Labour-use Efficiency of Assam Gramin Vikash Bank: Branch- and District-level Analysis

Abstract

The present study estimated labour-use efficiency of 48 branches of Assam Gramin Vikash Bank at its branch level, covering three districts of Barak Valley, which falls under Silchar region of the bank for the time period from 2010–2011 to 2017–2018. The study applied data envelopment analysis for estimating labour-use efficiency. In the second stage, the study applied censored Tobit regression for determining the impact of several contextual variables on efficiency. The study reveals that the mean labour-use efficiency score of the selected branches is 76% when averaged for the in-sample branches over the observation period. Results of the Tobit regression identified cluster 2 and total business of the branches as the significant factors for determining efficiency and the number of employees as a significant variable influencing inefficiency.

JEL Classifications: G2, G20, G21, J3

Keywords

Microeconomics producer choice rational choice and individual decision-making social choice

Introduction

In the 1970s and 1980s of the preceding millennium, Indian banking expanded rapidly with the goal of promoting mass banking. This policy was an integral part of the governmental objective of poverty alleviation, and the commercial banking system in the public sector was supposed to play a pivotal role in the process of goal achievement. Inter alia, the banking system tried to inculcate banking habits and replace rural money lenders by institutional credit delivery mechanism. The growth in rural sector lending was facilitated by the inclusion of agricultural credit in the priority sector lending mechanism. However, the need for having specialized institutions for promoting inclusive banking in the rural sector was felt. Thus, Regional Rural Banks (RRBs) were introduced as per the recommendation of the Narsimham Committee (Government of India, 1975). RRBs were established through the promulgation of RRB Act, 1976, involving joint shareholding of the central government, respective state government and the lead (sponsor) bank.

In the subsequent decades, RRBs have played a commendable role in the establishment of institutional financial intermediation in rural India in terms of branch network, coverage of rural households and the volume of business. Since the 1990s, the banking scenario in the Indian economy underwent major structural changes, involving greater degree of competition with the advent of microfinance, small finance banks and payment banks’ (Chadha, 2017; Kumar & Gulati, 2014) adoption of prudential operational norms, increased monitoring by the regulator and widespread use of information technology. In view of the above, rational utilization of labour force has become absolutely essential for the banking sector. This is because the cost of labour occupies a significant portion in every organization, and RRBs are no exception. Unfortunately, there are very few studies devoting their attention to the labour-use efficiency (LUE) of organizations. The literature on bank efficiency (Al-Jarrah, 2007; Ikhide, 2008; Ray, 2014; Sinha, 2006) has generally considered cost-efficiency of the bank or its branches. Only handful of efficiency studies (Das et al., 2009; Herwadkar et al., 2019; Jaffry et al., 2008; Shanmugam & Lakmanasamy, 2000) has estimated LUE of the banking sector. Further, there is no study of LUE relative to the rural banks. The present study intends to fill this gap. The study adopts a two-stage approach. The first-stage exercise finds out the level of LUE of the in-sample branches when measured against a global frontier (constructed from the data from all the selected branches under study) as well as against a local frontier (comprising branches within the district). In the second stage, we have estimated the influence of several contextual variables using Tobit regression. The paper is organized into five sections and proceeds as follows. The first section provides an overview of Assam Gramin Vikash Bank (AGVB). The second section includes the related research studies. The third section discusses the methodology. The fourth section contains a description of data, variables and results. The fifth section concludes the article.

Assam Gramin Vikash Bank: An Overview

The motive of RRBs is to uplift the rural masses as they are demonstrating a great impact on the lives of the rural people. The economy of Assam is mainly agriculture based as majority of the population of the state depend on agricultural income. According to the Indian census of 1971, out of the total workforce of Assam, 55.86% of the workers were cultivators, and 9.92% were agricultural labourers. The overwhelming majority of 91.18% of the population of Assam lives in rural areas, while only 8.82% lives in urban areas. Before the inception of rural banks in Assam, the primary source of financial lending was the private moneylenders. The establishment of a regional banking system was made possible after the introduction of RRBs in 1975 throughout the country.

AGVB was established in 2006 by merging four rural banks and is currently the only RRB in Assam. Because of the rural orientation, AGVB plays an important role in the upliftment of rural livelihood (Table 1). Various schemes of the bank like National Rural Livelihood Mission (NRLM), Pradhan Mantri Employment Generation Programme, Guaranteed Emergency Credit Line, agricultural loans, self-help group loan, etc., boost up the rural economy of the state.

Table 1.

Role of AGVB Towards Rural Livelihood for the Financial Year 2019–2020.

Sl. No.	Particulars	Amount (₹ in thousand)
1	Agriculture and allied sector	6,905,575
2	Small-scale industries	1,922,525
3	Self-help group (SHG)	2,512,985
4	Kishan credit card (KCC)	687,7021
5	Joint liability group (JLG)	5,828

Source: Bank Financial Report.

Efficiency in Labour Use: Related Research Studies

Most of the banking efficiency literature paid attention to bank-level cost-efficiency. In the present study, a brief review on LUE in banking sector is presented.

Battese et al. (2000) studied LUE in the Swedish banking sector with the help of stochastic frontier approach for the period from 1984 to 1995. The technical inefficiencies of labour use of Swedish banks were found to be significant, with average inefficiencies varying between 8% and 15% during the in-sample period. The inefficiency level was influenced by bank category, number of branches, total inventories and year of observation.

Shanmugam and Lakmanasamy (2000) estimated the LUE of the Indian banking sector for the period of 1999, using random coefficient frontier approach. The findings of their study indicated that the Indian banking sector used labour inputs less efficiently in terms of interest margin. Further, significant variations were observed in the mean efficiency score of labour use in terms of producing profit in different banking groups.

Das et al. (2009) analysed the LUE of a large public sector bank in India at its branch level, using data envelopment analysis (DEA). The study showed that efficiency can be increased by reduction in labour cost through downsizing in the staff strength, especially in clerical cadre, followed by subordinate staff. Moreover, the study also identified some efficient branches, which can be gainfully merged with other branches wherever possible.

Jaffry et al. (2008) applied flexible translog model for evaluating the LUE of 114 Indian and Pakistani commercial banks for the period from 1985 to 2003. The study found that there is scope for reduction in labour use to the tune of 34.7% relative to the frontier. The study confirmed an inter-temporal improvement in LUE for the entire subcontinent. Further, the foreign banks were the most efficient in terms of labour use, followed by private domestic and public domestic banks, respectively.

Ray et al. (2017) examined the labour cost-efficiency of Indian bank branches. They conducted the study on 536 branches of a major public sector bank for the period from 2007 to 2008. The result of their study showed that there exists significant labour cost-inefficiency in the operations of the selected branches. Moreover, the study found that the Chennai-based branches were most efficient than the other regions, while the Kolkata-based branches were least efficient. For reduction in labour cost for boosting up in LUE, their study suggested a reduction in the number of clerical staff.

Herwadkar et al. (2019) studied the labour cost-efficiency of Indian banks using non-parametric DEA methodology for the period from 2005 to 2018. The inputs and outputs for their study were selected using the production approach. The findings of their study revealed that the labour cost-efficiency of Indian banks has not improved during the study period. Further, they added that public sector banks are relatively better than the private and foreign banks. They concluded that larger banks being labour cost-efficient in relation to its counterparts, owing to economies of scale, lead to better efficiency outcome. The study found out the rationale for recent merger of banks and prescribed bank consolidation to reap the benefits from economies of scale.

The Methodology

Labour-use Efficiency

The performance of a productive unit is often expressed in terms of efficiency. A firm is efficient if no room for (a) reducing input usage for producing the given level of output or (b) expanding the level of output, given the current level of input. Thus, estimation of efficiency requires a comparison of observed performance with benchmark/best practice performance either from the input perspective (for a given level of output) or from the output perspective (for a given level of input usage).

The present study considers input-oriented efficiency in the short run. In the absence of price information, the input-oriented efficiency of a productive unit is determined by a comparison of observed input usage with benchmark input usage. However, in the presence of price data, we can measure cost-efficiency, which is derived by comparing the best practice cost with observed cost. The cost-efficiency was thus derived into a product of input-oriented technical efficiency and allocative efficiency for the multi-input case.

While conceptualizing efficiency, a distinction has to be made between short-run and long-run efficiency. In the long run, all inputs are variable. However, in the short run, some inputs (e.g., capital) are fixed, while some (e.g., labour) are variable. LUE corresponds to short-run efficiency of banks. At the branch level, labour-use inefficiency (LUIE) is present from the labour cost perspective, the respective branch is operating below the best practice frontier, that is, its labour cost is above the cost implied by the best practice level of cost. It refers to the extent to which individual branches can reduce labour cost, while still being able to produce the same level of output as that of the efficient branch. It requires a reallocation of the total labour cost of a bank over all of its branches with the objective of labour cost-minimization. Then, labour cost-efficiency of the branch can be obtained by comparing the actual/observed labour cost with that of the minimum cost. A branch will be considered as efficient if it can produce the given/target output bundle with the minimum labour cost without compromising with the service standard.

Efficiency Measurement Techniques

Inter alia, there are three major approaches for judging the LUE of the observed firms: the ratio approach, the econometric approach and the mathematical programming approach. In the ratio approach, business per employee or profit per employee can be good indicators of labour-use performance. However, these are partial measures of labour use. An efficiency measurement technique, which relies on the production function, is much more appropriate than the ratio-based approaches.

In the econometric approach, stochastic frontier analysis (SFA) can be used to estimate the frontier level of labour cost. Observed labour cost can deviate in the upward direction (relative to the frontier) because of stochastic deviations. However, SFA requires a parametric specification of labour cost function with the accompanying risk of possible misspecification of the labour cost function.

In the mathematical programming approach, inter alia, DEA or Free Disposal Hull (FDH) methods can be used, both of which are non-parametric approaches for measuring production/cost-efficiency. In comparison to SFA, DEA and FDH are more flexible as no such specification is needed. Further, the advantage of choosing DEA or FDH over SFA is that fewer assumptions are required in these two approaches as compared to SFA, and multiple inputs as well as outputs can be accommodated. However, the absence of convexity in case of FDH results in overestimation of performance in many cases.

In the present context, we have used DEA for estimating LUE. DEA was originally introduced for measuring technical efficiency of decision-making units (DMUs) in the absence of price data. In the context of a convex technology, DEA constructs a production/economic frontier, which can act as the reference technology for evaluating the performance of in-sample DMUs. Charnes et al. (1978) introduced DEA in the context of constant returns to scale (as the global technology). Banker et al. (1984) extended the analysis in the context of local technology (i.e., technology exhibiting variable returns to scale). The DEA approach can also be used to estimate the cost frontier, which, in turn, can be used as the benchmark for estimating cost-efficiency.

Estimation of Labour-use Efficiency Using Decision-making Unit

For explaining the methodology involved in the estimation of LUE, we consider an industry comprising N firms. The firms use n inputs to produce m outputs. The input vector is denoted by x (x₁, …, x) and the output vector by y (y₁, …, y). The technology set can be specified as

T = \{(x, y) : x can produce y\}

(1)

The technology faced by the firms in the industry can be described by the following production possibility set T = {(x, y): x can produce y}

The inputs and outputs satisfy free disposability. Convex combinations of observed inputs and outputs are also feasible.

For any specific output bundle y⁰, the input requirement set consists of all input bundles x that can produce y⁰ and can be specified as

V (y^{0}) = {x : (x, y^{0}) \in T}

(2)

For the given output bundle of the observed firm (y⁰), firm’s objective is to produce the target output bundle at the minimum cost.

Suppose the input price vector of the firm is w. Then, its actual cost is Then, the minimum cost of producing the target output is

C (w, y^{0}) = \min w^{'} x : x \in V (y^{0})

(3)

When the price vector is given, the inputs are the choice variables for the firm for the purpose of cost-minimization. However, this is possible in the long run where a firm can vary all of its inputs to achieve the fullest level of efficiency. In the short run, however, some inputs are quasi-fixed and only the other inputs are subject to variation. One needs to modify the relevant efficiency measure in order to take explicit account of the quasi-fixed inputs. Let us consider the input vector x to be partitioned as x = {v, K}, where v is an n₁ element vector of variable inputs, while K is an n₂ element vector of quasi-fixed inputs. We may define the conditional input requirement set for output y⁰, given the quasi-fixed input K⁰ is

V (y^{0} | K^{0}) = {v : (v, K^{0}) \in V (y^{0})

(4)

Let us now consider the short-run cost-efficiency of a firm. For example, the input price vectors are q and r for the variable and fixed inputs, respectively. The actual variable cost of the firm is $V C^{0} = q^{'} v^{0}$ and its fixed cost is $F C^{0} = r^{'} K^{0}$ . It is assumed that the fixed cost is a constant in the short run and plays no role in its cost-minimization. Thus, it is the variable cost, which ultimately determined the cost-minimization of the firm. Hence, the minimum variable cost of the firm is

VC (q, v, K^{0}) = \min q^{'} v : (v, K^{0}) \in V (y^{0})

(5)

The equation of cost-minimization (variable) in the DEA model is as follows:

{VC}^{*} = \min q^{'} v

(6)

\begin{matrix} s.t. \sum_{j = 1}^{N} λ_{j} y^{j} \geq y^{0}, \sum_{j = 1}^{N} λ_{j} v^{j} \leq v \\ \sum_{j = 1}^{N} λ_{j} K^{j} \leq K^{0}, \sum_{j = 1}^{N} λ_{j} = 1, λ_{j} \geq 0 : (j = 1, 2, ..., N) \end{matrix}

The variable cost efficiency of the firm is measured as

E_{VC} = V C (q, K^{0}, Y^{0}) / {VC}^{0}

(7)

Equation (6) not only estimates the minimum cost but also provides the optimum inputs, which can be compared with the actual inputs to determine to what extent a firm is over-/underutilizing its inputs.

In the present study of LUE, a comparison of the existing quantity of labour input L⁰ with the corresponding optimal quantity of labour L^* reveals whether a branch is using over-/underutilizing the labour. For the labour input I, define the ratio

α_{i} = \frac{L_{i}^{0}}{L_{i}^{*}}

(8)

The value of α greater/less than 1 implies over-/underutilization of the respective input. For a branch facing labour input price (vector) w using the labour input bundle L⁰ and capital input K⁰ to produce output y⁰, a measure of its LUE is

L_{e} = \frac{V C (w, K^{0}, y^{0})}{w L^{0}}

(9)

Determinants of Labour-use Efficiency: Impact of Contextual Variables

The efficiency level obtained by an observed DMU is influenced by several environmental/contextual variable, which are not included in the input–output framework. In order to estimate the impact of such efficiency determination process, the efficiency scores as obtained from the non-parametric estimation process are regressed on the relevant explanatory variables. However, since the DEA efficiency score lies in the interval 0 and 1, the dependent variable (efficiency score) is a limited dependent variable. Therefore, it is reasonable to use the Tobit regression model, which is a censored regression model, which is applied in cases where the dependent variable is censored in either left or right direction or in both the directions.

Data, Variables and Results

Description of Variables

This paper makes use of production approach for selection of input and output. As the present study is confined to LUE of the selected branches of AGVB, different categories of labour are considered as inputs, viz. number of officers, number of clerks and number of sub-staff. Branch-level overhead is taken as the capital, which is considered as fixed in the short run. The outputs included in the present study are deposit, advance and non-interest income. The input costs are salary of officers, salary of clerks and salary of sub-staff. Estimation is made under variable returns to scale.

Description of Data

The paper makes use of data pertaining to AGVB for 8 years, ranging from 2011–2012 to 2018–2019. The selection of the Silchar region out of the eight regions of the bank is carried out on the basis of convenience. As the bank entered into Core Banking Service platform in November 2011, the branch-level data for this year was not available in the system. Hence, data for the year 2011 are collected from the individual branch level. It was not possible to visit each and every branch of the bank individually; hence, the study area is limited to Silchar region only. The main reason behind selection of the period is the availability of data. For estimation purposes, the present study considered 44 branches (out of the 48 branches) which are operational in the southern part of Assam, that is, Barak Valley spread over the three districts of Cachar, Hailakandi and Karimganj and remained in existence for the entire study period. The study, thus, includes 19 branches under Cachar district, 8 branches under Hailakandi district and 17 branches under Karimganj district. Following Das et al. (2009), cluster-wise analysis has been made (Table 2).

Table 2.

Clustering of Branches.

Cluster Number (s)	District Name	Number of Branches
1	Cachar	19
2	Hailakandi	8
3	Karimganj	17

Source: The data on inputs and outputs have been obtained from secondary sources like the annual report of AGVB, published financial statements, etc.

Year-wise Labour-use Efficiency Scores of the In-sample Branches (2011–2018)

In the present study, LUE has been estimated on a stand-alone basis, that is, cost frontiers are determined separately for each observed year. However, the cost-efficiency does indicate the extent of variability in performance, and, thus, efficiency scores are inter-temporally comparable in a limited sense.

Table 3 exhibits the year-wise LUE scores of the selected branches. It can be seen that the mean LUE of all the branches is 76% with a standard deviation of 15%. This implies that the branches can produce the same level of output by reducing its labour cost to the tune of 24% (1–0.76) to be in efficient frontier. Thus, the overall LUIE of branches came out to be almost 24%. The branches exhibit a fluctuating trend in LUE for the study period, ranging from 2011 to 2018. Further, the year 2017 is marked by least efficiency score, whereas 2018 displays highest efficiency score. The e fficiency score ranges from 0.67 to 0.96. The highest efficiency score in 2018 may be due to the reduction in labour cost as the number of employees reduced due to more number of retirement/relinquishment of the existing staff as compared to a new recruitment drive.

Table 3.

Year-wise Branch-level LUE Scores.

Particulars	2011	2012	2013	2014	2015	2016	2017	2018	Mean (2011–2018)
Mean LUE	0.80	0.82	0.74	0.70	0.69	0.75	0.67	0.96	0.76
Standard deviation	0.14	0.15	0.19	0.15	0.16	0.16	0.19	0.08	0.15
Maximum LUE	1	1	1	1	1	1	1	1	1
Minimum LUE	0.54	0.47	0.42	0.48	0.41	0.52	0.43	0.74	0.50

Source: The authors.

Intra-district Labour-use Efficiency Scores of Branches Under Cachar District (2011–2018)

Table 4 presents the LUE scores of branches under Cachar district. It is seen that the mean LUE score of all the branches under Cachar district is 83%. Year-wise comparison reveals a fluctuating trend for the study period. The branch with ID 1 is the most efficient branch, whereas branch ID 14 is the least efficient branch in Cachar district.

Table 4.

Intra-district LUE Scores of Branches Under Cachar District.

Branch ID	2011	2012	2013	2014	2015	2016	2017	2018	Mean
1	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
2	0.878	0.932	1.000	0.963	0.596	0.751	1.000	1.000	0.890
3	0.834	0.900	1.000	0.717	0.667	0.992	0.592	0.786	0.811
4	0.780	0.785	1.000	0.850	0.635	0.773	0.646	1.000	0.808
5	0.802	0.793	0.989	0.812	1.000	0.934	0.601	0.805	0.842
6	0.890	0.944	1.000	0.831	0.659	0.788	0.705	1.000	0.852
7	0.840	0.794	0.946	0.806	0.754	0.785	0.803	0.863	0.824
8	0.763	1.000	1.000	0.923	0.703	0.921	0.582	1.000	0.861
9	0.685	0.779	1.000	0.935	0.866	0.710	0.680	0.085	0.717
10	0.820	0.957	1.000	0.655	0.895	0.850	0.733	1.000	0.864
11	1.000	1.000	1.000	1.000	0.697	0.841	0.612	1.000	0.894
12	0.936	0.919	0.982	0.867	0.577	0.663	0.565	0.913	0.803
13	0.679	0.610	0.898	0.884	0.552	0.881	0.500	1.000	0.751
14	0.809	0.768	0.949	0.745	0.727	0.703	0.509	0.339	0.694
15	0.668	0.721	1.000	0.886	0.872	0.782	1.000	0.946	0.859
16	0.859	0.851	0.991	0.648	0.618	0.854	0.498	1.000	0.790
17	0.579	0.479	0.828	0.578	0.568	0.945	0.711	0.990	0.710
18	0.938	1.000	1.000	1.000	1.000	1.000	1.000	0.997	0.992
19	1.000	0.869	0.869	0.774	0.572	1.000	0.800	1.000	0.860
Mean of 19 branches	0.830	0.847	0.971	0.835	0.735	0.851	0.712	0.880	0.833

Source: The authors.

Intra-district Labour-use Efficiency Scores of Branches Under Hailakandi District (2011–2018)

Table 5 provides the LUE scores of branches falling under Hailakandi district. The mean LUE score of all the branches under Hailakandi district is 92%. Year-wise mean LUE score depicts that branches are near the efficiency frontier as the efficiency score ranges from 87% to 96%. Further, branches under Hailakandi district reveal the fact that except the branch IDs 22 and 23, all other units were in efficient frontier in one or the other years of the study period. Moreover, on an average, branch ID 20 is the best performing branch in terms of labour use, and branch ID 23 is the worst performing branch.

Table 5.

Intra-district LUE Scores of Branches Under Hailakandi District.

Branch ID	2011	2012	2013	2014	2015	2016	2017	2018	Mean
20	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
21	1.000	1.000	1.000	1.000	1.000	1.000	0.725	1.000	0.966
22	0.890	0.773	0.869	0.809	0.830	0.809	0.866	0.959	0.851
23	0.918	0.795	0.866	0.791	0.710	0.712	0.820	0.778	0.799
24	0.656	1.000	1.000	1.000	1.000	1.000	1.000	1.000	0.957
25	1.000	1.000	1.000	1.000	1.000	1.000	0.994	1.000	0.999
26	0.789	0.738	0.935	0.778	0.929	0.832	0.923	1.000	0.866
27	1.000	0.663	0.663	1.000	1.000	1.000	1.000	1.000	0.916
Mean of 8 branches	0.907	0.871	0.917	0.922	0.934	0.919	0.916	0.967	0.919

Source: The authors.

Intra-district Labour-use Efficiency Scores of Branches Under Karimganj District (2011–2018)

Table 6 exhibits the intra-district LUE score of branches operating under Karimganj district. The mean LUE score is 86%, which reflects the cost-saving potentialities to the tune of 14% (1–0.86) to be in efficient frontier. Year-wise mean LUE score reveals a decreasing trend over the years of study, except for the years 2013 and 2018. Branch ID 42 is the most efficient, and branch ID 31 is the least efficient branch in its labour use when compared to all other branches under Karimganj district.

Table 6.

Intra-district LUE Scores of Branches Under Karimganj District.

Branch ID	2011	2012	2013	2014	2015	2016	2017	2018	Mean
28	0.657	0.758	1.000	0.975	0.953	0.662	0.664	0.964	0.829
29	1.000	1.000	1.000	1.000	1.000	1.000	1.000	0.979	0.997
30	0.889	0.804	0.905	1.000	0.800	1.000	0.625	0.969	0.874
31	0.629	0.495	0.787	0.656	0.697	0.928	0.491	0.832	0.689
32	0.898	0.983	1.000	0.782	0.925	0.697	0.880	1.000	0.896
33	0.821	0.812	0.988	0.705	0.814	0.797	0.598	1.000	0.817
34	0.933	0.979	1.000	0.651	0.675	0.626	0.493	0.838	0.774
35	1.000	0.942	0.942	0.886	0.857	0.557	0.568	1.000	0.844
36	0.836	0.925	1.000	0.683	0.614	0.705	0.490	0.844	0.762
37	1.000	1.000	1.000	1.000	1.000	1.000	0.701	1.000	0.963
38	1.000	0.883	0.883	0.905	1.000	1.000	1.000	1.000	0.959
39	0.817	0.815	0.998	1.000	1.000	1.000	1.000	1.000	0.954
40	0.896	0.804	0.898	0.896	0.835	0.824	0.853	1.000	0.876
41	0.932	0.806	0.864	0.739	0.879	0.705	0.948	0.827	0.838
42	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
43	1.000	1.000	1.000	1.000	0.625	0.838	0.493	0.970	0.866
44	1.000	0.576	0.576	0.500	0.607	0.863	0.500	1.000	0.703
Mean of 17 branches	0.900	0.858	0.932	0.846	0.840	0.836	0.724	0.954	0.861

Source: The authors.

Determinants of Labour-use Efficiency

Selected Contextual Variables for Tobit Regression

The explanatory variables, which are considered in the Tobit model (for the estimating the impact of contextual variables on LUE) are location (dummy variable), cluster 1 (dummy variable), cluster 2 (dummy variable), number of employees, the number of nearby branches of other financial institutions and the total business of the selected branches. For location, it is hypothesized that the urban branches are more efficient in its labour use than semi-urban and rural branches. Cluster 1 and cluster 2 are two district dummy variables. For the branches under Hailakandi district, the dummy variable assumes a value of 1, and 0 otherwise. Similarly, the cluster 2 dummy variable assumes a value of 1 for the branches under Karimganj district, and 0 otherwise. For avoiding the dummy variable–induced multicollinearity, we have considered only two cluster dummies. Since the location dummy assumes a value of 1 for urban branches, and 0 otherwise, it is expected to have a positive impact on branch-level performance.

For cluster 1 and cluster 2, there is a probability of two tails. In case of number of employees, it is assumed that the increase in number of employees will decrease the LUE score as it will increase in labour cost directly. For the number of nearby branches of other financial institutions, it is hypothesized that if there exists any other financial service provider in the vicinity of the in-sample branches, it will have a positive spillover effect on LUE. It is hypothesized that the total business of the branches will have a positive impact on efficiency score. An overview of the independent variables included in the study and their expected impact on the dependent variable is presented in Table 7.

Table 7.

Description of Variables for Tobit Model.

Variables	Category	Description	Expected Sign of the Co-efficient
Labour-use efficiency	Response variable	Estimated efficiency scores	-
Location	Contextual variable	Urban = 1, others = 0	Positive
Cluster 1	Contextual variable	Branches under Hailakandi district = 1, = 0 for Cachar and Karimganj	Positive/negative
Cluster 2	Contextual variable	Branches under Karimganj district = 1, Others = 0	Positive/negative
Number of employees	Contextual variable	Total staff strength of the branches	Negative
The number of branches of other financial service institutions	Contextual variable	Other financial service providers	Positive

Source: The authors.

Outcome of the Tobit Regression

Table 8 presents the efficiency score changes by the coefficient of each unit increase in the corresponding predictor variable. First, location, which indicates the categorization of branches as urban, semi-urban and rural, has no significant impact on LUE score. Likewise, second independent variable cluster 1 has no significant impact on LUE score. Third, cluster 2 (which indicates the performance of branches under Karimganj district in comparison to branches under Cachar district) has a positive impact on LUE. It means that the branches under Karimganj district are better than the branches under Cachar district in its labour use. Fourth, the number of employees, which denote the total staff strength of the branches, has a negative influence on LUE. It is indicative of the fact that with an increase in the number of employees, the cost on employees increased, which ultimately increased the overall expenditure of a branch. Number of employees can be considered as an asset only when the productivity from the employees is greater than their cost, that is, salary. Thus, it affects negatively on the overall efficiency of a bank. So in the present study, the total number of employees can be considered as a determinant of inefficiency. Fifth, the number of other financial institutions are nearby branches, which mean that other financial service providers have no significant impact on the LUE. Lastly, the total business of the selected branches (the total of deposit and advances) is found to have a positive impact on LUE. As increase in the total business of the bank means increases in deposit and advances, which is possible only when the labour is utilized in an efficient manner efficiently.

Table 8.

Outcome of Tobit Regression.

Contextual Variable	Observed Coefficient	Standard Error (SE)	Coefficient/SE	Probability of Type 1 Error	Coefficien t Lower Bound	Coefficien t Upper Bound
Intercept	0.5613	0.1131	8.50	0.000	0.7396	1.1130
Location	0.0292	0.0509	0.57	0.567	–0.0707	0.1290
Cluster 1	0.0132	0.0283	0.47	0.642	–0.0423	0.0687
Cluster 2	0.0509	0.0187	2.73	0.006	0.0143	0.0875
Number of employees	–0.1286	0.0127	–10.13	0.000	–0.1534	–0.1037
Number of other financial institutions	–0.0016	0.0069	–0.23	0.818	–0.0151	0.0119
Total business of the branch	1.33e-09	1.67e-10	7.96	0.000	1.00e-09	1.66e-09

Source: The authors.

Comparison with Other Labour-use Efficiency Studies

The extant literature on LUE of Indian banks includes two strands of literature—labour-use performance at the aggregate level and at the branch level. The research studies following the first strand include Jaffry et al. (2008) and Herwadkar et al. (2019), while the first study noted an inter-temporal improvement in labour cost-efficiency (for the period from 1985 to 2003. However, the second study confirms a moderation of such trend for the period from 2005 to 18. Das et al. (2009) followed the second approach and found considerable heterogeneity in labour-use performance across four regions of India, which can, at least partly, be attributed to differences in local work ambience. Although the branches chosen in our study are much more homogeneous (in terms of work environment), we also found significant variations in branch-level performance across the branches.

Conclusion

The study is expected to make a significant contribution to the existing research base and is likely to signal the bank managers about the possible cost reallocation in their branches. It cautions them about the possible cost curtailment prevailing in the branches under study following the best practice. From the present study, it can be said that the LUE may be achieved to its fullest extent by restructuring of the bank branches and reallocation of the employees among the existing branches, where the branches are understaffed from the branches which are overstaffed. Thus, it is of utmost importance to have optimal staffing in all the branches as per its volume of business as well as its requirement. Reallocation of labour over the branches based on the best practice branch under study may be a solution to the problem of LUE at branch level. Further, opening up of new branches in the most business potential areas may be the solution of adjustment of the excess labour force after occupying in the understaffed branches. Thus, the management has to take appropriate measures to improve their managerial efficiency in its labour-use management. The present study, however, leaves ample scope of research in this direction. Research is necessary to find out the lacunas with respect to LUE at branch level.

The study suffers from certain limitations as it covers only one region out of eight regions of the bank. The present study is confined to only 8 years and only one technique for efficiency measurement is used.

Footnotes

Declaration of Conflicting Interest

The authors declared no conflicting interest with respect to the research, authorship and publication of this article.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

ORCID iDs

Mohuya Deb Purkayastha

Joyeeta Deb

Ram Pratap Sinha

References

Al-Jarrah

I. M.

(2007). The use of DEA in measuring efficiency in Arabian banking. Banks and Bank Systems Journal , 2(4), 21–30.

Banker

R. D.

, Charnes

, & Cooper

W. W.

(1984). Some models for estimating technical and scale inefficiencies in Data Envelopment Analysis. Management Science Journal , 30(9), 1078–1092.

Battese

E. G.

, Heshmati

, & Hjalmarsson

(2000). Efficiency of labor use in the Swedish banking industry using a stochastic frontier approach. Empirical Economics , 25(4), 623–640.

Chadha

(2017). An analytical study of reforms and their impact on Indian banking sector. International Journal of Business Administration and Management , 7(1), 112–129.

Charnes

, Cooper

W. W.

, & Rhodes

(1978). Measuring efficiency of decision making unit. European Journal of Operational Research , 2(6), 429–444.

Das

, Ray

C. S.

, & Nag

(2009). Labor-use efficiency in Indian Banking: A branch level analysis. Omega - The International Journal of Management Science , 37(2), 411–425.

Government of India. (1975). Working group on Rural Banks, Narasimham Committee on RRBS.

Herwadkar

S. S.

, Neelima

K. M.

, Verma Radheshyam

, & Asthana

(2019, April). Labor cost efficiency of Indian banks: A non-parametric analysis. RBI Bulletin , 2019, 95–106.

Ikhide

S. I.

(2008). Measuring the operational efficiency of commercial banks in Namibia. South African Journal of Economics , 76(4), 586–595.

10.

Jaffry

, Ghulam

, & Cox

(2008). Labor use efficiency in the Indian and Pakistani commercial banks. Journal of Asian Economics , 19, 259–293.

11.

Kumar

, & Gulati

(2014). Deregulation and efficiency of Indian Banks . Springer.

12.

Ray

S. C.

(2014). Branching efficiency in Indian Banking: An analysis of a demand-constrained network (Working Papers 2014-34). University of Connecticut, Department of Economics.

13.

Ray

S. C.

, Das

, & Mukherjee

(2017). Labor-cost efficiency with indivisible outputs and inputs: A study of Indian Bank Branches (Working Papers 2017-04). University of Connecticut, Department of Economics.

14.

Shanmugam

K. R.

, & Lakmanasamy

(2000). Labor use efficiency of banking sector in India. Indian Journal of Labor Economics , 43(4), 989–999.

15.

Sinha

R. P.

(2006). Spread efficiency of Indian Commercial Banks. IUP Journal of Bank Management , 5(4), 39–58.