I have serious problems with several popular government economic numbers. The first of these is the unemployment rate, which several studies have clearly demonstrated under-report the degree of US unemployment. The second of these is the inflation number. My own experience simply buying and selling goods over the last few years combined with evidence from the financial press adds up to suspicion. After doing some digging on the internet, I realize my concerns are well-founded. The core CPI number has serious flaws that understate inflation, largely because it does not include health insurance premiums and the appreciation of house prices.
Prior to 1982, the housing cost numbers were based upon what you actually spent for the house and the related mortgage. After 1982, the Bureau of Labor Statistics (BLS) began to use an imputed number. They now use what is known as “owners’ equivalent rent of primary residence” for the housing portion of the CPI. This is based on an economic theory that says that homeowners are essentially leasing the houses from themselves and paying implied rent for that service.
In theory, they are trying to figure out what it would cost you to rent your home. There’s actually a rational reason for doing this and we will talk about that in a minute, but first let’s look at the numbers.
Why are these imputed rents so low? Dean Baker tells us, “The main factor holding down shelter costs is the overbuilding associated with the housing bubble. This has led to record nationwide rental vacancy rates, which is putting downward pressure on rental prices in many of the areas with the biggest bubbles in housing prices. For example, rents in the New York City area rose by just 1.9 percent over the last year. They rose by 1.8 percent in Tampa, Florida and by just 0.3 percent in both Boston and San Francisco. (This is the inflation rate for the owners’ equivalent rent index, which strips out utility prices.)”
How much does using imputed rent affect the CPI? Bill King wrote a few months ago, “In the Q1 GDP data, the US government has housing prices up only 1.1%, yet industry data shows double digit gains. And this week the June existing home sales data shows a 14.7% increase in the median house price. The BLS has ‘owners’ equivalent rent of primary residence’ up only 2.2%.
Since 1982, the housing component of CPI has not included the actual cost of houses. Simply eyeballing this chart from Prudent Bear indicates the average price of homes sold increased from $180,000 in January 2002 to just below $260,000 in mid-year 2005. That’s a hefty increase to not include in a price index.
Also of particular interest is the housing bubble’s effect on the CPI numbers. Because of overbuilding, rents are not escalating nearly as much as the actual price of the underlying property. In other words, the housing bubble is helping to understate CPI.
According to the BLS’ website:
The CPI does not publish a health insurance index, although BLS is testing its feasibility with an experimental index. The weights in the CPI do not include employer-paid health insurance premiums or tax-funded health care such as Medicare Part A and Medicaid. Currently, the index employs an indirect method for measuring price changes for health insurance premiums. Under this indirect method, the medical care index will not be affected by changes in policy characteristics, such as modifications to policy benefits and utilization changes. The approach implicitly assumes that the level of service from individual carriers is strictly a function of benefits paid. While other components may affect the index, such as more convenient claims handling or a 24-hour nurse line, their effects are probably small. This indirect approach factors medical insurance premiums into two parts:
Changes in the prices of medical care items covered by health insurance policies
Changes in the cost of administering policies, maintaining reserves and, as appropriate, profits.
Most expenditure for health insurance goes to the first item above, and reflects insurers’ payments for medical treatments. The CPI allocates this portion to the indexes that account for medical care items (i.e. physicians’ services). Thus, the weights for most of MCS indexes reflect out-of-pocket expenditures plus allocated health insurance benefit payments. (It is for this reason that provider’s reimbursements from insurance companies are valid prices for the CPI’s MCS indexes.)
Let’s backtrack through the above paragraph.
First, they don’t have a health insurance index. Here’s a big problem. If this was a small expense, there wouldn’t be a problem. However, health insurance is a large expense for most Americans. According to the most recent Kaiser Health survey, the average family yearly premium $2713. This is one of the largest expenses facing the average American family. And it’s cost has increased dramatically over the last 5 years.
Secondly, The weights in the CPI do not include employer-paid health insurance premiums. Here lies another problem. According to this Kaiser Health survey, health insurance premiums increased 10.9%, 12.9%, 13,9% and 11.2% and 9.2% in 2000-2005, respectively. Additionally, the average annual premiums for covered workers in 2005 was $10,880. In other words, this is a large expense the CPI does not consider.
So, two large expense are not included in the national CPI numbers.
So, what is the actual CPI? I don’t know. But I feel comfortable in saying it’s higher than reported by the BLS in their CPI numbers.