![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
ABSTRACT
This paper uses the theoretical underpinnings of the real options theory framework to investigate whether uncertainty affects a firms’ decision to directly export versus indirectly export and to determine whether indirect exports may offset the negative effects of uncertainty on export intensity. Using firm-level survey data of 40 emerging countries, we find that uncertainty and firms’ export intensity are negatively correlated. We also find that an increase in uncertainty increases a firms’ willingness to export through trade intermediaries. These results when considered holistically imply that trade intermediaries are able to countervail the negative effect of uncertainty on firms’ export intensity.
Disclosure Statement
No potential conflict of interest was reported by the author(s).
Notes
1 The data that support the findings of this study are openly available in Enterprise Surveys (http://www.enterprisesurveys.org), The World Bank.
2 Since monetary variables are reported in local currency units and span different fiscal years, all data were converted into U.S. dollars using the official exchange rate. They were also deflated to 2009 using the GDP deflator for the United States.
3 These firms come from 9 South and Central American countries (Brazil, Chile, Costa Rica, Ecuador, El Salvador, Guatemala, Guayana, Honduras, Nicaragua), 14 African countries (Algeria, Benin, Egypt, Ethiopia, Kenya, Madagascar, Malawi, Mali, Mauritius, Senegal, South Africa, Tanzania, Uganda, Zambia), 13 Asian countries (Bangladesh, Cambodia, China, India, Indonesia, Kyrgyz Republic, Oman, Philippines, Syrian Arab Republic, Sri Lanka, Tajikistan, Thailand, Vietnam) and 4 Eastern European countries (Lithuania, Moldova, Poland, Turkey).
4 To identify firms’ operating industries, we refer to the broad classification of industries reported in the data. Specifically, we make use of 18 industries of the enlarged manufacturing sector: textiles, leather, garments, agroindustry, food, beverages, metals and machinery, electronics, chemicals and pharmaceutics, construction, wood and furniture, non-metallic and plastic materials, paper, other manufacturing, mining and quarrying, auto and auto components, other transport equipment, other unclassified industries.
5 For instance, differences in the amount and the speed of the distribution of products in the international market can vary a lot among industries because of differences in consumers and producers’ behaviour, as well as because of different national trade obstacles and legal constraints. Thus, we restrict our uncertainty measure to industry and countries specificities. Nonetheless, this measure of uncertainty may also be able to capture both consumer demand and competitive pressures in the international market.
6 It is worth noting that we use both lagged values of time-varying variables (firms age and sales-per-employee) in order to treat regressors as predetermined and time-invariant regressors (the other covariates) to reduce the error variance (Wooldridge Citation2010). See Di Cintio and Grassi (Citation2017) for a similar approach.
7 Unfortunately, the information about years of experience as exporters is available only for few firms and consequently, the number of observations drops substantially.
8 Although our measure of uncertainty is constructed using sales-per-employee, the level of correlation between these two variables is not particularly high (0.139). Hence, we also include sales-per-employee as a regressor to account for how different cash flows may finance export activities. Moreover, the results still hold even if we drop sales-per-employee from the regressions.
9 We construct dummies for small (lower than 51 employees), medium (from 51 to 250 employees) and large firms (more than 250 employees).
10 The dummy is equal to 1 if a firm was established less than five years ago.
11 See footnote 4 for the list of industries included in the analysis.
12 We are aware that the identification of causal effects would require the availability of panel data, and that in the absence of such data (in particular panel data that includes the share of export modes over time), we will only be able to provide suggestive evidence of dependencies among variables. Nevertheless, we believe our results can be of interest to shape future research on this topic.
13 The inclusion of the instrumental variable shrinks the number of observations in each specification.
14 Since we focus on South and Central America, Africa, South and Central Asia, and Eastern Europe, in these countries especially corruption is widespread and may act as a barrier to operate across borders. So we assume that this is an extra-cost that could affect the decision to export or not, irrespective to the decision to export indirectly or directly. In particular, the variable corruption takes the value of 1 if a firm perceives corruption as a very severe obstacle for its current operations, 0 otherwise.
15 To account for sample selection, we also add the IMR to both stages.
16 We estimated also a heckprobit model to verify possible correlation between the likelihood of entering foreign markets and the probability to export through intermediaries. Since heckprobit shows correlation between two equations, we expect a significant mills ratio in fractional regression. Details on the first stage regression and on heckprobit estimation are available upon request.
17 The use of the instrumental variable slightly reduces the number of observations with respect to .
18 Also notice that this preliminary estimation account for sample selection into export market by including the IMR.
19 The use of the exclusion restrictions slightly reduces the number of observations with respect to .
20 We also consider the capital-per-employee and the costs of materials-per-employee to capture differences in the main supply inputs.
21 All regressors in equation A1 are highly significant and with the expected signs. Complete tables and further estimation details are available upon request.
22 Also, with respect to equation A2, all regressors are highly significant considering both the TFP and firm sales. Complete tables are available upon request.