Presentation is loading. Please wait.

Presentation is loading. Please wait.

Estimating the future demand for care

Similar presentations


Presentation on theme: "Estimating the future demand for care"— Presentation transcript:

1 Estimating the future demand for care
Elizabeth M. King*, Hannah Randolph and Maria Floro** *The Brookings Institution, Washington, DC ** American University, Washington, DC Please, do not distribute outside of conference. Work in progress only. Presented at 2018 Annual Meeting of the Research Project “Care Work and the Economy: advancing policy solutions with gender-aware macroeconomic models” in Berlin, October 21-23

2 Motivation Care needs are intensifying but diversely, due to different patterns of demographic change – how to meet those care needs? Women have been the primary provider of unpaid direct and indirect care in the household. How can women increase their labor force participation and earnings while also bearing this burden? This paper uses the perspective of the household to estimate the future demand for care in different populations Care provision fosters growth and sustainable development (SDG 3, 5, 8); it is necessary for enhancing quality of life and advancing gender equality. The connection between gender norms and the burden of unpaid care work creates persistent inequalities at home, in labor markets, and in decision-making processes. Point 3: high fertility in low- and middle-income countries, as they move to demographic transition also have aging Increased LFP for women

3 The scope of care work and data sources
 Type of care Unpaid care Paid care services Household unpaid Non-household Primary Secondary Home-based Center-based Direct (relational) care Indirect care Outside the household (e.g. transfers to other households

4 The scope of care work and data sources
 Type of care Unpaid care Paid care services Household unpaid Non-household Primary Secondary Home-based Center-based Direct (relational) care Time use data Limited time use data Very limited time use data Household/ consumption survey, labor force survey data Indirect care Limited time data (e.g., transfers of time from other households) No time data Outside the household (e.g. transfers to other households We focus on the relational care services demanded by dependents in the population, including young children, elderly, and the sick and disabled; Direct and relational care; Special attention to the provision of unpaid direct care given by household members.

5 Presentation in three parts
Aggregate demand: Projecting care needs by country income groups (“extensive margin of care”) Estimating future populations ill and disabled, by country income groups Part II Using time use data to estimate the burden of care on the household and to understand how household composition affects that burden (“intensive margin of care”) Estimates for Ghana and Mongolia using their time use data Part III Relative care intensity scale derived from analysis of time use data Projecting care needs using this data-based scale

6 Population structures determine the future need and demand for care
Data source UN population estimates and projections, 2015 and 2030, by income country group Diverse patterns across country income groups, but those patterns will intensify between 2015 and 2030 Middle-income countries have the bulk of the world’s population: young, but promises a “demographic dividend” because of falling fertility rates Low-income countries, mostly in Africa, are very young populations. Fertility rates are above 3 percent for many countries High-income countries are aging and will continue to do so Hannah

7 Emphasize age groups patterns, especially aging populations.
Calculate dependency ratios based on population projections (use ILO methodology) Low-middle and high-middle

8 Briefly describe population trends, particularly for low- and middle-income countries
MI countries can benefit from demographic dividends– large number of prime-age adults Dependency ratio worries in LI, not MI countries

9 An epidemiological transition is changing the sources of morbidity and mortality
Public health improvements, medical advances, and economic growth have reduced the toll from communicable diseases worldwide Noncommunicable diseases become more important as incomes grow and populations age. Seen most in middle-income countries Data sources Institute for Health Metrics and Evaluation (IHME) estimates of years lost due to disability (YLDs) YLDs represent the burden of disease on the population in a given year. Projected disability rates are a straight-line fit on yearly IHME YLD rate estimates from , projected forward to 2030 Beth Also insert slide with dependency ratios based on the healthy population (see ILO report ~p58 for methodology)

10 These four slides can go quickly– the main point is that the burden of communicable diseases is decreasing, and the burden of noncommunicable diseases is increasing. This will change the landscape of care demand, particularly for households in developing countries.

11 Non-communicable YLDs have a much larger impact than communicable;
Need for care is distributed throughout population, although some groups are more impacted than others (e.g. children by communicable diseases)

12 Unpaid household care: estimating the intensive margin of care demand and supply
Population and YLD projections provide an estimate of the extensive margin of care needs and demand – but what about the intensity of care for different types of dependents? Durán Heras (2012) and others have developed scales of care units that can be used to translate population projections into estimates of future care needs. (e.g. Durán 2012) Demographic definition of dependence: “Dependence indices do not weight, they only express a ratio. ...they weight the entire population with one unit of care“ (p. 424) Current scales to translate demographic statistics into care weights averaged from expert opinion from focus groups (e.g. Durán Heras, 2012). Weights relative to a central group Can probably be combined with following slide for simplicity.

13 Demand for care scales Scale Region 1-4 5-14 15-17 18-64 65-74 75-80
81-84 85+ Disability? Data Madrid I Spain; general use 2 1.5 1.2 1 1.7 Yes Madrid II 3 Freetown Africa Rogero García 2012 for Africa; care demand given by (A x (care weight))/10 Domínguez 2010 UN; TUS for Benin, South Africa, Madagascar, and Morocco Domínguez 2025 Same as above Domínguez 2050 4 Santiago Latin/South America 5.57 3.48 2.13 2.29 3.98 5.1 Estimates of relative care needs provided by 21 specialists

14 Relative care intensity scale approach
We use time use and household composition data in regression analysis to describe how household unpaid care is allocated among household members Aim is to construct a scale or index that weights different household members according to their age, relative to adults in a middle age group (e.g., ages 18-64) And the weights are derived from a regression analysis of household unpaid care time against the household composition (count of household members by age group) Age groups (e.g., 0-4, 5-17, 18-64, 65-74, 75+) but time use surveys differ with respect to age categories Note: Need to use time use data that include time from all household members, not just from 1-2 household members Basic regression (linear specification): estimate the marginal demand for unpaid care time provided by the household for each age group by regressing the care time (direct, indirect, child, etc) on relevant age groups and a vector of control variables (such as gender of the household head, etc.) The care index is then the ratio of marginal effects of the group in question and the comparison group (15-64 year-olds) Discuss other scales– Madrid, Santiago, etc.

15 Measurement issues common to time use data
Unpaid care work often includes only primary activities; secondary or overlapping activities are missed underestimates intensity of care time Unpaid direct care work might not be reported separately from other types of unpaid work underestimates direct care time Survey respondents may not include all household members underestimates household care time Care work may be received from a paid worker at home or in a center underestimates care needs unless specific data available Care work may be provided by an unpaid non-household member underestimates care needs Recipients of care work may not be adequately identified (important if care work is provided to another household as unpaid care) overestimates household need Time use data confound supply and demand “reduced form” analysis Will also discuss the specific data limitations of the countries we focus on in this section. Ghana: does not separate primary and secondary activities; does not include the sex composition of the household (cannot identify male/female yos)

16 Regression analysis of household unpaid care
Desirable to have aggregate primary and secondary activities, but usually only primary activities Separate estimates for direct care and indirect care Estimated linear and log-linear specifications, with quadratic term for count variables and pairwise interaction terms of count variables to estimate economies of scale and scope Control variables: Sex and age of household head, urban/rural residence, ownership of land or livestock 𝐶 𝑗 = 𝛼+ 𝛽 𝑖 𝑖=(0−4) (75+) 𝑁 𝑖 + 𝛽 𝑖 ′ 𝑖, 𝑘=(0−4) (75+) 𝑁 𝑖 𝑁 𝑘 +𝛾𝑿+ 𝜀 𝑗 These three specifications, the marginal effects, and the elasticities reveal whether there are economies of scope and/or scale. Add what exactly indicates the presence of economies of scale/scope

17 Direct and indirect care work of women and men
Ghana Mongolia

18 Estimates include elasticities of scale and scope
Direct CARE by all Direct CARE by women Direct CARE by men GHANA Num. age 0-4 0.233*** 0.247*** 0.176 Num. age 5-14 0.124** 0.111 -0.363 Num. age 15-64 0.504 1.149 7.930 Num. age 65-74 0.037 0.053 0.036 Num. age 75+ 0.302 0.186 4.651 Observations 4166 Adjusted R2 0.464 0.472 0.121 MONGOLIA Num. age 0-11 0.394*** 0.407 0.438 Num. age 12-14 0.063 0.042 -0.044 0.200 0.0986 0.639 -0.003 -0.037 -0.067 -0.164 0.001 -0.120 1322 0.307 0.285 0.180 Elasticity: Percent change in direct care time in response to a one percent change in household member of a given age, primary activities only Estimates from separate log-linear regressions for unpaid direct (primary) care for children and for elderly members Estimates include elasticities of scale and scope Control variables: Gender and age of head; rural or urban residence; own land or livestock

19 Estimates include elasticities of scale and scope
Direct CARE by all Direct CARE by women Direct CARE by men GHANA Num. age 0-4 0.263*** 0.269*** 0.257*** Num. age 5-14 0.176*** 0.181*** -0.070 Adjusted R2 0.495 0.491 0.124 Num. age 65-74 0.286* 0.216** 0.129 Num. age 75+ 0.672*** 0.502*** 0.178 0.020 0.028 0.015 MONGOLIA Num. age 0-11 0.456*** 0.474*** 0.503 Num. age 12-14 0.006 0.011 0.054 0.295 0.276 0.175 -0.136 -0.126 -0.039 -0.009 0.003 -0.165 0.009 0.007 Elasticity: Percent change in direct care time in response to a one percent change in household member of a given age, primary activities only Estimates from separate log-linear regressions for unpaid direct (primary) care for children and for elderly members Estimates include elasticities of scale and scope Control variables: Gender and age of head; rural or urban residence; own land or livestock

20 Estimates include elasticities of scale and scope
Indirect CARE by all Indirect CARE by women Indirect CARE by men GHANA Num. age 0-4 0.0770*** 0.101*** Num. age 5-14 0.109*** 0.110*** 0.172 Num. age 15-64 0.987*** 0.312 0.488 Num. age 65-74 Num. age 75+ 0.0552 0.101 Observations 4166 Adjusted R2 0.355 0.447 0.089 MONGOLIA Num. age 0-11 0.0266 -0.232 0.0454 Num. age 12-14 0.0806 0.0142 -0.167 -0.319 0.0103 0.217 0.0677 0.0756 0.0258 1322 0.148 0.874 0.481 Elasticity: Percent change in indirect care time given a percent change in household member of a given age, primary activities only Estimates from log-linear regression for unpaid indirect (primary) care, from all household members, all female members, and all male members Estimates include elasticities of scale and scope Control variables: Gender and age of head; rural or urban residence; own land or livestock

21 A relative care intensity scale based on time use data
Does not assume one care unit per household member Using regression analysis of time data, reflects economies of scale and scope within the household that may derive from social norms and household structure Independent of time units Still early days – need to explore how different such a scale would be in diverse contexts GHANA (0-4) (5-14) (15-64) (65-74) (75+) Linear 6.29 0.58 1.00 3.39 3.64 Log-linear 7.03 1.30 2.28 3.24 The goal is to produce results for at least two countries before the conference– likely South Korea and Ghana. MONGOLIA (0-11) (12-14) (15-64) (65-74) (75+) Linear 4.24 -1.91 1.00 5.58 1.66 Log-linear 6.35 0.87 4.54 2.34

22 Applying the relative care intensity index on population projections for 2030
Accounting for unpaid care in the household yields a different picture of the impact of demographic trends on the household economy and the burden of the absence of care services on its members, especially girls and women

23 Early days still … further work
Questions for further analysis How to separate demand and supply forces on observed unpaid care time: would be used to have more control variables at individual and household levels How to consider disability and illness when specific data are not available How to take account of use of paid care services when specific data are not available And of course Estimate for as different and as many countries as possible Apply index to aggregate population projections Which next steps do we want to focus on? How much do we want to add here– immediate next steps, or larger picture of where this could go?

24 Thank you

25 Household composition in Ghana and Mongolia
1-2 3-4 5-6 7+ Total >0 GHANA Num. age 0-4 2604 1486 85 4 4179 1575 62.3% 35.6% 2.0% 0.1% 0.0% 37.7% Num. age 5-14 2020 1598 499 56 6 2159 48.3% 38.2% 11.9% 1.3% 51.7% Num. age 15-64 277 2877 863 142 20 3902 6.6% 68.8% 20.6% 3.4% 0.5% 93.4% Num. age 65-74 3687 492 88.2% 11.8% Num. age 75+ 3872 307 92.6% 7.4% MONGOLIA Num. age 0-11 603 619 96 1322 719 45.6% 46.8% 7.3% 0.3% 54.4% Num. age 12-14 1079 242 1 243 81.6% 18.3% 18.4% 62 786 425 44 5 1260 4.7% 59.5% 32.2% 3.3% 0.4% 95.3% 1198 124 90.6% 9.4% 0.00% 1258 64 95.2% 4.8%


Download ppt "Estimating the future demand for care"

Similar presentations


Ads by Google