Determinants of Labor Productivity in Malian Firms

This article identifies the determinants of labor productivity of Malian firms using 2007-2015 panel data. These data were collected by the World Bank in 2016 from industrial companies and services. The Hausman test made it possible to choose the composite error panel model for the estimation. The results indicate that there is a positive correlation between capital intensity, commodity imports and labor productivity. Labor productivity decreases with the number of years the manager has spent at the head of the company, the use of new information and communication technologies, and the purchase of fixed assets. We also find that formal firms seem to have higher productivity when confronted with stronger local informal competition. Overall, 79.5 percent of companies reported power outages. Power outages have a negative impact on labor productivity.


Introduction
This article enriches the empirical literature on labor productivity in Mali by using information from the World Bank's 2016 business survey in Mali. Such a survey provides rich data, directly from companies, with a detailed breakdown by sector that is not available in existing statistics. Using this data set, the evolution of the firm's labor productivity in relation to the determinants is examined in a multivariate empirical framework that takes into account the number of years of existence of the firm, i.e., since its creation, the number of years of the director at the head of the firm, the capital intensity, imports of raw materials, power cuts, the use of ICT, purchases of fixed assets and unfair competition following the presence of informal or unregistered firms.
The literature is quite extensive on the determinants of labor productivity. There are several country-specific case studies that explore the determinants of labor productivity growth in a particular country (Aslam, 1983;De Fuentes et al., 2016;Dias, 1991;Fallahi et al., 2010;Mahmood, 2015;Nto et al., 2012;Velucchi and Viviani, 2011). Others have been interested in explaining the productivity differential between countries or regions (Belorgey et al., 2006;Bourles and Cette, 2007;Choudhry, 2009;Gust and Marquez, 2004;Mefford, 1986). In this study, we focus not only on the role of ICT in labor productivity growth but also on other socio-economic indicators that are essential for labor productivity growth. The "power cuts" factor is largely ignored in the literature, but is fundamental in the case of the Mali study.
The objective of this article is to identify the potential determinants of labor productivity for the different companies in Mali belonging to different sectors (manufacturing and service). To our knowledge, this is the first study to empirically examine the determinants of labor productivity in Mali using survey data to fill gaps imposed by severe data limitations, both at the micro and macroeconomic levels.
We find that capitalistic intensity, imports of raw materials and unfair competition have a positive influence on labor productivity. On the other hand, we observe a negative correlation between the number of years of the director at the head of the company, power cuts, the use of NICT by the company, the purchase of fixed assets.
The rest of the document is organized as follows. Section 2 provides a brief review of the literature on the main determinants of productivity. Section 3 deals with methodology. Section 4 presents the productivity results and their discussions. Section 5 concludes.
In Bangladesh, Islam (1990) analyses the determinants of labor productivity in two stages in the cotton textile industry. In the first step, aggregated, he finds that the size of the firm, the number of hours/persons lost compared to the actual number of hours worked and the man/capital ratio together explain 80.25 percent of the variation in labor productivity. The second step deals with labor productivity at the inter-company level. At this level, labor productivity change is associated with differential employment structures relative to work, capital equipment staffing, production management, preventive maintenance and capacity utilization.
For the year 2005-2006, Mahmood (2015) is working on 229 manufacturing industries in Pakistan. Using capital intensity as an independent variable and advertising expenditures as a mediator variable, he estimates the total direct and indirect effects on labor productivity and finds that about 18 percent of the total effects on labor productivity are attributable to advertising expenditures. Fallahi et al. (2010), examines the determinants of labor productivity at the firm level in the Iranian manufacturing sector. The analysis is based on descriptive statistics and cross-sectional regression models on a sample of 12,299 industrial companies. The results show that labor productivity is positively related to wages, fixed capital per employee, export orientation, R&D activities and labor force training. Nto et al. (2012), study the productivity determinants of 120 Nigerian manufacturing firms using cross-sectional data. The results indicate that at the 1 percent threshold, unskilled labor, raw material costs and net productive assets have a positive impact on productivity. Shinkai and Hossain (2011), analyze the productivity and performance of the IT sector in Bangladesh based on a survey of 202 companies of different sizes. They find that e-governance activities contribute to the total productivity of companies more than five years old. The presence of foreign executives in industries positively influences their productivity and performance. Kim and Loayza (2017), construct indices representing each major determinant of productivity as a linear combination of representative indicators and assess the relative contribution of the indices to productivity change in 65 countries over the period 1985-2011. The results show that productivity change is more sensitive to physical infrastructure, education, market efficiency, innovation and institutional infrastructure. The overall determinants index has a positive relationship with productivity growth. Frijters (1999) on South Africa finds that education has a very small effect on productivity.
Ortega and Marchante (2010), study the impact of temporary contracts on labor productivity in Spain over the period 1987-2000. The results indicate a slowdown in labor productivity growth due to the use of temporary contracts in manufacturing and energy industries. Giang et al. (2019), focused their study on total factor productivity and its determinants for 420 companies active in Vietnamese agriculture. They are broken down into very small, small, medium and large companies. Large companies have a positive productivity unlike small and very small companies. They find that foreign participation, export, access to bank loans and the Internet have a positive and significant influence on total factor productivity. The State's shareholding acts negatively. Decker et al. (2009), examines the determinants of labor productivity in 48 states over the period 1989-2000. Earle and Gehlbach (2014), demonstrated, based on panel data on 7,000 Ukrainian manufacturing companies, that political patronage in a context of weak institutions can have a significant redistributive impact on economic productivity. Innovation has a positive impact on the labor productivity of Ukrainian companies (Kostenko, 2014).
However, so far, no attempt has been made to study the systematic relationship between labor productivity and a relatively broad set of characteristics of Malian firms. Therefore, we aim to fill at least part of this gap in the literature. Our study is based on data at the level of Malian companies for the period 2007-2015. This allows us to assess the determinants of productivity in the manufacturing and service sectors.

Methodology and Methodology
To examine the determinants of productivity of Malian firms over the period, we use the Cobb-Douglas production function: Where Y, K, L are respectively production, capital and labor, α and β are elasticity parameters expressing the substitutability of inputs with each other.
Referring to studies by Nelson and Phelps (1966), Lucas (1988), (Corvers, 1997) and Niringiye et al. (2010), which extend and modify equation (1) to integrate individual effects i and longitudinal effects t, we have the following: Where Y, K are defined as previously in (1) and L* represents effective units of effective work. We can then write an equation for the efficiency of the work units as follows: Where Li is the number of employees of company i and r r L  is the number of employees whose level of education r = 1, 2 and 3 is respectively lower, intermediate and upper. The parameter θ r reflects the contribution of the respective levels of education to the units of work efficiency.
By substituting of equation (3) for its expression in equation (2) and dividing by the work performance (L), we have: According to equation (4), the level of labor productivity depends on the relative shares of the three levels of education in the firm's labor force.
Equation (4) can be used to calculate the dynamic effects of human capital on labor productivity. We modify and extend Corvers (1997) model by adding other human capital variables to it in order to minimize overestimation of the relative importance of the components of human capital that were originally specified in the original model. We therefore include the weighted average intensity of education and skills. So we have the following model: Applying the logarithm to equation 5, we have:

Data and Definition of Variables
The data used in this study are those of the World Bank-Mali. They are obtained from a survey of Malian companies in 2016. The survey covered companies located in the Bamako District, the Segou region, the Mopti region in the center of the country and the Sikasso region in the south. We use non-cylindrical panel data for the period 2007-2015. This choice is dictated by the fact that we want to have recent and original data. Mali is a weakly industrialized country. There are 25 companies that meet our objective.
These data provide a lot of information on manufacturing industries and service companies. They include the firm's output that has been related to the number of its employees to obtain the labor productivity that is the modeldependent variable.
The Director, through his experience or the number of years at the head of the company, must be able to better understand the workers and encourage them to work more. Capitalistic intensity is a factor that favors the company's production. We also assume that the number of years of existence of the company, i.e. from its creation to the date of the survey, should allow for a better integration into the sector. Imports of quality raw materials are also a factor that stimulates labor productivity. To date, Mali remains a country whose population as well as entrepreneurs suffer enormously from power cuts. While in most studies on the determinants of productivity, the absence of this variable seems obvious, this is far from being the case in Mali. For the study period, 79.5 percent of companies experienced power outages. The practice of the Internet is, in principle, observed throughout the production chain. The company buys fixed assets (cars, land, etc.) that are intended to support its production even if it is competing with informal companies.   Table 3 shows the gross correlation coefficients. The values of the correlation coefficients (r2) show that the firm's labor productivity is negatively associated with the age of the firm (r2 = -0.27), the power outage experienced by firms (r2 = -0.36), the use of NICTs (r2 = -0.39), and the purchases of fixed assets by firms (r2 = -0.14). On the other hand, the firm's labor productivity is positively associated with capital (r2 = 0.23) and the value of raw material imports (we have r2 = 0.23).

Results and Discussions
The Hausman test refutes the hypothesis that there is no correlation between the random term ui and the explanatory variables of the model. (P-value= 28.07 percent greater than 5 percent) with the Chi-2 test at 8 degrees of freedom. This means that the estimators of the fixed-error model are biased. It is preferable to retain those in the compound effects model that are unbiased.
We estimate the equation using the composite error panel model. The results are recorded in the table below. The number of years of the director at the head of the company or the director's experience has a negative influence on labor productivity. The 1 percent increase in the number of years of the director at the head of the company leads to a 0.55 percent decrease in labor productivity. This is explained by the fact that if the director lasts in his position, he becomes animated by a lack of spirit of innovation for having used all his know-how and a rupture of understanding between him and the workers.
There is a positive and significant correlation between capital and labor productivity. If the company's capital increases by 1 percent, labor productivity increases by 0.28 percent. These results confirm those of Dias (1991) on Sri Lanka Papadogonas and Voulgaris (2005), on Greece andFallahi et al. (2010) on Iran andCieślik et al. (2017) on Ukraine. Capital accumulation is seen as a means of increasing business productivity when combined with an appropriate mix of effective labor force. Macroeconomic theory therefore attaches great importance to investment demand, i.e. the demand for plants and equipment.
Imports of raw materials have a positive and significant impact on labor productivity at the 1 percent threshold. The 1 percent increase in the amount of imports of raw materials leads to an increase in labor productivity of 0.64 percent. This is due to the fact that raw materials are essential and fundamental in the production system. The positive labor productivity can also be explained by the quality of imported raw materials. These results confirm those of Ding et al. (2016) for China and Cieślik et al. (2017) for Ukraine.
Power cuts have a negative and significant impact on labor productivity at the reasonable threshold of 1 percent. In other words, 1.90 percent is the decrease in labor productivity induced by the increase in power outages of a unit. The power outage has negative effects on the company's production activity and prevents it from honoring its commitments (with suppliers, customers, etc.). It extinguishes the motivation to work among the workers by making them lose the beginning of a momentum. These effects are immeasurable. Electricity is a major concern for companies. 81.5 percent of them located in the Bamako District reported being victims of untimely power cuts that caused a temporary shutdown of production activity. 78.4 percent of those located in the Ségou, Mopti and Sikasso regions reported experiencing power cuts during the study period. Power outages resulted in an average loss as a percentage of sales of 11.08.
There is a negative correlation between NICT and labor productivity. The 1 percent increase in the use of NICT would lead to a 0.93 percent decrease in the company's labor productivity. Malian companies are not sufficiently connected to NICT. This is corroborated by the statistics, 64.1 percent of them are not connected. The use of NICT increases business productivity by increasing labor productivity. These results refute this assertion by Steffen and Stephan (2008) and Kılıçaslan et al. (2017).
The purchase of new or used fixed assets has a negative and significant impact on labor productivity. The 1.74 percent drop in labor productivity is due to a 1 percent increase in the purchase of fixed assets. Malian companies do not pay for good quality fixed assets or that they are certainly diverted from their functions.
The presence of informal or unregistered competitors in the country has a positive and significant impact on labor productivity. Formal enterprises seem to have higher productivity when confronted with stronger local informal competition. Formal enterprises facing low competition from the informal sector are certainly protected by regulations (labor regulations or state interventions). Also, due to the importance of the presence of informal competitors in the country, formal firms may not distinguish between sources of competition. In both situations, companies, in order to increase their competitiveness, strive to be more efficient and productive. These results confirm those of Ali and Najman (2016).

Conclusion
The study makes it possible to lay the foundations of the empirical and economic literature on labor productivity in the case of companies in Mali. Our results show that capital investment and the purchase of raw materials induce labor productivity growth. The use of NICT leads to a decrease in labor productivity. The study indicates that 64.1 percent of companies do not use NICT. Companies must train themselves by developing the use of ICT and make the information system a tool for employees' work. The purchase of fixed assets has a negative impact on productivity. Also, power cuts are a real brake on production. They have a negative impact on firms' labor productivity and cause an average loss, as a percentage of sales, of 11.8. Minimizing productivity losses caused by power outages requires improving the quality of electricity service.