Financing Social Health Insurance in Developing Countries: Impacts on Fiscal & Labor Market Outcomes

by

Adam Wagstaff (World Bank) IDB/PAHO Regional Workshop on “Fiscal Space and the Financing of Universal Health Care in the Americas”

Washington DC, 29 - 30 November 2007

Introduction 90% of OECD countries finance majority of health

expenditures publicly Half use general revenues. Other half have SHI

systems, dedicated earnings-related contributions for formal-sector workers

Among the non-OECD countries, 56% finance a majority of health spending publicly, and only 20% have SHI

Many countries are embracing SHI, often with the blessing—if not the encouragement—of donors

Yet this is happening at a time when Germany et al. trying to reduce their reliance on payroll

financing of health care LAC advised to follow suit by Baeza & Packard in their

Beyond Survival

Evidence is clear on some issues Revenues:

Revenues fall short of “theoretical” levels due to evasion and underreporting of earnings

MOF sometimes reduces govt. spending on health in line with theoretical SHI revenues

SHI financing less equitable than taxes in general Contribution ceilings limit progressivity Horizontal inequity: contrib. schedules vary by

scheme Coverage:

Gaps in coverage until countries reach high per capita income

Financing less progressive under SHI

TaiwanKorea

Japan

Hong Kong

Denmark

Finland France

Germany

Ireland

I taly

Netherlands

Portugal

Spain

SwedenSwitzerland

UKUSA

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

0% 20% 40% 60% 80% 100%SHI share of general govt. health financing

Prog

ress

ivity

of g

ener

al

govt

. hea

lth fi

nanc

ing

Gaps in coverage under SHIVietnam, 2004

0%10%20%30%40%50%60%70%80%90%

100%

1 2 3 4 5I ncome quintile

% in

sure

d

Volunt.

Student

Compuls.

Pers ofmeritFree

Colombia, 2005

0%10%20%30%40%50%60%70%80%90%

100%

1 2 3 4 5I ncome quintile

% in

sure

d

Contrib. Subs.

Argentina, 1996/7

0%10%20%30%40%50%60%70%80%90%

100%

1 2 3 4 5I ncome quintile

% in

sure

d OS+priv.

Private

Obrassociales

Chile, 1998

0%10%20%30%40%50%60%70%80%90%

100%

1 2 3 4 5I ncome quintile

% in

sure

d

FONASAOther ins.ISAPREs

Evidence is less clear on other issues—i

Do SHI systems spend more on health care? On the one hand:

People more willing to pay SHI contributions than taxes?

Guaranteed revenue stream? On the other hand:

Actual and theoretical revenues often diverge SHI revenues grow more slowly than tax revenues

due to ceilings

Evidence is less clear on other issues—ii

What are the impacts of SHI on health outcomes? On the plus side, SHI may:

Bring in additional resources to the health sector Stimulate efficiency in the delivery system, through

separation of purchasing and provision, and provider payment reform

On the negative side, SHI may result in: Gaps in effective coverage Focus on inpatient care, neglect of prevention,

early detection, etc. Higher wages in health sector, not necessarily

higher quality Purchaser-provider splits and payment reform

happen also in non-SHI countries, not always in SHI ones!

Evidence is less clear on other issues—iii

What’s the impact of SHI on employment? Argued that payroll financing reduces employment,

by raising cost of labor. But… Labor supply curve (for formal sector) shifts

rightwards because workers value SHI benefits—so, smaller disemployment effect and a larger reduction in the post-tax wage than in standard case

Relevant question is whether a health system financed through payroll taxes leads to lower employment than one financed through general revenues

Does SHI encourage informalization of the economy? Depends on incentives people face—ECA different

from LAC?

Europe & Central Asia’s SHI ‘experiment’—learning opportunities

Staggered and incomplete adoption of SHI in ECA countries during 1990s provides an opportunity to assess some of the aggregate effects of SHI adoption

Study design similar to multiple U.S. studies in many fields that exploit staggered and incomplete policy roll-out across the 50 states

Country-level analysis permits aggregate effects to be estimated. So, capture effects on all the relevant actors in the

health system, including new ones (e.g. new SHI agency, new entrants into provider market, etc.)

SHI adoption in ECA: A quick history—i 1945-1990, most ECA countries financed

health care through general revenues and delivered it though centrally-planned Semashko model

In early 1990s, as they shifted from Communism, many countries looked to SHI to help solve several emerging problems: Dramatic decline of govt. revenues as share of

GDP and falling GDP SHI thought likely to lead to better health delivery

system. SHI agency would sit at arms’ length from MOH and MOF, would develop purchasing capacity, promote competition within public sector and between it and private sector

Who adopted SHI when? And what share of spending was financed through SHI?

Source: HiTs and World Health Reports, various years

SHI adoption in ECA:A quick history—ii

SHI makes up a bigger share of revenues in E European countries, where contribution rates are high

Most countries do have a SHI agency, but so too do Poland and Latvia which use income taxes or general revenues

Often MOH still transfers some funds to providers, and SHI agency contracts have taken time to emerge, are often not competitive, and often do not involve private sector

SHI has often but not always led to switch from budgets to FFS or patient-based payments (e.g. DRGs). Some non-SHI countries also switched

Methods

Generalization of differences-in-differences (DID) estimator: includes zit and git

Also estimate eqn below using IV where have evidence that above eqn doesn’t address endogeneity of SHI adequately

itiiitittit utgSHIzy

itititit eSHIzy

Health sector outcome variablesVariables Sources

Health spending & resources

Total health spending per capita; salaries as % spending; physician numbers

WDI; WHO-Health-for-All

Hospital throughput & capacity

LOS; bed occupancy rate; # beds; inpatient admissions WHO-Health-for-All

Hospital discharges By diagnosis WHO-Health-for-AllImmunization By type WHO-Health-for-All

MortalityLife expectancy; U5MR & IMR; MMR; standardized death rates

WHO-Health-for-All; UNICEF TransMONEE

Avoidable deaths (quality proxy)

Deaths from appendicitis, hernia, surgery infections WHO-Health-for-All

Disease incidence By diagnosis WHO-Health-for-All

Health outcomes dataset is 77% non-missing. (69 outcome variables. 28 countries. 16 years. Maximum # observations = 30912. Actual # observations on health outcomes = 23680.)

Labor market outcome variables

Variables Sources

Wage rateTotal annual wages and salaries in constant PPP averages for the employed population aged 15-59

Own calculations based on data from WDI and UNICEF TransMONEE

Unemployment Unemployment rate; registered unemployed; long-term unemployed ILO

Employment % working-age population and population aged 15-59 employed ILO; UNICEF TransMONEE

Informal economy

Based on discrepancy between growth of GDP and electricity demand

Own calculations, based on Johnston et al. method

Informal employment

Self-employment; agricultural employment ILO

Labor force participation Whole population; women only ILO

Labor market outcomes dataset is only 55% non-missing. (8 outcome variables. 28 countries. 16 years. Maximum # observations = 3584. Actual # observations on health outcomes = 1987.)

z variables

Variables Sources

GDP GDP per capita, PPP (constant 2000 international US$) WDI

Public share of health spending

Health expenditure, public (% of total health expenditure) WDI

Elderly population* Population ages 65 and above (% of total) WDI

Urban population* Urban population (% of total) WDIHealth spending$ Total health spending per capita WDIHospital payment method*** EXCLUDED FROM BASIC MODEL ***

FFS, patient-based method (e.g. DRG). Budget is omitted category HiTs

* Excluded from labor models. $ Excluded from health models.

Basic model: SHI impacts on spending and hospitals

Basic model: SHI impacts on life expectancy and mortality

Basic model: SHI impacts on cause-specific mortality

Basic model: SHI impacts on disease incidence & immunization

Effects on SHI impacts of including provider-payment reforms variables

SHI impacts on labor market outcomes

Conclusions

Already known Coverage

Gaps often occur under SHI—often among poor, near-poor Less generous coverage translates into lower utilization but

not necessarily inferior financial protection Raising revenues: SHI vs. taxes

SHI revenues may be lower than expected due to evasion etc., are they more/less predictable?

SHI less progressive than tax finance New from ECA study

Health care delivery SHI neither necessary nor sufficient for separation of

purchasing & provision SHI systems more expensive but do not apparently achieve

better health outcomes despite higher spending SHI and the labor market

SHI raises (gross) wages, decreases employment Impacts on size of formal sector unclear