econ problem set

Home Production, Women’s Market Work, and Structural

Transformation∗

Taryn Dinkelman† L. Rachel Ngai‡

April 7, 2021

Abstract

This paper reviews the literature on how linkages between women’s home and market

work change through the structural transformation. We relate historical shifts in female

time between home production and market work in developed economies to more recent

changes in developing countries. We highlight forces that contribute to shifting women’s

time into the market in developed countries, and barriers that may get in the way of

this movement of time. We review available and required data that could be used to

uncover frictions retarding the movement of women’s time into the market in developing

countries. Several areas for future research addressing these frictions are proposed. To

inform policies that address macroeconomic misallocation of female labour resources and

persistent gender gaps in labour market outcomes in developing countries, we need (i)

more time use data to understand how women spend their time in home production and

in the market; more research on (ii) how infrastructure affects home labour productivity;

(iii) how child-care services and part-time or flexible work arrangements facilitate market

work; and (iv) whether social norms against women’s market work are malleable.

∗This paper was prepared as a Pathfinding Paper for the CEPR-led research programme on Structural Transformation and Economic Growth (STEG), funded by the United Kingdom’s Foreign, Commonwealth and Development Office (FCDO) as part of the UK aid effort. We thank Rachel Heath for excellent comments on a draft of this manuscript. Research assistance was provided by Brigid Meisenbacher and Vivek Moorthy.

†University of Notre Dame, CEPR, and NBER. Email: [email protected] ‡London School of Economics, CFM, and CEPR Email: [email protected]

1 Introduction

Economic growth is often measured as growth in output per market hour. But for many

countries, time spent in home production is almost as large as the amount of time spent on

market production. For the world as a whole, using national time use surveys from sixty-four

countries for the 2000s, the International Labour Organization reports that there are about

seven hours of home work for every ten hours of market work (Addati et al., 2018). Three

quarters of these unpaid home production hours fall on women.

Home production is defined as time spent on the production of goods and services for one’s

own use. For advanced economies, common home production activities are cleaning, cooking,

and childcare. In early stages of economic development, however, people also grow their own

crops, keep small farm animals, make clothes, and preserve food. These activities are gradually

taken over by the market as the economy develops and markets improve. This process of

marketisation is linked to the process of structural transformation in the market economy:

the fall in agriculture, hump-shaped manufacturing, and rising services; a process observed

both historically for advanced economies and across a large panel of developed and developing

countries.

The evolution of home production during the structural transformation has important im-

plications for the economic role of women. Traditionally women are the main source of labour

for home production and this tradition seems to persist into the 21st century, where women’s

home production hours are triple that of men’s. These demands of home production are the

main self-reported barriers to women entering the labour market (see Addati et al. (2018)).1

Cultural and social norms regarding women’s role at home are important too, but they seem

to be shifting in some places: a 2016 ILO-Gallup survey finds that 70% of women would prefer

to work in the market, and 66% of men agree with this preference.2

A central question is: “How can women increase their market hours – should they want to

– without diminishing the required output of home production, e.g. a clean home, meals, well

looked-after children?” The affordability and acceptability of marketised goods and services

1The home production hours are based on 64 national time use surveys and the main reasons for staying out of labour force are based on 84 national labour force surveys in 2000s. In contrast to women, the main reason that men report being out of the labour force is because they are “in education, sick or disabled”.

2The 2016 ILO-Gallup survey is based on 142 countries with 149,000 respondents ages 15 and older.

1

that allow women to outsource home production time while allowing families to still enjoy the

output seems to be the key. The development of corresponding market sectors that substitute

for home production also creates jobs for women because these sectors are usually intensive in

female labour. Thus, potentially two paid jobs for women can be created by outsourcing home

production: one for the woman who provides the home produced service in the market, and

one for the woman who chooses to leave home production and shift time towards market work.

This was the historical experience of many advanced economies.

When women face barriers to transitioning from home to market, or barriers to transitioning

across sectors during the structural transformation, the misallocation of time and talent that

results reduces aggregate output (Hsieh et al., 2019; Lee, 2020a). Addressing the barriers in

the way of shifting women’s time out of home production and into the market is therefore a

first order problem for many developing countries today.

Eliminating frictions that slow down the marketisation of home production or impact the

acceptability of women’s market work could lead to substantial gains in output in a post-COVID

world. Unfortunately, the centrality of marketised home production activities in supporting

the movement of women into the labour market is particularly salient in 2020. The COVID-19

pandemic has unleashed strong forces of demarketisation: the shutting down of many market

services such as schools, restaurants, cleaning services etc. Data from advanced economies show

that women took on the bulk of these demarketised home hours and there is growing worry

about the reversal of women’s role back to the home in all countries, e.g. (Alon et al., 2020;

Georgieva et al., 2020).

In this paper, we review existing theory and evidence from advanced countries on the role of

home production in the structural transformation. We address the following questions: does it

look like developing countries are following the historical path of advanced economies in home

and market work allocations and in gender gap outcomes in the labour market? Which factors

that were influential in the historical path of advanced economies are at play in developing

countries? Which new factors should be considered now? Are advanced country policies that

support the marketisation of home production relevant for developing countries? What types

of new data are needed to answer these questions? Our goal is to set out a research agenda that

will build evidence on opportunities for and constraints to marketisation of home production

2

sectors and the acceptability of women’s market work in developing countries.

2 Home Production and Development

What is home production? The book Economics of Household Production by Reid (1934)

explains:

It consists of those unpaid activities which are carried on, by and for the members,

which activities might be replaced by market goods, or paid services, if circumstances

such as income, market conditions, and personal inclinations permit the service

being delegated to someone outside the household group. (Reid 1934: 11)

Home production is defined as time spent on the production of goods and services for own

use.3 Typical modern day examples are cooking, cleaning, and child care around the globe;

and wood and water fetching in developing countries. There are two important features of

home production that distinguish it from leisure. First, one derives utility from the output of

home production but generally not from the time that she spends on performing this work.4

Second, as Reid (1934) observed, time spent on home production can potentially be outsourced

to the market while household members can still enjoy the output. This happens via a process

of marketisation of home production (Lebergott, 1993; Freeman and Schettkat., 2005). This

marketisation process depends on the development of sectors that produce close substitutes to

home production, such as those among service sectors, which is closely linked to the process

of structural transformation: the reallocation of labour across agricultural, manufacturing, and

services along economic development path (Kuznets, 1966; Maddison, 1980).

3In early development stages, both home production and working on the family farm or in the family business are important, and both are unpaid work. It is the emphasis on “own use” that differentiates home production from the latter which is referred as “market-oriented” establishment and is included as part of formal labour supply by the International Labour Organization (ILO). Although this distinction is conceptually clear, it is often difficult to measure in practice, since home businesses may produce goods and services for own use as well as for market.

4Positive utility from some aspects of home production are clearly possible but it is unlikely that all time in home production is enjoyed. For example when the people were asked to rank the enjoyability of all different activities during the day in the UK’s 2014-2015 time use survey, most home production activities were placed at the lowest end of the enjoyability spectrum. Childcare ranked below playing with one’s own child. See https://ourworldindata.org/time-use-living-conditions for more details.

3

2.1 Measurement of Home Production

The current standard for measuring input into home production is to use a time use survey.

Survey respondents are asked to enter activities in specific time increments (e.g. 15 minutes) for

each of the 24 hours in a day. In some surveys, respondents are asked flag whether an activity

completed in a given window was the primary, or secondary activity, while in others, only the

primary activity information is collected.5 Time diaries may be completed contemporaneously,

or recorded using the recall method for time use in the previous day. Often, one or two

individuals in each household are randomly selected to participate in these surveys. Participants

tend to be 15 years or older in developed country surveys, and 10 years or older in developing

countries.6

In describing aggregate patterns in time use, and how these patterns shift from home to

market work, macro-economists typically aggregate information from time spent on a given

day to time spent in a representative week by a representative person. Typically, this is done

by calculating how much time is spent on a given activity during an average weekday, as well

as during an average weekend day, and then adding up over days of the week to produce a

total weekly hours of work variable. To produce national averages, this aggregation occurs over

everyone: both those who spend any time on a given activity and those who spend zero time

in the activity.

Historical time use surveys from specific populations in developed countries reveal an im-

portant transformation of home activities as countries modernise. Based on 60 families in

Pennsylvania, Leeds (1917) documented most families at the turn of the century spent time

on keeping small farm animals, making clothes, and preserving food for their own use. Two

decades later, Reid (1934) wrote about the disappearance of these activities. Using various

time use surveys collected by US Bureau of Home Economics, Vanek (1973) documented the

transformation of home activities from 1924 to 1968, revealing the convergence to modern day

home activities in developed countries.

5For example, someone could be cooking dinner as the primary activity and watching kids as the secondary activity.

6This difference raises an issue of whether children work in home production and/or in the market. The issue of how children shift from being workers to being in school is beyond the scope of this paper: see Edmonds (2005) and Edmonds et al. (2009) for examples of how child time allocation and work shifts in the process of trade-induced economic growth.

4

While these early time use surveys were often based on small convenience samples, in recent

decades many countries have started conducting surveys with a more representative sample of

the whole country. The first nationally representative time use survey became available for

the US in 1965 and the American Time Use Survey began in 2000. Effort has been made to

harmonise time use surveys across countries such as those of the Harmonised European Time

Use Survey (HETUS) which includes 23 European countries and the Multinational Time Use

Survey (MTUS) which includes 24 countries across continents.

A second common way to calculate time use on home production activities is to ask questions

about how much total time per week is spent on specific activities, such as hours spent on

household management activities like water collection, or firewood collection. Such questions

are common in household surveys conducted in developing countries. While they may be subject

to more recall error in terms of actual time spent in the activity, they may more accurately

measure extensive margin participation in such activities by asking everyone in the household

(rather than two randomly selected individuals) about an activity that may not occur every

day of the week.

A central motivation for collecting time use data is to measure time spent on unpaid work,

particularly unpaid work in home production. Valuing this unpaid work and incorporating

this value into the national accounts is sometimes referred to as the extended GDP.7 For the

US, using wages of domestic workers and gross return on household durables to value unpaid

home production work, Bridgman (2016) finds that in 1929, the extended GDP is 42% higher

than measured GDP. Over time, this ratio declines due to a shift of unpaid home hours into

the market sector, but extended GDP is still 27% higher than measured GDP in 2010. In a

similar calculation that only values the labour time spent in home production using national

minimum wages, the ILO calculates that the value of the time spent on home production across

53 countries for the 2000s is equivalent to 9% of global GDP Addati et al. (2018). This is

undoubtedly an underestimate of the importance of home production, since it does not include

the value of household durables.

7There are two general approaches to measuring the value of home production, based on output or input evaluation. The output evaluation approach focuses on the market value of a home output such as how much the market will pay for a clean home or a cooked meal. The input evaluation approach instead puts value on the home hours based on wages, where it could be the wage of the person performing the home work, the average wage in the economy, the minimum wage, or the wage needed to hire someone to perform the work.

5

2.2 Women’s role in home production across time and space

What is the role of women in home production? What does home production look like in today’s

developing countries and how does it compare to home production in developed countries in

the past? Does the pattern of home production of today’s developing countries follow those of

developed countries?

To begin, we compare the time spent in different home production activities among indi-

viduals who might be referred to as housewives. In Table 1, we show the average weekly hours

spent on cooking/food preparation, cleaning/care of household and gardens, laundry and repair

of clothing, child and adult care, and general household management (purchasing food, travel

related to home production, etc). We do this using time use recorded among farmer house-

wives in the US in the 1920s, American housewives in 1965, and women who could be called

housewives in America (2010), South Africa (2010), and Ghana (2009).8 The 1920s and 1965

US data are from Ramey (2009); the other data are from national time use surveys.

As noted in the prior section, time use in different activities is collected with some mea-

surement error. For instance, in the US 2010 survey using for Table 1, except for time spent

on childcare, no time spent on secondary activities is collected. This means that some home

production time in secondary activities would be uncounted. In the other countries, secondary

activities are more likely to be included in these measures. For example, in South Africa,

respondents listed up to three activities per half hour and we assign a share (30 minutes, 15

minutes, or 10 minutes) of time to each activity based on the number of activities listed. In

Ghana, respondents listed up to five activities per hour along with the time spent on each ac-

tivity. Where activities were listed as simultaneous, we split the time allotment equally across

activities.

Despite these issues of measurement, the table provides a snapshot of how the time in and

composition of home production has changed in the US with increasing levels of development.

Among housewives in the US, time spent in home production has fallen by about 7 hours, or 15%

8The relevant sample for which we compute weekly hours of work in the 2009 and 2010 time use surveys includes women ages 15 to 59 (before they attain pension age), who report spending zero hours in market work and zero hours in education in the prior weekday. We do not restrict to married women, because households may be inter-generational arrangements of families or may contain multiple wives. In recent work on the US, Lee (2020a) documents that home production time jumps up when women transition from single to married status.

6

between the 1920s and 2010. The composition of home production has also shifted over time:

cooking and laundry hours have shrunk dramatically, while time spent in care activities and

household management have more than doubled. This shift is consistent with many household

activities shifting from the realm of home production into the market.

Table 1: Differences in Weekly Hours in Home Production by Housewives

U.S. U.S. U.S. South Africa Ghana 1920s 1965 2010 2010 2009

Total weekly hours 51.7 51.8 43.8 48.0 45.8

Cooking 23.5 16.5 8.6 16.4 24.3 Cleaning 9.6 9.5 7.8 11.1 2.6 Laundry 11.3 6.9 3.3 5.3 2.2 Care of children, adults 3.9 8.5 11.3 6.4 9.6 Household management 3.3 10.5 12.8 8.8 7.2

Household size 4.3 4.1 3.2 4.6 5.2 N. households 559 536 1,661 3,491 1,754

Notes: USA 1920s data are from Ramey (2009). We compute all other columns using national time use survey data. Weekly hours are weighted averages calculated for married women: farm housewives in 1920s, all housewives in 1965, all married women with zero minutes in education and zero minutes in work in Ghana (ages 15-64), the USA (ages 18-59), and South Africa (ages 15-59). Variable definitions: Cooking (food preparation, clean up, fetching wood and water in SA and Ghana); cleaning (care of house, gardens); laundry (mending, laundry, making clothes); care (of children and adults in the household); household management (buying food, shopping, home management, travel for home management, other).

Comparing the more recent data from the US to the time use averages in Ghana and South

Africa allow us to compare home production across high-income, upper-middle income (South

Africa) and lower-middle income (Ghana) countries. Overall, housewives in South Africa and

Ghana spend about as much total time in home production each week as American housewives

did in the 1920s and 1960s. In Ghana, these total home production hours are even closer to

total hours in the US in 2010. However, the composition of home production time in these two

countries differs from the composition of time among US housewives in 2010. In South Africa

and Ghana, the hours spent cooking and cleaning each week (South Africa: 27.5 hours, Ghana:

26.9 hours) is very close to the 26 hours spent on the same activities by American housewives

in the 1960s. Except in the 2010 US survey, cooking makes up the lion’s share of time spent

7

in home production. Cooking and cleaning together make up over half of the total weekly

time in home production among housewives. Household management – including shopping and

travel for home production – takes up a non-negligible 8-10 hours of work time each week in

US households in the 1960s and in South African and Ghanaian households in recent times.

Though similar in total time spent on household management, the composition may be quite

different: for example, fetching wood and water is an important component for rural South

Africans and Ghanaians.

Next, we show how weekly hours of work at home vary with a country’s level of development.

Using data collated by Bridgman et al. (2018), we plot data on women’s weekly hours of work in

home production activities against real GDP per capita in Figure 1.9 Markers of different color

and shape denote observations from different country groups: OECD (high-income), middle-

income countries, and lower-middle and low-income countries.10 The focus here is the average

time spent in home production each week per working-age person. This measure therefore

captures both the intensive margin of hours of work documented in Table 1 as well as the

extensive margin of any participation in any home production.

As countries grow richer, female time spent in home production falls, quite dramatically.

In Tanzania and Guatemala in the 2000s, the average female time spent in home production

was around 55 hours per week. In contrast, women in Norway spent about 25 hours per week

in home production in the 2000s. There are two striking facts here. First, as countries grow

richer, the work of home production moves from being more than a full-time job at the lower

end of GDP per capita, to a part-time job at the highest levels of GDP per capita. Second: in

no country do women spend less than 20 hours per week in home production. At minimum,

home production is the equivalent of one part-time job.

9Weekly hours of home work are computed by Bridgman et al. (2018) using time use surveys. The data cover years from 1960 to 2014 for 43 countries. We do not use any of the interpolated data computed in that paper. We also omit observations from Japan and Taiwan prior to 1990; Bridgman et al. (2018) note that these countries are outliers with very low levels of male home production hours (under 3 hours per week). Data on real GDP per capita are from the Penn World Tables version 9.1 and measured in 2011 International dollars. In all relevant figures, we use the log of real GDP per capita.

10The lower-middle and low-income countries (LMICs) in the sample include: Algeria, Bangladesh, Ghana, India, Kyrgyzstan, Nicaragua, Pakistan, Tanzania, Uganda. Upper-middle income countries include: Albania, Ecuador, Guatemala, Iraq, Russia, and South Africa. OECD countries include Australia, Austria, Canada, Colombia, Denmark, Estonia, Finland, France, Germany, Hungary, Italy, Japan, Korea, Mexico, Netherlands, New Zealand, Norway, Poland, Slovenia, Spain, UK, USA.

8

Figure 1: Weekly hours in home production by level of development

Notes: Weekly hours of work in home production computed for females age 15 and older from Bridgman et al. (2018). Real GDP per capita (2011 international dollars) from Penn World Tables v9.1

Wrapped up in Figure 1 is variation in home production hours across countries and within

countries over time. The decline in home production hours with GDP is most pronounced in

the OECD and middle-income countries, and especially within country over time. By contrast,

in the lower-middle and low-income developing countries, there is less variation in hours of work

across vast changes in GDP per capita.

Part of this difference reflects a lack of data. For many of the OECD countries and even

some of the middle-income countries in Figure 1, time use surveys exist for multiple decades,

so it is possible to trace out changes within countries over time. In the data we use, Tanzania

is the only country in the lower-middle and low-income group that has collected multiple time

use surveys.11 For the other countries, any changes in home production hours at different levels

of GDP (in different years) are not yet observable.

11Moreover, Tanzania’s data in the earliest period covers only one region of the country.

9

Figure 2: Gender gaps in weekly hours in home production by level of development

Notes: Weekly hours of work computed for individuals age 15 and older from Bridgman et al. (2018). Real GDP per capita (2011 international dollars) from Penn World Tables v9.1. The red horizontal line represents parity in home production hours.

Keeping this data gap in mind, the pattern in Figure 1 might suggest that we need to

see substantial increases in real per capita GDP in most developing countries before there are

meaningful declines in home production hours among women in these countries. This would be

in line with the historical trend for the US reported in (Ramey, 2009) that home production

hours of women ages 18-64 hardly changed during 1900-1965, fell by only 5 hours from 47 hours

per week, and then fell by 12 hours over the next 20 years.

It is also instructive to examine gender gaps in home production hours. We plot the ratio of

female/male hours in home production by real GDP per capita in Figure 2, again color coding

markers to visualise differences between high-, middle-, and lower-income countries. The red

dashed horizontal line is at 1, denoting gender parity in the time spent in home production.

The figure shows a wide dispersion of gender gaps in the allocation of home production time at

10

different levels of GDP per capita. For example, at the low range of GDP per capita, Tanzanian

and Ugandan women in the 2000s spent more than twice as much time in home production than

men did. These gaps widen for some countries at higher levels of income: several middle-income

countries (Guatemala, Iraq, Albania) see women doing more than four times as much home

production as men. In other countries, at even higher levels of GDP, the gaps gaps shrink once

more. By the 2000s, in Sweden and Norway, the female/male ratio of home hours falls below

2. At no point in time does the ratio of home production hours reach gender parity in any

country.

2.3 Data needs: Time use surveys

A number of developed and developing countries have time use survey data representative at na-

tional level. Many of them are harmonised and available at https://ipums.org/projects/ipums-

time-use. To really understand whether and how time used in home production has shifted

into the market over time, one needs repeated time use surveys on the same country over time.

We have found that repeated national time use surveys exist for only seven middle- and lower-

income developing countries: Benin (1998, 2015), Mexico (1996, 1998, 2002, 2009), Mongolia

(2007, 2011), South Africa (2000, 2010), Tanzania (2006 and 2014), Turkey (2006, 2014/15),

and the Occupied Palestinian Territory (1999/2000, 2012/13).

In the absence of time use data over time for given countries, it could be informative to look

at how the share of adults involved in key home production activities changes with the level

of GDP per capita. For example, one could use standard Living Standards and Measurement

Surveys (LSMS) surveys from the World Bank to examine participation in/time spent in specific

home production activities like collecting wood and water. These surveys exist for many more

countries, and are often collected in multiple years (e.g. see Bick et al. (2018) for use of these

types of data).

11

3 Women’s work and structural transformation

The evolution of home production during the structural transformation has important impli-

cations for the economic role of women.12 Traditionally women are the main source of labour

for home production and this tradition persists into the 21st century, where women still spend

three times as many hours on home production as men across the globe. Demands of home

production are the main self-reported barriers for women to enter the labour market despite

most women reporting that they would prefer to be in paid employment.13 These demands

affect the intensive margin of women’s work – how many hours they work in the market each

week – as well as the extensive margin of participation in the market at all. And, if women do

work for the market, the demands of home production may also affect the type of work they

choose: contributing family work without pay, self-employment, or paid work in the informal

or formal sectors?

Available data from around the world, both historically and in present days, show that

when women work in the market, they are much more likely than men to work as unpaid family

workers, or in the informal sector, or as part-time workers. One likely reason for this is that

these employment opportunities, relative to full-time jobs in the formal sector, are likely to be

more compatible with demands from home production.

3.1 Female labour force participation, job type, and market hours

How does women’s involvement in market work change along the development path? A U-

shaped relationship between women’s labour force participation and economic development has

been documented both historically for the US and other developed economies and over large

cross-sections of countries, see (Sinha, 1965; Boserup, 1970; Durand, 1975; Goldin, 1995; Mam-

men and Paxson, 2000; Olivetti, 2014). Durand (1975) is the first comprehensive study, which

reports on a world-wide study of labour force characteristics conducted at the Population Stud-

ies Center at the University of Pennsylvania. This study puts together Census data from 100

countries during 1946-66, covering 72% of the world’s population. Later studies in demography

12See Mincer (1962)’s pioneering work on the role home production for understanding female labour supply. 13See Addati et al. (2018) and footnote 1 and 2 in introduction for data source.

12

and economics extend this work with both historical data and more recent data.

The labour force participation (LFP) rate is the labour force relative to working age popu-

lation, where labour force includes both employed and unemployed. The concept of the labour

force (or the economically active population) aims to include individuals who supply labour

for production of economic goods and services which contribute to national accounts income.

This definition excludes home production from the labour force, as the output of this activ-

ity is for own use. Specifically, the International Labour Organization (ILO) data on labour

force is based on the 1993 International Classification of Status in Employment and it in-

cludes five main categories of jobs: paid employment jobs (employees) and self-employment

jobs (employers, own-account workers, contributing family workers, and members of producers’

cooperatives).

Figure 3 plots the female LFP rate against the level of development for all countries using

OECD and ILO data on labour force participation of women and Penn World Tables data

(version 9.1) for GDP per capita. 14 To summarise the average relationship between GDP

and female LFPR, we estimate regressions of FLFPR on a quadratic in the log of real GDP

per capita, with (dashed blue line) and without (solid blue line) country fixed effects and plot

resulting predictions in the figure.

The intuition behind the U-shape (Sinha, 1965; Boserup, 1970; Durand, 1975) traced out

by the solid blue line (and to a lesser extent, the dashed line) is that as the economy grows,

the farm sector and family-based production on family farms shrinks and households get richer.

Instead of moving into “dirty” factory work (which social norms suggest are not appropriate

for women), families buy back women’s time from market work into the home. As the economy

continues to grow, and as more “pleasant” (and culturally acceptable) service sector jobs expand

and women attain education, the opportunity cost of staying at home rises and women increase

their participation in the market.15

While others have documented a clear picture of the U-shaped female LFP rate using his-

14Labour force participation includes both employed and unemployed individuals. The ILO data are from their modelled estimates series for 1990-2017, and we supplement this with data on OECD countries from 1960-1990.

15Given Figure 3 is for age 15 and above, increases in education leading to women spending more time in school could also contribute to the downward portion of the U-shape. However, as reported in Durand (1975) Table 6.5, the U-shape of female LFP is also present among older age groups, indicating that there are other factors at play in driving up female market work as GDP per capita rises.

13

Figure 3: Female labour force participation and development, all countries

Notes: Female LFP is computed for individuals age 15 and older. OECD country data for 1960-2017 taken from OECD statistics. For all non-OECD countries, ILO data on LFP are for 1990-2017. Real GDP per capita (2011 international dollars) from Penn World Tables v9.1. Country income groups are from World Bank 2020 classifications. Blue lines show predicted values of female LFP from regressions of FLFPR on a quadratic in ln(Real GDP per capita) first excluding (solid blue line) and then controlling for controlling (dashed blue line) for country fixed effects.

torical data for the OECD countries, both across and within countries (e.g. Olivetti (2014)),

the orange and green markers in Figure 3 reveal that the picture for developing countries is still

incomplete. Many of today’s developing countries are on the declining part of the U-shape, a

phase which was completed in developed countries a few decades ago. One important question

for future research is whether developing countries will move onto the increasing part of the

U-shape as they develop.16 Yet another line of research would be in examining changes in the

relationship between home and market work for women in East Asian and Asian economies that

have experienced recent, rapid growth and structural transformation (e.g. Korea, China, Viet-

16The lowest female LFP in Figure 3 are mostly Middle Eastern/North African (MENA) countries where LFP was mostly flat during this period.

14

nam). A deep-dive into any of these economies may tell us more about whether the patterns of

female work through the structural transformation observed in the West were replicated during

these more modern transitions, or not.

Unfortunately, complicating the study of future changes in the female LFP rate in the

developing world is a recent change in how the ILO officially measures labour force participation.

Gaddis et al. (2020) note that in 2013, the ILO adopted a resolution that changed the way in

which individuals qualify to be counted as working. The change, which adjusts a prior definition

that has been stable since the early 1980s, essentially narrows the definition of work to include

only work done for pay or profit, or in the case of farming, based on the self-declared main

intended use of the farm output. To assess the impact of this change in measurement, Gaddis

et al. (2020) implement the new definition for a couple of sub-Saharan African countries with

the right labour force survey data. They show that the new measure substantially understates

employment in rural areas, and especially so for women. This is because women in rural areas

are more likely to be farming for own-use: such activity will be captured in a different category

called “one-use production”. It is not yet clear whether the ILO will continue to produce the

old measures of employment or female LFP rates to maintain continuity of these series over

time. While it should be possible to reconstruct comparable measures of female LFP rates

from original labour force survey data and possibly from some Census data, this change in

measurement raises the costs of conducting research in this space.

3.1.1 Unpaid family work

Although the concept of LFP is well-established, in practice and as we have just noted, mea-

surement can vary and impair the comparability of LFP rates over time and across countries.

This is especially the case for female LFP (see Durand (1975) for an early discussion on all

sources of discrepancies). One of the key difficulties of measuring who is doing market work

is how to treat female household members who live on family farms but who do not report a

non-farm gainful occupation. Some censuses consider them all to be employed in agriculture,

while others count them all as housewives and not in the labour force. Similar discrepancies

also apply to women who are involved in other non-farm family businesses.

As difficult as it is to measure (see Goldin (1995)), unpaid family work on family farms

15

and businesses was an important source of employment for women historically. According to

Ruggles (2015), half of American women aged 18 to 64 participated in unpaid family work in

the 19th century. Such work continues to be important for women in developing countries.

Figure 4 reports the percentage of female population by employment status (including out

of the labour force) and by country income group for the years 1990-2015. Types of employment

(shown as different colors in the figure) include wage work, contributing family workers (e.g.

family farm workers and family business workers), own-account workers, and employers. The

ILO defines contributing family workers as those workers who hold “self-employment jobs” as

own-account workers in a market-oriented establishment operated by a related person living in

the same household.

It is clear from these figures that while unpaid family work (contributing family worker, the

yellow area in each figure) is no longer a major source of female employment in high-income

countries, this type of work still makes up a substantial share of women’s market employment

in developing countries. Figure 4(b) shows that unpaid family work has been falling in middle-

income countries over the last 15 years. But the share of female work in lower-middle and

low-income countries has not shrunk to the same extent (Figures 4(c) and (d)). We also note

that the decline of the share of contributing family workers in the upper-middle income countries

contributes to an overall decline in the female LFPR and an increase in women being out of

the labour force. This same pattern is reported in Ruggles (2015) for the US during the early

part of the 20th century.

Measuring unpaid family work is relevant to documenting female labour force participation

during the process of the structural transformation. Using historical data from seventeen OECD

countries during 1840-2005, Ngai et al. (2020) provide evidence that the declining part of the

U-shape of FLFP is related to falling employment shares in agriculture, and specifically the

decline of family farms. The increasing part of the U-shape is related to the rising employment

share of services in the economy, an important force through which marketisation of home

production occurs.

16

Figure 4: Composition of female work by level of development

(a) High-income countries (b) Upper-middle income countries

(c) Lower-middle income countries (d) Low income countries

Notes: Data are from World Development Indicators: data series for wage work (SL.EMP.WORK.FE.ZS), employer (SL.EMP.MPYR.FE.ZS), own-account workers (SL.EMP.OWAC.FE.ZS), family workers (SL.FAM.WORK.FE.ZS), and self-employed (SL.EMP.SELF.FE.ZS).

3.1.2 Women’s work in the market and home

Looking at female LFP gives us a broad picture of women’s involvement in the labour market

but it misses some key elements in understanding the full picture. Does high female LFP

necessarily mean women devotes a lot of time to market work? Does a low female LFP reflect

that women are enjoying more leisure time? To address these issues we need time use data on

market hours and home production hours so as to get a sense of total working hours.

Using time use surveys for a set of OECD countries, Burda et al. (2013) document that to-

17

Figure 5: Weekly total hours of work (market + home) by gender and level of development

(a) Female total hours: Level (b) Female/Male total hours: Ratio

Notes: Weekly total hours include market and home production work time for women and men aged 15-64.

tal work hours (market hours plus home production hours) are similar across men and women,

especially in non-Catholic countries. They call this the “iso-work phenomenon”. This observa-

tion suggests that in high-income countries it is the distribution of work time across home and

market that lowers market work hours for women.

In lower-income countries, though, as we show in Figure 5, women work more total hours

than men, in some cases many more hours. At a range of different income levels, women report

working more than 40 hours per week across both home and market activities. The ratio of

female to male total hours of home and market work shown in Figure 5(b) is only around 1,

or below 1, for a minority of countries. Combining these figures with the patterns in Figure 1,

it is clear that women’s time in market work is substantially lower than men’s but their time

spent in home production more than exceeds the balance of time men spend in the market and

in home work. This leads to substantially more inequality in weekly total work hours relative

to developed countries.

As GDP per capita rises, though, total time worked by women does fall (Figure 5(a)), sug-

gesting that leisure is increasing with GDP per capita. In addition, growth in national income

is also accompanied by declining gender inequality in total hours of market plus home work.

18

The experience of developed countries therefore suggests that as home production “marketises”

through the structural transformation, inequality in work time across women and men may be

reduced.

In figure 6, we restrict our focus to countries that have both time use survey data from

Bridgman et al. (2018) and FLFP data from the ILO. We plot the FLFP rate and share of

market hours out of all work hours for women, highlighting different country income groups in

different colors. A key takeaway from this figure is that in the set of developing countries in

this sample, many women participate a little bit in the labour market each week. For example,

extensive margin labour market participation is high (over 80%) for women in Tanzania and

Uganda, but the share of total work hours spent in the market is only around 30%. This

low intensive margin activity is probably revealing some frictions that lead to women needing

to combine market work with home work. In some of the poorest countries, it may be more

important to address factors that constrain intensive margin increases in market work time.

Figure 6: Female LFP and distribution of hours in home and market sectors

(a) Female LFP (b) Female market hours/total hours

Notes: Countries included are those with female LFP data from the ILO and time use survey data from Bridgman et al. (2018).

19

3.2 Women’s market work and the service economy

As Reid (1934) observed and Lebergott (1993) documented with consumption expenditure data,

historically many home activities have been transferred to the market through the process of

marketisation.17

“As time went on, one form of production after another, spinning, weaving, sewing,

tailoring, baking, butchering, soap-making, candle-making, brewing, pre-serving,

laundering, dyeing, gardening, care of poultry, and other tasks have wholly or in

part been transferred to commercial production. In addition, child care, education,

and the care of the sick are now to a large extent carried on by paid workers.” (Reid

1934: 47)

This marketisation process is closely linked to the process of structural transformation,

namely the decline in agriculture, hump-shaped manufacturing, and rising services along the

development path. The process of structural transformation is often measured using value-

added shares or employment shares in each sector and is well documented both historically

for developed countries and over large cross-sections of advanced countries (Kuznets, 1966;

Maddison, 1980; Herrendorf et al., 2013).

One way to see the process of marketisation and structural transformation in developing

countries is to look at employment shares by broad sector of work. We zoom in on sub-Saharan

Africa for this exercise, and use a panel data set on sectoral employment shares collected for 11

African countries over forty years in the Gronigen Growth and Development Centre (GDCC)

dataset (Timmer et al., 2015).18

Figure 7 shows that as countries get richer, employment shifts strongly out of agriculture,

into manufacturing and services.19 Rodrik (2015) observes today’s developing countries see

their manufacturing sector shrink at lower levels of GDP per capita than happened in developed

17See Lebergott (1993) Chapter 8 on the transition from home to market using time use data and household expenditure data. He wrote But by 1990 they increasingly bought the goods and services they had produced in 1900. ‘Consumerism’ appeared when housewives began to buy goods they had once produced.

18The specific countries include Botswana, Ethiopia, Ghana, Kenya, Malawi, Mauritius, Nigeria, Senegal, South Africa, Tanzania, and Zambia. Details on data construction are in (Timmer et al., 2015).

19Nigeria’s pattern is a bit of an outlier; this may be related to the fact that it is an oil exporter. Also note that developing countries may experience decline in GDP.

20

Figure 7: Employment shares by level of development: SSA countries 1970-2010

Notes: Employment share data by sector for 11 countries in sub-Saharan Africa from the GDCC 10-sector database (Timmer et al., 2015). Real GDP per capita (2011 international dollars) from Penn World Tables v9.1.

countries historically, a feature termed premature de-industrialisation. This observation implies

an earlier movement from agriculture into services and may indicate a faster marketisation of

home production activities in currently developing countries. What does the shift into services

mean for home production? These typical figures of the structural transformation in a subset

of African countries show a reallocation of labour across sectors as economies get richer but

do not connect back to home production very clearly. Bridgman et al. (2018) draw out the

link between the growth of the service sector and changes in home production over time using

a panel based on 136 time–use surveys from 43 countries over the period 1960–2014. They

show that the growth in market services is associated with a decline in home production hours,

21

especially for women, during a period where most home production activities are services. Yet

as mentioned before, very few of the lower-income countries in their sample have time use data

over time, making it harder to study marketisation and the shift towards services within the

lowest income countries over time.

The literature has also measured the marketisation of home production by looking at the

share of jobs in the market that are doing home-substitutable activities. For example, childcare

facilities, health workers, cleaners, restaurants, hotels, all substitute for home produced services.

As economies grow, more and more jobs tend to be in these sectors, providing market substitutes

for what families used to spend time producing. The evidence on the growth of employment in

these home production substitute sectors (as a subset of the service sector) has mainly come

from developed countries, see for example Addati et al. (2018), but this is something that could

be an area of research in developing countries too.

We illustrate how one could study the extent of marketisation in developing countries by

comparing data on employment in home substitutable sectors in the US with similar measures

from South Africa, Kenya, Ghana, and Ethiopia. Using officially published statistics on market

work by sector in each country for the year 2015, we graph the share of all jobs and the share

of all service sector jobs (for both men and women) that are in home substitute activities in

Figure 8. We define home-substitutable sectors to include jobs in education, health and social

work, arts/entertainment, domestic work for private households, and all other services (e.g.

personal services including food and accommodation, and miscellaneous repairs) that are not

transportation, financial or real estate, professional services, defence, and public or private

administration.20 All service jobs include all home production substitute jobs and all other

service sector jobs not already counted. Public administration and defence are excluded from

both categories.

What we see is that in 2015, Ethiopia, Ghana, and South Africa, a larger share of all jobs

(for men and women) are in home substitute work relative to the US. In all countries, home

substitute jobs represent over 50% of all jobs, and make up the majority of service sector work.

A natural extension of this would be to consider how the share of marketised home production

20We included jobs in wholesale and retail trade as part of the home substitutable jobs because more aggregate categories meant we could not separate out these jobs from repair jobs in a few of the countries.

22

Figure 8: Extent of Marketisation of Home Production

Note: Share of employment (total, or in services) that can be classified as home production substitutes. See text for definition. Data are from official statistics based on national household or labour force surveys.

substitute jobs has changed over time within a set of developing countries for which there are

data.

3.3 Data needs: Measuring marketised home services

In order to document the shift into services in developing countries, and to describe the level

and process of marketisation of home production in developing countries, we need more graphs

like the ones above, for more countries, for more years. We need similar data on employment

shares by narrow sector of work for all lower- and middle-income countries to really see how far

marketisation has proceeded in non-OECD countries.

The GDCC data are excellent for the 11 SSA countries included, and an update and exten-

sion to the dataset to be released in early 2021 will expand the time range and set of African

23

countries included in the sample.21 Ideally, one could build GDCC-like datasets for as many

low- and middle-income countries as possible. Doing this in developing countries would shed

some light on the extent to which a market even exists for the necessary demands of running a

home.

National statistics offices may provide another way to measure employment numbers by

sector in many countries over longer periods of time. Other data sources like Census data could

provide employment by sector information over long periods of time, with information from

labour force surveys filling in the time gaps in more recent decades. World Bank LSMS surveys

may also be useful here.

There are two important challenges in measuring the level of marketisation in developing

countries. One is that many people are employed in the informal sector, and their jobs may

not be captured in official national statistics. The other is that many people may have multiple

jobs, in multiple sectors. Therefore describing the nature of “most” jobs requires a way of

characterising these multiple roles (e.g. choosing a primary job, or using time use data to

allocate time across market activities that are and are not home substitutes). It is reasonable

to think that ignoring these issues is likely to underestimate the extent of marketisation in

developing countries. Note that the GDCC database is built on population census data and

supplemented by nationally representative labour force surveys, making it more likely that the

informal sector work is captured to a better extent than in national statistical office reports on

formal (wage) work.

4 Marketisation and barriers to marketisation

If marketisation of home production provides opportunities for households to buy in home

production services rather than use female labour in the home, and if marketisation provides

job opportunities for women in the market, it is important to understand the forces that trigger

this marketisation. In this section, we discuss factors relevant for marketisation in a frictionless

world, set out some of the barriers that have slowed down the transition of female time from

21Relatedly, the UN has released a dataset called the Expanded Africa Sector Database which contains data for African 18 countries from 1960 to 2015.

24

home to market in developed economies, and discuss the macroeconomic implications and costs

of these barriers.

4.1 Forces that shape Marketisation

What forces are responsible for marketisation of home production? Both technological progress

at home and in the market play a role in altering the relative costs of producing home services

in the household versus in the market.22

When service sectors produce close substitutes to home production (for example, restaurants

that provide alternatives to cooking at home), faster productivity growth in these market

services triggers the process of marketisation: production will be outsourced to the market

(Rogerson, 2008; Ngai and Pissarides, 2008). This marketisation process creates labour demand

for women as these sectors have higher female intensity and at the same time allows those

released from home production to work in sectors that have higher female intensity. As home

production is outsourced, two jobs are created. This process contributes to a rise in female

market work (Ngai and Petrongolo, 2017; Rendall, 2018; Buera et al., 2019).

The development of market services is linked to the process of the structural transformation.

Broadly, market services expand during the process of structural transformation due to price

and income effects (Herrendorf et al., 2013). Price effects are based on the assumption that

services are complements to other consumption goods: a rise in the relative price of services, due

to its slower productivity growth, implies higher expenditure shares on services – the Baumol

effect. Income effects, in contrast, are due to preferences that imply that people have higher

income elasticities for services: higher income implies higher expenditure shares on services –

the Engel curve. 23

Certain triggers for structural transformation may be more or less likely to set in motion

income and price effects that lead to marketised home production. For example: improvements

such as occurred in the Georgian agricultural revolution paved the way for the industrial rev-

22See Gronau (1997) for a survey on the theory of home production. 23As noted by Kongsamut et al. (2001) income effects for services can also be related to the presence of home

production. Specifically, if home produced output is a good substitute for market services, then higher income is associated with a higher degree of marketisation, i.e. more home production is outsourced to the service sectors (Moro et al., 2017; Ngai and Petrongolo, 2017).

25

olution in 18th century England; or in India in the 1960s, the Green revolution led to large

increases in agricultural productivity. This type of productivity growth is sometimes referred

as the “push” mechanism that releases workers out of agriculture, and works to expand market

services through both the Baumol effect and the Engel Curve, once subsistence consumption

in agriculture is satisfied. 24.

International trade is another important trigger for structural transformation in developing

countries, affecting employment though productivity growth and/or through income growth.

If opening to trade grows the economy, this can lead to a higher demand for services (income

effect). If opening to trade generates productivity growth outside of the home sector, this can

also marketise home production (price effect). Trade is potentially a force for shifting time out

of the home and into the market at lower levels of GDP than happened in developed countries.

Another technological channel for marketising home production time is the falling relative

price of household durables, induced by faster productivity growth in sectors that produced

these durables. Greenwood et al. (2005) refer to this as “engines of liberation” releasing female

time from home production to market. Cavalcanti and Tavares (2008) provide some support

for this view by documenting a negative relationship between the relative price of household

appliances and female labour force participation, using a panel of ten OECD countries during

the period 1975-1999. An alternative view based on historical data for the US is that the

availability of household durables may create more work for women – with clothes being washed

more often and better meals being served (Mokyr, 2000; Vanek, 1973).25

4.2 Frictions slowing women’s transition from home to market

Around the world, developing countries continue to experience technological progress in agricul-

ture, opening borders to trade, and improved home production technologies, all of which could

set in motion structural shifts in the labour market. Against this backdrop, various frictions

may get in the way of women transitioning smoothly from home to market.

24Rapidly increasing agricultural productivity played an important role in driving the decline in family farms, and hence to the declining part of the U-shape female labour force participation(Mukhopadhyay, 1994; Ngai et al., 2020)

25See also Buera and Kaboski (2012) for the argument that more affordable household durables may lead to service activities being brought back into home production, for example the shift from using the laundromat to the washing machine at home.

26

We have seen in previous sections that home production easily represents at least a part-

time job, and at most more than a full-time job, in different countries. The requirements of

childcare and elder-care place formidable obstacles in the way of women’s work, in developing

and developed countries alike. As long as women are primarily responsible for child- and elder-

care – because there is no market for care work, or because these market services are too

expensive, or because there is a cultural norm against outsourcing this work to the market

– they will only be able to work on an intermittent basis, in activities that can be stopped

and started. For care work to get done, women need flexible, part-time work; or households

and families need to restructure; or there needs to be an infrastructure for childcare/elder-care

outside of the home (a marketised care sector).

As we noted in Figure 4, a large share of women’s work is in unpaid family work. Historically,

this was also the case in the US. However, as family farms and businesses disappeared in

developed countries, it became more difficult for women to combine home and market work.

The shrinking of work on family farms coincided with a return of women to the home in the

US. In developing countries, women may continue to need to work in family businesses, or as

self-employed workers in the informal sector, to cope with a schedule that facilitates childcare.

In addition, one of the major barriers to women moving their time into the market is the

social norm regarding women’s role in home production, especially for married women (Goldin,

2014; Erosa et al., 2020; Lee, 2020a). A traditional gendered view of time use in the family

is that women are mainly responsible for taking care of the household while men are the

main bread-winners in the market.26 Globally, men and women in several parts of the world

continue to hold strong beliefs about whether women can work outside of the home, and in what

industries women can work (Addati et al., 2018). One strand of work compares immigrants

from different origins (e.g. Fernandez and Fogli (2009); Alesina et al. (2013)) to show that

culture from their country of origin persists and has a large influence on female decision to

work in the market. Alesina et al. (2013) argue that these norms have deep roots linked to

gender-specific comparative advantage in agriculture production, providing empirical evidence

for Boserup’s thesis that male advantage in the labour market stems from male advantage in

driving the plough in plough-based cultivation (Boserup, 1970).

26See Fernandez (2009) for a review on the effect of culture on female labour supply.

27

In addition to norms and culture, legal constraints on the types of market work women can do

have been important in developed countries (e.g. marriage bars in rich countries in the past).

Today, legal restrictions constrain women’s work in certain sectors of developing countries:

Hyland et al. (2020) show that in the average country, women have three quarters the rights of

men in the workplace. Many of these restrictions are similar in flavor to restrictions on women

in the 19th century in rich countries: protection of weaker workers, especially around childbirth

time. Yet some restrictions are broad, covering all non-pregnant, non-nursing women and

many sectors of the economy. Figure 9 shows for a range of industries, the share of developing

countries in which women cannot work to the same extent as men. These data are taken from

the World Bank’s database on Women’s Work and Law. Gender-based restrictions on work are

most prevalent, although perhaps least binding, in the mining sector, where female employment

is typically low even in unconstrained settings. More surprisingly, 25% of developing countries

restrict women’s work in factories and around 20% do in construction.

4.3 Implications of women’s choices

When women face barriers to transition from home to market, or barriers to transitioning across

sectors during the structural transformation, there are macroeconomic consequences from this

misallocation of time and talent. Barriers to women’s transition from home to market generates

misallocation of talent, not just across home and market but also within occupations in the

market. Using data for the US, Hsieh et al. (2019) show that the decline in these barriers

between 1960 and 2010 has contributed to an improvement in the allocation of talent and

contributed significantly to economic growth in terms of market output and increasing total

output.

When social norms constrain women’s choices, requiring women to be the primary house-

keepers and unpaid caregivers, this imposes a constraint on women’s time allocation, and may

induce women to choose occupations with more flexible or shorter working hours and more

generous parental leave instead of following their comparative advantage. Preferences for these

types of job amenities have been shown to be important for women’s choice of which firms to

work in, using employer-employee matched data from Finland Xiao (2020) and Brazil Morchio

and Moser (2020). This home production constraint, affecting women’s choice in occupation

28

Figure 9: Share of developing countries where women face employment restrictions

Note: Graph shows for each sector of employment, the share of countries in which non-pregnant, non-nursing women cannot work in the same way as men can. Data are from the World Bank’s Women’s Work and the Law database. Means for developing countries are shown.

and firms, is also reflected in the gender pay gap (Goldin, 2014; Erosa et al., 2020; Xiao, 2020;

Morchio and Moser, 2020).

The importance of this form of barrier is also reflected in the fact that women tend to

work on family farms during the early development stage as this allows women to combine

home production and market production more easily. But as shown by (Lee, 2020b), using a

sample of 66 countries, this form of misallocation of female talent across sectors contributed

significantly to the relatively low agricultural productivity in developing countries. With the

process of structural transformation and the modernisation of agricultural sector, women’s work

on family farms is declining over time. The choice of working in the market now means leaving

home, making it incompatible to combine home production and work.

As shown in national time surveys across the globe, and discussed previously, a key com-

29

Figure 10: Total fertility rate by region and year

ponent associated with the home production time constraint on women is getting married and

having children. As women’s education increases around the globe and the improvement in

awareness in gender equality in terms of work at home, market, and pay, it is not surprising to

see fewer women choosing to get married, have children, or have as many children. Figure 10

shows the ongoing decline in the number of live births per women, using data from the United

Nations. According to the UN forecast, world population growth will be zero by 2100. This

will have important implications for the future of economic growth. As shown by Jones (2020),

humans are needed for knowledge production, since this is the driver of technological progress

and an engine of growth. Negative population growth implies a gradual disappearance of the

human race and the end of economic growth.

The challenge ahead is therefore to provide an environment, perhaps by removing some

of the existing barriers to combining childcare and market work, so that women can allocate

their talents according to their comparative advantage with a solution for the home production

requirement. The development of the service economy can facilitate the process of marketisation

of home production, and social norms regarding women’s role in home production also have the

potential to change in response to a changing economic environment.

30

5 Policies tackling barriers to women’s market work:

Evidence and knowledge gaps

In developing countries, there are many frictions we do not fully understand that keep women’s

market work low on either the extensive (participation) or intensive (hours) margins, and con-

tribute to keeping women’s overall time spent in home and market work higher than that of

men. These barriers likely contribute to misallocation in the labour market, and to large gender

gaps in labour market outcomes like hours of work, labour market participation, and wages.

How can developing countries address these frictions to promote growth and gender equality?

How can these countries position themselves well to facilitate a shift in women’s time towards

their highest productivity sector – whether that is home or market – once technology or trade-

induced structural transformation occurs? In this last section, we outline some broad areas for

future research on these questions. We cover policy areas that address frictions on the home

production side (public infrastructure, provision for child- and elder-care), and on the market

production side (addressing gender-based social norms and laws, and supporting productivity

improvements in part-time work and informal sector employment).

5.1 Labour productivity in the home sector: Public infrastructure

Infrastructure seems particularly important for developing countries. Reid (1934) made the

point forcefully about how broad economic changes happened when industrialisation began,

and where market productivity took off relative to home productivity:

After 1800 economic conditions changed rapidly. Roads improved steadily. Trade

increased. Modern inventions made the most efficient tools too expensive for small-

scale household use. Steam power possible only for centralised industries brought

about the withdrawal of much manufacturing from the home.Reid (1934, p. 45)

While developing countries may experience global trade (for example) as a force that ac-

celerates the shift to the market, the lack of industrial structure (roads, energy, water and

sewage systems) slows it down. The lack of public infrastructure is likely to create a brake on

marketisation. We show in Figure 11 that in many developing countries, large fractions of men

31

and women spend a considerable amount of time each week just fetching wood and/or water

for their homes.27

Figure 11: Collection of wood and water by GDP per capita

(a) Any collection (b) Time spent each week

Note: Data are from Bick et al. (2018)

What do we know about whether new access to basic utilities at the household-level releases

female time from home production or whether it translates into more market work for women?

Some direct evidence from randomised trials in urban Morocco (Devoto et al., 2012) and rural

Kenya (Kremer et al., 2011) shows that having better access to clean water reduced the time

spent in home work. But none of the released time translated into more market work. In

Kyrgyzstan, however, Meeks (2017) shows that new water infrastructure substantially reduces

time spent in home production, and that some of this time does move into the market, with the

rest taken as leisure. Dinkelman (2011) shows that grid electrification in South Africa changed

the nature of home production in rural areas. Newly electrified homes relied more on electrical

appliances for lighting, cooking, and heating, and new access to electrification also increased net

female employment. In that paper, evidence pointed towards a net increase in labour supply

driving increased female market work. Yet electricity could also change the demand for female

labour. In the historical US Vidart (2020) shows that electrification in the late 1800s and early

27The data for this figure are from Bick et al. (2018). Information on wood and water collection is assembled from various labour force surveys.

32

1900s affected female LFP by raising the return to hiring skilled workers into new types of jobs,

drawing women out of the home. In turn, these new jobs raised the return to education, leading

women to invest in more education over eight decades. This link between infrastructure and

investment in human capital, and then to employment in higher-skilled sectors – part of the

structural transformation – is an under-explored area in developing countries.

Some researchers have found small to no impacts of access to basic utilities on time spent in

market work (e.g. Lee et al. (2020)). One possibility is that there are complementary constraints

faced by women involved in home production. To release female time into the market, a number

of home production constraints may need to be lifted at the same time: for example, access to

efficient cookstoves may not reduce female time in home production if someone still needs to

take care of children. This point naturally leads to a consideration of another key constraint

on how women can use their time: the need for safe, quality child-care (or elder-care).

5.2 Supporting home-substitute market sectors: Care work

What types of policies might developing countries look to support the market provision of

child- and elder-care? Olivetti and Petrongolo (2017) provide a overview of the effects of family

policies put in place in many high-income countries.28 These policies target gender equity,

higher fertility, and child development. Their main findings are that the range of family-friendly

policies put in place in these countries have small to no effect on gender gaps in employment

and wages, but large positive effects on children. They note: “The one policy that across the

board improves gender outcomes (wage gaps, employment gaps, labour force participation, and

fertility) is spending on early childhood care and education”.

As we show in figure 10, fertility rates are falling everywhere, and also in developing coun-

tries. Declining fertility could release some of the frictions on women’s market work that are

related to the demands of having more children, although not the time demands for increasing

quality of investments in children. Even if labour market participation is more feasible for

women having fewer children, the presence of especially young children affects intensive margin

participation and may distort the type of work that women choose or are constrained to choose.

Working when children are young requires such work to be compatible with interruptions, have

28See for examples, Cascio (2009); Baker et al. (2008); Duval-Hernandez and Ngai (2018); Guner et al. (2020).

33

extreme flexibility, and be easy to access from a home base. Ironically, outside of subsistence

agriculture, these job characteristics almost perfectly describe self-employment in home-based

businesses in the informal sector. The majority of women working in developing countries are

already in these types of jobs (see section Figure 4).

Is the prevalence of self-employment among women in developing countries partly related

to dealing with young kids? A nice paper by Heath (2017) suggests some nuance in this

relationship. Using an urban Ghanaian household panel survey from 2004-2013, she shows

that some women drop out of market work when they have children while others increase their

market work time, working more in self-employment. Those who drop out have no other family

members who can help with child-care; those who remain and work more rely on older female

relatives or elder children for child care support. This paper suggests that household formation

is an important margin of adjustment and support system for working mothers. It also implies

that childcare constraints affect the market work of three generations of women – mothers,

grandmothers, and older daughters.

Can childcare subsidies work to promote women’s market work in developing countries?

Possibly. In urban Kenya, Clark et al. (2019) use a randomised controlled trial to show that

subsidised access to private childcare raises female employment by 8.5 percentage points. The

subsidy worked to raise the LFP of married women, and shift the type of work of unmarried

mothers to more regular/inflexible hour jobs. In this study, there was already a booming private

sector market for childcare. More evidence similar to these two cases, from other settings, would

be extremely useful in shedding light on how households cope with the arrival of young children

and which women are most constrained by the requirements of childcare.

In general, much more research is needed into whether marketising childcare (or eldercare)

could be an important avenue for generating female jobs and releasing female time to look for

market work in other sectors.

5.3 Shifting social norms and reducing legal barriers

Norms against women’s market work that operate on both the demand and supply side of the

labour market may be difficult to shift, but there is some evidence this is possible. Where social

norms against women working are particularly strong, for example in some Middle Eastern,

34

North African, and South East Asian countries, policy interventions may help.

Bernhardt et al. (2018) use an experiment in rural India to show that the perceived social

cost of women’s work falls on men and that husbands’ opposition to female labour is associated

with their wives’ lower take-up of employment. Addressing these perceived costs can cause

norms to shift in a very short period of time. Bursztyn et al. (2018) show that in Saudi Arabia,

a divergence between privately held opinions and publicly accepted norms about women working

outside the home may generate a friction that keeps women trapped at home. In their setting,

only one in five women above 15 years is working at all, and an even smaller share is working

outside of the home. In two experiments, they randomly correct both male and female and

beliefs about what others believe about the place of women outside of the home. They find

that this increases married men’s willingness to help their wives search for jobs, as measured

by their costly sign-up for a job-matching service for their wives. And they also show that this

information intervention increases the chances that women switch from an at-home temporary

enumerator job to a higher-paying, outside-the-home version of the job. More broadly, exposure

to other ways of life through cable television or through direct information provision may also

reduce the norm-based barriers to women’s work (e.g. Jensen (2012); Jensen and Oster (2009)).

Aside from operating directly on preferences, policy may be targeted at specific technical

solutions that can improve outcomes even in the presence of strongly persistent social norms.

For example, work by Kondylis et al. (2020) in Brazil and ongoing studies by some of the

same research team in Pakistan and in Saudia Arabia investigate the potential that providing

safe transportation – female-only bus transit to work – could have for addressing norm-based

constraints on women’s mobility outside of the home. Results from Brazil suggest that women

value these safe methods of getting to work. These studies will tell us more about whether

such transportation interventions could enable some women to shift work into the market in

the short-run, potentially feeding back into changing social norms about women working.

On the demand-side of the labour market, it is relevant to consider legal barriers to women

participating fully in the labour market, and legal protections specific to female-intensive in-

dustries. A handful of papers consider how minimum wage legislation can affect conditions of

women’s work in female-intensive market sectors (e.g. see (Dinkelman and Ranchhod, 2012) for

the case of domestic workers in South Africa McKenzie et al. (2014) for migrant workers from

35

the Philippines). Whether legal barriers to women working in certain industries, or a lack of

job protections in female-intensive sectors leads to misallocation at the macro-level is an open

area for research.

5.4 Creating jobs with home production in mind

Randomised controlled trials in labour markets of developing countries are a relatively new and

active area of research. Many randomised controlled trials have identified important frictions

in the informal and formal sectors of developing countries that inhibit job creation, and job

matching. Some of these frictions include: lack of access to training, to capital and assets, to

cash for job search, lack of access to credit, lack of information about relevant skills and jobs,

or a lack of information about worker characteristics. A variety of treatments have been able to

overcome these frictions and promote employment, matching, and small business performance;

but many of them have not been successful.

Direct access to new jobs matters for women’s market work in the short-run. For example:

Blattman and Dercon (2018) show that providing access to factory jobs improves employment in

the short-run, especially among women – but these effects die out one year, partly because of the

health-risks associated with these jobs (relative to self-employment/entrepreneurship). Groh et

al. (2016) show that providing wage subsidies to job applicants improves female employment

rates in Jordan, with large employment gains in the short run but no impacts on employment

in the longer-run once the subsidy ends.

Asset transfers and capital infusions for entrepreneurs have mixed results for employment.

Bandiera et al. (2017) show that large asset transfers and training programmes in village

economies in Bangladesh improve female labour market outcomes among for the poorest women

over time. On the other hand, capital infusions into small businesses in Sri Lanka (de Mel et

al., 2009) did not improve business outcomes for female entrepreneurs. Recent research in Sri

Lanka and Ghana suggests that this may be because capital provided for female businesses gets

diverted into higher productivity male businesses within the same household (Bernhardt et al.,

2019).

The role of information in inhibiting job creation and job matches seems to be important.

Carranza et al. (2020) randomly provide information to job seekers about their skills and to

36

firms about job seekers in South Africa; they find that two-sided frictions matter and reducing

these frictions improves job matches.29 Cash is also important for finding work: Franklin (2018)

shows that providing transport subsidies in Ethiopia allows unemployed youth to wait longer

to find better jobs, making more productive matches.

Many of these papers focus on women, or examine differential effects for men and women,

sometimes (although not always) finding large gender differences. Yet none of them explicitly

links the effects on women’s market work back to home production, or considers whether the

demands of home production are the reason for gender heterogeneity in treatment effects.

There are at least two ways in which these types of experiments could be used to shed more

light on the gendered nature of employment through the structural transformation. First, they

could be used to show that solving a specific job market friction for women (e.g. providing

access to capital) actually increases output at the aggregate level rather than only improving

welfare for the targeted individual. This would be evidence of misallocation of workers.30

Second, researchers could design interventions that combine access to market work for women

with interventions that affect the time constraints on home production at the household-level.

This could work either by changing household productivity directly (e.g. investments that

save female time in home production directly) or by stimulating the marketisation of home

production services. Time use survey data would be particularly relevant here for designing

the specific times of interventions on the household production side, for measuring the impact

of marketised home production services on time allocations, and for evaluating the impacts of

the various labour market interventions.

6 Conclusion

This paper reviews the literature and evidence on how linkages between women’s home and

market work change through the structural transformation. We relate historical shifts in female

time between home production and market work in developed economies to more recent changes

29In related work, Wheeler et al. (2020) show that training work seekers to use job search platforms also improves employment outcomes in the short-run.

30Bandiera et al. (2017) get at this somewhat by showing that the asset transfers did not hurt the employment prospects of non-treated individuals in the same villages.

37

in developing countries, showing that the picture for developing countries is far from complete.

We highlighted the forces that contribute to shifting women’s time into the market in developed

countries, and the barriers that get in the way of this movement of time.

Women in developing countries face many frictions to participating in the labour market

that we do not yet fully understand. We reviewed available and required data that could

be used to uncover some of the frictions retarding the movement of women’s time into the

market in developing countries. The relevance of these frictions is important not just at the

individual level, but at the macroeconomic level. Misallocation of workers to sectors in which

they do not have comparative advantage leads to economic losses which could be substantial.

Labour market frictions that are gender-specific also contribute to continued gender inequality

in economic outcomes.

We propose several areas for future policy-relevant research addressing these frictions are

proposed. On the data side, we need more time use data for more developing countries over

more years to understand how women spend their time in home production and in the market.

We also need more research on how access to infrastructure affects home labour productivity;

how child-care services and part-time or flexible work arrangements facilitate market work; and

whether social norms against women’s market work are malleable.

Although Reid (1934) talked about how work-time flexibility and availability of part-time

work were the main concerns facing US households in the 1930s, recent news articles reveal

that these concerns are still present: women are still struggling with balancing home produc-

tion and employment, in most countries. This balancing act has been made more apparent

through the experiences of the COVID crisis in 2020. The challenge ahead is to provide an

environment, perhaps by removing some existing barriers, so that women can allocate their

talent according to their comparative advantage with a technological or market-based solution

for the home production requirement. The development of the service economy can facilitate

the process of marketisation of home production, while social norms regarding women’s role of

home production may also change in response to a changing environment.

38

References

Addati, Laura, Umberto Cattaneo, Valeria Esquivel, and Isabel Valarino, “Care

work and care jobs: For the future of decent work,” Technical Report, International Labour

Organization 2018.

Alesina, Alberto, Paola Giuliano, and Nathan Nunn, “On the origins of gender roles:

Women and the plough,” Quarterly Journal of Economics, 2013, 128 (2), 469–530.

Alon, Titon, Matthias Doepke, Jane Olmstead-Rumsey, and Michèle Tertilt, “This

time it’s different: The Role of Women’s Employment in a Pandemic Recession,” 2020.

Baker, Michael, Jonathan Gruber, and Kevin Milligan, “Universal Child Care, Ma-

ternal Labor Supply, and Family Well-Being,” Journal of Political Economy, 2008, 116 (4),

709–745.

Bandiera, Oriana, Robin Burgess, Narayan Das, Selim Gulesci, Imran Rasul, and

Munshi Sulaiman, “Labor Markets and Poverty in Village Economies,” The Quarterly

Journal of Economics, 2017, 132 (2), 811–870.

Bernhardt, Arielle, Erica Field, Rohini Pande, and Natalia Rigol, “Household Matters:

Revisiting the Returns to Capital among Female Microentrepreneurs,” American Economic

Review: Insights, 2019, 1 (2), 141–60.

, , , , Simone Schaner, and Charity Troyer-Moore, “Male Social Status and

Women’s Work,” AEA Papers and Proceedings, 2018, 108, 363–67.

Bick, Alexander, Nicola Fuchs-Schündeln, and David Lagakos, “How Do Hours Worked

Vary with Income? Cross-Country Evidence and Implications,” American Economic Review,

2018, 108 (1), 170–99.

Blattman, Christopher and Stefan Dercon, “The Impacts of Industrial and En-

trepreneurial Work on Income and Health: Experimental Evidence from Ethiopia,” American

Economic Journal: Applied Economics, 2018, 10 (3), 1–38.

39

Boserup, E., Woman’s Role in Economic Development, London: George Allen and Unwin

Ltd., 1970.

Bridgman, Benjamin, Georg Duernecker, and Berthold Herrendorf, “Structural

Transformation, Marketization, and Household Production around the World,” Journal of

Development Economics,, 2018, 133, 102–126.

Bridgman, Benjamin R., “Home productivity,” Journal of Economic Dynamics and Control,

2016, 71 (C), 60–76.

Buera, Francisco and Joe Kaboski, “The Rise of the Service Economy,” American Eco-

nomic Review, 2012, 102, 2540–69.

, , and Kent Zhao, “The Rise of Services: The Role of Skills, Scale, and Female Labor

Supply,” Journal of Human Capital,, 2019, 13, 157–187.

Burda, Michael, Daniel Hamermesh, and Phillipe Weil., ““Total Work and Gender:

Facts and Possible Explanations,” Journal of Population Economics, 2013, 26, 239–261.

Bursztyn, Leonardo, Alessandra L. González, and David Yanagizawa-Drott, “Mis-

perceived Social Norms: Female Labor Force Participation in Saudi Arabia,” Working Paper

24736, National Bureau of Economic Research 2018.

Carranza, Eliana, Robert Garlick, Kate Orkin, and Neil Rankin, “ LinkedIn(to) Job

Opportunities: Experimental Evidence from Job Readiness Training,” 2020.

Cascio, Elizabeth U., “Maternal Labor Supply and the Introduction of Kindergartens into

American Public Schools,” Journal of Human Resources, 2009, 44 (1), 140–170.

Clark, Shelley, Caroline W. Kabiru, Sonia Laszlo, and Stella Muthuri, “The Impact

of Childcare on Poor Urban Women’s Economic Empowerment in Africa,” Demography, 2019,

56, 1247–1272.

de Mel, Suresh, David McKenzie, and Christopher Woodruff, “Are Women More

Credit Constrained? Experimental Evidence on Gender and Microenterprise Returns,”

American Economic Journal: Applied Economics, 2009, 1 (3), 1–32.

40

de V. Cavalcanti, Tiago V. and José Tavares, “Assessing the “Engines of Liberation”:

Home Appliances and Female Labor Force Participation,” The Review of Economics and

Statistics, 2008, 90 (1), 81–88.

Devoto, Florencia, Esther Duflo, Pascaline Dupas, William Parienté, and Vincent

Pons, “Happiness on Tap: Piped Water Adoption in Urban Morocco,” American Economic

Journal: Economic Policy, 2012, 4 (4), 68–99.

Dinkelman, Taryn., “The effects of rural electrification on employment: New evidence from

South Africa,” American Economic Review, 2011, 101 (7), 3078–3108.

Dinkelman, Taryn and Vimal Ranchhod, “Evidence on the impact of minimum wage

laws in an informal sector: Domestic workers in South Africa,” Journal of Development

Economics, 2012, 99 (1), 27–45.

Durand, John, The labor force in economic development: A comparison of international

census data, 1946-66., Princeton: Princeton University Press, 1975.

Edmonds, Eric, “Does child labor decline with improving economic status?,” Journal of

Human Resources, 2005, XL (1), 77–99.

, Nina Pavcnik, and Petra Topalova, “Child labor and schooling in a globalizing world:

Some evidence from urban India,” Journal of the European Economic Association, 2009, 7

(2-3), 498–507.

Erosa, Andrés, Luisa Fuster, Gueorgui Kambourov, and Richard Rogerson, “Hours,

Ocupations and Gender Differences in Labor Market Outcomes,” 2020.

Fernandez, Raquel, “Women, Work, and Culture,” Journal of the European Economic As-

sociation, 2009, 5, 305–322.

and Alessandra Fogli, “Culture: An Empirical Investigation of Beliefs, Work, and Fertil-

ity,” American Economic Journal: Macroeconomics, 2009, 1 (1), 146–177.

Franklin, Simon, “Location, Search Costs and Youth Unemployment: Experimental Evidence

from Transport Subsidies,” The Economic Journal, 2018.

41

Freeman, Richard and Ronald Schettkat., “Marketization of Household Production and

the EU-US Gap in Work,” Economic Policy, 2005, 41, 5–39.

Gaddis, Isis, Gbemisola Oseni, Amparo Palacios-Lopez, and Janneke Pieters, “Who

is employed? Evidence from Sub-Saharan Africa on Redefining Employment,” 2020. World

Bank Polcicy Research Working Paper No. 9370.

Georgieva, Kristalina, Stefania Fabrizio, Cheng Hoon Lim, and Marina M. Tavares,

“The Covid-19 Gender Gap,” https://blogs.imf.org/2020/07/21/the-covid-19-gender-gap/

2020. IMF Blog.

Goldin, Claudia, “The U-Shaped Female Labor Force Function in Economic Development

and Economic History,” in T. P. Schultz, ed., Investment in Women’s Human Capital and

Economic Development, Chicago: University of Chicago Press, 1995, pp. 61–90.

, “A Grand Gender Convergence: Its Last Chapter,” American Economic Review, 2014, 104

(4), 1091–1119.

Greenwood, Jeremy, Ananth Seshadri, and Mehmet Yorukoglu, “Engines of Libera-

tion,” Review of Economic Studies, 2005, 72, 109–133.

Groh, Matthew, Nandini Krishnan, David McKenzie, and Tara Vishwanath, “Do

wage subsidies provide a stepping-stone to employment for recent college graduates? Evidence

from a randomized experiment in Jordan,” The Review of Economics and Statistics, 2016,

98 (3), 488–502.

Gronau, Reuben, “The theory of home production: The past ten years,” Journal of Labor

Economics, 1997, 15, 197–205.

Guner, Nezih, Remzi Kaygusuz, and Gustavo Ventura, “Child-Related Transfers,

Household Labor Supply and Welfare,” Review of Economic Studies, 2020, 87, 2296–2321.

Heath, Rachel, “Fertility at Work: Children and Women’s Labor Market Outcomes in Urban

Ghana,” Journal of Development Economics, 2017, 126, 190–214.

42

Herrendorf, Berthold, Richard Rogerson, and Ákos Valentinyi, “Growth and Struc-

tural Transformation,” in P. Aghion and S. Durlauf, eds., Handbook of Economic Growth,

Amsterdam: Elsevier, 2013, pp. 855–941.

Hsieh, Chang-Tai, Erik Hurst, Charles I. Jones, and Peter J. Klenow, “The Allocation

of Talent and U.S. Economic Growth,” Econometrica, 2019, 87 (5).

Hyland, Marie, Simeon Djankov, and Pinelopi Koujianou Goldberg, “Gendered Laws

and Women in the Workforce,” American Economic Review: Insights, 2020, 2 (4), 475–90.

Jensen, Robert, “Do labor market opportunities affect young women’s work and family de-

cisions? Experimental evidence from India,” Quarterly Journal of Economics, 2012, 127 (2),

753–792.

and Emily Oster, “The Power of TV: Cable Television and Women’s Status in India,”

The Quarterly Journal of Economics, 2009, 124 (3), 1057–1094.

Jones, Charles I., “The End of Economic Growth? Unintended Consequences of a Declining

Population,” 2020.

Kondylis, Florence, Arianna Legovini, Kate Vyborny, A. M. T. Zwager, and Luiza

De Andrade Cardoso, “Demand for Safe Spaces : Avoiding Harassment and Stigma,”

World Bank Policy Research working paper s, 2020, (9269).

Kongsamut, Piyabha, Sergio Rebelo, and Danyang Xie, “Beyond Balanced Growth,”

Review of Economic Studies, 2001, 68, 869–82.

Kremer, Michael, Jessica Leino, Edward Miguel, and Alix Peterson Zwane, “Spring

Cleaning: Rural Water Impacts, Valuation, and Property Rights Institutions,” The Quarterly

Journal of Economics, 2011, 126 (1), 145–205.

Kuznets, Simon, Modern Economic Growth, Rate, Structure and Spread, New Haven: Yale

University Press, 1966.

Lebergott, Stanley, Pursuing Happiness, Princeton, NJ: Princeton University Press, 1993.

43

Lee, Jay Euijung, “Marriage and Misallocation: Evidence from 70 Years of U.S. History,”

2020.

Lee, Kenneth, Edward Miguel, and Catherine Wolfram, “Experimental Evidence on

the Economics of Rural Electrification,” Journal of Political Economy, 2020, 128 (4).

Lee, Munseob, “Allocation of Female Talent and Cross-Country Productivity Differences,”

2020.

Leeds, J., “The household budget: With a special inquiry into the amount and value of

household work.” PhD dissertation, Columbia University 1917.

Maddison, Angus, “Economic Growth and Structural Change in the Advanced Countries,”

in Irving Leveson and Jimmy W. Wheeler, eds., Western Economies in Transition: Structural

Change and Adjustment Policies in Industrial Countries, London: Croom Helm, 1980, pp. 41–

65.

Mammen, Kristin and Christina Paxson, “Women’s Work and Economic Development,”

Journal of Economic Perspectives, 2000, 14, 141–164.

McKenzie, David, Caroline Theoharides, and Dean Yang, “Distortions in the Inter-

national Migrant Labor Market: Evidence from Filipino Migration and Wage Responses to

Destination Country Economic Shocks,” American Economic Journal: Applied Economics,

2014, 6 (2), 49–75.

Meeks, Robyn, “Water Works: The Economic Impact of Water Infrastructure,” Journal of

Human Resources, 2017, 52 (4), 1119–1153.

Mincer, J., “Labor Force Participation of Married Women: A Study in Labor Supply.,”

in “National Bureau of Economic Research, Aspects of Labor Economics,” Princeton N.J.:

NBER, 1962, pp. 63–106.

Mokyr, Joel, “Why more work for mother? Knowledge and the household behavior,” Journal

of Economic History, 2000, 60, 1–40.

44

Morchio, Iacopo and Christian Moser, “The Gender Pay Gap: Micro Sources and Macro

Consequences,” 2020.

Moro, Alessio, Solmaz Moslehi, and Satonshi Tanaka, “Does Home Production Drive

Structural Transformation,” American Economic Journal: Macroeconomics, 2017, 90.

Mukhopadhyay, Sudhin K., “Adapting Household Behavior to Agricultural Technology in

West Bengal, India: Wage Labor, Fertility, and Child Schooling Determinants,” Economic

Development and Cultural Change, 1994, 43 (1), 91–115.

Ngai, L. Rachel and Barbara Petrongolo, “Gender Gaps and the Rise of the Service

Economy,” American Economic Journal – Macroeconomics, 2017, 9, 1–44.

and Christopher A. Pissarides, “Trends in Hours and Economic Growth,” Review of

Economic Dynamics, 2008, 11, 239–256.

, Claudia Olivetti, and Barbara Petrongolo, “Structural Transformation and the U-

shaped Female Labor Employment,” 2020.

Olivetti, C. and B. Petrongolo, “The economic consequences of family policies. Drawing

lessons from a century of legislation across OECD countries,” Journal of Economic Perspec-

tives, 2017, 31, 205–230.

Olivetti, Claudia, “The Female Labor Force and Long-run Development: The American

Experience in Comparative Perspective,” in C. Frydman L. Platt Boustan and R.A. Margo,

eds., Human Capital in History: The American Record, Chicago: University of Chicago Press,

2014, chapter 5.

Ramey, Valerie A., “Time Spent in Home Production in the Twentieth-Century United

States: New Estimates from Old Data,” The Journal of Economic History, 2009, 69 (1),

1–47.

Reid, Margaret, Economics of Household Production, New York: J. Wiley Sons, Incorporated,

1934.

45

Rendall, Michelle, “Female Market Work, Tax Regimes, and the Rise of the Service Sector,”

Review of Economic Dynamics, 2018, 28 (2), 269–289.

Robert, Lei Fang Duval-Hernandez and L. Rachel Ngai, “Social Subsidies and Marke-

tization – the Role of Gender and Skill,” 2018. Center for Macroeconomics DP 2018-04.

Rodrik, Dani, “Premature Deindustrialization,” Journal of Economic Growth, 2015, 21, 1–33.

Rogerson, Richard, “Structural Transformation and the Deterioration of European Labor

Market Outcomes,” Journal of Political Economy, 2008, 116 (2), 235–259.

Ruggles, Steven, “Patriarchy, Power, and Pay: The Transformation of American Families,

1800–2015,” Demography, 2015, 52, 1797–1823.

Sinha, J. N. (1965), “Dynamics of Female Participation in Economic Activity in a Devel-

oping Economy,” in “Proceedings of the World Population Conference,” New York: UN

Publications, 1965.

Timmer, Marcel P., Gaaitzen de Vries, and Klaas de Vries, “Patterns of Structural

Change in Developing Countries,” in J. Weiss and M. Tribe, eds., Routledge Handbook of

Industry and Development, Routledge, 2015, pp. 65–83.

Vanek, J., “Keeping busy: Time spent in housework, United States, 1920-1970.” PhD disser-

tation, University of Michigan 1973.

Vidart, Daniela, “Human Capital, Female Employment, and Electricity: Evidence from the

Early 20th Century United States,” 2020.

Wheeler, Laurel, Robert Garlick, Eric Johnson, Patrick Shaw, and Marissa

Gargano, “ Job Search and Hiring with Two-sided Limited Information about Workseekers’

Skills ,” 2020.

Xiao, PengPeng, “Wage and Employment Discrimination by Gender in Labor Market Equi-

librium,” 2020.

46