CHI Bangladesh Model Documentation

Population

Population simulates how many people are born and die each year:

The figure shows that as Population increases, births increase. And as Population increases, deaths increase but that causes Population to decrease. Thus, the “births loop” continuously increases Population, and the “deaths loop” brings Population back into balance.

The principal output of Population is the Population Insured Status. As the Population increases or decreases, it affects the number of people who are insured or noninsured.

The principal inputs to Population are birth and mortality rates and the “effect of effect of improved healthcare on mortality rate”.

There are two “initial conditions” that can be selected. One is called dynamic equilibrium in which the total population is distributed across age groups and the birth rate is computed to maintain the same total population over time. The other is called dynamic disequilibrium in which the initial value of each age group reflects the 2010 census. Although the dynamic equilibrium is purely hypothetical, it helps users to isolate the effect of changes in other parts of the healthcare system.

Enrolment

Because one of our goals is to understand how the market for healthcare insurance can emerge, we simulate enrolment in the proposed health plan. The target market is informal workers, most of whom are between the ages of 15 years and 60 years old. In the subject area, we estimate there are about 27,000 informal workers, and about twice as many men are classified informal workers as women.We assume people enroll because of two influences: advertising (solicitation) and word-of-mouth:

There are important questions to consider:

How much influence does advertising have?
How much influence does word-of-mouth have?
How does the attractiveness of the offering affect the size of the market?
What causes the rate of disenrollment to change?

Because there is no comparable marketplace to study in Bangladesh, the model allows us to explore a wide range of assumptions that can change outcomes. For example, we can change the influence of both advertising and word-of-mouth to see how differently the market can grow. Or we can consider how access to medical care might change the disenrollment rate.

The principal inputs to Enrolment are Population, Attractiveness, Accessibility and Availability.

The principal output is Primary Enrolment.

Insured Status

Insured status is highlighted in the model because we can assume that only informal workers are allowed in the health plan or we can assume that informal workers may enroll their dependents:

In the model, we distinguish between ages and sexes because the healthcare utilization rates are very different amongst ages and sexes. In the full model, we simulate by male and female gender and age groups (cohorts) for children (under age 5 and juveniles under age 15 years), people of child-bearing years (fecund), those who are economically productive but no longer having children (mature), and those over age 60 years.

The principal inputs are informal workers, fraction of informal workers enrolled, and the dependency ratio.

The principal output is the population by sex and insured status.

Dependencies (Dependents of informal workers)

The Group estimates there are 3.5 dependents for each informal worker – the dependency ratio. The target enrolment market comprises all informal workers. Thus,

total dependents of informal workers =

dependency ratio

* target enrolment market

Units: people

When switch on dependents of informal workers = 1, total dependents of informal workers are distributed to the age and gender cohorts in the proportion of the total population:

Demand

Demand is simulated as a perceived medical need. We do not have measured data to say what medical needs are, but we can look to Bangladesh or similar countries (in this case, Indonesia) to estimate how many times per year people of the same age and gender will seek care from a physician, nurse or community healthcare worker. We can assume the perceived need for medical care is the same regardless of insured status, but the likelihood of the people with needs is definitely influenced by their insured status. People with insured access to care seek it more frequently.

At the beginning of any large health plan, the most likely people to enroll are those with a perceived medical need. In economic terms, this is called “adverse selection”. That is, the selection is adverse to the operation of the health plan because the early insured people demand more care than those who sign up later.

As the Historical Fraction of Population Enrolled increases, the effect of adverse selection decreases. In essence, the healthier population is less likely to enroll early and only gradually the impact of adverse selection. As the fraction enrolled approaches 100%, adverse selection no longer effects the utilization rate.

The inputs to demand are the average (by age and gender) perceived medical need for physician or medic care, the likelihood of people to seek care based on their insured status, the modification we expect for adverse selection and the historical fraction of enrolled population.

The outputs are demand for physician or medic (nurse) care.

Adverse Selection Effect

The group assumes the possibility of adverse selection: low risk patients opt out of the scheme (or do not join in the first place) with the result that a scheme contains only patients with expensive needs, i.e. above average demand for healthcare. The group further assumes that those with the greatest perceived needs will enrol earliest. Hence, demand per enrolee can be estimated to be highest when enrolment is least, and demand per enrolee will decrease as more people enrol.

To implement these assumptions:

effect of adverse selection[sex] =

( 1

+ adverse selection modifier

* ( 1

- Historical Fraction of Population Enrolled ) )

Units: Dimensionless

Where:

Historical Fraction of Population Enrolled is a simple exponential smooth over 6 months of fraction of informal workers enrolled, and

fraction of informal workers enrolled =

insured status[insured] * Primary Enrolment / target enrolment market

Units: Dimensionless

The adverse selection modifier = 1, selected by the group to estimate that earliest enrolment will have twice the needs of the total population.

Therefore, as actual enrolment approaches total potential enrolment, the effect of adverse selection diminishes to zero.

Health Status

The model permits healthcare to affect health status. To approximate this effect, the group assumed that as more of the population maintains membership in the healthcare scheme, the population health status will improve over baseline. To that end, a coflow measures average enrolment time.

This structure tracks enrolment and rate of gaining time in the stock, less the number of those who disenrol times the average enrolment time to keep a rolling average, which is applied to the population:

total enrolment time of insured population =

average enrollment time

* population by insured status

Units: Month

The result is then compared to the total population experience:

fractional improvement in mortality rate =

total enrolment time of insured population

/ total population experience

Units: Dimensionless

Where:

total population experience =

total population

* average population life span

* months per person per year

Units: Month

The fractional improvement in mortality rate is applied to the insured population in a fixed delay of 60 months:

effect of improved healthcare on mortality rate =

fractional improvement in mortality rate

* max improvement in mortality rate from healthcare

Units: Dimensionless

Where:

max improvement in mortality rate from healthcare = 0.1

Units: Dimensionless

The effect of a more inclusive healthcare system is expected to be more productive work time. That is, actual morbidity is assumed to remain constant but the time away from economic production can increase over the baseline:

informal worker days per month =

initial informal worker days per month

+ max ave productive days gained

* fraction target enrolled

Units: day/(Month*person)

Where:

initial informal worker days per month = 24

Units: day/(Month*person)

max ave productive days gained = 2

Units: day/(Month*person)

fraction target enrolled =

Primary Enrolment

/ target enrolment market

Units: Dimensionless

Treatment Capacity

The capacity to meet the demand for healthcare is limited or constrained by the number of physicians and nurses. We make no assumption about the severity of medical conditions, but we assume people have a limit to their willingness to wait for care. Thus, as waiting times increase, the number of people who simply quit waiting increases too.

Patients waiting for treatment cannot know how much longer they will have to wait, but they can know how long they have waited. In the model we simulate this by comparing the size of the queue with the number of people currently being treated:

Receiving treatment and quitting the queue have the same effect on the Queue for Care: an increase in either reduces the Queue and future waiting time.

We assume that physician and nurse caregivers do not distinguish between insured and noninsured patients. That is, they allocate their visit time based on the presenting patient. However, demand by insured patients is likely to be greater than noninsured patients (measured patient by patient), so that available capacity will skew to insured patients.

The principal inputs are demand and physician visit capacity.The principal output is receiving treatment.

CHI Attractiveness

The marketplace judges whether a health plan is attractive. In the model, we assume several variables influence that judgement.

People perceive:

the health plan is available – Awareness
there is treatment capacity to meet their demand – Availability
treatment is within reach – they can get to treatment with ease – Accessibility
the health plan is affordable – it does not use too much monthly income – Affordability
expenditures to treat their medical conditions will be paid by the health plan – Coverage

We add one more element of attractiveness: Bank Attractiveness. That is, our plan is connected to a bank that offers e-payments by telephone text. The members make small monthly deposits to savings that permit them to borrow money for approved needs. The Bank may be highly attractive if it offers low cost loans, above-market savings rates, and financial stability.

Inputs include advertising impact, physician and nurse demand met, relative distance to treatment, portion of monthly income expended for premiums, and portion of medical conditions covered.

Outputs affect market size and enrolment, and in the case of the bank, membership in the bank customer community.

Expenditures Per Member Per Month

The principal expenditures per member are the sum of monthly visits to care venues times the prices charged for each visit:

total visit claims =

nurse visit claims BDT + physician visit claims BDT

Units: BDT/Month

Where:

nurse visit claims BDT =

nurse visit price

* receiving nurse treatment[insured]

Units: BDT/Month

physician visit claims BDT =

physician visit price

* receiving phys treatment[insured]

Units: BDT/Month

Claims Processing

Health plans provide structure to the health care system. Not every expenditure claimed as a covered expense is covered by the plan. Some noncovered expenditures include certain illnesses, say, very costly cancers or very common upper respiratory tract infections. Claims are usually submitted for payment by the provider and contain date of rendered service, identification of the insured person, a description of the condition and service provided, and amount of the claim. The claims are bundled by provider and the covered claims are paid.

The processing of claims can be entirely manual or electronically assisted. In large plans, claims are submitted electronically and processed without intervention by a claims processor. This automated claims processing is very inexpensive but can be prone to erroneous payments to providers. The question is whether the cost of auditing many claims is less than the amount that would be saved by catching overbillings.

Unacceptable claims are “pended”, meaning the provider is told the claim will not be paid without additional justification. This is a relatively time-consuming process because the human intervention is extensive. Some pended claims will be reprocessed and paid; some pended claims will not be pursued by the provider.

In considering the claims processing function, processor productivity is influenced by electronic assistance such as automated claims processing. The denial rate influences the number of times a claim will circulate through the system before it closes.

Principal inputs include claims made, claim processor productivity and claims denial rates.

The output is claims processed.

Monthly Premium

To breakeven the monthly premium covers medical and administrative costs. The challenge facing any healthcare plan is to set a stable premium in advance of incurring those expenses and hold that rate constant for a year. Health plans employ actuaries to analyze historical expenditures and guess at which trends will dominate in the coming year.

Our model simulates that activity by accumulating medical, general and administrative expenses that indicate a monthly premium which is then analyzed for its trend over time. The trend-adjusted monthly premium is then projected over a time horizon that represents how confident the forecaster is that trends will continue.

Inputs are claims from medical practice, claims processor wages paid and general expenses of running the business.The output is the monthly premium.

Insurance Firm Accounts

Accounting for health plan operations is relatively simple, although companies in the industry use sophisticated management tools to maximize invested sums.

Inputs:

The principal outputs are monthly insurance income, Cash and Short-term Investments.

Microfinance Structure

Banks are boxes of cash coming in and going out. The accounts reflect those flows.

Inputs:

Deposits may be mandatory for membership, and that is a choice yet to be made.

Outputs include loan defaults and equity that influence the Bank’s Reputation.

Bank Attractiveness

Bank Attractiveness is represented by perceived Awareness, Accessibility, Affordability and Reputation.

Awareness is a function of advertising and the word of mouth in the Health Plan. Accessibility is measured as the number of bank locations relative to the number of customers. Affordability is measured as the interest rate spread in its loans. Reputation is a function of loan defaults and the Bank Equity.

Inputs are target market awareness, bank locations, loan interest rate spread, loan defaults and bank equity:

Bank Attractiveness =

Bank Accessibility

* Bank Affordability

* Bank Awareness

* Bank Reputation ,

Units: Dimensionless