The Congressional Budget Office earlier this week released a new report on the distribution of household income and federal taxes in the United States. The report contains data on household income from 1979 to 2011, which shows the impact of taxes and transfers (government income-support programs) on incomes. The data paint a different picture from the U.S. Census Bureau income data released earlier this year: the CBO shows a 47 percent increase in the median household income from 1979 to 2011 compared to a 5.3 percent increase according to Census Bureau data.

But is this latest snapshot of the distribution of U.S. incomes necessarily the right picture?

The CBO data account for a variety of factors including federal taxes, the decline in household size and the increasing importance of health insurance as a form of compensation, and transfer programs. It’s the combination of the last two categories—health insurance and transfer programs—that causes the most difficulty when measuring income.

The problem boils down to this question: how much is government-provided health insurance, Medicaid and Medicare, worth to a household? (The Affordable Care Act does not factor into this calculation, as that program did not expand health insurance coverage in earnest until 2013.)

CBO answered that question using the so-called “average cost” method. The calculation is quite easy: take all the spending on, say, Medicaid, and then divide that number by the number of Medicaid enrollees. That “average cost” is designated as the value of Medicaid to the household. Any increase in spending on the program gets registered as an increase in the value.

But CBO didn’t always answer the question this way. Prior to 2012, the agency calculated the value of government-provided health insurance by looking at the “fungible value” of health insurance. This process looks at how much money a household has compared to its basic needs for food and housing. If the household has more income than its basic costs, then the valuation process assumes that government-provided insurance would free up this money for other purposes. So that amount of money (up to the average cost) is called the fungible value.

The difference between the two methods is quite large. In their 2012 report explaining the change from the fungible value process to the average cost method, CBO noted the average cost method increased the average income of a household in the bottom 20 percent of the before-tax distribution by $4,600, or by about 25 percent. Using data from the 2013 report, which covers 1979 to 2010 and was released last December, we can see the average income of the bottom 20 percent grew by 49 percent from 1979 to 2010 using the average cost method. But using the fungible value method, the average income of that group grew by only 23 percent.

Sadly, the new 2014 report (with the latest 2011 data) doesn’t include income data using the old methodology. This means that going forward economists and policymakers will be have one less data set by which to evaluate income inequality.

Now this isn’t to say that CBO is definitely using the wrong method by dropping the average cost method. Nor is it clear that the fungible value is the correct method—after all, calculating the value of health insurance based on how much money is freed up relative to the cost of food and housing isn’t exactly intuitive. The average cost method has the merit of being consistent with the method of valuing employer-sponsored health insurance, which calculates the full value of employer-paid insurance into household income. In both cases, though, every dollar that employers spend on insurance is assumed to increase the households’ income on a one-for-one basis.

The main takeaway here is that we need to be cognizant of the difficulty of measuring the world around us. Ewe Reinhardt, a health care economist at Princeton University, admits that he and his colleagues in the profession don’t have the answer to this question yet. Government programs have certainly boosted incomes at the bottom, but we still aren’t sure by how much. If we want to understand how to improve the lot of households at the low-end of the income ladder, it’d be useful to know exactly what rung they are standing on.