Abstract
We use data from the last Ecuador Economic Census, covering the universe of manufacturing firms, to study the relationship between firms’ R&D and workers’ training investments and ICT use and firms’ productivity and markups. These knowledge-related investments may affect productivity. Moreover, investments in both knowledge and productivity can affect the ability of firms to set prices above marginal costs. Whether R&D and workers’ training investments and ICT are important for productivity and the capacity to set higher markups in developing countries are interesting development policy questions. We find that good business practices, including access to internal capital markets or to external finance, encourage R&D and workers’ training investments, and ICT use. These investments affect positively firms’ productivity and markups. Their influence on markups operates in general through efficiency and prices. Finally, there is evidence about demand conditions to boost knowledge-related investments and markups.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
There are 24 provinces in the country (see “Appendix 1 or 2”), three of which (Pichincha, Guayas and Azuay) account for 53.01% of manufacturing firms.
Industries are food, beverages and tobacco; textiles and wearing apparel; leather and footwear; wood, paper and printing; chemicals and petroleum products; rubber and plastics; non-metallic mineral products; metal products; office machinery and electrical equipment; communication, precision, optical and medical equipment; transport equipment; and furniture and n.e.c. This classification is based on the International Standard Industrial Classification at the two-digit level for manufacturing.
Legal forms are Natural Persons, Non-profit Company, Private Company, Foreign Company, Public Company, Local Government Company, Cooperative, and Association. These correspond to one of the questions in the survey.
Our justification for the inclusion of this last variable in the set of variables that tries to characterize firms from the point of view of good business practices is that, once controlling for industry fixed effects in estimation, the answers to this question in the survey respond, at least in part, to the firm’s attitude about the environmental aspects of its activity.
See the “Data” section above and Appendix 1 or 2.
For the R&D and ICT decisions it was not possible to estimate coefficients associated with the legal form Foreign company, the reason being a perfect prediction of zeros in the corresponding firm’s choices when the Foreign company dummy also has a value of zero. Firms with this legal form account for only 0.01% of total manufacturing firms.
The two Heckman’s lambda terms, Lambda R&D and Lambda training, are calculated as the expected value of the error term in the corresponding equation of interest (log of R&D or log of workers’ training intensity equations) conditional to the explanatory variables \( x_{1i}^{{}} \) in Eq. (1) and \( x_{2i}^{{}} \) in Eq. (3), and conditional on the observability of positive values for the R&D or workers’ training investments, respectively. As in our case \( x_{2i}^{{}} \) is a subset of \( x_{1i}^{{}} \), for regressors we only need to condition on the vector of explanatory variables \( x_{1i}^{{}} \). The Heckman’s method assumes that the errors in the two equations involved for sample selection correction (in our case \( \varepsilon_{2i,j} \) in Eq. (3) and \( \varepsilon_{1i,j} \) in Eq. (1)) have a bivariate normal distribution with 0 means and standard deviations \( \sigma_{{\varepsilon_{2,j} }} \) and 1, respectively, being \( \rho_{21,j} = (\sigma_{21,j} /\sigma_{{\varepsilon_{2,j} }} \sigma_{{\varepsilon_{1,j} }} ) = \left( {(\sigma_{21,j} )/\sigma_{{\varepsilon_{2,j} }} } \right) \) the correlation coefficient between error terms. Under these assumptions, the calculous of the following conditional mean gives rise to the lambda term: \( E\left( {\varepsilon_{2i,j} \left| {x_{1i}^{{}} , \, y_{1i,j}^{{}} = 1} \right.} \right) = E\left( {\varepsilon_{2i,j} \left| {x_{1i}^{{}} , \, y_{1i,j}^{*} > 0} \right.} \right) = E\left( {\varepsilon_{2i,j} \left| {x_{1i}^{{}} , \, \beta_{1,j}^{'} x_{1i}^{{}} + \varepsilon_{1i,j} > 0} \right.} \right) = \sigma_{{\varepsilon_{2,j} }} E\left( {\frac{{\varepsilon_{2i,j} }}{{\sigma_{{\varepsilon_{2,j} }} }}\left| {x_{1i}^{{}} , \, \varepsilon_{1i,j} > - \beta_{1,j}^{'} x_{1i}^{{}} } \right.} \right) = \rho_{21,j} \sigma_{{\varepsilon_{2,j} }} E\left( {\varepsilon_{1i,j} \left| {x_{1i}^{{}} , \, \varepsilon_{1i,j} > - \beta_{1,j}^{'} x_{1i}^{{}} } \right.} \right) = \sigma_{21,j} \lambda \left( { - \beta_{1,j}^{'} x_{1i}^{{}} } \right), \) where \( \lambda \left( { - \beta_{1,j}^{'} x_{1i}^{{}} } \right) \) is the lambda term or the inverse Mill’s ratio and \( \sigma_{21,j} \) is the covariance between \( \varepsilon_{2i,j} \) and \( \varepsilon_{1i,j} \). According to the properties of truncated normal distributions, and since the lambda term is no more than the mean of a standard normal random variable truncated from below at \( - \beta_{1,j}^{'} x_{1i}^{{}} \), it can be calculated as the ratio of the density function \( \phi \) over the cumulative distribution function \( \varPhi \) of a standard normal distribution evaluated at \( \beta_{1,j}^{'} x_{1i}^{{}} \), that is \( \lambda \left( { - \beta_{1,j}^{'} x_{1i}^{{}} } \right) = \left\{ {{{\phi \left( { - \beta_{1,j}^{'} x_{1i}^{{}} } \right)} \mathord{\left/ {\vphantom {{\phi \left( { - \beta_{1,j}^{'} x_{1i}^{{}} } \right)} {\left[ {1 - \varPhi \left( { - \beta_{1,j}^{'} x_{1i}^{{}} } \right)} \right]}}} \right. \kern-0pt} {\left[ {1 - \varPhi \left( { - \beta_{1,j}^{'} x_{1i}^{{}} } \right)} \right]}}} \right\} = {{\phi \left( {\beta_{1,j}^{'} x_{1i}^{{}} } \right)} \mathord{\left/ {\vphantom {{\phi \left( {\beta_{1,j}^{'} x_{1i}^{{}} } \right)} {\varPhi \left( {\beta_{1,j}^{'} x_{1i}^{{}} } \right)}}} \right. \kern-0pt} {\varPhi \left( {\beta_{1,j}^{'} x_{1i}^{{}} } \right)}} = \lambda \left( {\beta_{1,j}^{'} x_{1i}^{{}} } \right) \). Notice that on its final expression we have used the symmetry property of the normal distribution.
In a recent survey on TFP estimation, Van Beveren (2012) performs an empirical evaluation of TFP estimation methods as regards yielding different conclusions when conducting policy or impact evaluations (e.g., trade liberalization, deregulation, etc.). He shows that comparing OLS estimates with more sophisticated methods available for panel data, high correlations between different estimated TFP measures emerge (higher than 0.8 or 0.95 depending on the methods) and, more importantly for us, similar conclusions are obtained when evaluating the effect of some policy change with different TFP measures.
The mean of log labor productivity in our sample is 8.858.
The means of the Cobb–Douglas and Translog TFP (in logs) in the sample are 3.231 and 5.141, respectively.
The variables acting as instruments are whether the firm is the mother company, declares to have access to finance, carries company accounting, and has a craft certification. Since the cross-sectional nature of our dataset limits our choice of instruments, they have been empirically selected to guarantee validity. That is, with our selection of exclusion restrictions, we accomplish simultaneously two objectives: (1) that the instruments used are significantly correlated with the R&D, workers’ training and ICT variables; and, (2) that they are not correlated with the error term in the productivity regressions. We have checked 1 by performing Wald tests of joint non-significance of instruments in the first stage estimates for the variables to be endogeneized. Accordingly, with the first stage estimates for ICT that are in column 3 of Table 1, we obtain a \( \chi_{\left( 4 \right)}^{2} \) = 602.91 (p value = 0.0000); with the first stage estimates for R&D and workers’ training intensities that are in columns 1 and 2, respectively, of Table 2, we obtain \( \chi_{\left( 4 \right)}^{2} \) = 18.52 (p value = 0.0010) and \( \chi_{\left( 4 \right)}^{2} \) = 49.10 (p value = 0.0000). Hence, we reject the null of non-significance. The results for the verification of 2 will be presented below when commenting the estimates from the productivity regressions. All papers in the CDM framework assume exclusion restrictions in order to identify and estimate the model. Typically, they estimate a more parsimonious productivity equation as regards regressions in previous stages (such as the estimation of innovation inputs effort and/or innovation output equations). For example, Arvanitis and Loukis (2009) besides the variables being instrumented include physical capital, R&D and controls (for size and sector). Griffith et al. (2006) and Crespi and Zuniga (2012), besides the predicted innovation variables only include in the productivity regressions physical capital investment per worker and the usual controls for firm size and industry dummies. Aboal and Tacsir (2018) further include the ratio of professionals and technicians in the workforce.
Estimated residuals for the two knowledge investment intensity variables come from the bivariate Heckman in “Firms’ Investments: R&D and workers’ training” section. The estimated residual for the ICT dichotomous decision comes from the difference between \( y_{1i,ICT}^{{}} - P\left( {y_{1i,ICT}^{*} > 0} \right) \) obtained from results in “The Firms’ Decisions on Knowledge Creation Activities: R&D, workers’ training and ICT Use” section. Similar results were obtained when alternatively using a generalized residual for ICT.
Hence, the estimates reported in Table 3 already eliminate the instrumentation of the ICT variable.
We have checked that the instruments are not correlated with the error term in the productivity regressions by performing Sargan—Hansen tests of overidentifying restrictions. Notice that we can perform such tests since we have four instruments to instrument R&D and workers’ training intensities. For the labor productivity regression in column 1 of Table 3, we get a \( \chi_{\left( 2 \right)}^{2} \) = 1.98 (p value = 0.3722); for the Cobb–Douglas TFP regression in column 2 of Table 3, we get a \( \chi_{\left( 2 \right)}^{2} \) = 2.52 (p value = 0.2838); and, for the Translog TFP regression in column 3 of Table 3, we get a \( \chi_{\left( 2 \right)}^{2} \) = 1.52 (p value = 0.4681). Hence, we do not reject the nulls of no correlation.
If we had a proper export dummy, it could also contain relevant demand side information when firm prices are set differently in domestic than in export markets (Aw et al. 2011).
De Loecker and Warzyinski (2012) compare in their empirical work estimation results obtained with a translog and with a Cobb–Douglas production technology, although their main empirical specification relies on a translog. In their online appendix it is shown that estimated percentage differences between exporters and non-exporters in terms of markups are very similar under both production technologies, although somewhat lower with a translog.
The variables acting as instruments are the same ones than in the productivity regressions. See footnote 12.
In column 2 of Table 5 we exclude a variable in \( x_{4i}^{{}} \) with respect to \( x_{3i}^{{}} \), log workers squared, which contributes additionally to the endogenization of TFP in this markups specification. This variable is significantly correlated with TFP (see its statistical significance in column 3 of Table 3).
In our baseline specification (column 1), the residual for ICT is non-significant. Instead, in column 2 is positive and significant. Hence, the estimates reported in Table 5 for column 1 already eliminate the instrumentation of the ICT variable.
We have checked that the instruments are not correlated with the error term in the markups regressions by performing Sargan–Hansen tests of overidentifying restrictions. In the regression in column 1 of Table 5, we have four instruments to instrument R&D and workers’ training intensities, and we get a \( \chi_{\left( 2 \right)}^{2} \) = 1.99 (p value = 0.3703). In the regression in column 2 of Table 5, we have the same four instruments plus the extra-instrument log workers squared to instrument R&D and workers’ training intensities, ICT use, and TFP. In this case, we obtain a \( \chi_{\left( 1 \right)}^{2} \) = 0.67 (p value = 0.4123). Hence, we do not reject the nulls of no correlation.
The reason is that ‘revenue’ productivity may still potentially capture differences in firms’ prices. In any case, De Loecker and Warzyinski (2012) show that using ‘revenue’ productivity affects only the level of the markup estimates, and not the correlation between markups and firm-level characteristics.
References
Aboal, D., and E. Tacsir. 2018. Innovation and productivity in services and manufacturing: The role of ICT. Industrial and Corporate Change 27 (2): 221–241.
Añón-Higón, D. 2012. The impact of ICT on innovative activites: Evidence for UK SMEs. International Small Business Journal 30: 684–699.
Arvanitis, S., and E.N. Loukis. 2009. Information and communication technologies, human capital, workplace organization and labor productivity: A comparative study based on firm-level data for Greece and Switzerland. Information Economics and Policy 21: 43–61.
Aw, B.Y., M.J. Roberts, and T. Winston. 2007. Export market participation, investments in R&D and worker training, and the evolution of firm productivity. The World Economy 30 (1): 83–104.
Aw, B.Y., M.J. Roberts, and D.Y. Xu. 2011. R&D investment, exporting and productivity dynamics. American Economic Review 101: 1313–1344.
Baldwin, R. 2016. The great convergence: Information technology and the new globalization. Cambridge, MA: Harvard University Press.
Benavente, J.M. 2006. The role of research and innovation in promoting productivity in Chile. Economics of Innovation and New Technology 55: 301–315.
Bernard, A.B., J. Eaton, J.B. Jensen, and S. Kortum. 2003. Plants and productivity in international trade. American Economic Review 93: 1268–1290.
Black, S.E., and L.M. Lynch. 2001. How to compete: The impact of workplace practices and information technology on productivity. Review of Economics and Statistics 83: 434–445.
Bloom, N., M. Darca, T. Kretschmer, R. Sadun, and J. Van Reenen. 2010. The economic impact of ICT. LSE: Center for Economic Performance.
Bratti, M., and G. Felice. 2012. Are exporters more likely to introduce product innovation? The World Economy 35: 1559–1598.
Bresnahan, T.F., and M. Trajtenberg. 1995. General purpose technologies ‘Engines of growth’? Journal of Econometrics 65: 83–108.
Bresnahan, T.F., E. Brynjolfsson, and L.M. Hitt. 2002. Information technololgy, workplace organization, and the demand for skilled labor: Firm-level evidence. The Quarterly Journal of Economics 117: 339–376.
Brynjolfsson, E., and L.M. Hitt. 2003. Computing productivity: Firm-level evidence. The Review of Economics and Statistics 85 (4): 793–808.
Castiglione, C. 2012. Technical efficiency and ICT investment in Italian manufacturing firms. Applied Economics 44: 1749–1763.
Cohen, W.M., and D.A. Levinthal. 1990. Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly 3: 128–152.
Colombo, E., and L. Stanca. 2014. The impact of training on productivity: Evidence from a panel of Italian firms. International Journal of Manpower 35: 1140–1158.
Commander, S., R. Harrison, and N. Menezes-Filho. 2011. ICT and productivity in developing countries: New firm-level evidence from Brazil and India. Review of Economics and Statistics 93: 528–541.
Corchón, L.C., and L. Moreno. 2010. Interpreting markups in Spanish manufacturing: The exponential model. MPRA paper 26609. Munich: University Library of Munich.
Corrado, C., C. Hulten, and D. Sichel. 2005. Intangible capital and economic growth measuring capital and technology: An expanded framework. In Measuring capital in the new economy, studies in income and wealth, vol. 65, ed. C. Corrado, C. Hulten, and D. Sichel. Chicago: University of Chicago Press.
Crépon, B., E. Duguet, and J. Mairesse. 1998. Research and development, innovation and productivity: An econometric analysis at the firm level. Economics of Innovation and New Technology 7 (2): 115–158.
Crespi, G., and P. Zuniga. 2012. Innovation and productivity: Evidence from six Latin American countries. World Development 40 (2): 273–290.
De Loecker, J., and F. Warzyinski. 2012. Markups and firm-level export status. American Economic Review 102: 2437–2471.
Greenan, N., A. Topiol-Bensaid, and J. Mairesse. 2001. Information technology and research and development impacts on productivity and skills: Looking for correlations on French firm-level data. In Information technology, productivity, and economic growth, ed. M. Pohjola. Oxford: Oxford University Press.
Greene, W.H. 2003. Econometric analysis, 5th ed. New York: Prentice Hall.
Griffith, R., E. Huergo, J. Mairesse, and B. Peters. 2006. Innovation and productivity across four European countries. Oxford Review of Economic Policy 22 (4): 483–498.
Hall, R.E. 1988. The relation between price and marginal cost in U.S. industry. Journal of Political Economy 96 (5): 921–947.
Hall, B. 2011. Innovation and productivity. Nordic Economic Policy Review 2: 165–203.
Hall, B.H., F. Lotti, and J. Mairesse. 2013. Evidence on the impact of R&D and ICT investments on innovation and productivity in Italian firms. Economics of Innovation and New Technology 22: 300–328.
Heckman, J. 1979. Sample selection bias as a specification error. Econometrica 47: 153–161.
INEC. 2010. Economic census of Ecuador. Quito: INEC.
ITU. 2017. Measuring the information society, vol. 1. Geneva: International Telecommunication Union. https://www.itu.int/en/ITU-D/Statistics/Pages/publications/mis2017.aspx. Accessed 20 Aug 2018.
Janz, N., H. Loof, and B. Peters. 2004. Innovation and productivity in German and Swedish manufacturing firms: Is there a common story? Problems and Perspectives in Management 2: 184–204.
Konings, J., and S. Vanormelingen. 2015. The impact of training on productivity and wages: Firm-level evidence. Review of Economics and Statistics 97: 485–497.
Levinsohn, J., and A. Petrin. 2003. Estimating production functions using inputs to control for unobservables. Review of Economic Studies 70: 317–342.
Melitz, M. 2003. The impact of trade on intra-industry reallocations and aggregate industry productivity. Econometrica 71: 1695–1725.
Melitz, M., and G. Ottaviano. 2008. Market size, trade and productivity. Review of Economic Studies 75: 295–316.
Ministerio de Industrias y Productividad. 2016. Política Industrial del Ecuador 2016-2025. https://www.industrias.gob.ec/wp-content/uploads/2017/01/politicaIndustrialweb-16-dic-16-baja.pdf. Accessed 20 Aug 2018.
Olley, G.S., and A. Pakes. 1996. The dynamics of productivity in the telecommunications equipment industry. Econometrica 64: 1263–1297.
Polder, M., G. van Leeuwen, P. Mohnen, and W. Raymond. 2010. Product, process and organizational innovation: Drivers, complementarity and productivity effects. UNU-MERIT Working Paper Series 035.
Rincón, A., M. Vecchi, and F.Venturini. 2013. ICT as a general purpose technology: Spillovers, absorptive capacity and productivity performance. Discussion Paper 416. London: NIESR.
Rivers, D., and Q.H. Vuong. 1988. Limited information estimators and exogeneity tests for simultaneous probit models. Journal of Econometrics 39: 347–366.
Roodman, D. 2011. Fitting fully observed recursive mixed-process models with cmp. Stata Journal 11: 159–206.
SENPLADES. 2017. Plan Nacional de Desarrollo 2017-2021. Quito-Ecuador. http://www.planificacion.gob.ec/wp-content/uploads/downloads/2017/10/PNBV-26-OCT-FINAL_0K.compressed1.pdf. Accessed 20 Aug 2018.
UNIDO. 2016. Industry 4.0 opportunities and challenges of the new industrial revolution for developing countries and economies in transition (Panel discussion). https://www.unido.org/sites/default/files/2017-01/Unido_industry-4_NEW_0.pdf. Accessed 20 Aug 2018.
Van Beveren, I. 2012. Total factor productivity estimation: A practical review. Journal of Economic Surveys 26 (1): 98–128.
Van Leeuwen, G., and H. van der Wiel. 2003. Do ICT spillovers matter? Evidence from Dutch firm-level data, 26. The Hague: CPB Netherlands Bureau for Economic Policy Analysis.
WIPO. 2017. The Global Innovation Index 2017: Innovation feeding the world. Geneva: WIPO. http://www.wipo.int/publications/es/details.jsp?id=4193. Accessed 20 Aug 2018.
Wooldridge, J.M. 2009. On estimating firm-level production function using proxy variables to control for unobservables. Economics Letters 104: 112–114.
Wooldridge, J.M. 2010. Econometric analysis of cross section and panel data. Cambridge, MA: MIT Press.
World Economic Forum. 2015. The global competitiveness report 2015-2016. Geneva: World Economic Forum. https://www.weforum.org/reports/global-competitiveness-report-2015. Accessed 20 Aug 2018.
Zwick, T. 2006. The impact of training intensity on establishment productivity. Industrial Relations: A Journal of Economy and Society 45: 26–46.
Acknowledgements
J.A. Rodríguez-Moreno gratefully acknowledges financial support from the Ecuadorian Secretary of Higher Education, Science, Technology and Innovation, under its Grants Program Open Call 2011; and research funds for the period 2014–2019 from the Universidad Santa María Guayaquil, Ecuador. M.E. Rochina-Barrachina also acknowledges financial support from the Spanish Ministerio de Economía, Industria y Competitividad and the Spanish Agencia Estatal de Investigación (reference number ECO2017-86793-R, co-financed with FEDER funds, European Union).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
On behalf of all authors, the corresponding author states that there is no conflict of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Appendix 1: Variables Description
Variable | Description |
|---|---|
R&D | Dummy variable taking value 1 if the firm does R&D activities and 0 otherwise |
Workers’ training | Dummy variable taking value 1 if the firm performs training programs for the employees and 0 otherwise |
ICT | Dummy variable taking value 1 if the firm uses Information and Communication Technologies, and 0 otherwise. |
R&D intensity | Expenditure in R&D per employee |
Training intensity | Expenditure in training programs per employee |
Provinces | Dummy variables taking value 1 if the firm is located in a particular province and 0 otherwise. Provinces are: Azuay, Bolívar, Cañar, Carchi, Cotopaxi, Chimborazo, El Oro, Esmeraldas, Guayas, Imbabura, Loja, Los Ríos, Manabí, Morona Santiago, Napo, Pastaza, Pichincha, Tungurahua, Zamora, Galápagos, Sucumbíos, Orellana, Santo Domingo, y Santa Elena |
Industries | Dummy variables taking value 1 if the firm belongs to a particular industry and 0 otherwise. Industries are Food, beberages and tobacco; textiles and wearing apparel; leather and footwear; wood, paper and printing; chemicals and petroleum products; rubber and plastics; non-metallic mineral products; metal products; office mach. and elect. equipment; communi./prec./optic./med. equipment; transport equipment; furniture and n.e.c. |
Natural person | Dummy variable taking value 1 if the business has a natural person recognition by the National Tax Service, and 0 otherwise |
Non-profit company | Dummy variable taking value 1 if the firm is a non-government, non-lucrative organization, and 0 otherwise |
Private company | Dummy variable taking value 1 if the firm is a private company and 0 otherwise |
Foreign company | Dummy variable taking value 1 if the firm has foreign control and 0 otherwise |
Public company | Dummy variable taking value 1 if the firm is under the central government control and 0 otherwise |
Local Gov. company | Dummy variable taking value 1 if the firm is under the local government control and 0 otherwise |
Cooperative | Dummy variable taking value 1 if the firm is a cooperative and 0 otherwise |
Association | Dummy variable taking value 1 if the firm is considered an association and 0 otherwise |
Enterprise network | Dummy variable taking value 1 if the firm is a member of an enterprise network or business group, and 0 otherwise |
Market research | Dummy variable taking value 1 if the firm does market research and 0 otherwise |
Accountancy | Dummy variable taking value 1 if the firm has accounting control and 0 otherwise |
Access to finance | Dummy variable taking value 1 if the firm has access to external finance and 0 otherwise |
Environment | Dummy variable taking value 1 if the firm does some activity to improve the environment and has environmental concerns, and 0 otherwise |
Main customer foreign | Dummy variable taking value 1 if the firm has a foreign main customer and 0 otherwise |
Craft certification | Dummy variable taking value 1 if the firm has a Craft Certification and 0 otherwise. It is giving by the government to ‘natural persons’ (self-employed) who demonstrate long experience (in years) with handmade or technician work (non-professional). The ‘natural persons’ enjoy a tax benefit with this type of certification |
Own local HQ | Dummy variable taking value 1 if the local of the firm is own by the firm, and 0 otherwise |
Mother company | Dummy variable taking value 1 if the firm is a mother company and 0 otherwise |
Male manager | Dummy variable taking value 1 if the firm manager is a male and 0 otherwise |
Log workers | Number of employees of the firm. This variable is in log form |
(Log workers)2 | Number of log employees squared |
Log age | Number of years since the firm was born. This variable is in log form |
(Log age)2 | Log age squared |
Log labor productivity | Sales per employee in log form |
Log capital | Stock of tangible fixed assets at book values. This variable is in log form |
Log capital/worker | Stock of tangible fixed assets at book values per worker. This variable is in log form |
Log material | Amount of materials. This variable is in log form |
Log materials/worker | Amount of materials per worker. This variable is in log form |
Appendix 2: Descriptive Statistics
Variable | Obs. | Mean | SD | Mean performers (4938 obs.) | Mean non-performers (37,354 obs.) |
|---|---|---|---|---|---|
R&D | 42,292 | 0.0097 | 0.0982 | 0.0834 | – |
Workers’ training | 42,292 | 0.0432 | 0.2033 | 0.3701 | – |
ICT | 42,292 | 0.0987 | 0.2982 | 0.8450 | – |
Log R&D intensity | 412 | 5.1555 | 1.7565 | 5.1555 | – |
Log training intensity | 1828 | 4.3227 | 1.4649 | 4.3227 | – |
Azuay | 42,292 | 0.1026 | 0.3034 | 0.1160 | 0.1008 |
Bolívar | 42,292 | 0.0080 | 0.0891 | 0.0026 | 0.0087 |
Cañar | 42,292 | 0.0194 | 0.1379 | 0.0101 | 0.0206 |
Carchi | 42,292 | 0.0078 | 0.0883 | 0.0022 | 0.0086 |
Cotopaxi | 42,292 | 0.0294 | 0.1690 | 0.0141 | 0.0314 |
Chimborazo | 42,292 | 0.0398 | 0.1955 | 0.0360 | 0.0403 |
El Oro | 42,292 | 0.0377 | 0.1905 | 0.0257 | 0.0393 |
Esmeraldas | 42,292 | 0.0166 | 0.1280 | 0.0068 | 0.0179 |
Guayas | 42,292 | 0.1887 | 0.3913 | 0.1769 | 0.1902 |
Imbabura | 42,292 | 0.0393 | 0.1944 | 0.0409 | 0.0391 |
Loja | 42,292 | 0.0382 | 0.1918 | 0.0269 | 0.0397 |
Los Ríos | 42,292 | 0.0292 | 0.1685 | 0.0145 | 0.0312 |
Manabí | 42,292 | 0.0561 | 0.2301 | 0.0348 | 0.0589 |
Morona Santiago | 42,292 | 0.0096 | 0.0977 | 0.0054 | 0.0101 |
Napo | 42,292 | 0.0046 | 0.0680 | 0.0028 | 0.0048 |
Pastaza | 42,292 | 0.0066 | 0.0815 | 0.0038 | 0.0070 |
Pichincha | 42,292 | 0.2388 | 0.4264 | 0.3772 | 0.2205 |
Tungurahua | 42,292 | 0.0605 | 0.2384 | 0.0641 | 0.0600 |
Zamora | 42,292 | 0.0069 | 0.0830 | 0.0016 | 0.0076 |
Galápagos | 42,292 | 0.0020 | 0.0447 | 0.0022 | 0.0019 |
Sucumbíos | 42,292 | 0.0078 | 0.0879 | 0.0050 | 0.0081 |
Orellana | 42,292 | 0.0050 | 0.0707 | 0.0040 | 0.0051 |
Santo Domingo | 42,292 | 0.0297 | 0.1698 | 0.0214 | 0.0308 |
Santa Elena | 42,292 | 0.0139 | 0.1170 | 0.0038 | 0.0152 |
Food, beverages and tobacco | 42,292 | 0.2211 | 0.4150 | 0.1462 | 0.2310 |
Textiles and wearing apparel | 42,292 | 0.2188 | 0.4134 | 0.2069 | 0.2204 |
Leather and footwear | 42,292 | 0.0277 | 0.1642 | 0.0330 | 0.0270 |
Wood, paper and printing | 42,292 | 0.0998 | 0.2997 | 0.1761 | 0.0897 |
Chemicals and petroleum products | 42,292 | 0.0091 | 0.0949 | 0.0449 | 0.0043 |
Rubber and plastics | 42,292 | 0.0111 | 0.1050 | 0.0469 | 0.0064 |
Non-metallic mineral products | 42,292 | 0.0578 | 0.2334 | 0.0415 | 0.0600 |
Metal products | 42,292 | 0.1655 | 0.3716 | 0.1188 | 0.1717 |
Office mach. and elect. equipment | 42,292 | 0.0129 | 0.1130 | 0.0332 | 0.0102 |
Communi./prec./optic./medic. equip. | 42,292 | 0.0053 | 0.0730 | 0.0107 | 0.0046 |
Transport equipment | 42,292 | 0.0101 | 0.1002 | 0.0164 | 0.0093 |
Furniture and n.e.c. | 42,292 | 0.1602 | 0.3668 | 0.1249 | 0.1649 |
Natural persons (self-employed) | 42,292 | 0.9591 | 0.1980 | 0.7300 | 0.9893 |
Non-profit company | 42,292 | 0.0007 | 0.0270 | 0.0028 | 0.0004 |
Private company | 42,292 | 0.0364 | 0.1874 | 0.2557 | 0.0074 |
Foreign company | 42,292 | 0.0001 | 0.0119 | 0.0012 | 0.0000 |
Public company | 42,292 | 0.0008 | 0.0299 | 0.0014 | 0.0008 |
Local gov. company | 42,292 | 0.0005 | 0.0233 | 0.0004 | 0.0005 |
Cooperative | 42,292 | 0.0002 | 0.0153 | 0.0016 | 0.00005 |
Association | 42,292 | 0.0018 | 0.0426 | 0.0064 | 0.0012 |
Enterprise network | 42,292 | 0.1914 | 0.3934 | 0.5907 | 0.1386 |
Market research | 42,292 | 0.0238 | 0.1526 | 0.1111 | 0.0123 |
Accountancy | 42,292 | 0.0794 | 0.2704 | 0.4277 | 0.0334 |
Access to finance | 42,292 | 0.2469 | 0.4312 | 0.3665 | 0.2311 |
Environment | 42,292 | 0.0170 | 0.1295 | 0.1091 | 0.0048 |
Main customer foreign | 42,292 | 0.0056 | 0.0744 | 0.0330 | 0.0019 |
Craft certification | 42,292 | 0.2949 | 0.4560 | 0.3566 | 0.2868 |
Own local HQ | 42,292 | 0.4768 | 0.4994 | 0.5028 | 0.4733 |
Mother company | 42,292 | 0.0342 | 0.1819 | 0.1318 | 0.0213 |
Male manager | 42,292 | 0.7422 | 0.4373 | 0.7624 | 0.7396 |
Log workers | 42,292 | 0.7496 | 0.8394 | 1.8234 | 0.6077 |
(Log workers)2 | 42,292 | 1.2666 | 3.2160 | 5.2440 | 0.7408 |
Log age | 42,292 | 1.7227 | 1.0917 | 2.1795 | 1.6624 |
(Log age)2 | 42,292 | 4.1600 | 3.9232 | 5.7758 | 3.9464 |
Log labor productivity | 41,665 | 8.8584 | 1.0648 | 9.7154 | 8.7315 |
Log capital | 41,647 | 8.1121 | 1.7503 | 10.2519 | 7.8233 |
Log capital/worker | 41,665 | 7.3610 | 1.4038 | 8.4293 | 7.2161 |
Log materials | 41,647 | 8.5995 | 1.6346 | 10.4658 | 8.3469 |
Log materials/worker | 41,665 | 7.8484 | 1.2371 | 8.6417 | 7.7382 |
Appendix 3: Correlation Coefficients
Top panel: among the variables in x1i and x2i | |||||||
|---|---|---|---|---|---|---|---|
Enterprise network | Market research | Accountancy | Access to finance | Environment | Main custom. foreign | Craft certification | |
Enterprise network | 1.0000 | ||||||
Market research | 0.1177*** | 1.0000 | |||||
Accountancy | 0.2661*** | 0.1775*** | 1.0000 | ||||
Access to finance | 0.0947*** | 0.0599*** | 0.0602*** | 1.0000 | |||
Environment | 0.1595*** | 0.1720*** | 0.2434*** | 0.0566*** | 1.0000 | ||
Main custom. foreign | 0.0870*** | 0.0673*** | 0.1517*** | 0.0278*** | 0.1322*** | 1.0000 | |
Craft certification | 0.3655*** | − 0.0023 | − 0.0340*** | 0.0455*** | − 0.0140*** | − 0.0060 | 1.0000 |
Own local | 0.0230*** | 0.0059 | 0.0526*** | 0.0083* | 0.0295*** | 0.0346*** | 0.0337*** |
Mother company | 0.1565*** | 0.1026*** | 0.2383*** | 0.0600*** | 0.1367*** | 0.0836*** | 0.0352*** |
Male manager | 0.0183*** | 0.0131*** | 0.0316*** | − 0.0197*** | 0.0288*** | 0.0049 | 0.0203*** |
Log workers | 0.3162*** | 0.1713*** | 0.5351*** | 0.1179*** | 0.2708*** | 0.1990*** | 0.0271*** |
(Log workers)2 | 0.3015*** | 0.1896*** | 0.5506*** | 0.0903*** | 0.3187*** | 0.2843*** | − 0.0280*** |
Log age | 0.2246*** | 0.0366*** | 0.1682*** | 0.0159*** | 0.0825*** | 0.0633*** | 0.2103*** |
(Log age)2 | 0.2192*** | 0.0426*** | 0.1740*** | − 0.0025 | 0.0907*** | 0.0733*** | 0.2004*** |
Top panel: among the variables in x1i and x2i | |||||||
|---|---|---|---|---|---|---|---|
Own local | Mother company | Male manager | Log workers | (Log workers)2 | Log age | (Log age)2 | |
Enterprise network | |||||||
Market research | |||||||
Accountancy | |||||||
Access to finance | |||||||
Environment | |||||||
Main custom. foreign | |||||||
Craft certification | |||||||
Own local | 1.0000 | ||||||
Mother company | 0.0047 | 1.0000 | |||||
Male manager | 0.0253*** | 0.0099** | 1.0000 | ||||
Log workers | 0.0730*** | 0.2785*** | 0.0622*** | 1.0000 | |||
(Log workers)2 | 0.0849*** | 0.3124*** | 0.0579*** | 0.8588*** | 1.0000 | ||
Log age | 0.2478*** | 0.1051*** | 0.0920*** | 0.2142*** | 0.2081*** | 1.0000 | |
(Log age)2 | 0.2502*** | 0.1091*** | 0.0932*** | 0.2160*** | 0.2295*** | 0.9491*** | 1.0000 |
Bottom panel. Among the variables in x3i and x4i | ||||||||
|---|---|---|---|---|---|---|---|---|
Log R&D intensity | Log training intensity | ICT use | Log labor productivity | Translog TFP | Market research | Main custom. Foreign | Log workers | |
Log R&D intensity | 1.0000 | |||||||
Log Training intensity | 0.3013*** | 1.0000 | ||||||
ICT use | 0.2210*** | 0.3391*** | 1.0000 | |||||
Log labor productivity | 0.1289*** | 0.2079*** | 0.3006*** | 1.0000 | ||||
Translog TFP | 0.0158*** | 0.0304*** | 0.0660*** | 0.6056*** | 1.0000 | |||
Market research | 0.2274*** | 0.2217*** | 0.1878*** | 0.1067*** | 0.0173*** | 1.0000 | ||
Main custom. foreign | 0.0979*** | 0.1156*** | 0.1402*** | 0.0942*** | 0.0061 | 0.0673*** | 1.0000 | |
Log workers | 0.1984*** | 0.2945*** | 0.4823*** | 0.2540*** | − 0.0008 | 0.1713*** | 0.1990*** | 1.0000 |
(Log workers)2 | 0.2396*** | 0.3304*** | 0.4720*** | 0.2746*** | 0.0002 | 0.1896*** | 0.2843*** | 0.8588*** |
Log age | 0.0556*** | 0.0959*** | 0.1554*** | 0.0855*** | 0.0219*** | 0.0366*** | 0.0633*** | 0.2142*** |
(Log age)2 | 0.0596*** | 0.0974*** | 0.1553*** | 0.0667*** | 0.0078 | 0.0426*** | 0.0733*** | 0.2160*** |
Log capital per worker | 0.1163*** | 0.1873*** | 0.2744*** | 0.4392*** | − 0.0036 | 0.0881*** | 0.0729*** | 0.1660*** |
Log mater. per worker | 0.1035*** | 0.1736*** | 0.2407*** | 0.7613*** | − 0.0012 | 0.0887*** | 0.0790*** | 0.2081*** |
Log capital | 0.1885*** | 0.2918*** | 0.4512*** | 0.4754*** | − 0.0033 | 0.1528*** | 0.1543*** | 0.6117*** |
Log material | 0.1805*** | 0.2832*** | 0.4301*** | 0.7092*** | − 0.0014 | 0.1555*** | 0.1623*** | 0.6704*** |
Bottom panel. Among the variables in x3i and x4i | |||||||
|---|---|---|---|---|---|---|---|
(Log workers)2 | Log age | (Log age)2 | Log capital per worker | Log mater. per worker | Log capital | Log material | |
Log R&D intensity | |||||||
Log Training intensity | |||||||
ICT use | |||||||
Log labor productivity | |||||||
Translog TFP | |||||||
Market research | |||||||
Main custom. foreign | |||||||
Log workers | |||||||
(Log workers)2 | 1.0000 | ||||||
Log age | 0.2081*** | 1.0000 | |||||
(Log age)2 | 0.2295*** | 0.9491*** | 1.0000 | ||||
Log capital per worker | 0.2057*** | 0.1147*** | 0.1026*** | 1.0000 | |||
Log mater. per worker | 0.2224*** | 0.0482*** | 0.0349*** | 0.3864*** | 1.0000 | ||
Log capital | 0.5758*** | 0.1946*** | 0.1859*** | 0.8817*** | 0.4095*** | 1.0000 | |
Log material | 0.6089*** | 0.1461*** | 0.1372*** | 0.3780*** | 0.8653*** | 0.6249*** | 1.0000 |
Appendix 4: Estimated Industry-Specific Input Elasticities from Translog and Cobb–Douglas Production Functions
Industry | Materials | Labor | Capital | |||
|---|---|---|---|---|---|---|
Translog (βm) | Cobb–Douglas (βm) | Translog (βl) | Cobb–Douglas (βl) | Translog (βk) | Cobb–Douglas (βk) | |
Food, beverages and tobacco | 0.62*** | 0.60*** | 0.32*** | 0.33*** | 0.14*** | 0.16*** |
Textiles and wearing apparel | 0.51*** | 0.52*** | 0.35*** | 0.41*** | 0.12*** | 0.12*** |
Leather and footwear | 0.62*** | 0.62*** | 0.36** | 0.35*** | 0.10*** | 0.10*** |
Wood, paper and printing | 0.57*** | 0.59*** | 0.36*** | 0.37*** | 0.15*** | 0.14*** |
Chemicals and petroleum product. | 0.61*** | 0.66*** | 0.42** | 0.31*** | 0.15*** | 0.18*** |
Rubber and plastics | 0.54*** | 0.53*** | 0.49*** | 0.52*** | 0.14* | 0.12*** |
Non-metallic mineral products | 0.62*** | 0.64*** | 0.33*** | 0.36*** | 0.06* | 0.07*** |
Metal products | 0.58*** | 0.58*** | 0.41*** | 0.45*** | 0.09** | 0.10*** |
Office machin. and electrical equip. | 0.56*** | 0.62*** | 0.42*** | 0.32*** | 0.16** | 0.18*** |
Communi./prec./optic./medic. equip. | 0.52*** | 0.56*** | 0.41* | 0.41*** | 0.07** | 0.08** |
Transport equipment | 0.51*** | 0.54*** | 0.50*** | 0.49*** | 0.08*** | 0.09*** |
Furniture and n.e.c. | 0.61*** | 0.61*** | 0.37*** | 0.38*** | 0.08*** | 0.08*** |
Totala | 0.58*** | 0.59*** | 0.36*** | 0.39*** | 0.11*** | 0.12*** |
Rights and permissions
About this article
Cite this article
Rodríguez-Moreno, J.A., Rochina-Barrachina, M.E. ICT Use, Investments in R&D and Workers’ Training, Firms’ Productivity and Markups: The Case of Ecuadorian Manufacturing. Eur J Dev Res 31, 1063–1106 (2019). https://doi.org/10.1057/s41287-019-0197-0
Published:
Issue Date:
DOI: https://doi.org/10.1057/s41287-019-0197-0