Comment by Shawn Fremstad to Office of Management and Budget on Directive No. 14 (Statistical Measurement of Poverty)

June 26, 2019

June 21, 2019, Shawn Fremstad

June 21, 2019

PDF pdf_small

Nancy Potok
Office of Management and Budget
725 17th St. NW
Washington, DC 20006

Re: CEPR Comments on Directive No. 14 (OMB-2019-0002; Statistical Measurement of Poverty)

Dear Ms. Potok:

As a Senior Fellow at the Center for Economic and Policy Research and an expert on poverty, I am writing to comment on OMB’s request for comments on Statistical Directive No. 14.

My main comment is that OMB has put the cart before the horse in its request for comments.

The “horse questions” that need to be answered—by OMB and the Interagency Technical Working Group—before hitching up the cart include:

  • Is the current poverty measure accurate and relevant as a statistical measure of poverty today?

  • If not, how should the federal government rebenchmark the measure so that it is accurate and relevant as a statistical measure of poverty today?

  • Should the federal government produce more than a single statistical measure of poverty and/or low-income? (In other words, as is the case with the statistical measurement of inequality by the Census Bureau, would it be more accurate and relevant to produce and report on two or more measures of poverty and/or low-income?)

The “cart questions” that need to be answered include:

  • Once the federal government has an accurate and relevant statistical measure(s) of poverty or low-income, how should the measure(s) be adjusted annually to ensure accuracy and relevance over a period of time?

On the horse questions, there is broad cross-ideological agreement—agreement that includes the current administration’s Council of Economic Advisers—that our statistical poverty measure is fundamentally inaccurate as a statistical measure of poverty.[1]

When it comes to rebenchmarking the measure to ensure its accuracy and relevance, there are both areas of broad agreement as well as areas of disagreement. Most notably, there seems to be broad agreement that income should be measured using disposable income rather than pre-tax money income.[2] There is less agreement on how a base poverty or low-income threshold should be set. Conservative experts and the current administration seem to prefer extremely low poverty thresholds, but they also often say that poverty thresholds are inherently “arbitrary”. Liberal experts tend to prefer poverty thresholds that are more consistent with broad social consensus about the minimum amount of income needed to basic needs, an amount that most Americans believe is substantially more than the current poverty line.[3]

] Some of these differences could be addressed: 1) by developing and using more than a single low-income threshold; and 2) more transparent labeling of low-income thresholds. Other wealthy nations and multi-national organizations like the OECD and Eurostat provide ready models along these lines that could be adapted for statistical purposes in the United States.[4]

Absent answers to these horse questions, it is premature to move on to the cart questions. Moreover, putting the cart before the horse (say by switching to the C-CPI-U without first addressing the more fundamental problems with the current statistical poverty line) will produce a statistical poverty line that is even less accurate and relevant over time, while implying that is has been improved.

Instead of putting the cart before the horse, OMB, Census, and other Executive agencies should take their professional and statutory duty to improve the quality of statistical information seriously. This means:

  • improving the accuracy, objectivity, and relevance of the poverty statistics produced by the federal government;

  • given the broad agreement on its inaccuracy, discontinuing the use of the current poverty measure for statistical purposes, rather than making marginal and controversial changes (with spillover impacts on public programs) to how a fundamentally inaccurate statistical measure is updated annually.

Specifically, OMB should:

  • amend Statistical Policy Directive 14 to discontinue the use of the current statistical poverty line for statistical purposes by the federal government;

  • specify that the current statistical poverty line will continue to be updated and published by HHS using the CPI-U, but for programmatic purposes only (e.g., in programs that cross-reference 42 U.S.C. §?9902(2));[5]

  • label this programmatic line in an accurate, publicly transparent way that identifies its base year and non-statistical purpose, e.g., “official base-1963 poverty line for programmatic purposes”;

  • in any guidance it provides on the use of price indices, acknowledge that:

    • the CPI-U, C-CPI-U, CPI-W, PCEPI, and other prices indexes all have biases;[6]

    • on balance, the current body of research suggests that the CPI-U understates inflation faced by low-income households;[7] and

    • no existing research finds that the C-CPI-U or PCEPI are more accurate indicators of inflation faced by low-income households than the CPI-U or CPI-W.   

The Census Bureau should:

  • annually calculate and report on the number and percentage of people living in households with disposable income below 60% of median equivalised disposable income;

  • track changes over time in this indicator using both an “anchored” (absolute) and “non-anchored” (relative) threshold:

    • anchored threshold: 60% of median income in base year and then updated for a period of time using the CPI-U;

    • relative threshold: 60% of median income in base year and updated each year for changes in median income;

  • label these measures in an accurate, publicly transparent way:

    • the measures should be identified as statistical measures of low income (rather than “the official poverty line”, absent designation as such by the President or Congress);

    • Anchored thresholds should have clear labels that identify their base year.

  • Acknowledge (including in annual Census reports on income and poverty) the likely misreporting of various forms of income by households, especially at the tails of the income distribution, and provide an assessment of how this may affect income statistics.

Submitted by:

Shawn Fremstad
Senior Policy Fellow
Center for Policy and Economic Research
1611 Connecticut Avenue NW, Suite 400
Washington, DC 20009

[1] According to the most recent Annual Report of the Council of Economic Advisers, a “full-income poverty measure” that is “anchored to 1963 standards” (and adjusted using only the PCEPI) would put the current poverty rate at 2.3 percent. According to CEA, the “War on Poverty” (which the current statistical poverty measure was initially designed for) is over and was as success. As a consequence, the current statistical poverty measure is no longer relevant. CEA goes on to note that the “task of establishing … new poverty thresholds [ones that account for economic growth since 1963] is the responsibility of elected policymakers rather than researchers” and that “policymakers should consider setting new, higher poverty standards than those defined by President Johnson over 50 years ago.” Unfortunately, the report fails to provide guidance or recommendations to President Trump on what that “new, higher” standard should be. Economic Report of the President Together with The Annual Report of the Council of Economic Advisers, March 2019,

[2] Some experts also argue that the poverty should be measured and reported on for official statistical purposes using both pre-tax/transfer and post-tax/transfer measures.

[3] See, e.g., responses to Q.22 in Robert Doar, Karlyn Bowman, and Eleanor O’Neil, 2016 poverty survey, American Enterprise Institute, August 16, 2016, (in a nationally representative survey, 61 percent of respondents said poverty line for a family of two adults and two children was $30,000 or more, including one-third who said it was $40,000 or more; percentages exclude 13 percent of respondents who volunteered that they didn’t know and 1 percent who refused).

[4] See, e.g., United Kingdom Department of Work and Pensions, National Statistics, Households Below Average Income: An analysis of the UK income distribution: 1994/95-2017/18, (using “widely used widely-used international standard measures of low income” to measure trends in low income in both an absolute and relative fashion); Ireland Central Statistics Office, Survey on Income and Living Conditions (SILC) 2017: Poverty and Deprivation, (using various at-risk-of-poverty and deprivation measures); Eurostat, People at risk of poverty or social exclusion, (reporting on various measures of at-risk-of-poverty and social exclusion).

[5] 42 USC §?9902(2) requires HHS to use the CPI-U for annual updating of the programmatic poverty line (the HHS poverty guidelines). Absent a statutory change, OMB should continue to use the CPI-U to update any poverty line that is used for programmatic purposes for consistency and separation-of-powers reasons alone. That said, if the current administration breaks with longstanding norms in this area—specifically the norm that the Executive Branch does not unilaterally cut (or expand) program eligibility by changing the statistical poverty line—then future administrations should not hesitate to substantially increase the base threshold of the statistical poverty line in a way that directly increases the programmatic poverty line.

[6] See, e.g., Dean Baker, “The ‘We’re Overstating Inflation Story’ is Back”, Center For Economic and Policy Research, May 13, 2019,

[7] See research discussed in Arloc Sherman and Paul Van De Water, “Reducing Cost-of-Living Adjustment Would Make Poverty Line a Less Accurate Measure of Basic Needs,” Center on Budget and Policy Priorities, June 11, 2019,

Support Cepr


If you value CEPR's work, support us by making a financial contribution.

Si valora el trabajo de CEPR, apóyenos haciendo una contribución financiera.

Donate Apóyanos

Keep up with our latest news