measuring digital inequality in sa - research ict africa...measuring digital inequality in sa alison...
TRANSCRIPT
Measuring digital inequality in SAAlison Gillwald (PhD)
Executive Director: Research ICT Africa
Data enquiries:Onkokame Mothobi (PhD) [email protected]
WEF Internet for All Data Working Group 13 February 2018, Johannesburg
Sampling Sampling is based on Census sample frames. Census divides a country in Enumerator Areas (EAs) with roughly a household density of 200
2
National Census Sample Frame or Census
Split enumerator areas (EAs) into rural EAs
Separate PPS sampling of EAs for rural and
Listing of all households (HH) and all
Simple Random Sampling of X
households from
Simple Random Sampling of Y
businesses from Simple Random Sampling of 1
SDG - ICT targets
‣ 7 ICT indicators covering 6 targets under Goals 4, 5, 9, and 17 • Target 4a: Proportion of schools with access to the Internet for pedagogical purposes • Target 4a: Proportion of schools with access to computers for pedagogical purposes • Target 4.4: Proportion of youth/adults with ICT skills, by type of skills • Target 5b: Proportion of individuals who own a mobile telephone, by sex (ITU) • Target 9c: Percentage of the population covered by a mobile network, broken down by technology
(ITU) • Target 17.6: Fixed Internet broadband subscriptions, broken down by speed (ITU) • Target 17.8: Proportion of individuals using the Internet (ITU)3
Most popular include the World Economic Forum’s Network Readiness Index (NRI), the ITU’s ICT Development Index (IDI), the Alliance for Affordable Internet’s Affordability Drivers Index (ADI), GSMA’s Mobile Connectivity Index and the Inclusive Internet Index (3i) from the Economist Intelligence Unit/ Facebook.
Indices - common problems
‣ Summing up indicators that explain the same factor (muti-colinearity)
‣ Summing up indicators that are all highly correlated to GDP per capita
‣ Measuring affordability as prices expressed as % of GDP per capita, masking inequality (SA one of highest GINI co-efficients in the world)
‣ Using first world indicators such as fixed- line penetration and wired broadband
4
Comparing rankings
Source: Esselaar S, Gillwald A & Stork C (2017) Analysis Instead of Summation: Why Indices Are Not Enough for ICT Policy and Regulation https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3043719
5
Table 1: Comparing rankings against
selected ICT indicators
Rankings ICT Indicators
ADI 3i IDI NRI MCI1 GB
prepaid data USD
Active SIM cards per
100
Fixed-line per 100
Nigeria 13 45 137 119 98 3,20 83 0,10
Kenya 30 51 129 86 105 5,00 82 0,19
Ghana 26 49 112 102 96 2,46 128 1,01
Namibia 31 NA 120 99 NA 5,89 99 7,42
Brazil 10 18 63 72 56 8,48 124 21,01
Sources A4AI 2017
EIU 2017
ITU 2016
WEF 2016
GSMA 2016
RIS / RIA Q4 2016
ITU 2016
ITU 2016
Sample
6
Table 1: SA EA sampling Rural Urban Total
Sample of households 720 1080 1800
HHs per EA 24 24 24
Sampled EAs 30 45 75
Two sampling frames were available for South Africa based on the 2011 census: Enumerator Areas (EAs) and Small Layers (SAL)
Out of 90, 425 EAs, a total of 75 EAs were sampled using SRS 47 urban areas and 32 rural areas were sampled ( 60% urban : 40% rural split)
The target sample of 1800 was spilt in to 60% urban and 40% rural, yielding a target of 720 rural and 1080 urban households
Household and individual weights were then constructed based on the census and world data on urban and rural, male and female proportions
Individual sample structure
7
Male Female Urban RuralRIA, 2017 45% 55% 65% 36%Census, 2011 49% 51% 60% 40%Student 18% 18% 19% 17%Unpaid housework 1% 13% 11% 6%Retired 10% 11% 9% 12%Unemployed, active 22% 24% 28% 20%Unemployed, not active 7% 9% 12% 6%Disabled, not active 1% 1% 1% 1%Employed 33% 21% 16% 32%Self employed 8% 3% 3% 6%
Household Indicators
8
Perc
enta
ges
0
20
40
60
80
Fixed line Computer Radio TV Internet
11
7967
87 20
7862
261835
7568
2115
2011/Census 2012/RIA 2017/RIA
Household access to electricity
9
National Urban RuralGhana 85% 96% 67%Kenya 42% 86% 19%
Mozambique 24% 46% 10%Nigeria 64% 84% 44%Rwanda 29% 60% 20%
South Africa 89% 90% 86%Tanzania 33% 61% 19%
By international standard South Africa is highly electrified and the disparity between urban and rural is insignificant. Ghana’s urban areas are more electrified than South Africa with one but rural areas in SA are more developed than Ghana.
Household access to electricity vs Mobile and Internet penetration - individual level
10
Electricity Mobile InternetGhana 85% 78% 27%Kenya 42% 87% 27%Mozambique 24% 40% 9%Nigeria 64% 65% 29%Rwanda 29% 46% 7%South Africa 89% 85% 48%Tanzania 33% 59% 14%
More developed countries have accelerated adoption rates than low income countries
Household access and use of the Internet
11
Smartphone effect
ADSL
Mobile phone
USB Dongle/MIFI
Fiber
Percentages0 10 20 30 40
2,1
34,71
36
36,5
12
Household reasons for not using the Internet
0
5
10
15
20
25
30Costofequipmenttoohigh
Costofservicetoohigh
DonotneedtheInternet
HaveaccesstotheInternetelsewhere
Internetnotavailable inthearea
Donotknowhowtouseit
Privacyorsecurityconcerns
Individual Adoption of ICT
13
National Male FemaleNo bank account 40% 37% 42%
Yes, my own 57% 60% 54%I use someone 3% 2% 4%
computer/laptop 14% 17% 12%Mobile phone 84% 84% 85%Basic Phone 36% 33% 39%
Feature Phone 8% 7% 10%Smartphone 56% 60% 52%
There is a slight gender gap in the adoption of computers and smartphones. In South Africa the adoption of smartphones has reached the critical mass and the
gap between male and females is diminishing
Comparison of adoption patterns between 2012 and 2017
14
Perc
enta
ges
0
25
50
75
100
Mobile phone Ownership Prepaid Smartphone More than 1 active SIM card
19,4
55,5
91,383,2
14,2
51
87,584,2
2012 2017
Smartphone penetration increasing while mobile phone adoption is saturated. Shifts only observed in the type of mobile phone with demand for smartphone
increasing.
Adoption of ICT - urban rural divide
15
National Urban Rural
Basic Phone 36% 25% 39%
Feature Phone 8% 7% 8%
Smartphone 56% 54% 33%
Adoption of ICT by age group
16
15-24 25-54 55-64 65+Sample structure 26% 53% 12% 9%
Basic Phone 14% 31% 51% 45%
Feature Phone 9% 6% 8% 6%
Smartphone 56% 51% 28% 19%
Internet 78% 72% 58% 37%
Smartphone driving Internet adoption
17
Yes With difficulty Not at allRead
National 78% 16% 6%Mobile phone 88% 74% 57%
Internet 73% 50% 0%Write
National 77% 16% 6%Mobile phone 88% 77% 55%
Internet 73% 16% 0%There is a slight gender gap in the adoption of smartphones in South Africa the adoption of smartphone has reached the critical mass/tipping point and the gap
between male and females is diminishing
Literacy rate and adoption of mobile phones and Internet
Device used to access the Internet for the first time
18
Perc
enta
ges
0
10
20
30
40
Male Female Urban Rural
0,41,20,816,1
16,5
10,915,1
30
39,434,6
37,9
Mobile phone Computer Tablet
Employment status, adoption and affordability
19
Mobile phone Internet Internet
expensiveStudent 79% 76% 56%Unpaid housework 84% 77% 59%Retired 76% 37% 41%Unemployed, active 84% 64% 44%Unemployed, not active 76% 50% 47%
Disabled, not active 70% 36% 36%Employed 91% 83% 43%Self employed 91% 77% 37%
Unemployment, Adoption and affordability
20
Price
(ZAR
)
0
50
100
150
200
2014Q2 2014Q4 2015Q2 2015Q4 2016Q2 2016Q4 2017Q2 2017Q4
Cell C MTN Vodacom Telkom Virgin Mobile
Source: RIA African Mobile Pricing (RAMP) Index
Price of 1G data
21
Data prices in South Africa are high and remain unaffordable to the poor. The reason behind the 50% of the poor that are still not connected.
Egypt
Ghana
Uganda
Tunisia
Nigeria
0 0,7 1,4 2,1 2,8
2,8
2,8
2,27
2,24
1,69
Source: RIA African Mobile Pricing (RAMP) Index
Price, Internet penetration
22
0
12,5
25
37,5
50
South Africa Nigeria Kenya Ghana Tanzania Mozambique Rwanda
Internet 1GB Price
79
14
272729
48
2,392,642,252,272,942,87,267,26
2,8 2,94 2,27 2,25 2,64 2,39
Saving mechanisms
23
Special data promotions
Public WiFi
Home Internet
None
0 12,5 25 37,5 50
7,67
21,67
31,09
49,72
Majority of South African residents access Internet on a mobile phone daily. The study shows that most South African Internet users use special data packages (49.72%) and Public WiFi (31%) as a saving mechanism.
Activities on the Internet
24
18%
10%
10%
10% 14%
37%
Social Media Game apps News EducationalSearch tools Weather
Activities on the Internet
25
Internet in South Africa is mostly used for social networking
2012 2017 Urban Rural Urban RuralEducational 9,8% 44,0% 44,7% 41,9% 45% 43%
Social networking 51,7% 73,2% 75,1% 67,7% 71% 76%Work related 26,9% 30,4% 17,0%
Shopping 3,2% 10,4% 12,8% 3,5%Government
services4,9% 8% 7% 8%
Job searching 26,1% 26,1% 26,2%Online banking 3,8% 16,8% 20,8% 5,5% 32% 22%
Reasons for not using the Internet
26
36% of those who do not use the Internet, stated that lack of devices as the main inhibitor to using the Internet. Affordability is also an issue with more females stating that Internet is more expensive
National Male FemaleNo access device 36% 38% 35%
No interest 38% 15% 16%Dont know how to use it 9% 12% 7%
Too expensive 15% 11% 18%Spouse or parents dont allow 5% 1% 8%
Reasons for no mobile phone
27
2012 2017I cannot afford it 63% 43%
No mobile coverage 2% 1%No electricity 9% 1%
Privacy concern 12%
National
Did you use mobile phone in the past three months 38%
How many active SIM cards do you have
0 1%
1 94%
2+ 5%
Do you plan to buy a phone in the next 6 months 39%
Mobile money vs Bank account ownership
28
Mobile money Bank accountGhana 54% 27%Kenya 83% 34%
Mozambique 11%Nigeria 3% 34%Rwanda 26% 28%
South Africa 8% 57%Tanzania 45% 11%
Mobile money is common in countries where majority of residents are unbanked. South Africa use more banking facilities than in other countries
Mobile money vs bank account ownership
29
Majority of those who subscribed to mobile money use it to purchase airtime voucher
Airtime top up
Bill payments
Salary payments
Insurance payments
Receive pension
0 12,5 25 37,5 50
3,89
7,67
12,14
22,58
48,49
Reasons for not using mobile money
30
Most people do not use mobile money in SA because there is no one to send to (and though large numbers underbanked, banked.
0510152025303540
Donothaveamobilephone
Noonetosendmoneytoorreceivemoney
from
Mobilemoneyisexpensive
Mobilemoneyistoocomplicated
Idonottrustmobilemoney
social media use and devices used for access
31
Most people use smartphones to access their social media
58%42%
Social media No
Perc
enta
ges
0
22,5
45
67,5
90
Computer Feature phone Smartphone Tablet
3,7
80,4
10,810,5
Information shared on social media / Privacy issues
32
National Male Female
Real name 73,3% 72,0% 74,4%
Gender 62,7% 62,1% 63,2%
Marital status 56,1% 57,1% 55,1%
Mobile number 46,3% 45,7% 46,9%
Pictures and video of family members 36,5% 38,0% 35,1%
Religion 15,0% 53,2% 52,9%
Political view 28,1% 24,4% 31,4%
Sexual orientation 11,1% 13,6% 8,9%
Social media activities
33
Social media is mostly used for chatting with friends and making online calls
0
10
20
30
40
50
60
70
80
90
100Chatting
Onlinecalls
Sharingvideo/pictures/music
makingnewfriends
Readingnews
Playinggameslookingforeducationalcontent
Gettingopinion/sharingexperience
Makingproffessionalandbusinesscontacts
Marketingproducts/services
Followinggovernmentpages
Microwork
34
Males are more likely to do online jobs than females except for shopping and ride hailing
Perc
enta
ges
0
12,5
25
37,5
50
Ride hailing Shopping for delivery Online tasks Cleaning and laundry
16,0711,259,32
5,37
12,68
42,55
7,374,34
Male Female
10%
34%
37%
20%
FlexibilityFill the gaps in flactuations of income to gain experienceJuly
Quality of service
35
Vodacom performs better in both measures followed by MTN. Telkom performance, however, is worse in the two measures. This provides some explanation for why consumers are not migrating to the cheapest Telkom services.
Data source: SpeedCheckers, 2018
Using the value for money index, constructed from 1GB data prices, download and upload speed, shows between 2014Q1 and 2015Q3 the two dominant operators Vodacom and MTN offered high quality. In the same period Telkom’s quality was the lowest. However, since 2016Q1, it seems that smaller operators improved their quality (reflected in increased investments) catching up with dominant operators in 2 0 1 6 Q 2 . Vo d a c o m S A’s h i g h p r i c e s a re accompanied by higher Internet speeds compared to MTN SA and Cell C, which are performing more poorly on the Value for Money Index, based on average download/upload speed (in Mbps) divided by 1GB basket costs.
Latency- The previous figures show the logarithmic latency to Measurement Lab servers around the world from South Africa as measured from our MySpeed Test probes since 2015.
- In the first figure, between 2015 and 2017, on average, all operators had relatively high latency levels, with MTN and Telkom mobile having the highest levels of variability.
- When we analyse latency on different years, in 2015 and 2017, both MTN and Vodacom had very variable latency levels, although they also had the lowest levels of latency.