Assignment Three

link to all assignments

modified: Tuesday, November 22, 2016

Using data on employment, education, occupation, and other relevant demographic characteristics, answer the following questions about your two case study locations:

1. What is the current occupational profile and labor market status of your locations (e.g., employment by occupation, unemployment, labor force participation rates)? How does this compare to the larger geographic context (e.g., the region, state or nation)? What can you say about skill and education levels?

Be sure to find all the relevant data (or note why it is not available), such as:

• employment by occupation
• unemployment
• labor force participation rates
• income levels
• measure(s) of skill and/or education levels.

2. Given the locations' current labor/occupational structures, what are their prospects for the future? That is, how well or poorly positioned are the two local economies to weather economic uncertainties and meet the demands of the changing labor market? Can you find evidence that suggests that the locations' occupational structures have led to the local economies doing either better or worse than the national economy as a whole?

Note: Q#1 involves a quantitative analysis of current (or recent past) conditions. Q#2 addresses the future (e.g., the next 1-10 years), so is predictive and involves intelligent speculation and interpretation. You have flexibility in how you answer Q#2 given the lack of one clear answer.

Presentation (Nov 22)

As with earlier assignments, each group will briefly present their findings in class. Here is the link where you can upload your slides.

Advice: don't just march through a long, numbing display of statistics; instead, use the statistics to illustrate your narrative analysis of your city's economy through the lens of its workers, labor force, education, skills, etc. Organize your presentation around several main points.

More advice:

Here are some comments/advice for Assignment 3 on labor markets/occupations:

The assignment is, in some ways, similar to Assignment 2, though this time you are looking through the framework/lens of occupation (what people do at work) rather than of industrial sector (what firms produce). However, rather than using several specific methods (e.g., shift-share, LQ, etc.) with Assignment 3 you have more flexibility & discretion about how you analyze and present your results. So, do cover the basics (Q1), but I encourage you to also be creative and exploratory -- the labor side (since it involves people, skills, education, culture, neighborhoods, race and gender politics, migration, etc.) is arguably more complex and rich than the firm side (Assignment 2).

Q#1 involves a quantitative analysis of current (or recent past) conditions. Q#2 addresses the future (e.g., the next 1-10 years), so is predictive and involves intelligent speculation and interpretation. You have flexibility in how you answer Q#2 given the lack of one clear answer.

Also: you can generally find employment-by-occupation data for the recent past (for Q1). But finding employment-by-occupation projections for local areas may be hard; projections (looking to the future) tend to be at the national level. One strategy is to take national-level projections by occupation, and then see how your cities occupational profiles compare: does your city have a concentration in fast or slow-growing occupations? (like shift-share analysis, only here a measure of "occupational mix

Looking for Differences / Looking for Similarities: You will likely find differences (e.g., in wages, occupational structure, unemployment rate, etc.) between your cities (or metro areas) and the U.S. as a whole.  This is a significant finding, but how to interpret?   Are these patterns specific to your case studies (e.g., they are distinctive metro areas?), or are these patterns general to all US metro areas?  (e.g., you're picking up, through your two cases, a more general metro vs. non-metro difference).

For examples of understanding local economies through analyzing occupations and labor markets, see the optional reading in ctools: Koo, J. 2005. How to analyze the regional economy with occupation data. Economic Development Quarterly 19, no 4: 356-72.

Please be sure to...

• provide a clear, logical presentation of the data
• properly label tables: full title (variables, geography, years, units, unit of analysis, etc.); label x and y axes; sources (full citations). If you are using BLS occupational categories, list BOTH the SOC code AND the occupational title.
• Be clear about geographic unit of analysis:  City?  County?  Metropolitan (or Micropolitan) Statistical Area?  Combined Statistical Area? etc. (Be consistent about using the precise geographic title/label, especially when the central city name is also the county and/or metro name as well.) You might include, in the corner of your graph page, a little map showing the relevant geography.   This would readily inform your readers (or viewers) the geographic areas. (for more details, see US Census geography)
  
• explain all unclear, unusual terms [and see the BLS glossary if you need definitions]
• [optional] provide a caption at the bottom of a table or chart if it helps the reader understand the graphics and/or quickly focus on the major trends/patterns in the data
• if showing monetary values over time, note if constant (adjusted for inflation) or current (NOT adjusted for inflation) values (e.g., dollars) [US Census explanation]
• To adjust time-series data for changes in prices (usually inflation, rarely deflation), user deflators such as the CPI:
  http://www.bls.gov/cpi/ (There are specific deflators for specific sectors;  the CPI is a useful, aggregated index.)   Remember that the CPI focuses on Consumers.
  
• Also:   the poverty thresholds do NOT adjust for variations regional cost of living (e.g., that rents etc. in New York City are higher than in Detroit).  But there are some exploratory efforts to do this.  e.g., https://www.census.gov/hhes/www/poverty/publications/PAA_Where_are_the_Poor_Do_prices_matter.pdf
• For income (and other values), clearly label if it is the MEAN income or the MEDIAN income.
  
• Clearly note whether income data etc. are measured at the per capita, household or family level.  Each of these three levels has value.   But these may be affected (i.e., distorted?) by demographic characteristics of the local population.  e.g., an unusually large or small hhd size will have an affect.  Communities with a lot of non-family households (e.g., Ann Arbor) will show a much higher median family income ($92K in 2013) than hhd income ($55K)  -- since students generally live in non-family hhds and have lower income.    The gap between hhd and family income will be smaller in communities without a lot of college students.  e.g., in Petoskey, MI:  median hhd income is $42K and median family income is $57K. (source: US Census. INCOME IN THE PAST 12 MONTHS (IN 2013 INFLATION-ADJUSTED DOLLARS), 2009-2013 American Community Survey 5-Year Estimates).
• Labor Force Participation Rates (LFPR). Underlying this rather dry statistic is the more poignant question:  why is someone not in the labor force?  Is it voluntary or not?  Are they raising children?  taking care of elderly parents?  discouraged workers?  students?  retired?  etc.    see: http://www.bls.gov/cps/cps_htgm.htm#nilf and http://www.bls.gov/cps/lfcharacteristics.htm#nlf Age-specific LFPR can be useful.
• Numbers in Data Tables and Graphs Generally, you don't need accuracy past 1/10 of a percent.   e.g., x.x%.  
  And be consistent with the number of decimal places you show (e.g., don't have a table with values such as 3.4;   6;   2.73; etc.).   Instead, round to:  3.4;  6.0;  2.7.
  You can easily have Excel do this for you:  Format > Cells > select "number" and set decimal places to "1".   Also:   my sense is that it is better to use negative signs to indicate negative numbers (rather than parentheses or red digits), but this convention may vary by discipline and region.
• Be sure to right-justify all numbers in tables (decimal points should line up vertically). This makes reading the table easier.
• Employed Residents vs. Jobs Located in Geographic Area: Be sure to differentiate between these two.  Keep in mind the difference between jobs located in a city (or county or metro area) vs. employed residents in a city.  The former are usually reported by the employer (e.g., the firm), along with other data about wages and salaries paid, number of employees, the type of business, etc.   The latter is reported by the resident (e.g, through the population census, ACS, etc.).    In closed labor markets the two should be the same (e.g., everyone who lives there works there and vice versa -- Hawaii might come close to this).   But most of you are looking at open systems:  e.g., people commute in and out of the location. In locations with a high net in-commuting pattern (e.g., Manhattan), jobs >> employed local residents.  In typical "bedroom suburbs", you will find the reverse:  employed local residents >> jobs.  Elsewhere you may find parity between the two (e.g., a jobs-housing balance).  
• What chart types to use? Bar and column charts can only handle a limited number of cases x variables before they get crowded and hard to interpret.  Try alternatives:  scatterplots, small multiples of charts; a well-constructed data table; etc.
  and be careful of excessive color-coding, 3D graphic effects, etc.  Have the reader see patterns in the data, not patterns in your graphic design.  
• If your table or figure includes "selected occupations," briefly note the selection criterion: e.g., "the ten largest occupational groups in the city," or "the ten fastest growing occupations in the US," etc.
  
  
  

some additional data sources:
http://www.bls.gov/data/
http://www.bls.gov/news.release/ocwage.toc.htm
http://www.bls.gov/bls/blswage.htm

and my page on data sources.

International Cases and Challenges in Finding Data:

Each country seems to have different practices and traditions about what and who is counted -- and not counted. In the US, we look at employment both by industry (defined by the sector) and occupation (defined by the nature of the individual's work tasks). But other countries often don't go into detailed occupational data. Sometimes they just compare blue-collar (manual labor), white-collar (mental labor) -- and sometimes have a third category of civil servant (which is often white collar, but some countries have a clear tradition of civil servants so they separate this category).

For this assignment (especially if you are looking at non-US cities), those if you can't find all the elements for ....
employment by occupation
unemployment
labor force participation rates
income levels
measure(s) of skill and/or education levels.

... then briefly explain what data you can and CANNOT find, and then do the best with the data you can find. If there are useful statistics about work and the labor force (beyond the ones listed above), feel free to substitute some other measures of the labor force.

So: I will not penalize assignments in cases where the data is not available, especially if you explain the limits to the available data -- and do your best to find and document measures of the labor market.

Discussing Data:

1. When presenting data, find a useful balance between description of the data itself and analysis/reflection.   Put the data in context.   Have the audience/reader see the connections between specific data points and the larger picture of understanding how the local economy works.  Articulate the key points.  Highlight what is both expected and unexpected in the data.  Develop a story line.  For example, do the data indicate economic strengths and vulnerabilities?  Do the data suggest that the local economy is doing better or worse than expected?  To what extent is the local economy typical or exceptional (compared to neighboring cities or to seemingly similar cities across the country)?  Do the data suggest a specific economic history and role for the city (e.g., as a deindustrializing city that hasn’t made a viable shift into new sectors?  as a place with a troubling low labor force participation rate?  as a place that has a high disparity between the rich and the poor, between the highly educated/skilled and the rest of the population?  etc.)
   
   

2. I would encourage you to effectively answer three questions:

1. How does it compare to the county, state or US, etc.?  (If, for example, if you say:  “12 percent of the city’s workforce is employed in the production occupations”, should we think of 12 percent as surprisingly big or small?)

2. Can you explain (or at least speculate) why this number is unusually big or small?

3. What is the impact/significance of this difference?

Table Formatting:

1. If the data is sorted in a specific way, explain.  e.g., from large to small, fast growing to fast declining, alphabetical, by SOC code, etc.

2. Right-justify data.  (easy to do in Excel).

3. Use a consistent number of digits to the right of the decimal point.  (In Excel, select “Format>Cells>Number rather than “General”.)  Generally, 1/10 of a percent (1 decimal place) is sufficient for percentages.   For employment numbers, whole numbers (0 decimal places) are appropriate.  (It is rare that 1/100 of a percent accuracy is useful.)

4. Use full titles:  variables, geography, units of analysis, year(s)

5. If you include a selection of places, occupations, sectors, etc., explain the selection criterion used (e.g., the ten largest cities, the 15 fastest growing occupations, etc.).

6. If you are listing occupational categories, include both the SOC code AND title.

7. Please see this updated updated one-page pdf file: "Data tables:  Common mistakes and fixes"

Charts:

1. To avoid distorting the pattern of the data, be sure that the x and y axes (e.g., on an age pyramid or a time-series chart) are linear.  (Unless of course you are intentionally using a log scale, which is generally not needed for this type of assignment.)

2. Don’t try to include so much data in a chart that the important data patterns are obscured.

3. Label the x and y axes.  Though the chart title might seem to imply x and y, it is better to explicitly label the axes.  (Anytime a reader or audience member asks, “what are the units?”, “do the column heights show the employment levels or the income”, etc., then that tells you to add more complete labels.)

Precision of Language:

1. Clearly state  if data is organized by sectors or occupations.
2. Are the employment levels based on employed residents of a city (e.g., from the ACS) or employees working at firms located in the city (e.g., the Economic Census)?
3. Income:   is it per capita, household or family?
4. What year(s) are the data collected?
5. If time series data, is it adjusted for inflation (constant dollars) or not (current dollars)?
6. Be clear about the geographic units of the data:  city, county, MSA?   Use the precise US Census titles for geography to avoid confusion.  (It’s ok to include various geographic scales in your report, but just be explicit about the geography.)
7. Be precise in your use of terms such as unemployment, labor force, labor force participation rates, poverty, etc.  [BLS Glossary]

What variables to include:

Be sure to find all the relevant data (or note why it is not available), such as:

• employment by occupation
• unemployment
• labor force participation rates
• income levels
• measure(s) of skill and/or education levels.
  

   

Level of Detail: In supplying employment occupation statistics, how deep and specific should you look? I.e. 6 digit SOCs codes, or more generalized? And how many occupations/classes of occupations should we present?

1. how much disaggregation (i.e., digits) for the SOC codes?   As you have discovered earlier with NAICS codes, there is a trade-off between using two-digit codes (small, concise tables but rather broad-brushed categories) and using 3,4,5,6-digit codes (nice detail and differentiation, but potentially massive tables).   You might try a two-tier approach:  do a table with all the two-digit SOC codes (which would have several dozen categories -- I think there are 23 groups at the two-digit level), and then do a second table that drills down deeper into selective 6-digit codes:  e.g., “the ten most common occupations (as defined by 6-digit SOC codes)”, or alternatively explore 6-digit SOC codes within several selected 2-digit occupation classes (selected either because they are large, they are growing, they are highly -- or poorly -- paid, they are interesting, etc.).
2. how many occupations/classes of occupations?
   As noted above, if you use a two-tier strategy, you could include ALL the 2-digit occupations (or at least all the ones with measurable employment in your city or region), and then selectively include some 6-digit occupational categories (e.g., the top 10-20, or some other selection criterion.)   There is no specific required number here:  include enough to give a good broad picture of the local labor market, but also be selective enough that you don’t overwhelm the audience/readers with huge tables.

From the “2010 SOC User Guide”

“To facilitate classification and presentation of data, the SOC is organized in a tiered system with four levels, ranging from major groups to detailed occupations. There are 23 major groups, broken into 97 minor groups. Each minor group is broken into broad groups, of which there are 461. There are, at the most specified level, 840 detailed occupations. Detailed occupations in the SOC with similar job duties, and in some cases skills, education, and/or training, are grouped together. Each worker is classified into only one of the 840 detailed occupations based on the tasks he or she performs.”  [ii]

“The 2010 SOC system contains 840 detailed occupations, aggregated into 461 broad occupations. In turn, the SOC combines these 461 broad occupations into 97 minor groups and 23 major groups. Of the 840 detailed occupations in the 2010 SOC, 359 remained exactly the same as in 2000, 453 had definition changes, 21 had a title change only, and 7 had a code change without a change in definition. Most of the definition changes (392) were editorial revisions that did not change occupational content. Therefore, no substantive changes occurred in occupational coverage for about 90 percent of the detailed occupations in the 2010 SOC.” [vi]

see also:  https://www.census.gov/people/io/about/faq.html#Q8

Sources:

1. provide full citations.  (see the UM Library’s Citing Sources: Giving Credit Where Credit is Due).   For example, rather than just citing “Bureau of Labor Statistics,” provide a more detailed citation:  e.g.,  "Bureau of Labor Statistics, U.S. Department of Labor, 2014–2015 Occupational Outlook Handbook, [date accessed] [http://www.bls.gov/ooh/]."

IMPORTANT FINAL WORD: Use complete and correct citations (really small footnotes or references fine -- or perhaps use footnotes on one page and have a separate "sources" page). Refer to all sources used (including data, maps, images, tables, graphs, course readings and materials found on the Internet). Please familiarize yourself with standard practice of academic integrity in coursework. --> See this link for complete information.