Page 108 - Effective healthcare cost containment policies Using the Netherlands as a case study - Niek W. Stadhouders

Chapter 5
The final dataset contains 1893 providers, of which 742 (96% of total turnover) have data for all years. This indicates that small providers are likely to have data for only a few years, which could cause selection bias if budget reallocations are biased towards small providers. To take into account reallocations to missing data, total sector spending (including missing data) is used as reference. Total sector spending is derived from annual financial statements of the Ministry of Health, Welfare and Sports of the Netherlands (2006- 2016). The most recent figures are used.
We define the market volatility index (MVI) as the part of the total budget that is reallocated between providers between years. In each year, provider market share is calculated by dividing the provider budget by the sector macro budget. Changes in market shares can be positive or negative, and sum up to zero. The MVI is defined as half the sum of absolute change in market shares, as a gain in market share of one provider automatically means a loss in market share for another provider8. Equation (1) displays our index of market volatility (MVI) mathematically, where S is the macro budget of sector and is the budget of each provider , ranging from 1 to .
(1)
(2)
(3)
8 E.g. If a payer reallocates 5% of the market, he takes 5% from providers and gives 5% to other providers. The sum of absolute changes in the market is 10%, but the percentage of the market that has been redistributed is half that (5%).
100
We standardize the MVI by assuming that the unobserved market volatility due to missing data is equal to the observed volatility. This may underestimate total volatility when small providers that do not file their annual statements experience higher volatility than large providers.
We calculate equation (2) for each sector: hospital care, elderly care and disability care (LTC), social care and personal budgets. Furthermore, in order to discern between annual fluctuations (e.g. increase in one year and reductions in the next) and structural reallocations (e.g. a trend of reductions in a row), we calculate structural volatility between 2007 and 2014.
Structural volatility ( ) is by definition equal or lower than annual volatility. For each market, we calculate the mean MVI over time, which consist of structural volatility and annual fluctuations.