1
00:00:01,480 --> 00:00:04,359
hello my name is Karen San and I am

2
00:00:04,359 --> 00:00:06,560
going to be presenting today the results

3
00:00:06,560 --> 00:00:09,160
of our study on the role of user agent

4
00:00:09,160 --> 00:00:11,719
interactions in Mobile money practices

5
00:00:11,719 --> 00:00:14,719
in Kenya and Tanzania this was work that

6
00:00:14,719 --> 00:00:17,039
was done in collaboration by researchers

7
00:00:17,039 --> 00:00:20,160
from kegi melon

8
00:00:21,680 --> 00:00:24,000
University so let's start with a quick

9
00:00:24,000 --> 00:00:26,840
motivation for studying mobile money

10
00:00:26,840 --> 00:00:28,720
Financial Services have traditionally

11
00:00:28,720 --> 00:00:30,880
been offered by Banks which has left

12
00:00:30,880 --> 00:00:32,000
many people

13
00:00:32,000 --> 00:00:34,800
unbounded in Africa and other emerging

14
00:00:34,800 --> 00:00:37,640
economies however mobile money is a

15
00:00:37,640 --> 00:00:39,800
service which is a service provided by

16
00:00:39,800 --> 00:00:42,520
Telos is transforming how millions of

17
00:00:42,520 --> 00:00:45,280
people can access various Banking and

18
00:00:45,280 --> 00:00:48,520
digital payment services mobile money is

19
00:00:48,520 --> 00:00:50,520
therefore helping to close the financial

20
00:00:50,520 --> 00:00:53,480
inclusion gaps for the unbanked many of

21
00:00:53,480 --> 00:00:55,879
whom have been from

22
00:00:55,879 --> 00:00:59,600
Africa mobile money which I will use the

23
00:00:59,600 --> 00:01:02,879
term mofo has reimagined what it means

24
00:01:02,879 --> 00:01:06,280
to be banked for example users do not

25
00:01:06,280 --> 00:01:09,840
need to have a bank account instead it's

26
00:01:09,840 --> 00:01:12,560
their phone number that acts as an

27
00:01:12,560 --> 00:01:15,560
account the services also leverages on

28
00:01:15,560 --> 00:01:20,119
the TCO infrastructure and relies on SMS

29
00:01:20,119 --> 00:01:22,880
and ussd which makes it accessible to

30
00:01:22,880 --> 00:01:25,920
feature phones which a large population

31
00:01:25,920 --> 00:01:28,280
still

32
00:01:28,439 --> 00:01:32,240
uses instead of Bank branches authorized

33
00:01:32,240 --> 00:01:34,759
third parties who are known as agents

34
00:01:34,759 --> 00:01:37,759
facilitate Last Mile access of mobile

35
00:01:37,759 --> 00:01:40,880
money agents are typically small

36
00:01:40,880 --> 00:01:43,200
businesses and shops within the

37
00:01:43,200 --> 00:01:46,560
community who often provide mobile Money

38
00:01:46,560 --> 00:01:48,719
Services in addition to their Core

39
00:01:48,719 --> 00:01:50,920
Business for example selling

40
00:01:50,920 --> 00:01:54,079
grocery agents earn a commission for

41
00:01:54,079 --> 00:01:57,280
extending mobile Money Services to the

42
00:01:57,280 --> 00:02:00,479
customers because of their accessibility

43
00:02:00,479 --> 00:02:02,680
they therefore offer a much greater

44
00:02:02,680 --> 00:02:06,439
reach that is at least 55 times that of

45
00:02:06,439 --> 00:02:10,479
banks in 20122 for example agents in

46
00:02:10,479 --> 00:02:13,160
Kenya digitized a value equating to a

47
00:02:13,160 --> 00:02:18,280
third of Kenya's GDP in just 6

48
00:02:18,519 --> 00:02:21,120
months so in essence we can think of

49
00:02:21,120 --> 00:02:24,160
mobile money as two categories of

50
00:02:24,160 --> 00:02:27,440
interactions the first is the user Momo

51
00:02:27,440 --> 00:02:30,000
interaction which you can think of as a

52
00:02:30,000 --> 00:02:33,080
adoption and the second is the user

53
00:02:33,080 --> 00:02:34,319
agent

54
00:02:34,319 --> 00:02:37,400
interaction to date the user Momo

55
00:02:37,400 --> 00:02:40,280
interaction that is adoption has

56
00:02:40,280 --> 00:02:42,599
received enormous attention and has been

57
00:02:42,599 --> 00:02:46,239
well researched in the adoption

58
00:02:46,239 --> 00:02:49,280
literature the user agent interaction

59
00:02:49,280 --> 00:02:51,760
however has not been studied much and

60
00:02:51,760 --> 00:02:54,120
especially not in the context of privacy

61
00:02:54,120 --> 00:02:56,720
and security this is particularly

62
00:02:56,720 --> 00:02:59,239
important because in addition to

63
00:02:59,239 --> 00:03:02,599
digitizing money agents end up wearing

64
00:03:02,599 --> 00:03:06,319
many other Hearts including onboarding

65
00:03:06,319 --> 00:03:09,519
which entails registering new users

66
00:03:09,519 --> 00:03:11,840
typically new users have to register

67
00:03:11,840 --> 00:03:15,360
their SIM card using an identification

68
00:03:15,360 --> 00:03:17,799
normally that would be a national ID

69
00:03:17,799 --> 00:03:21,680
card agents also end up educating new

70
00:03:21,680 --> 00:03:24,799
and old users on how to use Momo and

71
00:03:24,799 --> 00:03:27,959
ensure compliance with identity proofing

72
00:03:27,959 --> 00:03:30,799
requirements also known as know your

73
00:03:30,799 --> 00:03:34,480
customer that is kyc and kyc simply

74
00:03:34,480 --> 00:03:36,959
ensures that momu has a clear picture of

75
00:03:36,959 --> 00:03:40,640
who their customer is kyc is also

76
00:03:40,640 --> 00:03:43,319
essential in the fight against money

77
00:03:43,319 --> 00:03:47,280
laundering and anti-terrorism

78
00:03:47,280 --> 00:03:50,799
financing agents also provide customer

79
00:03:50,799 --> 00:03:56,239
support whenever users run into any

80
00:03:56,239 --> 00:03:58,640
problems so to understand the privacy

81
00:03:58,640 --> 00:04:00,360
and Security in one

82
00:04:00,360 --> 00:04:03,280
around user agent interaction we

83
00:04:03,280 --> 00:04:05,840
conducted qualitative studies in Kenya

84
00:04:05,840 --> 00:04:08,159
and Tanzania we chose these two

85
00:04:08,159 --> 00:04:10,640
countries because East Africa has been a

86
00:04:10,640 --> 00:04:13,599
Trailblazer for mobile money and so we

87
00:04:13,599 --> 00:04:16,680
interviewed 36 mobile money users in

88
00:04:16,680 --> 00:04:22,400
each country resulting in a total of 72

89
00:04:22,400 --> 00:04:25,960
users let's now take one step back so we

90
00:04:25,960 --> 00:04:28,600
can understand how mobile money Works

91
00:04:28,600 --> 00:04:30,919
before we highlight the findings of the

92
00:04:30,919 --> 00:04:34,000
study once registered a user can

93
00:04:34,000 --> 00:04:36,199
complete one of three types of

94
00:04:36,199 --> 00:04:39,360
transactions that is cashing in cashing

95
00:04:39,360 --> 00:04:44,320
out and P2P or pass to

96
00:04:45,160 --> 00:04:47,600
person let's start with the first one

97
00:04:47,600 --> 00:04:50,639
which is cashing in also called

98
00:04:50,639 --> 00:04:53,160
depositing this starts with the user

99
00:04:53,160 --> 00:04:56,639
handing physical cash to the agent the

100
00:04:56,639 --> 00:04:59,479
agent maintains a balance that is called

101
00:04:59,479 --> 00:05:03,160
a float and if the agent has sufficient

102
00:05:03,160 --> 00:05:06,280
float he will accept Alice's physical

103
00:05:06,280 --> 00:05:09,600
cash in exchange of mobile money that he

104
00:05:09,600 --> 00:05:13,880
will transfer to Alice's phone via

105
00:05:13,880 --> 00:05:18,720
SMS similar to cashing in cashing out

106
00:05:18,720 --> 00:05:21,639
which is withdrawing starts with Bob

107
00:05:21,639 --> 00:05:24,280
transferring monies from his phone to

108
00:05:24,280 --> 00:05:27,440
the agent's mobile money account and so

109
00:05:27,440 --> 00:05:30,400
long as the agent has sufficient flow

110
00:05:30,400 --> 00:05:33,600
he will accept the transfer and hand Bob

111
00:05:33,600 --> 00:05:35,919
the physical cash equivalent of the

112
00:05:35,919 --> 00:05:38,440
amount that he has

113
00:05:38,440 --> 00:05:40,919
transferred in personto person

114
00:05:40,919 --> 00:05:43,160
transactions agent mediation is

115
00:05:43,160 --> 00:05:46,319
typically not required Alice can simply

116
00:05:46,319 --> 00:05:49,240
initiate a transfer from her phone to

117
00:05:49,240 --> 00:05:52,160
Bob's phone via SMS or

118
00:05:52,160 --> 00:05:56,160
ussd and using Bob's phone number as a

119
00:05:56,160 --> 00:05:57,199
receiving

120
00:05:57,199 --> 00:06:00,199
account when transacting at the agents

121
00:06:00,199 --> 00:06:03,120
that is cashing in and Cashing Out users

122
00:06:03,120 --> 00:06:05,880
are required to show their ID in order

123
00:06:05,880 --> 00:06:09,639
to comply with kyc regulations which is

124
00:06:09,639 --> 00:06:12,280
know your

125
00:06:16,039 --> 00:06:18,720
customer the agent checks the physical

126
00:06:18,720 --> 00:06:21,000
ID to authenticate the customer by

127
00:06:21,000 --> 00:06:24,240
comparing the details on the ID with

128
00:06:24,240 --> 00:06:26,840
those returned from the Momo after they

129
00:06:26,840 --> 00:06:30,960
have initiated a transaction

130
00:06:32,680 --> 00:06:35,039
now that we have an understanding of how

131
00:06:35,039 --> 00:06:38,120
Momo works I will now focus on a subset

132
00:06:38,120 --> 00:06:40,479
of our findings by highlighting three

133
00:06:40,479 --> 00:06:43,400
important results first we found that

134
00:06:43,400 --> 00:06:46,199
users and agents design work arounds in

135
00:06:46,199 --> 00:06:49,479
response to challenges within the Momo

136
00:06:49,479 --> 00:06:52,400
ecosystem second whereas the choice of

137
00:06:52,400 --> 00:06:54,919
an agent had a lot to do with

138
00:06:54,919 --> 00:06:58,199
convenience we also observed that a key

139
00:06:58,199 --> 00:07:00,400
motivating factor was the need to

140
00:07:00,400 --> 00:07:03,840
mitigate any privacy and security risks

141
00:07:03,840 --> 00:07:07,639
arising from this ecosystem and third we

142
00:07:07,639 --> 00:07:10,280
found that workarounds in the long run

143
00:07:10,280 --> 00:07:12,840
introduced more privacy and security

144
00:07:12,840 --> 00:07:16,239
challenges in over

145
00:07:19,360 --> 00:07:22,360
overall let's dive into the first

146
00:07:22,360 --> 00:07:25,759
finding we observed that workarounds

147
00:07:25,759 --> 00:07:28,919
were designed by users and agents in all

148
00:07:28,919 --> 00:07:31,120
the phases of the

149
00:07:31,120 --> 00:07:33,479
transaction for example in the

150
00:07:33,479 --> 00:07:36,120
transaction execution phase the

151
00:07:36,120 --> 00:07:38,800
workarounds entailed changing the

152
00:07:38,800 --> 00:07:42,000
transaction execution process such by

153
00:07:42,000 --> 00:07:44,919
such as by involving proxies and

154
00:07:44,919 --> 00:07:48,080
modifying the role of the agent as well

155
00:07:48,080 --> 00:07:51,120
as modification of the transaction

156
00:07:51,120 --> 00:07:53,879
characteristics such as the size and the

157
00:07:53,879 --> 00:07:56,000
location of the

158
00:07:56,000 --> 00:07:59,840
transaction in the kyc phase agent and

159
00:07:59,840 --> 00:08:02,599
users collaboratively worked together to

160
00:08:02,599 --> 00:08:06,879
modify how kyc was completed as well as

161
00:08:06,879 --> 00:08:10,759
changing what was given for kyc that is

162
00:08:10,759 --> 00:08:15,039
in place of the user's physical

163
00:08:17,800 --> 00:08:21,319
ID we also note that the workarounds

164
00:08:21,319 --> 00:08:24,039
were motivated by challenges within the

165
00:08:24,039 --> 00:08:26,960
Momo ecosystem that were not limited to

166
00:08:26,960 --> 00:08:30,000
privacy and security but which in end

167
00:08:30,000 --> 00:08:32,039
had serious privacy and security

168
00:08:32,039 --> 00:08:34,279
implications that I will highlight later

169
00:08:34,279 --> 00:08:35,760
in this

170
00:08:35,760 --> 00:08:38,000
presentation in the transaction phase

171
00:08:38,000 --> 00:08:40,839
for example some of the motivators

172
00:08:40,839 --> 00:08:44,159
included navigating transaction costs

173
00:08:44,159 --> 00:08:47,120
seeking more convenience Network

174
00:08:47,120 --> 00:08:50,720
downtime insufficient fla security

175
00:08:50,720 --> 00:08:53,040
concerns and data

176
00:08:53,040 --> 00:08:56,000
privacy with regards to network downtime

177
00:08:56,000 --> 00:08:59,959
for example k01 said

178
00:08:59,959 --> 00:09:02,120
if there's a delay I will just have to

179
00:09:02,120 --> 00:09:04,240
leave the money with my details and

180
00:09:04,240 --> 00:09:07,640
phone number users often went ahead to

181
00:09:07,640 --> 00:09:09,760
explain that they wrote down this

182
00:09:09,760 --> 00:09:12,440
information on papers which they left

183
00:09:12,440 --> 00:09:15,240
with the agent and also left the money

184
00:09:15,240 --> 00:09:17,160
with the agent and trusted that they

185
00:09:17,160 --> 00:09:19,200
would complete the transaction into

186
00:09:19,200 --> 00:09:21,079
their account

187
00:09:21,079 --> 00:09:25,399
later TZ 32 was among those who were

188
00:09:25,399 --> 00:09:27,800
concerned about agents knowing their

189
00:09:27,800 --> 00:09:30,640
transaction details and using this

190
00:09:30,640 --> 00:09:34,079
information to perpetrate fraud I don't

191
00:09:34,079 --> 00:09:36,959
feel good because one can't know the

192
00:09:36,959 --> 00:09:39,800
agents intentions they know your

193
00:09:39,800 --> 00:09:42,040
transaction details they can tell

194
00:09:42,040 --> 00:09:45,279
someone else and then you end up being

195
00:09:45,279 --> 00:09:48,839
robbed with regards to data privacy we

196
00:09:48,839 --> 00:09:52,200
observed varied perceptions on the role

197
00:09:52,200 --> 00:09:55,600
and impact of data sharing with agents

198
00:09:55,600 --> 00:09:59,040
users like tz6 for example said they

199
00:09:59,040 --> 00:10:00,839
Trust Ed agents to keep their

200
00:10:00,839 --> 00:10:04,360
information safe while others like

201
00:10:04,360 --> 00:10:07,680
k25 mentioned that agents sometimes used

202
00:10:07,680 --> 00:10:11,200
people's IDs to register additional SIM

203
00:10:11,200 --> 00:10:13,480
cards which they sold to other

204
00:10:13,480 --> 00:10:16,440
individuals without an ID to register

205
00:10:16,440 --> 00:10:20,360
their own SIM card to use for

206
00:10:25,279 --> 00:10:29,040
M while M more users felt that kyc was

207
00:10:29,040 --> 00:10:32,320
necess necessary for security many also

208
00:10:32,320 --> 00:10:35,320
spoke of the challenges with using IDs

209
00:10:35,320 --> 00:10:36,920
for identity

210
00:10:36,920 --> 00:10:40,000
proofing with regards to kyc being

211
00:10:40,000 --> 00:10:45,680
essential for security K ke20 and

212
00:10:45,680 --> 00:10:49,360
k21 highlighted how providing an ID

213
00:10:49,360 --> 00:10:51,959
ensured that one was not feigning their

214
00:10:51,959 --> 00:10:55,040
identity and how this ensured the

215
00:10:55,040 --> 00:10:59,360
security of momo overall

216
00:10:59,360 --> 00:11:02,360
while many had similar sentiments most

217
00:11:02,360 --> 00:11:04,480
participants also expressed

218
00:11:04,480 --> 00:11:06,839
dissatisfaction for example the

219
00:11:06,839 --> 00:11:09,600
potential denial of service if if one

220
00:11:09,600 --> 00:11:11,800
forgot to carry their ID when they

221
00:11:11,800 --> 00:11:15,079
needed to transact participants also

222
00:11:15,079 --> 00:11:18,920
felt that checking each customer's ID

223
00:11:18,920 --> 00:11:21,880
was really not practical since agents

224
00:11:21,880 --> 00:11:27,600
facilitated many transactions in a day

225
00:11:31,600 --> 00:11:34,560
as a result of the kyc challenges we

226
00:11:34,560 --> 00:11:37,639
observed several workarounds related to

227
00:11:37,639 --> 00:11:41,079
kyc which as I mentioned before can be

228
00:11:41,079 --> 00:11:44,320
broadly categorized as changing how kyc

229
00:11:44,320 --> 00:11:48,560
was provided and what was provided for

230
00:11:48,560 --> 00:11:53,320
kyc K15 for example mentioned that they

231
00:11:53,320 --> 00:11:56,160
used a cousin's ID since they had

232
00:11:56,160 --> 00:11:58,680
registered their sim card with that ID

233
00:11:58,680 --> 00:12:00,760
this is really akin to using someone

234
00:12:00,760 --> 00:12:03,240
else's social security number and

235
00:12:03,240 --> 00:12:05,839
account as your own many other

236
00:12:05,839 --> 00:12:09,120
participants like K20 reported just

237
00:12:09,120 --> 00:12:12,320
giving the agent the ID number instead

238
00:12:12,320 --> 00:12:15,480
of the actual ID while some like

239
00:12:15,480 --> 00:12:18,480
k31 mentioned that the agents they used

240
00:12:18,480 --> 00:12:21,199
have memorized their ID numbers and so

241
00:12:21,199 --> 00:12:24,920
they do not need to provide

242
00:12:26,920 --> 00:12:30,959
that as a result of the in inherent and

243
00:12:30,959 --> 00:12:34,279
created risks we therefore observed in

244
00:12:34,279 --> 00:12:37,959
our second finding that how users chose

245
00:12:37,959 --> 00:12:41,560
agents was driven by both convenience

246
00:12:41,560 --> 00:12:44,079
and the desire to mitigate the privacy

247
00:12:44,079 --> 00:12:47,120
and security risks within the Momo

248
00:12:47,120 --> 00:12:51,079
ecosystem for example at the transaction

249
00:12:51,079 --> 00:12:54,399
execution stage participants sometimes

250
00:12:54,399 --> 00:12:57,720
modified the transaction size and the

251
00:12:57,720 --> 00:12:59,800
location where they

252
00:12:59,800 --> 00:13:01,560
transacted

253
00:13:01,560 --> 00:13:05,839
k08 said they divide the transaction to

254
00:13:05,839 --> 00:13:09,000
two maybe three times and do one

255
00:13:09,000 --> 00:13:12,000
transaction at One agent and another at

256
00:13:12,000 --> 00:13:14,720
a separate agent and they do so for

257
00:13:14,720 --> 00:13:16,920
security

258
00:13:16,920 --> 00:13:20,880
purposes TZ 33 also changed the location

259
00:13:20,880 --> 00:13:26,720
of his transactions for security reasons

260
00:13:32,959 --> 00:13:35,880
sending someone else as a proxy to

261
00:13:35,880 --> 00:13:39,480
transact threed the kyc procedures of

262
00:13:39,480 --> 00:13:43,279
authenticating the customer as a result

263
00:13:43,279 --> 00:13:44,680
like

264
00:13:44,680 --> 00:13:47,800
ke18 who said that the person he sent

265
00:13:47,800 --> 00:13:51,000
had to be known by the agent Mumu users

266
00:13:51,000 --> 00:13:53,120
tended to stick with agents who knew

267
00:13:53,120 --> 00:13:55,399
their elected proxies at least to an

268
00:13:55,399 --> 00:14:00,920
extent to facilitate proxy transactions

269
00:14:05,560 --> 00:14:09,199
our third and final finding suffices new

270
00:14:09,199 --> 00:14:12,800
privacy and security risks emerging from

271
00:14:12,800 --> 00:14:15,880
the different workarounds that M users

272
00:14:15,880 --> 00:14:18,680
adopted in general most of the

273
00:14:18,680 --> 00:14:21,360
workarounds thwarted any privacy and

274
00:14:21,360 --> 00:14:24,120
security measures that Mumu had put in

275
00:14:24,120 --> 00:14:26,079
place and especially

276
00:14:26,079 --> 00:14:29,199
kyc this left room for more vulnerable

277
00:14:29,199 --> 00:14:30,959
abilities such as

278
00:14:30,959 --> 00:14:35,360
repudiation for example T TZ 17 shared

279
00:14:35,360 --> 00:14:38,440
how someone who's using another person's

280
00:14:38,440 --> 00:14:41,959
SIM card may deny receiving money thus

281
00:14:41,959 --> 00:14:47,079
making P2P transactions difficult to

282
00:14:50,040 --> 00:14:52,600
authenticate another broad risk was

283
00:14:52,600 --> 00:14:55,079
related to higher chances of fraud

284
00:14:55,079 --> 00:14:59,199
because of the broken kyc

285
00:14:59,600 --> 00:15:01,680
as explained further by

286
00:15:01,680 --> 00:15:05,279
k17 the data Rich environment also made

287
00:15:05,279 --> 00:15:08,320
Momo users more vulnerable to fraud

288
00:15:08,320 --> 00:15:12,680
leading to a potential loss of their

289
00:15:13,320 --> 00:15:16,720
money in general this work highlights

290
00:15:16,720 --> 00:15:18,560
three important

291
00:15:18,560 --> 00:15:21,279
takeaways first we have seen that

292
00:15:21,279 --> 00:15:23,959
workarounds are often motivated by

293
00:15:23,959 --> 00:15:27,560
efforts to limit risk and uncertainity

294
00:15:27,560 --> 00:15:30,959
as well as to overcome challenges within

295
00:15:30,959 --> 00:15:34,759
Momo second we have seen that user work

296
00:15:34,759 --> 00:15:37,639
arounds are inconsistent with their

297
00:15:37,639 --> 00:15:40,240
security preferences especially with

298
00:15:40,240 --> 00:15:43,279
regards to kyc and this is likely

299
00:15:43,279 --> 00:15:45,480
because of the but and some

300
00:15:45,480 --> 00:15:49,120
processes and last we have seen how M

301
00:15:49,120 --> 00:15:52,120
users have varied perceptions that

302
00:15:52,120 --> 00:15:57,120
impact their data sharing with agents

303
00:16:00,079 --> 00:16:02,160
following these insights our

304
00:16:02,160 --> 00:16:04,040
recommendations are as

305
00:16:04,040 --> 00:16:07,000
follows one there is a need to design

306
00:16:07,000 --> 00:16:09,480
for privacy and security at the

307
00:16:09,480 --> 00:16:12,680
interface between agents and users

308
00:16:12,680 --> 00:16:15,160
especially given that agents continue to

309
00:16:15,160 --> 00:16:17,639
play a very important role in the

310
00:16:17,639 --> 00:16:20,759
financial inclusion of people who would

311
00:16:20,759 --> 00:16:22,480
otherwise be

312
00:16:22,480 --> 00:16:26,319
unbanked second since the importance of

313
00:16:26,319 --> 00:16:29,279
kyc cannot be underestimated

314
00:16:29,279 --> 00:16:31,519
there is a need to think about

315
00:16:31,519 --> 00:16:33,839
registration and identity proofing

316
00:16:33,839 --> 00:16:36,920
mechanisms for those who lack

317
00:16:36,920 --> 00:16:39,639
identification finally we highlight the

318
00:16:39,639 --> 00:16:42,680
need to improve the usability of momo to

319
00:16:42,680 --> 00:16:45,720
address some of the motivations for user

320
00:16:45,720 --> 00:16:48,240
workarounds with that I would like to

321
00:16:48,240 --> 00:16:50,360
thank you for listening and I encourage

322
00:16:50,360 --> 00:16:52,639
you to have a look at the paper if you

323
00:16:52,639 --> 00:16:56,680
would like to learn more