Privatization of the Taiwan Motor Transport Company

The TMTC was set up in 1980, since then, the Taiwanese intercity bus services have been provided by the nationwide TMTC in monopoly. However, a long-term trend decline about 10% per annum in intercity passenger journeys from 1980s has been seen, mainly due to the increasing use of private cars and illegal bus services following the opening of the first national highway in 1979. Subsequently, during 1990s, both endogeneous and exogeneous factors led to a major structural change in the Taiwanese intercity bus industry. Several

influences occurred within TMTC itself. First, public management was exceeding inefficient.

Under the operations of TMTC's 50 subsidiaries (stations) in 1990, there were 105 national highway lines and 102 provincial highway lines with a total of 3,070 vehicles. The number of TMTC employees was 13,000 by 1990. The number of employees per vehicle was more than four, approximately double the average of privately owned local bus operators. Secondly, the TMTC's quality of service severely deteriorated. Its fleet, where more than half the number of vehicles were more than 10 years old, was the least-maintained part of the system. Complaints about its poor services began to increase. Part of its operational inefficiency should attribute to both TMTC's management and its employees, and the rest of it was directly caused by governmental and politicians’ intervention in controlling the transportation industry. For example, without governmental approval (usually time consuming), TMTC could not determine its budget, fare, staff salaries and authority for new services (such as renew rolling stock). In addition, it was always required to serve cost-inefficient social goals and operate unprofitable lines. These aforementioned factors, not only combined to either increase the cost or reduce the productive efficiency but also resulted in a deficit of a million U.S. dollars accumulated in 1999. TMTC was unable to pay the debt service of its bonds and became a grant and subsidy soaking company, which allowed it to be relatively more irresponsible and inefficient.

Besides these pressures mentioned above for TMTC's privatization, there were three major exogeneous factors from the government's decision-making. First, intercity bus service provision within Taiwan has undergone fundamental change over the last decade, moving from a publicly owned and heavily regulated industry to a privately owned and partially deregulated market. Specifically, the intercity bus was still subject to quality control and price cap regulation because of its fare; however, restriction of entry was removed to a great extent.

The first POE was allowed to enter the national highway bus market running on 26 lines in 1990 and was able to provide alternative intercity service other than TMTC. Following the

partial deregulation of bus industry in 1995, many POEs successively set up entirely new services but almost parallel to the national highway lines, altogether 22 POEs with 32 lines, with those of TMTC's by 1999. TMTC continued to operate, lacking innovation in competition with these increasing new entrants, eroding its revenues, and making achievement out of sound financial condition impossible.

Secondly, in respond to a request from the government, TMTC began to implement an organization reform, mainly to both downsize its personnel and pass (to POEs) or close its less-attractive lines to reduce its increasing cost from 1995. However, the deficit was still increasing until 1998. The then government decided to privatize TMTC by 2001. Thirdly, in 2000, there was political enthusiasm for proceeding to reform public organizations under the newly-elected DPP (Democratic Progressive Party) government. Excessive cost due to inefficient management and thus suffering the problems of long-period loss making was another driving force for reexamining the performance of the TMTC. And lastly, the TMTC's financial crisis on the verge of bankrupt at the end of 1999 served as a further impetus for accelerating the privatization of TMTC.

All these factors led to the TMTC's privatization. The privatization has produced major structural changes in the intercity bus industry. Some of these important changes, characterizing its privatization, can be summarized into two points: First, the TMTC was fully privatized by transferring hundreds of vehicles, 53 of 62 national highway lines and 43 of 83 provincial lines, together with all the 15 stations and depots to some 1,100 employees (out of 3,100 employees), the resultant private enterprise became organized as Guo Guang Motor Transport Company (KKTC). The rest of national and provincial lines as well as remaining old vehicles were passed to existing local bus companies by means of tendering. Second, KKTC was awarded a public passenger transportation franchise for a period of five years.

Today, the KKTC is a good example of a POE operating in a similarly partially deregulated transportation market but almost entirely free from the government's restriction as a SOE.

Thirdly, subsequent to TMTC’s privatization, other three nationalized enterprises, including Taipei Municipal Bus Company, have followed this successful case to impletement their programme of privatization.

CHAPTER 3 Literature Review

It’s only quite recently that frontier studies have been developed as an appropriate methodologies to the transport sector, and the majority of studies have been published during the 1990s. A comprehensive survey of frontier methodologies and empirical results for public transit has been presented by De Borger et al. (2002). The existing frontier studies measuring urban transit performance have also been systematically summarized and critically assessed by them. In this chapter, frontier studies of transit systems are first reviewed, then the relevant researches concerning transit efficiency measurement, including frontier studies in Taiwan, input and output measures in transit, related studies of transit performance, as well as related studies of transit privatization are briefly reviewed.

3.1 Frontier studies of transit systems

Methods of measuring efficiency can be broadly classified into non-parametric and parametric. Non-parametric methods include indexes of partial and total factor productivity (TFP), and data envelopment analysis. The latter is essentially a linear programming based method. Parametric methods involve the estimation of neoclassical and stochastic cost and/or production functions (Gillen and Lall, 1997).

An overview of non-parametric and parametric frontier studies concerning bus transit systems is presented below.

3.1.1 Non-parametric approach

Regarding applications to transit efficiency studies, non-parametric approach have been used in the following cases.

Chu et al. (1992) used DEA to develop a single measure for the efficiency and a single measure for the effectiveness of a transit agency relative to other agencies within the same peer group. By using a single measure for each of these criteria, the paper provided a more robust indicator or transit performance than the widely used multiple ratio analysis performed in the Irvine Performance Evaluation Method (IPEM). Their analysis reinforced the notion that, for a public agency, measures of efficiency should be kept distinct from measures of effectiveness.

Obeng (1994) studied subsidy-induced technical inefficiencies in public transit systems using DEA in the United States. He found that subsidies improved technical efficiency in approximately 75% of the transit system studied. These efficiency improvement resulted in total cost savings of $13.66 million or $0.187 million per transit system. He argued that the type of subsidy given to the transit systems may determine its impact on technical efficiency.

He notes that an output-based subsidy and capital subsidy are important in determining transit efficiency.

Nolan (1996) used the DEA approach with a second stage regression analysis to study technical efficiency determinants in the United States transit sector. Among other things, he concluded that operating subsidies created significant and negative impacts on efficiency.

Agencies that received larger subsidies from state (but not federal) government had less incentive to produce efficient levels of output.

Kerstens (1996) evaluated the performance of a sample of French urban transit companies using a broad selection of nonparametric reference technologies for two specifications of the production process. In particular, the variable returns to scale DEA models with either strong or weak disposability in both inputs and outputs, and the Free Disposal Hull (FDH) are applied. An extensive comparison of the resulting radial output efficiency measures yields the following major methodological conclusions. First, the location of the efficiency distributions differs substantially depending on the methodology and

especially on the output specification considered. The latter differences vanish if the impact of outliers is eliminated. Second,, convexity has a stronger influence on the efficient-inefficient dichotomy than allowing for congestion by means of a weakly disposable DEA model. For policy purposes, these efficiency distributions are explained using a Tobit model. The findings corroborate results reported elsewhere: the harmful impact of subsidies, etc.

Furthermore, the network structure seems to account for some differences in performance.

Finally, a novelty in the urban transit context is the indirect monitoring effect of the French earmarked transportation tax.

Roy (1996) studied the productivity of the transport sector in Canada using the total factor productivity (TFP) index at the aggregate level. He found that total productivity of Canada’s transport sector grew by 15 per cent over the 1981 to 1993 period (1.1% per annum).

He also found that since the mid-eighties, productivity had been trending downwards in passenger carriers, while the gains of freight carriers had been accelerating since 1986.

However, he excluded urban carriers such as transit systems, taxicab operations, and special services such as school bus operators from the analysis.

Lyons (1997) indicated that there are many ways to look at productivity in the transit industry. The most commonly used indicators of performance are partial measures of efficiency. Recent studies have focused on developing a single measure of overall transit performance based on TFP and/or DEA models. However, these studies use different theoretical concepts, measures of output and input, and data sets to measure productivity. The objectives of this dissertation are: to determine whether the use of different single measure performance indicators yields consistent results; to determine whether there are significant differences in performance as measured by total and/or partial indicators; to explore which set of partial measures can best be used to predict overall performance; and to examine the influence of operating environments on overall performance. The overall measures used are TFP and DEA. Cross-sectional panel data are used for 93 urban transit firms which had 50

more buses in 1986, 1988, and 1990. Analysis revealed that the overall measures of productivity yield consistent results when the output variable remains fixed. However, the designation of firms as “best” or “worst” performers is substantially influended by the choice of the output variable. There are significant differences in performance as measured by overall and/or partial measures, but commonly-used partial measures are good predictors of overall productivity. While operating environment does matter, characteristics influenced by transit managers and policy-makers explain much of the variation in overall performance.

Viton (1997) studied the efficiency of U. S. multi-mode bus transit systems by asking whether they could expand their service (outputs) without requiring additional resources (inputs); or whether they could reduce input utilization without having to reduce service. He used the DEA technique in the study. The findings indicated that the degree of inefficiency present is small: decision-making units (DMUs) could reduce input usage by an average of only four percent without curtailing services, or could increase service by an average of only six percent without requiring additional inputs. Just under 80% of the sample was technically efficient, and about 20% of the industry was to some degree inefficient. The incidence of inefficiency was not strongly correlated with system input or output characteristics.

Viton (1998) examined the claim that US bus transit productivity had declined in recent years. It did so with reference to a piecewise-linear best-practice (DEA) production frontier, computed for multi-modal bus transit between 1988 and 1992. Efficiency was measured both by a Russell (Static) and Malmquist (dynamic) measure of productivity change. The principal finding was that, overall, bus transit efficiency had improved slightly over the period.

Button and Costa (1999) indicated that the regulatory framework under which the European transport network operates has changed significantly over the last 15 years. At the macro level the creation of the Single European Market has removed many of the institutional impediments to international transportation within European Union. At the meso level national governments have liberalized inter-city authorities have acted to introduce greater

market incentives in the provision of local public transport. This paper is primarily concerned with the effects on economic efficiency of measures which have resulted in more liberalized markets at the local levels. In particular, it focuses on expanding the relatively scant empirical literature in this field by quantifying the impacts of major regulatory changes in two major European cities. The input minimization version of DEA programming is used for the empirical analysis.

Nakanishi and Norsworthy (2000) stated that in the past few decades, the market share of bus passengers has declined, while many transit services have expanded. Furthermore, wage rates and other costs of labor such as benefits have been increasing, and regulations relating to the environment and Americans with disabilities have been enacted. This has fueled a systemic decline in productivity of agencies providing bus service. With increasing pressures on public agencies to be accountable to taxpayers and constrain resources, the efficiency of transit agencies must be addressed. The measurement of productivity is the initial step that must be taken toward improved performance. The authors use DEA, a linear programming technique, to estimate the relative efficiency of transit agencies providing motor bus service.

DEA is a nonparametric approach, used to and generate a best practice frontier and rank DMUs. The agencies that are efficient comprise the frontier and those that are not are ranked according to how far they are from their best practice counterpart on the frontier. The results generated from the DEA model are part of the first phase of the transit productivity study the authors have undertaken. Subsequent phases will examine other measurement techniques such as TFP and econometric estimation of scale and scope economies.

Odeck and Alkadi (2001) evaluated, from a productive efficiency point of view, the performance of Norwegian bus companies subsidized by the government. The framework was that of a deterministic non-parametric DEA approach to efficiency measurement. In this context several important issues were addressed: efficiency rankings, distribution and scale properties in the bus industry, potentials for efficiency improvements in the sector, the impact

of ownership, area of operation and scope, and ways of improving efficiency in the sector.

The findings showed that the average bus company exhibited increasing return to scale in production of its services. The extent of such returns however vary, with size and was more prevalent among smaller companies. The average bus company was found to have a considerable input saving potential of about 28 percent Neither economies of scope nor company ownership were found to have an influence on company performance. It was suggested that geographical factors needed a closer attention in future research.

Pina and Torres (2001) indicated that in recent years in the European Union (EU), they have witnessed an externalization process of the provision of local government services, in order to separate the political responsibility and the direct delivery of the service. The reasons that justify this process are focused on the belief that the private sector is more efficiency in carrying out economic activities, the pressure to reduce the public deficit and the public debt, the search for management systems that bypass public administration procedures, and the increase of control on local governments in auditing and accountability issues. The objective of this paper is to compare the efficiency of public and private sectors in the provision of urban transportation services. This paper shows the results of an empirical study commissioned by the Regional Audit Office of Catalonia (Spain), in order to evaluate the efficiency with which urban transportation services are delivered in the most important cities of this region. This efficiency study has been carried out using the DEA model, multiple linear regression and logit and cluster analysis. The results allow them to conclude that, in the cities studied, exogenous factors are not relevant and the private management of urban transport service is not more efficient than public management.

Cowie (2002) indicated that the British Bus industry had undergone considerable transformation since privatization. Five major operators had emerged to dominate the market, a position almost exclusively attained through acquisition. He reviewed the economies of scale argument commonly cited for this change and gave an overview of the acquisition

process. He questioned whether this argument gave a complete explanation for this industry development. For 58 individual companies, the level of technical efficiency attributable to firms operating at or near the optimum level of output was examined over 5 years to determine if mergers in practice had resulted in scale economies. Technical efficiency was estimated using data envelopment analysis, under assumptions of constant and variable returns to scale. Efficiency scores were then regressed on a time trend and a merger dummy to test whether acquired firms’ efficiency had significantly improved above the average. It was found that over the period, efficiency had improved. This improvement, however, could not be wholly attributed to the achievement of economies of scale. More specifically, there had been an improvement in the internal efficiency of acquired firms and some scale economies within group companies, the latter of which might have resulted from the eradication of competition.

Nolan et al. (2002) examined the extent to which subsidized urban transit agencies complied with the ISTEA (Intermodal Surface Transportation Efficiency Act) requirements of both technical and social efficiency. To do this, they proposed a method of measuring both technical and social efficiency using DEA. They found that in general, urban transit agencies did not pursue the social objectives (e.g. reduction of air pollution, increased safety and job creation) specified by ISTEA which was passed in the US in 1991.

Karlaftis (2003) developed an efficient frontier production function in a three-stage approach to investigate transit production and performance. First, efficiency rankings and efficient subsets of transit systems are obtained through DEA, a non-parametric linear programming based methodology. Second, based on the results of the DEA analysis, globally efficient frontier production functions, in the context of transit operations in the United States, are built. Third, convex programming is used to extract the aggregate production function for the transit systems examined. The results indicated that when jointly considered, there was an improvement on both the theoretical and empirical aspects of examining efficiency and

production in transit systems. Further, the results indicated that efficiency and returns to scale findings differed substantially depending on the evaluation methodology used.

Odeck (2003) examined the efficiency of Norwegian bus industry to gain insight about factors affecting it. Non-parametric DEA was used to examine and decompose efficiency differences into input saving, output increasing and scale efficiency scores. Further.

Mann-Whitney rank test was employed to test for efficiency differences with respect to ownership, region of operation and size. The results suggested that there was in general a potential for input saving in the whole sector of about 21 percent. No significant differences were found between urban and rural operators with respect to efficiency scores, neither were there any performance differences with respect to ownership. This latter result deviated were

Mann-Whitney rank test was employed to test for efficiency differences with respect to ownership, region of operation and size. The results suggested that there was in general a potential for input saving in the whole sector of about 21 percent. No significant differences were found between urban and rural operators with respect to efficiency scores, neither were there any performance differences with respect to ownership. This latter result deviated were