(AGENPARL) - Roma, 28 Gennaio 2026 - (AGENPARL) – Wed 28 January 2026 Statistics
Methods and Sources:
Special Topics
www.bancaditalia.it/statistiche/index.html
28 January 2026
Mobile phone data and the “Travel” item in the
Balance of Payments: the Bank of Italy’s experience
by Andrea Carboni, Claudio Doria and Valeria Valletta1
Key Messages
− Since the end of 2020, the Bank of Italy has used mobile phone data (MPD), alongside
administrative sources and manual counting operations, to estimate the number of
international travellers.
− Compared to traditional techniques, MPD offers continuous temporal coverage, broader
geographical reach, greater timeliness, and reduced costs relative to manual counting
operations.
− Effective integration of MPD into tourism statistics requires ongoing collaboration with
mobile network operators to ensure the adoption of procedures and methodologies that
filter raw data so as to align with official statistical definitions.
− Comparative analyses with administrative data from border authorities, considered
highly reliable, suggest that MPD provides high-quality estimates of international tourist
flows.
Abstract
Despite the potential of Mobile Phone Data (MPD) as a resource for international tourism
statistics, few countries use them in official statistics due to the difficulty in translating tourism
definitions into telephony terminology. Since 2020, the Bank of Italy has integrated MPD into
the estimation process of the “Travel” item of the Balance of Payments, resulting in more
comprehensive information on border crossings and lower costs compared to traditional
survey methodologies. This paper describes the advantages and challenges related to the
use of MPD and their performance through a comparison with administrative data sources.
1 External statistics division – Statistical analysis directorate.
1. Introduction
Technological advancements, the spread of the internet and the proliferation of digital products
and services have generated new data flows with heterogeneous characteristics. These data – which
involve large volumes of information, a high update frequency, and are complex to use – are
commonly referred to as “big data”.
One type of big data is mobile phone data (MPD, also called mobile positioning data), which
provide valuable information about individuals’ location and movements, representing a potentially
significant resource for the study of social mobility and, more specifically, of tourism.
Tourism is a complex phenomenon, with a significant impact on social, environmental, and
economic dimensions, being an important share of the global gross domestic product (Leng et al.,
2016). Analysing and understanding tourism is crucial for effectively managing its impact on
destinations and communities, and it enables informed decision-making regarding: a) the urban
space design (Xu et al., 2023), i.e., the sustainable development of transportation and infrastructure;
b) the enhancement and preservation of natural and cultural heritage; c) the promotion of local
culture and events, creating job opportunities and encouraging the development of tourism services
to support the local economy (Saluveer et al., 2020).
Until a few years ago, tourism statistics mainly relied on expensive and demanding surveys with
large sampling errors. More recently, research in this field has begun to leverage MPD, recognizing
the advantages of this data source in estimating traveller flows, including lower costs, timeliness,
continuous data collection and broad geographic coverage. In general, MPD enable full coverage of
the surveyed territory, and they reach a large share of the population due to the widespread use of
mobile phones; however, their reach is strictly dependent on the market share of the Mobile Network
Operator (MNO) providing the data and the extension of its network. Despite growing international
attention (Demunter, 2017; UN, 2019; UN-ESCAP, 2022; Eurostat, 2014, 2023), leveraging MPD for
official tourism statistics by individual countries remains very limited. On the one hand, the reason
for this limited use can be attributed to the strict privacy regulations that MNOs must adhere to in the
vast majority of countries; on the other hand, translating the definitions spelled out by international
statistical standards for official tourism statistics into telephony terminology is objectively complex
(Grassini & Dugheri, 2021; UN, 2019).
Against this background, the Bank of Italy, in the constant search for new sources and
methodologies to improve the statistics it publishes, has begun to integrate MPD into the estimation
process of the “Travel” item of the Balance of Payments (BoP) following an experimentation phase
started in 2018.
This study describes this experience, highlighting the main characteristics of the data used and
how they are being used.
Section 2 reviews the existing literature on mobile phone data, with a particular focus on studies
related to the estimation of international travel. Section 3 describes the survey used by the Bank of
Italy since 1996 for compiling the “Travel” item of the BoP. Section 4 describes how MPD have been
gradually integrated into the existing estimation process and the expected benefits. Section 5
outlines the current use of MPD, emphasizing their key characteristics, primary benefits and
significant challenges. In Section 6 some quality issues of MPD are assessed through a comparison
with survey and administrative data sources. Finally, in Section 7 sets out the conclusions.
2. Literature overview
2.1. Definition of MPD
MNOs collect a large amount of data in their systems for billing, monitoring, management, and
network planning purposes.
In the literature, two types of MPD are primarily used: active and passive mobile positioning data
(Ahas et al., 2008). In the former case, data are generated through a dedicated app, installed on the
phone; as they allow real-time tracking of the phone’s location, a user permission is required for
sending the data to the MNO. They are rarely used due to their limited volume.
Passive mobile positioning data, on the other hand, are collected automatically by MNOs,
without the need for specific user consent and are the most commonly used in tourism research
(Saluveer et al., 2020). They are further divided into event-driven data (such as Call Detail Records,
CDRs) and network-driven data (such as signalling or probing data). The former are recorded by the
MNO with detailed temporal and positional information whenever the mobile device user performs a
trigger event, such as making a call or sending an SMS or MMS. The latter are collected periodically
without a trigger event occurring.
Among the data collected by MNOs, CDRs have been the most commonly used in research,
thanks to their greater availability, relatively low cost, and high scalability (Chen et al., 2018; Grassini
& Dugheri, 2021; Rahmadian et al., 2023). However, the time frequency of CDRs is uncertain,
generally low, and further decreasing, considering that the use of instant messaging apps continues
to reduce the number of SMS, MMS and phone calls. Additionally, the time frequency of CDRs is
closely dependent on the users’ mobile device usage behaviour and, consequently, on their
economic and sociodemographic characteristics and habits (Saluveer et al., 2020). Nonetheless,
they are easily accessible and sometimes sufficient for research purposes.
Network-driven data are generally more efficient, detecting a higher number of SIM cards
compared to CDRs, with a much higher time frequency, typically in the order of seconds or minutes;
consequently, they produce a large amount of data that require specific technologies for their
management (Grassini & Dugheri, 2021). This kind of data may be affected by statistical noise due
to the presence of certain types of international travellers, such as those flying over the country,
crossing seas, or unintentionally connecting to the network while being near territorial borders. This
noise can generally be corrected, with good results, through appropriate techniques and algorithms
(Lestari et al. 2018).
2.2. The use of MPD in research on tourism
Multiple studies in the literature show the significant role of MPD in various fields, including
healthcare, socioeconomics, disaster response, urban management and more (UN, 2019).
In the field of travel, several analyses acknowledge the potential of such data in studying both
domestic and international tourism. Ahas and Mark (2005) were among the first to recognize both
the potential of MPD and the associated issues, primarily related to privacy. They introduced the
Social Positioning Method which investigates social flows over time and space through phone
localization and the social characteristics of owners.
Subsequently, interest in MPD grew along with the number of studies using them. In particular,
there is considerable interest in studying travel behaviours and identifying travel patterns
(Phithakkitnukoon et al., 2015; Chen et al., 2018; Xu et al., 2021; Park et al., 2020), assessing the
impact of various factors on tourism such as climate change (Kubo et al., 2020), and defining new
indicators (Leng et al., 2016).
In Ahas et al. (2007), different travel patterns are identified based on the season, as well as
differences in the predominant nationalities between summer and winter tourism areas.
In Ahas et al. (2008) a detailed description of MPD is provided and the main points of attention
are described, such as access issues due to privacy regulations, data collection methods, spatial
resolution, network density, statistical noise due to traffic near borders, and various
representativeness issues, such as different mobile phone usage among different population
subgroups (especially applicable to CDRs).
Raun et al. (2016) develop a methodology for measuring visitor flows to specific destinations
using space-time tracking data. The paper proposes that a tourism destination has five dimensions
– spatial, temporal, compositional, social and dynamic. Results show that smaller destination areas
within a country can be differentiated by the geographical, temporal and compositional parameters
of the visits.
The issue of MPD quality, which can normally be verified by comparison with more reliable
traditional sources, is of central importance. In several works, a comparison of MPD on incoming
flows with official statistics on accommodation is carried out (Ahas et al. 2007, 2008; Heerschap et
al., 2014; Saluveer et al., 2020). MPD show a high correlation and superiority for areas with few
structures and transit flows. For outgoing flows from Estonia Saluveer et al., 2020 report a low
availability of data from other countries.
MPD also work efficiently in analysing domestic tourism. Indeed, Indonesia’s national statistical
institute (Badan Pusat Statistik – BPS) has leveraged MPD since 2018 for estimating domestic
tourism and commuting.
The interest in MPD is part of the ongoing initiatives at BPS to develop big data governance
(Rahmadian, 2023). Noviyanti et al. (2020) propose a framework to support the use of big data in
the production of official statistics.
Finally, MPD enable the construction of origin-destination matrices with advantages relative to
those derived from conventional data sources. (Caceres et al., 2020; Bonnel et al., 2015).
2.3. MPD for official tourism statistics
The use of MPD for compiling official statistics has begun only recently. Notable examples
include contributions from France (Cousin & Hillarieu, 2020) and Netherlands (Heerschap et al.,
2014).
Outside Italy, MPD have been officially integrated into the estimation of international tourism
flows only in Estonia and Indonesia.
In particular, the Estonian Central Bank, Eesti Pank, has been using CDRs in combination with
payment data for the compilation of the “Travel” item of the BoP since 2008 (Eesti Pank, 2019). The
methodology used by Estonia is described and adopted by Saluveer et al. (2020), who, for the first
time, use MPD to derive statistics on Estonian inbound and outbound international tourism flows
(such as the number of travellers and nights spent). The authors emphasize the importance of
developing and harmonizing regulations and laws for data access.
BPS-Statistics has been using MPD (signaling data) for compiling Indonesian official statistics
on the number of arrivals of international travellers since 2016. Previously, administrative data on
immigration and a border survey conducted one month a year were used. However, both sources
faced significant issues of under-coverage, due to the vastness of the Indonesian territory, the variety
of border points, and the scarcity of control points. MPD have allowed to achieve better coverage
and timeliness compared to traditional sources (Lestari, 2018).
On a more general note, a limitation common to the different types of MPD is that they do not
contain qualitative information, such as travel characteristics (i.e., main reason of the trip, type of
accommodation, etc.) and, to some extent, traveller attributes. Since some of these characteristics
are mandatory details prescribed by the international guidelines for the BoP compilation,2 MPD need
to be integrated with data from traditional surveys in order to obtain more comprehensive tourism
statistics. Most importantly, it should be remarked that, while telephone data can provide information
on the size of tourism flows, they cannot provide any information on the expenditure associated with
these trips, which is the fundamental variable for BoP statistics.
3. The Bank of Italy’s Survey on International Tourism
The Survey on International Tourism has been conducted by the Bank of Italy since 1996, with
the main purpose of collecting information useful for the compilation of the “Travel” and “Transport passengers” items of the BoP. Simultaneously, it also collects additional information on tourism
flows, of interest for researchers, industry operators, and local authorities3.
In the survey a stratified sampling plan is used, which takes into account the following variables:
direction (inward vs outward), mode of transport, border crossing, type of day (weekday/holiday),
time slot, day of the week.
2 As outlined in the Balance of Payments and International Investment Position Manual Sixth Edition (BPM6) of the IMF.
3 The survey does not collect information on domestic tourism.
The survey takes place at international border points (ports, airports, railway and road crossings)
and relies on two main operations: interviews and counts. A subset of counts are “qualified” because
they also collect information on the country of residence of the travellers.4 Until 2019, about one
hundred thousand interviews and one million counting operations were carried out annually at more
than 60 border points.
The purpose of the qualified counts is to estimate the size of the reference population, namely
the total number of incoming non-resident international travellers and outgoing resident travellers,
by country of residence or destination, respectively. Counts are supplemented, where available, with
administrative data (e.g. information provided by the airport authorities about the number of travellers
entering or leaving Italy) provided by the authorities responsible for border points. These data provide
the “exact” population size of the reference universe but do not provide any information on the
residence of travellers.
The interviews, conducted using the Computer Assisted Personal Interview (CAPI) method,
serve to collect data on expenditure and other relevant aspects for compiling BoP items, such as the
counterparty country and the purpose of the trip. Secondly, they gather additional information useful
for a broader understanding of the tourism phenomenon. The interview takes place at the end of the
journey (when leaving Italy for non-residents and when entering Italy for residents), minimizing the
traveller’s difficulty in recalling expenditure and trip characteristics, as well as ensuring that the
expenditure is not hypothetical or planned but already incurred by the respondent. Each interview is
administered to a single individual per group of travellers (i.e., those who share travel expenses,
such as a family), who also responds on behalf of the other members.
The information acquired with the questionnaires is then grossed up to the reference population
as determined from the counting operations, administrative data sources and, as of 2020, MPD.
Indeed, information gathered through MPD since 2020 (see Sections 4 and 5 for the details) has
only complemented counting operations, reducing them by about two-thirds, while the number of
interviews – and the information they aim to collect – has remained unchanged.
4. The path toward the use of Mobile Phone Data
The integration, of MPD in the process of estimating the “travel” item took place in several
stages, starting with the general idea that they could be useful since a SIM card issued in Italy
disappearing from the network near a border may indicate a resident travelling abroad and
conversely a SIM card issued by a foreign operator appearing near a border may indicate a nonresident entering Italy.
With this idea in mind, a pilot study was conducted in 2018 (Carboni et al., 2023) that saw the
collaboration of the Bank of Italy with one of the main Italian telephone operators. Subsequently,
during the period when the COVID-19 emergency placed Italy in an almost complete lockdown, the
absence of traditional information sources required the use of alternative ones, including MPD,
although limited to non-residents detected in Italy (Carboni et al. 2020).
Finally, MPD were officially integrated into the Bank of Italy’s tourism statistics at the end of
2020.
4.1. The pilot study and the COVID-19 emergency
The pilot study was initiated in 2018, in order to evaluate the feasibility of permanently integrating
MPD into the estimation process for the “travel” item. This stage involved the collaboration between
the Bank of Italy and Vodafone, one of the leading Italian mobile network operators.
The pilot study focused on three border points, two airports (Roma Fiumicino and Milano
Malpensa) and a highway frontier point (Tarvisio), and concerned flow data, i.e., the number of
travellers leaving Italy (non-residents) or returning (residents) through these border points. MPD
were compared with time series derived from surveys and, specifically for the two airport
checkpoints, with administrative data.
4 The information on the destination is instead gathered through the interviews.
The results highlighted the informative potential of MPD and their ability to improve the
estimation of population size for border points where administrative data are not available. Moreover,
for all types of border points, MPD were able to improve the classification by counterpart country of
the reference population. Additionally, the data demonstrated high timeliness and the ability to
reduce costs by minimizing manual counts, which are relatively expensive.
On the other hand, the challenge to translate the definitions of official tourism statistics into
telephony terminology became evident, as well as the consequent need for constant interaction
between statisticians and MPD providers and the need to develop specific algorithms depending on
the type of border point.
During the period from March to June 2020, the survey was suspended, in compliance with
government measures aimed at containing and managing the COVID-19 health emergency and due
to the practical impossibility of conducting counts and interviews at borders. International tourist flows
for this period were therefore estimated based on the application of y-on-y variations 2020/2019
derived from alternative data sources, such as MPD for foreigners in Italy and payment card data for
Italians abroad, to the survey data referred to the corresponding period of 2019.
The number of foreign travellers in Italy was estimated using mobile phone data provided by the
same operator involved in the pilot study. The need to quickly obtain data led to the use of stock
data on the number of non-residents present in the national territory at a given point in time, rather
than flow data at border crossings. Indeed, flow statistics require tracking the behaviour of a SIM
card over time, whereas stock statistics are based on data collected at a specific point in time.
Specifically, information on the number of foreign SIM cards detected in the Italian territory at least
once a month was used, broken down by nationality of the issuing operator, as a proxy for the
traveller’s residence.
The quality of the estimates thus obtained was subsequently verified through comparisons with
other available information sources, such as administrative data, other official statistics, and mirror
data.5 These comparisons confirmed the reliability of the method used.
4.2. Adoption of MPD and expected benefits
The experience gained with MPD during the experimentation phase and the subsequent
emergency phase led to the final integration of this source into official tourism estimation procedures
as of the end of 2020. The integration occurred gradually: at first, MPD were only used as a reference
benchmark, then as a substitute for a part of the counting operations.
The main benefits expected from the adoption of MPDs can be summarized in the following
points:
− Airport crossing points: all international airports have administrative data and,
consequently, the size of the reference universe can be considered as known.
However, as stated above, these administrative data do not provide any information
on the residence of travellers, but only about the number of travellers who travelled
on international flights.
− The use of MPD should lead to a better qualification of the universe, especially
for large airports characterized by the presence of numerous gates, some of which
may be very distant from each other, making monitoring through counting operations
very complex.
− Road passes: In Italy there are dozens of road passes connecting it with neighbouring
countries often located in mountainous areas, far from major population centres and,
therefore, difficult to reach. Inevitably, unless a significant increase in the cost of the
survey is accepted, counting operations must be limited to a sample of crossings,
those that are more relevant in terms of traffic flows. Moreover, in terms of cost
sustainability, counting operations are conducted only for a certain number of days
5 “Mirror data” refer to the estimation of the same item provided by a different BoP compiler. Indeed, travel expenditures
from country A to country B are recorded by both economies: as debits for A and as credits for B. Mirror data play a crucial
role as they allow verifying, validating, and improving the reliability, accuracy and consistency of international data.
each month and for a daily time span usually ranging from 6 to 8 hours. Finally, it
should be noted that counting operations on roads, in the absence of checkpoints,
involve counting vehicles and the number of occupants, using the license plate as a
proxy for residency status. In some situations, counting operations may be
challenging as, due to safety concerns, they cannot take place near the road where
vehicles are passing. Additionally, information about road checkpoints is largely
missing from administrative data sources.
On the contrary, MPD can potentially cover all road checkpoints and monitor them
continuously. Therefore, an advantage is expected in terms of a estimation of both
the universe size and the traveller’s residency distribution.
Railroad crossing points: counting operations are conducted directly on sampled
trains, for which administrative data are available, and, as a rule, they manage to
cover all passengers on the train. Therefore, in this case the expected advantage of
MPD lies mainly in the fact that telephone data enable to monitor all trains and not
just a sample of them.
Port crossing points: counting operations are conducted in boarding and
disembarkation areas, which are subject to access restrictions for security reasons.
Additionally, it is often prohibited to approach docking areas during disembarkation
phases, as they must occur as quickly as possible.
MPD can monitor ports without the need to access restricted areas and with equal
effectiveness during both the boarding and disembarkation phases. Considering that
only a minority of ports provide administrative data with timing useful for BoP use,
MPD should lead to an improvement in both the estimation and qualification of the
universe.
5. The Use of Mobile Phone Data by the Bank of Italy for Compiling the
“Travel” Item of the Balance of Payments
5.1. The MPD used and the main characteristics of the MNO
MPD provided by Vodafone, the same MNO that supplied data during the experimental phase
and the pandemic period, have been integrated into the estimation process since the end of 2020.
In September 2023, the Italian market share of total SIM cards issued by this MNO was 27.5%
(Figure 1). At that time, the company had more than 200,000 cell towers and covered more than
99% of the Italian territory. Since the operator extrapolates the data about SIMs detected on its
network to the target population (total SIMs detected on Italian mobile networks) based on its market
share, this factor is crucial to ensure sufficient representativeness and adequate expansion of the
findings to the entire universe. In the case of non-residents, the proportion of users connecting to
the specific MNO is determined by considering the likelihood of their devices connecting to the
network of the mobile carrier in the specific cell near the border. This probability also depends on
the presence of competing operators’ transmitters in the area, as well as on existing commercial
agreements between mobile carriers across different countries.
The information provided by the MNO leverages on the data traffic that each SIM, limited to the
human-type6 ones, continuously maintains with the operator’s network. Billions of positions are
collected daily, with a frequency of one minute and a spatial resolution ranging from a few hundred
meters in urban areas to several kilometres in rural areas.
The data are compliant with the Italian privacy law. With specific reference to the data
transmitted to the Bank of Italy, they are irreversibly anonymized and aggregated into subsets that
cannot be formed by less than 15 SIM cards. If this is not the case, the information is not provided.
6 Human-type SIM cards are associated with physical persons, while machine-to-machine (M2M) SIM cards are used for
communication between devices.
Figure 1 – Total SIMs: market shares of the main Italian MNO.
Source: the Italian Regulatory Authority for Communications (AGCOM).
5.2. The structure of MPD: a comparison with traditional counting operations
The mobile phone data were requested from the MNO in such a way as to approximate the
structure of the information obtained from counting operations, which MPD are meant to
complement.
Data derived from counting operations differ for each type of border crossing (road, rail, etc.);
this is because different information is collected depending on the mode of transport used, such as
the type of vehicle boarded on a ship, the class of a train carriage, the type of flight (scheduled or
charter). All this detailed information cannot be captured through mobile telephony. The use of MPD
thus results in an objective loss of secondary information, which is counterbalanced, as will be
discussed later, by a greater reliability of the most relevant information.
Other advantages of MPD relate to the much higher frequency at which data are collected, as
well as to the wider spatial coverage. Indeed, counting operations are exclusively conducted at the
border points sampled in the survey, based on a monthly calendar which includes a certain number
of counting shifts at each crossing point for a certain number of hours. In this respect, MPD are much
more efficient, as they can continuously cover not only each sampled crossing point, but potentially
all of them (although smaller crossings are typically grouped in clusters based on type and
geographic location). As a second example, regarding the determination of the traveller’s residency,
counting operations only sample these data on a subset of counted individuals.7 In contrast, MPD
considers the nationality of the SIM card issuer for every SIMs detected on the network.
5.3. The impact of MPD on counting operations
Mobile phone data have been integrated into the estimation process, replacing part of the
counting operations conducted at border crossings. Consequently, the counting operations have
decreased from 1,200,000 to about 430,000 per year (Figure 2), thus allowing for a significant
reduction in the costs associated with the survey.
It is expected that, with the continuous improvement of methodologies for the exploitation of
MPD, the need for counting operations will further decrease over time. However, a certain number
of counts will always be necessary, mainly as a control tool of mobile data, especially for frontier
points without administrative data sources.
7 For example, at a road crossing with a count step of 5, the nationality of the vehicle’s license plate will be recorded
once for every five vehicles that pass through. The counting step varies, based on traveller flow intensity and the specific
crossing point characteristics.
Counting operations are also important for gathering evidence on specific secondary
phenomena, which are difficult or impossible to capture with MPD, such as travellers from smaller
countries without a resident MNO (San Marino, Monaco, Andorra, etc.); or to gather additional
information not detectable through MPD, such as the type of vehicle used at road border crossings.
Figure 2 – Sources for compiling the “Travel” item of the BoP,
before and after the introduction of MPD in the estimation process.
5.4. How to make the most of mobile phone data
In order to use telephony data for tourism statistics, it is necessary for the MNO to isolate the
SIMs that contribute to aggregates that correspond as closely as possible to the definitions of official
statistics.
Listed below are the main points that were worked on, in collaboration with the telephone
company, in order to limit the possibility of incorrect selection of relevant SIMs.
5.5. Identification of the destination country for Italians and of the origin country
for foreigners
In BoP statistics identifying the counterpart country is fundamental. In tourism statistics, this
means determining the foreign country (or countries) which Italian residents visited, and the country
of residence of non-resident travellers who visited Italy.
To determine the destination country of an Italian resident, the “Welcome message” received by
each SIM when connecting to a foreign operator’s network is used. If multiple countries are visited,
the number of welcome messages corresponds to the number of countries visited. The date of
receipt of these messages could also be used, in principle, to identify the period of stay in each of
the visited countries. The quality of this information is still being explored.
For the country of origin of non-resident inbound tourists, the residence of the SIM card issuer
is used as a proxy. This approximation is of good quality in most cases. However, certain specific
phenomena can be systematically distorting. A typical case is that of individuals originally from other
countries who are resident8 in Italy and use a SIM card issued in their home country. In Italy, there
are numerous and large communities of foreign residents, so this phenomenon cannot be
overlooked. To correct this type of false positive, an analysis of the SIM card’s behaviour over several
months is used: if a SIM card issued abroad is detected on the Italian network for more than six out
of the twelve months preceding the one under examination, then the SIM card is deemed to be
associated with a resident in Italy.
Regarding the use by foreign residents of SIMs issued in Italy, no specific filters have been
introduced at this stage.
5.6. Estimating Overnight and Same-Day Trips
The BoP records the monetary flows between two economies, therefore in travel estimation it is
crucial to distinguish between overnight and same-day travellers, as their spending profiles are
dramatically different. For instance, in 2023, same-day travellers represented around one third of
non-resident travellers visiting Italy but only 4% of the total expenditure. With appropriate measures,
MPDs can identify the two categories of travellers.
The general criterion adopted to associate a SIM issued by a non-resident MNO with an overnight
trip to Italy is that at least 24 hours must elapse between the time of entry into the Italian territory and
the time of exit. However, there are cases where a traveller can stay in Italy for less than 24 hours but
still spend a night in the country. We detect these situations by requiring that the traveller spend at
least 6 hours during the night in a single location in Italy (“6-hour-detection criterion”).
For SIMs issued by resident MNOs, a loss of signal for at least 24 hours is used to identify
overnight travellers abroad, while the 6-hour-detection criterion is not applied. This is because, unlike
non-resident SIMs, for which a journey is identified based on connection to the network, for resident
SIMs, the journey is defined by disconnection from the network. The methodology is continuously
monitored, and further adjustments may be considered in the future.
5.7. Determining the Province Visited or of Residence
Determining the province visited (for inflows) or of residence (for outflows) is not relevant for the
compilation of BoP statistics, but it is of considerable interest for analysing tourism-related phenomena.
To determine the province of residence associated with a (resident) SIM user, the following
procedure is followed: for each day, the municipality where the user spent the most time between 8
p.m. of the previous day and 8 a.m. is defined as the “prevalent night-time location”. The “residence”
is then defined as the municipality that most frequently in the previous twelve months (or in the
number of months available in the case of newly activated SIMs) has been the “prevalent night-time
location”. Although, at the level of MPD, information is collected at a more detailed territorial level
than the province, data are then aggregated to this last dimension for two reasons: a) to contain the
otherwise high number of masked records which the MNO cannot provide due to confidentiality
reasons;9 b) the impossibility, on the part of the MNO’s network, to reconstruct the administrative
subdivision into municipalities with precision, particularly for smaller municipalities.
As for determining the province visited by non-resident travellers, the “prevalent province” is
defined as the one where the travellers spent the most time during their trip to Italy. Due to
confidentiality reasons, we do not pursue the full list of visited provinces, since this would
considerably increase the number of “masked” data that do not meet the requirements of the Italian