MEETING NEEDS OF LOW-INCOME NEW YORKERS

R EPORT 2 RELEASED FALL 2014

PO VERTYTR ACK ER .R O B INHOOD.ORG

CHRISTOPHER WIMER CLAUDETTE BANNERMAN IRWIN GARFINKEL KATHRYN NECKERMAN FALL 2014

FALL 2014

|

P OVERTY TRACKER

2

BACKGROUND Robin Hood’s mission is to fight poverty in New York City. To do so effectively, we need an accurate picture of what it means to be poor in New York: current surveys and statistical models don’t tell the full story. To correct this, Robin Hood—in partnership with Columbia University—launched the Poverty Tracker, a survey of 2,300 New York City residents, to learn how they cope with financial poverty, material hardship and health challenges. The Poverty Tracker will revisit these same individuals every three months over two years, with the goal of providing a more accurate and dynamic picture of poverty in NYC, so that we can better target our resources to help those in need.

FALL 2014

|

P OVERTY TRACKER

3

INTRODUCTION The Fall 2014 Poverty Tracker report marks the second in a series of reports based on surveys of well-being in New York City. The first report in the series—the Spring 2014 report—was based on a survey of nearly 2,300 households across all five boroughs that began in December 2012 by Columbia University’s Population Research Center, in partnership with Robin Hood. Data provided for the Spring 2014 report indicated that an alarming number of New Yorkers struggle to meet their basic needs, including paying for food and household bills or getting help with physical or mental health problems. Indeed, the survey found that snapshots of New York City households, based on official “income only” measurements of poverty, painted a picture that was too optimistic and didn’t capture the magnitude of disadvantage, nor the true struggles New Yorkers face in trying to make ends meet.

FALL 2014

|

P OVERTY TRACKER

4

Data for the Fall 2014 Poverty Tracker report unearthed a troubling fact about private and public efforts to ameliorate poverty: many needy households don’t get the help from government programs for which they are eligible or from private philanthropy, and many of those who receive some help don’t obtain nearly enough support to fully resolve their problems. The 2014 report also identified which groups of impoverished New Yorkers get the services they need, either through private means or with government- or community-based social services—an analysis that could point the way to greater success in providing assistance in the future.

Among the key findings of the Fall 2014 Poverty Tracker report: Many individuals, including some of the most disadvantaged New Yorkers, don’t receive adequate medical attention, can’t afford their food or utility bills and are unable to find a job, yet they don’t seek assistance that could help them resolve these problems. Many individuals who seek help report that even with assistance, they are still unable to meet their basic needs. Suggestive, though not definitive, evidence indicates that those who seek help, though their problems may not be fully resolved, do far better than disadvantaged residents who do not seek assistance. And the finding that is perhaps most persuasive for rethinking anti-poverty interventions: New York’s most disadvantaged residents—those plagued by low incomes (e.g., a household of four earning less than $32,516 per year), severe material hardships and serious medical problems—are the least likely to receive all the private and public assistance needed to address their problems.1

We believe that public and private agencies need to collaborate to a greater degree and, in some cases, redesign their programs to address the needs of impoverished families, while better targeting their programs so that they assist the individuals with greatest need.

1

Please see Appendix A of the Spring 2014 report for more information on poverty threshold construction.

FALL 2014

|

P OVERTY TRACKER

5

FINDINGS The Spring 2014 Poverty Tracker report showed that measuring specific material hardship provides a more comprehensive tool for predicting household deprivation than traditional income-based measures of poverty. Survey results found that 23% of New Yorkers qualified as “poor” using the more sophisticated Supplemental Poverty Measure, 37% suffered at least one severe material hardship such as inability to pay for food or housing, and 23% suffered from a severe health problem. In total, over 50% of New York City residents faced at least one serious disadvantage; this compares to official government estimates of 21% of New Yorkers who are defined as “poor.”

NEW YORKERS

THE SPRING 2014 POVERTY TRACKER REPORT

QUALIFIED AS POOR

HARDSHIP

HEALTH PROBLEM

(using the more realistic Supplemental Poverty Measure)

(suffered at least one severe material hardship)

(suffered from at least one severe health problem)

23%

37%

23%

53% Experience at least one significant disadvantage (severe health problem, poverty or severe material hardship).

This compares to official government estimates of 21% of New Yorkers who are defined as “poor.”

21%

FALL 2014

|

P OVERTY TRACKER

6

The Fall 2014 Poverty Tracker report tracked respondents’ needs for services over the six months preceding the survey (this survey was fielded beginning in August 2013) and whether respondents had their needs met. Specifically, the Fall 2014 Poverty Tracker asked about needs in the following six domains:2

1. HEALTH OR MENTAL HEALTH 2. ADEQUATE OR AFFORDABLE HOUSING 3. PAYING FOR FOOD OR BILLS 4. GETTING OR KEEPING PUBLIC BENEFITS 5. ENCOUNTERING A MAJOR UNEXPECTED EXPENSE 6. FINDING A JOB

Respondents reporting a need for services in a given domain were asked what, if anything, they did to try to meet those needs. If they sought help, we asked whether they received all, some or

60

none of the help they needed. They also were asked whether, regardless of their actions, their situations improved, worsened or stayed the same.

50 1 NEW YORKERS REPORTING A NEED FOR SERVICES FIGURE 40 30 20

23%

17%

20%

10

17%

11% 9%

0 Health

2

Benefits

Food/Bills

Employment

Major Expenses

Housing

The survey additionally asks about three additional domains: caring for an elderly or disabled family member, legal issues and dealing with

finances or taxes. We do not focus on these domains in this report.

FALL 2014

|

P OVERTY TRACKER

7

What are current levels of needs for poor New Yorkers? Figure 1 shows the proportion of New Yorkers who identified a need for a particular type of help. Figure 2 shows the percentage of New Yorkers who had a need and sought help, and shows the proportion of those who received all or some of the help they needed, and the proportion of those who got all the help they required. The survey covered the six months immediately prior to the reporting; if the time span were expanded, the demand for services would almost certainly increase as well. We found that New Yorkers who had a medical problem were most likely to seek help— over 80% sought assistance—and virtually all of them got some amount of help. However, only 46% of people requiring medical help got all of the necessary assistance. Yet New Yorkers with medical issues fared far better than those who needed help paying for food or other bills. Only 70% of New Yorkers who could not afford to pay for food or their bills reported seeking help for these necessities, and only about 60% of those who sought help received some assistance. A meager 19% of those who needed help paying for food or bills received all of the relief they needed. Thus, fewer than one of five New Yorkers who could not otherwise pay for food or household bills reported getting all the help they needed. Respondents who required assistance finding a job fared even worse: only 64% of those with a stated need for help finding employment sought assistance, and only 47% who were in need got FIGURE 2

NEW YORKERS REPORTING A NEED FOR SERVICES WHO SOUGHT AND RECEIVED HELP

90 80 70

83%

79%

78%

60

64%

64%

70%

61%

47%

50 62%

46%

40

62% 57%

35%

30 20 19%

10 0

19%

14%

17% 8%

Health

Benefits

Food/Bills

Employment

Major Expenses

Housing

S OU GH T H E LP GOT S OM E H E LP GOT A LL H E LP

FALL 2014

|

P OVERTY TRACKER

8

help, such as a referral to a job training program. Unfortunately, only 14% of respondents got all of the assistance they needed to find a job. Similarly bleak results were seen in those requiring help obtaining public benefits: only 19% of respondents who needed help reported that they got all the help they needed. Housing issues appeared to be the most intractable, as only 8% of those who stated a need for housing assistance got all of the help required and just 35% reported that they got even some of the help they needed. As noted in our inaugural Spring 2014 report, a key purpose of the Poverty Tracker is to understand the implications of people facing more than one type of disadvantage, defined here as the combination of income poverty, severe material hardship and a severe health issue. For the purposes of the Poverty Tracker, we define those households that were not experiencing income poverty, severe hardship or a severe health issue as having no disadvantage and those experiencing either income poverty, severe hardship or a severe health issue as having one problem. A third category is reserved for those who reported having two or more of these problems. FIGURE 3 NEW YORKERS REPORTING A NEED FOR SERVICE BY NUMBER OF DISADVANTAGES

55 50

50%

45 40 35 34%

30 25

10

26%

24%

20 15

31%

28%

20%

19%

19%

18%

15% 11%

5

5%

0 Health

Public Benefits

9%

11%

9%

7% 2%

Food/Bills

Employment

Major Expenses

Housing

0 P ROBLE M S 1 P ROBLE M 2 + P ROBLE M S

FALL 2014

|

P OVERTY TRACKER

9

As expected, we found that there is a strong relationship between the degree to which a household is disadvantaged and the level of needs for that household. That is, respondents experiencing greater disadvantage on our first Poverty Tracker survey required more assistance than those with fewer problems (Figure 3). For instance, 50% of those respondents with two or more disadvantages needed help paying for food or bills, contrasted with 26% of those with one disadvantage. Perhaps most telling is that only 7% of those with no disadvantage expressed a need for help paying for food or bills. Likewise, 34% of those with two or more disadvantages reported needing assistance with a health issue, compared with 19% with one disadvantage and 15% with no disadvantages. Similar patterns emerged for other types of service needs. We found that among our three indicators of disadvantage—income poverty, severe material hardship, which is defined as a persistent shortage of critical resources or an episode of severe deprivation or a severe health issue—severe material hardship was the strongest predictor that respondents would report needs for assistance. This makes sense, given that these are the New Yorkers reporting recent chronic or acute inability to meet their day-to-day financial needs.

UNMET NEEDS: DOES IT GET BETTER? IF SO, FOR WHOM? Does the number of a person’s disadvantages make it more difficult for them to get their needs met? The answer to this question is shown in Figure 4 for New Yorkers surveyed with zero, one FIGURE 4 NEW YORKERS WITH UNMET NEEDS BY NUMBER OF DISADVANTAGES 40 35 35%

30 25 23%

20 18%

15

20%

18% 15%

15%

10 5 0

6%

5%

3%

Health

14%

12%

9%

Public Benefits

4%

Food/Bills

5%

Employment

9%

5%

2%

Major Expenses

Housing

0 P ROBLE M S 1 P ROBLE M 2 + P ROBLE M S

FALL 2014

|

P OV E RTY TRACKER

10

or two or more problems, as measured by the initial Poverty Tracker Survey. Across almost all domains, those with multiple disadvantages had substantially higher levels of unmet needs. For instance, 35% of those with two or more problems had an unmet need for help buying food or paying household bills, compared to 15% of those with only one problem and 4% of those with no disadvantage. And for help finding a job, nearly one in five New Yorkers with two or more problems still needed help finding a job, compared with 15% of those with only one disadvantage and only 5% of those with none. Similar patterns prevailed across almost every type of need for assistance, highlighting the need for government and social service providers to work together and better target their assistance to New York’s most distressed residents. It is an obvious point that more deeply disadvantaged New Yorkers will have more needs for various services such as help finding a job, help paying for food or household bills, or help finding housing. But it is less obvious whether, given a need for assistance, those New Yorkers are less likely to have their needs met. On the one hand, their needs may be more acute or severe, making it harder for service providers to adequately address those needs. On the other hand, many programs and services are already explicitly focused on the most disadvantaged New Yorkers, so this cohort may have better access to critical supports than other, less-obviously distressed residents. That we are still seeing significant levels of unmet needs among highly disadvantaged New Yorkers may point to a need to more effectively target programs for this group.

UNDERSTANDING WHY SOME PEOPLE IN NEED DON’T SEEK HELP A disturbing fact of deprivation is that even if one seeks and receives help for a particular need, the conditions that provoked the request still may prove intractable. For instance, imagine facing a chronic shortage of money to pay for housing, food and bills. You may go to the food pantry to secure free food for your family. If your experience at the pantry is pleasant, the food is of decent quality and the food satisfies your family’s hunger for the time being, you may well report that you got all the help you needed. But did your situation actually improve? Conversely, you might not get all the help you need, and indeed may not seek help at all, yet the situation may improve without intervention. As an example, an individual’s health issue may resolve without seeking medical care. Looking at whether someone’s situation actually improved, regardless of actions taken in response to their problems, is another way to examine New Yorkers’ unmet needs. Of course, not everyone who needs help actually seeks it. Reasons include a lack of knowledge about available community programs and government support services, a lack of geographical access to available community support, perceived stigma related to seeking help and fears related to an individual’s undocumented immigration status, among many other concerns. It should be noted, too, that many people simply don’t have the time or the skills needed to navigate the dauntingly decentralized social services system.

FALL 2014

|

P OV E RTY TRACKER

11

FIGURE 5 PREDICTED PROBABILITY THAT SITUATION GOT BETTER, BY HELP-SEEKING BEHAVIOR

70 60 59%

50

30

45%

44%

40

40% 37%

33% 26%

20

24%

22%

21%

20%

17%

10 0 Health

Public Benefits

Food/Bills

Employment

Major Expenses

Housing

DI D N OT S E E K H E LP S OU GH T H E LP

This situation begs the question: Do those who seek help for their problems actually see things improve, relative to others with the same types of problems? To test this, we ran a series of analyses predicting whether people’s situations improved, controlling for Poverty Tracker status and a set of key demographic characteristics. Whether someone’s situation improved is measured by responses to the question, “Since the time you needed help, did your situation get better, get worse or stay the same?” Based on this model, we predicted the probability of an individual’s situation getting better, both for those who sought help and those who did not. The results are presented in Figure 5.

FALL 2014

|

P OV E RTY TRACKER

12

Across every area of need, those who sought help were far more likely to report that their situation improved. For instance:

We estimate that

44% of those who

sought help for paying bills or buying food saw improvement,

33% of those who sought help finding employment

FOOD AND PAYING HOUSEHOLD BILLS

EMPLOYMENT

compared to only

26%

of those who did not seek help.

compared to only

20%

of those who did not seek help.

saw improvement,

59% of those who sought help for a health or mental health issue saw improvement,

HEALTH

compared to only

37%

of those who did not seek help.

Thus, across domains, vulnerable New Yorkers who sought help were much more likely to have their needs met and to report that their situation improved. Some improvement may be due to self-selection (for instance, people with more manageable problems were more likely to seek help). However, the results suggest that a potentially important avenue for improving the lives of vulnerable New Yorkers may be in better understanding the reasons some people with critical unmet needs do not seek help, and developing ways to persuade them to do so, such as placing more of the necessary support services in the neighborhoods where residents demonstrate the greatest need, and increasing outreach and education about the services available.

FALL 2014

|

P OV E RTY TRACKER

13

CONCLUSION The Fall 2014 Poverty Tracker report illustrates sizable gaps in the extent to which existing public and private services designed to help poor New Yorkers actually met the needs of these individuals. Across an entire spectrum of necessities, large numbers of city residents reported need for assistance in some form or another. Yet across these same domains, many fewer reported actually getting all the help they needed, suggesting considerable room for improvement and cooperation between government and social service providers. As expected, the city’s most vulnerable residents—those who were beset by more poverty, hardship and health issues—reported much greater levels of need. At the same time, these particularly disadvantaged New Yorkers were less likely to receive all of the help they needed. But far too many vulnerable citizens do not even seek help for their problems, though seeking help is one of the most powerful tools for predicting whether situations will improve. Those New Yorkers plagued by poverty, severe hardship and serious health problems need more assistance to improve their situations. Finding ways to connect non-users of help to critical support structures may be one promising vehicle for ensuring that when New Yorkers face needs for services and assistance, that assistance actually reaches them and, in so doing, makes a significant difference in improving the lives of the most vulnerable New Yorkers.

FALL 2014

|

P OV E RTY TRACKER

14

INCOME, HARDSHIP & HEALTH IN NYC

POV ERT YT RAC K E R . R O B I NH O O D.O R G

REPORT 1 RELEASED SPRING 2014

SUGGESTED CITATION: Wimer, Christopher, Irwin Garfinkel, Madeleine Gelblum, Narayani Lasala, Stephanie Phillips, Yajuan Si, Julien Teitler, Jane Waldfogel. Poverty Tracker — Monitoring Poverty and Well-Being in NYC, Report 1. New York, NY: Columbia Population Research Center and Robin Hood.

SP RIN G 2014

|

P OVERT Y TR ACKER

2

EXECUTIVE SUMMARY

The official poverty estimate for New York City, at about 21%, is well-known. It’s also wrong. In fact, we have found that 23% of New Yorkers were in poverty in 2012, meaning they lacked money to cover their basic food, medical and housing needs; 37% were affected by severe material hardship; and 23% were in poor health. Our survey suggests that more than half of New York City residents struggled across one of these dimensions at some point during 2012.

NYC HOUSEHOLDS EXPERIENCING SOME DISADVANTAGE THE R E AL I T Y IS WORS E T H AN THE O F F I CI AL STATI ST I CS

0%

23%

SUFFER AT LEAST ONE SEVERE HEALTH PROBLEM

10

%

20%

80%

23%

UNDER THE SUPPLEMENTAL POVERTY LINE

21%

UNDER THE OFFICIAL POVERTY LINE

37%

70%

30%

SUFFER SEVERE MATERIAL HARDSHIP

53% 60

%

%

40

EXPERIENCE AT LEAST ONE DISADVANTAGE: SEVERE HEALTH PROBLEM, POVERTY OR SEVERE MATERIAL HARDSHIP

50%

SP RIN G 2014

|

P OVERT Y TR ACKER

3

Experts have long pointed out that deep-seated flaws in the way the federal government calculates poverty distort our picture of life at the bottom of the economy. So Columbia University’s Population Research Center and Robin Hood have partnered to create a first-of-its-kind survey. It will be conducted by Columbia University with the same households four times each year over two years. After two years, a new panel of households will be recruited and quarterly surveys will start anew. The goal is to capture rich details on household expenditures, income and government supports. And — unique to this survey — it will serve to gather information about material hardships and family and child well-being that existing surveys simply overlook. Another novel feature of this survey is that it captures the dynamics of poverty: how low-income residents respond to difficult circumstances over time. With more accurate information in hand, public and private funders will be better equipped to combat poverty. The power of the survey will be fully realized after a year or two, but important findings have already emerged from the first survey. For example, more than half of New Yorkers suffer from one of the following: poverty, severe material hardship or a severe health issue. These numbers are far bleaker than official statistics; they represent more individuals and families than previously calculated who struggle on a regular basis. SP RIN G 2014

|

P OVERT Y TR ACKER

4

TABLE OF CONTENTS

SECTION 1

Examines LEVELS OF INCOME POVERTY in New York City in 2012, using both the Official Poverty Measure definition, as well as a revised, more nuanced measure developed by the U.S. Census Bureau, known as the Supplemental Poverty Measure.

SECTION 2

Illustrates MATERIAL DEPRIVATION in the city as evidenced by New Yorkers’ ability to purchase food, pay rent, mortgage and utilities, and generally make ends meet.

SECTION 3

Examines how families fare regarding GENERAL HEALTH AND

8 11 14

WELL-BEING.

SECTION 4

Considers how many New Yorkers suffer DISTRESS ACROSS MULTIPLE DIMENSIONS: income, material hardship and health. It also SYNTHESIZES THE RESULTS and briefly discusses the future of our sample and analyses.

SP RIN G 2014

|

16

P OVERT Y TR ACKER

5

SECTION 1

INTRODUCTION Official federal statistics tell us that over 21% of New York City residents — or 1.8 million New Yorkers — live in poverty. But a first-of-its-kind survey conducted by Columbia University’s Population Research Center in partnership with Robin Hood reveals that poverty and distress in New York City are even worse than these official measures indicate. According to the Columbia University/Robin Hood Poverty Tracker, nearly 4 in 10 New Yorkers suffered one or more severe material hardships during 2012, meaning that families faced a persistent shortage of critical resources or underwent an episode of acute deprivation (such as staying at a shelter, moving in with others or having utilities shut off). If we consider the number of New Yorkers who suffer moderate, if not truly severe, material adversity, the number climbs to 6 in 10 New Yorkers. These statistics emerge from the first of what will be a quarterly survey developed by Columbia University and Robin Hood to provide a more accurate picture of poverty in New York City. The survey pool will come from a fixed panel of approximately 2,300 residents from the five boroughs, many from families living in poverty (those with an annual income of less than $11,722 per year for individuals and $23,497 for a family of four). While typical surveys draw upon a different pool of individuals with each repetition, the Poverty Tracker’s fixed panel of respondents receives a greater array and more in-depth questions over time to provide a dynamic picture of poverty that illustrates how these individuals cope with unemployment, financial shocks and medical setbacks. The Poverty Tracker collects data — including the “hardship” data and the combination of health, hardship and poverty — far surpassing the scope of current government and private surveys in New York City. The first wave of Poverty Tracker surveys was conducted between December 2012 and March 2013, creating a baseline for comparison in future surveys. Follow-up surveys have been conducted by Columbia University every three months thereafter. Once several quarterly surveys are in hand, we will continue to identify meaningful trends — in hardships, health, financial well-being and much more — to develop a roadmap for effective solutions to fight poverty.

WHAT WE’LL LEARN FROM QUARTERLY SURVEYS Identifying only those New Yorkers who meet the technical definition of poverty undercounts (and thus underserves) those New Yorkers in greatest need. In its first 12 months, the Poverty Tracker revealed that a large segment of New York City residents who do not meet the official definition of poverty are often facing other substantial disadvantages in their lives. In addition to surveying self-reported income, the Poverty Tracker also includes participants’ reports of

SP RIN G 2014

|

P OVERT Y TR ACKER

6

their ability to meet their families’ critical expenses and the difficulties they may face in doing so. The first Poverty Tracker asked questions covering five areas of material well-being and potential deprivation over a 12-month period: food, housing, utilities, medical care and finances. The topics covered by these questions included: Did you run out of money for an adequate amount of food? Did you have enough money to pay rent or mortgage? Were you forced for financial reasons to stay in shelters or move in with family or friends? Could you afford your gas, electric and phone bills, or have these services been shut off? Did you forego necessary medical or dental care because of cost? If shortfalls on these measures are episodic and temporary, the household is deemed to experience moderate material hardship. If the shortfalls are chronic or acute (a family is forced to move to shelter), then the household is deemed to experience severe material hardship. The importance of tracking material hardship in addition to income can be seen by examining nearpoor households — those with incomes between 100 and 200% of the official federal poverty threshold. According to official definitions, these families don’t live in poverty. Yet, the first quarterly survey of the Poverty Tracker revealed that 50% of New Yorkers with incomes between 100 and 200% of the official federal poverty threshold suffered a severe material hardship. To better understand the effectiveness of programs funded by Robin Hood, as well as other philanthropies, the Poverty Tracker includes a unique group in its survey. Some of the panel’s 2,300 members were chosen because they participate in Robin Hood funded poverty-fighting programs (such as charter schools, soup kitchens and job training programs). We also asked these participants to refer other panel members who do not participate in these programs. By comparing demographically similar groups, Robin Hood can get a better read on the impact of its programs.

WHY HAVE COLUMBIA UNIVERSITY AND ROBIN HOOD UNDERTAKEN THIS PROJECT? Robin Hood’s mission is to fight poverty in New York City. To do so effectively requires an accurate picture of how New Yorkers in poverty live day-to-day; current surveys don’t provide appropriate levels of detail. Columbia University’s Population Research Center is devoted to research that addresses the health and well-being of vulnerable populations. It has a deep and abiding interest in better understanding what it means to live in poverty. By partnering, we can gather more precise information that public authorities, legislators, foundations, non-profits, philanthropists and private charities can use to more effectively solve the underlying problems that affect our city’s poorest residents.

SP RIN G 2014

|

P OVERT Y TR ACKER

7

SECTION 1

INCOME POVERTY IN NEW YORK CITY The income level deemed sufficient for meeting daily needs is called the poverty line or poverty threshold. In 2012, the official poverty line in the United States for a family of four was $23,497. A family with an income lower than the specific poverty line for its size is deemed poor or in poverty, since its income doesn’t cover the minimum costs of daily life in America. The Official Poverty Measure, referred to hereafter as the Official measure in this document, was developed in the 1960s and doesn’t reflect the massive changes to the American family over the past 50 years. In the 1960s, the poverty line was based only on the cost of a minimally adequate diet. Back then, families spent one-third of their incomes on food. Now, food has become a smaller share of families’ budgets while other factors, including housing, have become more costly. Living in some areas of the country, such as New York City, has become much more expensive than in other locations. Yet the Official poverty line does not vary by where one lives. The poverty line is the same in New York City as it is in any other part of the country. The definition of income used in the Official poverty statistics — pretax cash income — doesn’t capture the spectrum of income sources families use to meet their daily needs. For example, the Official poverty measure doesn’t capture in-kind resources such as food stamps and public housing, or after-tax resources, including income from the Earned Income Tax Credit. Based on decades of research on these problems with the Official poverty measure, the Census Bureau recently issued a revised measure of poverty: the Supplemental Poverty Measure, referred to hereafter in this document as the Supplemental measure. The Supplemental measure is based not only on food but on a range of goods that include food, clothing, shelter and utilities, plus a little more to cover other necessities. The Supplemental poverty line varies geographically with the cost of living so the poverty line is higher in expensive areas like New York City, and lower in less expensive areas. This Supplemental measure captures a wider variety of available resources, accounting for in-kind and after-tax sources of income, and adjusting for necessary but non-universal expenses to be subtracted from income, such as medical out-of-pocket expenses and work-related expenses like commuting costs and childcare. (For a fuller discussion of the technical differences between the Official and the Supplemental poverty measures, and how they were implemented in the current study, see Appendix A.)

SP RIN G 2014

|

P OVERT Y TR ACKER

8

-11 -11 35 35 -30 -30 degrees degrees

THE MAJOR FINDINGS Our survey used the Supplemental methodology to construct the poverty rate for New York City. FIGURE 2 illustrates both Official and Supplemental poverty rates for the city as a whole in 2012. The Supplemental measure provides a more realistic assessment of the number of the City’s residents living in poverty. FIGURE 2 NYC HOUSEHOLDS IN POVERTY

USING THE SUPPLEMENTAL POVERTY MEASURE, OUR SURVEY IDENTIFIED ANOTHER 170,000 NEW YORKERS WHO WERE NOT PREVIOUSLY COUNTED AS POOR.

21%

23%

OFFICIAL POVERTY MEASURE

SUPPLEMENTAL POVERTY MEASURE

BASED EXCLUSIVELY ON FOOD

INCLUDES FOOD, CLOTHING, SHELTER, UTILITIES

NATIONAL AVERAGE PRE-TAX CASH INCOME DEVELOPED IN 1960’S

POV E RT Y LI N E FO R FA M I LY O F 4

VARIES GEOGRAPHICALLY POST-TAX INCOME AND CAPTURES CERTAIN ADDITIONAL SOURCES OF INCOME AND CERTAIN EXPENSES

P OVERT Y LI N E FO R FA MI LY O F 4

N ATION W IDE

$ 23,497

IN NYC

$ 3 2 ,5 1 6 *

Figures from 2012. * Please see Appendix A for more information on poverty threshold construction.

THE DEPTH OF POVERTY Our survey explores not simply the level of poverty, but also the depth of poverty and the distribution of families’ incomes above the poverty line. This more detailed information illustrates how disadvantaged New Yorkers are and how many people are near poor, defined by falling under some percentage above the poverty line. In FIGURE 3, we show where New Yorkers’ incomes fall relative to the Official and Supplemental poverty lines.1 The Supplemental measure shows a larger overall pool of New Yorkers who fall into poverty, and it also alters our picture of how many New Yorkers are finding themselves close to the poverty line. 1

Cut points reflect the percentage under a given cut point. This means under 50%; 50% or above but under 100%; 100% or above but under 200%, etc.

SP RIN G 2014

|

P OVERT Y TR ACKER

9

ees

FIGURE 3 NYC HOUSEHOLDS INCOME DISTRUBUTION (PERCENTAGE OF HOUSEHOLDS)

10

47

30

11 12

20

7* 24% 17

15

I NCOME RE L AT IV E TO T H E P OV E RTY LIN E

31

< 1 /2 1 /2 TO 1 X

OFFICIAL POVERTY MEASURE

SUPPLEMENTAL POVERTY MEASURE

POV E RT Y LI N E FOR FA M I LY O F 4

P OVERT Y LI N E FO R FA MI LY O F 4

$ 23,497

1 X TO 2 X 2X TO 3 X

$ 3 2 ,5 1 6 **

> 3X

Figures from 2012. * Percentages under the poverty line do not sum to 23% due to rounding. Actual estimated percentages for under 50% of the SPM threshold and 50 –100% of the SPM threshold are 6.54% and 16.74%, respectively. ** Please see Appendix A for more information on poverty threshold construction.

THE FINDINGS The Supplemental measure significantly alters the picture of income distribution in New York City and provides a more realistic picture of what it takes to meet daily needs here. It also considers the broader range of resources and expenses families face. Supplemental poverty statistics therefore show many more people skating close to the poverty line than would be seen under the outdated official method of calculating poverty status. The remainder of this report focuses on the Supplemental measure because it is a more accurate reflection of poverty in the City. But income poverty is only one dimension in which New Yorkers might be experiencing difficult times. In the next section, we examine through the lens of self-reported levels of material hardship how New Yorkers are actually faring.

SP RIN G 2014

|

P OVERT Y T R ACKER

10

SECTION 2

MATERIAL HARDSHIP IN NEW YORK CITY We asked questions to measure New Yorkers’ actual ability to make ends meet. Our survey asked respondents to share their experience during the 2012 calendar year in five categories of hardship. FOOD How often had the respondent run out of food without money to buy more? How often had they worried about doing so? Had their household generally enough food to eat? HOUSING Was the respondent unable to pay rent or mortgage owed in the last year? Had they stayed in a shelter, or moved in with others for financial reasons? UTILITIES Was the respondent unable to pay gas, electric or phone bills? Had these services been shut off? MEDICAL Had anyone in the household avoided seeking necessary medical or dental care because of the cost? GENERAL FINANCIAL DIFFICULTIES How often had the respondent run out of money between income cycles? FIGURE 4 shows the prevalence of material hardship in each domain and clearly shows that residents

living in poverty frequently find their resources no match for the expense of living in New York City. For most domains, we construct two levels of hardship: moderate, indicating that family members at times cannot fulfill basic needs; and severe, meaning that families face a persistent shortage of critical resources or have undergone an episode of acute deprivation. Overall, 37% of New Yorkers encountered at least one severe hardship in 2012, while an additional 26% suffered from moderate hardship in at least one area. Our survey suggests that nearly two-thirds of New York City residents struggled to make ends meet at some point during 2012. Here (and in the sections that follow), we look at dimensions of hardship by the new Supplemental measure. As expected, income poverty and material deprivation tend to go hand in hand. Fully 58% of those under the poverty line reported at least one severe hardship, compared to 31% of those above the poverty line. FIGURE 5 shows, again, that income poverty is an important predictor of material hardship. This pattern

prevailed across each indicator of severe hardship. It may seem surprising that so many New Yorkers who are not officially in poverty report severe material hardships. But the poverty line is just an arbitrary threshold. The circumstances of households just above the poverty line are not very different from those just below it. FIGURE 6 shows the presence of severe material hardships according to income distribution. Over half of those below the poverty line experienced at least one severe hardship. FIGURE 6 also shows that hardship above the poverty line is concentrated in the lower income brackets.

SP RIN G 2014

|

P OVERT Y TR ACKER

11

FIGURE 4 NYC HOUSEHOLDS EXPERIENCING TYPES OF



MATERIAL HARDSHIP 0%

%

90

A

T

A LE

ST

ON

FIN

A

UT

NCIAL

18

FOOD

12

TIES

14

G SIN

6

ILI

O

U

C

AL

10

%

26 18

18

38

16

70%

30%

MED

I

H

80%

37

20%

DSHIP

AR E H

SEVERE

26 60

MOD E R ATE NON E

%

%

40

50%

The fact that nearly 1 in 4 higher-income New Yorkers experience a severe hardship is surprising, and makes it clear that income insufficiency is not the only source of hardship. Excessive debt may play a role. Poor financial management skills, depression or other mental illness, and a variety of other factors could all contribute to running out of money before the next paycheck arrives or some other material privations. FIGURE 5 NYC HOUSEHOLDS SUFFERING SEVERE MATERIAL



HARDSHIP BY POVERTY STATUS NOT I N P OV E RT Y *

(PERCENTAGE)

MEDI CAL

15  29

FOOD

10  20

UTI LI TI ES

11  22

HOUSI NG

5  11

FI NANCI AL

16  28

AT LEAST O NE SEVERE HARDSHI P

31  58

I N P OV E RT Y *

* As defined by the Supplemental Poverty Measure.

SP RIN G 2014

|

P OVERT Y T R ACKER

12

In a future report we will explore the role of debt in accounting for material hardship among New Yorkers who appear to be relatively high income. We will also explore the role of assets and savings in buffering some apparently lower-income New Yorkers from experiencing such difficulties. FIGURE 6 NYC HOUSEHOLDS EXPERIENCING AT LEAST



ONE SEVERE MATERIAL HARDSHIP BY INCOME 0%

%

61 57 40 33 23

10

%

20%

80%

90

I NCOME RE L AT I V E TO T H E S U P P L E ME NTAL P OV E RT Y L I NE

< 1 /2

70%

30%

1 /2 TO 1 X 1 X TO 2 X 2 X TO 3X

> 3X

60

%

%

40 50%

The prevalence of material hardship among New Yorkers is substantially greater than the rate of Official poverty and even poverty measured by the Supplemental method, underscoring the need for broader measures of disadvantage to assess the real needs of New York City families. Our measures of material hardship complement our poverty estimates, revealing another dimension of disadvantage that would otherwise remain invisible.

SP RIN G 2014

|

P OVERT Y TR ACKER

13

SECTION 3

HEALTH IN NEW YORK CITY While people may suffer from health and related challenges regardless of their economic situation, indicators of well-being are often tied to poverty and material hardship. Family well-being includes broader aspects of healthy family functioning than are tapped by income poverty or material hardship alone. In this initial report, we focus on two indicators of well-being: physical health and mental health. Physical health problems were captured with two survey questions designed to determine respondents’ general health and to identify health problems or disabilities that either prevented work or limited the kind or amount of work they could complete. Mental health survey questions sought to determine if respondents had experienced an extended bout (two weeks or more) of feeling sad or depressed in the past 12 months, and if respondents had experienced an extended bout (one month or more) of feeling worried, tense or anxious in the past 12 months. From these results, we created a measure of severe and moderate problematic health comparable to our material hardship measure. FIGURE 7 NYC ADULTS SUFFERING HEALTH ISSUES 0%

A

A

G

ES

65+

36

64

20

8–

AG E S

1

80%

23

10

%

20%

S

AG E LL

29

31

70%

30%

%

90

23

S E V E RE MODE RAT E

60

%

%

NONE

40 50%

SP RIN G 2014

|

P OVERT Y T R ACKER

14

Respondents were classified as having a severe health issue if they reported they were in poor health, or if they had a chronic condition that limited their ability to complete a specific type or amount of work. Respondents were listed as having a potentially moderate family health problem if they reported they were in only fair health. Respondents who answered affirmatively to either of the mental health questions were also coded as having a potentially moderate family health problem. As shown in FIGURE 7, 23% of our sample reported a severe health problem. Perhaps not surprisingly, severe health problems were more common among those 65 or older than among working-age adults. However, poverty status in and of itself was also associated with having a health problem. Of those in poverty, 29% reported having a severe health problem, compared with 21% of those above the poverty line. An additional 36% of the poor had a moderate health problem, compared to 28% of people not in poverty. Taken together, 65% of the poor reported health problems, as compared to 48% of the non-poor. FIGURE 8 NYC HOUSEHOLDS EXPERIENCING



SEVERE HEALTH PROBLEMS BY INCOME 0%

%

38 26 26 17 17

10

%

20%

80%

90

I NCOME RE L AT I V E TO T H E S U P P L E ME NTAL P OV E RT Y L I NE

< 1 /2

70%

30%

1 /2 TO 1 X 1 X TO 2 X 2 X TO 3X

60

%

%

> 3X

40 50%

Looking at health problems by one’s place in the income distribution (FIGURE 8) reveals that 38% of those in deep poverty have a severe health problem. Rates fall among those higher up in the income distribution, indicating considerable overlap between income and health.

SP RIN G 2014

|

P OVERT Y TR ACKER

15

SECTION 4

PUTTING IT ALL TOGETHER So far we have looked separately at three domains of disadvantage — income poverty, material hardship and health. In this section, we examine the same domains of disadvantage jointly, seeking to identify how many New Yorkers are suffering across multiple domains. For hardship and health, we focus only on those experiencing severe levels of hardship or health problems.

THE FINDINGS FOR 2012 53% of the City’s residents report that they suffer in at least one of the domains: income poverty, severe hardship or severe health problems. It was most common — 30% of New Yorkers — to have only one problem. About 19%, or nearly 1 in 5 New Yorkers, experienced two forms of disadvantage. And 4% of New Yorkers reported suffering from all three types of disadvantage. Taken together, these results paint a portrait of New Yorkers struggling to get by. More than half of the City’s residents are stressed in at least one dimension of disadvantage — even when we restrict our definition of disadvantage to measures that represent severe distress.

CONCLUSION This report has taken a broad look at poverty and disadvantage in New York City. By using the improved Supplemental poverty measure, we have found that: 23% of New Yorkers were in poverty in 2012 (slightly higher than the 21% the Official measure found to be in poverty). Severe material hardship affected 37% of the city’s residents, while 23% were in poor health. We found a slim majority (53%) of New Yorkers challenged in at least one of the three dimensions of disadvantage. A smaller percentage (19%) struggles in more than one area. And a much smaller percentage still (4%) was in exceptionally dire straits, exhibiting all three signs of severe disadvantage. This report is a beginning. At the end of the current survey, we asked our respondents to join a panel that we can follow over time. In subsequent surveys, we will gather more detailed information on the health and well-being of respondents and their children, as well as reports of various shocks to their home and family lives. We will also inquire about assets and debts, experiences with New York City’s government and social services, and many related topics. We will take a closer look at New York City residents affected by Hurricane Sandy to examine how those residents are faring. Though the content of these reports sometimes can be quite disconcerting, we’re confident that our in-depth, multilayered tracking of the well-being of New York City residents will be an important, positive step in addressing both the underlying causes of poverty and ways to devise more effective programs to help New Yorkers in need.

SP RIN G 2014

|

P OVERT Y TR ACKER

16

ACKNOWLEDGMENTS We are grateful to Robin Hood for providing funding for this study, and for their valuable input into the study design and questionnaires. We would also like to thank our terrific field staff, including Lolita Roy, Claudette Bannerman and Lily Bushman-Copp, as well as an outstanding group of research assistants (Taylor Clarke, Deshonay Dozier, Ernesto Hernandez, Shaynah Jones, Emma Liebman, Camilla Mazzolini, Ethan Raker, Natalee Rivera, Neha Sundaram, Hilary Szot and Lisa Tan) for their hard work in gathering data for this report. Lastly, we are grateful to Mariela Alburgues, M.K. Babcock and Abhishek Joshi for providing support. We would also like to thank Andrew Gelman for his assistance in constructing survey weights, and Jeanne Brooks-Gunn, Janet Currie, Sandro Galea and Sherry Glied for their input into our survey development. FOR MORE INFORMATION OR TO SPEAK WITH THE RESEARCHERS FROM THE COLUMBIA POPULATION RESEARCH CENTER OR WITH REPRESENTATIVES OF ROBIN HOOD, PLEASE CONTACT [email protected].

SP RIN G 2014

|

P OVERT Y T R ACKER

17

APPENDIX SECTION 1 A

KEY MEASURES

POVERTY UNITS The U.S. Census Bureau’s Official Poverty Measure (OPM) uses the family as the resource-sharing unit. Families are defined as people who are related by blood, marriage or adoption. The Supplemental Poverty Measure (SPM) departs from this definition in a number of key ways. First, unmarried domestic partners, as well as any relatives thereof, are treated as members of the poverty unit. Unrelated children under age 15 and foster children under age 22 are also included in the primary resource-sharing unit. In all of our poverty analyses, we adopt a poverty unit definition very close to that of the SPM unit. Our survey asks each respondent, a randomly selected adult in the household over the age of 18, to complete a household roster. This household roster lists the age, gender and relationship of everyone in the household (capped at 9 others, or 10 people overall). Everyone’s relationship is derived relative to the sampled adult. This allows us to include unmarried partners, unrelated children under 15 and foster children under 22 in the sampled adults’ poverty unit. We have no detailed information on non-family members and their interrelationships. Rather than attempt to collect detailed information on the families, and incomes of secondary (and tertiary, etc.) unrelated families within a household, we weight our primary poverty units to demographic, socioeconomic and geographic characteristics of the universe of poverty units available in the 2011 American Community Survey (ACS) for New York City. See the section on sample weights below for additional details on the weighting procedure used here.

POVERTY THRESHOLDS To construct the appropriate poverty thresholds for our sampled poverty unit members, we use information from the household roster to tabulate the number of children and number of adults in the poverty unit. The relevant OPM thresholds are then assigned to each sample member based on published Census tables of 2012 official poverty thresholds. Constructing the SPM poverty threshold is more involved. Current SPM poverty thresholds are available up to 2012; however, at the time of analysis only the 2011 thresholds were available. We consider these national-level SPM thresholds as our starting point. These thresholds are published separately by housing type (i.e., renters, owners who hold mortgages and owners who do not hold mortgages). We ask our respondents their housing status and if they own their home, the amount of their mortgage payment, if any. From this information we split sample members into three mutually exclusive housing groups. A very small minority of respondents reports an unclear housing situation. For example, a few respondents report living in a house owned by another family member and do not know the mortgage status of the house. These cases were included in the group of owners with mortgages. Others reported living in a park or on the street and were put in the rental group. Thresholds for each of the three groups (renters, owners who hold mortgages and owners who

SP RIN G 2014

|

P OVERT Y T RACKER

A1

do not hold mortgages) were applied. These thresholds are then adjusted for poverty unit size and composition using the same three-parameter equivalence scale used by the Census in the national-level SPM. We next perform two adjustments to these national thresholds; one brings the thresholds to 2012 from 2011, the latest year that Census-published thresholds are available to us; while the second geographically adjusts for NYC’s high cost of living. We first adjust the Census’s 2011 thresholds upward using the overall Consumer Price Index (CPI) for 2012. We explored an alternative inflation adjustment using the CPI for New York City, but the results were virtually identical given the nearly identical CPI adjustments one sees over one year in the two indices (1.019 for the U.S., 1.020 for New York City). The second adjustment involves adjusting our 2012 thresholds for differences in the geographic cost of living. Along with each year’s thresholds, the Census also publishes (for each housing type) the share of the SPM threshold devoted to shelter and utilities. This is the portion of the threshold that is inflated or deflated for cost of living. The Census and New York City’s Center for Economic Opportunity (CEO), however, differ in how they perform this geographic adjustment. The Census uses 5-year averages in median rent for 2-bedroom units by metro area relative to the nation as a whole. CEO uses a 5-year average of Fair Market Rents in New York City relative to the nation as a whole, again for 2-bedroom units. The key difference is that these Fair Market Rents are more likely to reflect recent housing market characteristics, and not include rents faced by people who have been in subsidized rent-controlled or rent-regulated apartments for a long time. This is consequential for New York City, where a sizable portion of the population lives in such units. We therefore follow CEO’s methodology and adjust the shelter and utilities portion of the threshold by New York City Fair Market Rents relative to the nation as a whole, using a 2008–2012 5-year average. This adjustment factor is 1.5002.

INCOME Our measure of income, both official poverty resources and supplemental poverty resources, is derived from a series of questions asked to sample adults. Sampled adults are first asked if they worked and for how many months in the past 12 months they were employed. If they reported any work they were asked their total earnings over the past 12 months. If the sampled adult had a spouse or domestic partner in the household roster, these questions were repeated for that spouse/partner. We then asked a series of seven income questions for the sampled adult and their spouse/partner (jointly) if one existed: 1. Retirement income from any source 2. Disability income 3. Cash welfare or public assistance 4. Supplemental Nutrition Assistance Program (SNAP) or food stamp income 5. Unemployment or workman’s compensation income 6. Regular assistance from family or friends outside the household 7. Any other income If respondents indicated that they or their partner had some income from one of these sources, they were asked the amount in the frequency that was most easy for them to report (e.g., annually, monthly, weekly). All of these

SP RIN G 2014

|

P OVERT Y T RACKER

A2

income sources except for SNAP were included in the OPM definition of resources. SNAP was included in the SPM definition of resources. After this series of income questions, respondents were asked a single question about income from all other sources in their family (i.e., from sources not mentioned or from other family members in the household, if applicable). This latter question deviates the most from traditional sources of income reporting in that surveys like the Current Population Survey (CPS) or ACS will generally record individual income components for each individual in the household. Because of time constraints, we were not able to record individual income sources and amounts for every poverty unit member in the household. It is also possible that our general Other Income measure for spouses and partners, and for all other members of the family, may contain income sources that are not counted under the OPM. This may partially explain why our estimates of deep poverty do not decline under the supplemental measure as they do under other poverty estimates, such as the national estimates produced by the Census or the New York City estimates produced by CEO. Another reason is that our methodology may have simply yielded less income than a more thorough accounting by family members might have revealed.

OTHER RESOURCES The supplemental measure’s definition captures some resources that we do not directly measure. These are detailed below: HOUSING ASSISTANCE We do not have a direct measure of the value of housing assistance, which is treated as income in the supplemental measure. We do ask our respondents if they are in public housing or receive government-subsidized housing assistance. We also ask respondents their rent and whether they are in a rent-controlled or rent-regulated apartment. Using affirmative responses to these questions, we flag those respondents for whom we need to create an income value for housing assistance. For these respondents, depending on their housing unit’s number of bedrooms (which we also directly collect), we read in their 2012 Fair Market Rent for their apartment’s bedroom number. To get the value for this assistance, we take their reported rent outlay (also directly collected) and subtract this from the Fair Market Rent for their apartment size. We cap this amount, following the Census, at the lowest earnings between the respondent and the respondent’s spouse/partner, if available. SCHOOL LUNCH We did not ask whether respondents with children participated in the Federal School Lunch Program (though we will have such information in the first follow-up). We used our report of OPM resources and thresholds to calculate eligibility for the program if respondents’ income-to-needs ratios were under 185% of the Federal Poverty Guidelines (for reduced-price lunch) or 130% of Federal Poverty Guidelines (for free-price lunch). Based on recent CEO reports that 75% of eligible students participate in the program, we randomly selected 75% of respondents to receive a benefit from the program. We then obtained valuation details from the Census (short, personal communication) to value the typical number of meals received over the school year and the dollar value of these meals. WIC We asked all respondents whether they or their partner received benefits from the Women, Infants and Children program over the past 12 months. We did not, however, ask the value of these benefits. To calculate the value, we obtained data from the US Department of Agriculture on the average weekly benefit amount for New York City, and assumed full participation over the past 12 months. TAXES AND TAX CREDITS As with most income surveys, we did not attempt to collect after-tax income data. Based on our respondents’ answers to income questions, we performed after-tax calculations using publicly available tax-calculator software. We use the National Bureau of Economic Research’s Taxsim software,

SP RIN G 2014

|

P OVERT Y T RACKER

A3

for both federal and state taxes. One limitation of the current analyses is that not all of our sampled respondents are likely to be tax filers (e.g., young adults in their late teens or early 20s who still may be claimed as dependents of their parents). Given our sample size we found the proportion of such cases among our data to be small. We also attempted to include other family members’ income in tax calculations when available and appropriate, but in future reports we will continue to refine our tax calculation. LIHEAP The supplemental measure includes the value of Low-Income Heating and Energy Assistance Program benefits in the calculation of resources. Because of time constraints, and the minuscule nature of LIHEAP benefits, we did not include LIHEAP in our analyses.

EXPENSES The SPM also deducts from resources two (mostly) non-discretionary but non-universal expenses: a) Medical Out-of-Pocket expenses (MOOP); and b) Work and Child Care expenses. MOOP Prior to their inclusion in the 2010 CPS for the calendar year 2009 , MOOP expenses were not directly available in the national data used to calculate poverty statistics. MOOP data also are not available in the ACS, which is used to calculate New York City and other sub-state poverty measures. We attempted to calculate MOOP expenses directly in our survey. In the CPS, MOOP estimates are gathered from questions about health care premiums, routine and over-the-counter expenses and all other MOOP expenses (e.g., prescription co-pays). Because of time limitations, we chose to consolidate these three questions into one general question about all MOOP expenses over the past 12 months. As expected, this choice led to substantial numbers of Don’t Know responses in our data. But, as with income sources, we imputed missing responses based on valid responses from the rest of our sample for this question. CHILD CARE Prior to 2010, childcare expenses were not directly collected in the Current Population Survey (and are still not collected in the ACS). As with MOOP, we attempted to collect childcare expenses directly in our survey. For those respondents with children, they were asked whether they paid for child-care during the past 12 months, and if so, the amount. These estimates were combined with work expenses (described below) and then capped to subtract from SPM resources. WORK EXPENSES Work expenses (aside from child-care) are typically not collected, either in the Current Population Survey or American Community Survey. Instead, researchers use either aggregated expenses (e.g., a percentage of weekly work expenses) from alternative sources applied to estimates of weeks worked, or a combination of weeks worked and administrative data on commuting costs, to estimate a non-child-care work expenses estimate. Rather than rely on similar approaches, we asked our respondents directly about their regular non-child-care work-related expenses. Respondents could choose their frequency of work expenses (e.g., annual, monthly, weekly) and the amount of their expenses.

ACRONYM GLOSSARY

ACS New York City American Community Survey

MOOP Medical Out-of-Pocket Expenses



CEO New York City Center for Economic Opportunity



OPM Official Poverty Measure



RDD Random Digit Dialing



CPI Consumer Price Index



CPS Current Population Survey



EITC Earned Income Tax Credit

L IHEAP Low-Income Heating and Energy Assistance Program

SNAP Supplemental Nutrition Assistance Program

SPM Supplemental Poverty Measure



WIC Women, Infants and Children program

SP RIN G 2014

|

P OVERT Y T RACKER

A4

APPENDIX SECTION 1 B

SAMPLING AND WEIGHTS

SAMPLE Our sample for the initial report consists of phone survey interviews with 2,002 New York City residents based on random digit dialing (RDD) of a phone sample of residents across the city, and an oversample from poor neighborhoods (defined as ZIP codes with poverty rates above 20%). The sample includes 500 cell phone calls, 751 RDD landline telephone calls and 751 RDD landline samples from the poorest ZIP codes in NYC. We also collected interviews with respondents sampled from Robin Hood funded social-service agencies, referrals of some of these agency users, and city residents of ZIP codes hard hit by Hurricane Sandy. Because devising weights that include these subsamples, while still being representative of New York City, is a complicated task, this report is based solely on the telephone sample.

SAMPLE WEIGHTING The target population is defined as New York City residents who are at least 18 years old. Our weighting process adjusts for the unequal selection probability for each respondent into our sample, primarily due to oversampling from the poorest ZIP codes, and non-response (i.e., declining to participate when contacted by the survey research firm). The weights include person-level weights and poverty-unit weights (see below for a discussion of poverty unit creation in our data). We use the 2011 NYC American Community Survey (ACS) to post-stratify our data in the creation of sample weights. We first calculate the marginal distributions of post-stratification variables from the ACS. Note that the ACS data is only a sample that has its own person- and household-level weights. When we calculate the marginal distributions, we take account of the ACS’s person-level weights and obtain the weighted frequency for each post-stratification variable from the ACS data. We construct the design weights, nonresponse weights and poststratification weights sequentially, and the final weights are the combination of these factors. STEP 1 We consider the selection probability of the respondent. First, we upweight cases by the number of adults in the household. The classic design weight equals the inverse selection probability for the respondent, which is the number of adults in the household. Second, we adjust weights by phone availability in the household. We weight down cases with multiple phones and weight up cases with interrupted service. For the cases with dual service, we construct a frame-integration weight as the compositing factor. The frame-integration weights for the landline phone sample are defined as the ratio of the number of dual-service landline cases to the total number of dual-service landline and cell-phone cases. Then the frame-integration weights for the cell-phone samplers are defined as the ratio of the number of dual-service cell-phone cases to the total number of dual-service landline and cell-phone cases.

SP RIN G 2014

|

P OVERT Y T RACKER

B1

STEP 2 We adjust for screener nonresponse using information on all calls made for landline RDD samples, landline oversampled cases and cell-phone samples, respectively. First, we calculate the eligible rates for each group. Second, we adjust the nonresponse rate by the ratio of the number of interviewed cases to the number of completed cases. Third, we multiply the design weights by the eligible rates and the inverse value of the complete rates. The eligible rates for the RDD landline, landline oversampled cases and cell-phone samples are .261, .292 and .609. The complete rates are 751/21869, 752/18230 and 514/8062. STEP 3 We implement a raking procedure and post-stratify the weights to adjust for noncoverage and the over-coverage of poor households. The initial information used for post-stratification includes ZIP codes, gender, age, education, race and income. Because the poverty rates of different age subgroups are also of interest — for example, the percentage of children/elderly/working-age people living under poverty — we add the subgroup numbers, each separately, for children, elderly and working-aged people in the poverty unit, to the list of post-stratification variables. We use ranking rather than cross-tabulation for the post-stratification, and trim the weights at the value of their mean, plus four standard error terms. Finally, we combine all the factors described above and create our person-level weights. The weights for the poverty units are defined as the personal weights decided by the number of adults in the poverty unit.

SP RIN G 2014

|

P OVERT Y T R ACKER

B2