INTRODUCTION

Benchmarking in seaports is an area that has attracted some research in the past (Cullinane et al, 2002). The aim of this paper is to identify 'best practices' in the management of seaports, which could be used as a management tool for the least efficient seaports to improve their efficiency through the emulation of the best practice identified. Previous research on seaports has analysed economic efficiency (Cullinane et al, 2002). Technical efficiency growth, also known as technological progress, which is a component of efficiency, has been cited widely as one of the major sources of change in European banking (Berger, 2003). Technical efficiency growth is a decomposition of economic efficiency in terms of pure technical change, non-neutral change and scale augmenting change.

However, until now, no studies have analysed seaports in this light. Most of the empirical literature on seaport efficiency has employed data envelopment analysis (DEA), stochastic production frontier and/or cost frontier with a Cobb–Douglas model (Cullinane et al, 2002). Furthermore, no previous study has decomposed cost growth into technical change and efficient change with a stochastic frontier model.

The motivation for the present research stems from several issues related to seaports. One of the issues is that the European Union's Single Market Program, established in 1992 with the aim of facilitating free movement of goods and services across member states, increases competition among European seaports, thus obliging the least performing seaports to improve efficiency (Barros and Athanassiou, 2004). Another issue arises from successive Portuguese governments' declared intention and subsequent efforts to modernise the country's public administration. This clearly implies that the ways and means by which seaports can improve their performance need to be found and implemented, since a modern, efficient seaport system plays a key role in the transport and distribution infrastructure, to the benefit of all sectors of the economy. However, no coherent plan of modernisation has yet been applied by the seaports' regulatory agency. With this paper, we endeavour to clarify some issues related to Portuguese seaport efficiency, with the purpose of providing findings that may inform the process of determining State policy in this respect. We analyse the performance of Portuguese seaports with a translog cost frontier model, decomposing total technical change.

The paper is organised as follows: in the following section, we describe the institutional setting; next, a review of the existing literature in this field is presented; following this, the theoretical framework is explained and the data and results are presented; subsequently, the seaports' efficiency scores are estimated, followed by the decomposition of technical efficiency; finally, the conclusions are given in the last section.

INSTITUTIONAL SETTING

The Portuguese seaports are run as State enterprises, regulated in a dual system: small seaports are regulated by a public body, the Instituto Maritimo Portuário (Maritime Port Agency), which comes under the direct control of the Ministry of Transport. Large seaports are public enterprises, directly under control of the Ministry of Transportation. The authorities, Ministry of transportation and Instituto Marítimo Portuário, manage those seaports that are situated along the Portuguese Atlantic coast and on the islands of Madeira and the Azores. Table 1 lists the main seaports and the annual total freight handled (loaded and unloaded) by each, in geographical order from north to south.

Table 1 - Movement of freight through portuguese ports.

Full table

The sample used in the analysis represents almost 100% of the total movement of freight through Portuguese ports. Some of the seaport authorities manage more than one port, as is the case for the Madeira Seaport Authority, which manages the ports of Funchal, Porto Santo and Caniçal. Similarly, the Azores Seaport Authority manages the ports of Praia da Vitória, Ponta Delgada, Horta and other minor ports. The Algarve Seaport Authority manages Portimão and Faro. Leixões seaport includes the minor fishing port of Douro, and Setubal seaport includes the minor fishing port of Sesimbra. The data from the Portuguese Transport Statistics are desegregated in Table 1, which prevents analysis by port, but allows analysis by seaport authority.

LITERATURE REVIEW

While there is extensive literature on benchmarking, applied to a wide variety of economic areas, the scarcity with regard to seaports bears testimony to the fact that this is a relatively under-researched area. The literature embraces three scientific methods of quantitative efficiency analysis, namely, ratio analysis, the econometric frontier and DEA.

Song and Cullinane (2001) apply ratio analysis to Asian container terminals. Among the papers using DEA, Roll and Hayuth (1993) present a theoretical exposition and propose the use of cross-sectional data from financial reports in order to render the DEA approach operational; Tronzon (2001) uses cross-section data from 1996 covering four Australian ports and 12 other ports from around the world; Martinez et al (1999) estimate the efficiency of Spanish ports; Barros (2003a) analyses the technical and allocative efficiency of Portuguese seaports; and Barros (2003b) analyses the total productivity change in the Portuguese seaports in two stages: in the first stage, a Malmquist index is estimated, followed by a Tobit regression model estimation in the second stage; Barros and Athanassiou (2004) compare the efficiency of Portuguese and Greek seaports with DEA; finally, Park and De (2004) analyse the efficiency of 11 Korean seaports with DEA.

Among papers using the econometric frontier Baños et al (1999) apply a translog function to Spanish ports; Liu (1995) compares the efficiency of public and private ownership in Britain with a translog function; Coto et al (2000) estimate a translog cost frontier for Spanish ports; Estache et al (2001) estimate a Cobb–Douglas and a translog production frontier for Mexican seaports; Cullinane et al (2002) estimate a Cobb–Douglas production function for major Asian container terminals; and Cullinane and Song (2003) estimate a production function for Korean container terminals.

The variables used in the literature cited are listed in Table 2.

Table 2 - Literature review.

Full table

The general conclusions that emerge from this research are (1) dimensions are important; that location is important too but less so; (2) capital intensity has no significant impact; and (3) private ownership has no significant advantage (Liu, 1995). Moreover, small ports are more efficient than larger ones and autonomy does not make any difference (Coto et al, 2000; Trongzon, 2001); there is overcapitalisation in Spanish seaports (Baños et al, 1999). Finally, action intended to improve the rate of total productivity growth is to be welcomed, as long as it is focused on capital accumulation and on the rate of innovation to shift the frontier of technology, that is, technical change (Barros, 2003b).

Comparing the above-mentioned research with that undertaken in other areas, ports are one of the main areas of economics where frontier models have been applied, with such diverse methods that range from DEA to econometrics, displaying an openness to different approaches that we do not see in other fields. On the one hand, eight papers on DEA and five on econometric frontiers is a small number, compared with other areas of economic research. In addition, there are too many papers that replicate previous research yet making scant methodological improvements.

On the other hand, we observe a growing number of papers with international comparisons, which seems a sound step forward, reflecting globalisation. Finally, we have not yet seen papers applying Fourrier frontiers (Altunbas et al, 2001), input distance functions (Coelli and Perelman, 1999, 2000) or papers using non-traditional DEA models such as the Cone-ratio DEA model of Charnes et al (1990) and the Assurance Region DEA model of Thompson et al (1986, 1990). In the light of the above observations, the present paper is a methodological improvement in this field, since it estimates the efficiency scores similarly to the effect of the Malmquist index in DEA, in order to decompose growth in seaports.

THEORETICAL FRAMEWORK

In this paper, we adopt the stochastic cost econometric frontier approach. This approach, first proposed by Farrell (1957), came to prominence in the late 1970s as a result of the work of Aigner et al (1977), Battese and Corra (1977) and Meeusen and Van den Broeck (1977).

The econometric frontier measures the difference between the inefficient units and the frontier through the residuals. This is an intuitive approach based on traditional econometrics. However, when we assume that the residuals have two components (noise and inefficiency), we have the stochastic frontier model. Therefore, the main issue is the decomposition of the error terms. Let us present the model more formally. The general frontier cost function proposed by Aigner et al (1977) and Meeusen and van den Broeck (1977) is the following:

where C_it represents a scalar cost of the i decision-unit under analysis in period t; Y_it is a vector of output measures, P_it is a vector of input prices and Z_it is a vector of output descriptors used by the ith outlet in period t.

The error term V_it is assumed to be independently and identically distributed and represents the effect of random shocks (noise); it is independent of U_it.

The inefficient term U_it represents the technical inefficiencies and is assumed to be positive and distributed normally with zero mean and variance _U². The U_it positive disturbance is represented by a half-normal independent distribution truncated at zero, signifying that each seaport's cost must lie on or above its cost frontier. This implies that any deviation from the frontier is caused by management factors controlled by the seaport.

The total variance is defined as ²=_V²+_U². The contribution of the error term to the total variation is as follows: _V²=²/(1+²). The contribution of the inefficient term is: _U²=²²/(1+²). Where _V² is the variance of the error term v, _U² is the variance of the inefficient term u and is defined as , providing an indication of the relative contribution of u and v to .

Because estimation procedures of equation (1) yield merely the residual , rather than the inefficiency term u, the latter must be observed indirectly (Greene, 2000). In the case of panel data, such as that used in this paper, Battese and Coelli (1988) used the conditional expectation of u_it, conditioned on the realised value of the error term _it=(v_it-u_it), as an estimator of u_it. In other words, E[u_it/_it] is the mean productive inefficiency of the ith outlet in the chain at any time t. Under the half-normal assumption:

where _i^*=_i+(1-_i)(-_i), and , where is the mean value of the distribution and t is the time period of the panel, is the standard normal distribution and is the respective cumulative distribution function.

Based on the panel data, the maximum likelihood estimators of model (1) are presented in Table 3.

Table 3 - Descriptive statistics of the data.

Full table

With regard to the measurement of technical change, most previous work was based on two methodologies: econometric estimation or index numbers, namely the Malmquist indexes (Baltagi and Griffin, 1988; Fox, 1996).

In this paper, we use the econometric approach, in which we include a deterministic time trend (T) in the estimation of the cost frontier function. The trend may be linear or non-linear and the specification can allow for interactions between time and other explanatory variables, output (Y) and factor prices (P). The coefficients of the time trend are interpreted as measures of technical change. We specify a translog total cost frontier model:

The inclusion of a trend enables the measurement of time-dependent effects in costs, such as the pure technical change, where C is the natural logarithm of total costs; T is a trend; ln Y_i is the natural logarithm of seaport outputs (ships and cargo); and ln P_h is the natural logarithm of the ith input price (wage rate and physical capital rate).

The rate of technical change (TCH) can be measured by the partial derivative of the estimated cost function with respect to the time trend, as follows:

Adopting the terminology of Baltagi and Griffin (1988), equation (4) shows that the rate of technical progress can be broken down into three components: (1) pure technical progress, t₁+t₁₁T; (2) non-neutral technical progress, ; and (3) scale-augmenting technical progress, .

Pure technical change accounts for reductions in total costs, achievable by holding constant the efficient scale of production of any specific mix of outputs and the shares of each of the inputs in total cost. It represents the shift of the cost function geometrically towards the origin and is equivalent to Hick's neutral technical change, where the marginal rate of substitution among factors of production remains unchanged. Pure technical change captures the effects of technical factors, such as learning by doing, managerial and organisational changes and institutional regulation. Non-neutral technical change accounts for changes in the sensitivity of total cost to variations in input prices. This reflects changes in the shares of each of the inputs in total cost, so if the parameter _i is negative, the share of the cost of input 1 towards total cost is decreasing over time. Finally, scale-augmenting technical change reflects changes in the sensitivity of total cost to variations in the quantities of outputs produced. If the _i are negative, the scale of production which minimises average costs for any given output mix is increasing over time.

DATA AND RESULTS

To estimate the production frontier, we used balanced panel data on port authorities in the years from 1990 to 2000 (10 units 11 years=110 observations). The ports considered in the analysis are those listed in Table 1.

The data were obtained from two sources: The Portuguese Transport Statistics and the annual financial reports of the port authorities.

Frontier models require the identification of inputs (resources) and outputs (transformation of resources). It is important for the applicability of the model results and management 'buy in' to the process that the measures of inputs and outputs be relevant, adequately measurable and that appropriate historical data is available. Several criteria can be used in selecting inputs and outputs. The first of these, an empirical criterion, is availability. The second, the literature survey, is a way to ensure the validity of the research and therefore, a criterion to take into account. A further criterion is the professional opinion of managers. In the present paper, we abide by the first two criteria.

We estimate a stochastic generalised Translog cost function. According to the economic theory, this procedure requires information on input prices as well as on total costs and outputs (Varian, 1987). Therefore, we include two input prices (price of labour, price of capital) and two outputs (number of ships and total cargo). A trend is included to capture technical progress, according to equation 4.

We transform the variables according to the description column in Table 3. The logarithmisation of the variables allows for a possible non-linearity of the data set.

Table 3 presents the characteristics of the variables.

The characteristics of the Portuguese seaports are the following: the average number of ships handled is 4,322; the average number of employees is 361; the average cargo handled (loaded and unloaded) is 11,364,205 tonnes. The number of containers handled is 570,934; the average weight of containers handled (loaded and unloaded) amounts to 1,079,689 tonnes; and the average fixed capital (book values) amounts to 24,668,663 (1999=100) euros. These are categorised as small seaports by European standards.

RESULTS

Table 4 presents the results obtained for the stochastic frontier. The variables have been defined and characterised in Table 3. A first step in the estimation procedure is to check the sign of the third moment and the skewedness of the OLS residuals associated with the sample data (Waldman, 1982). The third moment for the terminal frontier is -0.150; the negative sign implying that the residuals of the sample data possess the correct pattern for the implementation of the maximum likelihood estimation procedure used in the frontier models. In the literature survey, we verify that for the most part, the papers on seaport efficiency use the traditional Cobb–Douglas model. A common reason for using this model is the data span used in the analysis. An additional reason stems from statistical tests (usually a likelihood test) that compare the likelihood function value for each model. However, it is recognised that the Cobb–Douglas model possesses several limiting characteristics that make it undesirable whenever we have a data span that allows for the estimation of a less restrictive model such the Translog. Among these undesirable characteristics, the Cobb–Douglas assumes that all seaports have the same production elasticities, the same scale elasticities and unitary elasticities of substitution. The Translog model overcomes these restrictions, being a more flexible functional form. However, the final decision between Translog versus Cobb–Douglas is based on the likelihood test of both functional forms, with the Translog doing a better job in explaining variations in the sample data relative to the Cobb–Douglas function. Table 4 presents the results obtained for the stochastic frontier.

Table 4 - Stochastic translog panel cost frontier (dependent variable log of total cost).

Full table

We verify that the Translog cost function specified above fits the data well, as the R² from the initial ordinary least-squares estimation that was used to obtain the starting values for the maximum-likelihood estimation is in excess of 93% and the overall F-statistic is 364.44. In order to test the suitability of the functional Translog form chosen, we perform a likelihood ratio test with a value of 45.54. This allows the rejection of the null hypothesis that the Cobb–Douglas functional form is preferable to the Translog function.

The fact that the Lagrange test-statistic of 16.539 (critical value equal to 2.5) is significant confirms that we can specify and estimate a stochastic frontier analysis using the Translog cost function.

The variance of the error term ² is large, signifying that it varies along the sample and because the gamma value is defined as: and is high, the inefficiency disturbance u makes a more important contribution to the total variation represented in the error component than do the uncontrolled shocks denoted by v. This result gives rise to the assertion that cost functions on seaports are not supported by the data.

We also verify that the variables have the expected signs, with total cost increasing with the price of labour and price of capital, as well as with the number of ships. This is an intuitive economic result, supported by microeconomic theory (Varian, 1987). Total costs decrease with cargo handled, reflecting the economies of scale in seaport costs. In relation to the trend, we verify that it is not statistically significant, but its square value is, in fact, statistically significant. However, the cross variables with the trend are also not significant, indicating that time-varying inefficiency effects are not the appropriate specification of the stochastic frontier model.

Efficiency rankings

Table 5 presents the results of the efficiency scores computed from the residuals for the years from 1999 to 2000. The individual efficient scores (TE) are calculated according to the following formula, which decomposes inefficiency from statistical noise as follows:

Equation 2 measures technical efficiency scores as the cost of the decision-unit relative to the ideal, or best-practice cost, using the same input mix. The estimation of the denominator needs the decomposition of the individual residuals into their component parts. Jondrow et al (1982) and Battese and Coelli (1988) proposed the decomposition of the residuals for panel data.

Table 5 - Average technical efficiency scores for Portuguese seaports, 1990–2000.

Full table

Technical efficiency is achieved, in a broad economic sense, by the unit which allocates resources without waste and thus, the concept refers to a movement towards, or away from, the best-practice production frontier activity. A movement towards this type of production is improvement, while a movement away from it represents deterioration.

We verify that the mean score is 39.6%. This score suggests that there is waste in the management of seaports. The maximum seaport efficiency score (Sines) is 100%, signifying that it is on the cost frontier, while the minimum efficiency score (Figueira da Foz) is 7.9%, denoting a high level of relative inefficiency. The mean efficiency is 39.6%, while the median inefficiency is 23.3%. Since the median is lower than the mean, we conclude that there is a majority of inefficient seaports below the mean. The standard deviation is 33.89%, which is higher than the mean. Therefore, we conclude that there is some dispersion in the efficiency scores.

These efficiency scores are high when compared with what is found in other industries, such as banking and insurance, but similar to what is found in similar activities, for example, airports, Barros and Sampaio (2004) <R: such as

What is the explanation for such dispersion of the efficiency scores among Portuguese seaports? There are several reasons that are not captured by the frontier model, since it is a classical economic cost regression which explains the traditional factors of production. However, knowledge of reality and simulations allow us to explain the causes that differentiate efficiency scores. The first reason is the scale of operations. Seaports with low scores (Viana do Castelo, Algarve and Figueira da Foz) are of small dimension, with a low level of maritime traffic, with almost no container traffic, but with similar dimensions to the larger ports in terms of administrative and managerial structure, since they are public entities. The seaports of Madeira and Azores quote higher prices than the mainland ports for similar traffic, due to the distance between mainland Portugal and the Atlantic islands. These island ports also have less traffic than the main seaports, while their administrative and managerial structures are, once again, the same. Consequently, these ports attain lower efficiency scores.

DECOMPOSING THE TECHNICAL EFFICIENCY

Table 6 presents the descriptive statistics of total technical change and their components for all seaports according to the expressions presented in Data and Results section.

Table 6 - Descriptive statistics for technical change decomposition.

Full table

We verify that total technical change over the period studied contributed to reductions in seaport costs. Total technical change is explained by its components. First, over time, scale-augmenting technical change has had the largest impact in reducing the Portuguese seaports' costs in the period, signifying that the changes in output mix resulted in a decrease in costs. Second, non-neutral technical change had the second-most important impact in reducing costs, signifying that change in relative prices brought about by such technology has decreased costs, presumably because it has helped facilitate the shift in the inputs mix, with a greater weight on costs. Finally, the pure technical scale change caused the costs to increase, signifying that costs are increasing over time.

In seeking to interpret the results of this study, we can only compare them with results related to banking (Ashton, 1998; Humphrey, 1993), the only sector for which papers in this tradition are found. The results are, in fact, similar, with scale and non-neutral technical change contributing negatively to costs and pure technical change contributing positively. Factors determining this outcome reside, first, in the small level of investment in new technologies; second, an excessively high level of attention to factor prices, namely labour salaries; and finally, the lack of application of new technologies, such as containers, which modified the output mix. However, we conclude that scale is the main factor driving the costs down, with a lesser contribution from factor prices. Pure technical change, captured by the trend coefficient, is driving the costs up.

DISCUSSION AND ECONOMIC IMPLICATIONS

At this juncture, it is appropriate to consider the implication of our study for the public policy towards seaports. Given the lack of scrutiny that the Government and other stakeholders usually place on a public seaport's performance, the seaport managers are not under pressure to show positive financial results. Therefore, the improvement in productivity that could result from greater scrutiny is lost. However, the increasing pressure on seaports, arising from European integration, could enforce the adoption of a more productivity-oriented managerial culture. The key contribution of this paper towards public policy is first to highlight the need to upgrade technical innovation in seaports in order for pure technical change to induce improvements in efficiency. While Portuguese seaports may have a long history, they demonstrate in their modus operandae a resistance to modernisation and innovation. This remains characteristic of much of the country's public and commercial practices, which are the socio-economic and cultural legacies of its 20th century political history. A greater receptiveness to change and initiative, along with the fostering of an awareness and willingness to embrace an enterprise culture, would encourage the abandonment of the traditional resistance and would build strengths necessary to compete with other modes of transportation, as well as with other European seaports. Second, the low level of contribution of factor prices (non-neutral technical change) to costs requires a qualitative investigation into the causes of such dynamics, since it is known that in the recent past, there has been a huge shrinkage in the labour force (non-administrative and non-managerial) of the Portuguese seaports. Given this phenomenon, we would expect a more significant contribution to cost reductions. A cause for concern should be the manner in which the seaport senior managers are appointed by the government, As Mueller (1979) among others, has noted, parties affected by the actions of a public entity may have a strong incentive to influence those actions. The resulting influence induces the public enterprise to act inefficiently in the interest of these parties. Third, related to the contribution of scale-augmenting technical change to cost reduction, the continuous references in the media to mergers of seaport activities are recognition that this is a driving force in seaport management. However, there is a limit to this policy, so innovation should not be delayed. Fourth, the ranking of the seaports demonstrates that there is room for improvement in their management in order for the inefficient seaports to upgrade their performance through benchmarking. Naturally, the political regionalisation of the country can restrict this policy. What is the motivation for the regulatory agency not to oversee rigorously the performance of a public enterprise? Rent-seeking theories (Mueller, 1979) suggest that a government may maintain an inefficient enterprise in order to nominate its 'cronies', thereby wasting public funds.

Finally, there is the question of bad management. Recent evidence to confirm the perceptions of incompetence and inefficiency among Portuguese managers has emerged from an exhaustive survey in this area, carried out jointly by Ad-Capita Recruitment and Research and the Cranfield School of Management, UK. (see PDF report: 'Can Portuguese Management Compete?' at www.adcapita.com).

CONTRIBUTIONS AND LIMITATIONS

In the light of the extensive literature of productivity in seaports, it is useful to consider the potential contributions of the current research. The first contribution is the estimation of a translog frontier model with a trend that allows the desegregation of technical change, disentangled in pure, non-neutral and scale technical change, alongside the inefficient scores. Past studies have only estimated efficient scores mainly with Cobb–Douglas cost functions, or have relied extensively on DEA. This is an improvement over previous research. Our model also lends support to similar works observed elsewhere (Ashton, 1998).

The major limitation of this study concerns the model. Efficiency measurement assumes that the cost function of the fully efficient seaport is known. In practice, this is not the case, and the efficient isoquant must be estimated from the sample data. In these conditions, the frontier is relative to the sample considered in the analysis. Nonetheless, this is valid information because the inputs and outputs that contribute to this inefficiency are also identified (Bessent and Bessent, 1980).

Another limitation is the assumption that seaports have homogenous cost functions. This assumption is disputable, since it implies that the same mix of inputs should be used across different seaports. However, we can always claim that the units are not comparable, and therefore, a ratio analysis equally could not be carried out. Moreover, the data set is relatively short, thus the conclusions are somewhat limited. In order for the conclusions to be generalised, we would need to have a larger panel data set.

CONCLUSION

This article has proposed a simple framework for the evaluation of Portuguese seaports and the rationalisation of their operational activities. The analysis is based on a stochastic frontier model with technical change. Benchmarks are provided for improving the operations of less performing seaports. The general conclusion is that the majority of the seaports analysed are inefficient, therefore adjustment is needed in order to achieve the efficient frontier. Furthermore, scale economies (output mix), as well as non-neutral technical progress (factor price mix), are found to be determinant factors of efficiency in this sector, with pure technical change contributing to the increase of costs.

Turning to the significance of our results, firstly, almost half of the seaports have higher efficiency scores than the median, signifying that the sample does not have a normal distribution. We also verify that the least efficient units are minor seaports. The general conclusion is that the Portuguese Seaport Authority management is failing to improve the efficiency of seaports under its control and that structural limitations exist for the least performing seaports, since they are of small dimensions and consequently, cannot benefit from the same potential flow of outputs as the larger ports. Nevertheless, even certain relatively large ports, such as Lisbon and Leixões, display a high level of waste. The small seaports should be obliged to adjust the level of inputs, since their inefficiency is due to the imbalance between the low level of flow of outputs and a high level of resources allocated to the activity.

Explanations for this mixed evolution may be found in the recognition that public enterprises are prone to rent-seeking behaviour (Mueller, 1979). Rent seeking suggests that a government may maintain seaports in the public sector because of pressure to retain the traditional practice of nominating cronies to senior management posts, thereby wasting vital public funds. In such an environment, the implementation of new technology into the running of the ports may appear to suggest that this falls within the scope of government policy. It is to be remarked that in Portugal, governments routinely fail to oversee properly the performance of public enterprises, and in this context, inefficiencies are bound to be maintained for an extended period, such as the 11 years that this paper has examined.

Further research would serve to clarify these issues.

References

1. Aigner, DJ, Lovell, CAK and Schmidt, P. 1977: Formulation and estimation of stochastic frontier production function models. Journal of Econometrics 6: 21–37. | Article |
2. Altunbas, Y, Gardener, EPM, Molyneux, P and Moore, B. 2001: Efficiency in European Banking. European Economic Review 45: 1931–1955. | Article |
3. Ashton, J. 1998: Technical change in the UK retail banking sector: 1984–1995. Applied Economics 5: 737–740. | Article |
4. Baltagi, BH and Griffin, JM. 1988: A general index of technical change. Journal of Political Economy 96: 20–41. | Article |
5. Baños, PJ, Coto, MP and Rodriguez, AA. 1999: Allocative efficiency and over-capitalisation: An application. International Journal of Transport Economics 2: 181–199.
6. Barros, CP. 2003a: Incentive regulation and efficiency of Portuguese port authorities. Maritime Economics & Logistics 5: 55–69. | Article |
7. Barros, CP. 2003b: The measurement of efficiency of Portuguese seaport authorities with DEA. International Journal of Transport Economics 30: 335–354.
8. Barros, CP and Athanassiou, M. 2004: Efficiency in European seaports with DEA: Evidence from Greece and Portugal. Maritime Economics & Logistics 6: 122–140. | Article |
9. Barros, CP and Sampaio, A. (2004): Technical and allocative efficiency in airports. International Journal of Transport Economics 31: 355–377.
10. Battese, GE and Coelli, TJ. 1988: Prediction of firm-level technical efficiencies with a generalised frontier production function and panel data. Journal of Econometrics 38: 387–399. | Article |
11. Battese, GE and Corra, GS. 1977: Estimation of a production frontier model: With application to the pastoral zone of Eastern Australia. Australian Journal of Agricultural Economics 21: 169–179.
12. Berger, AN. 2003: The economic effects of technical progress: Evidence from the banking industry. Journal of Money, Credit and Banking 35: 141–176.
13. Bessent, AM and Bessent, EW. 1980: Determining the comparative efficiency of schools through data envelopment analysis. Educational Administration Quarterly 16: 57–75.
14. Charnes, A, Cooper, WW and Huang, ZM. 1990: Polyhedral Cone-ratio DEA with an illustrative application to large commercial banks. Journal of Econometrics 46: 73–91. | Article |
15. Coelli, TJ and Perelman, S. 1999: A comparison of parametric and non-parametric distance functions: With application to European railways. European Journal of Operations Research 117: 326–339. | Article |
16. Coelli, TJ and Perelman, S. 2000: Technical efficiency of European railways: A distance function approach. Applied Economics 32: 1967–1976. | Article |
17. Coto, MP, Baños, PJ and Rodrigues, AA. 2000: Economic efficiency in Spanish ports: Some empirical evidence. Maritime Policy & Management 27: 169–175. | Article |
18. Cullinane, KPB, Song, DW and Gray, R. 2002: A Stochastic frontier model of the efficiency of major container terminals in Asia: Assessing the influence of administrative and ownership structures. Transportation Research Part A 36: 734–762.
19. Cullinane, KPB and Song, DW. 2003: A stochastic frontier model of the productive efficiency of Korean container terminals. Applied Economics 35: 251–267. | Article |
20. Estache, A, Gonzaléz, M and Trujillo, L. 2001: Efficiency gains from port reforms and the potential for yardstick competition: Lessons from México. World Development 30: 545–560. | Article |
21. Farrell, MJ. 1957: The measurement of productive efficiency. Journal of the Royal Statistical Society Series A 120: 253–290.
22. Fox, KJ. 1996: Specification of functional form and the estimation of technical progress. Applied Economics 28: 947–956. | Article |
23. Greene, W. 2000: Econometric analysis. 4th Edition. Prentice-Hall: London.
24. Humphrey, DB. 1993: Costs and technical change: Effect of bank regulation. The Journal of Productivity Analysis 4: 9–34. | Article |
25. Jondrow, JC, Lovell, CAK, Materov, IS and Schmidt, P. 1982: On the estimation of technical inefficiency in the stochastic frontier production function model. Journal of Econometrics 19: 233–238. | Article |
26. Liu, Z. 1995: The comparative performance of public and private enterprises. Journal of Transport Economics and Policy 29: 263–274.
27. Martinez, BE, Diaz, AR, Navarro, IM and Ravelo, MT. 1999: A study of the efficiency of Spanish port authorities using data envelopment analysis. International Journal of Transport Economics 2: 237–253.
28. Meeusen, W and van den Broeck, J. 1977: Efficiency estimation from a Cobb–Douglas production function with composed error. International Economic Review 18: 435–444.
29. Mueller, DC. 1979: Public choice. Cambridge University Press: Cambridge.
30. Park, RK and De, P. 2004: An alternative approach to efficiency measurement of seaports. Maritime Economics & Logistics 6: 53–69. | Article |
31. Roll, Y and Hayuth, Y. 1993: Port performance comparison applying data envelopment analysis (DEA). Maritime Policy and Management 20: 153–161.
32. Song, DW and Cullinane, KPB. 2001: Asian container ports and productivity: implication for the new millennium. Singapore and Port Journal Maritime and Port Authority of Singapore. 146–161.
33. Thompson, RG, Langemeier, LN, Lee, C and Thrall, RM. 1990: The role of multiplier bounds in efficiency analysis with application to Kansas farming. Journal of Econometrics 46: 93–108. | Article |
34. Thompson, RG, Singleton, FD, Thrall, RM and Smith, BA. 1986: Comparative site evaluation for locating a high-energy physics lab in Texas. Interfaces 16: 35–49.
35. Tronzon, J. 2001: Efficiency measurement of selected Australian and other international ports using data envelopment analysis. Transportation Research Part A: 107–122.
36. Valentine, VC and Gray, R. 2001: The measurement of port efficiency using data envelopment analysis. Proceedings of the Ninth world conference on transportation Research, Seoul, 22–27 July.
37. Varian, HR. 1987: Intermediate microeconomics: A modern approach. W. W. Norton & Co.: New York.
38. Waldman, D. 1982: A stationary point for the stochastic frontier likelihood. Journal of Econometrics 18: 275–279. | Article |