Telecommunications reform and efficiency performance: Do good institutions matter?

Telecommunications Policy 38 (2014) 49–65

Contents lists available at ScienceDirect

Telecommunications Policy URL: www.elsevier.com/locate/telpol

Telecommunications reform and efficiency performance: Do good institutions matter? Noorihsan Mohamad n Economics Department, Faculty of Economics & Management Sciences, International Islamic University Malaysia, P.O. Box 10, 50728 Kuala Lumpur, Malaysia

a r t i c l e i n f o

abstract

Available online 16 October 2013

Until recently, most studies investigating telecommunication reforms performance have failed to incorporate the importance of institutions into the empirical analysis. This study highlights the importance of institutional governance on telecommunications efficiency and provides empirical results for the impact of institutions on reform outcomes. It provides significant evidence that the institutional environment in which reform progress takes place is an important determinant for successful reform. This study uses the stochastic distance function approach to capture the role of institutions in explaining efficiency differences across 70 countries. The empirical analysis reveals that policy stability in the form of substantive checks and balances on executive power is the most important aspect for successful reform. Independently, legal integrity improves telecommunications efficiency through privatization, while greater freedom from corruption influences the effectiveness of a regulatory body. & 2013 Elsevier Ltd. All rights reserved.

Keywords: Telecommunications Institutions Reform Efficiency Distance function

1. Introduction The world telecommunications industry has undergone significant physical and structural transformation since the early 1980s. Driven by unrelenting technological change and market forces, telecommunications today is one of the world's most dynamic economic sectors (Mayer, Butkevicius, Kadri, & Pizarro, 2003). Equally important is the liberalization and privatization of the telecommunications sector, which has become a veritable worldwide wave of change. To date, the market structure as well as the regulatory framework for the telecommunications sector continues to evolve. The 1990s witnessed the greatest period of policy reform in the telecommunications industry since its inception. National carriers were privatized, new competitors licensed, and an increasing number of new services introduced. During these years, traditional orders of state monopolies were overthrown by the wave of change in the new structure of the telecommunications sector, which gave rise to the convergence of the telecommunications, broadcasting, and information technology industries. Advances in information and communication technology have also brought about new challenges in the regulatory and legislative regimes. Sectoral reform, which began in only a few countries in the early 1980s, intensified and became widespread in the 1990s. The wave of telecommunications reform that started to emerge in a few highly developed economies spread quickly and has reached worldwide proportion. Since the early 1990s, more than 150 countries have at least modified pre-existing telecommunication legislation or introduced new telecommunication acts (ITU, 2008). These various new legislations not only established a new structure for the industry but they also underline the idealistic way in which the telecommunications n

Tel.: +603 6196 4607. E-mail address: [email protected]

0308-5961/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.telpol.2013.04.006

50

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

market should work. In addition, the new legislations also led to the creation of new regulatory bodies as part of the commitment to monitor the sector as well as to protect the welfare of consumers and competitors. As of the early 1990s, less than 40 countries had partially privatized their incumbent telecommunications operator. However, by the beginning of 2000, almost half of the countries in the world had fully or partly privatized their incumbent telecommunications operators. As of 2008, more than 123 out of 191 countries had a private or privatized national fixed-line incumbent (ITU, 2008).1 In essence, the monopoly-based system, which has dominated the world's telecommunications markets for over three-quarters of a century, continues to decline in popularity. Apart from encouraging the trend for a competitive market, the establishment of separate regulators is also considered as one of the most visible signs of sector reform. As market conditions grow tougher and more new services are offered, regulators are needed not only to monitor fair competition, but also to ensure affordable and quality services. In the beginning of the 1990s, there were only 10 independent telecommunications regulators in the world. After a decade, the number of countries that had created a regulatory body exceeded the number of countries that had allowed private investment in fixed-line incumbent operators.2 In spite of the growing consensus that reform is desirable, experience in developing countries, however, has shown that implementing and sustaining workable restructuring is more complicated than in the more developed counterparts (Gasmi & Virto, 2010; Jamasb, 2006).3 Due to over expectations that private sector involvement and reforms will bring increased resources and efficiency for the sector, some delusions concerning these policies has set in (Hoffman, 2008). Indeed, for some developing countries, economic and sectoral performances have even been more disappointing after reforms. A wide array of studies has been done in exploring the relationship between reform and sectoral performance; surprisingly, there is dearth of research which assesses the impact of reforms from the dimensions of efficiency and productivity. While, on the one hand, existing literature continue to support the global pace of reform in promoting teledensity growth and quality services (Fink, Mattoo, & Randeep, 2002; Li & Xu, 2004; Ros, 1999; Wallsten, 2001), all these studies have failed to acknowledge the relevance of diversity in governance and regulatory capacity to sectoral achievement. Since institutions goes hand in hand with the reform process, it is therefore critical to differentiate sources of improvement as well as to explicitly understand its impact in a dynamic setting. Focusing on the impact of reforms and diversity of governance and regulatory capacity, this study addresses contemporary issues concerning success or failures of reforms efforts by looking at how these components affect the varying efficiency levels within the industry and across borders. While concerted efforts have been applied to study the role of reforms and more recently the impact of reforms on private participation and investment, little emphasis has been given to the role of regulatory settings and governance in explaining diversity in efficiency performance, particularly in the telecommunications industry. Considering these neglected issues, this study attempts to complement studies of types of telecommunication reforms from the efficiency performance perspectives. In addition, given the significant contribution of the telecommunications sector, it is therefore proposed that understanding the forces behind telecommunications reform success and failure will prove significant for the industry as well as for other infrastructure industries. In short, this study seeks to answer two vital questions: (1) What benefits, if any, do reforms provide to the productive efficiency of the industry?; and (2) To what extent do variations in regulatory governance and institutional settings affect reforms performance and industrial efficiency? The rest of this paper is organized as follows. Section 2 provides a discussion of reform practices and obstacles in developing countries. Sections 3 and 4 describe the methodology and data used, respectively. Section 5 reports the empirical findings, which comprise the distance function and efficiency estimates. Finally, Section 6 concludes the paper. 2. Reform practices and obstacles in developing countries The privatization, deregulation, and liberalization movement of the 1980s, which started in the United Kingdom and the United States and then extended to Europe and some Latin American countries, has provided a significant amount of useful experience. Though the immediate impetus for reform has come mainly from the inadequate performance of old institutional arrangements, a closer look at some developing countries reveals few obstacles and other amplified motivations. While developed countries have reformed mainly in response to the pressures of large business users who are responding to globalization of the world economy, reform in many developing countries was generally initiated by the government as part of larger structural economic adjustment programs aimed at battling fiscal crises (Petrazzini, 1995). Rather than well-structured reforms with the possibility of market competition, privatization was given the highest priority 1 In the late 1990s, more than 60 percent of the global mobile market and 72 percent of the Internet market were open to competition. These figures jumped to almost 90 and 93 percent, respectively, in 2007 (ITU, 2008), and show no indication of slowing down. 2 To date, the world has more than 150 national agencies to regulate the sector, a figure that has increased by more than 40 percent since the late 1990s (ITU, 2008). In total, this figure corresponds to over 75 percent of 191 International Telecommunication Union (ITU) member countries. 3 Within this background, it is perhaps not surprising that many reforming countries have encountered unexpected problems and have achieved their goals to a varied degree (Estache et al., 2006; Fink et al., 2002).

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

51

to ease the fiscal imbalance (Ramamurti, 1996; Whincop & Rowland, 1998). Sectoral competition, on the other hand, was only limited to a number of value-added services, while domestic and long-distance services remained in the hands of private monopolies (Pisciotta, 1997). In Mexico's case, for example, the privatization of Telmex, which took place in December 1990, generated total revenue of $5.4 billion, making Telmex one of the most hotly traded stocks in Mexico (Singh, 1999). Similarly, the Argentine Government received a total of $214 million in cash and $380 million in 6-year notes (Campo-Flores, 1994) from the privatization of ENTel. Part of this money was used for debt reduction. Nevertheless, the privatization of Telmex was followed by 6 years of monopoly status in domestic and long-distance international services, while ENTel Argentina retained the exclusive right to provide local and long-distance services for a period of 7 years (Ramamurti, 1996). Apart from economic forces, pressures for reform in developing countries continue to mount as a consequence of innovation and increased international trade in telecom services and equipment (Wellenius & Stern, 1994). The pressures which came directly through trade negotiations and indirectly through the recommendations of multilateral aid agencies have compelled many developing economies to open their telecommunications markets. The combined effects of GATTS and WTO negotiations on telecommunications, for example, have led to over 40 developing countries favoring liberalization as part of the plan to modernize their telecommunications sectors (Pisciotta, 1997; Wellenius, 1999).4 Despite the liberalization trends, in many developing countries satisfaction with the reforms is far from uniform. In several cases, liberalization and privatization have been severely criticized for having led to vertically unregulated private monopolies that charged unduly high prices despite minimal quality improvement. In Mexico and Argentina, for example, the first few years of privatization were accompanied by a series of tariff hikes motivated by political pressures (Ramamurti, 1996). Similarly, Telkom of South Africa benefited from a tariff of over 160 percent in the first 5 years following its privatization in 1997 (Gillwald, 2004). Although the number of telecom regulators is on the rise, Petrazzini (1997) argues that regulatory agencies in most developing economies, however, are still heavily involved in the pursuit of more basic socio-economic goals, such as the expansion of basic services, rather than stimulating and fine tuning competition in order to improve quality and expand the sophistication of services. Despite its physical existence in many developing countries, Petrazzini (1997) also contends that politicians or the executive branch still play an important role in the life of regulatory bodies through the appointment of officials in politically sensitive regulatory decisions. Laffont (2005), in contrast, highlights the limitations of accounting and auditing capacity of many regulatory agencies among developing countries. He asserts that the lack of up-to-date technology such as computerized systems and the absence of well-developed auditing systems, has weakened the regulator's role. This is further undermined by the scarcity many regulators in developing countries face with regard to technical and professional expertise, especially in the area of accounting, economic policy, and law (Stern, 2000). Furthermore, reforms have encountered more significant difficulties than anticipated. Poorly designed regulatory frameworks and lack of government commitment are among the problems many developing countries face. In Africa, for instance, the reforms are often obscured by regional political conflicts that limit a government's ability to make policy commitments to promote telecommunications development (Berg, 2002). Likewise, in many Latin American countries, the reform policies are usually incoherent and often made in an ad hoc and decentralized manner due to the high-rotation of ministerial and secretarial posts in offices (Spiller & Tommasi, 2003). With fragile institutional endowment, such as lack of checks and balances, low legal integrity, and widespread corruption that commonly persist in many developing countries, the effectiveness of reform becomes a serious predicaments (Smith & Wellenius, 1999). Notwithstanding these limitations, Kessides (2004) argues that many countries have hastily imported and adopted universal reform templates, assuming implicitly that they would work locally as they do in their developed counterparts. Although reform strategy might be viable for some, this approach, however, makes less sense to many in developing countries where the necessary institutions do not exist or are poorly developed. Given that telecommunications reform takes place within diverse institutional settings and frameworks across borders, it is, therefore, essential to explore the link between regulatory reforms and governance as well as the latter's contribution to the efficiency of sectoral performance.

3. Methodology The traditional approach to econometric measurement of efficiency has been to estimate production technology. In the case of telecommunications, the majority of studies have either used production, cost, or profit functions as alternative methods. In view of the inherent feature of the telecommunications industry that offers multiple services, this study uses the distance function developed by Shephard (1970) to describe production technology in multi-output and multi-input settings. 4 While 69 countries pledged to liberalize the market under the WTO Basic Telecommunications Agreement, not all governments agreed to the same form of liberalization; neither did all governments agree to liberalize at the same pace. For instance, of the 69 nations making commitments on voice telephony services, only 36 governments made immediate commitments while the rest opted for phased-in entry, some until as late as 2013.

52

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

As in the case of infrastructure industries in general and the telecommunications industry in particular, operators and providers of telecommunications services have more control over inputs and are particularly obliged to fulfill the required level of demand.5 In other words, the supply level of outputs or services has little influence on production decision making (Coelli, Rao, O'Donnell, & Battese, 1998). Consequently, the best option for service providers is to control their inputs to achieve the optimal production level. Hence, input orientation is chosen to represent distance function in measuring efficiency and productivity of the telecommunications industry for this particular study. This modeling choice is consistent with Coelli, Estache, Perelman, and Trujillo (2003), who argue that input distance functions are the appropriate specification in network industries where the demand is outside of the control of managers. Formally, the input set consists of all input vectors xt that can produce a given output vector yt which can be expressed as LðyÞ ¼ fx∈RM þ : x can produce yg

ð1Þ

The value of the input distance function is exactly the inverse of Farrell's (1957) input-oriented technical efficiency measure, which can be defined on the input set Lt(yt) as 
  x ∈LðyÞ DI ðy; xÞ ¼ max ρ : ρ

ð2Þ

where the subscript I indicates an input distance function and ρ denotes the extent to which the input vector can be proportionally contracted while holding the output vector constant. It follows that DtI ðyt ; xt Þ ≥1 if xt ∈ Lðyt Þ and DtI ðyt ; xt Þ ¼ 1 if xt is located on the inner boundary of Lðyt Þ. In other words, the input distance function will take a value of unity if x is located on the inner boundary of the input requirement set Lðyt Þ.6 The deviation of the input distance function from (1) is attributed to technical inefficiency. The deviation can be adjusted through the specification of a function hð⋅Þ, such that the stochastic input distance function can be expressed as follows: 1 ¼ DI ðx; yÞ⋅hð⋅Þ

ð3Þ

As described in Lovell (1993), the inverse of the input distance function can be represented as a technical inefficiency measure. Literature on stochastic frontier analysis has usually described the inefficiency term, hð⋅Þ, by exp (−uit), and it is an element of the two-part error term. Since the input distance function has a general property of homogeneity of degree one in inputs, the original distance function can be transformed by imposing the homogeneity restriction for computational convenience. This is specifically achieved by normalizing all the inputs by an arbitrarily chosen input in the numerator. When a symmetric (noise) error term, exp (v), which is another component of the two part error term, is added to the function in a multiplicative formulation, Eq. (3) is transformed to yield the following estimable equation:
 1 x ¼ DI ; y ⋅ev−u xk xk

ð4Þ

Given that this study employs a panel dataset with I countries (i ¼1, …, I) and T periods (t¼1, …, T), vit is an error term that is assumed to be independently and identically distributed as normal with a zero mean and variance s2v . Following Battese and Coelli (1995), the technical inefficiency term uit is assumed to be a non-negative random variable, independently distributed as truncations at zero of the Nþ ðμ; s2u Þ and its mean, μ, is explained by a set of explanatory variables. More specifically, μ is specified as S

μit ¼ δ0 þ ∑ δs zit

ð5Þ

s¼1

where z is a vector of explanatory variables, e.g. institutional and organizational variables that are hypothesized to influence the relative efficiency, and δ is a parameter vector. A predictor of technical efficiency of each entity can be obtained using the conditional expectation as outlined in Battese and Coelli (1993), which is equivalent to TEit ¼ E½expð−uit j vit −uit Þ

ð6Þ

For this study, translog functional form has been selected to estimate the distance function since it fulfills several desirable properties, including flexibility, ease of calculation, and the possibility of imposing homogeneity (Coelli et al., 1998). Following Coelli, Perelman, and Romano (1999), and allowing for exogenous factors, the general formulation of the 5

This is consistent with the ever-growing requirement for universal service. Assuming that the feasible input set Lðyt Þ conforms to basic regularity axioms as outlined by Fare and Primont (1995), it can be readily shown that the input distance function is linearly homogeneous in input x, monotonically decreasing in outputs and monotonically increasing in inputs as well as convex in outputs and concave in inputs. 6

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

53

stochastic input distance function for three inputs and three outputs can be expressed as DIit ln xKit

!

3

¼ α0 þ ∑ αm lnymit þ m¼1

2

þ ∑

3

∑ ηkm

k¼1m¼1

3 2 1 3 lnxkit 1 2 2 lnxkit lnxlit þ ∑ ∑ βkl ∑ ∑ αmn lnynit lnymit þ ∑ βk 2m¼1n¼1 2 lnx lnx Kit Kit lnxKit k¼1 k¼1l¼1

2 3 5 lnxkit lnxkit 1 lnymit þ ∑ ϕkt t þ ∑ ϕmt lnymit t þ θ1 t þ θ11 t 2 þ ∑ κr Reg 2 lnxKit lnx Kit m¼1 r¼1 k¼1


 x −lnxKit ¼ TL yit ; kit ; t; Reg −lnDIit xKit

ð7Þ

where k and m indexes inputs (total investment, number of radio base stations and number of labor) and outputs (fixed-line subscribers, mobile subscribers and internet subscribers) respectively, while the Greek letters denote the coefficients to be estimated. Note that the additional variable Reg refers to the vector of regional dummies, t refers to time trend variable to capture non-neutral technological change and TL corresponds to translog functional form.7 In other words, the empirical model for this study includes an additional variable for regional dummies as part of the distance frontier. This is done to account for potential differences coming from regional-specific aspects of the telecommunications technology. Given that –lnDIit corresponds to the level of inefficiency, –lnDIit can then be replaced by uit (inefficiency term), and adding a random error term vit , Eq. (7) can thus be written as
 x −lnxKit ¼ TL yit ; kit ; t; Reg þ vit −uit xKit

ð8Þ

which is a standard stochastic frontier specification whereby the distributional assumptions on the error term, uit and vit , have been defined previously and the function is estimated using the maximum likelihood method.8 One important extension of the stochastic frontier literature has been the ability to incorporate determinants of inefficiency. For this study, this technique is particularly useful in determining the relative inefficiency among countries with regard to reform process and diversity in institutional practices. The inefficiency effect framework is specified as follows: uit ¼ δ0 þ δ1 Priit þ δ2 Ent it þ δ3 Bodyit þ δ4 Polit þ δ5 Leg it þ δ6 Cor it þ δ7 ðPriit  Bodyit Þ þ δ8 ðPriit  Polit Þ þδ9 ðPriit  Legit Þ þ δ10 ðPriit  Cor it Þ þ δ11 ðBodyit  Polit Þ þ δ12 ðBodyit  Legit Þ þ δ13 ðBodyit  Cor it Þ þδ14 nonOECD þ δ15 OECD þ δ16 t þ δ17 Roadit þ δ18 Urbanit

ð9Þ

where Pri, Ent, and Body refer to the reform variables which correspond to privatization, share of state enterprises, and independent regulator, respectively. On the other hand, institutional features are captured by Pol (policy constraint), Leg (legal integrity), and Cor (freedom from corruption). Non-OECD and OECD are dummy variables for high-income countries, with middle income countries being the reference group. Road and Urban, on the other hand, refer to road density and percentage of urban population, which act as control variables, while δi are the respective inefficiency coefficients to be estimated.9 The model parameters are estimated using the FRONTIER 4.1 program developed by Coelli (1996). Prior to estimation, all the relevant data are expressed in deviations from the geometric mean as this makes it possible to interpret the estimated first order parameters as production elasticities evaluated at the sample mean (Coelli et al., 2003). Moreover, according to Kumbhakar and Lovell (2000), this transformation provides additional convenience in reporting without altering the performance measure. Detailed descriptions of the data are presented in the next section. 7 The region refers to the geographic location of the country, i.e. East Asia and the Pacific, Europe and Central Asia, Latin America and the Caribbean, the Middle East and North Africa, or North America, with Sub-Saharan Africa being the reference group. 8 Certain regularity conditions as required by economic theory must be met. Linear homogeneity in inputs is held by imposing the adding-up restrictions as follows:

K

∑ βk ¼ 1;

k¼1 K

∑ βkl ¼ 0;

l¼1 K

∑ βkm ¼ 0;

k¼1

k ¼ 1; 2; …; K; m ¼ 1; 2; …; M

Conditions for symmetry of the cross-effects are imposed by restricting the parameters as αmn ¼ αnm; βkl ¼ βlk

m; n ¼ 1; 2; …; M k; l ¼ 1; 2; …; K

9 Observe that the inefficiency effect model in Eq. (9) includes interaction terms between reform and institutional features. This is done to determine how the institutional supporting structure may impinge on the credibility of the reform process.

54

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

4. Data 4.1. Sample The data used in this study consists of a panel of 70 countries’ annual data over a 25-year period from 1980 to 2004. The sample countries selected cover six continents and include developing countries as well as their developed counterparts. This wide country coverage, according to Carrington, Coelli, and Groom (2002), gives an advantage especially in providing a closer approximation to the ‘world best practice’ production frontier. For estimation purposes, the sample countries are determined by the criterion that consistent telecommunications data as well as other statistical data representing the inefficiency effect variables were available throughout the period of the study. These include data on reform, governance, and institutions as well as demographic indicators. The primary data source for this study is the World Telecommunications Indicators 2006 (ITU, 2006), which reports various telecommunications statistics. The database was compiled and collected from an annual questionnaire sent out by the Telecommunication Development Bureau (BDT) of the International Telecommunications Union (ITU) to each member country's telecommunications authority. To date, the database contains time-series data for around 100 sets of telecommunications statistics for over 200 countries from as early as 1975. These include statistical data on telephone network size and dimension, mobile and Internet services, quality of services, number of total telecommunications staff, tariffs, revenue, as well as total telecommunication investment. 4.2. Inputs and outputs This study employs three outputs and three inputs in modeling the stochastic frontier multi-output/multi-input distance function. The chosen output specification focuses on the three main telecommunications services, reflecting the heterogeneity of the telecommunications industry. In particular, total number of subscribers has been chosen to act as the most pivotal indicator for telecoms’ sectoral development (Estache, Goicoechea, & Manacorda, 2006).10 The outputs adopted for this study are (i) total fixedline subscribers ðy1 Þ; (ii) total mobile subscribers ðy2 Þ; and (iii) total Internet subscribers ðy3 Þ. The input specifications on the other hand, are (i) total capital investment in telecommunications ðx1 Þ; (ii) total capacity of local public switching exchanges ðx2 Þ; and (iii) total labor ðx3 Þ. Total capital investment refers to the expenditure associated with acquiring ownership of a telecommunications equipment infrastructure (including supporting land and buildings and intellectual and non-tangible property such as computer software). These also include expenditure on initial installations as well as additions to existing installations.11 The total capacity of public switching exchanges corresponds to the maximum number of fixed and mobile lines that can be connected. This number includes fixed and mobile telephone lines already connected and telephone lines available for future connection. As reported by the Telecommunications Industry Association (TIA, 2000), these crucial telecommunications equipments not only act as switching and transmission mechanisms but have also been regarded as the primary source of growth for telecommunications industries. Finally, total labor represents total full-time staff employed by telecommunications network operators in the country for the provision of public and mobile telecommunication services. However, this does not include employees working in postal services or broadcast operations, as they are not working principally for the provision of telecommunications services. Appendix A presents the summary statistics of outputs and inputs for the overall sample over the period 1980–2004 as well as the descriptive statistics over time (within variation) and across countries (between variation). 4.3. Inefficiency factors In order to explore the impact of reforms on efficiency performance and to determine how diversity in institutional and political settings affects the overall efficiency level, this study adopts three distinct categories of inefficiency factors.12 The following subsection details the inefficiency variables used for this study as highlighted in the methodology section. 4.3.1. Reform indicators The first set of inefficiency variables is meant to capture reform and the liberalization process each sample country has undertaken. Following the literature on telecommunications reform, two components of reform events were considered for this study: privatization and the establishment of a regulatory body to monitor the industry. Both variables are presented as a binary variable corresponding to a value of 1 since the year of inception, if a country undertook the reform process 10 Alternative measures for total number of subscribers are network penetration and telephone usage (in minutes). Wallsten (2001), however, points out that the former measure may over or underestimate network access because it is not possible to differentiate if one has multiple lines or if one line is used by groups of people. Although the usage of telephones (in minutes) presents a good indicator for network traffic and congestion, this alternative measure, however, is best to measure efficiency at the retail level rather than for international comparison (Lupi, Manenti, Sciala’, & Varin, 2011). 11 These costs are deflated using purchasing power parity with year 1996 as the base year obtained from Heston, Summers, and Aten (2006). 12 All the inefficiency variables used in this study are neither inputs nor outputs to the production process. Rather, all the inefficiency variables used in this study characterize the environment in which production occurs. Hence, they are all exogeneous.

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

55

respectively. All information and data regarding reform and the liberalization process were compiled and obtained from the Regulatory Knowledge Centre, Telecommunications Development Bureau of ITU.13 Moreover, to accommodate the presence of competition, which is believed to act as a catalyst for successful reform, a sub-component of the Economic Freedom Index (EFI) is used to proxy the level of competition.14 In particular, this study employs the degree of state-operated enterprises index as a proxy for competition obtained from Gwartney, Lawson, Sobel, and Leeson (2007).15 The degree of state enterprises index reflects the level of protectionism that shields industries from competition. Hence, a higher degree of state intervention indicates less competition and vice versa. The variable is placed on a scale from zero to 10, with a greater value being higher economic freedom. 4.3.2. Institutional indicators The second set of inefficiency variables reflects the institutional and governance quality of sample countries. The choices of institutional variables are based largely on the framework of institutional endowment Levy and Spiller (1996) proposed. Similarly, to ensure that the institutional features resemble the important nature of telecommunications reform, this study closely follows the elements of institutional governance highlighted by the telecommunications literature. In sum, three institutional variables are considered for this study. First, the study takes into account the importance of a credible policy regime as proposed by Levy and Spiller (1996). This indicator is proxied by the data obtained from the political constraint (POLCON) database developed by Henisz (2000). In essence, the POLCON index identifies underlying political structures and measures the political ability to support credible policy commitments. The index ranges from 0 to a theoretical maximum of 1. The highest index value indicates extreme constraints on the executive to change policies autonomously, while zero index values correspond to political systems with unchecked executive power (dictatorships, absolute monarchies, and other autocracies). Secondly, this study considers the importance of a credible judiciary system as part of the institutional features. This institutional variable is crucial, particularly in providing some insights on how the process of upholding contracts and securing property rights affects overall productivity and the reform process (Estache & Wren-Lewis, 2009; Sinha, 1995). To capture this feature, the study adopts the integrity of the legal system index obtained from the Fraser Institute database, which published the EFI. On a scale of zero to 10, high index values correspond to a strong legal foundation, while low index values represent weak constitutional control or delicate rule of law. Given the preponderance of empirical studies that highlight the negative implications of corruption on economic outcomes, this study also takes into account the cost of corruption to explain the inefficiency effects.16 In particular, this study adopts the component of freedom from corruption obtained from the Index of Economic Freedom database published by the Heritage Foundation. The database reports the freedom of corruption index on a scale from zero to 100. Essentially, a high index value corresponds to greater economic freedom. Similarly, a higher value can be interpreted as having a low level of corruption or anti corruption. On the other hand, a low value for the index represents a high degree of corruption.17 4.3.3. Demographic indicators In addition to the reform and institutional indicators, several other control variables are also included in the study. Essentially, to capture how mobility and access to markets affect efficiency of the sector, two demographic variables are included. These are road density and urban population. An increase in both road density and the share of the population living in urban areas can be expected to be associated with lower cost of network deployment, thus inducing higher investment by the service providers. The data for both variables are from the World Development Indicator 2006, published by the World Bank for the period 1980–2004 (The World Bank, 2006). Appendix B presents a descriptive summary of all the inefficiency variables discussed above. 5. Findings 5.1. Discussions of parameter estimates A parametric input distance function was estimated assuming a stochastic translog technology, as indicated in Section 3. Homogeneity of degree +1 was imposed by selecting one of the inputs, total telecommunications investment ðx1 Þ as the 13

The regulatory database can be accessed at http://www.itu.int/ITU-D/ICTEYE/Regulators/Regulators.aspx. Guiterrez and Berg (2000) viewed the EFI as a proxy for serious reform initiatives for his study in measuring telecommunications performance. Moreover, there have been relatively large numbers of literature on telecommunications that employ Economic Freedom Index to measure sectoral efficiency and growth (see for example Andova (2006), Hamilton (2001), and Uhlenbruck, Rodriguez, Doh, and Eden (2006)). 15 Initial attempt to estimate an alternative model taking into account the number of operators (as a proxy to capture level of competition in telecommunications industry) has resulted in the inconformity of the monotonicity conditions. This was indicated by the incorrect signs of the first order outputs and inputs which may lead to perverse conclusions concerning the effects of relative efficiency levels (O’Donnell & Coelli, 2005). 16 Example of studies which analyze the effect of corruption on telecommunications performance include the work by Estache, Goicoechea, and Trujillo (2009), Bailard (2009), Estache et al. (2006), and Uhlenbruck et al. (2006). 17 Given the diverse statistical scale of the variables considered, the index for POLCON and freedom from corruption has been rescaled on a scale from zero to 10. 14

56

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

dependent variable, and the ratio of public switching exchanges to total telecommunications investment ðx2 =x1 Þ and total labor to total telecommunications investment ðx3 =x1 Þ as the explanatory variables as described by Eqs. (7) and (8). Table 1 presents the parameter estimates using maximum likelihood. The empirical results indicate that the model is well-behaved.18 All the first-order outputs and inputs have correct signs. In particular, the output coefficients are negative and the inputs coefficients are positive, which conforms to the theoretical framework.19 In essence, this implies that the distance functions satisfy the monotonicity condition with respect to the outputs and also the monotonicity and concavity conditions with respect to inputs at the means, respectively. Moreover, the estimates of s2 are significant, supporting the theory that stochastic distance function, rather than the deterministic ones, is appropriate. A significant value for the estimated γ, on the other hand, signals the importance of inefficiency effects for the performance measurement. Recalling that γ ¼ ðs2u =s2v Þ þ s2u , the highly significant estimate of 0.9048 for this parameter suggests that the estimated deviation from the frontier is mainly due to inefficiency.

5.2. Distance elasticities Various measures that summarize production processes and act as performance indicators can be constructed from the estimated model. One of the most useful for the interpretation of the empirical estimates is the duality between the cost and input distance functions. Formally, the individual input elasticity corresponds to the input expansion required for increase in output, while the derivative of the input distance function with respect to each output implies the negative cost elasticity of specific output. Similarly, the former can be interpreted as the input share of the output, relative to a specific input and the latter refers to cost shares or the relative importance of each output. As indicated by Table 1, the input elasticity estimates imply that the elasticity with respect to capital investment ðx1 Þ is the highest (0.4462). This implies that the cost of capital expenditure represents nearly 45 percent of total cost at the sample mean.20 This is not an unexpected result given that the telecommunications sector is considered a capital-intensive industry that requires huge initial capital as well as ongoing infrastructure investment. On the other hand, significant share of labor cost bears more concern. In particular, a huge share of labor cost (0.3394) relative to total capacity of local public switching exchanges (0.2144) might imply unnecessary input redundancy or over staffing. The output elasticities of the distance function are highly significant for all the outputs. Essentially, these results suggest that an increase in the number of subscribers of any of the three services is expected to increase the cost. The estimates of the output elasticities in Table 1 indicate that the cost elasticity of fixed-line subscribers (0.4614) is almost three times larger than the corresponding elasticity for mobile (0.1739) and Internet (0.1995) services. In essence, this result means that a 10 percent increase in fixed-line services, results in a 4.6 percent increase in total cost, whereas the corresponding figure for mobile services is only 1.74 percent. Given the differences in capital for these two services, this result is not surprising.21 Likewise, for Internet service provision, a 10 percent increase in subscribers results in an increase in total cost of almost 2 percent. The implications for the tests on the estimated parameters for technological change, on the other hand, analyze how the technologies evolved over time. Theoretically, the derivative of the input distance function with respect to time is equal to the elasticity of cost reduction, and provides a dual measure of the speed of technological progress. As the estimate of θ1 from Table 1 is 0.0158 and statistically significant, this suggests that the sample average country has experienced an annual rate of technological change of almost 1.6 percent in the mid-period 1980–2004. In other words, the estimated value implies that less input per unit of output over time was used over these periods. Likewise, the inclusion of regional dummies allows the estimated distance function of each country's telecommunications sector to shift in relation to the distance functions of the arbitrarily chosen base or reference group. Three out of five estimated coefficients for regional dummies are statistically significant, indicating that the intercept of each estimated distance function is shifted by the country's geographical location vis-a-vis the reference group (Sub-Saharan Africa). These regions are East Asia and the Pacific, Europe and Central Asia, and North America. Contrary to this result, the estimated coefficients for Latin America and the Caribbean as well as the Middle East and North Africa are statistically insignificant, justifying the assumption that these regions are not distinguishable from Sub-Saharan Africa.

5.3. Inefficiency effects The estimated coefficients in the inefficiency model are of particular interest to this study. The results of the inefficiency effects model are used to explore the relationship between reform and institutional governance and its implication for telecommunications’ efficiency. The estimated coefficients of the inefficiency effects variables are listed in the second part of 18

A series of reliability tests are also implemented using the generalized likelihood ratio statistic ðλÞ proposed by Battese and Coelli (1995). Note that the dependent variable for the distance function is negative. Elasticity of input for total capital investment is calculated based on homogeneity conditions. 21 In network industries, the sunk cost of an installed base is very high. As the size of the overall physical network increases with the addition of mobile telephone interconnections, the marginal cost declines. 19

20

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

57

Table 1 Maximum likelihood estimation of the input distance function. Parameter Frontier α0 Outputs α1 α2 α3 α11 α22 α33 α12 α13 α23 Inputs β2 β3 β22 β33 β23 Inputs–outputs η21 η22 η23 Inefficiency δ0 Reforms δ1 δ2 δ3 Institutional δ4 δ5 δ6 Interactions δ7 δ8

Coefficient

t-Ratio

Constant

−0.5402

−10.8400

y1 y2 y3 y11 y22 y33 y12 y13 y23

−0.4614 −0.1739 −0.1995 −0.0129 −0.0623 0.2055 0.0955 −0.1618 −0.0407

−24.8961 −7.9677 −12.3657 −1.9879 −5.4761 9.7104 8.9021 −13.9533 −3.8938

x2 x3 x22 x33 x23

0.2144 0.3394 −0.0835 −0.2016 0.1365

16.9913 21.6856 −8.3821 −11.5470 11.3704

x2y1 x2y2 x2y3

−0.0351 −0.0284 0.0020

−4.1443 −3.9533 0.1917

Constant

−6.3481

−10.2586

Pri Ent Body

1.9945 −2.0515 0.3504

7.3156 −8.9570 2.8257

Pol Leg Cor

0.1657 −1.8634 0.0426

1.1748 −7.2031 1.8457

PrinBody PrinPol

−0.2661 −0.5563

−2.7007 −3.2085

4.5666 0.9048

11.9111 76.6315

Other parameters s2 Sigma-square γ Gamma

Parameter

Coefficient

t-Ratio

η31 η32 η33 Technical change φ1x φ2x φ1y φ2y φ3y θ1 θ11 Regional κ1 κ2 κ3 κ4 κ5

x3y1 x3y2 x3y3

0.0372 0.0124 −0.0100

4.8600 1.4935 −2.2370

x2t x3t y1t y2t y3t t t2

0.0084 −0.0046 −0.0048 0.0039 0.0217 0.0158 0.0045

3.5583 −1.5567 −0.1670 1.2414 5.6554 3.1994 3.3738

Reg1 Reg2 Reg3 Reg4 Reg5

−0.0226 0.0575 0.0164 −0.0915 0.2533

−3.0556 2.1297 0.2853 −1.4270 3.1618

δ9 δ10 δ11 δ12 δ13 Others δ14 δ15 δ16 δ17 δ18

PrinLeg PrinCor BodynPol BodynLeg BodynCor

−0.2008 −0.0445 −0.5598 −0.0159 −0.4933

−3.1957 −0.7819 −4.2057 −1.7896 −11.8366

NOECD OECD t Road Urban

0.0919 −0.4984 −0.4133 −0.2491 −0.4262

1.3488 −6.6734 −15.1778 −9.6460 −7.0707

Log-likelihood LR test of one-sided

−1957.9237 401.6775

Table 1. It is important to note that in the context of Battese and Coelli's (1995) model, a negative (positive) sign coefficient implies improvement (deterioration) in efficiency.22 While the information provided in Table 1 is informative, it remains somewhat limited. Simply looking at the coefficient of the reform policy of interest would not provide a correct assessment of the average total effect of reform policies across countries.23 Therefore, for ease of interpretation all variables are algebraically transformed to center around their means equal to zero prior to the estimation. Thus, the centered marginal effect captures the marginal effect of reforms when the interacted institutional variable is at its mean.

5.4. Impact of reforms The result of the estimated coefficient for Priv indicates that privatization has a significant negative effect on efficiency. Assuming no regulatory body, the result indicates that a hypothetical country which privatized is empirically associated with a 1.9945 point decrease in efficiency. Unlike privatization, the effect of competition as proxied by the share of state 22 Given the nature of the model, which involves multiplicative interactions between variables, the estimated coefficients should be interpreted with caution. In essence, one should refrain from interpreting the constitutive elements of interaction terms as unconditional or average effects. For instance, the coefficients on the constitutive terms, privatization or regulatory body, must not be interpreted as the average effect of a change in reform policy on efficiency as it can in a linear additive regression model. Instead, the effects of reforms on efficiency vary depending on the coefficients with which the privatization and regulatory body variables are interacted. 23 Looking at the reform coefficients at the bottom of Table 1 would only convey information about the effect of reform policies when the institutional indexes are zero.

58

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

enterprises (higher values correspond to greater competition) contributes to improvement in efficiency with an increase of approximately two points. On the other hand, restructuring the telecommunications industry by establishing a regulatory body implies better performance in terms of efficiency compared to privatization. Considering other variables at their respective means, this finding indicates that a hypothetical country with a regulatory body would reduce efficiency by only 0.3504 points, nearly 1.65 points lower than the case of privatization. Despite the distinctive negative impact of either reform on efficiency, the coefficient on the interaction between privatization and regulatory body ðPri  BodyÞ suggests that together they exert greater beneficial effects on telecommunications efficiency.24 Recognizing that the traditional table of results appearing in Table 1 can throw only limited light on the hypothesis of how distinctive institutional features have influences across different countries, the paper also present figures that illustrate how the marginal effect of privatization (Fig. 1) and regulatory body (Fig. 2) on efficiency changes across the observed range (individual values) of institutional factors.25 As displayed by Fig. 1, all subfigures for the marginal effects of privatization have significant reductive effects on inefficiency over the entire range of the respective institutional variables, with the exception of panel (a). As Fig. 1(a) depicts, privatization stops having a statistically significant reductive effect on inefficiency once the POLCON index is greater than seven. Given that the marginal effect of privatization is significant over the policy constraint values from zero to seven, this implies that a political system based on checks and balances reduces sectoral inefficiency through the privatization process. Nevertheless, no conclusion can be drawn with respect to countries classified as high in political constraint with two legislative chambers. Fig. 1(b) and (c) illustrate the influence of legal integrity and freedom from corruption on the marginal effect of privatization, respectively. As predicted, the effect on inefficiency declines as the index of legal integrity increases. This is indicated by the statistically significant negative sloping of the marginal effect line over the entire range of the legal integrity index. This result implies that transparent and effective rule of law, particularly on contracts and property rights, plays a significant role in determining the effectiveness of privatization. This finding supports the theoretical framework highlighted by Sinha (1995) and Cherry and Wildman (1999) that highlight the importance of judicial factors in achieving fruitful telecommunications liberalization. Similarly, the marginal effect of privatization is positive and significant across the range of freedom from corruption as indicated by Fig. 1(c). Initial investigation indicates that this supports the prediction that countries with high freedom from corruption correspond to lower inefficiency compared to countries seen to be highly corrupt. Although the estimated coefficient is negative as predicted, it lacks statistical significance. In other words, the nearly flat marginal effect line indicates that the magnitude of ðPri  CorÞ is substantively trivial and thus there is no appreciable interaction between privatization and freedom from corruption. The findings of this study regarding corruption and privatization are similar to those of Estache et al. (2006). In their study, they report an insignificant effect of corruption on privatization, particularly on labor productivity and quality of the telecommunications industry. Interestingly, they also highlight that countries that had not reformed, perform significantly better in the presence of corruption. Few other studies also report conflicting statistical evidence with regard to this. For instance, while studies such as Manzetti (1999) and Black, Kraakman, and Tarassova (2000) report that lower corruption leads to greater telecommunications access, Estache et al. (2009) reports otherwise. In contrast, Fig. 2 reports the marginal effect of a regulatory body over the entire range of institutional variations. Fig. 2(a) depicts the marginal impact of a regulatory body on efficiency over a range of political constraints index. Although the marginal effect of a regulatory body appears to improve in efficiency as policy constraints increase, it should be noted that this marginal effect is statistically significant only for a particular group of countries. The first group applies to countries that have an extremely low POLCON index (in the range of zero to one), and the second group corresponds to countries where the POLCON index is higher than five. Likewise, while the former group refers to countries with unchecked executive power (i.e. a dictatorship or absolute monarchy), the latter refers to countries with checks and balances with at least one legislative chamber. Fig. 2(b) illustrates the findings with respect to the impact of legal integrity on the marginal effect of regulatory body. Despite the fact that the marginal effects of regulatory body are statistically significant over the entire range of the legal integrity index, the estimated coefficient for ðBody  LegÞ, however, is statistically insignificant. While the interaction between regulatory body and legal integrity is negative as predicted, the nearly flat marginal effect line indicates that the magnitude of the interaction is barely important. In other words, there is no clear evidence of apparent interaction between regulatory body and legal integrity. This finding is quite surprising given the fact that a number of theoretical studies on regulatory reforms emphasize the existence of reliable law and practices for an effective regulatory framework. Wellenius and Stern (1994) and Kessides (2004), for instance, reiterate the vital elements of effective legal institutions for the effective development of a regulatory body for successful reform. Nevertheless, rather than highlighting the importance of legal integrity, many other authors focus more on the internal capacity and organizational framework of a regulatory body. For example, Stern (2000) presents

24 Note that these results represent only the case of a hypothetical country. The results do not indicate whether reforms have a statistically significant impact on efficiency when the effective institutional features are greater or less than the means. 25 Note that the interaction coefficient has a ‘two-way’ interpretation. This is always true as the interaction term has a symmetric properties.

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

a

59

b

c

Fig. 1. The effect of privatization on telecommunications’ efficiency. (a) Impact of political constraint on the marginal effect of privatization, (b) impact of legal integrity on the marginal effect of privatization and (c) impact of corruption on the marginal effect of privatization.

Fig. 2. The effect of regulatory body on telecommunications’ efficiency. (a) Impact of political constraint on the marginal effect of regulatory body, (b) impact of legal integrity on the marginal effect of regulatory body and (c) impact of corruption on the marginal effect of regulatory body.

the case emphasizing the capacity of legal skills rather than the rule of law itself. He highlights the case of many developing countries, such as Botswana and Romania, where the very limited availability of people with the appropriate skills remains a major issue for an effective regulatory body.26

26 Apart from emphasizing the problems of recruiting and retaining sufficient regulatory staff with the appropriate skills, Stern (2000) also acknowledges other factors that may impede regulatory effectiveness such as the difficulties in separation of powers and regulation for very small countries, inconsistent reform policies, and unreliable law of courts.

60

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

On the other hand, Berg (2002, 2009), Smith and Wellenius (1999), and Intven, Oliver, and Sepulveda (2000) elaborate on the systematic organizational structure of a regulatory body for effective reforms. They highlight that clear priorities and objectives are the main elements of a competent regulatory institution and suggest that the agency should be independent of the government in power. Moreover, their studies also suggest that keeping processes transparent and consistent for the entire public, with regard to issuing licenses as well as monitoring the service providers’ performance, would further improve the telecommunications sector. Finally, Fig. 2(c) portrays the influence of freedom from corruption on the marginal effect of regulatory body. The marginal effects are declining with greater freedom from corruption. Note that the marginal effects of regulatory body remain statistically significant until the freedom from corruption index is greater than four. For countries having a freedom from corruption index greater than six onwards, the marginal effect becomes increasingly negative as freedom from corruption increases. While the former indicates that inefficiency is declining for countries with less corruption, the latter implies improvement in efficiency as countries are more transparent in dealing with and combating corruption. In contrast, there is no clear evidence of the effect of regulatory body on efficiency for countries with a freedom from corruption index in the range of 4.5–6. This finding is in line with Estache et al. (2009, 2006), which indicates that corruption offsets the effect of regulatory body on telecommunication network and access. 5.5. Impact of institutions The constitutive elements of each institutional variable could also shed some light on the effect of institutional governance on the efficiency of telecommunications. The results from Table 1 indicate that only legal integrity has a positive statistically significant effect on the efficiency of the telecommunications industry. This implies that, on average, an increase in legal integrity contributes to an increase in efficiency of the telecommunications sector by 1.87 points. On the other hand, the two other institutional variables, policy constraint and freedom from corruption, are statistically insignificant. Of particular interest are statistical significance and signs of the marginal effects over another possible range of reforms such as privatization and the presence of a regulatory body. Despite the limitation presented by the constitutive institutional coefficients, Table 1 is quite useful in assessing how policies interact with institutional governance. In essence, the coefficient for the product term between institutions and reforms (i.e. Pri  Leg, Body  Cor, etc.) is always meaningful. For instance, the interacted coefficient between privatization and legal integrity shows that privatization offsets the effect of legal uncertainty on the measure of telecommunications efficiency. Similarly, privatization also has a negative impact on the effect of policy stability on efficiency. However, it has no significant effect on the effect of corruption on telecommunications’ efficiency. With respect to the interaction between regulatory body and institutions, it is interesting to note that, all other things being equal, the introduction of regulatory body reduces the effect of corruption on telecommunications’ efficiency. This is indicated by the estimated coefficient between regulatory body and corruption ðBody  CorÞ, which has a significant negative sign. Also, the presence of regulatory body offsets the effect of policy stability on efficiency. On the other hand, the presence of regulatory body has no significant effect on the impact of legal integrity on efficiency. Figs. 3 and 4 illustrate these institutional relations on telecommunications’ efficiency over specific reform arguments, namely privatization and the presence of regulatory body. In particular, the two figures depict a total of four separate panels which focus on explaining the relationship of reforms and institutions corresponding to each significant interacted coefficient presented above. Note that the small square in the figure represents the estimated marginal effect, while the surrounding lines portray the 95 percent confidence intervals. The top panel in Fig. 3 shows that the confidence interval for both cases, either a country has privatized or has not privatized its telecommunications sector, lies outside the value of zero. This suggests that the marginal effects of policy constraint on telecommunications’ efficiency based on the two distinctive scenarios are statistically significant.27 To justify whether there is a significant difference between the two types of ownership, the hypotheses were tested by setting H 0 : δnp −δp ¼ 0 and H 1 : δnp −δp ≠0. Given that the calculated test-statistic value is 4.10, the null hypothesis is rejected which indicates that there is a significant difference in terms of how types of ownership influence the marginal effect of policy constraint. Similar interpretations can be made with regard to Fig. 3(b) on the impact of ownership on the marginal effects of legal integrity. In essence, the marginal effect of legal integrity on telecommunications’ efficiency is significant for both types of ownership. However, given that the confidence intervals for both cases, privatized or not privatized, overlap across the values of privatization, this suggests that there is no high certainty that the marginal effects of legal integrity in the case of privatization are statistically distinguishable from the case without privatization (Kam & Franzese, 2007). Correspondingly, the test of the hypotheses indicates that the computed test-statistic (t¼1.87) cannot be rejected and thus there is no clear difference on the influence of ownership structure on the marginal effect of legal integrity. Fig. 4, on the other hand, exhibits the marginal effects of policy constraint and corruption on efficiency of the telecommunications industry conditioned upon the presence of regulatory body. As indicated by Fig. 4(a), the marginal 27

Note that comparisons made between univariate confidence intervals ignore potential covariance between the estimated parameters.

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

61

Fig. 3. Marginal effects of institutions on efficiency (ownership). (a) Impact of ownership on the marginal effect of policy constraint and (b) impact of ownership on the marginal effect of legal integrity.

effect of policy constraint remains reductively significant despite the non-existence of regulatory body. More importantly, the figure also manifests that when a regulatory body is introduced, the effect of policy constraint increases the impact on efficiency of the telecommunications sector. This finding is similar to the result presented in Fig. 3(a) with respect to ownership, which essentially indicates reforms remain the best option. Fig. 4(b) reveals that the confidence interval for the marginal effect of corruption without regulatory body includes the value of zero whereas the presence of regulatory body does not. In essence, the graph shows that the effect of corruption with the absence of regulatory body does not statistically differ from zero but the effect of corruption on efficiency with the presence of regulatory body does. Furthermore, the computed test-statistic (t ¼3.56) for the test of the hypotheses in checking the significant difference between the two scenarios is rejected. This result indicates that the effect of corruption differs significantly when a regulatory body exists to monitor the industry vis-a-vis the absence of a regulatory body.

5.6. Demographic implications At the more general level, Table 1 also indicates the effect of two demographic variables considered in the model. The estimated coefficient for road density is negative and significant, which implies that higher road density corresponds to higher efficiency for the telecommunications sector. This result is not surprising given the fact that higher road density implies higher probability of network expansion due to lower network deployment cost for telecom operators to expand network access and improve the quality of services. An urban population is also associated with improved efficiency in the telecommunications industry. In essence and on average, a one-percent increase in the share of the urban population corresponds to an improvement in efficiency of 0.43 percent. Similarly, this implies that the higher the percentage of the urban population, the higher the efficiency of the sector would be. The inclusion of dummy variables representing countries’ income not only enables this study to determine the effect of country-level development, but it also provides an indicator to explain the differences between the organized and established group of economies (e.g. OECD) vis-a-vis other countries. As reported in Table 1, high-income countries under the flagship of the Organization for Economic Co-operation and Development (OECD), perform relatively better in terms of efficiency compared to middle-income countries and other high-income nations. More interestingly, the estimated coefficient for non-OECD countries is statistically insignificant, indicating that there are no subtle differences in terms of efficiency for middle-income countries and high-income non-OECD countries.

62

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

a

b

Fig. 4. Marginal effects of institutions on efficiency (regulatory body). (a) Impact of regulatory body on the marginal effect of policy constraint and (b) impact of regulatory body on the marginal effect of freedom from corruption.

6. Summary and conclusion This research set out to examine the evolution of telecommunications sector performance in the context of regulatory reform and institutional governance. Using a stochastic distance function model, this study has estimated the technical efficiency in a sample of 70 countries for the period 1980–2004. Unlike many any other telecommunications performance studies, the model estimated here incorporates institutional governance in measuring telecommunications industry technical efficiency to determine how reforms and governance affect each other in determining the implications of the reforms for the sector. The findings of the study present a number of important points that are worth highlighting for corporate strategists as well as policy makers. First, given the nature of cost elasticity for mobile telecommunications, which is much smaller than fixed-line services, it is beneficial for telecommunication providers to concentrate more on developing wireless infrastructure for greater network coverage. This in turn, will improve efficiency and bridge the information and digital gap between urban and rural areas, at the same time fulfilling the commitment towards universal service. From a broader perspective, this initiative would also reduce the network penetration disparity between developing and developed nations. Conversely, reducing the number of employees should be viewed as one of the imperative strategies to enhance technical efficiency in the telecommunications industry. This problem is apparent with the huge share of labor cost relative to other inputs. This issue has to be addressed in line with the long-term infrastructure policy to avoid redundancy or duplication of inputs in the foreseeable future. Secondly, the findings also offer potential explanations for why some previous studies have found that telecommunications reform has little effect (Gillwald, 2005; Giray, 2007; Hoffmann, 2008; Horwitz & Currie, 2007; Mariscal, 2004), while others have found that reform actually increases the performance of the telecommunications industry (Boylaud & Nicoletti, 2001; Li & Xu, 2004; Ros, 1999; Wallsten, 2001). Generally, the result implies that both findings are possible, depending on the effective nature of institutional governance in place. Likewise, the findings provide insights into the importance of institutional governance on reform process and its impact on telecommunications productivity and efficiency. These findings largely agree with other studies, for instance, by Cordova-Novion and Hanlon (2002) and Stern and Holder (1999), which emphasize the integrity of a regulatory body vis-avis institutional governance such as transparency and clear guidelines for successful reform. Moreover, the findings also support the arguments by Bergara et al. (1998) and Levy and Spiller (1996), which emphasize policy stability for effective reforms.

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

63

Furthermore, there are clear differences of marginal effects for institutional governance on efficiency performance with regard to countries that experienced reforms and those that did not. In particular, countries that have had reforms are expected to have greater influence over institutions to improve performance. This is indicated by the higher negative marginal effect values for the respective institutional governance over the two types of reforms. This finding corroborates the models developed by Laffont and Guessan (1999), who showed that competition and corruption are negatively related. Thirdly, the independent effect of reforms, either privatization or the presence of a regulatory body, does not necessarily contribute to higher efficiency, contrary to reports by many earlier telecommunications studies. On the contrary, the findings suggest that having these two reform events together exerts greater beneficial effects on telecommunications efficiency than they would separately. Further analysis also reveals that reforms do not guarantee improved performance if such reform lacks the characteristics of good institutional governance. Taken together, the empirical analysis reveals that policy stability in the form of substantive checks and balances on executive power is the most important aspect for successful reform. Independently, legal integrity improves telecommunications efficiency through privatization, while greater freedom from corruption influences the effectiveness of a regulatory body. This conclusion has important policy implications. The most obvious is that market forces by themselves cannot guarantee improved performance. Although the universal structure of reform recipe could provide a partial platform for liberalization, the more important element is state readiness to revamp its institutional governance. Embarking on a universal remedy through reforms and regulatory restructuring may be detrimental and harmful for telecommunications development if the country under consideration does not have satisfactory institutional capacity. Finally, the importance of governance is paramount, particularly in sustaining workable and effective market liberalization. While credible rule of law is essential in promoting sectoral efficiency through privatization, transparent procedures need to be given the highest priority for the regulatory body to be effective in order to limit corrupt practices. This is to ensure a fair playing field for the service providers to contribute towards the attainment of telecommunications development.

Appendix A Please see Appendix Table A1.

Table A1 Descriptive statistics for outputs and inputs, 1980–2004. Variable

Mean

Std. Dev.

Min

Max

Observations

y1

Fixed-line subscribers (in million)

Overall Between Within

7.015797

19.800 3.104405 19.600

0.001210 1.681799 −6.999501

19.300 14.000 186.00

N ¼ 1750 n ¼70 T ¼ 25

y2

Mobile subscribers (in million)

Overall Between Within

2.936233

11.900 4.803645 10.900

0 0 −17.400

185.00 20.400 167.00

N ¼ 1750 n ¼70 T ¼ 25

y3

Internet subscribers (in million)

Overall Between Within

125302.1

0.227461 0.035825 0.224662

0 0.067913 −0.111077

2.12730 0.23691 2.01565

N ¼ 1750 n ¼70 T ¼ 25

x1

Investment (US$) (in million)

Overall Between Within

1740.000

4980.000 994.000 4880.000

0.04361 392.000 −2990.00

74200 4730 71200

N ¼ 1750 n ¼70 T ¼ 25

x2

Switching exchanges (in million)

Overall Between Within

5.205122

11.700 2.15161 11.500

0.000038 1.44754 −4.97099

140.00 10.200 136.00

N ¼ 1750 n ¼70 T ¼ 25

x3

Total staff (in million)

Overall Between Within

51570.34

0.134741 0.019385 0.133358

0.000115 0.019499 −0.045835

1.30210 0.09757 1.27001

N ¼ 1750 n ¼70 T ¼ 25

64

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

Appendix B Please see Appendix Table B1.

Table B1 Descriptive statistics of inefficiency variables, 1980–2004. Variable

Mean

Std. dev.

Min

Max

Observations

1 1 1.27486

N ¼1750 n¼ 70 T ¼ 25

10 9.2 13.23029

N ¼1750 n¼ 70 T ¼ 25

0 0 −0.51714

1 1 1.28285

N ¼1750 n¼ 70 T ¼ 25

2.27538 2.09468 0.92188

0.89592 1.436 0.18515

8.93591 8.91366 9.91812

N ¼1750 n¼ 70 T ¼ 25

7.75307

1.80374 1.69922 0.63698

0.17560 3.71669 2.26140

10 10 11.03066

N ¼1750 n¼ 70 T ¼ 25

Overall Between Within

5.92047

1.96511 1.91430 0.49741

1.63 2.55920 2.50327

10 9.36560 10.03407

N ¼1750 n¼ 70 T ¼ 25

Road density (Road)

Overall Between Within

1.17594

1.48398 1.47974 0.20640

0.00336 0.02194 −1.15417

8 8 2.52944

N ¼1750 n¼ 70 T ¼ 25

Urban population (%) (Urban)

Overall Between Within

69.61371

17.87484 17.59405 3.76921

9 10.04 46.09371

100 100 86.17371

N ¼1750 n¼ 70 T ¼ 25

z1

Privatization (Pri)

Overall Between Within

0.43486

0.49588 0.26375 0.42106

z2

Role of state enterprises (Ent)

Overall Between Within

4.59029

2.83461 2.39283 1.54529

z3

Regulatory body (Body)

Overall Between Within

0.32286

0.46770 0.26413 0.38721

z4

Policy constraints (Pol)

Overall Between Within

6.04148

z5

Legal integrity (Leg)

Overall Between Within

z6

Freedom of corruption (Cor)

z7

z8

0 0 −0.52514 0 0.08 1.470286

References Andova, V. (2006). Mobile phones, the internet and the institutional environment. Telecommunications Policy, 30, 29–45. Bailard, C. (2009). Mobile phone diffussion and corruption in Africa. Political Communication, 26(3), 333–353. Battese, G., & Coelli, T. (1993). A stochastic frontier production function incorporating a model for technical inefficiency effects. Department of Econometrics, University of New England. Battese, G., & Coelli, T. (1995). A model for technical inefficiency effects in a stochastic frontier production function for panel data. Empirical Economics, 20(2), 325–332. Berg, S. (2002). Institutional requirements for second-generation infrastructure reform: Processes and performance in developing countries. Public Utility Research Center, University of Florida. Berg, S. (2009). Characterizing the efficiency and effectiveness of regulatory institutions. Public Utility Research Center, University of Florida. Bergara, M., Henisz, W., & Spiller, P. (1998). Political institutions and electric utility investment: A cross-nation analysis. California Management Review, 40 (2), 18–35. Black, B., Kraakman, R., & Tarassova, A. (2000). Russian privatization and corporate governance: What went wrong?. Stanford Law Review, 52(6), 1731–1808. Boylaud, O., & Nicoletti, G. (2001). Regulation, market structure and performance in telecommunications. OECD Economic Studies, 32, 100–142. Campo-Flores, F. (1994). Lessons from privatization of Argentina's national telephone company. Policy Studies Review, 13(3/4), 235–248. Carrington, R., Coelli, T., & Groom, E. (2002). International benchmarking for monopoly price regulation: The case of Australian gas distribution. Journal of Regulatory Economics, 21(2), 191–216. Cherry, B., & Wildman, S. (1999). Institutional endowment as foundation for regulatory performance and regime transitions: The role of the US constitution in telecommunications regulation in the United States. Telecommunications Policy, 23(9), 607–623. Coelli, T. (1996). A guide to FRONTIER 4.1: A computer program for stochastic frontier production and cost estimation. University of New England, Armidale: University of New England. Coelli, T., Estache, A., Perelman, S., & Trujillo, L. (2003). A primer on efficiency measurement for utilities and transport regulators. Washington DC: World Bank Institute. Coelli, T., Perelman, S., & Romano, E. (1999). Accounting for environmental influences in stochastic frontier models: With application to international airlines. Journal of Productivity Analysis, 11, 251–273. Coelli, T., Rao, P., O'Donnell, C., & Battese, G. (1998). An introduction to efficiency and productivity analysis. New York: Springer. Cordova-Novion, C., & Hanlon, D. (2002). Regulatory governance: Improving the institutional basis for sectoral regulators. OECD Journal on Budgeting, 2(3), 57–118. Estache, A., Goicoechea, A., & Manacorda, M. (2006). Telecommunications performance, reforms and governance. Washington DC: The World Bank. Estache, A., Goicoechea, A., & Trujillo, L. (2009). Utilities reforms and corruption in developing countries. Utilities Policy, 17(2), 191–202. Estache, A., & Wren-Lewis, L. (2009). Towards a theory of regulation for developing countries: Following Jean–Jacques Laffont's lead. Journal of Economic Literature, 47(3), 729–770.

N. Mohamad / Telecommunications Policy 38 (2014) 49–65

65

Fare, R., & Primont, D. (1995). Multi-output production and duality: Theory and applications. Massachusetts: Kluwer Academic. Farrell, M. (1957). The measurement of productive efficiency. Journal of the Royal Statistical Society, 120(3), 253–290. Fink, C., Mattoo, A., & Randeep, R. (2002). An assessment of telecommunications reform in developing countries. Washington DC: The World Bank. Gasmi, F., & Virto, L. R. (2010). The determinants and impact of telecommunications reforms in developing countries. Journal of Development Economics, 93, 275–286. Gillwald, A. (2004). Transforming telecom reform for development. South Africa: LINK Centre, Graduate School of Public and Development Management, Witwatersrand University. Gillwald, A. (2005). Good intentions, poor outcomes: Telecommunications reform in South Africa. Telecommunications Policy, 29, 469–491. Giray, F. (2007). The privatization and liberalization of telecommunications: A Comparative analysis for Turkish telecommunication sector. European Journal of Scientific Research, 17(4), 546–560. Gutierrez, L., & Berg, S. (2000). Telecommunications liberalization and regulatory governance: Lessons from Latin America. Telecommunications Policy, 24 (10–11), 865–884. Gwartney, J., Lawson, R., Sobel, R., & Leeson, P. (2007). Economic freedom of the world: 2007 Annual Report. Canada: Fraser Institute. Hamilton, J. (2001). Institutions political regime and access to telecommunications infrastructure in Africa. Public Utilities Research Centre, University of Florida. Henisz, W. (2000). The institutional environment for economic growth. Economics and Politics, 12(1), 1–32. Heston, A., Summers, R., & Aten, B. (2006). Penn world table version 6.2. Center for International Comparisons of Production, Income and Prices: University of Pennsylvenia Hoffmann, B. (2008). Why reform fails: The ‘Politics of policies’ in Costa Rican telecommunications liberalization. European Review of Latin American and Caribbean Studies, 84(84), 3–20. Horwitz, R., & Currie, W. (2007). Another instance where privatization trumped liberalization: The politics of telecommunications reform in South Africa. Telecommunications Policy, 31, 445–462. Intven, H., Oliver, J., & Sepulveda, E. (2000). In H. Intven (Ed.), Telecommunications regulation handbook. Washington DC: The World Bank. ITU (2006). World telecommunications indicator database. Geneva: International Telecommunications Union. ITU (2008). Trends in telecommunication reform 2008: Six degrees of sharing. Geneva: International Telecommunication Union. Jamasb, T. (2006). Between the state and market: Electricity sector reform in developing countries. Utilities Policy, 14, 14–30. Kam, C., & Franzese, R. (2007). Modeling and interpreting interactive hypotheses in regression analysis. Michigan: The University of Michigan Press. Kessides, I. (2004). Reforming infrastructure: Privatization, regulation and competition. Washington DC: The World Bank. Kumbhakar, S., & Lovell, K. (2000). Stochastic frontier analysis. Cambridge: Cambridge University Press. Laffont, J., & Guessan, T. (1999). Competition and corruption in an agency relationship. Journal of Development Economics, 60(2), 271–295. Laffont, J. (2005). Regulation and development. Cambridge: Cambridge University Press. Levy, B., & Spiller, P. (Eds.). (1996). Cambridge: Cambridge University Press. Li, W., & Xu, C. (2004). The impact of privatization and competition in the telecommunications sector around the world. Journal of Law and Economics, 47(2), 395–429. Lovell, K. (1993). Production frontiers and productive efficiency. In H. Fried, K. Lovell, & S. Schmidt (Eds.), The Measurement of productive efficiency: Techniques and applications (pp. 3–67). Oxford University Press: Oxford University Press, 1993. Lupi, P., Manenti, F., Sciala', A., & Varin, C. (2011). On the assessment of regulators’ efficiency: An Applications to European telecommunications. Info, 13(1), 61–73. Manzetti, L. (1999). Privatization South American style. New York: Oxford University Press. Mariscal, J. (2004). Telecommunications reform in Mexico from a comparative perspective. Latin American Politics and Society, 46(3), 83–114. Mayer, J., Butkevicius, A., Kadri, A., & Pizarro, J. (2003). Dynamic products in world exports. Review of World Economics, 139(4), 762–795. O'Donnell, C., & Coelli, T. (2005). A Bayesian approach to imposing curvature on distance functions. Journal of Econometrics, 126(2), 493–523. Petrazzini, B. (1995). The political economy of telecommunications reform in developing countries. London: Praeger Publisher. Petrazzini, B. (1997). In W. Melody (Ed.), Telecom reform: Principles, policies and regulatory practices. Lyngby: Technical University of Denmark. Pisciotta, A. (1997). In W. Melody (Ed.), Telecom reform: Principles, policies and regulatory practices. Lyngby: Technical University of Denmark. Ramamurti, R. (1996). In R. Ramamurti (Ed.), Privatizing monopolies. London: John Hopkins University Press. Ros, A. (1999). Does ownership or competition matter?: The effects of telecommunications reform on network expansion and efficiency. Journal of Regulatory Economics, 15, 65–92. Shephard, R. (1970). Theory of cost and production functions. New Jersey: Princeton University Press. Singh, J. P. (1999). Leapfrogging development? The political economy of telecommunications restructuring. New York: SUNY Press. Sinha, N. (1995). In B. Mody, J. Bauer, & J. Straubhaar (Eds.), Telecommunications politics: Ownership and control of the information highway in developing countries. New Jersey: Lawrence Erlbaum Associates. Smith, P., & Wellenius, B. (1999). Strategies for successful telecommunications regulation in weak governance environments. Washington DC: The World Bank. Spiller, P., & Tommasi, M. (2003). The institutional foundations of public policy: A Transactions approach with application to Argentina. Journal of Law, Economics and Organization, 19(2), 281–306. Stern, J. (2000). Electricity and telecommunications regulatory institutions in small and developing countries. Utilities Policy, 9, 131–157. Stern, J., & Holder, S. (1999). Regulatory governance: Criteria for assessing the performance of regulatory systems an application to infrastructure industries in the developing countries of Asia. Utilities Policy, 8, 33–50. The World Bank (2006). World development indicators (p. 2006)Washington DC: The World Bank2006. TIA (2000). Multimedia telecommunications market review and forecast. Arlington: Telecommunication Industry Association, MMMTA Market Research. Uhlenbruck, K., Rodriguez, P., Doh, J., & Eden, L. (2006). The impact of corruption on entry strategy: Evidence from telecommunication projects in emerging economies. Organization Science, 17(3), 402–414. Wallsten, S. (2001). Econometric analysis of telecommunications competition, privatization and regulation in Africa and Latin America. Journal of Industrial Economics, 49, 1–19. Wellenius, B. (1999). In L. Manzetti (Ed.), Regulatory policy in Latin America: Post-privatization realities. Miami: North-South Center Press. Wellenius, B., & Stern, P. (1994). In B. Wellenius, & P. Stern (Eds.), Implementing reforms in the telecommunications sector: Lessons from experience. Washington DC: The World Bank. Whincop, M., & Rowland, S. (1998). In H. Moazzem, & J. Malbon (Eds.), Who benefits from privatization?. New York: Routledge.