GBA+ [Gender-Based Analysis plus] is an analytical tool used to assess the potential impacts of policies, programs, services, and other initiatives on diverse groups of women and men, taking into account gender and other identity factors. The "plus" in the name highlights that GBA+ goes beyond gender, and includes the examination of a range of other intersecting identity factors (such as age, education, language, geography, culture and income). [Status of Women Canada]
Over 20 years ago, the Government of Canada committed to implementing Gender-Based Analysis, or GBA, in policies, programs, and legislation. Status of Women Canada was tasked with creating and championing the framework for GBA, although all federal departments and agencies had a shared responsibility for implementation.
I believe understanding how policies may effect men and women differently is a worthwhile goal. I believe achieving gender equity is a worthwhile goal. But I have serious reservations about an overly strong reliance on the current GBA framework - GBA+ - to achieve that goal. The GBA+ framework developed by Status of Women Canada is not readily integrated with prevailing methods of policy analysis. Doing GBA+ properly often requires data that is often either unavailable, or else hard to obtain in a timely fashion. The lofty ambitions embedded in the current GBA+ analysis framework can make doing GBA+ seem overwhelming or impractical. Moreover, the very concept of Gender-Based Analysis locates gender considerations in the analysis phase of policy making. However effective integration of gender in policy formation requires a holistic approach, with gender being considered at every stage and level of the policy process, from data collection to setting of overall government priorities.
GBA and Cost-Benefit Analysis Compared
GBA+ fits somewhat awkwardly into an economist's toolkit. The philosophical underpinning cost-benefit analysis, as set out in, for example, Jean Drèze and Nick Stern's classic (1987) article, is enlightened utilitarianism. The objective of the policy maker is to maximize social welfare, V, which depends upon all people in society's "utility" (or satisfaction or happiness) Ui. Simplifying the Drèze and Stern framework slightly, each person's utility depends upon their consumption, xi, broadly defined to include market and non-market goods, as well as leisure. In sum:
V= V(U1(x1),U2(x2),...,Un(xn))
The process of cost-benefit evaluation involves estimating the change in social welfare associated with a given project - how much better off a project makes society as a whole. Formally, the costs and benefits of a project, dz, can be found by figuring out how much better off or worse off that project makes people:
dV=∑(δV/δUi)(δUi/δxi)(δxi/δz)dz
A project is worth doing if dV>0, that is, if it improves social welfare. If many different projects are possible, the best project is the one that leads to the greatest improvement in social welfare.
It could be argued that GBA+ is nothing more than a reminder to do cost-benefit analysis properly. Any given government policy will affect different people differently. Take, for example, a policy such as implementing a Universal Child Care Benefit (UCCB), funded by increased levels of income taxation. Depending upon the details of how the policy is implemented (is the payment made to mothers? fathers? is it taxable? how does it interact with other benefit programs?) the UCCB will have different effects on men, women, parents, non-parents, high income people, low income people, and so on. Any good cost-benefit analysis, by examining the impacts of a policy on individuals, must inevitably take into account a policy's impact on "diverse groups of men and women" - because those diverse groups are made up of individuals.
GBA+ is also about ensuring "that measures are taken to meet the needs of a group with a particular disadvantage, including a historical disadvantage." Yet cost-benefit analysis can take disadvantage into account. Recall that every cost benefit analysis starts - either explicitly or implicitly - with a social welfare function. That social welfare function can be set up to value gains to the disadvantaged more than gains to the advantaged. For example, a social welfare function of the form:
V=logU1+logU2+...+logUn
explicitly values gains to people with low levels of utility more than gains to people with high utility. A well-done cost-benefit analysis, that incorporates distributional factors, can take into account disadvantage.
The Auditor General's evaluation of federal government departments' and agencies' compliance with GBA+ fits with the idea that GBA+ is simply cost-benefit analysis done right. From the Auditor General's point of view, a GBA is complete if it passes three tests: it reviews all relevant data sources, it obtains perspectives of affected gender groups, and it examines all GBA considerations. It turns out that, as often as not, policy initiatives are not accompanied by completed gender-based analyses:
The Auditor General's report is valuable, yet it fails to recognize the key aspect of GBA+ that makes carrying out GBA+ is so challenging for many policy analysts. Gender-based analysis involves something other than mere disaggregation, or stakeholder analysis. As the Status of Women Canada web page says, "It is simply a different approach, focused on diversity and inclusion." It is about recognizing the importance of identity:
....women, men, boys and girls are not homogeneous groups. A variety of factors, sometimes referred to as identities, such as age, culture, language, sexual orientation, education, ability, geographic location, migration status, faith, ethnicity, and socio-economic status, intersect with sex and gender to create someone’s experiences (emphasis added).
The Auditor General's 2009 report does not use the word "identity" once. There is one mention of the word "identity" in the 2015 report, in a direct quote from Status of Women Canada. Yet without recognizing that GBA+ is about identity - which is a fundamentally, methodologically different approach to policy analysis - it is impossible to realize why incorporating GBA+ considerations can be such a struggle.
For social psychologists and sociologists, identity is a key analytical concept, has a (reasonably) precise and clearly understood meaning, and is a valuable tool for explaining behaviour. So, for example, Arrighi and Maume use men's desire to maintain a strong masculine identity to explain men's avoidance of housework - doing household chores would threaten their sense of themselves as men, so they resist undertaking such tasks. GBA+'s embrace of identity is an embrace of a key sociological and social psychological concept. It thereby signals a willingness - even a desire - to embrace a sociological and social psychological methodologies and research paradigms. Yet many government policy analysts will have come through standard Masters in Public Policy or MA - Economics degrees, where they would receive little or no training in these areas. "Identity" just sounds like blah blah blah - and a GBA+ framework that is premised upon an understanding of identity is incomprehensible.
Incorporating GBA+ into a formal economic framework.
So let me try to translate the idea of identity into terms that economists may be more familiar or comfortable with. Rachel Kranton and George Akerlof have a theory they call "economics of identity". In the Akerlof-Kranton framework, a person's utility depends upon their own actions, the actions of others, and their sense of identity. Re-working their model to fit with the cost-benefit framework set out above, let's say:
Ui=Ui(xi(Ii),Ii)
Utility, Ui, is a function of consumption, xi. But now the utility a person gets from these goods xi is mediated by their sense of identity, Ii. In addition, identity is an independent source of utility or disutility. With the Akerlof-Kranton framework, cost-benefit analysis then becomes a matter of estimating:
dV=∑[(δV/δUi)(δUi/δxi)(δxi/δIi)+(δV/δUi)(δUi/δIi)](δIi/δz)dz
A project, z, impacts utility in two ways. First, as is shown by the first term in square brackets, a project impacts utility by changing consumption possibilities - but the happiness or satisfaction a person derives from the changed consumption possibilities is mediated by identity. Second, a project will affect utility directly to the extent that it corresponds to, conflicts with, reaffirms, threatens or otherwise affects people's sense of their own identity, as shown by the second term in square brackets.
So, for example, a GBA+ analysis of child care subsidies, like a conventional cost-benefit analysis, would need to consider the impact of child care subsidies on parental labour supply and family incomes using a big, quantitative, dataset - the (δV/δUi)(δUi/δxi) part of the cost-benefit calculation. GBA+ would also demand that the possibility that different groups of men and women react differently to childcare subsidies, depending upon their identities, or their sense of themselves - the (δxi/δIi)(δIi/δz) component of the calculation above. Moreover, GBA+ demands in addition that the analyst think about how diverse groups of women and men might feel differently about child care subsidies, depending upon how they see themselves - the (δV/δUi)(δUi/δIi) term above.
Just how much diversity must be considered? Consider, for example, this advice from the Status of Women Canada website:
When conducting GBA+ it is important to understand the complexity of different aspects of identity or different socio-economic factors. For example, research that does not distinguish between the experiences of diverse Aboriginal women (including Inuit, Métis and First Nations, on and off-reserve) could mask important differences.
I chose this quote because it captures beautifully the differences between the aspirations of the Status of Women Canada's GBA+ framework and the reality of what is possible with readily available Canadian data. It's really tough to obtain information on the experiences of diverse Aboriginal women. The National Household Survey's response rate was too low in many remote areas to be of much use. The Canadian Community Health Survey (CCHS) does not sample the provincial on-reserve Aboriginal population, and neither does the Aboriginal Peoples Survey. Et cetera. Even when relevant data is collected, as in, say, the 2006 or 2016 Census, confidentiality demands that fine-grained details be masked in the public use microfiles.
Moreover, even if it was possible to tease out the impacts of policy on various groups of people with sufficiently disaggregated data, this leave unanswered questions about identity and experience - questions that are supposed to be part of a GBA+ analysis such as "Are there differences in how women and men or diverse groups access or experience the program or service?" (Source) or "Have the experiences of diverse groups of men and women been considered in defining the issue?" (Source). Implicit in the GBA+ framework is a call to go beyond a go-out-and-run-some-regressions approach to policy evaluation. Rather, women's and men's experiences matter. Culture matters. Qualitative data is o.k. Diverse stakeholders should be consulted, and their voices should be heard.
But where does the budget for focus groups or other types of consultations come from? Where can representative voices - as opposed to special interests- be found? Who has the knowledge and tools and resources to properly understand and analyze qualitative information, like focus group responses? I find it unsurprising that the Auditor General found that the federal government's GBA analyses "were not always complete and that the quality of the analyses was not consistent". When I go onto the Status of Women Canada on-line training module, and read that a GBA+ analysis should "Consider how identity factors intersect with" the issue under discussion I'm tempted to just give up because I don't understand how to do that.
GBA+ with an objective measure of well-being
It is telling that the practice scenario in Status of Women Canada's on-line GBA+ course involves health outcomes, rather than redistributive or labour force policies. Here is a screenshot setting up the scenario:
After the analyst chooses the right answer (C), they go to the Statistics Canada website, and find some gender-disaggregated data on heart disease. In the briefing note, they include the following:
When it comes to health, identity and experience matter, but in ways that are, to some extent, observable, predictable, and quantifiable. Epidemiological data can be used to show that a person's identity - ethnicity, gender, immigrant status, and so on - has a measurable effect on his or her risk of disease. Health utilization data shows that some groups more are likely to seek treatment than others. As the GBA+ scenario screenshot above describes, there is evidence that a person's identity affects how their symptoms treated, and the health outcomes they experience. In a scenario like this, doing a GBA+ analysis seems to make good policy sense. That's because we can tie things like "identity" and "experience" to measurable outcomes that matter, like the rate of mortality from heart disease.
GBA+ applied to tax-benefit policy
Let's contrast the heart disease scenario in the GBA+ training module with the kind of policy problem that an economist at the Department of Finance might be asked to analyze: a change in the method of determining eligibility for the GST credit.
It used to be that, in order to receive the refundable Goods and Services Tax (GST) credit, a tax-filer had to tick a box indicating that they wished to apply for the credits. In a married or cohabiting couple, only one person could tick the box. Any benefits would be paid to the box-ticker. Starting with the 2014 tax year, the GST credit delivery system changed. Now eligibility for the GST credit is calculated automatically for every tax filer. For married or cohabiting couples, the GST credit for the entire family is paid to the person whose tax return is assessed first.
The assessment of eligibility for GST credits was changed because of concerns about take-up rates (see, for example, this old Auditor General report). A gender-blind analysis of the reform might use, say, the Longitudinal Administrative Database (LAD) to identify people who look like they should be eligible for GST credit on the basis of their income, but who did not receive it - presumably because they failed to tick the "apply for GST credit" box. The analyst would then estimate how many people could potentially benefit from the change to a "tick the box" system to an automatic payment system, and by how much. A gender-blind analysis might even break out the potential impacts of the change in GST credit eligibility assessment by age, family type, location, immigrant status, and so on.
An analysis that incorporated GBA-type considerations would go further, and estimate how the changed method of determining GST eligibility impacted men, women, and the gender distribution of GST credits. The first step would be to gather baseline information: before the change was implemented, how often did women receive the GST credit? How often did men? Who receives GST credits on behalf of families?
The next step would be to try to predict the impact of the change. Again, it would be necessary to estimate how many women appeared to be eligible for GST credit, and did not receive it, and how many men were in that same category. But then, for multi-adult households, there are additional intra-household dynamics to consider: how does going from a box-ticking system to a first-assessed system change who gets the GST credit?
To answer this question, it is necessary to know how Canada Revenue Agency assesses tax-returns. I would guess that CRA assesses tax returns in the order that they are filed - so returns that are filed first get assessed first. Which order are returns filed in? This is a revealing question for two reasons. First, it shows - again - the inadequacy of available data for gender based analysis. Second, it shows that identity matters.
Imagine what happens when a couple - or a married or cohabiting person - visits a tax professional. One person's name has to be entered into the tax software first; one person's return has to be submitted first. Based on my experiences with just about every financial institution I have ever dealt with, my guess is that generally men's names get entered into the tax software before women's. I can understand why. Imagine the situation of tax preparer making a split-section decision about which name to start out with. The "Mr and Mrs" ordering is a strong social norm, and so it just feels right. It is unlikely to offend. Putting a woman's name before her husband's is risky.
This matters because if men's names are entered into the tax software first, then men's tax returns will be filed first, and assessed first - so the switch to automatic assessment will lead to men receiving the GST credit more often. Perhaps. Unfortunately we have no data. There is very little data available on the gender break-down of GST credit recipients. There is very little data available on how couples divide up the work of filing their income taxes. There is no data I know of on the gender-role attitudes of tax preparers.
This is what happens so often when one tries to incorporate gender norms and roles and identities into the analysis. Even when - as in this GST credit example - it is clear that gender norms will matter, there may be little or no evidence that can be used to predict how these norms will play out in the policy scenario under consideration. So what can we do? Rely on one or two tangentially related studies and whole heaps of guesswork, supplemented by stereotypes about gender roles and behaviour? Well, at least we could say that we've "reviewed relevant data sources."
This GST credit example illustrates a number of crucial points: even something that seems like a straightforward technical tax policy can have gendered impacts; getting proper data is both critical, but also extremely challenging; in the absence of data, it is easy to rely on stereotypes and suppositions, and these are not necessarily an improvement over business as usual. And there is still more to be said.
Defining good policy without an objective measure of well-being
Up to now I have been assuming, implicitly, that a "good" method for delivering the GST credit is one that results in more credits being paid to women. There is evidence to support this view: studies in Britain, Canada, Ghana, South Africa, and other countries find that funds received by women are more likely to be spent on good things such as education, food, or clothing.
Yet the GBA+ framework calls upon the analyst to question his or her assumptions:
GBA+ helps us to recognize and move beyond our assumptions, to uncover the realities of women's and men's lives in all their diversity, and find ways to address their needs. Source.
GBA+ reminds the analyst that people's identities shape their experiences, and other people's experiences may be different from those of the analyst. A good GBA+ analysis involves obtaining the perspectives of affected groups, and listening to their views - not just crunching numbers and examining data.
But if we were to go out and obtain these other perspectives, we might call into question the assumption that the best policy is the one that empowers women by giving them as much control over money as possible. The majority of Canadians, when asked, characterize their financial relationships with their partners in terms of sharing. For example, when the 2014 Canadian Financial Capability Survey asked married or cohabiting people "Overall, who is mainly responsible for making financial investment and planning decisions on behalf of the family?" over half of respondents reported that they shared the responsibility equally (my calculations). It is entirely possible that, if diverse groups of men and women were how they felt about the payment of the GST credit, many would say that it did not matter who received the credit.
Even if people were to admit that the payment of GST credit mattered, they might be unwilling to support a payment system that undermined traditional gender norms. As documented by Bertrand, Kamenica and Pan there are powerful gender identity norms that "induce an aversion to a situation where the wife earns more than her husband." What would we do with a GBA+ analysis that consulted with diverse groups of women, and found that some thought that refundable tax credits should be paid to male family members, in an attempt shape the messy reality of their lives to some idealized gender norms?
Feminists - and data driven scientists - can design more effective and responsive policy by listening to what people want. Perhaps people feel differently credits, and allocated the money differently, depending how the credits are labelled ("child benefits" versus "tax refunds"), how they are timed (lump sum versus monthly payments), and so on. Yet it's also important to acknowledge the limitations of consultation. People can't know how they will feel about a policy until its in place. A trivial example is the elimination of the penny: people thought they'd miss it; they didn't. Moreover, the views people voice will be shaped by their perceptions of appropriate roles and behaviour.
Serious feminist scholars have struggled with this issue. Jennifer Nedelsky's work on reconceiving autonomy neatly summarizes the dilemma:
Autonomy is a capacity, but it is unimaginable in the absence of the feeling or experience of being autonomous.... To be autonomous a person must feel a sense of her own power (which does not mean power over others), and that feeling is only possible within a structure of relationships conducive to autonomy.
Does this mean that would-be liberators should ignore identity? No. But it explains why GBA+ is easier to apply to, say, health policy than to economic policy. Better health is pretty universally considered to be a good thing, regardless of how one sees one's identity. Identity matters to the achievement of outcomes, but not the articulation of what a desirable outcome would be. It is possible to achieve a general consensus that the ultimate aim of policy is, say, to improve health. By way of contrast, the aim of refundable tax credits like the GST credit is to make people in some sense better off. There is inevitably fuzziness and ambiguity about what "better off" means. A feminist may interpret "better off" as "conducive to the achievement of autonomy." But as Nedelsky points out, autonomy is unimaginable to those who have not experienced it. Hence even the groups that the policy is intended to help may not share the goal of autonomy (or goals of empowerment or other similar ends) - and even if they share the goals, they may not understand how to achieve them.
Policy Options
One of the frustrating aspects of the GST credit example is that there are some many other ways that the GST credit payments could have been delivered, at least some of which would have been more conducive to gender equity. The credit could have been paid to the lower income partner. It could have been divided equally between the partners. It could have gone to the primary caregiver, for couples with children. And so on. But none of these options were chosen.
GBA+ part of a type of process known as "gender mainstreaming", defined by the United Nations as follows:
Mainstreaming a gender perspective is the process of assessing the implications for women and men of any planned action, including legislation, policies or programmes, in all areas and at all levels. It is a strategy for making women’s as well as men’s concerns and experiences an integral dimension of the design, implementation, monitoring and evaluation of policies and programmes in all political, economic and societal spheres so that women and men benefit equally and inequality is not perpetuated. The ultimate goal is to achieve gender equality.
Gender mainstreaming goes beyond policy analysis to policy design. For gender mainstreaming to be effective, gender considerations have to be taken into account at the stage in the policy process when ideas are being put on the table, when options and alternatives are being considered. Gender mainstreaming will be effective when powerful people, people who are in positions of influence, people who are making policy are aware of, and take into account, gender considerations.
There is nothing about the GBA+ framework that stops it from being applied to the design of policies - in fact it is pretty clear from reading the Auditor General's Reports and the Status of Women Canada website that, in an ideal situation, GBA+ would be incorporated into policy design. So why isn't it? Part of it comes down to what senior people in the organization think matter. For example, the Auditor General's report hints that one of the barriers to implementation of GBA+ is "limited senior management review of the completeness of gender-based analysis" - i.e. some senior management people are basically ignoring GBA+ requirements.
Yet I would argue that the GBA+ approach is part of the problem also. Looking at other countries, "gender-based analysis" does not seem to be as widely used an approach to gender mainstreaming as "gender-responsive budgets" (see, for example, the last chapter of this report). An advantage of "gender budgeting" is that it specifically locates gender considerations at the point in the policy cycle during which policies are being formed and debated.
Gender budgets can be produced within government, but as often they involve consultation with - and indeed are produced by - outside stakeholders - much like the gender equality chapter of the Alternative Federal Budget produced by the Canadian Centre for Policy Alternatives. One advantage of bringing stakeholders to the table is that it may help focus the minds of people in leadership positions: if gender-responsive policy making is seen to be a vote winner, it will happen.
A key part of having wider consultations and wider involvement in the policy formation process is more widely available data - for example, making more gender-disaggregated data available (it would be wonderful to have a completely updated and expanded version of Statistics Canada's Women in Canada publication, for example), more extensive reporting of summary administrative data, such as the income statistics on the CRA website, more consistent production of public use microfiles for statistical datasets, and an expansion of the Data Liberation Initiative to incorporate non-profits and community groups.
Conclusions
Canada's gender-based analysis framework is highly ambitious. It asks people to use a nuanced and sophisticated understanding of gender to inform policy making and evaluation. It demands finely disaggregated data, and widespread consultation.
However if GBA+ is ever going to be widely done in government, it will end up being done by lots and lots of policy analysts. Some of those people won't have a nuanced and sophisticated understanding of gender. GBA+ will end up being done by people who are working to tight deadlines with inadequate funding. What we need is not just GBA+, but GBA+KIS - or gender based analysis plus keep it simple.
Do all policies require a full consultation with affected groups? Or is it possible to have a stripped down version of GBA for small policy initiatives, and a fuller version for large ones? What are the minimum, essential steps for GBA? How can GBA be framed in a way that makes sense to people coming from a diverse range of methodological and disciplinary perspectives?
I'd also like to see GBA+DATA. Good data is the starting point; without that we risk being in the world of stereotypes and handwaving and vague assumptions.
Finally, GBA+C, or GBA plus cynicism. Gender-based analysis can be used all sorts of ways, and to reach all sorts of conclusions. For example, the Economist recently ran an article arguing that feminist economics deserved recognition as a distinct branch of the discipline. The article notes that "Economics as commonly practised often misses out another important element of inequality between the sexes: unpaid work." And what happens when unpaid work is taken into account? Well, it turns out that women working outside the home isn't quite as good for the economy as some might suggest: taking unpaid work into account "boosted GDP overall, but lowered the growth rate: as women have moved into paid work, they have been doing less unpaid work at home, so total production has not been rising quite as quickly as official figures suggest." So a feminist economic analysis suggests that women's labour force participation isn't all it's cracked up to be?
GBA+. It could be an exercise in box ticking. It could be meaningless blah blah blah. But right now there is a government in power that is committed to gender equality and there is a chance that GBA might, just might, become a valuable tool for creating policies that promote gender equality.
This post was written for M and S, who have invited me to talk about GBA at the Department of Finance later this week. K and D listened patiently, and gave wise counsel.