“Fix It or Junk It”

Editor’s note:  As long-time MCFE research director Aaron Twait leaves to embark on his new career, I asked him if there is was one parting recommendation he wanted to leave to the world of state government.  The answer: some frank advice regarding the Minnesota Price of Government.

It has been described as “the state’s main transparency method and tool for communicating the total cost of state and local government to citizens.”[1] And as a regular feature of state government press releases, editorial page commentary, Twitter fights, and legislative budget debates, for over a quarter century the Minnesota Price of Government (POG) report has lived up to the billing.

Created with broad-based political support, including the endorsement of Minnesota’s business community, this measure of state and local government’s claim on Minnesotans’ income has served as a major evidence point in debates over budgets and acceptable levels of government taxation.  However, design and maintenance problems with the report call into question whether it deserves the attention it receives.  “Fixing it” involves two relatively simple repairs.  Failure to do so likely does more harm than good by rendering long-term comparisons and trends analysis that is distorting at best and misleading at worst.

Fix #1: Change the definition of income

“It answers the question: How much do Minnesotans pay to state and local governments in total?”

Minnesota Management and Budget, Price of Government Report

One of the most valuable and useful features of the POG report is perhaps its least discussed feature – the breakdown of own source revenue collections by state and local governments.  The POG offers important insights into growth rates and comparative trends in state and local revenue collections over time showing how public finance in the state is evolving.  But this detail plays second fiddle to the POG’s headline measure – total state and local own source revenue divided by state personal income – which provides perspective on how big a claim state and local revenue collections have relative to one measure of Minnesotans’ income.[2]

As our readers know, the use of personal income in the denominator as the basis for the POG, as mandated by state law, has troubled us for a long time.  Over the years we have reiterated our concerns with the current practice in special reports, blog posts and many editorial page responses/rebuttals.  Rather than rehash this whole topic once again, the following is a reprise of our main points:

  • Contrary to widely held belief, the federal government’s “personal income” figure isn’t a simple measure of what people earn.  In fact, it includes over $70 billion of Minnesota non-money income[3] that cannot be used to pay government taxes and fees.  Only slightly more than 75% of state personal income is money income that can be used to finance government spending (see accompanying table).
  • Conversely, “personal income” excludes capital gains and retirement plan withdrawals which are largely subject to the state income tax regime.  According to the IRS, this totaled $25.7 billion for Minnesotans in calendar year 2016.  In other words, taxes paid for with capital gains and pensions are included in the POG, while that income itself generally isn’t.
  • The government transfers component of Minnesota personal income has been by far its fastest growing piece over the last 30 years.  Put that together with the fact that large government transfer programs provide considerable non-money “income” (e.g. the value of Medicaid and Medicare benefits paid to Minnesotans), and this creates a problematic and disturbing circularity in interpreting and employing the POG in budget debates.  Relatively faster growth in government transfer payments boosts personal income growth which creates the appearance of government getting “cheaper” – and becomes state-sanctioned evidence used to ask more from taxpayers to spend on, among other things, bigger and more generous government transfer programs.

For these reasons, a modified personal income base that excludes non-money income that can’t be remitted to government, and which also includes those income streams excluded from personal income that can be used to pay taxes and fees, should be the denominator in price of government calculations. 

Fix #2: Take greater methodological care in maintaining the POG

What we have not previously discussed are the research faux pas embedded in existing POG reporting.  These invisible, highly technical matters can have significant impacts on how POG trends appear.  In fact, we find methodological inconsistencies introduced into the Price of Government report over time explain somewhat more than half of the decline in the POG since the metric was first reported for FY 1991.

The first irregularity concerns a decision to switch the personal income denominator from a calendar year to a fiscal year basis several years ago.  Through FY 2011 the POG calculations were based on personal income data running on a calendar year basis.  Beginning with FY 2012 the POG calculations have been based on personal income running on a fiscal year basis.

Changing from a calendar year to a fiscal year basis is not in and of itself problematic, and there may be good reasons for doing so, (although we don’t know what the rationale was since the POG reports do not disclose methodological changes).  However, we now have one set of POG calculations (through FY 2011) that are calculated using income from January 1 through December 31, and a second set of calculations (beginning FY 2012) that are calculated using income from July 1 through June 30.  Because the annual personal income starting point is now six months later, this change has created a discontinuity in the personal income time series data that is distorting long-term trends in the measure.  To maintain comparability across the entire time series, the new methodology should have been retrofitted to the entire POG series, but MMB to date has not done this.[4]  As a result, personal income is artificially increased by about 2% for the post-FY 2011 POG calculations – thereby artificially lowering those POG results relative to the earlier period. 

The second concern relates to the federal Bureau of Economic Analysis’ (BEA) processes for ensuring the reliability of its data.  The Bureau regularly revises the statistics it publishes on items including gross domestic product, personal consumption expenditures, and – notably for purposes of the POG – personal income.  These revisions are generally made because the Bureau’s initial figures are based on incomplete or preliminary data, and therefore are subject to change as final data becomes available.

However, every five years the Bureau issues “comprehensive” updates to these statistical measures.  According to the BEA, “[the] comprehensive updates and, to a lesser extent, annual updates provide the opportunity to introduce data revisions, changes in definitions, and updated statistical methods from the comprehensive update of the National Income and Product Accounts (NIPAs) and the Industry Economic Accounts; to incorporate newly available and revised regional source data; and to introduce improvements in regional statistical methods.”[5]  Importantly, the Bureau applies changes in definitions, classifications, and data sources that comprehensive revisions initiate retroactively, often to the entire time series data set (which for personal income, dates back to 1929).  Therefore, these revisions can make noticeable changes even to decades-old personal income amounts.

To maintain comparability in the time series Price of Government results, which date back to FY 1991, it is important to apply any changes BEA’s revisions – comprehensive or otherwise – make to the personal income data for all years.  MMB does not appear to do this.  In the February 2019 POG report, for example, the personal income data being used for the FY 2012 through FY 2016 calculations matches the BEA quarterly data published in September 2018.  For years prior to 2012, however, the data used in the POG does not match what the BEA has published, and is likely data from some earlier revision(s).

At one time MMB’s practice was to retrofit revisions to the historical data to the entire Price of Government results to maintain comparability across the time series.  It is unclear when MMB abandoned this practice, but based on a then-Department of Finance memo in our files, we know it was sometime after 1997.

Figure 1 compares how the Price of Government, as of the February 2019 edition of the report, has changed relative to the base year of FY 1991 as published by MMB, and how it would look if the personal income denominator had consistent definitions all the way through the time series.[6]  Using published personal income data the Price of Government for FY 2018 is 92% of its FY 1991 level (15.2% compared to 16.6%).  Using consistent time series personal income data, the Price of Government for FY 2018 is 97% of its FY 1991 level (15.9% compared to 16.4%). 

For something that has so much influence on public perception about government, it seems to us more care and attention to detail is warranted in constructing the POG than it has received.  We also strongly encourage MMB to return to its prior practice of using calendar year personal income data when calculating the Price of Government report as part of this fix.  The calendar year (annual) data BEA produces provides much more underlying detail on the makeup of personal income than does the quarterly data currently used to establish fiscal year personal income.  This detail allows for a more thorough analysis of income components which, as we noted earlier, also affect the interpretation of the measure.  If the Price of Government calculations cannot be dissected to throw additional light onto why they are changing, the usefulness of the report becomes seriously limited.

And for Consumers: Recognize its Limitations

Several years ago, following a presentation by MMB on the latest Price of Government report, a senior senator paused and asked, “What exactly are we supposed to do with this information?”  It’s an excellent question.  Like any measure, the Price of Government is simply a number.  It’s up to policymakers, advocates, and the general public to interpret and understand what the number, and its changes over the long-term, mean.

Offering interpretations of the report’s findings and its policy implications has never been a problem.  It may be intended as a transparency tool but it functions as weapons-grade political ammunition in debates over state tax and spending policies.  Recently it’s been used as prima facie evidence for underinvestment in public goods and services, the affordability of higher taxes, the lack of harm imposed on the state economy for raising taxes, and for questioning the justification for reducing taxes.  The problem is at any point in time the POG is not just about government revenues, it has just as much to do with macroeconomic conditions and the role the federal government is playing. 

To illustrate this problem, consider another possible “standalone measure” that might be used to interpret and evaluate state and local own source revenues in a different context: the change in spending capacity created, rather than the claim on income.  If we control state and local governments’ ability to spend for two significant cost drivers – population growth and inflation[7] – inflation-adjusted per capita spending grew by 19% between FY 1991 and FY 2018.  In other words, after accounting for a 27% jump in population and a 122% increase in government inflation, Minnesota state and local governments have 19% more money to spend on initiatives implemented since that base year.  (This does not include federal government transfers and grants which, if added to this analysis results in a 26% increase in state and local real per capita spending capacity.)  State and local government representatives would undoubtedly object to this being the singular and definitive measure to perceive and evaluate fiscal policy, and rightly so, because the nature of spending capacity and government spending has changed dramatically over time due to changing demographics and many other issues.

As enticing and tempting as it is to make big policy conclusions based on a single benchmark measure like the POG, it is no substitute for careful and thoughtful vetting of the policy merits of state tax and spending proposals.  The POG may inform debate but it is in no way a replacement for it.  And without a fix, these informed debates will be based on a foundation communicating a fiscal illusion.


[1] Minnesota Management and Budget staff testimony before the Legislative Commission on Fiscal Policy, February 22, 2013 meeting.

[2] According to the federal Bureau of Economic Analysis, “personal income” includes the income Minnesotans’ receive from paychecks, employer-provided supplements such as insurance, business ownership, rental property, Social Security and other government benefits, interest, and dividends.

[3] This and other instances of the use of “money income” are not references to the Census Bureau’s “money income” measure.

[4] When MCFE switched several years ago from “personal income” to a modified personal income basis as advocated here in assembling our own How Does Minnesota Compare report, we created a similar discontinuity in trend reporting.  Resources and time prevent us from going back 50 years to adjust the entire series.  However, in HDMC, we highlight the rationale for the change, specifically do not provide a historical trend analysis dating before the change, and the tables themselves indicate to anyone attempting to compare current and historical reporting that the metrics have changed.

[5] “Preview of the 2018 Comprehensive Update of the Regional Economic Accounts”, Survey of Current Business, U.S. Bureau of Economic Analysis, August 2018.

[6] In this case, we use calendar year data from the September 2018 revision, since that is the revision in effect as of the February 2019 economic forecast date.

[7] Using the implicit price deflator for state and local governments, which averaged 3.9% annual growth during this period.