January 13, 2005

Why TABOR’s Growth Formula Doesn’t Work
By David H. Bradley, Nicholas Johnson, and Iris J. Lav

PDF of full report

Related Report:
  Is Colorado’s TABOR Creating Jobs?
  Public Services and TABOR In Colorado

View Related Reports

If you cannot access the files through the links, right-click on the underlined text, click "Save Link As," download to your directory, and open the document in Adobe Acrobat Reader.

Limiting the growth of state revenue collections and state expenditures to a population-growth-plus-inflation formula is a central provision of a new generation of “tax and expenditure limits” — or TELs — now being promoted by national anti-government groups.  Under such proposals, state constitutions would be amended to bar state and local expenditures from rising at a percentage rate that exceeds the rate of growth of state population plus an inflation factor.  A population-growth-plus-inflation limit is central to Colorado’s “Taxpayer Bill of Rights,” arguably the strictest TEL in the country and the model for TELs currently under consideration in several states.

Such formulaic limitations may sound reasonable, but are actually a recipe for sharply reduced public services and an impaired ability to respond effectively to public needs, federal mandates, and changing circumstances.

There are a number of reasons that a population plus inflation limit leads to such a dramatic scale-back in government services.

The “Ratchet Effect”:  Increasing the Severity of the Population-plus-Inflation Formula

Colorado’s TABOR and many current TEL proposals in other states contain a feature that causes public services to be cut even more than they would under a simple population-growth-plus-inflation formula.  This feature is known as the “ratchet effect.”  With the ratchet effect, the population growth plus inflation adjustment is applied to the amount of actual expenditures or revenue in the prior year (rather than to the amount of allowable expenditures or revenues).  When state budgets grow slowly or fall, as in the recent fiscal crisis, actual spending or revenues are likely to be lower than the level permitted by the formula.  If this lower level becomes the new base to which the population growth and inflation adjustment is applied, then the level of public services is permanently ratcheted down. 

Consider a hypothetical state with $1 billion in revenues in 2001 and with population growth plus inflation equaling 5 percent annually.  With no ratchet effect, by 2005 allowable revenue would be $1.22 billion (reflecting four years of 5 percent growth, compounded).  But if actual revenue collections declined in the first year by 5 percent, allowable revenue collections in years thereafter would be calculated from that new, lower base.  As a result, by 2005, even if the economy and tax base fully recovered, allowable revenues could not exceed $1.1 billion — a very significant difference of 10 percent.

The ratchet effect typically is not explicit, but rather hidden within the wording of a TABOR proposal.  Any TABOR proposal must be carefully scrutinized to determine whether it contains a ratchet effect.

Population Growth Plus Inflation in Colorado

The population-growth-plus-inflation formula is a key reason why Colorado’s TABOR amendment — a 1992 constitutional provision — has caused severe problems for the provision of public services in that state.

After 12 years of TABOR, Colorado has among the nation’s poorest-funded — and poorest-performing — public services.  For example, Colorado ranks 47th in K-12 education funding as a share of state income.  The ratio of teacher salaries to average private-sector earnings is lower in Colorado than any other state.  Child immunization rates are also lowest in the nation.  More poor children lack health insurance in Colorado than in all but five other states, and three-quarters of Colorado pediatricians won’t treat Medicaid patients because reimbursement rates are so low.  High school graduation rates, support for public colleges and universities, and access to prenatal care all have fallen since 1992. 

Just since 2001, as a result of the formula (combined with the ratchet effect described in the box on the previous page), Colorado policymakers have been forced to make $1 billion in spending cuts.  Unlike in other states, TABOR’s limit will not allow those services to be restored;  in fiscal year 2006 Colorado is scheduled to refund $459 million to taxpayers while being required to cut an additional $263 million out of its budget. 

Source: Nicholas Johnson and David Bradley, Public Services and TABOR in Colorado, Center on Budget and Policy Priorities, January 2005


The Population-Plus-Inflation Benchmark and the Cost of Government

Central to the strictest TELs, such as Colorado’s TABOR amendment, and to the new TEL proposals in other states, is a provision limiting the rate of growth of government spending to the sum of the population growth rate and the inflation rate, as measured by the Consumer Price Index-All Urban Consumers (CPI-U).  Mathematically, this is equivalent to saying that government spending per state resident, adjusted for inflation, cannot increase.  In calling for the adoption of strict tax and expenditure limits as a way to limit government, Barry Poulson of the conservative Americans for Prosperity Foundation (AFPF) notes that to be effective a TEL should “impose a stringent limit on the revenue that governments can keep and spend, such as the sum of inflation and population growth.”[1]  The model TEL legislation from the American Legislative Exchange Council, which urges lawmakers to enact conservative policies that reduce the size of government, also uses a spending growth formula of change in population plus CPI-U.

A casual observer of state finances might believe that such a formula is quite reasonable.  After all, this formula not only allows the public sector to maintain its current level of expenditures, but allows for some growth in nominal terms.  Nevertheless, a one-size-fits-all population-growth-plus-inflation formula is deeply flawed.  Such a formula is likely to hinder the ability of state governments to continue to provide services at their current levels.  A state that truly limited growth in its budget to population growth plus inflation would need to cut back or eliminate a range of services in most years.

There are several reasons why states cannot provide a constant level of public services under a population-plus-inflation formula.

It is important to note that all state programs — not just those with cost pressures exceeding the population-growth-plus-inflation level — are threatened by a rigid population-growth-plus-inflation limit.  This is because such limits typically cover nearly all areas of state and local spending.  So, if one spending area is forced to grow faster than the rate allowed under the limit (for instance due to court order, federal mandate or popular demand), then another spending area must grow at a slower pace — which is to say that in terms of the level of service provided, that second spending area must actually shrink.

Local Governments and TABOR

Most current state TEL proposals would apply the population plus inflation formula to local government budgets as well as to those of the state government.  This causes two problems for local governments. 

  • Local governments, like state governments, face costs that tend to rise faster than population plus inflation, for many of the same reasons:  rising health care costs, changing demographics, and changing and unanticipated needs.
  • A population-plus-inflation formula at the state level may increase the likelihood that states will cut aid to local governments, or pass on mandates and responsibilities to localities without adequate funding, in order to make room for other fast-growing budget items.

For these reasons, organizations of local governments— which represent elected officials across the political spectrum — have expressed particular concern about such proposals.  For instance, the National League of Cities has taken a strong stance against Colorado-style TABOR proposals and the Wisconsin Alliance of Cities has been among the leading voices against a TEL proposal in that state.


Realistic Budgeting for Public Services

Proponents of TABOR-type tax and expenditure limits sometimes contend that a growth formula based on population plus inflation would be adequate to maintain public services at a roughly constant level.  But researchers long have recognized that the services provided in the public sector, such as education, health care, and law enforcement, tend to rise in cost faster than many other goods and services in the economy in general.  This analysis was first put forward by economist William Baumol, who pointed out that technology and productivity gains may make goods cheaper to produce, but the services that government provides are different.  Baumol said public services typically rely heavily on well-trained professionals — teachers, police officers, doctors and nurses, and so on — and technology gains do not make these services cheaper to provide.[3]  It may take far fewer workers to build an automobile than it did 30 years ago, but it still takes one teacher to lead a classroom of children.  (In fact, as education has become increasingly important, the trend is toward more teachers per pupil, not fewer.)  Doctors generally still see patients one by one, and nursing care remains labor intensive despite technology.

For this and other reasons, “inflation” as commonly measured does not accurately reflect growth in the costs to government.  The most commonly used measure of inflation and the measure preferred by those pushing strict tax and expenditure limits is the “Consumer Price Index-All Urban Consumers (CPI-U),” which is calculated by the U.S. Bureau of Labor Statistics.  The CPI-U measures change in the total cost of a “market basket” of goods and services purchased by a typical urban consumer.  Since a typical urban consumer spends a majority of his or her income on housing, transportation, and food and beverages, those items are the primary drivers of the CPI-U.  By contrast, governments spend revenue primarily on education, health care, transportation, and corrections.  In short, the market baskets of spending are entirely different.  (See Table 1.)

Table 1
Differences in the CPI and Actual State Expenditures

Relative Importance of Major CPI Components, 2003

Spending Categories as a Share of Total State Own-Source Expenditures, FY 2003







K-12 Education






Food and Beverages


Higher Education


Medical Care










Public Assistance










Source: U.S. Bureau of Labor Statistics, http://www.bls.gov/cpi/cpiri2003.pdf; NASBO 2003 State Expenditure Report.  “Other” state spending includes the State Children’s Health Insurance Program.

Because governments purchase services in economic sectors that are less likely to reap efficiency and productivity gains found in other sectors of the economy, they face a different cost structure than urban consumers.  Table 2 below compares the change in the CPI-U for all items to components of the CPI-U that are roughly analogous – though not strictly comparable — to the major categories of states’ purchases.[4]  While the overall CPI-U increased by 27.3 percent from 1993 to 2003, costs in the three of the largest categories of government spending increased by much more.  The basket of goods that governments purchase is heavily weighted with items that have seen significantly greater cost increases in the past decade than the items in the basket of goods that consumers purchase.

Limiting the growth in government spending to the rate of growth in general inflation will not push down the rate of growth in the cost of medical care or education.  Many factors beyond the control of any one state are causing the rapid growth in these areas.  Instead, limiting state expenditures will cause a sharp drop in the amount of medical care and education services a state can support, and reduce its quality.  This is what has happened in Colorado.[5

Caps Won’t Stop Growth in Health Care Costs

When people say state spending is growing “out of control,” they often are referring to the rapid rise in the cost of Medicaid in recent years.  But Medicaid is not growing out of control.

Medicaid costs are growing at less than half the rate of growth of private health insurance on a per-person basis.  Medicaid is also more efficient that private insurance, with lower per-person costs.

Growth in Medicaid spending reflects the problem of rising health care costs that is affecting all health insurance, both private and public.  The growth in health care costs is largely driven by advances in medical technology.  It is a major national policy challenge, and only system-wide approaches to health care costs will address it.

Moreover, Medicaid faces two challenges that private insurers do not.

  • Increased enrollment during a weak economy, as more people become unemployed or lose employer coverage.

  • Filling in gaps in Medicare coverage.  More than 40 percent of Medicaid spending is on services for Medicare beneficiaries — primarily elderly and disabled people.  Costs have been shifting from the wholly federally paid Medicare program to Medicaid as medical practice has changed, reliance on hospital care has decreased, and patients increasingly use other services like prescription drugs.  (Under the new Medicare law, states will remain responsible for most of the cost of prescription drugs for the population that is eligible for Medicare and Medicaid.)  

No state can solve these larger problems on its own.  Only the federal government can stop the shift in costs from Medicare to Medicaid.  And it will require some combination of the federal government and the health care industry to bring down the annual rate of increase in health care costs in the country.

Many states already have, however, taken actions to improve the efficiency of their Medicaid programs through promoting disease management, managed care, prescription drug cost containment and similar policies.  Beyond that, states have few good options to bring down the rate of growth in Medicaid costs.

Policies such as reducing eligibility or imposing cost sharing on families living at the poverty line unfairly shift the burden of a system-wide problem onto the most vulnerable members of society.  And such policies are often counterproductive because they increase the ranks of the uninsured.  Sick people cost more when they are uninsured — and receiving care in emergency rooms — than when they are covered by Medicaid.

Just as the general rate of inflation fails to capture actual price changes faced by governments, overall population growth fails to capture changes in the level of need for services. Table 3 provides a comparison of growth factors affecting government costs.  While total population grew by a total of 15.4 percent from 1990 to 2002, other areas of state budgets that are the determinants of the need for services and expenditures grew by much larger percentages during this period.  For example, Medicaid enrollment of the expensive-to-serve population of elderly, blind, and disabled rose nearly 70 percent from 1990 through 2002.

Table 2

Inflation Factor

Percent Change, 1993-2003

CPI-U, All Items


CPI-U, Medical Care


CPI-U, Medical Care Services


CPI-U, Prescription Drugs and Medical Supplies


CPI-U, Education


CPI-U, Transportation Services


Source: U.S. Bureau of Labor Statistics.
Note: To provide comparable analysis, comparisons start in 1993, the first year for which education inflation data are available.

In the future, the growth of subpopulations may take on even greater significance in determining the cost of providing state services.  In particular, the elderly population will be rising rapidly.  According to Census Bureau projections, the share of the population aged 65 and older is projected to rise from 12.4 percent in 2000 to 20.4 percent in 2040.  While the total population is projected to increase by 39 percent from 2000 to 2040, the population aged 65 and older is projected to increase by 128 percent during this period.  Put another way, while total population growth is expected to increase at an average annual rate of 0.8 percent from 2000 to 2040, the elderly population is projected to increase at a rate more than double that — 1.9 percent.

Table 3

Determinants of Budget Costs

Percent Change, 1990-2002



State Prison Population


Vehicle Miles Traveled


K-12 Enrollment


Disabled Children in Schools (IDEA)


Medicaid Enrollment – Elderly and Blind/Disabled


Medicaid Enrollment – Children and Adults


Public University Enrollment*


Source: U.S. Bureau of the Census, U.S. Bureau of Justice Statistics, U.S. Federal Highway Administration, National Center for Education Statistics, Centers for Medicare and Medicaid Services. *Data through 2001.

When TABOR proponents argue that spending limitations will not shrink government, they imply that the limits will allow government to continue providing services at current levels.  The reality is that spending increases limited to increases in population and general inflation will result in a reduction of service provision over time.  For example, if caseload in the average program grows just 1 percent faster than general population and the cost of providing public services on average grows just 1 percent faster than general inflation, within ten years state government would be providing services to 20 percent fewer people than it would without the limit, or would be providing a level or quality of services 20 percent lower.

A formula that fails to take into account the cost factors in Tables 2 and 3 would force a state to scale back public services.  Figure 1 shows trends in state own-source spending under two scenarios — actual spending and spending limited to the growth of population plus CPI.  From 1990 to 2004, total state own-source spending rose slightly, from 6.6 to 7.0 percent of gross domestic product (GDP).  If state own-source spending had been limited to changes in population plus inflation over this same time period, total spending would have declined from 6.6 to 5.6 percent of GDP.[6]

How Might States Cut $162.7 Billion?

If a population-plus-inflation spending limit had been in effect for total state spending since 1990, state spending in 2004 would have been $631 billion rather than the actual figure of $793.7 billion.  In other words, states would have had to cut $162.7 billion from their annual budgets.  To place that amount in context, $162.7 billion equals:

  • 78 percent of all state elementary and secondary school spending, or

  • All state Medicaid and transportation spending combined, or

  • 1.6 times all state higher education spending, or

  • 13 times all state public assistance spending, or

  • 4 times all state corrections spending, or

  • 60 percent of all other state spending.

Source:  CBPP calculations from NASBO data.
Source:  CBPP calculation of NASBO State Expenditure Report data, various years.

If a population-plus-inflation growth formula had been in place in all states from 1990 to 2004, total state own-source expenditure in 2004 would have been $631 billion, which is $162.7 billion, or 21%, less than actual spending in 2004.  Public services would have been cut far below their current levels.  (See box.)



There is another basic flaw with the population-plus-growth formula:  It fails to take into account the possibility that court, voter, or federal mandates may require a state to take on new responsibilities.  Mandates, whether internal or external, may increase the responsibilities and costs borne by the state without any proactive policy change on the part of state lawmakers.  Rigid tax and expenditure limits, such as those based on population change and inflation, inadequately reflect the potential costs that may be imposed on state and local governments.

Court mandates, particularly for K-12 education funding, in numerous states have increased costs to state governments in recent years.  In the past 35 years, there have been court challenges to the adequacy and equity of K-12 education funding in 44 states.  Courts in Connecticut, Kentucky, New Hampshire, New Jersey, Ohio, Texas, and Vermont, among others, required states to make education funding more equitable and/or adequate.  These court mandates resulted in increased state spending on education as states invested greater resources in poorer school districts and shifted part of the financing of schools from localities to the state.  For example:

Voter mandates and initiatives have also increased costs to the several states, both through funded and unfunded initiatives.  Ballot measures often do not have dedicated funding sources or even specific cost estimates.  By definition a population-plus-inflation TEL does not provide funds for policy changes, but only for the current service levels.  When confronted with new voter demands, policymakers must accommodate voter mandates by squeezing funding for existing programs or by getting voter approval for a tax increase.

States are also shouldering a larger share of health care costs for low-income people because of the cost shift from Medicare (a federal program) to Medicaid (a joint program in which states bear an average of 43 percent of the costs).  States are expected to spend around $28 billion in state and local funds from fiscal year 2002 through fiscal year 2005 to provide prescription drugs to low-income elderly and disabled beneficiaries eligible for Medicare and Medicaid.  While the new Medicare law enacted in 2003 will cover under Medicare some of the cost of prescription drugs for all Medicare beneficiaries beginning in 2006, states will not realize substantial savings.  The new law includes an unprecedented provision that will require states to reimburse the federal government for most of the cost of prescription drug coverage for individuals who are eligible for Medicaid as well as Medicare.

The interaction between federal unfunded mandates and underfunded expectations and strict state tax and expenditure limitations may impede the ability of states to implement their own citizens’ policy priorities.  If states do not exempt all federal mandates from TEL limitations, any additional cost imposed from the U.S. government will necessarily cut the revenue available for in-state programs.  As noted above, if one spending area must grow faster than the rate allowed under the limit due to a federal mandate or “underfunded expectations,” then another spending area must grow at a slower pace to accommodate the TEL caps.


New Initiatives

State governments need to be able to change with the times.  What is considered an adequate level of public services at one point in time may seem inadequate 30 years later.  Voters and legislators may adjust their policy preferences to improve or expand government investment and program coverage.  In the 1990s, there were a number of examples of such changing policy preferences:

The Slow Squeeze of a TEL Formula

It may be a period of time — years, even decades — before the reductions in public services shown elsewhere in this paper as a consequence of a strict TEL actually result in reduced services — and even longer, perhaps, before a decline in the quality of life becomes apparent.  There are several reasons to expect such delayed impacts.

  • Spending limits typically rise at a rate only slightly lower than the cost of providing services, perhaps a difference of one or two percentage points.  Over time, however, the difference grows dramatically.  A one- or two-percentage point difference every year can translate into a 13 to 26 percent gap over the course of a dozen years.

  • Short-term deferrals can become long-term problems.  Any time that a state needs to cut spending, it looks first to cut items that can be deleted without a short-run impact on the provision of services.  For example, states may defer routine maintenance items, capital improvements, staff training, or other investments in infrastructure or workforce.  Such changes may help balance budgets in the short run, but can be costly in the long run.

  • Accounting maneuvers are only effective in the short-run.   Even under a tightly worded limitation, policymakers may be able to find a way to push spending into a future fiscal year in order to avoid the caps — a strategy that defers but cannot avert the financing squeeze.  For several years under its revenue limit, for instance, Colorado accounted for tax rebates in the fiscal year following the year in which revenues exceeded the limit — a strategy that worked as long as revenue collections remained strong, but which led to even greater budget cuts when revenue collections slowed in the recession.

  • Fluctuations in the formula itself may provide unexpected “room” in some years.  Colorado’s spending formula is based on population growth plus the Denver-area inflation rate.  As it happens, local consumer prices — particularly the cost of housing — rose rapidly in the mid-1990s due to a Denver-area housing crunch.  So it was not until 1997 that state revenues even began to hit up against the limit.  Of course, the opposite can happen also:  in a housing slump, consumer prices might hardly rise at all, further holding down the limit.

  • When public expenditures are investments, it may take many years for the harm from lack of investment to be evident.  Much state spending is intended to have long-term impacts.  Studies show, for instance, that early childhood spending has great benefits for a state that do not begin to show up for many years.  Infrastructure spending is another example.  Certain spending, therefore may not affect a state’s quality of life for decades.


Unanticipated Needs

State governments inevitably face spending needs that cannot be anticipated.  Natural disasters, public health emergencies, economic changes, and other such occurrences place expensive but unexpected demands on state and local governments.  In the last few years, for instance, states have faced:



Formulaic tax and expenditure limits are bad public policy.  Tax and spending limits based on population changes and inflation will hinder significantly the ability of state and local governments to cope with unanticipated changes, initiate policy changes, accommodate voter and court mandates, or even maintain current service levels.  Evidence in this paper should give policymakers pause before replacing deliberative, legislative budget processes with inflexible, constitutionally-based formulas.

End Notes:

[1] Barry Poulson, “The Next Generation of Tax and Expenditure Limits,” Americans for Prosperity Foundation, May 12, 2004.

[2] More than two-thirds of Medicaid expenditures are for elderly and disabled beneficiaries (Kaiser Family Foundation, “United States: Distribution of Medicaid Spending (Federal and State) by Enrollment Group, FFY 2000, available at www.statehealthfacts.org ).  Over 40 percent of Medicaid spending is on services for Medicare beneficiaries — primarily elderly and disabled (Brian K. Bruen and John Holahan, Shifting the Cost of Dual Eligibles, Kaiser Commission on Medicaid and the Uninsured, November 2003).

[3] William J. Baumol, 1967.  “Macroeconomics of Unbalanced Growth:  The Anatomy of Urban Crisis.”  American Economic Review 57(2), 415-426.

[4] For example, the education component of the CPI-U is primarily composed of tuition and fixed fees, while government spending on education goes toward teacher salaries, student transportation, computer equipment, etc.

[5] Nicholas Johnson and David Bradley, Public Services and TABOR in Colorado, Center on Budget and Policy Priorities, January, 2005.

[6] Spending data for Figure 1 are from the National Association of State Budget Officers’ (NASBO) State Expenditure Report, various years.  “State own-source” spending is the sum of general fund and “other” state spending.  This sum was divided by GDP adjusted for the normal state fiscal year (July-June) to determine state spending as a percent of GDP.  To calculate state spending under a population-plus-inflation TEL, the previous year’s allowable spending limit was multiplied by the sum of the percent growth in population plus inflation.  Since the model starts in 1990, actual (not allowable) spending was used to determine the spending limit in 1991.  Specifically, actual state own-source spending was $375.3 billion in 1990 and the growth in population plus inflation from 1990 to 1991 was 5.6 percent.  Thus the estimated 1991 total state spending under a population-plus-inflation growth formula was $375.3 billion x 1.056, or $396 billion.  The estimated 1992 figure was then calculated by multiplying $396 billion by the change in population plus inflation from 1991 to 1992.  By 2004, actual state own-source spending was $793.7 billion while spending under a TEL growth formula would have been $631 billion — a difference of $162.7 billion.  Note that the calculation does not include a ratchet effect such as the one in Colorado’s TABOR.

[7] New Jersey Office of Legislative Services, See http://www.njleg.state.nj.us/legislativepub/budget/educ04.pdf.