ABSTRACT
We investigate the association between research and development (R&D) engagement by small and medium enterprises (SMEs) and the institutional environment in Laos. We employed multivariate imputation by chained equations and full Bayesian inference to analyse data from the 2016 and 2018 World Bank Enterprise Surveys for Laos. Our findings show that Lao SMEs that engage in R&D are likely to experience more frequent tax inspections and more solicitation of bribes by government authorities, than Lao SMEs that do not engage in R&D. Firms that perceive political uncertainty as an obstacle to doing business are 28% less likely to engage in R&D activities. These findings raise concerns about the effectiveness of institutions in supporting intellectual property rights in Laos, and have policy implications for promoting productivity of small and medium-sized firms.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. See, for example, Egbetokun et al. (2008) for a study of industry-wide innovation in India, Sivak et al. (2011) for a study of firm-level R&D and innovation in Eastern European countries, de Waldemar (2012) for a study of R&D and innovation among Indian firms, and Sharma and Mitra (2015) for a study of innovation among Indian firms.
2. Lin et al. (2010) conducted a study of R&D investment decisions among Chinese firms. China is a rapidly developing economy that is no longer near the bottom of the world’s wealth distribution. Sivak et al. (2011) included a binary R&D engagement variable in their analysis, but the focus of the study was primarily on innovation incentives rather than innovative input.
3. World Bank, available at https://www.worldbank.org/en/topic/smefinance.
4. SMEs that are not registered under the Value Added Tax (VAT) system are subject to lump-sum tax instead of profit tax. This applies to enterprises which have annual revenue of less than LAK 12 million.
5. Lao PDR: Economic Recovery Challenged by Debt and Rising Prices, available at https://www.worldbank.org/en/news/press-release/2022/05/12/lao-pdr-economic-recovery-challenged-by-debt-and-rising-prices.
6. Extensive review about Lao PDR political leadership is provided by Pratt and Yongvanit (2016).
7. Although there exist officially eight political parties in China compared to the one political party in Laos, all the eight political parties in China are subservient to Chinese Communist Party.
8. Endogenous growth literature (see for example G. M. Grossman & E. Helpman, 1991; Jones, 1995; P. M. Romer, 1986; P. Romer, 1990) emphasises the endogenous technological component achievable through research and development in achieving the long-run economic growth. In a steady state, this means that with the increased input in research and development, the output of that R&D should grow proportionally to input. However, this is not the case in most modern industrialised economies. The research and development output has grown proportionally slower than the inputs have increased (Jones, 1995). Jones and Williams (2000) argue that such underinvestment in research and development compared to the available inputs occurs because of the appropriability problems, knowledge spillovers, creative destruction, and congestion externalities related to the research and development.
9. We used the MICE package in R to perform multiple imputations. See Van Buuren and Groothuis-Oudshoorn (2011).
10. We used the BRMS package in R for Bayesian estimations. See Bürkner (2017).
11. For a vague prior, we applied a prior that conformed to a normal distribution with a mean of zero and a standard deviation of 10, formulated mathematically as , and a Student’s t distribution prior with a location of zero, a scale of 10, and 3 degrees of freedom, formulated as
.
12. The generic weakly informative prior can be formulated as a prior with a mean of zero and a standard deviation of 1, formulated mathematically as (Gelman et al., 2008).
13. which is equivalent to Gelman et al. (2008)’s
prior.
14. We used a weakly informative normal distribution prior with a standard deviation of 1 and mean of 1 or−1, depending on the expected association.