationship, Griliches (1979) introduces a knowledge production function according to which investments into R&D increase the stock of knowledge, leading to innovation and, ultimately, to higher productivity. At the same time, such investments bear the risk of failure, as it might not be possible to realize positive returns on such investments (see inter alia Peters etal., 2017). The uncertainty from exogenous shocks may lead firms to delay or even abandon R&D projects, but uncertainty may also induce the introduction of cost-saving process innovations, thus acting as a hedge against risks (Goel & Nelson, 2021). There is also extensive research that empirically investigates—based on the Griliches (1979) knowledge production function and making use of the so called CDM model, a structural model introduced by Crépon etal. (1998)—the relationship between R&D, innovation, and labor productivity (see Hall, 2011, and Lööf, etal., 2017 for surveys). Existing research also focuses on the question to what extent are smaller firms similarly able to manage R&D efforts to improve their stock of knowledge and to transfer this improved knowledge into higher productivity? Reasons for firm size differences are the two conditions driving this R&D decision: opportunity and appropriability (Cohen & Klepper, 1996). From related empirical research, we know that firm size is indeed positively associated with the decision to invest in R&D. However, smaller firms still substantially engage in R&D activities. The question driving this research is whether or not smaller firms benefit in a comparable way from innovation processes: do they increase their labor productivity in a way that is similar to large firms (see Hall etal., 2009; Baumann & Kritikos, 2016)? However, the impact of innovation activities on SME performance is a priori unclear, since process innovation may be cost-saving with respect to the production inputs or labor may exhibit strong complementarities with other inputs (Goel & Nelson, 2022). In this contribution, we investigate how the triad relationship between innovation input, innovation output, and productivity develops during a prolonged economic crisis. When major exogenous shocks jeopardize markets, smaller businesses tend to be more vulnerable than their larger counterparts due to lack of resources, known as the liability of smallness (Eggers, 2020). In a lasting crisis, smaller firms may be reluctant, if not unable, to invest their limited resources into innovative projects with an uncertain outcome (Lee etal., 2015) 1164 I.Giotopoulos et al. 1 3 or other activities that will increase their financial risks (Thorgren & Williams, 2020). This holds even more if firms will struggle to manage high levels of debt. Therefore, we aim to determine if smaller firms tend to refrain from investing in innovation activities during such long lasting crises. 2.2 Data andcrisis measurement The data used to empirically investigate our main research question stem from an extensive field survey conducted through CATI method. The first wave took place in 2011, the second in 2013, with the same group of firms being surveyed. We should emphasize that both observation years refer to a crisis period that hit only the Greek economy particularly strong. The final sample used in this paper contains 524 Greek manufacturing firms that participated in both survey waves. Table1 describes in detail the examined variables and presents per wave their frequency distributions for binary and 5-point Likert scale variables as well as some summary statistics for the continuous variables. We use the same set of firms, the firm size distribution of which is shown in Table2, in both waves enabling thus to identify possible differences over time. As shown in Table1, about 67% of the manufacturing firms1 of the sample have introduced a product or process innovation within the last two years of wave 1 (2011), whereas this rate falls to 58% in wave 2 (2013). About 25% of the sample indicated the existence of in-house R&D activities in 2011, which increased to 31% in 2013. Employee training is widely used, reaching 73% of the firms in both waves. Training costs seem to be unaffected and are not reduced despite the sharp increase in liquidity constraints. Liquidity constraints are substantial as the crisis continues and the percentage of firms that indicate a very high degree of bank credit difficulties, as it doubles between the two waves (from approximately 20% to 40%). Finally, the average values of labor productivity and capital investment remain almost stable in both waves (Table1). In this analysis, the crisis continuation variable is formulated with the value of 0 for the responses of 2011 and the value of 1 for the responses of 2013, the latter incorporating the peak of the Greek economic crisis.2 As a matter of fact, the recessionary cycle of the Greek economy began in 2008, along with the burst of the global economic crisis, when a first negative growth rate in the GDP was recorded (− 0.3%). By the end of 2011, the accumulative recession was − 18% of the Greek GDP, while at the end of 2013 Greece had lost 26.4% of its GDP. 2013 was the year when the Greek GDP was at its lowest level (measured in constant prices of 2015) since the outbreak of the crisis in 2008, thus representing the trough of the Greek experience. This is why we consider 2013 as a crucial milestone representing the worst moment of the Greek economic crisis (European Commission, 2017). The other important factor in this context is that over the following years (from 2014 onward) the Greek economy grew only slightly, if at all. Thus, after five years of strongly negative signs, the economy did not recover, rather it remained at a low level in 1 Note that our empirical work focuses on the manufacturing sector. The Oslo Manual and several studies emphasize that fundamental differences in the innovation process exist between manufacturing and the service sectors (Audretsch etal., 2020; Becheikh etal., 2006; Ettlie and Rosenthal 2011). 2 In this context, we need to emphasize one limitation of our study. There are no sufficient data on the Greek manufacturing sector from the previous non-crisis period (prior to 2008) that do allow to make a comparison to these years. The Greek data from the (typically used) community innovation survey (CIS) miss information on the labor force so that it is not possible to estimate effects on labor productivity. 1165 A lasting crisis affects R&D decisions ofsmaller firms: the… 1 3 Table 1 Descriptive statistics of the examined variables Variables Description 2011 wave 2013 wave In-house R&D Firms indicated whether they have organized or developed an R&D department during the last two years Frequency 1. Yes 25.96% 31.48% 0. No 74.04% 68.52% Innovation Output Firms indicated whether they were engaged in new or significantly improved product or process innovations within the last two years Frequency 1. Yes 67.37% 58.40% 0. No 32.63% 41.60% Training Firms indicated whether they provided external or internal training programs to their employees within the last two years Frequency 1. Yes 73.85% 72.85% 0. No 26.15% 27.15% Liquidity Constraints Firms indicated (on a 1–5 Likert scale), the level of credit crunch conditions they face due to banks inability to provide loans Frequency 1. None credit difficulties 25.96% 11.95% 2. Low degree of credit difficulties 18.27% 9.06% 3. Moderate degree of credit difficulties 19.42% 15.99% 4. Relatively high degree of credit difficulties 17.12% 22.16% 5. Very high degree of credit difficulties 19.23% 40.85% Metropolitan Firms indicated their location and based on this information a regional dummy was constructed referring to the two metropolitan areas of Greece, i.e. Athens and Thessaloniki Frequency 1. The firm is located in the metropolitan areas of Greece 37.02% 0. The firm is located in the rest regions of Greece (i.e. non-metropolitan areas) 62.98% Labor Productivity Sales per full time equivalent employees (in logs) Summary Statistics Mean 11.873 11.807 Std Dev 0.833 0.925 Max 14.685 15.174 Min 6.463 7.875 1166 I.Giotopoulos et al. 1 3 Table 1 (continued) Variables Description 2011 wave 2013 wave Investment intensity Capital investment per full time equivalent employees (in logs) Summary Statistics Mean 9.383 9.061 Std Dev 1.435 1.313 Max 15.807 13.074 Min 3.912 4.855 Table 2 Frequencies per Size Group Size Group Micro Firms (firms that employ fewer than 10 persons) Small Firms (firms that employ 10–49 persons) Medium Firms (firms that employ 50–249 persons) Large Firms (firms that employ 250 or more persons) % of firms 6.58% 40.94% 42.18% 10.31% 1167 A lasting crisis affects R&D decisions ofsmaller firms: the… 1 3 economic stagnation before dropping by another 9% in 2020 in the wake of the pandemic. Overall the use of the crisis continuation dummy allows us to identify potential changes in firms’ innovation activities during a prolonged crisis, and especially in the Greek case as the crisis is deepening. 3 Empirical strategy To explore the relationship between a firm’s decision to invest in R&D, its innovation output and productivity, we apply the well-established three-stage CDM model (Crépon etal. 1998) by a variant developed by Mairesse etal. (2005). The general benefits of this framework are extensively described in various approaches (see Lööf et al., 2017), while the benefits of the variant by Mairesse etal. (2005) with respect to selection bias and endogeneity issues are discussed in Audretsch et al. (2020). The important difference of the model provided by Mairesse etal. (2005) is that it refers to the use of occurrence instead of intensity for R&D engagement and innovation. Hence, the selection bias for R&D intensity does not hold, for which Crépon etal. (1998) had to correct for in their specification. Thus, the Heckman selection approach is not necessary in the first stage of the CDM model when the variant of Mairesse etal. (2005) is applied. For the sake of brevity, we keep the model description short. In the first stage, we use a bivariate probit model to estimate the innovation input; i.e., the probability of undertaking R&D activities (Mairesse etal., 2005). The decision of firm i to invest in R&D at time t ( r∗ i,t ) can be specified as follows: where ri,t represents the observed binary variable for the R&D decision, r∗ i,t denotes an unobserved latent variable that captures the probability of undertaking R&D activities, X ′ i,t is a vector of possible factors influencing the decision of firms to engage in R&D, and ei,t is the error term. When the unobserved latent variable exceeds a certain threshold level c , then the observed ri,t takes the value of 1, and 0 otherwise. Dt denotes the crisis continuation dummy where in our analysis the first observation year (2011) takes the value of 0, while the second observation year (2013), where the crisis deepened, takes the value of 1. In the second stage, the specification of the knowledge production focuses on the link between innovation input and innovation output. We use a probit model to estimate the probability of introducing an innovation output, where product and process innovation are merged to one variable of innovation output (Hall, 2011), by including the predicted R&D decision obtained from stage 1 as the explanatory variable. To this end, the knowledge production is modeled as: where the observed binary variable for innovation output is denoted by ii,t and the latent R&D decision predicted in the first stage is represented by r∗ i,t . Z′ i,t is a vector of factors that may influence the innovation output and ui is the error term. The third stage of the CDM approach makes use of a productivity function including the predicted innovation output derived from stage two as the explanatory variable, as a (1) r i,t= { 1, if r∗ i,t=X � i,ta+Dt𝜌+ei,t >c 0, if r∗ i,t =X� i,t a+Dt𝜌+ei,t≤ c (2) i i,t= { 1, if i∗ i,t=r∗ i,t𝛽+Z� i,t𝛿+Dt𝜆+ui,t >c 0, if i∗ i,t=r∗ i,t𝛽+Z� i,t𝛿+Dt𝜆+ui,t≤ c 1168 I.Giotopoulos et al. 1 3 proxy for knowledge input. To estimate the productivity, we use a Cobb–Douglas production function extended with the use of knowledge stock (Griliches, 1979). The equation of the OLS estimation is expressed in logs as follows: where the dependent variable yi , t denotes the labor productivity measured in sales per employees in logs. The explanatory variables of primary interest in the production function are the knowledge input ( i∗ i,t ) derived from the estimated innovation output in stage 2, and the capital input ( ki,t ) measured by the investment intensity in logs. Finally, Wi,t is a vector of control variables, and vi , t is the observed error term. 4 Results 4.1 First stage: R&D engagement We estimate the panel probit model expressed by Eq.(1) for the full sample and separately for the size groups,3 as defined above. Table3 presents the marginal effects of the explanatory variables on the probability of firms’ engagement in R&D activities. Focusing on the total sample (Column 1), we find that micro and small firms are less likely to engage in R&D activities than the reference group4 of large-sized firms (confirming earlier findings of Hall etal., 2009 and Baumann & Kritikos, 2016). To further explore whether the examined factors influence in a different way the R&D engagement of micro and small firms, as compared to their larger counterparts, we discuss the empirical results for the two size groups separately (Columns 2 and 3). In particular, the continuation of the crisis has a negative effect (significant at the 5% level) on the probability of micro and small firms engaging in R&D activities, while a positive and strong association (at the 1% level of significance) emerges in the case of larger firms. The coefficients’ values obtained from the marginal effects indicate that the continuation of the crisis is associated with an 8pp decrease in the probability of smaller firms to become involve in R&D activities, while there is a 20% increase in the probability of larger firms to engage in R&D. 4.2 Second stage: knowledge production Table4 presents the results from the second stage on the full sample and on the two examined size groups. For the full sample, we reveal a strong link between the predicted R&D (obtained from the previous stage) and innovation output in terms of probability (3) yi , t = 𝛼 1+ a 2 ki , t + a 3 i∗ i,t + a 4 Wi , t + Dt𝜇 + vi , t 3 Following Baumann and Kritikos (2016) we split the sample to provide separate estimations per size group (small versus large firms), since this enables the exploration of heterogenous effects between small and large firms. 4 As a common practice in estimations with k group dummies, we include k-1 variables to avoid perfect multicollinearity and the missing group is considered as the reference group in the interpretation of the estimated coefficients of the k-1 dummies (in our case the group dummies refer to size and age groups). In reporting the estimation results from the probit regressions in the first two stages we use marginal effects and the results referring to the missing reference group are discussed implicitly. The constant term and subsequently more direct results for the missing dummy group can be extracted from the initial probit estimations, i.e. before the computation of the marginal effects. These results are not reported since they are not of interest given the scope of the current study; however, they are available from the authors upon request. 1175 A lasting crisis affects R&D decisions ofsmaller firms: the… 1 3 Kritikos, A. S., Handrich, L., & Mattes, A. (2018). The Greek private sector remains full of untapped potential. DIW Weekly Report, 29, 263–272. Le Mouel, M., & Schiersch, A. (2020). Knowledge-based capital and productivity divergence. DIW Disc. Paper No. 1868. Lee, N., Sameen, H., & Cowling, M. (2015). Access to finance for innovative SMEs since the financial crisis. Research Policy, 44(2), 370–380. Lööf, H., Mairesse, J., & Mohnen, P. (2017). CDM 20 years after. Economics of Innovation and New Technology, 26(1–2), 1–4. Lustig, N., & Mariscal, J. (2020). How COVID-19 could be like the Global Financial Crisis (or worse). In R. Baldwin & B. Weder di Mauro (Eds.), Mitigating the COVID economic crisis: Act fast and do whatever it takes (pp. 185–190). CEPR. Mairesse, J., Mohnen, P., & Kremp, E. (2005). The importance of R&D and innovation for productivity: A reexamination in light of the French innovation survey. Annales d’Economie et de Statistique, 487–527. Matt, M., Robin, S., & Wolff, S. (2012). The influence of public programs on inter-firm R&D collaboration strategies: Project-level evidence from EU FP5 and FP6. The Journal of Technology Transfer, 37, 885–916. Nickell, S., Nicolitsas, D., & Patterson, M. (2001). Does doing badly encourage management innovation? Oxford Bulletin of Economics and Statistics, 63(1), 5–28. Pellens, M., Peters, B., Hud, M., Rammer, C., & Licht, G. (2020). Public R&D investment in economic crises, ZEW Discussion Paper No. 20-088, Mannheim. Peters, B., Roberts, M., Vuong, V. A., & Fryges, H. (2017). Estimating dynamic R&D choice: An analysis of costs and long-run benefits. RAND Joural of Economics, 48(2), 409–437. Petrin, T., & Radicic, D. (2021). Instrument policy mix and firm size: is there complementarity between R&D subsidies and R&D tax credits? The Journal of Technology Transfer. Roper, S., & Turner, J. (2020). R&D and innovation after COVID-19: What can we expect? A review of prior research and data trends after the great financial crisis. International Small Business Journal, 38(6), 504–514. Thorgren, S., & Williams, T. A. (2020). Staying alive during an unfolding crisis: How SMEs ward off impending disaster. Journal of Business Venturing Insights, 14, e00187. Thrane, S., Blaabjerg, S., & Møller, R. H. (2010). Innovative path dependence: Making sense of product and service innovation in path dependent innovation processes. Research Policy, 39(7), 932–944. Williams, N., & Vorley, T. (2015). The impact of institutional change on entrepreneurship in a crisis-hit economy: The case of Greece. Entrepreneurship & Regional Development, 27(1–2), 28–49. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.