Recent Updates to LODES and OnTheMap

Recent Updates to LODES and OnTheMap


Coordinator: Welcome and thank you for standing
by. At this time all participants are in a listen-only
mode until the question-and-answer session of today’s conference. At that time you may press star then the number
1 on your phone to ask a question. I would like to inform all parties that today’s
conference is being recorded. If you have any objections, you may disconnect
at this time. I would now like to turn the conference over
to Earlene Dowell. Thank you, you may begin. Earlene Dowell: Thank you (Jennifer) and thank
you Greg Pewett from the U.S. Census Bureau for hosting our Webinar today. On behalf of the U.S. Census Bureau and a
partnership with the Council for Community and Economic Research and the Labor Market
Information Institute, welcome to our October LED Webinar. It is with great pleasure that I introduce
my friend and colleague, Matthew Graham as he presents recent updates to LEHD origin
destinations employment statistics for OnTheMap. In August the LEHD released two years — 2016
and 2017 — of data for the LODES datasets. These data were also added to the OnTheMap
Web application. This Webinar will discuss the recent data
update, associated application changes, and work through a series of demonstrations using
OnTheMap and the LODES to answer specific questions about where workers are employed
and where they live, and the relationship between the two. Matthew Graham is the Lead of the development
and application innovation group within the LEHD program in the Center for Economic Studies
at the U.S. Census Bureau. Over the last decade, he has led teams to
develop and implement new confidentiality protection systems, new public use datasets
such as LODES that use those privacy mechanisms and Web-based data dissemination exploration
tools such as OnTheMap which was a winner in 2010 of the U.S. Department of Commerce’s
gold medal for scientific engineering achievement. Matt has an MA in Urban Planning from UCLA
as well as an MS in Mechanical Engineering and a BS in Physics from MIT. With that, I welcome Matt. Matthew Graham: Thanks, Earlene. Again my name is Matthew Graham. I’m here to talk about the recent updates
to LODES, that is the LEHD Origin Destination Employment Statistics, and our Web application
that distributes LODES which is called OnTheMap. I do just want to say quickly that all of
the data I will discuss today is public use data. I will not be discussing any confidential
data and I will be speaking really only about LODES, the LEHD program and the products and
tools that we produce. This is a rough agenda for what we will be
doing today. The times that are listed are estimates. I will be taking questions. I encourage you during this Webinar to chat
your questions in using the chat feature in the Webinar. I will pause at several points to read and
answer those questions as best as possible. At the end of the session, if there is time
we will open-up the phone line and take questions via phone but again please do feel free to
chat your questions in and likely that I’ll get to them more quickly that way. So I’ll do a quick overview of the LEHD program
today, then I’ll spend most of the time talking about LODES and OnTheMap. Okay, the goal with this section is to give
us all a baseline. I see that there are some long-time friends
of the LEHD program and people who are relatively knowledgeable about our data and tools on
the Webinar today. But I’m also guessing there are folks out
there who are not as familiar with the program and our data and tools and so I’ll start by
addressing some of the basic topics. So, what are LEHD and LED? So, LEHD, and there will not be a test on
this particular point, LEHD stands for Longitudinal Employer-Household Dynamics. It is the program within the Census Bureau
and it connects administrative records with Census and survey data to produce new data
products without significant additional respondent burden. LED is the Local Employment Dynamics Partnership. It is through that partnership that the LEHD
program gets access to significant sets of administrative records on workers and firms
to produce the data that we then release to the public. So in a pictorial sense, I wanted to give
people an idea of how we do this so LEHD works mainly in the areas of workers, jobs, and
businesses and so we collect quite a few different administrative Census and survey data on these
different things, jobs, businesses and workers. And this is a visual representation of that
so firm data we get from a number of sources. One of those is the Quarterly Census of Employment
and Wages, QCEW that comes to us through our state partnership. We also get information from economic survey
data as well as the Business Register which is a Census Bureau internal product that is
sourced out of tax data and other sources as well. Jobs data come to us from two main sources. The first is unemployment insurance wage records
that come to us again from our state partners. These cover private employment and local and
state government employment and the coverage includes all those workers who are covered
by unemployment insurance within the state. That specific coverage can vary from state
to state but in general what we’re talking about is private employment as well as state
and local government. We also get federal employment data from the
Office of Personnel Management OPM and I will discuss a little bit more about that specific
source later in this talk. Finally we get person data from a couple of
different sources. These include federal records so from different
programs, different agencies we get access to administrative data on people and we also
have access to the demographic Census and survey such as from the Decennial Census or
the American Community Survey. We bring all of that together into a linked
national jobs dataset so what we’re really focused on here is jobs. We’re looking to understand employment and
jobs in the United States and produce data products out of that. In order to produce those data products, we
employ cutting-edge confidentiality protection mechanisms. Once we have the Linked National Jobs Data,
we can’t release it as is, it is confidential and so we protect that confidentiality and
through that produce public use data products. In general the coverage for this infrastructure
is about 97% of private employment and most state, local, and federal jobs. Right now the data availability is 1990 through
2018 although that may now be 2019 as we’re starting to roll out 2019 data for some of
our data products. Okay, the data products that we produce here
in LEHD there are four main ones and I’ll just highlight a few features of those so
the first is Quarterly Workforce Indicators. This is a data product that includes 32 indicators
on the workforce. It includes things like employment, hirings,
separations, turnovers, that kind of thing. It covers about 150 million job records or
more that are processed each quarter and for some states the data series starts in 1990
although for other states, the data series is much shorter because that was the available
history that we were able to access. The next data product is LODES, the LEHD origin
destination employment statistics. That will be the main data product we’ll be
talking about today, but in a nutshell it connects a job or a worker’s employment location
to their residential location. It’s at the census block detail. It has less characteristic detail than the
Quarterly Workforce Indicators but much, much, more spatial detail. The third data product that we produce is
called Job-to-Job Flows. This is also an origin destination data product,
but it’s one that connects jobs. When you leave a job, where do you go? Do you go to another job or do you go to a
spell of non-employment? And when you change jobs, do you also change
industries? Do you change geography? Those are the kinds of questions that Job-to-Job
Flows can answer. And, finally our most recent data product
is Post-Secondary Employment Outcomes, PSEO. This is a pilot project right now. We have these data for a small handful of
university systems but it looks at the question of if you get a degree from this university
at 1, 5, and 10 years, what are your employment outcomes? What do your earnings look like and where
are you working, what industry are you working in? This is another way to think about the data
products and I will say at this point that these slides along with the recording of this
Webinar and a transcript will be made available I believe about a week after this Webinar
is finished so you will have access to these slides for reference purposes. But this slide here shows just a different
way of saying what I just described about each of the data products and it’s really
a question of what do you want and this lets you know which data product might be able
to help you. LEHD also produces a series of dissemination
tools and applications on the Internet so these are all Web-based tools that anyone
can access. They’re free and available to all and they
make the data products that we produce available for exploration as well as download. So the tools I’ll mention just quickly, J2J
Explorer, OnTheMap, we’ll be talking about OnTheMap much more today. OnTheMap for Emergency Management, QWI Explorer,
the LED Extraction Tool, and the PSEO Data Viz which again is a pilot program, and each
of the Web sites is listed in this slide as well. Here’s another way of understanding the different
access points for the LEHD datasets. So, if you want for example QWI data, you
have a number of places to go to get access to that data and I will note that all of the
text here that is in blue and underlined are links to the places you can go so again when
you get access to this slide deck, you can click on these things and it will take you,
your browser will take you there. So QWI can be accessed via QWI Explorer. If you want to explore the data, answer some
basic questions or get some basic visualizations of the QWI data. If instead you prefer to download the QWI
data in bulk so that you can put it through your own software or your own analytical process,
you can get access to that data in two forms. One is through our raw data download where
we just give you the so-called flat files. They’re just CSVs comma-separated variable
files and you can download them in whole or in part and then you can also access the raw
data through the LED Extraction Tool. Finally QWI is also available through the
Census Bureau’s API. I will note for those of you familiar with
the API and who have already started using QWI through the API, we have been having a
glitch with the API and QWI the last week or two. My understanding and hope is that that glitch
will be resolved today if it’s not already resolved. It’s my understanding that the issue is now
understood by the API staff and that will be fixed shortly. But this is, the API is a way to get access
to the data if you’re building your own Web applications and need to make live queries
to the data. The other datasets, LODES, J2J, and PSEO all
have exploration applications so those are OnTheMap, OnTheMap for Emergency Management,
Job to Job Explorer, J2J Explorer, the PSEO Visualization Tool and you can also get all
of those datasets through Raw data download. We are working to develop API access to those
other datasets for the future. Okay, I’m going to pause and see if there
are any questions so far about what I’ve spoken on. So there is a question on LODES and it’s how
is home-based employment treated in LODES? I’m going to hold that question for just a
minute because we will get there. That’s a great question and I’m just going
to hold-off and see if there are any other general questions. Oh, this is great. Are employees at religious institutions included
in the LODES dataset? I’m actually going to expand that question
a bit and say are employees at religious institutions included in any of the LEHD datasets? The answer there is maybe. It depends if that religious institution is
covered under a state’s unemployment insurance laws. If for example the religious institution every
quarter pays into the state unemployment insurance system for that employee, then the answer
is yes. We will have information on them there. I’ll talk a bit more as we get into the definition
of what the LODES dataset is. About exactly what kind of information we
need to see to make it into LODES but so that’s really the test is are you covered by unemployment
insurance law? If you for example were fired or let go, or
laid off, would you be able to go to your state to claim unemployment insurance? That’s the question. Let’s see, if a person works two part-time
jobs, are they counted twice in the LEHD data? I’ll just expand from LODES and the answer
is potentially yes. We count jobs and I’ll discuss what qualifies
a job but if you are in the data twice, then we see you twice. In LODES we have some specific definitions
for job that help you as a user determine whether you want to see those additional jobs
and I’ll talk about those in a minute. So there’s a question here, what tool would
be good when trying to determine the age demographics in specific counties? Okay, so when if the geographic entity is
a county and you’re looking for age distribution and you’re focused on workers, then you should
be looking at the Quarterly Workforce Indicators, QWI. Those have data at the county level. They have relatively good age distributions,
they’re not as good as for example what you’ll see in the decennial data, or ACS, but again
this is focused on workers. If you’re interested in the demographics of
the population as a whole, you’ll want to look at datasets that the Census Bureau produces
about the whole population so again for example decennial or the American Community Survey. Is, let’s see here, is Census Business Builder
also being integrated with OnTheMap? Census Business Builder is another tool that
is produced by the U.S. Census Bureau. We currently do not have plans to integrate
those two applications. That said, there is a long-term strategy within
the Census Bureau to bring together lots of the functionality of the tools we already
have and to integrate that functionality widely. So maybe in the long run, functionality that
you see in Census Business Builder and the LEHD applications will be put together. All right, I’m going to pause there on answering
questions for now and I’m going to move-on to LODES because I know there are questions
specifically about LODES that I’ve already deferred a bit and I’d like to answer them
now. Okay, the first thing I want to talk about
is the LODES update. This will be very brief but for some of you
it may be the most important part of this presentation so the big news and Earlene mentioned
this at the beginning of the session is that LODES for 2016 and 2017 were released at the
end of August this year. There had been a delay on 2016 and 2017 data. We were able to put them out at the same time. Both years are now available in OnTheMap and
in the raw data files. Additionally we backfilled Wyoming data for
2014 and 2015. Occasionally because of delays in signing
or extending or renewing agreements with states, we do see lapses in state data. When we see that, we eventually, well, we
are hopeful that we will eventually be able to fill-in that data and that was the case
with Wyoming. There was not data for Wyoming in 2014 and
2015, the last few years but with this release in August, we were able to go back and fill
that in. And I’ll come back to that point in one second. The other thing we did with this data release
is update to TIGER 2018. For those of you who are not familiar with
TIGER, it is the base geography set that the Census Bureau uses and the Census Bureau keeps
that up-to-date to keep in line with current political and legal boundaries. And so that’s what we have done with this
latest release is update to TIGER 2018 to get the most recent political and legal boundaries
as the Census Bureau knows them. If the Census Bureau doesn’t know about them,
for example, there’s a recent change in the boundary of a city, then that is something
that will come later when one of the geography division’s processes picks that up and puts
it into TIGER. A few other key points, data for Alaska and
South Dakota are not currently available in 2017. Hopefully in the future we will be able to
backfill those state year combinations. We also do not have right now federal workers
in 2016 and 2017. I will talk about that in just another second
and because we do not have the federal workers in for 2016 and 2017. We made a choice to change the default job
type of OnTheMap to private primary jobs. Previously and this will make more sense for
those of you not super familiar with OnTheMap when I do some demos. Previously the default had been all primary
jobs which included federal employment. However, we did not want users to come to
the application and do the analyses they were used to doing but not know that federal jobs
were not available and see dramatic changes in their numbers and so we decided to move
to the default of private primary jobs so this includes only jobs in the private sector. And I’ll point this out when we do the demos
in OnTheMap. So federal jobs are not available again because
we are working to renew our agreement with the Office of Personnel Management. However, we didn’t have the data in hand and
we knew that we were delayed in releasing it and we finally decided to go ahead and
release LODES without federal employment. As soon as we renew the agreement with OPM,
get the data and process it, we will backfill that data into LODES and OnTheMap so that
will appear there as soon as we are able to access the raw data and process it and get
it out the door but right now federal data is not available for 2016 and 2017. Okay, the next couple of slides hopefully
will answer some more questions people have about LODES. It was developed in the mid-2000s to get at
spatially-detailed commuting-like information based out of the LEHD infrastructure, again
the dataset based on administrative records largely. It provides a number of geographic patterns
of jobs so one by the employment location so we can see the distribution of employment
down to the block level. Two, by residential location we can see the
distribution of workers’ residences at the block level and then most importantly here
is we can see the connections between the two. So we can ask questions like for workers who
are employed in this location, what is their residential pattern? Where do they come from when they get up in
the morning and go to work we suspect where do they leave from? Where do they arrive at to do their jobs. And I’ll discuss more of that as we work through
the application. These data are tabulated by several categorical
variables, age, earnings and industry, sex, race, ethnicity and education, firm age and
firm size, ownership of the business, job dominance and job type. So, ownership is for us is private sector
industry, a local or state government entity or a federal entity. Those are three different ownership classes
we express in LODES. The job dominance gets back to this question
about whether or not more than one job for a person appears in the data. As I said the answer is yes, we do see more
than one job. If a person works two covered jobs, we will
see them. If a person works three UI-covered jobs, we
will see them. To help folks who are interested more in people
and workers than a count of jobs, we have created this sense of job dominance or job
primacy and that is simply defined as the job that earned the worker the most income
during the reference period. And at this point it’s good to mention the
reference period for LODES is Quarter 2 of each year so that’s April through June. And then job type is a combination that we
use in LODES and OnTheMap of ownership and job dominance so it’s just a simple crossing
of ownership and job dominance so you’re not selecting them independent from each other. They’re mixed together in what we call our
job type so job dominance is either all jobs or primary/dominant job. We use primary and dominant interchangeably. All of these data, all of the raw data files
are available at this location and that link is embedded elsewhere in this presentation. So a few more details about LODES before we
start exploring the data. Employment for LODES purposes is through our
beginning of quarter job definition. When we get data from businesses about their
employees, we get information such as who the worker is and how much earnings did they
receive during the quarter? If we see a person has positive earnings that
is greater than zero earnings from a business in a reference quarter which is Quarter 2
and we see they also have earnings from the same business in Quarter 1, then we make the
inference that they were employed at that business on the first day of the 2nd Quarter. So at the beginning of Quarter 2 because we
see them having earnings last quarter and this quarter. We say it’s reasonable to think that they
were employed at the beginning of this quarter and that’s what we call a beginning-of-quarter
job. And that’s all that LODES covers. The quarterly workforce indicators covers
a wide variety of different job definitions but LODES only covers a single job definition,
those beginning-of-quarter jobs. I already mentioned the job type, a cross
of ownership and job dominance. We have the sense of labor market segments. These are 10 categories. There are three earnings categories, three
industry categories, three worker age, and a total. The origin-destination and the residence and
workplace margins can be tabulated by these 10 different categories. We also have some additional characteristics:
three earnings, 20 industry, the NAICS sectors, three worker age, two worker sex, six worker
race, two worker ethnicity, five firm age, five firm size, and one total. These characteristics are only available on
the residence or workplace margins. They’re not available on the origin-destination
data. All right, this slide discusses how the raw
data files are broken-up. I’m not going to spend really too much time
on this at all except to say that the files are broken-up by residence margin, workplace
margin and origin-destination. We also provide some geography files as well
that are not mentioned here. Okay, last couple bits on LODES and then we’ll
get into OnTheMap. If you’re looking to make sense of all these
different coverage questions and when if federal jobs available, what variables are available,
when, what states? This slide is a quick reference guide for
that so you can see on the left the years of the data so from 2002 when we first started
producing LODES through the 2017 data, the number of states that are available in those
years, which states are not available during that time, whether or not federal jobs are
available, whether or not the race, ethnicity, education and sex variables are available,
and whether or not firm age and firm size variables are available. Right now the heart of the dataset where everything
is available is 2011 through 2015. All states are in and all the variables are
available for these five years. Okay, and then this slide gets at how to understand
the different crossings, when you have OD data, that means you get a residence and a
workplace location that you get labor market segments but you don’t get any characteristics. Similar arrangement for the residence margins
and the workplace margins. Okay, and then what is OnTheMap, quickly OnTheMap
is a Web application that lets you play around with LODES and ask questions and get answers. I can talk about OnTheMap but it’s better
to show you so that’s what we’ll do next. So here is the OnTheMap application. You can get to it in a number of ways. One of them is by doing a Google search for
OnTheMap or if depending on, you know, which search engine you may want to type in OnTheMap
Census. Either way, you should end-up at onthemap.ces.census.gov
and your browser should look roughly like this. I will note that if you are using an older
browser specifically, IE, an older version of Internet Explorer, you may have some challenges
with OnTheMap. We suggest that you switch to a newer browser,
if possible, either Microsoft Edge, Chrome or Firefox all seem to work pretty well. So I’m going to start-out just by doing a
few analyses. I’m going to show you around the application
a bit but I will point-out that there is significant help and documentation about using OnTheMap. You can always access that by clicking on
the Help and Documentation link at the top right. And I have assume additional links embedded
in the slides that I’ll show you at the end of this presentation. So I’m going to search for an area that I
want to do an analysis about. I’m going to do an analysis on a place named
Sparks. If you’re, there are several Sparks. I’m going to choose the one in Nevada. If you’re not familiar with this part of the
country, this is the city right next door to Reno, Nevada, in Northern Nevada. I’ll move my mouse my map around a bit. Here is Reno, Sparks is over here and I’ll
back-out just to give you all a sense of roughly where it is. There is Lake Tahoe, Carson City is down here
and then the Bay Area is to the west so I’ll zoom back in and then we’ll start working
our analyses so the first thing I did was I searched for the name of something I was
interested in. I found it in my list of geography here and
there’s lots of different geographies that are supported and then I clicked on Sparks
City, Nevada. That zoomed the map to where I am now and
to move forward I just click on this link that says perform analysis on selection area. That brings me to my analysis settings window. This is where you get to choose everything
that will determine your analysis. So the first question I ask is do I want to
consider the jobs where the work area is in Sparks or the home location is in Sparks because
remember a job has two parts to it. We’ve got a worker who has a residence and
we have a business that has an employment location and the job connects the two and
so we can from a geography perspective we can start by considering either side of this. For now I’m going to leave it on the default
work. We’ll focus-on the jobs where the employment
location is in Sparks. Then I get this panel analysis type. I’m not going to talk too much about this
right now except to say I’m going to leave it on default, area profile. We’ll come back and I’ll explain a bit more
about this. Essentially though it tells us what information
we’re going to see at the end of this path of the data. Then I get to choose what year I’m looking
at. I’m going to look at the most recent year
2017 and I also get to choose which job type. As I mentioned before, private primary jobs
is the default job type. We used to have default job type set to primary
jobs which it does include federal employment. For now we’re going to leave it at private
primary jobs and only look at the private sector. I’m going to go ahead and click go. After a few seconds, the window will fill-in
and we will see employment patterns within Sparks, Nevada. So I’m going to turn, there are two different
visual representations of this. I going to zoom-in just a bit here and you
can see a little better so we have what is called a thermal overlay. This is like a density so job density in effect
and we also have a point overlay. I turned-off the thermal overlay. I use it in different situations but for right
now I’m turning it off. And what we see are a series of blue dots,
all within the boundaries of Sparks. These dots each represent one census block
and the number of jobs where the employment location is in that census block is given
by the size and coloring and is shown over here in the legend on the left-hand side. So we can see there’s quite a lot of employment
at the southern end of Sparks along Interstate 80. There is other employment distributed throughout
Sparks but it’s not this dominant job center that is here along Route 80. On the right-hand side we see some notional
charts by age, earnings, industry sector, and race. And we also see some tables and we can scroll
through these tables. First we see that the total employment in
private primary jobs is about 37,700 in 2017. We can see an age breakdown of that employment. We can see an earnings breakdown of employment
and then we can see an industry breakdown. These are by NAICS industry sectors, and I’ll
just point-out some of the dominant sectors in Sparks: construction, manufacturing, wholesale
trade, resale trade, transportation and warehousing, and also accommodation and food services. So one thing that’s quite neat about OnTheMap
is we can dig-in a little deeper here. If I say well there’s almost 12% of employment
in Sparks is accommodation and food services, where is that located? I can click on accommodation and food services. The map will update and we can see that there
are some dominant locations here along Route 80 and then there are other locations distributed
throughout. These are particularly interesting to me and
I do know a little bit about Sparks. That’s where my wife grew up, because there
is a big, there’s a cluster of casinos here. There’s one really big casino and then there’s
some other casinos and casinos tend to fall into accommodation and food services. That’s not always the case but that’s in terms
of an industry classification, that tends to be where they fall. And I do know there are other food and restaurant
options and hotel options around Sparks. So, that’s some of it but I seem to be missing
a lot of the employment that was going-on at the very southern end, so I can go looking
for that and there’s an interesting cluster here, transportation, retail trade, wholesale
trade, and manufacturing, construction. Let’s take a look at that so there we see
a bunch, transportation and warehousing and I know just from reading the papers that there’s
a lot of transportation and warehousing employment in the Reno area in general which is useful
for serving the Bay Area in California. It’s straight down Route 80 a few hours and
so this is actually not too surprising to me. This is a relatively light industrial area
here. It’s near a railyard, the railway comes through
here and Interstate 80 comes through here so it’s not surprising to see transportation
and warehousing jobs as well as wholesale trade also clustered in this area so none
of this is terribly surprising for me. So now what I’m going to do is go back and
look at a different aspect of employment in and around Sparks. So the first thing I’ll do is go down to change
settings. That brings me back to the analysis settings
panel and what I’m going to do is I’m going to keep the same set of workers, people who
are employed in Sparks in 2017 private primary job but now I’m going to do a different analysis. I’m going to do the distance direction analysis. This will show me the other side of the origin-destination
relationship. So we had been looking at the employment distribution. Now for this same set of workers we’re going
to be looking at the residence distribution. So, I hit go. That will spin for just a minute and then
it will bring the map back and show me the residence distribution for workers employed
in Sparks and there we are. I’m going to turn-off the points this time. I prefer to look at this in the thermal view
and I’m going to back-out just a bit. So I can see the greater Reno area and so
now we can see that people who work in Sparks also do seem to live in Sparks but they live
throughout the Reno area as well as far down as Carson out to Fernley, other parts of the
Washoe Valley, right, and we have over here a distribution of those workers by the distance
between their home block and their work block, right? So we see that over 70% live within 10 miles
of their employment location. Another 12% 10 to 24 miles and then there’s
about 10% here who are listed as greater than 50 miles and I know we’ve already received
a question about this and I’ll use this opportunity to talk about it and the question is do people
really drive more than 50 miles to their work. The answer might be yes but the answer also
might be no. We have employment location from the businesses
which is not always the same thing as a work site. If you telework your employment location is
where your employer is but your work site is probably your home. We do not have information on work sites. We have to make an inference based on the
information we’ve been given from the businesses. This can also be an issue for people who work
in jobs that are transitory in their location. So construction workers, if they’re working
at a different job site every day, we are not able to adequately capture their work
location but we do probably have the main office of their employment or drivers who
are on the road all day, clearly their work site is wherever their vehicle is and just
in general, long-distance relationships that people may have with their employers. I had a friend who lived on the East Coast
and was technically employed on the West Coast. She did not make that commute every day. That was just a relationship with her employer
and in fact her work site was she was a sales person and so her work site was even variable
during the day. So this may capture that sort of relationship. It may also capture cases in which businesses
are not reporting all of their work sites. That also happens although it’s very hard
for us to detect those things a priori. The last thing I want to point-out is this
diagram up here. This is what we call our radar chart. This is a distribution of distance and direction
and then we wrap it around a compass rose so the darkest wedges are the closest distances
and the size of the wedge represents how many people work for example live to the north
less than 10 miles from their work location. So this gives you a spatial sense of where
these workers are coming from so they tend to live to the north, to the west, somewhat
to the south and southwest but not really to the east very much. Mostly once you head out of Sparks, you are
in the desert and there are a few towns along Route 80 but not a lot so it’s not surprising
to see that there’s very little to the east and southeast. We can go back and for comparison we can flip
this around and ask about a different set of workers. By changing from work to home, now I’m going
to look at workers who live in Sparks but work somewhere else so these are the residents
of Sparks and we’re going to look at the distribution of their work locations. And now I’ll direct you back to the radar
chart because we get a very different distribution. People who live in Sparks dominantly work
to the southwest and south of their residence locations, okay? Dominantly, I mean, there’s almost nothing
to the north and very little to the west or east or the southeast, right? And that’s because most of the employment
in Sparks was at the very southern edge and employment outside of Sparks tends to be in
Reno or south in the Washoe Valley all the way down to Carson and so folks living in
Sparks tend to travel to the southwest or the south when they go to work. We can see this small pie wedge here to the
southeast. If I zoom the map out a bit more and moved
it a bit, you could see that to the southeast of Sparks is Las Vegas so I suspect that there
are some long-distance relationships between people living in Sparks and having some kind
of employment representation in Las Vegas. The last analysis I’d like to do quickly before
I go to questions is the inflow-outflow analysis and this will pull together both of these
perspective in one view so it will ask the question of who’s working and who’s living
in Sparks and how are those things related? And so what we see is that about 27,000 workers
are employed in Sparks but live outside the selection area, outside of Sparks. Thirty thousand workers live in Sparks but
are employed somewhere else and about 10,000 live and work in Sparks. So, this is relatively balanced although if
you’re a transportation planner for example, you might wish that these circles were a bit
more overlapping, that is people lived and worked closer to each other. But this is the inflow-outflow so it gives
you a sense of movement across the Sparks boundary, okay? So, I’m going to pause from this and go back
to the slide presentation. We talked about some of the details of OnTheMap,
you saw some of these things. Here are links for references to LODES and
OnTheMap. Again these will be in the slide deck that
you’ll have available to you. And now I’ll pause for questions and answers
and I’ll go back to the chat and have a look. I know there were quite a lot of questions
that came-in. I’ll try to go through them quickly. Can I use LODES to get employment by NAICS
industry at the block level? The answer is yes but only on the work margin
or residence margin. You do not have OD origin-destination access
to the NAICS industry sector at the block level. And when I say NAICS industry, that sector,
that’s two digit for those of you who are somewhat familiar with that term. So there’s a
question about Massachusetts, Alaska and South Dakota. Massachusetts has signed on in that one slide
I pointed out so we do have data for all states from 2011 to 2015 and in the future we may
be able to fill-in Alaska and South Dakota going forward. Right now I don’t expect us to be filling-in
any more states before 2011. That’s simply a matter of historical data
availability. When will the federal worker data on 2016
become available? Unfortunately I don’t have an answer to that
question. We are waiting for the legal agreement to
be approved so we can get access to that data. Let’s see, in LEHD is it possible to see the
size of the firm at the work location? Specifically would it be possible to see where
employees with small businesses live related to workplace as opposed to large businesses? So, we do have firm age and firm size variables. So, a couple of facets to this question. The firm age and firm size variables first
are only available on the work margin so they are not available on the origin-destination
dataset so it would not be possible to see where workers who are employed at small firms
live. I will also mention that the definitions for
firm age and firm size are a little different than you may assume them to be. Most of our data comes to us at the state
level and is within sort of a state scope. Firm age and firm size are nationally defined
so if you recently got a Walmart built in your community,
the firm age of that Walmart would be the age of Walmart as a national company, not
the age of that particular establishment. We do not have establishment age or the specific
site age. The same is true of the size so we have the
national size. Okay, so what is the best way to work around
headquartering for total employment data? I work with small geographies, typically downtowns
and center cities that sometimes have overrepresented figures in specific industries. What is the most effective way to work around
this? So, headquartering is another term we have
used to describe cases where firms do not report all their work sites. They may only report their headquarters and
so because we don’t have any other information we allocate all their workers to their headquarters
when they should be distributed at their work sites. I have seen a number of ways to do this. All of them require local data and local knowledge. That is the best way to combat this issue. In one case I saw someone who had access to
data on office square footage and they identified outliers from that as in there’s no way 10,000
people can be working at this place with 1000 square foot of office space. That’s just not possible and so they identified
the outliers and then if I recall correctly, they scaled them down, they rescaled them. That was for their particular purpose. That may not work for you but that’s one example. In general, local data, local knowledge are
really the best ways to combat this issue, external datasets that you can marry-up to
LODES and identify outliers and then appropriately deal with them. Are the jobs shown-in OnTheMap in the Sparks
demo the block level, is the job count available tabularly with spatial relationships? So, all the circles that you see in OnTheMap
are block representations so they are block totals. The raw LODES data is at the block level so
all of the residence, workplace margins and the origin-destination data are all enumerated
by census blocks. Can you select multiple places for analysis? Yes you can. I will go back here quickly to this slide. There are tutorials here if you go to this
tutorials page which you can get through the Help and Documentation link in OnTheMap. There is one about geographic selection. I strongly encourage all of you to have a
look at that because there is a great deal of power to select all kinds of arbitrary
areas in OnTheMap that I just did not have time to go through today. But we do have a tutorial that will help you
understand how to do all different kinds of selection. If we’re okay, I know we’re at 2:30 but I’ll
stick on here, answer a couple of more questions and what I’ll say is that if I didn’t answer
your question, I think what we’re going to try and do is get a printout of all these
questions and then try and provide some kind of answer that will be made available along
with the slides and other materials. So if I didn’t get to your question, I’ll
make sure we get to it afterwards. Any plans to integrate ACS commuting information
into OnTheMap? That’s a great question. Right now there are no plans but that is a
question we think about quite a lot so I’ll say there are no plans right now but maybe
in the future. Just to confirm exclusions 2016 to 2017 federal
workers affects OnTheMap only — not all LEHD data. LODES in OnTheMap is the only dataset from
LEHD that includes federal workers right now. The other datasets do not include federal
workers so those other datasets are not affected by that issue. If your region has a large number of college
students residing out of state and working part-time, will it show in the data as a large
population living beyond 50 miles and working for low wages? Maybe. So, college students are a good question. It depends how the data represent them. College students quite often work in jobs
that are not unemployment insurance wage covered, you know, they’re like temporary workers and
they can’t go claim unemployment insurance but if they are employed in an unemployment
insurance sense, then we’ll see the job and then it depends how we capture their residence. We get residence information through other
administrative records from other federal data sources. One big source of that data are tax records
so for example if a college student still shows-up on their parents’ tax forms, we’ll
probably identify them at their parents’ home location as opposed to their campus dormitory. But again it depends on the specific data. Is the data available for Puerto Rico? No, not right now. Puerto Rico has been a partner with us in
the past and we have explored producing data for Puerto Rico. There are some data quality questions we are
still trying to address related to that both spatial data quality and other administrative
issues but we will continue to work on that issue. Do you have any plans to expand the earnings
category? The current maximum range really limits the
usefulness of that data point. I’m going to make this the last question because
this is a good one and I do want to give it a meaningful answer. The earnings categories that we have in OnTheMap
and in LODES were defined 15 years ago or so. And they were defined in a way that was meant
to support analysis of low-income workers or low-earnings workers. In the 15 years since, inflation has eaten
away at those categories and I agree, now they’re relatively difficult to understand
a good chunk of the working population. We have been thinking about redefining the
earnings categories for almost 15 years and it is still an open question but one we’d
like to answer sooner than later. I would encourage you if you have ideas about
what kinds of earning categories would be useful for your specific case, please send
those to us. Feel free to e-mail me and the more you explain
your case, the better I’ll understand it because we are trying to figure-out what we might
do for the future with those earnings categories. So, I’m going to stop there. I don’t know if I need to turn it over to
Earlene at the end of this but thank you for listening and staying on the call for a couple
extra minutes and if we didn’t answer your question, we’ll get to it in writing. Thank you again. Earlene Dowell: So I would like to thank Matt
for his great presentation. Join us next month on November 20 when (Heath
Hayward) presents updates to another LEHD dataset and data tool, Job to Job Flows, and
Job to Job Flows Explorer. Thank you so much and we hope you enjoy the
rest of your day.

Leave a Reply

Your email address will not be published. Required fields are marked *