Transcript El control fiscal en la Economía Digital

Roma
Fiscal Control in the
Digital Economy l
Raul Zambrano - CIAT
September 2015
1
Digital Economy
• Devices connected to the Internet. PC
and mobile.
• App stores
• 27 average applications
• 900 million users of a single
messaging application
• 12 among the first one hundred most
valuable companies are from the
technology sector. Including the first
3.
• Global markets
2
B2C models
• On-line sales of traditional goods (Amazon, virtual stores of brick and mortar stores,
airline tickets)
• On-line sales of digital goods (iTunes)
• Streaming services (Netflix, Spotify)
• Low/no cost applications
• Free up-to
• “Freemium”. Functionality, capacity, virtual goods, features.
• Ads sold to third parties.
• Selling information, untreated or otherwise.
• E-mail, storage, note taking, virtual machines, photography, IM, micro-blogging,
video, travel agencies.
3
B2B models
• Platform interconnection
• Integration of logistics
• Corporate activities
• Cloud Services
• SaaS, PaaS, IaaS
• Ej. (Office 365, Salesforce;
• Operations based on off-the-country
sites
• Operations with related parts.
4
C2C models
• Operations supported by third party apps or services
• Sales of physical goods (eBay, Urgente24)
• Apps to support services in the sharing economy (Uber, Airbnb)
• Some major characteristics
• True global markets
• Potentially different jurisdictions for each participant
• Regulation frameworks
• Different psychology for C2C operations.
5
Some new things
• The value of data
• Specialization of content and advertisement
• Contribution to value created within processes
• Large network effect
• ¿What is the actual value of data ? ¿When can it be determined?
• ¿Can they constitute some kind of primary resource?
• The value of content
• Free applications with in-app purchases
• 3D printers
• “Downloaded goods”
6
Digital money and
cryptocurrencies
• Intermediary payment systems. (Ej. Paypal)
• Money-like items within environments
(Games, Tokens within entertainment apps
Bliive)
• Cryptocurrencies
• No central point of control
• Limit of the number of total possible
bitcoins
• Anonymity is a primary goal
• Usually identified with illegal operations
7
A “small” evolutions
8
Internet of things
9
Challenges for the Tax Administrations
• The usual problems. Exacerbated.
• Companies created to “operate” fictitious transactions
• Fraudulent mechanisms to get refunds
• Difficulties with informal operations.
• International Taxation and BEPS issues
• Gray borders and dark sites
• Difficulties to determine geographic locations.
• VPNs, Dark Web, Cryptocurrencies
10
Options and alternatives
• More data, closer to transactions.
• Third – party reported information
• Electronic documents (Electronic invoice and withholding certificates)
• EFOS and EDOS in SAT Mexico
• Concurrent control in Brazil
• Transparent existent information
• Pre-filled returns
• Actions triggered by returns processing
• Electronic mailboxes
• Truly anonymous lead systems
• Specialized applications for auditors
11
Network analysis
12
Visualization
18116682928982363
75673531375849428
04118798115170382
43644131679600413
78843744243269849
64471229780824531
13
Visualization
18116682928982363756735313758494280411879
81151703824364413167960041378843744243269
84964471229780824531
14
Non-compliance prediction
• Linear and logistic regression
15
Non-compliance prediction (2)
• Machine learning and data mining
• Exploratory analysis
• Pattern recognition
• Decision-trees prediction
• Neural networks
• Clustering
16
Human resources
• New skills and knowledge
• Change management
• Deal with new and sophisticated
risks
• Digital economy.
• Great responsibilities
• For example, data from
electronic invoices.
17
A simple practical example
• Sent emails: 300 000. Opened 84%
• The time when the mail was sent matters? Hour, day of the week, month?
• Does a different sender has a different effect.
• Kind of domain? (specific vs. general)
• Length of text?
18
A simple but practical example (2)
19
A simple but practical example (3)
20
A simple practical example
• Sent emails: 300 000. Opened 84%
• The time when the mail was sent matters? Hour, day of the week, month?
• Does a different sender has a different effect.
• Kind of domain? (specific vs. general)
• Length of text?
• Develop a fitted model for each instance
• Maximize the function to reach each person at the best moment.
21
Thanks.
www.ciat.org
@rulizv
linkedIn: Raul Zambrano
22