Welcome to Alumni & Corporate Relations
Kerala State Road Transport Corporation
A Relook at Its Efficiency and Potential
This paper is a modified version of the MA dissertation submitted to the Indian Institute of Technology Madras by Sreelakshmi P. The authors gratefully acknowledge the helpful comments and suggestions from an anonymous referee, which helped them to Improve the paper. The authors are also grateful for the various comments they received during the review meetings of the project.

This paper attempts a depot-wise efficiency analysis of the Kerala State Road Transport Corporation and its regional variations for the period 1988–97 using the data envelopment analysis methodology. It then links the findings to the KSRTC’s overall financial position for the subsequent period from 2003–04 to 2014–15. The study finally concludes that the day-to-day operational position of the KSRTC can be substantially bettered if it can utilise its prevailing efficiency improvement potentials to some extent.

State Road Transport Undertakings (SRTUs) owe their origin to the Road Transport Corporations Act, 1950. Currently, there are 54 SRTUs in India, providing connectivity to people across rural and urban areas. SRTUs collectively operate 0.15 million buses and serve 70 million passengers, generating on average 1,480 million passenger kilometres each day. Public road transport corporations are instrumental in providing a low-cost alternative mode of transport to the people. Being a more economical alternative to the other modes of transport, they successfully fulfil some social obligations and simultaneously provide efficient, economic, safe and reliable public transport facilities across the country.

However, SRTUs face many problems, which were accentuated further after the passing of the Motor Vehicles Act, 1989 that liberalised private bus operations. With declining financial support from both central and state governments, after the liberalisation of the economy and in the wake of intense competition from private operators, these organisations have had to largely fend for themselves (Karne and Venketash 2005). On the other hand, SRTUs have been incurring huge financial losses from the very beginning of their operations and their financial performance has further deteriorated in recent years. High operational and maintenance costs, concessions provided to various sections of the people (due to government policies), high share of old vehicles in their fleets, high staff to bus ratio,1 low fleet utilisation and low fuel efficiency are some of the major reasons cited for the poor financial performance of public sector transport undertakings. All these result in huge accumulated losses, which pose a major hurdle for their viable operations. According to a report published by Ministry of Road Transport and Highways, most of the SRTUs in India are unprofitable. In 2015–16, 47 SRTUs incurred a combined net loss of `11,349 crore, which was 7.2% higher than the loss they made in 2014–15. Only five SRTUs made profits during that period. These are the Bangalore Metropolitan Transport Corporation (BMTC), Karnataka State Road Transport Corporation (SRTC), Odisha SRTC, Uttar Pradesh SRTC and Himachal Road Transport Corporation. Delhi Transport Corporation was the most unprofitable SRTU, with a loss of `3,411 crore. The Kerala SRTC (KSRTC) has been listed third with a loss of `738 crore. Nonetheless, since public sector undertakings (PSUs), such as SRTUs offer services with some social obligations, financial losses should not be the only yardstick to measure their performance, rather efficiency and effectiveness are more important than mere profitability (Singh and Jha 2017).

Given this backdrop, we study the overall performance of the KSRTC with the focus on three specific objectives: (i) measure the relative performance across the depots; (ii) measure the regional variation of such performances; and (iii) investigate if there is any systematic correlation between the performance vis-à-vis size and quality of services provided by a depot. Since depot-wise information of the KSRTC is available only until 1999, we would subsequently analyse the implication of our efficiency analyses results on the overall financial performance of the KSRTC for the later years. The paper unfolds as follows. An overview of the KSRTC comes first, which is followed by a review of the concerned literature. Then the paper discusses the details on the methodology, database used and selection of the variables used in the analyses. This is followed up by a discussion on the empirical findings and conclusions.

An Overview of the KSRTC

The KSRTC is one of the oldest transport undertakings in India. The major objective of the corporation is to provide convenient and safe travel experience to the public. Covering both urban and rural areas alike, it operates services to the people in cities like Thiruvananthapuram to the remotest villages in Wayanad. It provides some services to a few cities in the neighbouring states as well. The origin of the corporation dates back to the pre-independence era. The erstwhile king of the Travancore state Chithira Thirunal Bala Rama Varma took the initiative to establish the Travancore State Transport Department (TSTD) to improve the public transport system in the princely state. The state road transport service began its operations on 21 February 1938, with its inaugural service from Trivandrum to Kanyakumari. The enactment of the Road Transport Corporations Act later in 1950, paved the way for the establishment of the KSRTC in 1965. Now, its total fleet of 6,241 buses provides services in 6,389 routes, carrying more than three million passengers a day.

The KSRTC has been in deep financial crisis for a good part of its existence. It incurs the highest loss among 58 underperforming PSUs in Kerala. According to a report of the Comptroller and Auditor General of India (CAG) in 2016, it incurs a loss of around `123 crore on an average per month. The mismanagement of the corporation is often cited as a reason for its poor financial health. Protests by the KSRTC employees and pensioners demanding their dues is a regular event in Kerala. The corporation is dependent on the financial support of the state government to pay even its employees. Several attempts to revitalise it have not been successful. In 2015, the Kerala High Court asked the KSRTC to shut down its operations and sell its assets, if it is unable to pay its employees and pensioners. In an affidavit submitted to the Supreme Court in 2019, the corporation informed that it is incurring a loss of `4,000 crore to which the Supreme Court also suggested that the corporation be shut down, if it is unable to stem its losses.

The Sushil Khanna Committee appointed by the Government of Kerala (2017) also observed that, not only are the total revenues of the KSRTC way below its total expenses, but this gap is widening at an increasing rate since 2009–10. The committee identifies this to be the one major factor that puts the KSRTC at a constant risk of default on salary and pension payments. Its average fleet utilisation is also found to be way below that of other SRTUs in neighbouring states. The committee opined that the crisis in the KSRTC is largely a result of its operational inefficiency, which is also reflected in the widening gap between its earning per kilometre (EPKM) and cost per kilometre (CPKM). Figure 1 depicts these trends, including the widening (negative) margin per kilometre (MPKM) for last few years.

Hence, finding a viable way to make KSRTC operationally efficient is an important issue. We attempt this by making a depot-wise efficiency analysis of the KSRTC during 1988–97. The dated data is because the KSRTC had stopped publishing depot-wise information since 2000. This will allow us to make an assessment of the overall performance and identify potential areas of improvement. Subsequently, we draw upon the depot-wise efficiency scenario to assess its possible implications on the overall financial performance of the KSRTC. Administrative reports of the KSRTC are the source of data used for the study. Data for two years are, however, not available. The data envelopment analysis (DEA) methodology is used for our study.

Review of Literature

A number of recent studies analyse the performance of the road transport sector. Although the stochastic frontier analysis (SFA) is used in some cases, DEA is still the most commonly used method. Using both variable returns to scale (VRS) and free disposal hull DEA methods, Kerstens (1996) examine efficiencies of the French urban transit companies and observes that the location of the efficiency distributions differs substantially depending on the methodology and especially on the output specification considered.2 As far as the studies evaluating performance of the SRTUs are concerned, these mostly involve the comparative analysis of performance across different SRTUs and identifying factors affecting them. For instance, using SFA, Kumbhakar and Bhattacharyya (1996) study the efficiency of 31 selected SRTUs from 1983 to 1987 and observes that the total factor productivity (TFP) has actually gone down for 11 SRTUs.3

A few studies have also been undertaken to evaluate the performance of specific SRTUs. For instance, Karne and Venkatesh (2003) study the technical efficiency (TE) and TFP of Maharashtra SRTC from 1996–97 to 2001–02 using DEA-based Malmquist productivity index and observes that while there is a marginal improvement in the overall TFP scenario during this period, the increment in TE is negligible. Depot-wise efficiency analysis of SRTUs are also available in the literature. For instance, using DEA, Bishnoi and Sujata (2007) measure the TE of 20 depots of Haryana SRTC for the period 2006–07 and observes that the inefficient depots are in a position to reduce, on an average, 9.25% of fuel consumption, 25.6% of the staff and ensure a 13.7% increase of fleet usage. Using DEA, Nagadevara and Ramanayya (2010) study the depot-wise efficiency of 25 depots of Karnataka SRTC for the period from 2004–05 to 2008–09 to identify the depots that were initially on an efficient frontier but eventually fell and became inefficient and vice versa. They also analyse the factors contributing to the inter-temporal efficiency movement of the depots. Hanumappa et al (2016) measure the performance of premium bus services of BMTC operating in Bengaluru city. They observe that while most of the depots are efficient, some depots have significant potential for further improvements. Using DEA, Venkatesh and Kushwaha (2018) examine short- and long-term cost efficiencies (CEs) of selected SRTUs for the 2004–13 period. CE is further decomposed into allocative efficiency (AE) and TE and all these efficiency scores are computed for four different types of SRTUs: company, corporation, government department and municipal transport undertaking. They observe that municipal transport undertakings are the worst performers among these, while the company types are found to be the best in terms of all these three indicators. However, there is no study worth mentioning on performance evaluation of the KSRTC.

Methodology, Data and Variables Selection

As we have already noted, both SFA and DEA4 methods have been used to measure the performance of transport operators. While SFA is the latest significant development among the class of parametric/econometric methods and corrected ordinary least squares, goal programming, modified ordinary least squares, thick frontier approach, etc, are few other popular methods that come under this broader umbrella. On the other hand, DEA is a mathematical programming-based method. Since there is apparently no parameter estimated in the DEA using the observed data points, its use is termed in the literature as a non-parametric method, and hence, the efficiency scores obtained through the analysis are more of calculated than estimated nature. However, later developments like bootstrap-based DEA is an outcome of the dedicated efforts to bring some statistical properties even with DEA efficiency scores.

Efficiency can be defined as the ability of an organisation or decision-making unit (DMU) to produce maximum output with minimum inputs. Hence, one may have two alternative conventional ways of defining TE: (i) reciprocal of the maximum possible proportional expansion of the output vector, keeping the input usage vector unchanged; and (ii) feasible minimum proportion of actual input usage vector, keeping the output vector unchanged. Hence, the former is termed as an output-oriented measure of TE, while the latter is termed as an input-oriented measure. And, needless to mention that both of these lie between zero and unity by definition. In relative performance measuring methods like SFA and DEA, benchmark production technology is defined on the basis of the best-performing DMUs, and each DMU is compared vis-à-vis to the so-constructed benchmark technology to obtain efficiency score as a by-product. We have used conventional DEA here for our analyses. It is easy to understand that output-oriented DEA is used when the DMUs desire to maximise the output without changing the level of input use. Alternatively, input-oriented DEA is used when their aim is to minimise the input use to produce some given level of output. Unlike few extreme cases where it is well defined before the management that which of these two is the prioritised objective, it is not so obvious before the public transport corporations like the KSRTC, and hence, each of these two has their own significance. Thus, both the methods are used in our study and efficiency distribution of the depots calculated with both the methods5 remain more or less the same.

The DEA model is based on the following fairly general assumptions about the production technology: (i) all observed input–output combinations are feasible; (ii) the production possibility set is convex; (iii) all inputs are freely disposable; and (iv) all outputs are freely disposable.

Output-oriented model: In the output-oriented model, the aim is to obtain a factor by which the output vector can be expanded from the same level of input use, and hence, one has to perform the following linear programming problem (LPP), once for each DMU.

Maximise φ for a particular firm p,

Subject to the constraints


Xpj ≥ ∑ λi Xij ; j = 1, 2, …, l



φ Ypr ≤ ∑ λi Yir ; r = 1, 2, …, k



∑ λi = 1


λ≥ 0 ” i = 1, 2, …, n

Here, φ is the maximum possible increase in the output vector of a firm, keeping the input at the same level so long the resultant input–output vector remains feasible. The output-oriented TE score is given by the reciprocal of the optimal φ

TEp = 1/optimal φ


Input-oriented model: In the input-oriented model, the aim is to obtain a factor by which the input vector can be contracted for the same level of output production, and hence, one has to perform the following LPP, once for each DMU.

Minimise Θ for a particular firm p,

Subject to the constraints,


Θ Xpj ≥ ∑ λi Xij; j = 1, 2, …, l



Ypr ≤ ∑ λi Yir ; r = 1, 2, …, k



∑ λi = 1


λ≥ 0 ” i = 1, 2, …, n

Here, (1 – optimal Θ) is the maximum proportional contraction in the input vector of a firm, keeping the output at the same level so long as the resultant input–output vector remains feasible. Hence, the input-oriented TE score is measured by the optimal Θ itself.

Data and choice of variables: We have used the administrative reports of the KSRTC from 1988 to 1997 (with the exception of 1992 and 1996 for which reports were unavailable). We have information across 56 depots6 of the KSRTC, geographically scattered all over Kerala. We have used (i) total number of buses; and (ii) total number of staff as two inputs and total revenue as the output variable. Input and output variables have also been similarly chosen in the literature (see, for example, Kerstens 1996; Viton 1998; Venketesh and Kushwaha 2016; Singh and Jha 2017; and others in this regard). Moreover, although the actual output KSRTC produces is the passenger services, since bus fares are decided centrally by the government agency, one can assume that the price of passenger services are same across depots, and hence, total revenue can be considered as a valid proxy of the services output. For the subsequent analyses (that is, for the post-1997 period), we have collected information from the KSRTC Annual Accounts and Audit Report of various years.

Empirical Findings and Discussions

TE score is computed for all the depots of KSRTC for eight sample years. As already noted, we have considered one output and two inputs (variable returns to scale [VRS]) production technology and have used DEA in this regard. We would like to clarify our choice of orientation of measurement of TE here. Only for few rare, extreme cases where it is quite straight forward, given the nature of the concerned production relation in the backdrop, it is subjective (that is, it depends on the choice of the researcher) in an overwhelming majority of the real-life cases. For instance, the primary objective of any PSU is to serve the broader society, and the KSRTC is not an exception. Unlike the private transport companies, profit maximisation cannot be its sole objective, rather to provide commutation facility to people, even in the uneconomic routes and during the odd hours. Inputs like number of buses and staff under a depot would largely be fixed, while it can use different strategies like advertising, time and route management, and so on to expand revenue. Hence, output-oriented measurement of TE seems a better choice. Alternatively, staff per bus ratio of many of the depots are very high while fleet utilisation rate shows that the fleets are not fully utilised in most of the depots of the KSRTC. An input-oriented analysis may be useful in such cases to know the extent of contraction of
resource utilisation to achieve the already attained level of revenue. Hence, we have followed both the methods and have made a comparison between them. Nonetheless, distribution of TEs across various depots remains largely the same.

Depot-wise analysis: Depots are clubbed under four quartiles to get a distributional picture as per their performance. Similar analysis has also been done by Saxena (2017). Tables 1 (p 60) (for output-oriented TE scores) and 2 (p 60) (for input-oriented TE scores) show these distributions. Thiruvananthapuram central turns out to be the only efficient depot. Among the marginally efficient depots, six are from south, six are from central and two are from north Kerala. More than half (eight) of the most inefficient depots are in south Kerala. Thiruvananthapuram city, Peroorkada and Pappanamcode are the three depots which come under the highly inefficient category in all the years. These depots need special attention as they are the worst performers among all.

Figures 2 and 3 (p 60) show the movement of average TE scores and average total revenue over the years. It shows an increase in TE till 1991, then a slight dip till 1995 and a subsequent revival thereafter. However, a continuous increase in average total revenue is observed over the years until 1994, a slight dip in 1995 and revival thereafter. Hence, we conjecture that 1995 perhaps is an important year for the KSRTC.

Regional variations: Kerala is divided into three regions: north Kerala, central Kerala and south Kerala. The north Kerala districts are Kasaragod, Kannur, Wayanad and Kozhikode. Central Kerala includes districts of Malappuram, Palakkad, Thrissur, Ernakulum, Idukki and Kottayam. South Kerala includes districts of Thiruvananthapuram, Kollam, Pathanamthitta and Alappuzha. Of the 56 depots in this study, 10 are in north Kerala, 20 are in central Kerala and 26 are in south Kerala. We have clubbed the depots within each of these three regions and calculated the (arithmetic) average of the TE scores to get an overall performance indicator of the regions in question. Figures 4 and 5 show this distribution. It is observed that the central Kerala region outperforms the two other regions. To be specific, while output-oriented (input-oriented) overall TE score for the central Kerala is 73% (75%), north and south Kerala’s scores are 62% (63%) and 64% (61%) respectively. Hence, although south region is slightly more efficient than the north as far as output expansion is concerned, it is the opposite from the input contraction point of view.

Quality of Services vis-à-vis Efficiency

We have also investigated the quality of services offered by the KSRTC. The literature offers many such studies. For instance, the study by Navarro-Espigares et al (2011) on efficiency and quality of the healthcare sector in Andalusian hospitals in Spain shows a weak positive association between efficiency and service quality indicators. Laine et al (2005) also observe a weak positive association between quality of care and efficiency in institutional long-term care wards for elderly. As already noted, public transportation is primarily aimed to satisfy people’s demand for commutation needs. Quality of services in this regard, hence, may also be considered as an important factor in determining the overall performance of the corporation. We have taken three indicators for that: (i) percentage of irregularity of services; (ii) break down rate; and (iii) accident rate. Percentage of irregularity of services is calculated as the percentage of reported irregular time of arrival. Break down and accident rates are measured as the number of breakdowns per 10,000 effective kilometres and the number of accidents per 1,00,000 gross kilometres respectively. The (arithmetic) average of each of these three measures are calculated against each quarterly measure of the efficiency levels. As discussed earlier, Level 1 stands for the most efficient depots and so on until Level 4 which stands for the least efficient depots. Tables 3 and 4 (p 61) shows this distribution vis-à-vis output-oriented and input-oriented efficiency measures respectively.

The figures show an apparent negative relation between (economic) performance and quality of services across the KSRTC depots. For further rigour in this regard, we form an overall quality index, by computing (geometric) average of these three quality indicators, and calculate its pairwise correlation coefficient with technical efficiency scores of the depots, as reported in Table 5. It shows that the correlation coefficient between overall quality of services delivered by a depot and its technical efficiency is negatively7 significant in five out of total 18 alternative cases, whereas it is positively significant in only one case. So, in totality, we observe some evidence that the overall quality of services provided by a depot is negatively correlated with its performance. In fact, we have noted earlier that the depots scattered in the central Kerala region are more efficient than the otherwise identical depots located elsewhere, possibly owing to relatively larger volume of businesses around Kochi, the financial capital of Kerala. High population density around this business hub calls for higher frequency of services in general and transportation services in particular, which may be a reason for reported higher percentage of irregularities and accidents associated with the better (economically) performing depots. In fact, both the average passenger kilometre per bus and the staff per bus are higher for the depots in central Kerala than in the others.8

Size of Depot vis-à-vis Efficiency

We have also investigated the relation between size of a depot and its performance. Such an analysis is very common in the productive efficiency-related literature on various sectors of the Indian economy (see, for instance, Bhandari and Maiti 2007, 2012; Bhandari and Ray 2012). We have taken the number of buses as an indicator of the size of a depot and observe that size and productive efficiency are positively correlated. Figure 6 summarises our findings in this regard. Such positive correlation between the size of a productive unit and its technical efficiency is also observed in all the three studies mentioned above.

Bridging the Gap

We are quite aware that the analyses so far are for a period which is almost two decades old. However, this is mainly because the KSRTC stopped publishing such depot-wise information since the early 2000s. Rather, these days, it publishes information for the KSRTC as a whole. Using that, we would like to make an effort here to bridge the gap between its productive performances since 1988 with its income-expenditure performance for the later periods. Table 6 shows the KSRTC’s overall income, expenditure and profit scenario from 2003–04 to 2014–15.9 Two immediate features that emerge from these figures: (i) although it faces cumulative losses for the entire period, it manages to make some profits through its regular operational activities in eight out of 12 years; and (ii) non-operating expenditures, like interest payments on debt and other expenses besides superannuation and welfare activities, seems too large to bear on its own. Nonetheless, its overall level of technical efficiency during the period of study is 67.3% and 66.7%, respectively for the output- and input-oriented measures.10 We compare these figures with its income–expenditure figures for the later years to explore whether it can sustain operational viability on its own, possibly through some productive efficiency gains. We define attainable operating revenue in column 2 of Table 7a (p 62) by dividing operating revenue by the overall average (output-oriented) technical efficiency score of 0.673. Columns 4, 8 and 10 of Table 7a are adjusted accordingly. It implies that the KSRTC may be able to increase its operating revenue by 49% (that is (   1— 0.673– 1) × 100% on an average, without incurring any additional expenditure. The last column of Table 7a shows that it can very well sustain on its own (except for the last two years) by improving its productive efficiency. Table 7b is a parallel counterpart of Table 7a if we consider a maximum (output-oriented) technical efficiency score of 0.799 for 1991 in this context. Even in that case of the smallest possible revenue expansion scenario, the overall financial performance does not look so dismal. Similarly, we define attainable operating expenditure in column 5 of Table 8a by multiplying its actual operating expenditure with the overall average (input-oriented) technical efficiency score of 0.667. Columns 7, 8 and 10 of this table are adjusted accordingly. Column 10 of Table 8a implies that the KSRTC can very well operate sustainably, by reducing its operating expenditure to the average extent without compromising its achieved revenue. Table 8b is the parallel counterpart of Table 8a where we use maximum (input-oriented) technical efficiency score of 0.791 for 1991 in this regard. We also observe here that the scenario before the debt-ridden KSRTC is not as bleak as it seems. Some bit of improvement in its productive efficiency, therefore, is inevitable for its sustainable continuance. Needless to mention here that we have presented the hypothetical scenario here by considering only the average and minimum possible expansion in revenue (contraction in expenditure). In fact, its overall profit scenario is really encouraging for its sustained growth if we consider maximum possible expansion in revenue (contraction in expenditure) in this regard.11


Concluding Remarks

The study highlights that many depots of the KSRTC have substantial scope for improvement of productive efficiency. It also observed that (i) the depots located in central Kerala are, on an average, better performers than those located elsewhere, possibly owing to the larger volume of business because of their proximity to Kochi city, the financial capital of Kerala; (ii) larger depots are performing better than their smaller counterparts; and (iii) there exists a negative correlation between productive performance of a depot and the quality of services it provides.

Investigating the possible implication of our findings on the KSRTC’s overall financial performance for the later period from 2003–04 to 2014–15, we note that the day-to-day operations of the KSRTC are not too unprofitable. Rather, its debt burden and other welfare activities-related obligations are too large to bear on its own. Moreover, the overall financial position of the KSRTC would be vastly better if it improves its operational efficiency to some extent. In fact, the scenario would even be rather encouraging if it utilise its efficiency improvement potential to the maximum extent possible.


1 The average staff to bus ratio is observed to be as high as 5.17.

2 Similarly, studies of Jorgensen et al (1997) and Odeck (2006) for Norway; Viton (1998) and Karlaftis (2004) for the United States; Boame (2004) for Canada; Sampaio et al (2008) for Brazil and Europe; and Carvalho et al (2015) across 21 Brazilian cities are worth mentioning.

3 Studies of Ramanathan (1999), Jha and Singh (2001), Singh and Venkatesh (2004), Anjaneyulu et al (2006), Agarwal (2009), Kumar (2011), Singh and Jha (2017), among others, may also be mentioned in this regard.

4 Intellectual origin of the DEA goes all the way back to Farrell (1957) who proposed a single-input—single-output model to measure productive efficiency of individual producers. It has been subsequently generalised to accommodate multiple inputs as well as multiple outputs and termed as DEA by Charnes et al (1978) (for CRS technological specification, assuming each of the DMUs operates with the optimal production scale) and further by Banker et al (1984) (for general VRS technological specification). Of course, it can easily be understood that the CRS can be considered as a special case of VRS.

5 One can easily understand that the TE score of a DMU would be exactly the same under both these methods, if the CRS technological specification is considered.

6 Our prior analysis shows that Erattupetta depot is an outlying observation in 1990. Hence, we drop it from our sample of that year to eliminate the outlier effect.

7 Since a larger value of the overall quality index stands for poorer quality of services a depot delivers (by virtue of the way we define it), a positive correlation coefficient between it and technical efficiency implies that the quality of services provided by a depot is negatively related with its performance.\

8 Relative share of average passenger kilometre per bus (staff) for southern, central and northern regions are 32.74% (31.10%), 33.74% (34.64%), and 33.53% (34.26%), respectively.

9 We could collect data until 2014–15 and are in the process of collecting it for the later years now.

10 To be specific, yearly average output-oriented (input-oriented) technical efficiency values are 70.9 (68.9), 72.2 (71.5), 76.7 (76.6), 79.9 (79.1), 57.1 (67.1), 44.1 (55.4), 70.0 (40.9) and 67.2% (73.6%) between 1988 and 1997, with exception for the years 1992 and 1996.

11 However, we do not show the corresponding figures here and would be happy to share it on demand.