Saturday, June 23, 2018

Moving Beyond Access: The Landscape of Internet Use and Digital Inequality in Nepal

[This research brief sketches the information characterization of the “have-less” in Nepal by unveiling Internet (data) consumption and online activity quantitatively. You can find the brief  at]

Conventionally, digital inequality has been seen as a deviance that can be fixed by increasing access. We argue that such framing shifts our focus away from critical issues about how Internet is used, who benefits and who is excluded in the end or even during the drive for increasing connectivity. We suggest that in an evolving landscape of digital expansion, the distinct in-between tier of the information have-less appears. The formation of this have-less group is a result of particular social transformations where communication technologies start to play crucial roles in the lives of the marginalized groups such as migrant workers, unemployed youth, and the old. We show that in utilizing the binary categories of “haves” vs. “have-nots,” the dominant narrative of Nepali Internet misses out the information stratification in Nepali society. Consequently, Internet policies, based on inaccurate ground assessment, will have no or limited impact on the lives and livelihood of the very users for whom the massive investment in the digital infrastructure is being rolled out. We call for a new set of policies that explicitly recognize the information stratification and make use of distinct technosocial characteristics of the information have-less in striving for the universal connectivity in the country. 

Wednesday, January 3, 2018

Changing connectivities and renewed priorities: Status and challenges facing Nepali Internet


Evidence available after the devastating April 2015 Nepal earthquake (Gorkha earthquake) illustrates uneven coverage and poor data consumption in Nepal in spite of impressive mobile Internet subscription numbers. Places with favourable terrain, higher population densities, and higher income have better connectivity. Online activity levels, on the other hand, do not always correspond with these factors. Overall, ownership of digital technologies and its use exhibit clear regional unevenness and a large urban-rural inequality. These geographical factors reflect differences in socio-demographic characteristics. Unfortunately, in Nepal, dominant discourses on the Internet brush aside these linkages. With deep structural inequalities, a resource-scarce economy, and a track record of poor governance, broadband connectivity will not reduce this development chasm. This paper calls for Nepali Internet discourses to be grounded in reality, detaching from a ‘self-evident’ development vision of connectivity.

Sunday, April 2, 2017

The bandwidth bottleneck - Inclusive internet policies have to acknowledge the inequality in income distribution

Internet policies of developing countries have long dreamt of an inclusive internet that can bridge the proverbial digital divide and noticeably contribute to economic growth. The path towards realizing this dream warrants broadband internet access to 45 percent of the Nepali households. This becomes probable when the entry-level home broadband connection is priced at less than 5 percent monthly income (ie, GNI per capita) of an individual. Broadband policy and Information and Communication Technology (ICT) policy envisage these to provide impetus for the sector to contribute 2.5 percent to GDP. The latest versions of the policies were drafted in 2015. A 45 percent broadband penetration rate is expected by 2018, with visible economic growth by 2020. These are the rates at which the necessary conditions translate to observable growth statistics. To understand the arguments, it suffices to equate broadband internet with ‘wide’ (bandwidth) and ‘fast’ connections.
Policy planning
Affordability of bandwidth is at the heart of broadband policy planning. In Nepal Rastra Bank’s fifth budget survey, the average income of households that fall in the richest 20 percent bracket is Rs53,578. This roughly equates to Rs11,906 per person. An entry level fibre broadband connection costs around Rs1,200 which is double the 5 percent affordability requirement. This cost is six times the affordability requirement for the poorest 20 percent bracket. What’s more, Rs1,200 is actually a low cost for advertised bandwidth and speed. Nepal has just above 1 percent broadband subscription.
Policies stipulate 512 Kbps (Kilo-bits per second) as the minimum download speed for a broadband internet connection. Interestingly, this minimum is the same in India, where the recent decision to classify 512 Kbps as broadband has been roundly criticised, with some labelling it “a joke”, “meagre”, “archaic and useless”, “out of touch with reality” and a “move back to the dark ages”. Such criticism comes due to the telecom regulators buckling under pressure from the telecom operators, which are also internet service providers (ISPs), for a U-turn on the original 2Mbps (Mega-bits per second) recommendation. The ISPs stated that without throttling, they cannot provide fast internet at an affordable price. Two options were suggested to the regulators: either set the minimum at 64 Kbps or allow unrestricted throttling. Regular home internet users would regard 64Kbps speed as a relic from the dial-up era. A crawling internet will also not contribute to the sustainable development goals (SDGs). The only realistic option on the table was to allow unrestricted throttling.
Throttling internet speed
Like their Indian counterpart, Nepali ISPs can throttle a fast internet speed as part of their ‘fair usage policy’ (FUP). Historically, the internet bandwidth was shared among six to eight subscribers. Sharing-ratios allowed ISPs to guarantee unlimited data volume. Upon the arrival of volume based packages (data cap), FUP mechanisms and tools such as throttling, sharing-ratio became an obsolete mechanism for bandwidth control. Throttling has been a thorny issue worldwide with frequent reports
of intentional throttling by the ISPs. Such measures raise questions on the nation’s net neutrality ambitions. A popular Nepali ISP defines ‘fair’ use: “Day to day surfing, checking emails and occasional downloading will not get you into trouble. However, downloading a 200-300 MB movie clip/data every day is almost certain to. The key is to keep an eye on the amount that you are downloading. Then if you’re classed as a heavy or excessive user, restructure your internet usage pattern so that you are not uploading or downloading at peak times. If you don’t heed their warnings, the system will slow down the speed of your connection after your data limit.” Worryingly, there is nothing a user can do about this.
Broadband internet and data cap with throttling is an oxymoron, considering high speed internet connection is being sold for high bandwidth applications. An internet user with a 10 Mbps connection will exhaust the 80 GB data cap after just 18 hours of usage. A four-hour per-day usage can extend access to this ‘fast’ internet for only five days, leaving 25 days for a buffered browsing experience. For some users, the data cap might seem to last longer. There are two likely explanations: ISPs give the promised 10 Mbps at certain time slots, the rest are throttled; or customers use the internet for low bandwidth applications such as browsing and VoIP (Viber/Skype).
There is no quick fix to the affordability conundrum. First, Nepali ISPs presently buy bandwidth from Indian ISPs at their dictated price. But as the Indian experience has recently shown, owning state of the art infrastructure in a competitive market does not necessarily translate into broadband affordability. Second, if affordability was to be defined in line with the meaningful use of the web, such as data usage of regular internet users, 512 Kbps is a ‘something is better than nothing’ target. Third, affordability demands a revision from the perspective of all income groups. Inclusive internet policies have to acknowledge the inequality in income distribution. Equipped with this new understanding, the entry-level broadband might have to be priced at 2 percent or less of the average monthly income.
Pandey is a researcher affiliated with Martin Chautari

Free Float Internet Policies (Findings from access and use survey)

Here we list the findings from the quantitative analysis of the household surveys conducted in 2014/15 to understand the access and use of the Internet technologies, including the fixed and mobile broadband. The research was carried out by Martin Chautari (2014-2016) with support from the Ford Foundation. This project would not have been possible without the generous support of the local partners. We would like to thank Nasancha Pucha: (Panauti), Himalaya Milan Secondary School (Tangting) and Progressive Women Society (Changunarayan) for their facilitation, assistance and support. We are also grateful to Bhagabati Gautam for her assistance and volunteering work in Pragatinagar.
Note: The paper is available in the journal “Studies in Nepali History and Society” (SINHAS), Vol. 21 No. 1. It is also available online here. The full set of briefs and papers can be found in our Universal Connectivity page]
  • A family spends 7 percent of their household income on mobile phone bills.
  • Though reported penetration of mobile phone has well crossed the cent percent mark (120%), the five locations had 72 percent household ownership (and similar individual ownership among the respondents). Therefore, NTAs subscriber count is likely to be sim counts.
  • Only half of the mobile phone owners had ever used internet from their sets.
  • Only 8 percent of the households had a fixed internet connection.
  • Less than 3 percent households in Dapcha own a computer (laptop or desktop). Except Panauti which has around 35 percent household ownership of computers, the four site averages to around 8 percent penetration. This is similar to statistics reported by the World Bank (8.9% in 2015).
  • Anything better than a weak statistical association among the Internet use and socio-demographic variables could not be found. For example, people with higher education did not exhibit a tendency to use the Internet for education and employment activities.
  • Income groups were indistinguishable in their use of mobile Internet.
  • Results on the ethnicity-Internet link show that in a particular settlement the ICT penetration and ownership of devices seem to reinforce the socio-economic exclusion. Across the sites, however, people belonging to the single ethnic group have varied access levels of ownership such as the Newars in Dapcha and Panauti or Gurungs in Changunarayan and Tangting.
  • Location advantage offered a better explanation of Internet adoption than chasing the ethnicity and income threads.
  • Over two-third of the people in the survey said home Internet was “not required”. The indistinguishable usage pattern across income groups and age levels indicate main reason for non-use could be the absence of any compelling need for such technology.
  • In our survey sites, only Panauti and Pragatinagar had households with more fixed broadband penetration than the unflattering national average.
  • Only 12 percent of the total Internet users have used the Internet or mobile-apps for private or government electronic-services (e-services). Only 27 percent of the respondents say they have ever used Internet for obtaining government information.
  • Four out of five mobile phone using respondents, including the broadband Internet users, complained about regular “network problems.”
  • There is dearth of evidence to unlock association between local content, Internet infrastructure and affordability to forge meaningful ICT policies.

The IT policy documents will do better if they accept that the outcome of a technological intervention depends on the use people find for the technology. There is no denying that the relationships among the digital divide, poverty and education are enormously complex, particularly when, for instance, the difference in relative and absolute poverty will make available studies about the relationship questionable. The complexity also becomes obvious when policy interventions could frame questions of the divide variously as problems of access, or skill, or content. Our survey reinforces the understanding that adoption decisions are primarily need-driven and based on cost-effectiveness of the investment. Farmers of Dapcha, for instance, are likely to revert to existing information channels if Internet do not offer content tailored to their information requests.

Wednesday, March 29, 2017

Research Bites: Universal Connectivity (information and communication technologies)

This post contains short summaries of research briefs broadly related to Information Technology and Communication. The research was carried out by Martin Chautari during 2014-2016. The homepage of the research is

Universal Connectivity in Nepal: A Policy Review

UC-related policies assume a levelling effect of the ICTs, and little consider that technologies are themselves socially constructed artefacts. They have provided a rationality for mobilizing public resources, for erecting new institutions and facilitating the sustaining of certain business interests, particularly that of the IT elite in Nepal. The solution lies in formulating evidence based UC policies while openly acknowledging the limitations of the technologies in mainstreaming the marginalized and vulnerable section of the population.

Stakeholders for Universal Connectivity in Nepal 

UC-related policies in Nepal should focus on developing the scientific and technological core and not simply on facilitating acquisition and diffusion of new Internet-based technologies. As a topmost priority, the design of digital ecosystem should address particularly ways to manage immense power demand. It should not be left as an issue belonging to another ecosystem or to be managed by yet another ministry. Past endeavours have sufficiently demonstrated ambitions to transform the country with diffusion of imported technologies have not worked.

Deliver Through Mobiles First 

Lack of benchmark studies such as on speed, penetration and price hinder setting achievable targets. But the real problem of ‘digital divide’ can only be dealt meaningfully by situating it in the context of broader socio-economic divide in the country. Widespread diffusion of the mobile phones provides an opportunity to direct, shape and fine-tune existing policies. Instead of leaning unreliably on the capacity-centric development model, Nepal’s IT policies need to frame ways to integrate user’s everyday experience of mobile phones into the drive towards universal connectivity.

A Regression Analysis into Nepali ICT's Energy Consumption and its Implications

Even with the most lenient assumptions regarding the behaviour of the ICT sector, it is a significant consumer of energy at the national level. The chase to parallel the energy demands of the transportation sector will see gains when large data centers are established to support e-governance, e-commerce and other data intensive always-online services. An energy audit of the ICT sector along with large scale studies on the context of technology has to be done simultaneously for Nepal.

Thursday, March 2, 2017

ICT and Development – Is there a connection?

The idea that Information and Communications technologies (ICTs) are instrumental in national development has become the faith of the policy circle in the last decade.  It is upon this premise that a robust ICT infrastructure would lower communication and transportation overheads, catalyze creative activities and eventually catalyze development activities. The impact of ICT on economic growth in the industrialized nations has been deeply explored and heatedly debated as an academic topic. Yet we are no closer to a formula linking characteristics of ICT technologies, their experience and inclusive development. The idea that ICT connects to growth is still a hypothesis. It remains unproven on scales, replication and its universality. There are several criticisms on the findings of ICT and Development (ICTD) literature, specifically in context of the economically least developed countries (LDCs).

There is a methodological criticism of the ICTD research, which is applicable to a wider range of empirical research. The major findings are based on the statistical technique called regression analysis and on a concept known as ‘p-value’. There have been critiques of the empirical paradigm from the very beginning. The mainstream empirical research relies heavily on the notion of p-values because it makes quick inferences possible, without necessitating a thorough understanding of the underlying phenomena. Researches that employ p-value without well-reasoned scientific argument are available in a significant volume. Owing to misuse and misinterpretation in the social sciences and psychology research, the American Statistical Association floated a notice in 2016 that inferences based on the p-value should not be used specifically in the policy making processes. A leading journal of psychology has completely banned the use of p-values.  The ICT research needs to be re-examined from the newer perspective; a perspective that does not rely on the short-cut of p-values.
There are other methodological issues regarding the findings of ICTD research. As the ICT and economic growth relation has been examined chiefly on the dataset of the industrialized OECD countries, one can also raise doubts regarding the generalization of the results to the LDCs. Similarly, the reliability of dataset is in the LDCs is itself questionable. For instance, the MIS reports of the Nepal Telecommunication Authority claim a hundred percent penetration of mobile phones, which clearly is an overstatement. 

Besides the doubts on the methodology of ICTD research, several socioeconomic arguments can be provided that cast doubts on the universality of the “ICT for development” phenomenon. I will state a few. The strongest argument is regarding the affordability of Internet. A popular criterion considers that an Internet connection affordable if its tariff is within the five percent of per capita income. This criterion renders even an entry level Internet unaffordable to a vast majority of the Nepali population. The slow economic growth creates a pessimism that broadband would remain unaffordable to the majority of Nepalis in the decades to come. The proponents of “ICT for development” would argue that investments in ICT will eventually boost the economic growth and Internet will be affordable to the masses in the LDCs. But such a miracle has not been observed. The World Bank mentions that the benefits of ICT are not observable in the countries that lack so-called ‘analog compliments’ for a digital economy, i.e. the triplet of favorable business climate, strong human capital, and good governance. Imprecise as these terms are, they nonetheless imply that the relation between ICT and development is far from being universal. 

Another criticism on the relevance of ICTD research to the LDCs is based on the interdependencies of the critical infrastructure. The dependencies between critical infrastructures are observed during both the development and execution phases. The underdeveloped roads have posed challenges to Nepal Telecom’s ongoing project of laying optical fibers in the mid-hills. On the execution phase, a fully fledged usage of ICT demands a robust electricity infrastructure as it prerequisite. A research by Martin Chautari has shown that the Nepali ICT has already become a significant consumer of electrical energy despite Internet penetration is still low in the country.

One could argue that the ICT infrastructure can be run on imported electricity or diesel based generators. However, a dependence on imported energy would widen Nepal’s trade deficit. The trade deficit would also rise due to the import of ICT goods and related repair and maintenance materials as Nepal does not produce them. Thus any would-be benefit of a Nepal wide ICT infrastructure has to be seriously analyzed and contrasted with the trade deficits incurred by ICT goods imports.  Such an analysis has not been done hitherto by the policy makers.

My intention is to convey that ICT is not a mystical mantra that cures the illness of the society; the findings of ICTD research are based on several controversial methodological assumptions and socioeconomic premises.  It is despairing that the popular media has been highlighting the one sided version of the ICTD research. The views of the critiques also require acknowledgements.  In summary, investments for national ICT infrastructure should be preceded by a series of thorough investigations at the user end. “ICT for development” could be a target, but it is not an accepted principle.  

Thursday, December 1, 2016

What’s with the 512 Kbps?

512 is a number that we are likely to encounter if we have been using internet in Nepal. Nepali homes with internet pre-fibre and post-dial-up are likely to have installed a 512 Kbps unlimited cable internet connection. If you have subscribed to Nepal Telecom’s volume based ADSL internet, you have 512 Kbps speed. We can find 512 multiple times in current Nepali information technology narratives, visions and policies. Here are a few ways to look at 512 Kbps in the context of Nepal:
  1.  It is the proposed minimum download speed for wired-internet. For mobile-phone internet there is no minimum set in the documents.
  2. Connections offering download speed of 512 Kbps and above are labelled as broadband in the IT/ICT/Broadband literature.
  3. It is the common default fallback speed on high-speed fibre internet plans after the data volume has been exhausted.
  4. ISPs can also throttle your Mbps internet speed down to 512 Kbps anytime as part of their ‘fair usage policy’. If you are wondering why your 10 Mbps internet is delivering no faster than a 512 Kbps connection, data limit and throttling are usually the two main culprits.
  5.  It will take you 1 minute to download a smart-phone quality photograph from the Internet or 21 hours for a DVD-movie with a constant speed of 512Kbps.
  6. You will be charged Rs. 1 every 17 seconds on a Rs. 1 per MB mobile-phone Internet tariff delivering the data at a stable speed of 512 Kbps. If you are not paying at that rate means you are getting data at a much lower speed or browsing mostly text content.

  •  In IT related policy documents you can find targets mentioning 512 Kbps such as: “Broadband access will be expanded across the country with the goal of achieving a broadband Internet user penetration rate of 30% at a minimum of 512 kbps and making available at least 10 Mbps download speed on demand in urban areas by 2018” (Source: It is same as in India where the decision to classify 512 Kbps as broadband has been heavily criticized with some labelling it as ‘a joke’, ‘archaic and useless’, ‘out of touch with reality’ and a ‘move back to the dark ages’ (Too many to list here. Readers can google or bing: trai India broadband 512 and find out for themselves). The criticism comes due to the telecom regulators buckling to pressure from the telecom operators and a u-turn from the original 2Mbps recommendation given a few months earlier. Currently India’s average Internet speed puts it at the slowest in the Asia-Pacific region. A minimum of 512 Kbps now is akin to a dial-up connection where most countries in the world are far ahead. Such actions do not match with Nepal’s digital ambitions expressed in the policy documents. The (speed) gap is wide as ever.
  • The data cap and throttling creates a paradox. High speed internet is meant to be used for high bandwidth applications such as high-definition video and live TV streaming but data cap restricts such possibilities. For example, an internet user on a 10 Mbps connection will exhaust the 80 GB data cap after just 18 hours of full usage. A 4 hour per day usage can extend access to ‘fast’ internet for 5 days only, leaving 25 days on the buffered browsing experience. The reasons to why it may seem to last longer to a customer are because: (i) ISPs give the promised 10 Mbps on certain time slots only as an realization of their ‘fair usage policy’; and (ii) customers mostly use the internet for low bandwidth applications such as web browsing and VoIP (viber/skype voice chat) services anyways.
  •  Regulations do not cover the minimum speed at which ISPs can throttle the internet in Nepal. Currently ISPs actively throttle the bandwidth under the ‘fair usage policy’ (FUP). For example, ISPs can penalize the users that are responsible for heavy use of the bandwidth by slowing their connections to those applications. Throttling has been a thorny issue worldwide with frequent reports of intentional throttling by the ISPs. Such measure also raises questions on the nation’s net neutrality ambitions. The usual response from the ISPs, as is the case in India, is that without throttling they cannot provide fast internet at an affordable price. Therefore, though India wants to set the minimum at 512 Kbps, ISPs want it to be set at 64 Kbps at best. Nepal has no minimum but we can rest assured that we are being actively throttled at promised speeds and will be throttled down to a lower speed after data cap is hit. Worryingly, there is nothing we can do about it. Here is an example definition of what ‘fair’ is: “Day to day surfing, checking emails and occasional downloading will not get you into trouble. However, downloading a 200-300 MB movie clip/data every day is almost certain to. The key is to keep an eye on the amount that you are downloading. Then if you’re classed as a heavy or excessive user, restructure your internet usage pattern so that you are not uploading or downloading at peak times. If you don’t take heed of their warnings, system will slow down the speed of your connection after your data limit.” (Source:

    To put it simply, ISPs and telecoms want the users’ to stick to text-based website browsing. Ironically they themselves sell fast internet so that you can enjoy the multimedia experience. Indian telecom operators such as Airtel recently recommended a minimum of 64 Kbps, or an alternative, not set minimum speed on throttling in order for internet to be affordable. The original recommendation from TRAI (telecom regulator in India) was for a minimum of 2 Mbps broadband speed, which was quickly pulled down to a neither-here-nor-there 512 Kbps. The original recommendations were based on studies of internet traffic and applications which suggested a minimum of 1 Mbps for India. Given that our fibre internet connections comes from such Indian operators at their dictated price, throttling and data cap is even severe in Nepal. And its your fault for not adhering to the 'fair usage policy' if your connection speed slows down. Damned if you do, damned if you don’t.

A bit on a bit:
The space taken by a file stored in a computer is usually mentioned in Kilo Bytes (1 KB = about 1000 Bytes) or Mega Bytes (1 MB = about 1000 KB) or Giga Bytes (1 GB = about 1000 MB). From a users’ perspective, measurement in KB, MB and GB is commonly associated with average sizes of text documents (e.g. word document), images (e.g. pictures taken from a smart-phone) and movies (e.g. DVDs) respectively. It therefore makes intuitive sense for Internet data tariff to be mentioned in ‘per KB’ or ‘per MB’ usage. But download speed of an internet connection is historically stated in ‘bits per second’ (bps) because ‘a bit’ is the smallest piece of information that can be transmitted over the network. Since a bit is a very small unit we are familiar in seeing interned speed expressed in kilobits per second (Kbps) or megabits per second (Mbps).

For those who worry that 1000 is not 1024:
“It's convenient within the computer to organize things in groups of powers of 2. For example, 2^10 is 1024, and so a program might group 1024 items together, as a sort of "round" number of things within the computer. The term "kilobyte" above refers to this group size of 1024 things. However, people also group things by thousands -- 1 thousand or 1 million items.

There's this problem with the word "megabyte" .. does it mean 1024 * 1024 bytes, i.e. 2^20 which is 1,048,576, or does it mean exactly 1 million, 1000 * 1000. It's just a 5% difference, but marketers tend to prefer the 1 million, interpretation, since it makes their hard drives etc. appear to hold a little bit more. Also, the difference grows larger and larger for the gigabyte and terabyte sizes. In an attempt to fix this, the terms "kibibyte" "mebibyte" "gibibyte" "tebibyte" have been introduced to specifically mean the 1024 based units. These terms do not seem to have caught on very strongly thus far. If nothing else, remember that terms like "megabyte" have this little wiggle room in them between the 1024 and 1000 based meanings. We will never grade off for this distinction .. "about a million" will be our close-enough interpretation for "megabyte". The "error" at the megabyte level is about 5%. At the terabyte level the error is about 10%.” (Source: