The Big Divorce

For decades a prominent and influential measure of government inflation mirrored general consumer inflation.  Then something happened.  In this article from our Nov-Dec 2013 edition of Fiscal Focus, we take a closer look at the split and why it matters for Minnesota.

Look under the hood of every major government fiscal study or report in Minnesota and you will find technical issues or assumptions of major consequence. By themselves, these issues may seem arcane and insignificant. But as with a car engine, tweaking these elements can have a major effect on output – and ultimately impact how the state operates and runs.

Measuring government inflation is one of those issues. In addition to their roles in and contributions to economic forecasts, pension valuation studies, and other fiscal reports, inflation measures are also pervasive in statute. In a 2009 information brief, the always-helpful House Research Department identified 44 statutes that reference which specific price index to use when adjusting dollar values in various state programs and laws. The brief also summarizes the three classes of inflation indexes and provides guidance on the appropriate applications for each. However, political squabbles still occasionally break out about which inflation measure ought to be used for this or that purpose. It’s definitive proof (if you needed it) that there truly is no topic, no matter how technical or mundane, that can’t be politically controversial under the right circumstances.

When evaluating government activity within the overall economy, analysts most commonly employ the “implicit price deflator for state and local government” (or mercifully “IPD” for short) as an inflation measure. Like the far more familiar Consumer Price Index or CPI produced by the federal Bureau of Labor Statistics (BLS), the Bureau of Economic Analysis (BEA – also a federal agency) calculates the IPD using national data. Unlike the CPI though, the calculations aren’t based on a “basket” of purchases made by government that’s held constant over time, but instead is based on actual government spending data.

The IPD is generally regarded as a “better” measure of government inflation for that simple reason: it captures actual government purchases. For a very long time, though, that distinction didn’t really matter. As Figure 1 shows, for three decades the CPI and IPD were largely indistinguishable, tracking each other exceptionally closely. But beginning around 1999, the trajectory of the IPD relative to consumer inflation trends began to change dramatically. Since then, the average annual growth rate in the government inflation measure has been nearly 1.5 times that of the measure for consumers (3.65% vs. 2.49%). In 13 of the last 15 years, the IPD has grown faster than the CPI and sometimes almost twice as fast. Over time, compounding that differential really starts to add up – as the figure also demonstrates.

Figure 1

Cumulative Change in Consumer Price Index and Implicit Price Deflator for State and Local Governments: 1973 to 2012

The questions are obvious. Why is this happening? Is this differential temporary? Or have we entered a new era where inflation in the cost of government services grows faster than overall inflation? If so, why and what might be the consequences for the future of state and local government budgets and spending?

Two Roads Diverged

Answering the basic question “why” isn’t easy. A big drawback – at least from a policy analyst’s point of view – of the IPD compared to the CPI is the lack of additional detail. Robust CPI data is available for regions of the United States and selected large metropolitan areas, such as Minneapolis- St. Paul. IPD data is only available for the country as a whole – creating a “one size fits all” measure that makes comparisons to national averages. Given that state and local governments across the country face varying degrees of inflationary pressures, local inflation rates can differ substantially from national averages – limiting the usefulness of the measure.

Moreover, data is available for the individual components within the CPI – for instance, if you want know the annual inflation rate between 1999 and 2012 for women’s footwear (0.35%), chicken (2.52%), pet food (3.01%), or practically any other consumer good – you can find it in the CPI data. In contrast, no one will drown in IPD-related data. In fact, you’d be lucky to fill your glass. Based on our own best efforts, the most anyone can do with the information the BEA makes available is to break the IPD down into its major components back to 1999.

Table 1 presents that perspective. The overall change in government inflation between 1999 and 2012 was 58.2% – or 3.6% per year. The single biggest component of the IPD – employee compensation – acted as an “anchor” and sat almost exactly at the overall average. Two of the other individual components, purchase of nondurable goods and gross investment in structures, have grown faster than the overall average.

Table 1

Changes in the Implicit Price Deflator for State and Local Governments and Major Components: 1999-2012

  Component CumulativeAnnualized
  Purchase of nondurable goods 100.1%5.5%
  Gross investment in structures 82.2% 4.7%
  Compensation of general government employees 58.3%3.6%
  Purchase of services 47.4%3.0%
  Consumption of government fixed capital 46.1%3.0%
  Purchase of durable goods (2.3%)(0.2%)
  Gross investment in equipment and software (8.6%)(0.7%)
  Less own account investment and sales to other sectors* (56.4%)(3.5%)
  Overall Change 58.2%
*Underlying numbers are negative, percentage change represents growth away from zero.
Source: Bureau of Economic Analysis, calculations by MCFE

The most difficult numbers in the table to interpret are the changes in “own account investment and sales to other sectors”, a term which requires a little unpacking. “Own account investment” represents spending by government on structures and software that it will use in future production. “Sales to other sectors” are largely charges for tuition and health/hospital services, but include other unspecified sales as well. BEA statisticians subtract these amounts when calculating net state and local government spending, and so growth in user fees and other charges for public goods actually act as a brake on the overall growth in government inflation. Put another way, growth in tuition and health care-related charges are working to keep government inflation in check.

How do government purchases match up with consumer purchases of the same general classification? As previously noted, IPD data limitations mean that comparisons of its components to other inflation measures can only go back to 1999, but conveniently that happens to be the era of concern. And the comparison provides some insights into the CPI-IPD differential. Inflation in the purchase of nondurable goods rose twice as fast for governments while inflation for the purchase of government durable goods fell more slowly than for consumers. Inflation in government purchases of services also topped growth in similar purchases for consumers, but here the differential is much smaller.

Obviously interpreting these figures requires major caution. After all, governments have different purchasing patterns than consumers do. But even if we throw food and beverages out of the CPI change– representing a very large portion of consumer purchases with very low inflation rates – the change in nondurable goods over this period climbs only to 55.4%. The differentials in these inflation figures provide real concern that inflation in government purchases of goods and services may be growing at faster rates than elsewhere in the economy. And the results do mirror related research findings of others. Noted labor economist Barry Bluestone found that between 2000 and 2008, the price of state and local public services nationally increased 50 percent faster than private services.

The People Factor

IPD figures also offer some insight into the effects of public employee compensation on government inflation, which is by far the largest component of the IPD, with somewhere between 50% to 60% of total state and local spending. As Table 1 indicates, inflation for compensation expenses – which includes not just salaries but also the cost of benefits such as insurance, pensions, and Social Security/Medicare – has grown at an average rate of 3.6% per year since 1999 for state and local governments across the country. During the period of relatively high IPD growth (3.5% or higher in every year but one) between 1999-2000 and 2007-08, compensation-related inflation was growing at a clip of 3.5% to 5.6% annually. Unfortunately, the BEA data doesn’t provide any insights into how much of the growth is attributable to wage pressures and how much to higher costs for health care and the effects of significantly higher contributions to public pensions. However, it’s hard to believe health care for active and retired employees and pension issues didn’t come into play.

All of this leaves more questions than answers with respect to the “big divorce” in inflationary measures. One plausible explanation for the differential is that a handful of very large (and not well-managed) states have dramatically skewed the overall government inflation numbers upwards. Remember, the IPD is a national data series based on aggregate spending totals in all 50 states. So actions by fiscally dysfunctional states like California can have a disproportionately major effect on the IPD. As an example, California’s 172% increase in pension expenditures per full-time employee between 2002 and 2008 was over 10 times Minnesota’s own growth rate. If efforts to restore public pensions to sound financial footing across the country as well as fund other post employment benefit liabilities are the primary force behind the divorce in trends, we may well be looking at substantially higher IPD inflationary measures for many years to come if costs per employee continue to climb at relatively high rates.

For many years we have argued on these pages and elsewhere that the Tax Incidence Study presents a much more complicated “tax fairness” story than most people think. Moving from calculations of effective tax rates to subjective assessments of fairness still includes assumptions and normative judgments despite the appearance of mathematical precision.

To illustrate our point, we offer three questions meriting consideration when mining incidence study findings for tax fairness conclusions. These questions involve both the nature of the taxes themselves and how Minnesotans actually think about fairness.

Why it Matters

One thing we can say with 100% certainty is that this issue has a major influence on tax and spending decisions in Minnesota. The use of the IPD has direct implications for levels of taxation and indirect implications for how Minnesotans perceive the adequacy of state and local government tax and spending levels. We highlight two areas.

Business and Cabin Property Tax Burdens – Since its inception the annual levy for the statewide property tax (the “state general tax”) on business and cabin properties has been adjusted based on the IPD. Based on our calculations, business and cabin owners have collectively paid $463 million more in property taxes since 2002 under an IPD adjusted levy than they would have under a CPI adjusted levy. If the past average inflation measure differentials remain for intact another 10 years, business and cabins would pay an estimated $651 million more in property taxes than under a CPI adjusted levy.

Public Perception of Spending, Budgets and Forecasts – Government inflation measures are also commonly used as discount rates to determine inflation-adjusted or real government spending changes over time. The use of the IPD, especially in the last decade, compared to the use of other inflation measures has had major implications for how Minnesotans perceive the adequacy of government spending and investments. For example, as we discussed back in 2010, the widely-reported State Auditor’s annual report on city finances found a 6.7% decline in real government spending over the period 1999-2008 using the IPD as an inflationary discount rate. If the CPI had been used the report would have found a 6.5% increase in real government spending over those same years.

In communicating information and perspective on government spending to citizens and the media, this is not a trivial matter. Nor is it limited to local government finance. Unsettling presentations of declining trends in real government spending in areas ranging from education to public safety often have a small footnote on the slide noting the use of the IPD as a discount rate. For proponents of higher levels of government spending, the trend divorce has been one of the most important and influential advocacy and communications-related developments of the past decade.

The issue may eventually find its way into debates about future levels of taxation and spending as well. Since 2002 inflation formally has not been formally included in the state’s economic forecasts but rather has been shown as a separate line item. Many, including the State Council of Economic Advisors, have criticized this practice arguing it is intellectually dishonest, misleading, and inconsistent with sound budgetary practices.1 It also has a major budgetary impact. For example, the state’s February 2013 economic forecast projected $854 million in estimated inflation (based on CPI) for the current FY 2014-15 biennium. Had legislators needed to raise an additional $854 million in revenues to balance the FY 14-15 budget, we estimate they would have needed to set the new fourth tier income tax rate at 11.38%, assuming no change in the income thresholds. Yet some argue that the use of the CPI actually understates a significant amount of government inflation and recommend application of the IPD to large chunks of the state general fund budget.

To us, the data suggests that policy makers and budget officials ought to get a better understanding of the reasons for the recent divergence between CPI and IPD growth, determine whether these trends are likely to persist going forward, determine how much of government cost inflation pressures are actually within government’s own control, and build this understanding into a reassessment of the use of inflation measures in the state budget generally and the use of the IPD specifically. We don’t have the specific formula for a “right” inflation measure but it seems clear to us that something tailored to a Minnesota-specific context that recognizes that some inflationary cost pressures are directly within government control are vitally important pieces.

  • 1 We believe a reasonable to fiscally responsible case can be made for continuing to exclude inflation from the economic forecast.  See "Inflation: In or Out?"  Fiscal Focus January-February 2007