The Trump administration’s higher education budget would eliminate the loan-interest waiver given to college borrowers. Ending the subsidy, it is argued, would reduce tuition inflation: A Federal Reserve study finds that subsidized interest hikes tuition. Between 1985 and 2011, average tuition nationwide increased 498 percent—more than four times the rate of general inflation (114 percent) as measured by the Consumer Price Index (CPI).

But is tuition inflation as big a problem as these statistics suggest? Some in higher education deny it through using, and misusing, their own “CPI”—the Higher Education Price Index (HEPI). Thus, the following was recently reported about the tuition decision of the Texas A&M System: “Regents voted to increase tuition at their 11 universities by 3.7 percent. That amount, they said, matched the rate of inflation in the higher education world as measured by the Commonfund Institute” (emphasis supplied).

The qualifier in the assertion that the increase matched “inflation” is what follows—“as measured by the Commonfund Institute.” Commonfund took on the task of maintaining HEPI in 2005. Commonfund argues that HEPI yields a “more accurate indicator of cost changes for colleges and universities than the Consumer Price Index.”

HEPI clearly seems helpful for a school’s internal budgeting purposes in ways that the CPI is not. However, HEPI (along with a later index, the Higher Education Cost Adjustment [HECA]) is “often misused” according to a study by Andrew Gillen and Jonathan Robe.

Gillen and Robe highlight the issue raised at Texas A&M. They find that HEPI and HECA “were not designed to measure inflation in general [like the CPI] but rather just inflation of higher education inputs.” Hence, they label it “unfortunate” that the “most frequent public use of HEPI and HECA is to discount tuition rates over time and then to argue that real tuition has not been increasing as much as other price indices, such as the CPI, would indicate” (emphasis supplied). Such a line of reasoning, they argue, is “completely inappropriate.”

More simply stated, economists adjust college tuition for inflation in order to glean how trends in tuition compare to the overall economy. Doing so requires an inflation measure that takes into account the whole economy, such as the CPI. HEPI and HECA, in contrast, deliberately look at prices for education goods solely. For this reason, HEPI and HECA cannot tell us how tuition trends compare to general price trends. Instead, these two indices were crafted to compare tuition trends to cost-of-college trends. This is doubtless a valuable measure, but it is not a gauge of inflation. Only after adjusting college tuitions for inflation using the CPI can HEPI or HECA be employed “to explain why tuition grew more or less slowly than other variables that were also adjusted using the CPI.” The CPI establishes “what happened (e.g., tuition went up by five percent)”; only then can we use HEPI or HECA to argue “why it happened (e.g., tuition went up five percent in part because the price of inputs for higher education went up four percent).”

Gillen and Robe demonstrate that these inappropriate uses of HEPI and HECA add up over time. According to the CPI, average annual inflation from 1990 through 2009, was 2.80 percent. But for HEPI, it was 3.68 percent. HECA is not quite as far off, standing at 3.23 percent. We see here that HEPI and HECA “systematically understate the true increase in tuition over time. . . . Over the entire twenty year period, while the CPI indicates real tuition increased by 86 percent, the HECA indicates it went up by a significantly lower 72 percent and the HEPI by only 59 percent.”

This is why it is illegitimate to employ HEPI and HECA in an attempt to argue, for example, that “a five percent increase in costs somehow requires universities to obtain five percent more revenue from tuition and state appropriations.” Gillen and Robe rejoin that measuring the “the cost of maintaining the status quo does not in any way legitimize the status quo. If a university is spending money inappropriately, forecasts of future revenue needs using HECA and HEPI will merely report an estimate of the budget needed to continue wasting money.”

While the CPI is not without its own shortcomings, its merit lies in the fact that it measures real prices through sampling from a wide basket of goods and services purchased by households, whereas HEPI does not. Thus, HEPI cannot provide us the real price of college tuition.

When we try to use HEPI to measure tuition inflation—as the Texas A&M leaders, quoted above, do—we end up with a dollar’s value measured only in higher education goods. This would be valid if the dollars used to pay tuition could be spent only for college. But they are not. The dollars used for tuition are spendable on many goods and services, and derive their value from those purchases.

How, then, would HEPI be properly used? HEPI reveals certain prices changing over time, just as a newspaper might go from one dollar in 1950 to two in 2017. But, as is the case with the newspaper, we need to make sure that HEPI’s prices are all the same unit in order to make proper comparisons. We would use the CPI to do that with newspapers—and everything else. Only then can we properly use HEPI, which was developed to provide a convenient bunching of higher education costs. It was not developed to tell us what a dollar is worth.

Finally, Gillen and Robe find an additional danger in the use of HEPI: It is “self-referential in the sense that it relies upon labor costs (faculty, administrative, and clerical salaries) which are “influenced by university policy decisions.” Citing economist Richard Vedder’s work, they report that HEPI’s self-referentiality undermines clarity, “because if administrative salaries rise, then the Higher Education Price Index rises. Colleges can give their employees huge salary increases, claim that ’higher education costs are soaring,’ and demand larger government subsidies, etc., as a consequence.”

The CPI does not suffer from this defect in measuring the prices consumers pay “because consumers cannot change the prices of the goods and services they purchase.” As a consequence, using HEPI disincentivizes schools “to control costs because colleges can decide to increase staff salaries and, by so doing, cause their overall price index (as measured by the HEPI) to rise as well. Their bottom line: “HEPI increases in part because university leaders decide to pay themselves more.”

To be sure, Gillen and Robe place no blame for this confusion on HEPI and HECA’s creators. Both indices are helpful when limited to the internal budgeting purposes for which they were originally created. Rather, the blame lies with those who seek to use these indices beyond their purposes in deceived and/or deceiving fashion.