ESTIMATING AND ANALYSING DETERMINANTS OF COST EFFICIENCY OF SAMILL INDUSTRIES IN NIGERIA: A STOCHASTIC FRONTIER APPROACH
OGUNDARI KOLAWOLE
Department für Agrarökonomie und Rurale Entwicklung, Georg- August Universität, Platz der Göttinger Sieben 5, D-37073, Göttingen, Germany.
Email: .
Tel. No. +49551394845. Fax: +495513912177.
Abstract
This paper employed stochastic frontier models to examine cost efficiency and its determinants in Nigeria sawmill industries. A total of 120 mills were sampled in 2006. The result of the empirical analysis shows that, an average cost efficiency of 1.262 was obtained. This, however, suggests that, an average mill from the study, incurs when processing planks from logs, about 26% costs above the frontier cost. Further analysis indicates that significant level of cost inefficiency is present in Nigeria sawmilling industry. Analysis of determinants of cost efficiency shows that, more years of education, experience, age, and level of investment of the mill´s operators reduce cost inefficiencies between 7% and 16%. Economies of scale are found among the mills. The findings, therefore, suggest that measures that will help address policy variables (determinants of cost efficiency) considered in the study needs to be introduced. Among such measures suggested include; policies that will enhance sawmilling skills and promote investments in scale improvements in the sawmilling sector of Nigeria economy.
Key words: Sawmill, cost efficiency, determinants, economies of scale, Nigeria.
1. Introduction
Forest resources increasingly constitute a significant element in the economies of many tropical countries. Nigeria is one example where sawmilling is one of the leading industries in the manufacturing sector (Ajibefun and Daramola, 2004). It employs a significant proportion of the rural population in clerical, and technical jobs such as; power saw operators, saw doctors, and mechanics, as well, as in distribution, and wholesale of the wood products (Alviar, 1983). The industry provides the wood requirement for construction development of residential and commercial buildings, engineering work, home, and office furniture, and road construction. The sawmilling industry contribution to the country’s Gross Domestic Product (GDP) surpassed that from other sectors of the economy in the early 60s except the petroleum industry (Adenuga and Omoluabi, 1975).
The industry also serves as an important source of revenue for government, through fees and royalties paid on licenses given to saw millers for operating the mills. The amount paid for the licenses depends on the type and capacity of machine to operate such mills (Igugu, 1981). According to Aruofor (2000), Nigerian sawmill capacity is estimated at 11, 684, 000m3 per annum in log equivalent while capacity utilisation was 5,422,000m3 per annum. The industry has a few large integrated mills in Nigeria. Mills in the country using their estimated install capacity are distributed between small, medium, and large scale with small scale operated mills dominates over 85% of the mills in the country.
Sawmilling industry is an integral part of micro-enterprises in Nigeria economy. Others include; block-making, carpentry, metal-fabricating enterprises to soap-making, and tailoring, among others. Micro-enterprises are currently taking the lead in the effort to reposition Nigeria for rapid economic growth. This is evidenced in the current macro-economic framework of the Federal Government of Nigeria under the National Economic Empowerment Development Strategy (NEEDs) (Ajibefun, 2007).
The policy blueprint of NEEDS adequately reflected the government philosophy of minimum administrative control of economic activities. Others include; wide scope of free market forces in the economy, a greater role for the private sector, and more emphasis on efficiency and productivity.
One of the reasons often cited by development economist that trade protection brings down the level of industrial sector efficiency is that, absence of foreign competition, allows local firms to enjoy monopoly power and excess profit. The consequence is that domestic producers usually fail to produce at minimum cost. Hence, lack of competitive market generally induces costs to rise above the minimum levels, due to imperfect agency relationship within the firm.
To achieve long-term sustainable development coupled with increase in productivity and profit for the entrepreneurs in Nigerian manufacturing sector under present policy (NEEDs), studies have shown that, resources need to be use more efficiently, with more attention paid to eliminating waste at firm level (Ajibefun and Daramola, 2004 and Ajibefun, 2007).
With regard to sawmilling, efficiency gains in resource allocation at individual mill level have important implications for investment and employment at the national level. This in turn will facilitate technological progress, leading to large shifts in the supply of sawn timber, and greater contribution to national income. It is therefore important to gain good understanding, and knowledge of the current inefficiency inherent in the industry, and the factors responsible for such low efficiency levels.
The challenge of this study, therefore, is the estimation of cost efficiency of saw mill industry in Nigeria. In order to achieve the foregoing end, it is important to understand clearly, the factors that are responsible for efficiency differential (determinants) at individual firm levels with the aims to distil important policy lessons about key constraints to improve production efficiency in this sector.
The outcome of the study is expected to serve as a guide to public policy design and implementation in the country.
2. Theoretical framework
Stochastic frontier models, pioneered by Aigner et al (1977) and Meensen and Van Den Broeck (1977) has been extensively used for modelling, measurement, and estimation of industry, as well as, firm level inefficiency. The novelty of this model lies in the decomposition of observational error into two unobservable stochastic components viz, uncontrollable error representing statistical noise and controllable error representing inefficiency. While the statistical noise (vi) is unrestricted in sign and symmetric about its mean, the inefficiency error (ui) is assumed to be asymmetric and restricted in sign (ui ≥ 1). Consequently, the observed cost of a firm always lies above its stochastic cost frontier. A firm is termed “cost inefficient” when its observed cost is above its stochastic cost frontier otherwise known as minimum cost or optimum cost of production.
A typical stochastic cost frontier specification would be
1
Where Ci denotes the observed total variable costs, represents minimum costs, wi is a vector of variable of input prices, yi is the output of the firm under consideration while c represents functional form. The error is composed of two parts vi and ui earlier defined as .
Cost efficiency (CEi) is defined as the ratio of observed cost (Ci) to the corresponding minimum cost () given the available technology as
≥ 1 2
If Ci = c (yi, wi; b)∙exp{vi}, then firm i is fully efficient and CEi = 1. Otherwise, actual cost exceeds the minimum so CE≥ 1. The higher the value of CE, the more cost inefficient the firm is.
3. Analytical framework
Model specification
For the specification of cost model, we considered a mill which uses 4 inputs viz, labour, materials, electricity, and capital to produce a single output (volume of sawn-wood in cubic meter). If we assumed the mills are in a long run static equilibrium with respect to all input employed and that they minimize total variable costs, a stochastic cost frontier function in equation 1 for the mills can be written as
Ci = f (W1, W2, W3, F, Q) 3
Where, Ci denotes the total variable costs. W1, W2, W3, F are prices of labour, materials, electricity, and capital respectively. Q is the quantity of planks processed from logs (sawn wood) in cubic meters.
The properties of cost function (3) are that; it is concave and linearly homogeneous in input prices, non-decreasing in input prices and non-decreasing in output. Hence, we need a functional form that is locally flexible to maintain these properties. Studies have shown that, translog functional form offers an appropriate flexibility to fulfil the consistency properties in cost function by simply imposed constraints on the homogeneity of the function (Sauer et al. 2006). For that reason, the translog functional form is applied for the study.[I] The trasnlog form of (3) can be written in the following way
4
Through symmetry, it should be noted that,;,,,, ,, ,,,,.
These constraints are usually imposed through estimating a model where the costs and prices of the k-1 inputs are normalized by the price of the k-nth inputs. The normalisation of the costs and input prices is to impose linear homogeneity in input prices. For this study price of labour (w1) was selected to normalize all prices of variable inputs (ws), as well as, the total variable costs by simply divide all the prices and total variable costs by price of labour as shown below
5
The imposed constraints ensure that
; ;; ;
Since, the cost function (5) is homogeneous of degree 1 in variable input prices (ws), the parameter representing the coefficient of price of labour can be calculated as
6
To estimate equation 5 using maximum likelihood method, the composed error term, follows a distributional assumption such that: vi ~iid N (0,σ2v), ui ~iid N (μi,σ2u). vi and ui are distributed independently of each other and of the regressors. The two error terms are proceeded by positive signs, because inefficiencies are always assumed to increase cost. Hence, any error of optimization is taken to translate into higher cost to the producer.
Besides the parameters of cost frontier (5) and the cost efficiency estimates (2), the main objective of this paper is to examine the marginal effects of variables explaining ( determinants) of cost inefficiency of the mills.
To analyze determinants of cost efficiency of the mill operators for the study, we follow the model proposed by Battese and Coelli (1995). Battese and Coelli proposed a model that allows inefficiency to be a function of certain explicative variables whose parameters are estimated jointly with the stochastic frontier. This can explain inefficiency of firms through exogenous variables and which do not form part of the cost frontier functional form. To use this model we assume the distribution of mean cost inefficiency (μi,) is related to the mill operator´s socio-economic variables by allowing heterogeneity in the mean cost inefficiency term to investigate sources of differences in cost efficiencies among the operators as
7
Where Z1, Z2, Z3, Z4 and Z5 are respectively sawmill operator´s number of years of formal education attainment, years of experience in sawmilling business, age of the mill operators, age of the business/mill, and level of investment. The estimated parameters of the factors responsible for efficiency differential among the firms (πs) indicate the direction of the effects that included variables have on inefficiency levels (where a negative parameter suggests that the variable reduces cost inefficiency and vice versa).
The stochastic frontier variance parameters are expressed in terms of and . The lager value of implies that the variance of the inefficiency effects represent larger proportion of the total variance of the terms, u and v. The restriction that equals zero can be tested to determine if stochastic frontier regression is appropriate for the data set.[II]
The marginal effects of variables explaining cost efficiency are calculated from the parameter estimates of equation (2) and (7), using the procedure outlined in Wilson et al. (2001), estimated at their mean values.
The quantification of the marginal effects is possible by partial differentiation of the cost efficiency predictor with respect to the variables in the inefficiency model (Wilson et al. (2001).
The study further ascertains the presence of economies of scale (ES) among the mills. Economics of scale measures the reaction of costs when the output increases proportionally. It is defined as the reduction in cost of production of a given output level while holding all other input prices constant. This measure becomes important when analyzing whether or not it is beneficial to expand production capacity.
ES can be computed from result of equation 5 in the following:
9
The economies of scale exist when ES >1 and conversely diseconomies of scale exist, when ES <1.
The parameters of the cost frontier (5) as well as determinants of cost efficiency (7) were simultaneously estimated using FRONTIER 4.1 program (Coelli, 1996).[III]
4. Data description and sources
This study was based on a survey of sawmill industries in Southwest Nigeria carried out in 2006. The region is made up of 6 states viz, Oyo, Ogun, Osun, Ondo, Ekiti and Lagos states. A total of 43, 40, 38, 23 and 16 mills were sampled in Ogun, Oyo, Ondo, Ekiti and Osun states respectively. The sample is a single period observation consisting 160 mills. The region has a tropical climate with moderate temperature all the year round, which suggests why there are many forest reserves in the region. Most of the wood products in the country are supply from the region as many of the sawmill industries in Nigeria are located in this region. Forest resources from the region include: Iroko, African mahogany, Obeche, Opepe, and African Walnut.
Data were collected via a well structured questionnaire administered by trained enumerators. The descriptive statistics of information collected are presented in Table 1. Total variable costs equals operating expenses which include cost of labour, materials, and electricity. Price of labour is equal to the average annual wages, estimated as labour expenditure divided by the average number of employees. Price of materials is obtained by simply add all costs of materials such as saw-blade, lubricant, and fuel used in course of production. Price of electricity is equivalent to the amount of energy used in kilowatt while depreciation costs of the machines as a proxy for the price of capital.
The demographic characteristics of the mills and mills-owner´s use in the analysis include age of the business, level of investment, age of the operators, years of schooling, as well as, years of experience of the saw millers.
Table 1: Descriptive statistics
Variables / Variables / Mean / SDTotal annual variable cost [N/firm]
Price of labour [N/employees]
price of materials [N/firm]
price of electricity [N/firm]
price of capital [N/firm]
processed logs (sawn-wood) [M3/firm]
Years of Schooling [yrs/firm]
Experience in sawmilling [yrs/firm]
Age of the operators [yrs]
Age of the business [yrs/firm]
Level of investment [N /firm] / Ci
W1
W2
W3
Fi
Qi
Z1
Z2
Z3
Z4
Z5 / 2,301,031
284,534
545,719
176,158
212,209
3,839,679
13.38
11.65
46.50
17.12
2,953,521 / 2,963,634
106,386.1
583,729
132,578
164,734
5,108,790
16.58
7.43
41.93
22.60
5,837,281
Exchange rate as at time of this study: 1US $ = N 125