Following 
the norms

In some parts of the world, deep gender inequalities in labour market participation persist despite decades of strong economic growth and other social improvements. This persistence of women’s low participation can be explained by introducing social norms in economic analyses.

We talked about this with Maurizio Bussolo, economist, on the occasion of the thirty-second annual conference of the International Association for Feminist Economics (Iaffe). inGenere was media partner of the conference, which took place in Rome from 3 to 5 July 2024. Bussolo analyses gender inequalities in South Asia – Afghanistan, Pakistan, India, Bangladesh, Nepal, Sri Lanka, Maldives and Bhutan – for the World Bank. The aim of his work is to identify on the ground, particularly in Nepal, Bangladesh and Pakistan, the norms and attitudes that hinder women's access to the labour market.

What is meant by social norms and how can they be measured?

To answer this question, it is necessary to first consider the situation of the countries of South Asia, where we can observe a deep gender disparity in the level of participation in the labour market: in fact, only about a third of women of working age participate, against more than three quarters of men. On average, in the rest of the world, female participation in the labour market is about 55%. It is interesting to note that this gap has not changed in the last two decades or more, despite high economic growth and the fact that women have increased their education, while the number of children per woman has decreased.

Why, then, does this disparity persist?

The literature has recently begun to analyze culture and ‘social norms’, subjective variables that have not yet been widely used in economic studies, but which can explain why women’s participation in the labour market has not increased in South Asian countries. In fact, if we rely on the few quantitative data we have on the ‘attitudes’, we see that there is a strong correlation between negative attitudes towards gender equality and women’s participation in the labour market.

Can you give us a definition of social norms?

According to the definition contained in studies on the subject, social norms “are informal rules based on beliefs that people have regarding the behaviors to follow in a given social context and that are approved or disapproved by a specific group of people, the reference group”.[1]

Can you explain what does this mean, exactly?

The definition includes several elements: on the one hand, beliefs about the behaviors to adopt are ‘personal’, that is, based on each person’s attitudes and preferences. On the other hand, these beliefs are also derived from the perception that individuals have regarding the behaviours that other people (the reference group) expect from them in a given situation. It is important to underline the difference between what we think is right to do, and what we think society expects as the correct behaviour. Until recently, most of the available data concerned exclusively personal beliefs: what I, a man or a woman, think is right to do in a certain situation. A typical question in the World Values ​​Survey, one of the most widely used social surveys globally, is: “if jobs are scarce, do men or women have priority to get those jobs?”. The answer to this question reveals a personal belief or attitude.

Are there alternative ways of asking the question to collect different information?

In the new survey we did, we asked: “In your opinion, among 10 of your peers, how many think it is right to give priority to men over women?”. Detecting the difference between personal attitudes and social expectations is very important for quantifying social norms. We started measuring them separately in Nepal, but the survey was also done in Bangladesh, and a variant – not on women’s labour market participation but on women’s education – was done in Pakistan.

Is it possible to explain the gap between female and male labour market participation using standard economic models, which do not take social norms into account?

The collective or bargaining model of the household, used to study the decisions of a family where the husband and wife do not necessarily have identical preferences, can be used to explain the specialisation of men in paid work activities outside the home (breadwinning) and of women in unpaid domestic work (homemaking, for example raising children) when there is a difference in relative productivity in the two activities. In fact, if there is a difference in wages, whereby the man earns more (and wages correctly measure productivity) in the labour market, while the woman is more productive in caring for children, the prediction of this model is that women specialise in homemaking and men in breadwinning, and this maximising the utility (or well-being) of the family. According to this model, if these wage or productivity gender gaps are reduced, the specialisation is reduced and thus the difference in participation decreases. But when gender gaps are reduced, as in the case of South Asia, participation does not increase, contradicting what the model predicts. In fact, in South Asia, female labour participation is higher in rural areas where the wage (or productivity) gap is higher, compared to urban areas. In other words, the model predicts higher female labour participation in urban areas, but the data contradict this prediction.

How can this phenomenon be explained?

By introducing social norms into the standard model. Social norms balance the model’s tendency to predict more female participation when the wage gap decreases. As we said, a very common social norm is that relating to the role of the man as breadwinner which can be inferred by the fact that the husband has a higher income than the wife. Let’s start with an example – which comes from a completely different part of the world but which helps us understand: the scholar Marianne Bertrand and her research group have written an interesting article on the increase in divorces in the United States in situations where women reach earnings close to 50% of the family total income, therefore contributing equally.[2] This shows that, when the wage gap within the family is relatively high, the man feels empowered, his role as breadwinner is secure. When the income gap narrows, however, a ‘disutility’ is created, that is, a reduction in the overall utility (which in the extended model includes adherence to social norms) to for the family: women, therefore, in order not to decrease the overall well-being of the family, even if the opportunity cost of not participating in the labour market increases, because they could earn more, decide not to do so. Adding social norms to the standard model therefore allows us to resolve this apparent contradiction.

In practice, how did you collect information to measure social norms?

We identified a series of themes relevant to labor participation and for each theme we measured various elements. The themes were identified through a qualitative analysis, which consists of a series of unstructured interviews where both women and men ‘told’ the interviewers what, in their opinion, are the key issues related to the difficulties for women in accessing the labor market.

What are the themes you identified and the elements you measured?

In addition to traditional gender roles (breadwinner husband, homemaker wife), we identified more specific themes related to male honour; at work, in situations where women and men work together, or working hours; mobility, considering that in some cases travel can be long and dangerous for women; finally, sexual harassment. For each of these themes we measured various elements: personal attitudes (or first-order opinions); social expectations (or second-order opinions, i.e. opinions on what other people, reference groups, may think); the reference group whose opinions are relevant; sanctions (if I do not respect the social norm, what are the possible social sanctions to which I expose myself?). Another important element is the measurement of the variation in personal attitudes and social expectations depending on the context. For example, for a woman, being a widow justifies her participation in the labour market more than that of a woman who is not, as does having a husband who is an alcoholic or has health problems (in Nepal we encountered these phenomena).

What do we learn from these observations and measurements?

Using a factor analysis, we try to measure the latent variable that interests us: social norms, which are not directly observable. With factor analysis we can summarise the information obtained from the survey in two main factors (or indices): one refers to gender roles (for men and women) and the other to the issue of mobility and work outside the home, and they summarise all this data (the different elements for the different themes) both from the point of view of personal attitudes and from that of social expectations. In a next step, we measure whether these indices can predict female participation in the labour market. That is, whether variations in the level of restrictiveness of the social norms (summarised by the indices) correspond to similar variations in female participation.

What are your findings?

A regression shows a strong negative relationship between the levels of the index and female participation (the more conservative the norm is the lower the rate of participation). This is also true when controlling for fixed effects, geographic region, and socioeconomic status. This is a key result we obtain in Nepal. Two other results are interesting: one is that normative social expectations (i.e., what I think others think, or second-order beliefs) have a greater explanatory power than personal attitudes, confirming that if we approximate social norms with personal attitudes, we risk underestimating their potential impact on female participation. The other result is that sanctions have almost the same correlation with female participation as social norms.

Why is this finding on sanctions an interesting one?

Because measuring sanctions is simpler as it is possible with a question that is easy to ask and well understood. If the correlations between social norms and female participation on the one hand, and between sanctions and female participation on the other are similar also in other countries, it might be possible to replace the very complex questionnaire to measure social norms with the simple question used to measure the intensity of sanctions. This would allow us to introduce (at low cost) the measurement of social norms in many more surveys, and thus to accumulate more data on social norms across countries and across time. This is all the more important if we consider that the starting point was the observation that we have very few measurements of social norms.

Notes 

[1] The definition of social norms as “informal rules that embody beliefs about which behaviors are approved or disapproved of in a specific context by a given social group” can be found in Görges, Nosenzo, 2020; Cislaghi, Heise, 2019; and Bicchieri, 2006, 2017. 

[2] M. Bertrand, E. Kamenica, J. Pan, “Gender Identity and Relative Income within Households.” Quarterly Journal of Economics, 130 (2): 571–614, 2015.

References

B. Cislaghi, L. Heise, “Using Social Norms Theory for Health Promotion in Low-Income Countries.” Health Promotion International, 34 (3): 616–23, 2019.

D. Nosenzo, L. Görges, “Measuring Social Norms in Economics: Why It Is Important and How It Is Done.” Analyse & Kritik, 42 (2): 285–312, 2020.