Why is ‘real world’ financial risk so high?

Abstract

Over the past three years a series of articles in BJHCM have sought to unravel the complex issues behind ‘real world’ financial risk in healthcare commissioning. These have included the difficulties associated with forecasting future demand. It has been observed that the risk calculated by computer simulation (i.e. the minimum possible case scenario) is already unacceptably high, and that adding in the effects of the environment and high-cost individuals leads to ‘real world’ risk which is up to threetimes higher than the minimum case scenario. How do we understand the practical effects of risk? Risk and volatility are interlinked. A useful way to view risk is to look at the year-to-year volatility in healthcare expenditure—does expenditure jump about erratically? Figure 1 displays the average yearto-year volatility in total healthcare expenditure experienced by English primary care trusts (PCTs) between 1994/95 to 2010/11 after adjusting for the underlying growth in expenditure. The data is presented as a log-log plot since this reflects the role of Poisson statistics in determining the relationship between volatility and size (Jones, 2009). Over this period healthcare funding grew at a roughly constant rate (approximately 6% per annum) and the volatility has been adjusted for underlying linear growth. Average national volatility is 2% and we firstly need to sense check this. Computer simulation of variation in costs using a set of limiting assumptions (i.e. known case mix, excluding any environmental effects (‘bad’ winters), 22 high-cost HRGs and high-cost individuals) shows that the minimum theoretical volatility for PCT-sized organisations is around 1.1% (Jones, 2008). Given that winter falls at the end of the financial year, giving high volatility in medical costs, and adding back high-cost patients will easily give average volatility of 2% and hence the results in Figure 1 are indeed a reflection of ‘real world’ volatility in expenditure rather than data errors. Further consideration of Figure 1 leads to a number of key conclusions. The first key observation is that the bulk of PCTs experience higher than 2% average volatility, yet the national average was 2%. To reconcile this apparent discrepancy, PCTs were ranked and the resulting average volatility calculated for a risk pool where one extra PCT is incrementally added at a time. The line for the minimum risk pool shows an amazing minimum risk of only 0.3% when the four (geographically disperse) PCTs having the lowest risk are aggregated together. What has happened is that peaks and troughs in expenditure have almost exactly cancelled out giving the financial stability equivalent to winning the lottery. This is counter balanced by another group of four ‘unfortunate’ PCTs whose combined average risk was around 5%. Figure 1. Volatility in total healthcare expenditure