About

The Cornell ILR Wage Atlas for New York State was developed with data from three sources: (1) The MIT Living Wage Calculator; (2) The U.S. Census Five-Year American Community Survey (ACS) Public Use Microdata Samples (PUMS); and (3) The U.S. Bureau of Economic Analysis (BEA) Regional Input-Output Modeling System (RIMS II).

The tools collected on this website were originally developed throughout the summer of 2022 by ILR Buffalo Co-Lab staff, with support from Cornell ILR undergraduate student and High Road Fellow Lincy Chen. Following development, a series of interactive demonstrations with key stakeholders from the academic, government, and nonprofit sectors led to further refinements and additions. When new PUMS data were released in January 2024, Co-Lab staff updated the key data and developed additional functionalities based on further conversations with users and stakeholders (see “What’s New” below).

We strongly recommend that the Atlas be accessed via a web browser on a desktop or laptop computer that has a screen resolution of at least 1366 x 768 pixels.

On this page, we answer several Frequently Asked Questions to provide users with essential technical details about the tools and the data behind them.

What’s New? (Start Here!)

The Cornell ILR Wage Atlas, which launched in January 2023, was comprehensively updated and expanded in February 2024 following the January 2024 release of the 2018-22 Five-Year U.S. Census ACS PUMS dataset and the 14 February 2014 update to the MIT Living Wage Atlas. Please read the following release notes prior to using the Atlas.

Release Notes for Current (February 2024) Version

Geography Changes

Prior to the current, 2018-22 PUMS dataset, the “PUMA” attribute in all recent PUMS releases corresponded to a worker’s residential Public Use Microdata Area (PUMA), as defined by states following the 2010 decennial census. The year 2022 marked the first year that the PUMS reported a respondent’s updated (post-2020 census) PUMA, rather than their 2010 PUMA. This observation leads to some internal inconsistency in the current PUMS dataset. Because the current Five-Year PUMS pools all respondents who participated in the ACS between the years 2018 and 2022 into a single sample, four-fifths of the records in the sample (respondents who participated between 2018 and 2021) can be linked only to their 2010 PUMA (but not their 2020 PUMA — the Census Bureau does not report both for reasons of privacy), while the remaining one-fifth of records can only be linked to their current (2020) PUMA (but, again, not their 2010 PUMA). To overcome this issue and create a geographically consistent dataset, the Co-Lab team generated a set of 61 mutually exclusive and exhaustive combined PUMAs, which, as the description suggests, combine adjacent PUMAs (using both 2010 and 2020 boundaries) to create PUMAs that are consistent across time. These are the geographic units that are visualized in the Wage Atlas’s various maps.

Living Wage Changes

New MIT Living Wage data were obtained from the Calculator in February 2024, following a systemwide update for 2024. Crucially, there was a general tendency in NYS for living wage levels to increase dramatically relative to both the June 2022 levels that featured in the original version of this Atlas and the late 2023 levels that were current prior to MIT’s most recent (Feb. 2024) update. For example, as of Februrary 2024, the 2024 MIT estimate for a statewide living wage for a single adult with no children is $26.86 per hour. For comparison, in late 2023, the MIT Calculator reported a substantially lower living wage figure of just $21.46 per hour for a single adult with no children — a year-over-year increase of more than $5 per hour. The implication, therefore, is that cost of living in NYS has surged in recent months. County-level changes are even more extreme than this statewide difference. As shown below using just the MIT Living Wage estimate for a single adult with no children, compared to 2023, current (February 2024) MIT living wages for this group of workers have risen rapidly across the state. Hamilton County in the North Country, for instance, is associated with a $7.93 per hour increase in the local living wage for single adults with no dependents (an increase of roughly $16,500 per year). With only a handful of exceptions, the living wage for this type of worker increased by between $4 and $8 per hour virtually everywhere in NYS. Whether cost of living has actually increased that rapidly in NYS is beyond the scope of this tool. Likewise, users who wish to know more about changes in the methodology of and ongoing updates to the MIT Calculator should engage with that resource directly. For present purposes, the patterns illustrated below — as well as the changes in PUMA geography described above — have critical implications for who earns a living wage in NYS.

Why Did the Estimated Probability of Earning a Living Wage Appear to Change So Much Since the Launch of the Cornell ILR Wage Atlas?

When the Cornell ILR Wage Atlas first launched in January 2023, powered by the 2016-20 vintage of the Five-Year ACS PUMS dataset, the data showed that only about 40% of New York workers earned at or above their personalized MIT living wages. By contrast, the current (February 2024) version of the Atlas estimates this number to be 50.9%. At face value, that change appears to be quite substantial in magnitude. However, before concluding that wages in NYS are on the rise, observe that at least three key factors are playing a significant role in the current, higher living wage probability:

1. Changes in geography. The process of creating consistent, combined PUMAs to overcome the inconsistent geography problem described above also affects the probability of earning a living wage. In short, the consistent (combined) PUMA boundaries shown in the current Wage Atlas maps integrate data from areas with different living wage levels. The method that the Co-Lab research team used to assign each worker their living wage in these cases was to use the average living wage for all of the counties that are combined into new, spatio-temporally consistent PUMAs. To illustrate how this works, consider the case of the large Southern Tier geography described in the Wage Atlas maps as “Steuben, Schuyler, Chemung, Broome, Delaware, Tioga, and Chenango Counties”. Because of changes in PUMA boundaries that occurred between 2018 and 2022 (the period during which the PUMS data were collected), the PUMAs found in these seven counties were combined into a single area so that data could be analyzed consistently over time. The table below shows the current (February 2024) MIT Living Wage values for a single adult with no children in these seven counties, as well as the average for the combined area. Within the Wage Atlas, any single worker (with no children) living in this combined area would be assigned the average value from the table. An analogous approach is used for all other household situations (e.g., with respect to number of adults, working adults, and children in the home) covered by the MIT Calculator.
   

   County	MIT Living Wage for a Single Adult with No Children
   Steuben	20.41
   Schuyler	19.94
   Chemung	21.28
   Broome	20.61
   Delaware	21.30
   Tioga	21.60
   Chenango	19.72
   Steuben, Schuyler, Chemung, Broome, Delaware, Tioga, and Chenango Counties (Combined PUMA) Average	$20.69

   Observe from above that a living wage in this combined area exhibits some noteworthy variation. Per the MIT Calculator, a single worker (with no children) in Chenango County, for instance, could meet their basic needs by earning just $19.72 per hour, but a similarly situated worker in Chemung County would need to earn $21.28 per hour. However, because of the PUMS geography changes and the accompanying need to combine these counties into a single region for reasons of spatio-temporal consistency, and whereby within these regions no worker’s county of residence is identifiable, the “living wage” for both of these workers would be $20.69 per hour. Indeed, living wages in most combined areas are effectively “watered down”. In these cases, a worker’s estimated hourly wage is being compared to a slightly lower number, which places upward pressure on the estimated probability of earning a living wage. (For example, if a hypothetical Tioga County worker currently earns $21 per hour, they would be coded in the Wage Atlas as earning a living wage [which is $20.69 per hour], despite them earning below the living wage level for their home county [which is $21.60 per hour]. That being said, because it is not possible to know the precise residential county of workers in areas where combining PUMAs is necessary to bring about spatio-temporal consistency, the Atlas cannot make these fine-resolution distinctions.) The net result of these geography issues is to place upward pressure on a worker’s probability of earning a living wage in their (combined) geographic area.

2. Changes to the NYS minimum wage, combined with changes in the PUMS sample. The PUMS dataset that powered the original version of the Wage Atlas was the 2016-20 five-year vintage. That means that the sample contained workers who were interviewed at any time between those years. Namely, roughly one-fifth of the sample respondents were interviewed in 2016, another one-fifth in 2017, and so on. In 2016, the statewide minimum wage in New York was just $9.00 per hour. In 2017, legislation raised the minimum wage to $11 per hour for large employers in New York City (NYC), $10.50 per hour for small NYC employers, $10 per hour for Long Island and Westchester County, and $9.70 per hour for the rest of the State. Thus, roughly two-fifths of the workers represented in the original version of the Wage Atlas (i.e., those interviewed for the ACS in either 2016 or 2017) were working in an economy that allowed for wages which, in 2023, were far less than living wages anywhere in the state, even after adjusting for inflation. Consider the case of a minimum wage worker who was earning $9 per hour in 2016. After adjusting for inflation, that worker’s wage in 2022$ (i.e., at the launch of the Wage Atlas) would have been just $10.97 per hour — which was not a living wage anywhere in NYS at the time this website launched. The same can be said for a minimum wage worker earning $10 per hour in 2017, whose wage in 2022$ would have been just $11.94 per hour. Because the NYS minimum wage was so low in 2016 and 2017 relative to current levels ($16 per hour downstate and $15 per hour everywhere else), the 2016-20 ACS PUMS dataset contained disproportionately many extremely-low-wage workers compared to the current (2018-22) dataset. In the latter, workers interviewed in 2016 or 2017 were removed from the sample and replaced with workers who were interviewed in 2021 or 2022, when statewide minimum wage was well on its way to $15 per hour. The combination of an escalating minimum wage and replacing older worker data with more recent worker data (where the latter operate in an economy with higher minimum wage standards) placed strong upward pressure on the likelihood of earning a living wage when compared to the original version of the Wage Atlas.
3. Upward pressure on wages. It has been well-documented that, for many, especially lower-income workers, wages have experienced notable growth since the start of the COVID-19 pandemic. Whereas the prior two sources of upward pressure on the probability of earning a living wage are assumed to be the stronger forces at work, documented wage growth plausibly accounts for at least some of the changes in earning a living wage in January 2024 relative to when the Wage Atlas first launched.

What New Features Are Included in the February 2024 Edition of the Wage Atlas?

The February 2024 edition of the Cornell ILR Wage Atlas brings a significant expansion to the tools available for studying minimum wage in NYS. First, users are now able to generate basic answers to the question of “Who are minimum wage workers” at statewide and regional levels of analysis. Tools show the estimated probability of earning minimum wage for NYS workers as a whole, as well as by race-ethnicity, gender, age, and industry. Further, the tools breakdown the universe of likely minimum wage earners by race-ethnicity and age. These tools bust the myth that most minimum wage workers are teenagers working part-time or seasonal jobs. Rather, the data show that only 23.7% of NYS’s minimum wage workers are Gen Zers, born in 1997 or later. The plurality of minimum wage workers are millennials in their 30s or 40s, and most (46.1%) minimum wage workers are white.

In addition to exploring the characteristics of New York’s minimum wage workforce, the Atlas continues to give users the ability to simulate a new (single-tier) minimum wage for the state of New York. Unlike the earlier version of the Wage Atlas, though, which only reported the number and racial-ethnic breakdown of workers who would benefit from the user’s proposed NYS minimum wage, the February 2024 Wage Atlas enables users to estimate the broader impacts of raising statewide minimum wage to the proposed level. With the help of statewide economic multiplier data from the BEA RIMS II program, the tools first quantify the level of investment, by industry, that would need to happen to raise workers’ wages to the user-proposed minimum; from there, BEA multipliers allow users to estimate how much total earnings in the NYS economy would increase following the implementation of the new minimum wage, as well as the number of jobs expected to be created to accommodate the growth in consumer spending/demand expected to arise from growth in total earnings. As an example, if one were to set the NYS minimum wage to be $21.25, which was the level proposed in a bill that was introduced to the NYS legislature last year, then the tools reveal that employees currently earning below that level would see their aggregate earnings increase by roughly $49 billion. Factoring in earnings increases that would be expected to occur throughout the economy as a result of this investment, total earnings in the NYS economy are estimated to increase by $80.6 billion. The growth in consumer spending/demand that this increase would set off is expected to give rise to the creation of nearly 76,000 new jobs across NYS. Under this new minimum wage, the probability of earning a living wage is estimated to increase from 50.9% to 65.4%.

Final note on these changes: The new tools for measuring the impact of changes to the statewide minimum wage use statewide Type II multipliers from the BEA RIMS II program. Although these tools allow users to filter the data by NYS Economic Development region (REDC), keep in mind that the multipliers are statewide. In conducting economic impact analyses, it is best practice to use multipliers for the specific geographic region under investigation. Thus, to generate estimates that are as accurate as possible, it would be helpful to obtain REDC-level Type II multipliers from the BEA. Unfortunately, RIMS II datasets are not freely available to the public. Each regional file costs $275 to purchase. Because the Wage Atlas’s budget does not currently include funds for purchasing data, the Co-Lab team only purchased the statewide (NYS) multiplier data from the BEA. As such, whereas statewide impact estimates will be reliable, regional estimates may end up being slightly inflated. Consequently, regional estimates might not sum to the statewide totals that are shown by default.

All content below this line is original to the Wage Atlas, though references to specific datasets were updated as needed. Readers who are already familiar with the Atlas and the data that power it can skip this section and start using the tools.

Frequently Asked Questions

Why a Wage Atlas?

Although existing tools, such as the New York State (NYS) Department of Labor’s (DOL’s) interactive Occupational Wages data visualization, allow users to find typical wages and wage ranges by job and labor market region in NYS, such tools do not offer opportunities to explore and quantify wage disparities by race-ethnicity, gender, age, or related demographic and household characteristics. The Cornell ILR Wage Atlas seeks to begin filling this gap with tools that enable users — policymakers, advocates, grassroots organizations, researchers, planners, and economic development practitioners, among others — to identify disparities in earnings, and disparities in earning a living wage, by a host of demographic and geographic variables.

Where Does the Atlas Get Its Data?

As suggested above, all wage-related data come from two sources: (1) The MIT Living Wage Calculator; and (2) The U.S. Census Five-Year ACS PUMS. With respect to the latter, at the time of the last Wage Atlas update (January 2024), the current PUMS dataset was the 2022 vintage, covering the period 2018-22. That dataset was obtained on 24 January 2024 and subsequently used to build the current version of the Wage Atlas. See the “What’s New” section above for current release notes and changes. Concerning the MIT Living Wage Calculator, data on living wages (see below) were downloaded in February 2024.

What is a “Living Wage”?

According to MIT researchers,

The living wage model is an alternative measure of basic needs. It is a market-based approach that draws upon geographically specific expenditure data related to a family’s likely minimum food, childcare, health insurance, housing, transportation, and other basic necessities (e.g. clothing, personal care items, etc.) costs. The living wage draws on these cost elements and the rough effects of income and payroll taxes to determine the minimum employment earnings necessary to meet a family’s basic needs while also maintaining self-sufficiency.

In short, a living wage is an hourly wage that allows a worker to meet the basic needs of their household, given their household composition (e.g., number of adults, workers, and children) and county of residence.

How Does a Living Wage Relate to the Minimum Wage?

According to an analysis by World Population Review, there is no state in the U.S. where the statewide minimum wage reaches the living wage for that state. Put another way, regardless of one’s state, a minimum wage earner in the U.S. does not earn enough per hour to meet the costs of basic needs where they live.

In addition to being lower than a living wage, a state’s minimum wage is flat — it applies to everyone, regardless of household circumstances. A living wage, by contrast, considers a worker’s household composition. For example, the MIT Living Wage Calculator reports that, as of February 15, 2024, a living wage for a single adult with no children in New York State is $26.86 per hour, up significantly from the $21.99 statewide Living Wage reported for NYS by MIT when this tool first launched in January 2023. For a single adult with one child, however, the living wage jumps to $48.16 per hour, reflecting the fact that, on balance, the costs of meeting basic needs in a household with a child are higher than they are for a similar household with no children. Similarly, these statewide living wage figures change considerably depending on where one lives in a state. Thus, a living wage is something that is constantly changing and responsive to a worker’s life circumstance; whereas a statewide minimum wage is flat, slow to change, and does not cover basic needs.

Why Are So Many Tools in the Cornell ILR Wage Atlas Focused on a Living Wage?

Between June and December of 2023, the Cornell University ILR School administered the 2023 Empire State Poll (ESP), which surveys adult New Yorkers and asks numerous questions related to work and employment. One item on the ESP asked respondents to report their level of agreement with the following statement: “The minimum wage for adults should be sufficient to enable a full-time worker to afford essential needs for the worker and one child. Essential needs include food, water, housing, education, health care, transportation, clothing, childcare, and related needs including provision for unexpected events.” In other words, we asked New Yorkers if they thought the minimum wage should be a living wage. More than six out of every ten respondents (64.5%) agreed or strongly agreed with this sentiment. In fact, the most common response (32.7% of respondents) was that New Yorkers strongly agree that the minimum wage should be a living wage, followed by 31.8% of respondents who said they agree. Only 15.7% of respondents disagreed or strongly disagreed with this statement.

On that backdrop, our Wage Atlas functions as a planning tool. In strategic planning, and in campaigns that seek transformational/systemic change, one of the first steps in the process is to clarify the current reality by developing a shared understanding of “where we are” relative to “where we want to be”. Based on the responses to the ESP described above, where New Yorkers “want to be” is ostensibly in a situation where the minimum wage is a living wage. The  Wage Atlas provides numerous tools that allow users to better understand “where we are”, and how large the gap between that current reality and the vision of all earners (including minimum wage earners) receiving a living wage.

How Can the Atlas Know Who Does or Does Not Earn a Living Wage?

The answer to this question gets to the root of how the Wage Atlas tools were developed and the data behind them. As such, the question is separated into a few subquestions:

What are the ACS PUMS Data?

Whereas data from the Bureau of Labor Statistics, which NYS DOL uses in its Occupational Wages dashboard, provide up-to-date information on work in the United States, they do not offer much information on workers. Arguably, the premier data source for learning more about the demographic and socioeconomic characteristics of the latter is the Census Bureau’s ACS. The ACS is a rolling survey that asks each respondent about their occupation, income, and many other demographic, employment, and housing-related questions. ACS data come in three “vintages”: (1) one-year, (2) three-year, and (3) five-year. The different vintages reflect different compromises between geographic precision, data accuracy, and data currency. Namely, whereas one-year ACS estimates are always the most current (insofar as they are published annually), they are generally the least accurate. This accuracy issue stems from the fact that one-year estimates are derived from relatively small samples. The one-year program therefore only publishes data for larger geographies (i.e., places that meet a minimum population threshold), where economies of scale in sampling make it possible to obtain sufficient sample sizes in the course of a single year. For lower population geographies like small counties, towns, villages, or neighborhoods, the ACS combines annual survey responses into multi-year increments to generate usable sample sizes. Because the vintage with the widest time increment (five years) brings together the largest number of responses (i.e., the largest sample sizes), five-year estimates tend to have the highest reliability of all ACS estimates, meaning that they can be provided for all geographic units from fine resolution census block groups and tracts (often proxies for neighborhoods) up to counties and beyond. The price paid for that added reliability is currency, as the data are collected over a longer time horizon.

The point of the preceding paragraph is that to study attributes of workers across New York State, five-year ACS estimates unlock the greatest number of possibilities and should therefore have the most value. As such, unless otherwise noted, all ACS data used in this report come from the most recent publicly available five-year estimates.

That being said, ACS data are aggregated to political or statistical geographic units to protect the privacy of survey respondents. The Census Bureau uses a standard approach for publishing these aggregated data, so that metrics are reported consistently across the nation. While both privacy protection and standardized reporting protocols are invaluable, one byproduct of these practices is that they limit one’s ability to analyze and describe workers’ economic conditions in nuanced ways. For example, standardized reports of ACS data do not reveal how wages differ by race-ethnicity or gender for people in the same occupation. Moreover, although the ACS does include median income by generalized economic industry among its standard outputs, these conventional data do not allow analysts to examine intersections between earnings, occupation, and demographic characteristics. Thus, standard ACS data products have limited utility for building detailed profiles of the workers in a given place.

Fortunately, a powerful, but less common, product of the ACS program makes it possible to overcome some of these challenges. The ACS Public Use Microdata Samples (PUMS) “enable data users to create custom estimates and tables…that are not available through ACS pretabulated data products. The ACS PUMS files are a set of records from individual people…with disclosure protection enabled so that individuals…cannot be identified.” In other words, ACS PUMS datasets contain anonymized records for individual survey respondents – the data are not aggregated.

The rich person- (worker-) level information contained in PUMS records allows researchers to construct detailed pictures of worker and economic conditions for numerous locations across the United States. With respect to geography, however, to protect respondents’ privacy, PUMS data are not provided at conventional “small area” units of analysis like census tracts or even places (e.g., towns and villages). Instead, the finest resolution geographic units to which individual respondents can be linked are called Public Use Microdata Areas, or PUMAs. The decision to use PUMS data to analyze worker characteristics, then, involves a trade-off between geographic and informational resolution. By sacrificing the geographic resolution that comes with standard ACS products (which are published for small areas like census tracts), it is possible to gain a wealth of new information on the intersections between occupation, industry, income, demographic characteristics, and socioeconomic status. The Wage Atlas makes this trade-off and reports detailed information workers’ wages at the PUMA level of analysis.

How Does the ACS PUMS Report Worker Wages?

Respondents to the long-form ACS provide four key pieces of information that allow for estimates of their effective hourly wages: (1) wage or salary income in the past twelve months; (2) self-employment income in the past 12 months; (3) weeks worked during the last twelve months; and (4) usual hours worked per week during the past twelve months.

Unfortunately, the way these data are recorded do not allow for straightforward computations of a worker’s hourly wages. In the first place, self-reported hours worked often include uncompensated hours and/or hours worked outside of one’s regular job in the form of self-employment. Second, up until 2019, the number of weeks a survey respondent reported working was collected into bins (e.g., 40 to 47 weeks, 48 to 49 weeks, 50 to 52 weeks). Therefore, one cannot simply divide annual earnings by weeks worked by usual hours worked. To overcome this issue, the Cornell ILR Buffalo Co-Lab Wage Atlas relied on a new ACS PUMS feature that began in 2019, which reports the exact number of weeks that a person reported working. From the sample of New York State residents who were surveyed between 2019 and 2022, the Wage Atlas team computed the average number of self-reported weeks worked in each of the previous (pre-2019) “bins” used by the ACS. The results were as follows: (1) average of 52 weeks worked in bin 1; (2) 48 weeks in bin 2; (3) 42 weeks in bin 3; (4) 33 weeks in bin 4; (5) 21 weeks in bin 5; and (6) 6 weeks in bin 6. These averages were applied to workers surveyed during 2018, the only pre-2019 portion of the ACS sample, according to their respective bin. Next, for workers who report that they work for an employer, workers employment income was defined as their income earned through wages, salaries, tips, etc. For workers identifying as self-employed, employment income was defined as income earned through self-employment. Finally, for each worker, the worker’s total employment income was adjusted to 2022$ using Census Bureau-provided adjustment factors. These figures were then inflated to 2023$ using the Minneapolis Federal Reserve Bank’s online calculator, to account for the record inflation (and coincident upward pressure on wages) that occurred over the course of 2022. Finally, each worker’s effective wage was computed as:

where Hours Worked is a self-reported value between 0 and 40, and Overtime Hours Worked is a self-reported value defined as: (1) 0, for workers whose self-reported hours worked are less than or equal to 40; or (2) (Hours Worked – 40) for workers whose self-reported hours worked exceed 40.

Because of self-reporting, some workers will inevitably have “effective” hourly wages that are less than state and local minimum wages; however, these effective wages still offer a useful proxy for studying patterns of wages as reported by workers.

The final point in the preceding paragraph is an important one. Most existing tools for studying occupational wages rely exclusively on employer-side data. One feature that gives the monthly BLS Employment Situation reports their power and increases their reliability is that they supplement employer-side data with data obtained directly from workers (households). By using PUMS data, the Wage Atlas therefore brings data self-reported by workers into a space where employer-reported data has mostly stood alone.

What About the Time Lag Associated with PUMS Data?

The Five-Year ACS PUMS data that currently feature in the Wage Atlas were collected between 2018 and 2022. Consequently, relative to employer-side data that are often collected quarterly by the BLS, worker-reported PUMS data are somewhat dated. However, to at least partially overcome this temporal mismatch, all earnings figures calculated from PUMS data were inflated to 2023$. (Note: The 2018-22 ACS PUMS data include adjustment factors that allow for all earnings data to be reported in 2022$. According to the Minneapolis Federal Reserve Bank’s inflation calculator, $1 in 2022 was worth $1.04 in 2023. As such, all PUMS earnings data were multiplied by 1.04 to express wages in 2023$.)

How Does the Atlas Compare a Worker’s Effective Wage to a Living Wage?

Because records in the PUMS dataset are people, and not geographic areas, it is possible to know exactly what a worker’s household composition is — i.e., the PUMS provides data on the number of adults in a household, the number of adults who are employed, and the number of children. In addition, the PUMS data report the Public Use Microdata Area (PUMA) in which a worker lives. With a handful of exceptions, PUMAs fall wholly within, or coincide with, county boundaries in New York State.

In February 2024, the Wage Atlas team collected county-level data from the MIT Living Wage Calculator (note: prior to updating the Atlas in February 2024, MIT Living Wage data were from June 2022), which provides living wage levels, by county, for various scenarios based on how many adults, working adults, and children live in a household. From there, the Cornell team matched each worker from the PUMS dataset to their individual living wage based on: (1) their PUMA of residence (**See “What’s New” for more information on PUMAs in the current PUMS dataset**); (2) the number of adults in their household; (3) the number of those adults who work; and (4) the number of children in the household. If a worker reported having more than three children in their household — the maximum number of children considered by the MIT Living Wage Calculator — then they were assigned the living wage from the MIT Calculator associated with three children. In other words, the final column in the MIT Calculator matrix (for three children) was treated as “three or more” children. Analogous decisions were made when considering the number of adults and working adults in a worker’s household.

The bottom line is that each New York State worker represented in the PUMS was assigned a personalized living wage based on where they live and who lives in their household. Thus, it becomes possible to compare each worker’s effective wage (from self-reported data) to the living wage associated with their geography and household circumstances. That comparison is the basis for how the Wage Atlas is able to estimate who does and does not earn a living wage.

What Else Should We Know About the Data?

To limit the influence of outlying observations and questionable self-reported data, the Cornell team, for the wage schedule tool and map on the Home page — and for the “Living Wage Jobs” tool — filtered out workers whose effective hourly wages (based on self-reported data) were identified as “outliers” given their occupation and geographic area of residence (>3 standard deviations from their respective means). Additionally, the tool is limited to full-time workers who work at least 35 hours per week and are not self-employed. When using filters, please note that any occupation that is not associated with at least 25 unique records in the PUMS dataset are automatically filtered out to further limit the influence of outlying observations (e.g., many jobs are, at face value, associated with either 100% or 0% living wage earning status; however, these extreme values occur because the results correspond to only a handful of workers). The choice of 25 unique records is somewhat arbitrary, but was arrived at after a crude sensitivity analysis that used various thresholds (10, 25, 50, 100). When thresholds below 25 were applied, the problem of extreme results were widespread in tests of various filters. When thresholds above 25 were applied, many less-populated areas of the state were essentially excluded and almost wholly lacking in usable data. The adopted 25 threshold acted as something of a compromise between these two tendencies.

The “Compute Living Wage Probability” tool removes all of the aforementioned filters except for the one on workers whose wages lie 3 or more standard deviations away from their respective occupational-geographic means. Other than that, all workers — including self-employed and part-time workers — are represented in the tool. The “Living Wage Industries” tool includes part-time workers but not self-employed workers.

However, the ACS PUMS data are sample data. The Cornell ILR Wage Atlas tools offer users many ways to filter these data. The more filters that are used in combination, and hence the more specific breakdowns created by filtering, the less reliable are the results — insofar as they are based on smaller and smaller samples. To lessen instances of unstable results, some tools only allow filtering by aggregated groups (e.g., for race, white workers or workers of color). An additional potential issue with sample data is that their utility hinges on the accuracy of the information provided by respondents. It is entirely possible (and likely), for example, that some respondents misreport data related to their occupations, earnings, hours worked, and related variables that feature on this website. For these reasons, the Wage Atlas tools must be approached with an appropriate level of caution.

What Are the Plans to Update and/or Expand the Atlas?

A new vintage of the Five-Year ACS PUMS dataset releases every year in late January. The Cornell team will immediately set to work collecting, cleaning, and manipulating newly released datasets to make them compatible with the Atlas tools and update the Atlas tools with the latest vintage data. Such updates should take effect in February of each year. In addition to the data source update, the Cornell team intends to create a space on this website for data stories and blog posts. While the team will create some of this content internally, we also expect to build a submission system for external users to submit and publish stories and blog posts based on findings from and explorations of the data provided in the Wage Atlas tools. Bookmark the home page to follow the progress of this project.

Can I Cite the Atlas and Its Data?

Absolutely! Here are the recommended citations:

Weaver, Russell. (2024) [2023]. Cornell ILR Wage Atlas. Cornell University ILR School. Available at: https://blogs.cornell.edu/livingwage.

Amy K. Glasmeier, “Living Wage Calculator,” Massachusetts Institute of Technology, 2024. Accessed on [Insert Date Accessed], https://doi.org/10.18128/D010.V14.0