Estimating and Projecting Health Expenditures on the Elderly in Low - - PowerPoint PPT Presentation

Estimating and Projecting Health Expenditures on the Elderly in Low and Middle Income Counties: An Econometric approach

Abt Associates Inc. In collaboration with:

By By A.K. Nandakumar, Ph.D. A.K. Nandakumar, Ph.D. Brandeis University Brandeis University Jonathan Wilwerding, PH.D. Jonathan Wilwerding, PH.D. Abt Associates, Inc. Abt Associates, Inc.

IAAHS Colloquium IAAHS Colloquium Dresden, Germany Dresden, Germany April 2004 April 2004

“Old age is the most unexpected of all things that happen to a man.” (Tolstoy)

Demographic Trends Demographic Trends

Life Expectancy

10 20 30 40 50 60 70 80 90 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 Year World More developed regions Less developed regions Least developed countries

Average age

0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Year World Less developed regions More developed regions Least developed countries

source: UNPOP 98. Post '95 data are projections.

Population share 65+

0.00 0.05 0.10 0.15 0.20 0.25 0.30 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Year World Less developed regions More developed regions Least developed countries

source: UNPOP 98. Post '95 data are projections.

Population share 80+

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Year World Less developed regions More developed regions Least developed countries

source: UNPOP 98. Post '95 data are projections.

Share of 80+ in the old population

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Year World Less developed regions More developed regions Least developed countries

source: UNPOP 98. Post '95 data are projections.

Dependency ratio

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Year Least developed countries Less developed regions More developed regions World

source: UNPOP 98. Post '95 data are projections.

Elder share of dependent population

0.1 0.2 0.3 0.4 0.5 0.6 0.7 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Year Least developed countries Less developed regions More developed regions World

source: UNPOP 98. Post '95 data are projections

Trends in Global Aging

L Rapid L Predictable L Pattern -- Oldest Old L Distribution across countries L Lack of preparedness

The Elderly in 2020: In the Next 50 Years

L Share of world population 65+ will double L Average age increases from 26.5 years to 36.2 L Share in 80+ age group will quadruple L By 2025, 75% of world’s elderly will be living in

developing countries

L Aging populations severely pressure public

finances; policy responses must be radical and swift

See: See: Bloom, D.E., Nandakumar, A.K., Bhawalkar, M., “The Demography of Bloom, D.E., Nandakumar, A.K., Bhawalkar, M., “The Demography of Aging in Japan Aging in Japan and the US,” and the US,”

Estimating Expenditures on Elderly Estimating Expenditures on Elderly

Current State of Knowledge

L Almost all the work done in developed countries L Approaches fall into following categories

Actuarial Micro-simulation

L A number of challenges in low and middle income

countries

Lack of longitudinal data Health systems that are not transparent Correlations observed in developed countries do not

always hold

See Research Paper No. 01.23, GBD 2000 in Aging Populations by A See Research Paper No. 01.23, GBD 2000 in Aging Populations by Ajay Mahal and Peter Berman jay Mahal and Peter Berman that reviews estimation methodologies. Report published by WHO that reviews estimation methodologies. Report published by WHO

The Proposed Methodology

L Uses a National Health Accounts framework

Classification of sources and uses of health

expenditures in developing countries approved by the World Bank, WHO and USAID

L Four Step Process

Estimate the base case Model macro-economic growth Age Population Project Expenditures (to 2015 in this case)

Sources of Data Sources of Data

National Health Accounts Household Health Care Utilization and

Expenditure Survey

Costing studies Annual Reports GDP Information Population Data from UN Sources Government budgets

Estimating the Base Case

L Public Expenditures (g) = the sum of (quantity

x unit cost) across all functions, j, and public entities, l.

( )

( )

∑ ∑

= =

× =

L l J j jkl jkl g k

c Q Y

1

Estimating the Base Case

L Step 1: Estimate total outpatient visits and

inpatient admissions at public entities for year k

k N i ijk i jk

population q w N Q ×       =

∑

=1

/ 1

Estimating the Base Case

L Step 2: Allocate visits and admissions across

public entities.

jk jkl jkl

Q p Q × =

( )

∑

= ′ ′

+ = =

L l i ijkl

e e x m p P

2

1 |

l i m i â x â x

Multinomial Logit Multinomial Logit

Estimating the Base Case

L Estimate private out of pocket expenditures,

using household survey data

∑

=

=

N i ik i k

y w N y

1

/ 1

k k private k

populaiton y Y × =

Estimating the Base Case

L Step 4: Estimate expenditures by donors

The best way to get donor assistance information is

through a survey of donors

This is because donors give assistance in both cash

and kind

Projecting Expenditures

L We follow a four step process

• Estimate econometric models of private

Estimate econometric models of private expenditures, visits, admissions, and choice of expenditures, visits, admissions, and choice of provider using data for the base year provider using data for the base year

• “Age” the base year population data

“Age” the base year population data

• Model the macro

Model the macro-

• economic growth

economic growth

• Apply econometric models to aged data to predict

Apply econometric models to aged data to predict private expenditures, utilization and out private expenditures, utilization and out-

• of
• f-
• pocket

pocket expenditures for the forecast year expenditures for the forecast year

Estimating Utilization in Projection Year

( ) ( ) ( )

1 | 1

, =

⋅ = =

k k j k k j

S Q E S P Q E

Visits, admissions = (the probability that an Visits, admissions = (the probability that an individual had an illness x the expected individual had an illness x the expected number of visits, admissions) number of visits, admissions) Estimate probability of illness by logistic Estimate probability of illness by logistic regression regression Estimate expected visits, admissions by Estimate expected visits, admissions by econometric model econometric model

Visits, admissions generated by two-stage Poisson Process

• - Separate individuals who must have zero

counts

• - Determine number of visits for others

Estimate utilization by zero-inflated negative binomial regression model (ZINB) Other models underpredict zeros; in low, middle income countries utilization drops for older ages

Estimating Outpatient Visits in Estimating Outpatient Visits in Projection Year Projection Year

Estimating Out-of-Pocket Expenditures and Choice of Provider in Projection Year

L Use OLS to predict private expenditures in

projection year, conditional on having > 0 expenditures.

L Tobit model and Sample Selection models

predict poorly within sample

L Use multinomial logit models to predict the

choice of provider

Aging the Survey Data to Get the Data Set for Projection Year

L Age only data for variables that appear in

econometric models used to predict use and expenditures for the forecast year

L Add fifteen years to each individual in the

survey data (projection year was 2015)

L Create observations for age group 0-15 L Use econometric methods to assign predicted

values to relevant variables

L To account for the effects of mortality, weight

the data using 15-year survival rates, by age

Creating the Data Set for Projection Year

L Variables whose predicted values were

modeled included

Marital Status Education Income Employment Health Status Insurance Cost per visit and cost per inpatient admission

Modeling Macro-economic growth

L Model of economic growth reveals complex

relationship:

L Assumes that every country has a ceiling on the

level of per capita income determined by geography, natural resources, public policies, and human capital

L Assumes that every country is out of

equilibrium with respect to its attainable income, but is tending toward that level.

See: Demographic transition and economic development: The case o See: Demographic transition and economic development: The case of Jordan f Jordan by David Bloom, David Canning, A.K. Nandakumar, Jaypee Sevilla, by David Bloom, David Canning, A.K. Nandakumar, Jaypee Sevilla, Kinga Kinga Huzarski, David Levi, and Manjiri Bhawalkar Huzarski, David Levi, and Manjiri Bhawalkar

Modeling Macro-economic growth

L Compare two countries with the same starting

level of income: country with the higher potential income would grow more quickly

L If all countries had the same underlying

characteristics per capita incomes would converge

Application to Jordan Application to Jordan

Distribution of Population: 2000-2015

4686 348 5034 6384 597 6981 1000 2000 3000 4000 5000 6000 7000 8000 Non-Elderly Elderly Total In thousands 2000 2015

Population Distribution: 2000-2015

Elderly as Percent of Population increases from 6.9% in 2000 to 8.9% in 2025

Percent Change in Choice of Provider Outpatient Visits: 2000-2015

• 47%
• 6%

51%

• 28%
• 14%

34%

• 21%
• 16%

28%

• 60%
• 40%
• 20%

0% 20% 40% 60% MOH RMS Private 5.6% Growth rate 3% Growth rate 2% Growth rate

Change in Choice of Provider Inpatient Care:2000- 2015

• 36%
• 35%

99%

• 33%

55%

• 8%
• 33%

40%

• 15%
• 60%
• 40%
• 20%

0% 20% 40% 60% 80% 100% 120% MOH RMS Private 5.6% Growth rate 3% Growth rate 2% Growth rate

Percent Change in Annual Per Capita Out-of-Pocket Expenditures Elderly:2000-2015

0% 20% 40% 60% 80% 100% 120% 140% Outpatient Inpatient Medicines 5.6% Growth rate 3% Growth rate 2% Growth rate

Expenditures on Elderly as Percent of Total Health Expenditures: 2000-2015

20.2% 23.2% 32.7% 37.9% 0% 5% 10% 15% 20% 25% 30% 35% 40% Year 2000 5.6% GDP Growth 3% GDP Growth 2% GDP Growth Year 2015

Concluding Comments

L As countries age and development the

demand for health care services will grow rapidly

L The demand for private sector services vis-à-

vis public services will grow

L Competition for scarce health resources will

increase

L Pressure on the non-health public sector will

also increase

L The burden of out-of-pocket expenditures on

the elderly will rise over time

Conclusion

L Globalization will force insurance companies to

enter health markets in low and middle income countries

L Lack of longitudinal data L Lack of transparency in health system L This methodology permits estimating and

projecting expenditures using econometric tools

Conclusion

L Can be used until such time as reliable

longitudinal data becomes available

L In developing countries it is necessary to model

both public and private expenditures

L The entire population has to be modeled not

just one sub-population

L This work is supported by USAID through the

PHRplus project

Collaborative effort with East-West Center in

Hawaii and counterparts in J

• rdan and the

Philippines

PHRplus is funded by the U.S. Agency for International Development and implemented by Abt Associates Inc. and partners:

Development Associates, Inc. Emory University Rollins School of Public Health Philoxenia International Travel, Inc. Program for Appropriate Technology in Health Social Sectors Development Strategies, Inc. Training Resources Group Tulane University School of Public Health and Tropical

Medicine

University Research Co., LLC

Thank You Thank You