A. Network Externalities

Telecommunications is a network industry in which new users generate positive externalities for existing users by broadening the network. Since the presence of positive or negative externalities can be an important reason for distinctive tax treatment of an industry or product, it is important to evaluate this aspect of cell phone use.

The value of a communication network lies in its ability to provide access to its members (Economides, 1996). Network expansion therefore provides a free benefit to those already in the network: Although individuals decide to buy and use a cell phone based on their perception of the existing network’s value, each decision to join benefits all existing users. This positive externality suggests that free markets could yield a sub-optimal level of cell phone communications, since the marginal user takes account only of its own benefits, but not of those on others. This might indicate that a subsidy (e.g., for handsets) that lowers the cost of joining a network would be efficient.⁷ This is especially likely to be the case when networks are first being established, when the externalities from their expansion will be large relative to the private benefits of joining. This is borne out by the common practice of cross-subsidizing network expansion to remote areas with revenues from established markets.

Evidence on the presence of network effects in the mobile sector is strong. For example, Doganoglu and Grzybowski (2007), estimating a demand model for Germany, conclude that the 700 percent expansion of cell phone communications over 1998-2003 was largely due to network effects (measured by lagged network size), and less so to the 41 percent price decrease.⁸ Grajek (2010) reaches similar conclusions regarding Poland (1996-2001), finding that intra-operator network effects count most.⁹ Other authors, such as Wu and Chu (2009) and Rouvinen (2006), rely on diffusion models¹⁰ and conclude that both network effects (existing penetration) and potential user base have a significant impact on diffusion.

Despite strong growth in recent years, data suggests that cell phone access is still suboptimal in some areas. The ratio of global phone subscriptions (both fixed and mobile) to population exceeds 100 percent (107 percent in developing countries and 165 percent in developed countries). However, unique mobile phone subscribers as a percentage of global population (i.e., counting multiple phones for the same person as a single subscription) is only 67 percent worldwide—62 percent in developing countries (Figure 3).¹¹ Given the increasing importance of cell phones and the internet to accessing basic public and private services, full coverage for adult subscribers is arguably desirable. It should therefore be expected that public policy will thus continue to support network expansion to underserved areas through such policies as “buildout requirements” in telecom license agreements. Universal service funds, which are typically financed with charges on telecom activities, are sometimes also authorized for this purpose; however, they are prone to governance problems and the funds are often under- or mis-utilized.

Global Cell Phone Penetration Rates

(Subscription per 100 habitants)

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: International Telecommunications Union and GSMA

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Global Cell Phone Penetration Rates

(Subscription per 100 habitants)

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: International Telecommunications Union and GSMA

• Download Figure
• Download figure as PowerPoint slide

Global Cell Phone Penetration Rates

(Subscription per 100 habitants)

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: International Telecommunications Union and GSMA

• Download Figure
• Download figure as PowerPoint slide

B. Contribution to Economic Growth

Information is essential to the function of any market economy and has been intimately associated with positive economic externalities. By decreasing the cost of acquiring information, cell phones reduce transaction costs, creating opportunities for additional transactions, and therefore contribute to economic efficiency and growth. In this respect, tax measures that decrease or slow down cell phone penetration could bring short term revenue benefits for the government at the expense of long term benefits for all.

There is a vast literature on the relationship between general information technology and growth, but a relatively small subset deals specifically with cell phone use.¹² Macroeconomic studies generally rely on cross-country panel data to assess the relationship between cell phone use and growth.¹³ These studies rely on a host of cell phone use measures and explanatory factors, which can make comparing results challenging. Microeconomic studies document specific examples of improved efficiency, often at the village or community level.

Measuring the macroeconomic impact of telecoms is difficult due to the bidirectional causality between cell phone usage and growth. Following the pioneering work of Hardy (1980) on the impact of fixed lines on growth, Cronin, et al. (1991) find that growth is driven by investments in telecoms, as well as the reverse.¹⁴ Röller and Waverman (2001), working on fixed lines, address this causality problem using simultaneous equations, thus endogenizing telecom investments.¹⁵ Finally, Waverman, Meschi and Fuss (2005) apply a similar model, as well as an endogenous growth model, to cell phone data, which ushered in a wave of studies on cell phones and growth.

The main conclusion of the macroeconomic literature has been that cell phones have a statistically significant and potentially large impact on growth; however, the magnitude of this effect varies widely depending on data, methods and study period. In a study of 92 countries over the 1980-2003 period, Waverman, Meschi and Fuss (2005) found that among low income countries, 10 more mobile phones per 100 people increased per capita GDP growth by 0.6 percentage points—roughly twice as much as in high income countries. Sridhar and Sridhar (2007) confirmed these findings for 63 developing countries over 1990-2001 (the early stage of mobile use in developing countries), and found that an increase of 1 percent in mobile penetration increased GDP by 0.01 percent; thus, the average cumulative growth in cell phone use of 248 percent increased GDP by 2.48 percent over the study period.¹⁶

The recent literature includes more data on low-income countries. Lee, Levendis and Gutierrez (2012) use 1975-2006 data from 44 sub-Saharan countries and find that cell phone penetration had a significant impact on growth recently (2000-2006) and that this impact is stronger where landline penetration is lowest. Using data on 44 African countries over 1988-2007, Andrianaivo and Kpodar (2012) find results consistent with Waverman, Meschi and Fuss (2005) that a 10-percentage point increase in the mobile penetration rate increases the growth rate by 0.6 percentage points. Interestingly, they also find that the price of a 3-minute conversation is negatively related to growth. Some of these studies also find that the impact on growth is stronger with higher penetration levels, which is consistent with network effects (see next section).¹⁷

Microeconomic studies regarding the benefits of mobile communication are reviewed by Duncombe (2016) and by Acker and Mbiti (2010), who point to 5 specific sources of enhanced welfare: (1) access to and use of information; (2) productive efficiency; (3) new jobs for mobile-related services; (4) reduced exposure to shocks and risk; and (5) delivery platform for key development services (e.g., financial, agricultural, educational). These factors are highlighted in a series of detailed case studies. For example: Jensen (2007) documents the reduced price dispersion, lower prices, higher profits, and the elimination of waste following the introduction of cell phones. Muto and Yamano (2009) find that the expansion of mobile phone networks is associated with greater market participation of remote producers of perishable goods. Acker (2010) finds that the introduction of cell phones in Niger between 2001 and 2006 lowered price dispersion by 10 to 16 percent on grain markets. Acker and Fafchamps (2015) measure a 6 percent reduction in producer price dispersion in Niger following the introduction of cell phones.

Research also shows a strong link between cell phone use and financial inclusion (e.g., Andrianaivo and Kpodar, 2012). However, the expansion of e-banking also depends on more than cell phone use (ADB, 2013): well-designed regulation, inter-operability of networks and related institutions, the presence of retail-level agents, etc. are some of the many building blocks of an effective e-banking system. In addition, it remains unclear how the current tax structures impact e-banking. For example, transfer of money through SMSs (as in the original M-PESA system in Kenya) in a system where excises apply only to voice communications should not have any impact at the margin. In this respect, more micro-evidence on the exact impact of tax levels and structures is needed.

The evidence, especially at the microeconomic level, generally suggests a positive causal relationship between cell phone use and growth, which tends to confirm that lower information costs improve economic efficiency. This does not, however, imply that there are externalities from telecoms, as transactions may be properly priced and all social costs and benefits thus internalized by the market. Thus, contribution to growth might not warrant special government intervention, although it might advocate against taxes that deter business (as distinct from consumer), and telecommunications in particular. This issue will be further discussed in the section on taxes (excise vs. VAT surcharge).

C. Demand Elasticity

Telecommunication excises on both services and handsets are sometimes justified on the ground that the demand for telecom services is inelastic, offering a good “tax handle”. This section examines the empirical evidence regarding the own-price elasticity of telecom service demand. Discussion of whether demand inelasticity constitutes a sound rationale for excise taxation is postponed until the following section.

Empirical studies of the price elasticity of cellular communication find widely variant results across time and countries (Table 2). Thirteen early studies reviewed in Commerce Commission (2003) indeed found the demand for cell phone service to be inelastic: The own-price elasticity ranged from −0.06 to −0.54 for mobile subscription/network access, and from −0.09 to −0.8 for mobile originated calls.¹⁸ By contrast, studies using more recent data including more developing countries show a wider range of estimates: Garbacz and Thompson (2007), examining panel data for 1996-2003 for 53 developing countries, conclude that though connection charges have little impact on penetration rates (price elasticity of −0.029 to −0.37), monthly prices have a stronger impact (price-elasticity of −0.195 to −1.268). The high end of that range of estimates is comparable to the −1.5 price elasticity that Waverman, Meschi, and Fuss (2005) find for a sample of developing countries over the years 1996-2003. Price elasticity estimates can also be found in several country-level studies for both developed and developing countries—for example, Caves (2011), Koutroumpis et al. (2011), and Hakim and Neaime (2014). Nevertheless, some country-level studies, such as Karacuka et al.’s 2011 analysis of the Turkish market, find inelastic demand in developing countries. Overall, the diversity of estimates for the elasticity of cell phone use remains a striking feature of this industry, although the price-elasticity of network access seems to be generally lower.

Table 2.

Empirical Studies of Telecommunication Price Elasticities

Table 2.

Empirical Studies of Telecommunication Price Elasticities

Authors and country of study	Estimated price elasticity
Commerce Commission (2003): Review of 13 papers	Network access: −0.06 to −0.54 Mobile originated calls: −0.09 to −0.8
Waverman, Meschi and Fuss (2005): Aggregate production function for 38 low- and lower middle income-countries, including a demand for telecoms (penetration rate)	−1.5
Garbacz and Thompson (2007): Penetration data for 53 developing countries (1996 to 2003)	Connection charges: −0.37 to −0.029 Monthly price: −1.268 to −0.195
Dewenter and Haucap (2008): Austria traffic (01/1998 to 03/2002)	Short-term, businesses: −0.33 Short-term, private consumers: −0.14 Long term, businesses: −0.74 Long term, private consumers: −0.37
Gasmi et al. (2009): South Africa: sample of 6936 individuals in 2005	Between −3.8 and −1.3
Caves (2011): USA: traffic for 38 states (2001 to 2007)	−2.1
Koutroumpis et al. (2011): Greece: Quarterly traffic data (2005 to 2010)	−1.6
Hakim and Neaime (2014) (Middle East and North Africa: country-level traffic data (1995 to 2007)	Between −1.2 and −1.0
Karacuka et al. (2011): Turkey: operator data (01/2002 to 12/2006)	Short term, pre-paid: −0.20 Short term, post-paid: −0.36 Long term, pre-paid: −0.33 Long term, post-paid: −0.72

View Table

Table 2.

Empirical Studies of Telecommunication Price Elasticities

Authors and country of study	Estimated price elasticity
Commerce Commission (2003): Review of 13 papers	Network access: −0.06 to −0.54 Mobile originated calls: −0.09 to −0.8
Waverman, Meschi and Fuss (2005): Aggregate production function for 38 low- and lower middle income-countries, including a demand for telecoms (penetration rate)	−1.5
Garbacz and Thompson (2007): Penetration data for 53 developing countries (1996 to 2003)	Connection charges: −0.37 to −0.029 Monthly price: −1.268 to −0.195
Dewenter and Haucap (2008): Austria traffic (01/1998 to 03/2002)	Short-term, businesses: −0.33 Short-term, private consumers: −0.14 Long term, businesses: −0.74 Long term, private consumers: −0.37
Gasmi et al. (2009): South Africa: sample of 6936 individuals in 2005	Between −3.8 and −1.3
Caves (2011): USA: traffic for 38 states (2001 to 2007)	−2.1
Koutroumpis et al. (2011): Greece: Quarterly traffic data (2005 to 2010)	−1.6
Hakim and Neaime (2014) (Middle East and North Africa: country-level traffic data (1995 to 2007)	Between −1.2 and −1.0
Karacuka et al. (2011): Turkey: operator data (01/2002 to 12/2006)	Short term, pre-paid: −0.20 Short term, post-paid: −0.36 Long term, pre-paid: −0.33 Long term, post-paid: −0.72

D. Technology, Industry Structure, and Market Power

The telecoms sector has evolved rapidly in recent decades, not only technologically but also in terms of its industrial organization and regulation. Until the 1980s, telephony depended on fixed-wire technology that required large-scale infrastructure investment. Like other network utilities (e.g., electricity and natural gas), telephone service was thus viewed as a “natural monopoly”—an industry with high fixed costs that creates increasing returns to scale such that the average cost of service provision is declining over the relevant demand range. Under these conditions, licensing two providers to establish two separate infrastructures would be inefficient; a single provider could offer universal service at the lowest cost. Since a private telecoms monopoly would charge excessive prices, most countries organized telephone service provision as a state-owned monopoly; however, a few countries including the United States licensed private companies subject to government price regulation. In both cases, telephone service prices were aligned with the average cost of service provision: marginal cost plus a normal return to invested capital (Averch and Johnson, 1962).¹⁹

In recent decades, the telecoms sector has opened to greater competition due to both technological and organizational innovations. In the 1980s, governments began to liberalize telecommunication markets to reap the benefits of greater competition, such as increased efficiency, lower prices and innovation (Economides, 2004). Early market liberalization focused on privatization and unbundling of trunk-line (“back-hauling”) connections and other services. The rapid spread of cellular phone networks in the 1990s introduced a major new source of competition for fixed-line telephony. Cellular phone penetration has been particularly strong in developing countries, where fixed-line networks initially covered a much smaller percentage of the population. The rise of the internet offers yet another form of competition for voice telephony: voice-over-internet-protocol (VoIP).²⁰ VoIP relies on internet data (as opposed to a voice signal) and can be delivered over telephone, cable and cellular networks, but it requires at least 3G technology. Thus, technological and organizational innovation have transformed telephony from a “natural monopoly” into a far more contested market.

Nonetheless, several factors still constrain competition in telecommunication markets, particularly in developing countries. First, though the fixed cost of establishing a cell phone network is lower than that of a fixed-line network, it is still quite substantial (GSMA, 2008). High entry costs thus continue to restrict telephone competition, particularly in more thinly populated or low-income areas where the return to investment is lower. In higher density areas, infrastructure duplication by competing cell phone providers suggests another means by which fixed costs could be lowered: Some governments are experimenting with decoupling physical infrastructure and user service provision for cellular communications, as they previously did for fixed-line communications.

Second, limited electromagnetic spectrum for the transmission of cellular communications constrains the number of potential service providers in any given market. The spectrum has numerous wavelengths, each with its own technical characteristics best suited to different types of transmission: Generally, higher frequencies have clearer reception but carry over a shorter range than lower frequencies. Given current technology, wavelengths in the 800-900 megahertz (MHz) range are best suited for cell phone communications, although frequencies as high as 2+ gigahertz (GHz) are sometimes used.²¹ Limited spectrum availability requires that the spectrum authority can only license a few cell phone operators to serve a particular area. ²²

Together, these factors tend to limit cellular phone service provision to a small number of companies. Most countries have 2-4 providers, while some retain a monopoly structure, usually with at least partial state-ownership (Figure 4). Less than one fifth of countries have at least 5 providers—the level that Li and Lyons (2012) identify as the threshold for full effective competition. Limited competition facilitates implicit or explicit collusion among providers, leading to higher service prices than in a fully competitive market. Studying the telecoms industry in the OECD, Sung (2014) finds that industry concentration relates positively to both service prices and operator profitability, and Boniecki and others (2016) provide strong evidence of high industry profitability, which is positively related to industry concentration.

Number of Mobile Operators per Country

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: GSMAintelligence.com.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Number of Mobile Operators per Country

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: GSMAintelligence.com.

• Download Figure
• Download figure as PowerPoint slide

Number of Mobile Operators per Country

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: GSMAintelligence.com.

• Download Figure
• Download figure as PowerPoint slide

Theoretically, limited competition is also likely to generate operator rents, or returns to investment in excess of the risk-adjusted market rate. For example, the standard Cournot model of oligopoly suggests that prices are a negative function of the number of operators; however, other models, such as Bertrand (1883), suggest that competitive prices could be reached with as little as two operators. Empirical research is needed to determine which model best describes the effective competition among telecoms providers in a given market.

Since rents exceed the rate of return necessary to induce telecommunication companies to invest and provide services, they can in theory be redistributed to consumers or the public treasury without undermining service provision. The existence of significant sectoral rents may thus provide a rationale for distinctive taxation of the telecommunications sector.

Major cellular operators’ financial results confirm a relatively high level of profitability. For example, since 2009 the average return on assets of large US telecom service companies exceeded that of large-capitalization companies in general by 70 percent; telecom return on investment and return on equity were 16 percent and 15 percent higher than the market, respectively.²³ However, looking at individual markets, profitability varies significantly among operators depending on their market share. Analyzing telecom markets in the Middle East and Africa, Boniecki and others (2016) show that dominant operators have a ratio of earnings before interest, tax, depreciation and amortization to total revenue of 41-47 percent, vs. 30-37 percent for the second-largest operators and only 15-25 percent for the third and fourth operators.

As in the recent past, technological and/or organizational change may alter the competitive landscape for telecommunications, shifting the amount and locus of rents. A major trend in developed country markets, which is steadily trickling down to developing countries, is the shift from voice telephony to data-based communications (i.e., VoIP). Internet communications are usually channeled via local unlicensed spectrum (i.e., wifi) onto fixed-wire networks (DSL or cable), although data may also be carried over cellular broadband (3G+) networks.

The trend toward VoIP thus presents new competition for cellular operators that may erode cellular rents—or reallocate them to owners of fixed-wire and cable networks. This dynamic will vary by market, depending on consumer demand, industrial organization and regulation, and is an issue which requires further study. Internet service provision, which provides the link between data networks and end users, tends to be more competitive than cellular telephony, and therefore less likely to generate rents. Organizational innovations to extend the separation of infrastructure and end-user services - for example, the licensing of “virtual mobile network operators - may lead to greater effective competition among telecom providers, driving down prices and reducing rents for the entire sector. The extent to which new technologies and organizational forms augment effective competition and lower consumer prices of course depends in great part on regulatory and licensing policies.

E. Regulation

The magnitude of telecommunication rents depends integrally on regulatory policy, including spectrum licensing. Regulatory charges such as license fees increase operator costs, while price regulation, conversely, constrains revenues, both reducing profits. Regulations aimed at spurring competition, such as licensing additional operators, may also lead to lower service prices, also have the potential to reduce rents.

Spectrum management policy determines the number of mobile telecom licenses in each market, with obvious implications for competition and profitability. In addition to limited spectrum and entry costs, market depth—a function of population size and density as well as income levels—is an important constraint on the number of operators that can feasibly be licensed in a given market. Examining 139 countries, Symeou (2011) finds that country size positively influences the number of operators as well as the efficiency of telecoms provision.

Not just the number of operators, but their relative market power is also important for effective competition. Even where multiple providers are licensed, telecom markets are often dominated by just one or two firms: the average market share of second providers is less than one third, while those of third and fourth providers are less than 20 percent and 10 percent, respectively.²⁴ The dominant firm is often the incumbent telecoms operator, which may still be partially or fully owned by the government. Providing for effective competition in telecom services may thus require restructuring of state-owned dominant firms and/or rebalancing of existing licenses.

Despite innovations that permit greater competition in the telecoms sector, price regulation may still be necessary in some markets. In many countries, the incumbent operator still controls key infrastructure assets, such as “trunk” telephone, cable or cellular networks. Unless access charges are limited, it could use this power to weaken its rivals. Further, every provider has a monopoly over connections to its own network, so regulation may be necessary to prevent companies from overcharging their competitors for connecting calls on their network.

Another cost that regulation may impose on telecom providers is service coverage extension to new areas, or “buildout”. Regulators may set minimum buildout requirements in a license contract, or allow operators to bid on buildout levels as part of a license auction. Since buildout requirements usually extends service into more thinly populated areas, it usually reduces operator profitability. An alternative means of financing buildout is through a government “universal service fund” financed through fees or charges on telecom services.²⁵ Universal service fund charges also reduce profitability by reducing output and/or operator margins.

Regulatory agencies levy numerous charges on telecoms companies, including not only spectrum and license fees but also fees for public services such as industry monitoring (e.g., minute counts). Spectrum and license charges are usually quite substantial and have a significant impact on providers’ profitability, while other regulatory fees are generally much smaller.

The function of telecoms regulation should be to ensure the widespread provision of good quality service at competitive prices, not to raise revenue. With the exception of spectrum licenses, which will be discussed in greater detail in the following section, regulatory charges should therefore be limited to the cost of providing the regulatory service. Charges that just cover service costs can be considered user fees, but any charge above (below) cost recovery is a tax (subsidy).

While tax revenues flow to the public treasury, license fees typically flow to telecoms regulators to finance their activities (although they may also be shared with the treasury). Ideally, all government bodies act as agents of the commonwealth, and thus the entity collecting various charges is irrelevant. In practice, however, there is likely to be divergence of interest among fiscal and regulatory bodies. Vesting taxing authority in both fiscal and regulatory interests may therefore result in their “overgrazing the commons” of the telecoms industry. Fiscal authority should therefore be concentrated in the treasury: telecoms regulatory agencies should have sufficient funds to carry out their important activities, but any residual revenue that they raise from spectrum licenses or other fees should flow to the treasury.

A. General Considerations

The main goal of modern tax systems is to finance critical public goods and services, including infrastructure, education and health care. Key criteria for assessing a tax system are its efficiency (i.e., minimum impact on economic behavior), equity (i.e., a tax burden proportional to or increasing with income), and simplicity (i.e., low administrative and compliance costs). Together, these criteria prescribe a system consisting of a few taxes with low rates that apply to wide, appropriately defined bases²⁶ uniformly across the entire economy.

While this prescription appears to preclude special tax treatment for any sector, certain economic factors may merit exceptions. One notable exception is industries that generate economic rents, which can in theory be efficiently taxed with no distortion to investment incentives. A salient example of this is extractive industries such as mining, which are often subject to special rent taxes as well as royalties. Another justification for differential taxation is the presence of externalities, where the private costs (or benefits) of some activity differ from the social costs (or benefits) of that activity, resulting in too much or too little of it. Activities with negative externalities, such as consumption of petroleum products (which cause pollution), should be taxed to bring their private cost in line with their social cost, while activities with positive externalities should be subsidized. Finally, it may be administratively infeasible to apply standard taxes to certain activities. A prime example of this is applying a credit-invoice VAT to margin-based financial services (such as bank loans and deposits). In these cases, a second-best feasible solution (such as VAT exemption) must be sought.

The previous section’s discussion of the economics of cellular telecommunications suggests that the sector does have distinctive features that may warrant special tax treatment. Most notably, the existence of technological and regulatory barriers to entry, coupled with strong consumer demand that must be served by local infrastructure, may give rise to rents. Ideally, regulatory solutions to promote greater competition should be sought, but high fixed costs, limited spectrum and regulatory capacity, and the small size of some developing country markets constrain this option. Since rents can be efficiently taxed without discouraging investment, some form of rent tax may therefore be appropriate. However, since telecom services are generally provided by multinational enterprises (MNEs), collection of profit-based taxes, such as CIT and rent taxes, may be difficult due to profit-shifting, particularly in countries with low administrative capacity. Simpler (albeit more distortive) revenue instruments, such as excises, may therefore be justified. As discussed in the previous section, positive externalities suggest that extension of the telecom network to unserved areas or populations should be (cross-) subsidized, and that taxes that specifically limit access should be avoided.

Public finance theory also posits that taxes on productive inputs should be avoided to prevent distortions in production efficiency that reduce output below the production possibility frontier (Diamond and Mirrlees, 1971). Revenue can be raised more efficiently with taxes imposed at the consumer level, which do not distort production. As a corollary, customs duties on capital equipment should ideally be zero.

The following sections review the various taxes that are or might be applied to the telecoms sector and assess them based on their capacity to capture rents without deterring investment or impeding consumer access.

B. Spectrum License Fees

Spectrum license fees (which in some countries are split into separate spectrum and operator license fees) are usually fixed payments levied up-front and/or periodically over the term of the license²⁷. Once agreed, license fees form part of the operator’s fixed cost structure, in addition to investment and overhead, etc. To incentivize investment, licenses must be long-term, with a typical duration of about 15 years.

Countries apply various methods for allocating spectrum licenses among potential providers (McMillan, 1995). Prior to the advent of cell phone technology, much of the spectrum was unlicensed and marginal values were low (if not zero). Licenses were thus often allocated on a first-come, first-served basis or by lottery. The spread of cell phone technology precipitated a sharp rise in demand for usable spectrum, and the resulting rise in its value required finding a more efficient means of allocation.

Since the 1990s, competitive auction has become the preferred means of license allocation in developed countries, although some countries set prices administratively based on various pricing algorithms (Bauer, 2003). Auctions ensure that licenses are allocated to the most efficient providers and procure public revenues. In theory, a bidder for a license would be willing to pay up to the amount by which the returns to the license exceed the cost of exploiting it (including a normal return to invested capital)—that is, up to the amount of the rents derived from holding the license (Cramton, 2002). The bidder with the lowest cost of provision and/or best ability to maximize revenues would therefore have the highest valuation and should win the bidding, even if the winning bid does not fully reflect its expected rents. The extent to which the bid falls short of those rents will depend on various factors including the level of bidding competition and uncertainty regarding future profits.

Spectrum license auctions have yielded considerable government revenue, to the extent that cell phone companies have complained that spiraling license fees are undermining their investment in the sector and forcing them to raise prices. If fees set by auction constitute a tax on the pure economic rents arising from the limited supply of spectrum, however, they should neither deter investment nor affect output pricing (provided that they don’t drive operators into insolvency²⁸). Morris (2005), Kwerel (2000) and Bauer (2003) test this hypothesis and find no evidence that license fees have any impact on service pricing, which supports the theory that auctioned license fees indeed function like a rent tax. However, more recent work on auctions which take into account their effects on debt financing and post-auction competition show that auctions may indeed affect pricing and penetration rates. Janssen and Karamychev (2009) show that less risk-averse bidders are more likely to win auctions and to price their products higher post-auction. Haan and Toolsema (2011) show that, where bidders finance their bids through the debt market, debt levels may affect post-auction pricing. Studying 47 countries, Kuroda and del Pilar Baquero Forero (2017) find that penetration rates are lower in countries that conduct auctions.

Seeking to maximize spectrum auction revenues can create perverse incentives for regulators in designing spectrum licenses. Hazlett, Munoz and Avanzini (2011) stress that the most important function of auctions should not be to raise revenue but to ensure that the most technically efficient providers receive licenses. Misplaced emphasis on revenue-raising gives regulators an incentive to limit the size and number of spectrum licenses to increase their value.²⁹ Since rents are maximized under a monopoly, seeking to maximize license fees is thus likely to result in reduced competition and higher service prices.

A prominent concern regarding the use of auctions to allocate spectrum in developing countries is the possibility of collusion among bidders. The number of large multinational telecoms operators competing for licenses globally is relatively small, and their repeated interaction across different markets increases the risk of collusion to suppress license prices. Developing country spectrum authorities that have held auctions often complain of a lack of bidding competition, which frequently causes them to fall back on pricing licenses by negotiation.

Another possible source of auction failure is faulty auction design, including excessive reserve values. The general trend in developing countries has been from undervaluation of licenses toward full or overvaluation. In many countries, early cellular licenses were awarded by “beauty contest” and granted to winner free of charge. This early undervaluation—in some cases accompanied by tax incentives—later led to aggressive taxation in some countries: When governments realized that telecom licenses were highly lucrative and concessions were therefore not necessary to attract investors, they imposed telecom excises or other charges ex-post to recoup the value of earlier concessions. Thus, in more recent years, developing countries began to impose significant fees for spectrum licenses—whether by auction or negotiation—as regulators’ understanding of their market value increased. Pricing was in some cases sufficiently aggressive that auctions failed due to excessive reserve values.

Improving government capacity to value spectrum licenses accurately would clearly facilitate the design of a reasonable fiscal regime for the telecom sector. Valuation of telecom licenses is complex, depending not only on the amount of spectrum allocated and market size and depth, but also on the license’s many terms. ³⁰ These may include the types of services that licensees should provide, the technologies they should employ, price caps for certain services, build-out, or a government equity stake. Further, the licensee may or may not have the right to sell the license to a third party. The more flexible and fungible the license, the greater will be its value and the higher the license fees it will elicit. Multilateral collaboration on spectrum license databases and valuation tools could help ensure reasonable revenue recovery from spectrum license allocation.

Ex-ante fixed license fees place all financial risks on telecoms providers, who will accordingly shade downward their bids during acquisition.³¹ To distribute risk more evenly between government and provider, license fees could also be set in ex-post terms, with providers remitting a certain share of profits or gross revenues to the government.³² Kwon, Lee and Oh (2010), who compare ex-ante fixed fees with ex-post fees based on profit-sharing or “royalties” (gross revenue sharing—the method used in Hong Kong) favor royalties. Although royalties, unlike fixed fees and profit sharing, do raise consumer prices, they create better incentives for regulators to allocate more spectrum to telecoms (in order to enhance revenues) and for operators to increase investment (due to risk-sharing with the government).

C. Rent Taxes

Spectrum fees (whether auctioned or not) are not the only way to tax an economic rent. Resource rent taxes (RRTs), which are observed mostly in natural resource taxation (Daniel et al., 2012), tax profits above the “normal rate of return” for a given industry—the minimum return required to compensate investors. For example, if the normal rate of return on the capital market is 15 percent, an equity-financed investment of US$100 that yields a profit of US$25 generates a rent of US$10. The tax base of the CIT is the whole US$25, but the tax base of the RRT is only US$10. To the authors’ knowledge, RRTs have not yet been applied to the telecom industry.

RRTs take many forms (Box 1), each with its own pros and cons. Calculation of rents is generally based on the same accounting elements as the CIT, with certain key differences: Under most RRTs, interest expense is disallowed, and investment is fully expensed rather than depreciated. To preserve the real value of investment expensing, losses are carried forward with interest against future revenues. For this reason, the revenue profile of an RRT is more back-loaded than that of a CIT, entailing higher public revenue risks. RRTs in extractive industries typically have statutory rates of 15-30 percent and are applied in addition to—not instead of—the standard CIT.³³

Because RRTs tax only rent and not the normal return to investment, they are less distortive than the CIT. Like well-structured spectrum auctions, they should have no impact on marginal investment, production or pricing, and should therefore not impede network access. Insofar as rent taxes increase the average tax rate on investment—that is, the ratio of total taxes paid to total pre-tax profits—they may nonetheless affect an operator’s decision on whether to enter a certain market. Since provision of telecom services requires local incorporation and infrastructure investment, a heavier average tax burden may not deter entry but would nonetheless reduce valuation of spectrum licenses.

Rent taxes are arguably more equitable than spectrum auctions, since they are based on actual rents earned rather than expected future rents at the time of bidding. If an operator earned no rents (for example, due to greater than expected price competition or technological obsolescence), then its liability under a rent tax would be zero. And unlike weakly structured auctions, rent taxes are not subject to collusion. However, like the CIT, RRTs are vulnerable to accounting manipulation and aggressive transfer pricing schemes aimed at shifting profits to lower-tax jurisdictions. Designing and implementing a new tax also entails significant administrative costs that must be weighed against the potential revenue and efficiency gains from introducing a rent tax. Also, a rent tax is most easily introduced at the outset of an investment project, so like spectrum auctions it should ideally be timed in accordance with the licensing and investment cycle.

D. Corporate Income Tax

Several countries, including Jamaica, Côte-d’Ivoire, Jordan and Yemen, subject telecom companies to higher CIT rates. The typical rationale for this policy is that the sector is perceived to have above-normal profits, so a higher CIT rate is applied to capture a larger share of them. Like spectrum fees and rent taxes, the corporate income tax (CIT) taxes rents, so a CIT surcharge may be a reasonable instrument to capture telecom rents. In a short-run, static context, the CIT is unlikely to directly affect consumer pricing: If operators are maximizing profits, then no proportional tax levied on profits would cause them to alter their input/output behavior.

Unlike spectrum fees and rent taxes, however, the CIT taxes the normal return to capital, which raises the cost of capital and depresses investment.³⁴ Output is therefore likely to be lower under an elevated CIT rate, and prices correspondingly higher. Furthermore, Davidson and Martin (1985) demonstrate that higher CIT rates may augment oligopolistic collusion to raise prices by restricting output: A decrease in the net-of-tax corporate discount rate raises the cost of future retaliation for cheating, which makes more restrictive equilibria possible.³⁵

Like imposition of a rent tax, imposing a CIT surcharge increases incentives for MNEs to shift profits across borders through more aggressive tax planning. ³⁶ For example, in a meta study, Heckemeyer and Overesch (2013) find an average semi-elasticity of multinational corporate profits to the CIT rate of −0.8, i.e. a 1 percentage-point higher CIT rate will reduce reported profits of an affiliate of a foreign company by 0.8 percent. Another method of tax avoidance, where the CIT base includes capital gains (or a separate capital gains tax is imposed), is transferring telecom licenses indirectly through offshore mergers and acquisitions rather than direct sales.³⁷

Spectrum auctions, rent taxes and CIT surcharges all have the potential to tax rents accruing to telecom operators, and each instrument has its own strengths and weaknesses. Spectrum auctions reflect expected rather than actual rents and are vulnerable to collusion. Rent taxes are likely the most efficient and equitable of the three instruments, but defer revenues to varying degrees. CIT surcharges are more distortive than rent taxes or spectrum fees; however, given that countries already have CITs in place, imposing a higher CIT rate on telecom companies may be administratively less costly than introducing a rent tax in addition to the standard CIT. Which instrument (or instruments) are best suited to a particular market depends on various factors, notably administrative capacity as well as the stage of the licensing and investment cycle.

E. Excises

An important weakness of spectrum fees, rent taxes and corporate income taxes is that, as noted above, their performance depends in part on administrative capacity. Furthermore, profit-based taxes on capital-intensive industries yield revenues with a lag due to large up-front deductions for depreciation or expensing. Many countries, especially in Sub-Saharan Africa, have therefore introduced telecom excises to ensure a significant, up-front revenue yield from the sector. The most important excises in revenue terms are on telephone services (purchased by both businesses and final consumers), but some countries also impose excises on handsets and initial connection (e.g., purchase of SIM cards).

Levying an excise on a particular good raises its price relative to other goods, thereby suppressing its consumption. This effect is less pronounced, the less the own-price elasticity of the excised good.³⁸ Even if the demand for a good is highly inelastic, however, taxing it more heavily than other goods distorts consumption due to cross-price elasticities. For this reason, excises are usually confined to goods that are not only price-inelastic, but also have negative externalities: e.g., alcohol, tobacco, and petroleum products. As shown in the previous section, telecom demand varies greatly across samples and is not always inelastic; moreover, telecom connectivity has positive rather than negative externalities. Telecom service thus does not conform with the standard requirements for an excisable good.

The remaining justification for telecom excises is that they provide a good “tax handle” on an industry than can otherwise be difficult to tax. As such, excises have the attractive property of being relatively easy to administer, making them especially important to countries with low administrative capacity. Telecom service excises function similarly to royalties in the extractive industries: Levied as a percentage of output value or fixed charge per unit, royalties are used – even in the presence of rent taxes and CIT – to raise revenue from the start of production. Because they tax gross revenues, rather than net income, royalties are much less vulnerable to tax evasion than CIT or rent taxes. However, they therefore run the risk of rendering marginal projects unprofitable and thus reducing investment and output (Daniel et al., 2012). Nonetheless, moderate royalties can play an important role as part of a balanced mix of tax instruments aimed at sharing fiscal and economic risks between government and investors.

The remainder of this section reviews the various types of telecom excises currently in use, considering their capacity to capture rent, their impact on externalities, and their administrability.

Excises on domestic calls

Most excise revenues from the telecom industry are from excises on domestic phone calls. In Malawi, for example, domestic excises account for almost half of total fiscal and regulatory revenues generated by the telecoms sector, which averaged 1.3 percent of GDP in 2013-2015 (Figure 5).³⁹ In Jamaica, telephone excises plus the VAT surcharge account for more than one third (Figure 6) of total telecom revenues, which averaged almost 1.8 percent of GDP in 20132015.⁴⁰

Malawi Telecommunications Revenues (% of GDP)

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: Authorities, IMF staff calculation.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Malawi Telecommunications Revenues (% of GDP)

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: Authorities, IMF staff calculation.

• Download Figure
• Download figure as PowerPoint slide

Malawi Telecommunications Revenues (% of GDP)

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: Authorities, IMF staff calculation.

• Download Figure
• Download figure as PowerPoint slide

Jamaica Telecommunications Revenues (% of GDP)

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: Authorities, IMF staff calculation.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Jamaica Telecommunications Revenues (% of GDP)

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: Authorities, IMF staff calculation.

• Download Figure
• Download figure as PowerPoint slide

Jamaica Telecommunications Revenues (% of GDP)

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: Authorities, IMF staff calculation.

• Download Figure
• Download figure as PowerPoint slide

Telephony excises may be either specific (a fixed charge per minute) or ad valorem (a percentage of service charges). Since both types of excises are proportional to output and gross revenues, they should also be roughly proportional to profits⁴¹; however, the relationship between output and profits, which depend on costs as well as gross revenues, will generally vary considerably among products as well as operators. Excises are thus less effective than a CIT or rent tax at capturing a share of rents.

Unlike a rent tax or CIT, excises generally affect the price of the excised good. The extent to which an excise changes the price of the good depends on the relative price elasticities of supply and demand: The more elastic supply (demand), the more the burden of an excise will be borne by the consumer (producer). An excise would only fall entirely on producer rents in the unlikely case that telecommunications supply was completely inelastic. Although this might be the case in the short run, given existing infrastructure, over the medium-term disinvestment due to reduced profitability would reduce capacity and drive up consumer prices, shifting some of the burden back onto consumers.

In general, therefore, some part of the burden of telecom excises can be expected to fall on consumers through higher service prices and therefore result in lower traffic (Figure 7). In Guinea, for example, significant budget difficulties led to the adoption of 1 GNF per second tax (so 60 GNF per minute, or nearly 15 percent of the average price per minute) in 2015, which led to an immediate equivalent fall in traffic. While the upward growth path immediately resumed, this new path remained roughly 15 percent below the pre-tax trajectory (Figure 8). The impact of the SMS component of the same tax was even more dramatic.

Impact of A New Excise: The Case of Guinea^*

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: Agence de Régulation des Postes et Télécommunications.^* The tax amounts to 1 GNF per second, so 14.2 percent of the Q2-2015 average on-net voice tariff. Traffic decreased by 16.1 percent in Q3-2015. Quarterly revenue from the tax is roughly 12 M USD (exchange rate of 10,000 GNF/USD).

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Impact of A New Excise: The Case of Guinea^*

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: Agence de Régulation des Postes et Télécommunications.^* The tax amounts to 1 GNF per second, so 14.2 percent of the Q2-2015 average on-net voice tariff. Traffic decreased by 16.1 percent in Q3-2015. Quarterly revenue from the tax is roughly 12 M USD (exchange rate of 10,000 GNF/USD).

• Download Figure
• Download figure as PowerPoint slide

Impact of A New Excise: The Case of Guinea^*

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

Source: Agence de Régulation des Postes et Télécommunications.^* The tax amounts to 1 GNF per second, so 14.2 percent of the Q2-2015 average on-net voice tariff. Traffic decreased by 16.1 percent in Q3-2015. Quarterly revenue from the tax is roughly 12 M USD (exchange rate of 10,000 GNF/USD).

• Download Figure
• Download figure as PowerPoint slide

Specific vs. Ad Valorem Excise

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Specific vs. Ad Valorem Excise

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

• Download Figure
• Download figure as PowerPoint slide

Specific vs. Ad Valorem Excise

Citation: IMF Working Papers 2017, 247; 10.5089/9781484324981.001.A001

• Download Figure
• Download figure as PowerPoint slide

In contrast to a VAT surcharge (discussed in following section), excises increase telecom prices for both consumers and businesses. Taxes on business inputs distort production and lead to “casacading” when business outputs are subject to further taxes (such as VAT or sales tax). This aspect of telecom excises is of particular concern, as it may reduce productivity.

In countries that impose telephone excises, the resulting higher service prices have accelerated the trend toward VoIP. This illustrates the fact that excise taxation is less effective where substitution options increase tax base elasticity. In response, many countries have therefore extended the telecom excise base to include internet data in order to forestall base erosion (e.g., Turkey, Guinea). As noted previously, the impact of the trend to VoIP on telecom rents is unclear, but it may undermine sectoral rents and thus the justification for special sectoral taxes.

Where telecom competition is limited, service excises may increase the market power of the dominant firm—an effect often seen with high royalties in the extractive industries. Highly profitable dominant firms will better withstand imposition of a sizeable consumer excise than marginal (higher-cost) operators, who may be driven out of the market. Excessive consumer excises could thus serve to bolster the market power of dominant operators to the detriment of consumers. Excise taxes can also facilitate collusion among the providers, causing a greater contraction in output than the price increase alone would warrant, so that the excise is actually “overshifted” to the consumer. Higher corporate profits may increase CIT revenue somewhat, but at the cost of reduced excise revenue, productivity and welfare.

The literature on the effect of excises on prices in markets with limited competition⁴² generally concludes that ad valorem excises dominate specific excises in terms of welfare effects. Under oligopoly, equilibrium prices will be lower under an ad valorem excise than under an equivalent specific excise. An ad valorem excise in effect flattens the effective demand curve that operators face, while a specific excise shifts the curve downward (Figure 8). Facing a more elastic effective demand curve nudges operators’ behavior toward what it would be in a perfectly competitive market. With an ad valorem excise, part of the cost of a price cut is borne by the government in terms of lower tax revenues, so producers have more incentive to reduce prices. Further, if extension of the telephony excise to internet data becomes necessary, ad valorem excises are more straightforward to extend, since internet service has different consumption units than telephony (megabytes vs. calling minutes).

F. Value-added Tax

Most countries appropriately apply a VAT to telephone services at the standard rate. However, several countries apply a higher VAT rate, for the same reasons that countries apply telephone excises. The arguments for and against doing this are thus the same as for excises, with one significant difference: Due to input VAT crediting, the VAT is not a cost for registered VAT payers, so the production distortion created by telecom excises is alleviated—at least for formal-sector companies above the VAT threshold. Informal companies, however, would still experience the higher VAT rate as a cost, although taxing these companies indirectly may be a deliberate policy goal. To raise the same amount of revenue from a VAT surcharge as from an excise, the surcharge would have to be higher to offset the cost of crediting for registered businesses, which would increase the price distortion for other users.

On the downside, a VAT surcharge could potentially complicate tax policy and administration. Dual VAT rates, once in place, may be difficult to contain to a single industry. For example, many countries introduce a reduced VAT rate on a few staple items to relieve poverty; these reduced rates have a tendency to creep to other goods and services as businesses selling directly to consumers (e.g., tourism) lobby to have their output taxed at the lower rate. Reduced VAT rates are an inefficient subsidy to the poor, and can greatly complicate VAT administration. The IMF has therefore consistently supported a single-rate VAT. Clearly, businesses would not lobby to broaden the base of a VAT surcharge, but there could nonetheless be political pressure to use it to increase revenues, ultimately leading to a multiple-rate VAT. Also, if telecom expenses were a large enough to generate excess VAT credits, refunds would be required, which are a weak point of VAT administration in many low- and middle-income countries.

As for any industry, implementing a VAT may require practical adaptations to local market conditions and institutions (Box 3). However, a more complex issue is to properly and consistently implement the VAT on international calls. If the base case (Box 3) is rather straightforward, within- or cross-country confusion or inconsistencies on the definition of the place of consumption may create significant departures from basic VAT principles (hence economic inefficiencies), double taxation, or no taxation at all.⁴⁹ Such difficulties may be compounded by other VAT-related goals, such as regional harmonization, hence at times the need for arbitrary rules, especially for complex transactions involving multiple countries.

Box 4.

Prepaid cards, and international calls: who pays the VAT?

Generally, and in line with standard principles, VAT on telephone calls should apply at the point of consumption (destination principle – hence exports are zero-rated), and all VAT paid on inputs should be offset against VAT collected on sales, as per the credit-invoice method (except for non-registered entities – i.e., the final consumers). In practice, however, a conversation involves two individuals and potentially many operators, and payment systems may differ (e.g., caller pays vs. shared payment), and determining the point of consumption and VAT liability can be problematic, hence the need for a common agreement on the application of VAT for cellular communications. A common agreement, however, is not just an arbitrary decision, as it can have significant administrative implications. This notably led to the recent revisions of the EU Directive on the Value-added Tax^* which now recommends that the country of origin of the calling SIM card be considered the place of consumption.

This norm echoes existing practice and has numerous implications in many low- and middle-income countries, where VAT-able prepaid card/“caller pays” system are in place, and where the treatment of the VAT between operators has often been the source of much disagreement between authorities and operators, especially regarding international calls. The table below summarizes the implications of this standard for the application of the VAT, depending on the type of call. The base case in which a local call is made between two local (Home) SIM cards is simple and the caller is liable for VAT (taken from his prepaid account by the operator and paid to the government upon completion of the call). Complications arise in other cases. When, for example, a “Home” SIM card calls from “Abroad” this call should in principle be subject to VAT in Abroad (i.e., where the call is “consumed”). However, this would require that the caller registers his SIM in Abroad to pay the VAT, or that the Home operator transfers the VAT to the Abroad operator (at the Abroad rate). Because of such administrative difficulties, assuming that a Home SIM card always calls from Home - wherever it is located - and hence determines the place of consumption greatly simplifies the VAT treatment. This, however, has a series of implications for the VAT treatment of various types of calls. The table below presents possible cases, depending on the origin of the SIM card and the location of the caller and receiver, and provides the VAT treatment of inter-operator transactions. The latter notably implies that VAT cannot be applied to the net income from transactions between Home and Abroad operators, as current practice in some countries.

Country of originating SIM card	Location of originating SIM card	Location of receiving SIM card	Type of call	VAT treatment
Home	Home	Home	Base case	Originating Home SIM pays VAT The caller’s operator pays a VAT to the receiver’s operator and credits it as input VAT
Home	Home	Abroad	Outgoing	Originating Home SIM pays VAT Home operator pays access fee to foreign operator (import of service) and reverse charges the VAT (charges the VAT to itself and remits it to Home tax authorities) at Home rate. This input VAT can be credited
Home	Abroad	Home or Abroad	Outgoing roaming	Although the service is consumed Abroad, it is deemed consumed at Home, because the call is from a Home SIM card. Home VAT therefore applies Home operator pays access fee to foreign operator (import of service) and reverse charges the VAT (charges the VAT to itself and remits it to Home tax authorities) at Home rate. This input VAT can be credited
Abroad	Home	Home or Abroad	Incoming roaming	Although the service is consumed at Home, it is deemed consumed Abroad, because the call is from an Abroad SIM card. The foreign operator pays an access fee to the Home operator, who exports a service, and the transaction is zero-rated
Abroad	Abroad	Home	Incoming	The Home receiver does not pay the VAT
Abroad	Abroad	Abroad	NA	NA

View Table

Country of originating SIM card	Location of originating SIM card	Location of receiving SIM card	Type of call	VAT treatment
Home	Home	Home	Base case	Originating Home SIM pays VAT The caller’s operator pays a VAT to the receiver’s operator and credits it as input VAT
Home	Home	Abroad	Outgoing	Originating Home SIM pays VAT Home operator pays access fee to foreign operator (import of service) and reverse charges the VAT (charges the VAT to itself and remits it to Home tax authorities) at Home rate. This input VAT can be credited
Home	Abroad	Home or Abroad	Outgoing roaming	Although the service is consumed Abroad, it is deemed consumed at Home, because the call is from a Home SIM card. Home VAT therefore applies Home operator pays access fee to foreign operator (import of service) and reverse charges the VAT (charges the VAT to itself and remits it to Home tax authorities) at Home rate. This input VAT can be credited
Abroad	Home	Home or Abroad	Incoming roaming	Although the service is consumed at Home, it is deemed consumed Abroad, because the call is from an Abroad SIM card. The foreign operator pays an access fee to the Home operator, who exports a service, and the transaction is zero-rated
Abroad	Abroad	Home	Incoming	The Home receiver does not pay the VAT
Abroad	Abroad	Abroad	NA	NA

^* See: Council Implementing Regulation (EU) No 1042/2013 of 7 October 2013 amending Implementing Regulation (EU) No 282/2011 as regards the place of supply of services. See also: OECD (2017), International VAT/GST Guidelines, OECD Publishing, Paris.

G. Input Taxes and User Fees

Like the excises discussed above, countries sometimes levy charges on telecom inputs to try to capture a share of sectoral profits. This practice is highly variable across countries and mainly takes the form of elevated customs duties on telecoms capital equipment and/or user fees on public inputs. Examples of the latter include charges for use of public infrastructure (e.g., trunk networks) or land (e.g., for antennas), issuance and management of phone numbers, and industry monitoring (minute counts).

It is economically efficient to charge for the cost of public services whenever the cost of the service can be calculated with reasonable accuracy and an exclusive user can be identified. This is clearly the case for many regulatory services, including those mentioned above. However, to the extent that fees exceed the cost of providing a service, they comprise an input tax. In many cases regarding telecoms, the service provided by public authorities is ill-defined or even non-existent, or its cost is overstated or reflects inefficient public service provision.⁵⁰

Like consumer excises, input taxes have the potential to worsen the effects of limited competition. Katz and Rosen (1985) demonstrate using a Cournot oligopoly model that taxing inputs to a firm in an oligopolistic industry alters equilibrium prices, output and profits in a manner that may differ from that under perfect competition. If marginal costs are nondecreasing in the input tax—a reasonable assumption—then it will increase market prices and reduce output and, like for excises, the input tax can be overshifted to the consumer.

Since input taxes distort production and increase prices, they should be avoided in favor of more efficient types of taxes. Regulatory fees should be restricted to covering the cost of service provision.

H. Tax Incentives

Many governments offer telecom operators tax incentives that are also provided to other sectors, often through an investment code. The goal of such incentives is generally to attract foreign investment or to reallocate domestic investment towards specific economic sectors. Incentives may also be used to attract investment to geographic regions with low incomes or high unemployment. These incentives typically take the form of reduced taxes on profits (e.g., CIT holidays or reduced rates), investment (e.g., accelerated depreciation, expensing or investment tax credits) or inputs (e.g., customs duty waivers on capital goods and other inputs). Niger, for example, offered an initial 8-year holiday for CIT and some other taxes to four operators, and the Central African Republic has similar concessions.

Many of the business tax incentives offered in developing country investment codes are ill-advised, sacrificing scarce revenues on investment that would have taken place even without a special incentive (IMF et al., 2015). Typical investment incentives include tax holidays or reduced CIT rate, accelerated depreciation or expensing, and customs waivers on imported inputs. As discussed in the previous section, customs waivers on capital goods can be justified in the name of productive efficiency, but a preferable policy is to specify a zero tariff on capital goods in the customs code in order to minimize discretion.

Extending investment incentives to the telecom sector is difficult to justify, given that telecom operator licenses confer exclusive rights to exploit “locational rents” from selling services into the domestic market. Incentives are sometimes necessary to attract foreign direct investment in export industries that depend on cheap inputs elastically supplied by various competing jurisdictions. Tapping this market necessarily requires a local presence to build the necessary infrastructure. Telecom companies therefore should not need special incentives to enter a country and should be excluded from any incentive scheme, since telecom tax breaks would likely result in revenue losses with minimal impact on investment. The existence of network externalities may, however, justify lenient terms for establishing networks in new areas. This was seen in many countries when initial cellular licenses were granted at no charge.

Since telecommunications is not a labor-intensive industry,⁵¹ it is arguably not an appropriate candidate for incentives aimed at boosting regional employment. Host country governments do, however, have an interest in ensuring telecoms coverage in remote, thinly populated regions, which are not profitable for telecoms companies to cover. Approaches to ensuring this coverage vary across countries, encompassing methods such as “buildout” requirements in operator licenses and universal service funds. As in developed countries, these methods usually involve an operator’s cross-subsidizing rural service provision from high-margin urban service provision, rather than a net government tax break or subsidy.