Evaluation of travellers processing through a GBA+ lens:
2. Assessment of the travellers continuum through a GBA+ lens

2.1.1 Gender

The evaluation examined the impact of traveller processing based on gender. Overall, the evaluation results did not indicate that [*] travellers were disproportionately referred or searched; [*] travellers were more likely to be selectively referred for a customs examination in the air mode.

Due to data limitations and system integration challenges, the agency is currently not equipped to assess the impacts of these activities on smaller subpopulations.

A note on data disaggregated by gender: Across the travellers continuum, data is not consistently recorded to allow for a comprehensive GBA+ by travellers' gender through all stages of traveller processing. For example, the NTC does not actively record gender for the targets that are issued. Therefore, evaluators were unable to assess the impact of targeting on travellers based on gender.

Between FY 2014-2015 and FY 2020-2021^{Footnote 9}, [*] of scenarios used by the NTC to risk assess incoming travellers specified a gender category of male [*], female [*], or male and female [*]. The majority of scenarios that specified a gender category were related to national security concerns, with [*] of all national security targets issued being based on a scenario specifying a [*] gender category. According to NTC subject matter experts, this trend may have been the result of increased risk of illicit migration, as well as the threats posed by serious transnational organized crime groups, [*].

However, in response to a survey conducted as part of the evaluation, [*] NTC Targeting Officer respondents reported that a traveller's gender was "not at all important," as a risk indicator when deciding to issue a target for contraband. Additionally, [*] targets issued for contraband were based on scenarios that did not specify a gender category. Furthermore, NTC officers generally do not input the gender variable contained in certain scenarios into the ICES when issuing a target. The intent is to improve the likelihood that the target will result in a match. However, this process makes it difficult to assess the impacts of scenarios by travellers' gender. As certain GBA+ factors, such as gender, are not recorded in the NTC's target tracking data set (for Scenario Based and Flight List targets), the agency is unable to conduct a comprehensive GBA+ of the NTC's targeting activities.

In addition to targeting, once a traveller arrives into Canada, a frontline officer may refer a traveller for secondary examination. The overall referral rate (for all referral subjects, types, and sources) in the air mode was [*] to the gender distribution of incoming travellers. However, this trend was not consistent across all continents. For example, while there was a higher proportion of [*] travellers coming to Canada by air from [*], the referral rate was [*] from these continents (refer to Table 3).

When analyzing only selective referrals for customs examination, by referral source, [*] proportion of [*] travellers were also selectively referred for customs examination by PIL Officers. Customs referrals from other sources, such as Roving Officers, also presented [*] distribution of selective referrals of [*] travellers. For example, referrals by Roving Officers [*]. [*] travellers were also more likely to be the subject of a more intrusive search when compared to [*].

[*] of frontline evaluation survey respondents [*] indicated that gender identity or expression are not at all important as risk indicators when deciding to refer a traveller to secondary for customs-related concerns. However, interviews with program representatives revealed that [*] travellers are believed to be [*]. While decisions are based on a combination of experience, enforcement trends, training, and other sources of information, the trends presented here should be examined further, as the resultant rate of male and female travellers are [*].

[*] to the distribution of incoming passage volumes and the risk factors associated with [*] travellers being used by the NTC and BSOs to identify travellers for secondary examinations.

2.1.2 Socioeconomic status

The economic status of a traveller is not a data element that is tracked or recorded in CBSA databases. However, using the World Bank model for Income Groupings, the evaluation was able to broadly categorize travellers from countries of high, upper middle, lower middle, and low income groupings. For example, citizens from U.S and Canada would fall into the high income grouping. This enabled the comparative analysis of travellers from countries categorized by socioeconomic status. In general, travellers from [*] countries were [*].

However, these broad categories of socioeconomic status do not fully represent the characteristics of the travelling population of any given country. For example, a traveller coming from a country in the [*] socioeconomic grouping could still belong to a [*] income group. Therefore, without specific data on travellers' individual socioeconomic statuses, the analysis based on World Bank grouping is only indicative.

A note on data disaggregated by socioeconomic status (based on citizenship): The broad categories used to place travellers by citizenship into socioeconomic groupings do not fully represent the characteristics of the travelling population of any given country. For example, a traveller coming from a country in the [*] socioeconomic grouping could still belong to a [*] income group. Therefore, without specific data on travellers' individual socioeconomic statuses, the analysis based on World Bank grouping is only indicative.

Based on evaluation survey results, [*] of frontline respondents [*] reported that a traveller's perceived economic status is [*] as a risk indicator when deciding to refer a traveller for customs-related concerns. However, it is challenging to draw conclusions on whether respondents perceive the economic status of a traveller as a risk indicator. Based on open-text survey responses, the perceived economic status of a traveller may be considered in conjunction with other factors, [*].

When examining primary referral rates at the continent level, the evaluation results indicated that citizens of countries [*], which are in lower socioeconomic classifications, [*].

Further assessment of the countries within these continents reinforced these trends. When exploring the primary and secondary processing activities of frontline personnel at the border, evaluation results revealed that citizens of countries in [*] socioeconomic classifications [*], when compared to incoming passage volumes. [*]

For example, between FY 2018-2019 and FY 2020-2021,^{Footnote 12} citizens of [*] had a resultant rate of less than one percent. However, the referral rate of citizens of [*].

One of the possible explanations for the [*] referral rate might be a [*] level of concern over the risk posed by travelling [*] citizens (e.g. immigration related issues, such as visa requirements and [*] concerns). However, when compared to [*] citizens, which account for a relatively similar percentage of total passages and may present a similar level of border-related risk, citizens of [*] were [*] referred for secondary examination. It is unclear why citizens of [*] were [*] referred for secondary examinations and to such a degree.

When assessing NTC practices, [*] listed in combination with other factors when the NTC issues a target. Between FY 2014-2015 and FY 2020-2021, [*] of NTC scenarios specified [*] while [*] specified [*], and [*].

In comparing NTC Scenario Based Targeting (SBT) and Flight List Targeting (FLT) practices between FY 2015-2016 and FY 2019-2020,^{Footnote 13} both targeting activities demonstrated similar trends. Namely, citizens of certain countries are [*] targeted in comparison with incoming passage volumes.

[*] NTC targeting rates of citizens of specific countries appeared to be partly explained by the resultant rates generated by targets. [*]. As illustrated below, citizens of [*] (which fall into the [*] grouping) comprise [*] of the traveller volume in the air mode and they comprise [*]. This appeared to be [*] in comparison to high income countries such as Canada and the U.S., [*] which may be due to higher immigration and other border related risks.

Note: In [*], Canada lifted the visa requirement for all [*] citizens. This may have contributed to [*] which can be observed when examining trends in immigration enforcement actions year over year.

If a high resultant rate is an indicator of risk and partially explains a higher targeting rate, [*] population.^{Footnote 14} This difference in targeting rate between [*] and [*] citizens may warrant further analysis to determine if the variances can be explained by the level of risk and other policy considerations.

The evaluation also compared NTC targets and primary referrals, based on the distribution of targets and referrals by World Bank socioeconomic grouping. To do so, the evaluation isolated NTC targets issued for contraband (using Scenario Based Targeting and Flight List Targeting) and selective referrals for customs examination, issued by Roving Officers and Point Officers.^{Footnote 15} This was compared with the overall incoming traveller population, by socioeconomic grouping.

Reminder: The resultant rate is only one metric that can be used to examine level of risk among a multiplicity of risk indicators. The value for duty (VFD) and/or quantity of each seizure, for example, were not the focus of this evaluation. As with resultant rates, these are also imperfect measure of risk. Refer to Appendix E for more details.

Based on this analysis, the distribution of NTC targets issued for contraband more closely mirrored incoming passage volumes between FY 2017-2018 and FY 2019-2020, in comparison to referrals by Point and Roving Officers. In the referrals of citizens of [*] and [*] countries, Point Officer referrals were the most [*] when compared to incoming passage volumes. While these trends are affected by the unique reporting regulations, operating procedures, and volumes, they also indicate that, frontline travellers processing and frontline personnel should be the focus of future efforts to integrate GBA+ related training and intersectional analyses (such as GBA+), in the travellers stream.

This analysis demonstrates that while there is no specific direction to refer travellers solely, based on their income level, there is consistent evidence to suggest that, generally, travellers [*] were examined [*] at the border when compared to incoming passage volumes. The agency could do further analysis to examine how risk analysis and assessment practices contribute to this trend.

2.1.3 Race or ethnicity

Based on document and literature review, it is widely recognized that conducting research on discrimination based on race or ethnicity is difficult. Many national statistical agencies do not collect information on race or ethnicity for social and statistical/research reasons (e.g. France, Denmark, and Germany prohibit collection of data on race or ethnicity). Only Canada, the UK, the U.S., and Colombia have official data collection mandates and legal definitions alluding to race or ethnicity. Further, Canada's definition of visible minority groups, under the Employment Equity Act, has come under scrutiny by the United Nations.

According to the Ontario Human Rights Commission, "the process of social construction of race is called racialization: 'the process by which societies construct races as real, different and unequal in ways that matter to economic, political and social life.' Recognizing that race is a social construct, the Commission describes people as 'racialized person' or 'racialized group' instead of the more outdated and inaccurate terms 'racial minority,' 'visible minority,' 'person of colour,' or 'non-White'."^{Footnote 16}

Literature that sought evidence of profiling practices in law enforcement was often based on surveys of the general population. These surveys asked respondents if they felt they had been the victims of profiling. While some studies used proxy measures, others explained why proxies (e.g. citizenship or country of birth) are problematic for determining ethnicity and race, as they do not account for the diversity of a country's population (e.g. as a result of global trends in migration). For this reason, the use of this methodology was limited and citizenship was only viewed as a general indication of a traveller's race, ethnicity, or ethnic/cultural origin. Furthermore, it is important to note that the racial/ethnic/cultural composition of a country, based on its reported census data, may not always be a direct indication of the composition of those travelling to Canada in the air mode.

The CBSA's traveller processing activities, such as NTC targeting, do not intentionally set out to target travellers based on perceptions around their race or ethnicity. In its Scenario Based Targeting, for example, the NTC uses a combination of information sources, such as global trends and reports (e.g. World Customs Organization drug trend reports), in the development of scenarios, which are systematically reviewed for Human Rights and other considerations.

However, certain practices can have unintended consequences that result in the overrepresentation of racialized communities in the law enforcement context. For example, when targeting rates are higher for certain origin countries (largely representative of a traveller's citizenship), there could be unintended consequences for travellers of certain racial/ethnic groups when those groups make up a larger proportion of incoming travellers from those countries.

The examples used in this section were selected based on citizenship groups referenced and explored in other areas of the evaluation. A United Nations (UN) world census database^{Footnote 17} that contains aggregated datasets of national, racial and/or ethnic groups in each country was also used to support the analysis. For example, according to the UN world census data, citizens of [*] may be considered as belonging to a racialized group.

When keeping target type and departure country constant, contraband targets^{Footnote 18} issued on flights departing [*], by citizenship, were proportional to incoming passage volumes. This indicates that citizenship was not a defining factor in issuing contraband targets for travellers departing [*].

The table below illustrates that citizens of [*], when targeted using SBT, were targeted proportionally when compared to incoming passage volumes. In contrast, citizens of the [*] on incoming flights from [*] were not issued targets proportionally when compared to their incoming passage volumes. However, this analysis does not account for the cultural diversity of [*] travellers on those flights, as there is no further demographic information available for these travellers.

Similar to trends in SBT data, FLT contraband targets issued for travellers on flights departing [*] were generally proportionate to incoming passage volumes, when examining targets by travellers' citizenship (refer to Table 8). Similarly, this does not account for diverse sub-populations of travellers within and across these citizenship groups.

Based on this preliminary analysis, on flights departing from, [*] there is no indication that citizens of [*], who may belong to a racialized group, were disproportionally targeted by the CBSA's NTC for contraband.

With consideration for the associated limitations, the above methodology could be replicated for other citizenship groups, which may comprise larger proportions of persons belonging to racialized groups, to indicate if there are potential issues requiring further assessment.

It is also important to note that the CBSA currently collects data on travellers' perceived race for searches and arrests. This information is collected to identify individuals without access to and the use of biometric data. It is not collected for analytical or risk assessment purposes. In ICES, a description of a traveller's perceived race may be entered following a search, using a drop-down list. Further, Enterprise Information Data Architecture has not included racial descriptions as an attribute within the agency Collaborative Platform (ACP), explaining that the lack of defined business requirements for capturing and using information on race could not be satisfied by other data attributes. Relevant branches within the CBSA are examining options to address the concerns raised over data accuracy, privacy, and consent.

Most frontline survey respondents were satisfied (to a very large extent or to a large extent) with the CBSA's efforts to prevent discrimination and eliminate barriers encountered by diverse groups of incoming travellers.^{Footnote 21} However, in examining complaints information collected by the agency's Recourse Directorate between June 2016 and June 2018, 11% (n=71) of complaints (633) in the air mode were related to reports of unfair or disrespectful treatment based on travellers' race, ethnicity, or ethnic/national origin. Additionally, a quarter (n=227) of frontline respondents (n=922) indicated that they had directly witnessed a colleague discriminate against a traveller in the past two years. Of these respondents, 71% (n=162) suggested the discrimination they witnessed was based, in full or in part, on the travellers' race and 76% (n=173) on the travellers' national or ethnic origin(s).

2.1.4 Intersectional identities

Intersectionality refers to "the interconnected nature of social categorizations and identity factors…" as they apply to a given individual or group, which are "regarded as creating overlapping and interdependent systems of discrimination" or inequity.^{Footnote 22} An intersectional lens is important for assessing the potential impacts of the agency's policies and practices on diverse subpopulations of travellers. Examining the CBSA's operational data through travellers' intersecting identity factors adds further complexity to GBA+. The following section examines the combination of citizenship, socioeconomic status, and gender identities within the context of travellers processing.

The evaluation results indicated that [*] travellers, particularly citizens of [*] income or [*] income countries, were [*] referred when compared to incoming passage volumes and [*] travellers of the same citizenship groups. [*] travellers from certain [*] countries were [*] to be referred when compared to [*] travellers from [*] countries. The Agency may need to further assess whether this degree of difference is considered within an "acceptable" range.

The availability of traveller demographic information across the travellers continuum limited the extent and depth of intersectional analysis that could be completed. For example, the agency does not collect information on diverse gender identities. Operational data on the gender identities of travellers is categorized in four ways: male, female, unspecified, and unknown.^{Footnote 24} A traveller with a gender neutral passport may be categorized as "gender unspecified," but this category may not be unique to persons with gender neutral travel documentation. Nonetheless, it is recognized that border processing may be experienced differently, and may disadvantage those with diverse gender identities.

As part of its modernization efforts, the agency has created a new resource responsible for developing an agency-wide data analytics strategy. This includes foundational pieces aligned with all stakeholders and partners, as well as enhancing access to information for Canadians while protecting the personal information of Canadians. As this function matures, the CBSA anticipates being able to provide more consolidated data reports in the future.

In summary, GBA+ can require significant amounts of demographic data or an analysis that goes beyond the use of operational and quantitative data, particularly when attempting to assess the potential disproportionate impacts of programs, policies, and practices on smaller subpopulations of travellers.