Contents

“Don Juan’s task, as a practitioner making his system available to me, was to disarrange a particular certainty which I share with everyone else, the certainty that our “common-sense” views of the world are final.  Through … well-directed contacts between the alien system and myself, he succeeded in pointing out to me that my view of the world cannot be final because it is only an interpretation.”

                        A Separate Reality; Further Conversations with Don Juan

                                                            --- Carlos Castaneda

I.  The Consequences of Being a Down-hill State

State LMI Offices generally do not have access to the types of data needed to respond to many customer needs and expectations for information and analysis.  Unemployment Insurance wage records merged with other administrative records and integrated with survey data represent a comparatively inexpensive way to expand the inventory of information and add dimension to what we know about the labor market. This chapter provides a brief description of the development of the wage records program in Wyoming and presents an analysis of some of the barriers and opportunities accounting for its current scope and development.

In order to evolve effective ways of responding to customer needs, the Department of Employment’s Research and Planning (R&P) began downloading and archiving UI wage records from the Department’s main frame in 1993.  The first download provided us with data for all of 1992 and the first half of 1993.  We first published analysis of wage records in May of 1995.  In that year, we also published our first analysis of educational outcomes using data from the University of Wyoming in cooperation with the University’s Department of Sociology and Statistics.

By mid-decade, we had moved our administrative data sets from a standalone personal computer to a server network configuration.  And in 1996, we published our first analysis of earnings by age, gender, and industry based on wage records data merged to ES-202 files and Wyoming Department of Motor Vehicles (DMV) drivers license records.  As can be seen in Figure 1, today we have data sharing agreements with seven State LMI Offices, several state agencies and state and local educational institutions.  We are also in negotiations with other state agencies to obtain copies of the new-hires directory and monthly (rather than annual access for OES purposes) access to state employee occupational, earnings, work status, and place of work and residence records.

Figure 1: Inventory and Selected Characteristics of Databases Maintained by Research and Planning 

Gaining access to DMV records represented a comparatively strait-forward process.  Our Department’s UI work unit was accessing individual records for fraud and enforcement purposes.  With the support of the Executive Director, we were able to negotiate a low-cost quarterly download of DMV records from the state’s mainframe.  Other than coping with confidentiality issues, gaining access to wage records and DMV information and archiving them has not been problematic.  Moreover, we began working with another work unit within the Department, Vocational Rehabilitation, to explore outcome reporting strategies.

However, progress in the development of Wyoming’s administrative data program has been neither steady nor linear.  In 1995, the Governor used an Executive Order to implement Workforce Development.  The Order called for a comprehensive inventory and analysis of the current workforce development system and a description of the manner in which the system interfaced with the market.  Despite the fact that Council staff and members understood the potential represented by the wage records approach, and we were able to display the potential through cooperative arrangement with Vocational Rehabilitation, our efforts were largely unsuccessful.

Our failure to produce a comprehensive analysis of the workforce development system faced several barriers.  First, an administrative records program in an LMI Office represents a potential threat to some agencies (and potentially, Sections within the SESA).  Program administrators historically controlled all of the program processes and output information for which they are held accountable.  Under the Executive Order, given the fact that wage records may not be released outside of the SESA, access to information would no longer be controlled by program managers and it would be made available to the Workforce Development Council.  Second, many state agencies, the University, and the colleges lack an organizational feature as part of their administrative structure which parallels the research function of a SESA.  These organizations lack an internal entity that can work with the LMI Office on co-equal terms.  For example, colleges, as part of their administrative structure, lack research staff who can inform administrators about the viability of data sharing proposals, identify how the information may be of use to the college donating information, and identifying how colleges can exercise some control over the output.  Finally, the Executive Order failed to provide a source of funding for R&P’s work; the agencies and colleges suspicious of our work  were the same entities that were expected to fund it.

DMV and UI records, records that were not intended for use by the LMI Office to describe the business processes of the donating agency, could be obtained by us for statistical purposes.  However, records that were intended for use in the description and analysis of the donating entities’ business practices and labor market outputs and outcomes were held close to the vest.

During this entire process, and continuing to today, R&P has had the strong support of the Department’s Executive Director and Workforce Development staff.  This support has taken the form of resources, marketing, guidance, and crafting Workforce Development Council issues in a manner amenable to research using wage records.  This support is essential but not sufficient to explaining the current development of the wage records program.  One solution to this problem has been a willing partner in the form of a local Community College, and a small amount of Council funding to treat them as a customer.  In this environment, the SESA LMI Office fills the research function for college administrators that is missing from the College’s organizational structure.

Passage of the Workforce Investment Act changed the impetus for some organizations and educational entities in terms of their willingness, and sometimes the requirement, to work with us to produce performance reports.  But resistance is still formidable.  For example, R&P prepares part of the annual report for vocational education using wage records.  However, they were initially unwilling to provide us with the minimum demographic and educational information required to meet the system goals identified in the Executive Order. 

All of the agencies and educational entities with whom we work are required to provide the basic variables to R&P that we need to produce the comprehensive system information to the Council as required by the Executive Order.   A useful strategy has been to incorporate into the Memorandum of Understanding (MOU) between a state agency and ourselves the requirement that any reports generated by us be developed at the request of the Workforce Development Council.  This provision of the MOU (see Appendix A) provides the donating agency with at least some sense of control over what will be done in the way of reporting program process, output, or outcome information.    

In the absence of an Executive Order, a mission statement, or statement of goals from the WIA five year plan legitimizing the LMI Office access to confidential information, the continued development of the wage records program would be more difficult.  The successful expansion of the system appears to depend upon:  exposure of the system’s potential to users, marketing to imaginative users with analytical skills, using a strategy of learning about customer business functions and how they envision using information to address their problems, and resources for research.  Obtaining successful results from customer relationships, and consequently word of mouth marketing through informal customer channels, requires selecting a strategy for working with customers.    

Wyoming’s natural resource based economy means that it is a state whose economy tends to experience rapid expansions and contractions.  Expansions are generally inhibited by a lack of human resources and contractions are generally occasioned by public concern over the loss of human resources. Public concern seems most acute regarding youth, namely the loss of college graduates.  It is believed that once these human resources leave, they are very unlikely to return.  The strong national economy of the 1990s was matched in Wyoming by relatively slow growth during most of the decade, with stagnation during late 1995 and much of 1996.  These underlying economic and demographic currents tended to produce an R&P client profile highly focused on the need for answers relating to migration and labor retention.  The character of Wyoming’s economy has been a key driver in the types of problems we have been asked to work on as well as a factor in developing our partnerships with other states.

Understanding a particular labor market depends upon the capacity to develop information about the context within which it exists.  The term “down-hill” state refers to states that typically export people.  This means that if the strengths and weaknesses of ones own market is to be understood, information is needed about what draws people away.   At the same time, understanding limitations to expansion means at least identifying where labor entering the state comes from.  Entering into data sharing agreements with other states, in particular neighboring states that either provide labor and/or serve as destination states for out-migration, has become an essential part of our wage records program.  A copy of the reciprocal data agreement we have with South Dakota is found in Appendix B.  This is a second generation Memorandum of Understanding (MOU).  Our first generation MOUs were time limited, contained more narrowly defined statements of purpose, and defined fewer variables.

Data sharing agreements between LMI Offices in the Plains and Rocky Mountain States tend to be designed primarily around facilitating labor market analysis.  They facilitate the analysis of inter-state labor market areas, labor migration, the policy pertinent analysis of student, teacher, nursing, and prison guards, the competition for labor in general and the development and inter-state tracking of control groups.  Although initially designed for performance reporting for specific programs, inter-state MOUs  facilitate market analysis in general rather than serving limited program purposes of dubious value.

While the character of Wyoming’s economy and the scale of population movement have been a factor influencing the manner in which the wage records program has developed, there is another important reason to establish partnerships with other states.  Not only can the experience of other State LMI Offices prove highly valuable, these partnerships are essential to the overall research strategy.  Customers often want answers only to their own questions and/or focus only on their program specific population segment.  However, our ability to conduct applied research depends upon the little appreciated fact that a prerequisite is the basic understanding of the behavior of wage records and linked databases.   This is why it is critically important to each state’s wage records program to become involved with neighboring states, especially those with a similar view about the value of a wage records program.  Sharing common research projects and/or sharing pilot and specialized results provides each state with the capacity to obtain basic knowledge without necessarily having to incur the cost of others mistakes.  Through this vehicle, we can replicate research and learn to distinguish results that are an artifact of research applications rather than a function of the market processes of interest.

Replication in other states represents a strategy in which we can obtain some assurance that produced research results are a function of market forces and the workforce development system rather than a function of state specific theoretical or computational idiosyncrasies.   From the customer’s perspective, replication represents an assurance that the research results produced by each LMI Office are “real” rather than an artifact of something unique to a particular Research Office. 

Late in the last decade we had acquired a sufficient number of observations across DMV files and wage records, and enough experience to identify error patterns that we could realistically engage in the model based imputation of missing demographic and geographic data.  All administrative data sets are characterized by update lags and time gaps between observations.  Given enough auxiliary information and history among the files, it is possible to identify place of residence between the three year DMV renewal dates.   With these developments, practical applications became more geographically relevant and therefore locally marketable.  At that point, when the informed user accepted the modeled elements of the databases as a reasonable trade off between no or little reliable information, and a reasonable approach to analysis, we entered a new period of financial support.

As we crossed into the new decade, we can produce demographics and work experience information by place of work and place of residence.  While there are some limitations to our strategies in terms of verifying decisions about who can be defined as a member of Wyoming’s workforce, the issues of geography and residency should prove themselves to be resolvable.  That is, we now seem to have most of the pieces needed to incorporate variables associated with substate locality in dynamic descriptions of economic – demographic relationships.  In addition, we have been able to incorporate more segments and data elements from the workforce development system into our archives and seem to be expanding the number of policy areas in which our research assistance is sought.

A final and most difficult barrier to overcome in the development of a wage records program at the state and local level has to do with the unquestioned assumptions we make about the way the labor market and workforce development systems work and why they function as they do.  The concepts and language of labor economics are based on traditional static measures of market levels in systems in which time is taken as an implicit given and interpretation of these statistics is fully institutionalized.  On the other hand, when working with wage records, the present is almost always defined and specified explicitly relative to the past and future.  Measuring the current stock of labor only becomes possible to the extent that we can measure the flow of labor between the present, the past, and the future.  In practice, this means that we develop the pieces for subsequent full analysis in discrete units.  We then identify the characteristics and typical behavior pattern of each element.  And then we assemble these items in relationship to one another and set them in motion.  Describing parts of the analytical framework while holding in abeyance traditional background assumptions governing the process to which these concepts will eventually be applied is a necessary prerequisite to the successful use of wage records.  

While habit makes it tempting, simply transferring the language of the CPS or other familiar systems to wage records based descriptions tends to confuse rather than clarify the new territory administrative records represents.  One of the first tasks anyone using wage records needs to undertake is to unlearn the concepts associated with the analysis and measurement of stocks, or levels, in a time series.  Most of us developed our labor market analysis careers by using a count or an estimate of labor supply, or demand, by linking together monthly or quarterly snap-shots of employment establishment or household survey data.  When we change the unit of measure to wage records, we still maintain an interest in profiling each quarter’s level of employment.  However, the far more interesting potential is the analysis of the flow information between time periods.

Shifting from stocks analysis to stocks and flows analysis in the context of inter-state data sharing agreements has the effect of adding enormous dimension to the research.  However, we initially made the mistake of pirating the language, and therefore concepts, from existing systems in an attempt to force what is at first alien phenomenon into a familiar language.  Merging alien and familiar concepts into a familiar language simply leads to confusion.  We seem to communicate more clearly with each other and customers if we adopt new language strategies which distinguish wage records analysis from anything else.

II.  Field Engineering

The process of building an administrative records system is a process of field engineering in an opportunistic setting where position is everything.  There are few formal models[1] for the development of integrated systems and little time to build one because the justification for acquiring data is often based on how quickly one can respond to the data suppliers’ need for tabulations and analysis.  The overriding effect of developing this strategy is the change in the number and types of customers we work with.

The goal of our Research Office is developing the capacity to move from descriptive research and analysis to causal analysis.  Accomplishing this goal means:

* accessing and controlling the data elements encompassing the population of human            resources, firms, and events making up the universe of interest

* having the physical capacity to store and protect those data sets

* acquiring the human capital and supporting resources in the Research Office who can competently utilize the elements we have assembled

Because our goal is the capacity to conduct descriptive analysis, and subsequently produce causal analysis, successful development of the system does not depend upon the willingness of any particular program (e.g. Carl Perkins, ABE, WIA, etc.) to partner with the LMI Office.  Rather, it depends upon the ability to gather universe information on the supply of workers and employers, and their characteristics.

As we acquire the ability to control more and more competing explanations for market effects (e.g. interstate data sharing agreements that allow us to account for wage differentials) the reasons for working with us increase.  For example, our inability to track former prison guards to work destinations in other states was at one point the announced reason for the Department of Corrections (DOC) not working with us in monitoring turnover.  Our data sharing agreements with seven states now invalidate this concern.  In effect, we have positioned ourselves to address DOC’s personnel issues as well as the interests of many other customers.

Positioning also means marketing, and networking with decision-makers and other researchers within similar arenas --- especially those who have the capacity to adapt the examples of our published work to their particular area of need.  This informal network of analysts and decision-makers, it appears, have become advocates.  In one way or another, they convey the message that the LMI Office can address their problems (see Appendix C).

Marketing also means publication of results from our research.  All of the contract agreements we have with consultants, colleges, and program operators permit the publication of results.  The process of developing research and evaluation results for publication is probably best developed in a cooperative environment with the data supplier.  This strategy appears to be effective in trust building.

Until we begin to see more LMI Offices step forward with an account of what they are learning and how they proceed, we are pretty much stuck with one set of anecdotes.  The constant that binds these anecdotes, and that distinguishes them from many other research activities LMI Offices engage in, is that they always involve the use of confidential information, and they nearly always have a political consequence.

We like to think that what LMI Offices produce raises the overall stock of knowledge and that everyone benefits.  This implies that everyone has an equal stake in changing the current state of affairs based on new knowledge.  However, it must be remembered that some institution always has a vested interest in the current state of knowledge.  And sometimes, institutions have a greater interest in the current state of knowledge, and who has access to it, than they do in the unregulated development of more information.  In the initial study of prison guard turnover (conducted in 2000), for example, it was found that guards who went to work outside the prison setting were often earning less than they had when working for the State.  This information, along with guard exit interviews conducted by the Legislative Service Office (LSO), led to the recommendation that DOC make major management changes.  In effect, the increase in knowledge resulted in a shift in authority away from the DOC in the Executive branch, to the Legislature and LSO.  A shift in authority and a loss of control is not something every agency manager is comfortable with.

Analysis we completed with the Nursing Association during the last legislative session could be viewed as having the effect of changing the locus of authority as a function of more information.  In the matter of perhaps a month, working with the Association, we produced more factual information about the work experiences of nurses in Wyoming than was previously available from all institutions affiliated with the health care field combined.  Historically, these entities were the authoritative source of information about the shortage of nurses, enjoyed the prestige of that authority, and because of it, were able to influence the manner in which the issue was dealt with. 

Wage records based analysis, and its distribution to the legislature, changed the institutionalized understanding about where information concerning a critical labor shortage could, and potentially should come from.  These entities may have lost some authority, and potentially the control over how state government will deal with an issue in which they have a vested interest.  New knowledge can challenge the organizational structure of existing authority and influence.  (Consequently, part of the positioning strategy is to always make the overture to work collaboratively with any vested interest as often as the opportunity presents itself.)  By changing access to, and control over, information about a critical market problem, wage records analysis frequently changes the balance of (or democratizes information about) influence.  This consequence is best planned for.

The strategy of developing universe information permits the development of several outputs (as distinct from the quantifiable outcomes or benefits from it) sought by our Workforce Development Council.  Wage records linked to the ES 202 and DMV files were used to create the profile of workers in Wyoming found in Table 1.  Program participants in a state incumbent worker program supported by the Workforce Development Training Fund are found in Table 2.  And Table 3 provides a tabular description of JTPA/WIA participants.  Tables 2 and 3 are developmental tables and are intended simply to portray who is served across programs in the context of comparable information about the larger workforce.

Multiple reporting platforms for each federal and state program encompassing varying time frames, from fiscal to program and school years, prevent our Council from gaining a consistent overview of who it is that the workforce development system serves and where the people it serves fit in to the overall context.  Eventually each client group for the system as a whole (including employers) will be grouped in a standard way and this information shared with the Council. 

Ultimately, this strategy serves a second purpose.  The second purpose is to lay the foundation for causal analysis.  Normally, we develop component pieces individually, and as you will see in the papers focusing on turnover, and quasi-experimental design and multivariate analysis, we subsequently set the pieces in motion.

Table 2:  1999 and 2000 WDTF Participants Appearing in 2000 Wage Records (859 Total Participants)
		Age Group						Total
		<=24	25-34	35-44	45-54	55+	N/A
Agriculture	Count	1	1	5	5	2		14
	Row %	7.1%	7.1%	35.7%	35.7%	14.3%	0.0%	100.0%
Mining
Total	Count	3				1		4
	Row %	75.0%	0.0%	0.0%	0.0%	25.0%	0.0%	100.0%
Construction
Total	Count	3	4					7
	Row %	42.9%	57.1%	0.0%	0.0%	0.0%	0.0%	100.0%
Manufacturing
Total	Count	11	26	21	6	2	1	67
	Row %	16.4%	38.8%	31.3%	9.0%	3.0%	1.5%	100.0%
Transportation, Communications & Public Utilities (TCPU)
Total	Count	8	10	25	5		2	50
	Row %	16.0%	20.0%	50.0%	10.0%	0.0%	4.0%	100.0%
Wholesale Trade
Total	Count	3	2	1				6
	Row %	50.0%	33.3%	16.7%	0.0%	0.0%	0.0%	100.0%
Retail Trade
Total	Count	104	108	71	64	18	1	366
	Row %	28.4%	29.5%	19.4%	17.5%	4.9%	0.3%	100.0%
Finance, Insurance & Real Estate
Total	Count	7	15	6	10	2		40
	Row %	17.5%	37.5%	15.0%	25.0%	5.0%	0.0%	100.0%
Services
Total	Count	51	45	28	13	4		141
	Row %	36.2%	31.9%	19.9%	9.2%	2.8%	0.0%	100.0%
Government
Total	Count	15	20	3	2	1		41
	Row %	36.6%	48.8%	7.3%	4.9%	2.4%	0.0%	100.0%
Not Available	Count
	Row %	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%	0.0%
Grand Total	Count	206	231	160	105	30	4	736
	Row %	28.0%	31.4%	21.7%	14.3%	4.1%	0.5%	100.0%
Prepared by Douglas Leonard - Wyoming Department of Employment - Research and Planning Section, 3/18/2002.

Table 3:  WIA Participants (1999 & 2000 Calendar Years) Appearing in 2000 Wage Records (895 Total Participants)
		Age Group						Total
		<=24	25-34	35-44	45-54	55+	N/A
Agriculture	Count	5	2					7
	Row %	71.4%	28.6%	0.0%	0.0%	0.0%	0.0%	100.0%
Mining
Total	Count	22	10	6	7			45
	Row %	48.9%	22.2%	13.3%	15.6%	0.0%	0.0%	100.0%
Construction
Total	Count	29	13	12	4	2	1	61
	Row %	47.5%	21.3%	19.7%	6.6%	3.3%	1.6%	100.0%
Manufacturing
Total	Count	22	6	4	1	2		35
	Row %	62.9%	17.1%	11.4%	2.9%	5.7%	0.0%	100.0%
Transportation, Communications & Public Utilities (TCPU)
Total	Count	6	11	8	10	1		36
	Row %	16.7%	30.6%	22.2%	27.8%	2.8%	0.0%	100.0%
Wholesale Trade
Total	Count	12	3	3	3	2		23
	Row %	52.2%	13.0%	13.0%	13.0%	8.7%	0.0%	100.0%
Retail Trade
Total	Count	176	22	14	10	4	2	228
	Row %	77.2%	9.6%	6.1%	4.4%	1.8%	0.9%	100.0%
Finance, Insurance & Real Estate
Total	Count	4	3	1	4			12
	Row %	33.3%	25.0%	8.3%	33.3%	0.0%	0.0%	100.0%
Services
Total	Count	136	33	35	20	11	4	239
	Row %	56.9%	13.8%	14.6%	8.4%	4.6%	1.7%	100.0%
Government
Total	Count	56	9	13	7	2	3	90
	Row %	62.2%	10.0%	14.4%	7.8%	2.2%	3.3%	100.0%
Not Available	Count		1					1
	Row %	0.0%	100.0%	0.0%	0.0%	0.0%	0.0%	100.0%
Grand Total	Count	468	113	96	66	24	10	777
	Row %	60.2%	14.5%	12.4%	8.5%	3.1%	1.3%	100.0%
Prepared by Douglas Leonard - Wyoming Department of Employment - Research and Planning Section, 3/18/2002.

III.  User Characteristics and the Field Situation

From the series of encounters with customers, it seems possible to identify key variables and types of situations that re-appear with regularity and that, given the latitude, guide our decisions in how we work with them.  This guidance has remained implicit in the development of our administrative records program.  We will try here to codify these elements of the “field” in which we conduct our engineering.

One of the critical elements in any customer relationship, whether they are donating program data or not, is to obtain an identification of the concepts of interest – the “what” if you will, rather than the “how.”  Some customers approach a research problem in terms of satisfying themselves that a question could be answered by mentally designing the components, and the manner in which they are to be manipulated.  This problem tends to arise with all customers whether Ph.D. consultant or city planner.  Illustrative examples of customers grappling with the issue of stating their interests in a useful way are found in the second half of this chapter.

The customers we most often encounter generally represent a client group found in the first column of Figure 2.  Or, as is often the case with economic development, they represent an industry segment within a particular geographic area.  Sometimes they are interested in the effect of educational inputs on market interactions, identified in the second and third columns of Figure 2.  We are frequently simultaneously conducting several analyses across a number of paths and different levels of interaction represented in Figure 2.

Almost uniformly, customers are interested only in their own target segment or client group across a single path of analysis.  Comparative analysis, the use of analytical controls, or the use of control or comparison groups to account for competing explanations for observed outcomes, are for the most part foreign concepts, and certainly not additional costs worth incurring.  Given the opportunity, and we have had a few, we always elect to work first with customers who understand the purposes of research design.  For others interested in outcomes analysis, rather than simply description, we usually try to explain the advantages of including controls.  Their receptivity to these proposals generally provides us with an indication about how realistic the customer’s expectations are.

Each of the following characteristics, issues, and questions about the customer and/or the field situation is used to set our research priorities.

1.  Are the goals of the project consistent with our primary roles in data collection and estimation, in support of labor market analysis, or the workforce development system?

2.  Is the customer or client willing?  Willing customers are more open to providing documentation, interpretive insights, and making themselves available to answer questions.

3.  Is there an organizational imperative for the client to produce an administrative records based result?

4.  Is there an organizational imperative (contract, legislative mandate) real or implied for our Section to produce the same or related outputs?

5.      Is the project data rich both in content and quality?

6.      Is the customer a good manager of data and related systems?

7.  Does the project present us with the opportunity to showcase our capabilities or extend our capacity to conduct new types of research?

8.  Does the project allow us to maintain momentum?  The project may not be the most attractive.  However, given the balance of activity in the field, it may allow us to maintain our direction and expand the scope of our experience.

9.  Is the client’s decision making centralized?  Some projects involve program staff with little research background but decision-making authority while the technical background resides in staff with little authority.  In other situations, the project lead for the customer’s interests is not empowered to make critical decisions.

10.  Do any of the principals have an overwhelming background?  Sometimes the consultant is too knowledgeable about survey research or the way analysis should be conducted to learn about the strengths and weaknesses of administrative databases.

11.  Does the client have clear articulated goals? (Are they familiar with our previous work?)

12.  Does the customer provide funding, and to what extent is its use restricted? 

13.  Is the customer interested in applying the results to produce an identifiable outcome?

14.  Does the customer realize that the scope and scale of the data provided to us represents the limit of the services we can provide to them?

15.  What is the probability that the customer represents a long-term and stable relationship as part of the workforce development system or a key source of data? 

Given the fact that our “field” is simultaneously filled with a number of different customers and data partners at any given point in time, we need some mechanism of organizing our efforts.  The fifteen points listed above play an important role in setting our priorities.  In the balance of this chapter, we review elements of three case histories that illustrate selected problems associated with some of the issues enumerated above.

While customers come to us with many different characteristics, some customers represent longer-term relationships.  As a result R&P is involved in an increasingly complex set of inter-organizational relationships.  If LMI is to serve as the common language among components of the workforce development system, then administrative records represents one area in which the development of that common language is likely to occur.

The major customer for wage records analysis is the workforce development system.  In Wyoming, the Workforce Development Council has responsibility for all One-Stop programs as well as all other components of the workforce development system.   Section 136 (c)(B) of WIA states that local levels of performance shall be determined relative to “the specific economic, demographic, and other characteristics of the populations to be served in the local area.”  The focus on local information is mirrored throughout the balance of the workforce development system.  Since performance measures for WIA, and other programs, are increasingly based on wage records, it seems prudent to measure local economic and demographic dynamics based upon a wage records strategy.  Further, it is difficult to determine how local LMI could efficiently be developed through other techniques.

Some customers have goals that are highly consistent with WIA’s focus on local LMI. Some of these customers have financed small-scale research projects that have the additional benefit of assisting us in developing our capacity to produce localized LMI.  One of these entities is local economic development.  Publishing commuting pattern data for a transportation planning consulting firm[1] led to additional contract work for a local Northeastern Wyoming economic development agency.  Staff from this local economic development entity recently sent the following email:

Hi Tom -

I just wanted to let you know that I have talked to Randy Brxxxs in Cheyenne

and Tim Stxxxx in Laramie about the commuting study.  Hopefully they will be

contacting you about completing one for their areas.  Also if you all

complete the commuting patterns for the state will you please let us know.

If you think of any other data that might help us in our grant and economic

development efforts please contact me.

Thank you,

Erin Alsxxx

CCEDC/NEWEDC

ena@vcn.

307-686-0000

www.ccedc.net

www.newedc.net

The Northeastern economic development agency has a very specific agenda of applying for Economic Development Administration (EDA, US Department of Commerce) grants.  Successful applications are based, in part, on the ability of applicants to demonstrate that a potential grant recipient county is a destination county for workers commuting from economically “distressed” counties.   Our ability to anticipate this customer’s needs, and those to whom she is recommending our services, is partly a function of our ability to keep up with information about EDA grant requirements, and partly a function of the customer base’s ability to communicate their needs to us.  Subsequent steps have been taken with Northeastern Wyoming economic development and additional contract work is in negotiation.

The publication of the commuting analysis for the Teton County labor market area also seems to have sparked the following request from their local housing authority:

Teton County Housing Authority 2/15/02 Request

Ok...

This is probably very convoluted... So please call me if you need more

information. I tried to include our "objective" (very grammar school, I

know) at the end of all the ballyhoo so you could see what our purpose

is.... Hopefully this will help you. Sources of data are going to be

important... as are certain limitations of the data, so if you could flag

anything you know is going to be missing or misleading in any data sources

... just let me know when you write back with a quote.

Number 1 we definately need so could we get a price on that separately?

We definitely need income quartiles for the purpose of seeing the

distribution under and around the imaginary median income number that is

skewed by those people with billions.

1. Part 1: income in 5 income breaks (0r ten?)

                by number of people (all Teton County)

                for most recent year

Part II: (all Teton County) each person by SSN tallied by their total

combined multi-job or single job              salary broken out....

 Objective: to see what HH income is for buyers so we can set prices.

2. and then Sic by the 5 income breaks by year

                for instance

1999

                                                less than 10k           10-20       21-30                       Total

construction                            2(1%)                      4 (3%)     20(10%)                  26 (100%)

FIRE

etc.

etc.

Percentages                            7%                          10%

by Quintiles/???

(or what ever)

Objective: to see what the largest industries (ie retail, service and

construction) are paying

3.    the number of people who worked in Teton County then left (time series

on SSN's)

                and if possible could we see how may jobs those people work?

                like... 5000 people left between 1999 -2000 and 60% had one job, 20% had

two jobs, 20% had                 three jobs ??? So what number of in and out people...how

many jobs by people?

                Objective: Are transient workers more likely to work more than one job? (my

guess is no)

                And then the same analysis by SSN's of people that are employed year after

year in TC...

                This would help us chip away at the huge "job" number and low people

number... we need                 to fill in the discrepancy between jobs and people...

commuters are one piece, but                numbers commuting in from Idaho are

nonexistent, illegal workers?...an estimate at best,  and the                large number

of people who live year in and year out with four different seasonal

jobs.... this is what we want to know.

                Objective: What percentage and number can we subtract from the CBP count

for multi-job             holders...

4.             Pursuant to above analysis what are the real incomes then of the human

beings in Teton        County THAT WORK? We look at  data that shows x number of

people employed in retail, at x                average wage? What does the yearly salaries

from multi-job combinations of people add up        to? This is actually

complicated because if you don't know the time on these jobs you       won't know

if they were             concurrent or new jobs.  People who worked 3 jobs in 2000

account for X number of people, X         percentage of labor force, and made an

average of x$...     Is it possible to run what field                 constantly overlap, X

percent of people with two jobs             worked in service and construction         and had

an average yearly combined wage of X????

                If we could run it off tax reports that would be great.

5.             We still are interested in what percentage of our population is

commuting.              (I know    impossible because of Idaho)

Thanks,

Cindy Norxxx

TCHA

307-732-0000 (p)

307-732-0000 (f)

As the result of this email, and several subsequent discussions and emails, we initiated a small contract with the Teton Housing Authority.  The first three issues were addressed in the contract. 

However, this email is noteworthy on two other accounts.  First, had we in place an agreement equivalent to the MOU found in Appendix B with Idaho.  Had Idaho been using DMV files in conjunction with wage records, we could have involved the Idaho LMI Office in a joint analysis of the Teton County, Wyoming, and Teton County, Idaho, labor market areas.  This arrangement would have allowed us to address issue number five in the Housing Authority’s list of requests.

Second, although the customer is still specifying relationships between variables, she has also been maneuvered from only specifying how the LMI Office should manipulate data, to explaining the purposes of their request.  She also has begun to identify the assumptions they are using to configure relationships among variables, and the concepts they wish measured.  Once we move the customer from specifying “how” the work might be accomplished to “what” they want to achieve, we are in a much better position to design the product that they can use.

Maneuvering the customer to work at the conceptual level in specifying their concerns and questions is an essential step in permitting Researchers to develop viable and acceptable ways to answer user questions.  A purpose in enumerating user and field characteristics used in setting our work priorities is to develop ways to better serve them.  Customers presenting us with computational proposals often seem not to have worked through at the conceptual level the type of information that would address their questions.  At this stage in our development, we are trying to outline a user manual or check list for customers to facilitate description of the problem they want solved.

Clearly, part of that manual needs to be comprised of the basic definitions of occupation, firm, industry and many of the other building blocks we tend to take for granted. 

Sometimes it appears that customers purposely seek to conceal their overall agenda.  In other cases, it appears that they need to work out the manner in which the measures and variables should be operationalized and arranged to answer the often-implicit questions they are seeking to answer.  The client, in some cases, is interested in playing the role of analyst rather than the role of customer.  Teaching people to be good customers is essential, but sometimes very problematic.

Some customers are so focused by a certain set of background assumptions that the role of student is no longer part of their repertoire.  Recently, we provided services to two sets of consultants for the analysis of interstate teacher migration and turnover. This work was largely conducted through merging of confidential teacher files, wage records, ES 202 data, and inter-state matches.   The State Department of Education financed our work.   However, the analysis proved unsatisfactory in many respects. 

Despite providing the consultants with background orientation materials, the consultants developed all of the cross tabulations strategies with little apparent understanding of the strengths and weaknesses of the wage records approach.  Description and hypothesis testing were conducted using strategies grounded in conventional, conversational views about how the world works, rather than on exploratory strategies in which measurement results are used inductively to shape theory and subsequent data manipulation.   

For the consultants, in most cases, if the results of our matches and tabulations did not fit with the proposition or hypothesis implicit in consultant specified variables, the results were ignored.  Alternative cross tabulation schemes we developed were adopted by one consultant but not the other.  Earnings gains for out-migrating teachers were specific to a particular gender and age group.  Middle aged male (high school) teachers had materially higher earnings following out-migration.  Earnings for females of middle age were largely unchanged or decreased following out-migration.  These results were apparently not within the scope of the analysis nor part of policy development because they were never published.  This finding, however, raises the following consideration:  If migration, except for the brief period after college graduation, is a function of household decisions made relative to the female age period associated with child bearing and career choice, and the age period associated with male based career choice, then we would expect to see the same earnings differences between the genders at middle age following migration regardless of occupation that we found for teachers.  The test of this proposition requires a control or comparison group.  And an answer is possible because we have the universe of workers and many of their characteristics.  Finding the time and resources to conduct the research, however, is a problem.

The elements of these three case studies illustrate how some of the questions enumerated at the beginning of this chapter are sometimes answered in the context of working with several customers at the same time.  If we recognize these issues, make them explicit, and realize that we are using the answers to guide our actions during the process of field engineering the system, we are more likely to rationally approach customers and manage our responses to attain common ends. 

IV.  Confidentiality and Public Accountability

Above and beyond the normal security conditions associated with confidentiality issues, there is a need for a conscious strategy relating to privacy and confidentiality when obtaining and working with administrative data from other agencies and other Sections within a SESA.  However, there are few institutionalized safeguards or protocols within the State Research structure providing for either security or accountability.  These failures both inhibit the development of data sharing agreements and could prove fatal to LMI programs based on the use of administrative records.  In addition to involving the concern over the privacy of individuals, data sharing agreements imply a joint liability between the sending and receiving entities. Therefore, there is a need to develop, articulate, and institutionalize confidentiality and privacy protections in Research Offices beyond those already in place.

Educational institutions and State and local agencies have their own confidentiality protections for the data they may share with  a Research Office.  Each entity is quite likely to be represented by an attorney versed in the institution’s particular confidentiality statutes and, in the case of federally funded agencies, the related federal statutes and regulations.  Due simply to the function of the job, the attorney is unlikely to be as knowledgeable about the agency or institution’s requirements for human resource planning or accountability that can be addressed through research strategies based on records matching.  Opening an initiative with another state agency means conducting the appropriate research into governing statutes and planning and accountability requirements, as a first step.

Overriding all data sharing discussions between state agencies are concerns for the protection of the privacy of employer information, and in the case of persons, their civil liberties as well.   When opening discussions with another institution it is often useful to have available information about the confidentiality and security requirements Research Offices and their parent agency must meet as well as copies of contracts with any similar institution.  As our Research Office obtains more agreements between agencies, colleges, and commissions overseeing professional licensure, these issues have become less often the focus of overt concern. 

One strategy that appears to facilitate the ongoing acceptance of data sharing agreements with the Research Office in Wyoming is publication of research results and an explanation of the data matching mechanisms used to achieve these results.  Publication of results is not simply treated as information dissemination, we also attempt to communicate the value-added justification for data sharing as an aspect of public accountability.  If we are to claim that data sharing benefits outweigh data sharing risks, it seems that we need to constantly demonstrate beneficial outcomes.

Unfortunately, beyond State UI statutory requirements and the BLS Commissioner’s Orders on Confidentiality, there is no institutional structure supporting awareness among Research Offices of current or historic issues related to confidentiality and privacy. The concepts of statistical standards of performance, including the obligations for public accountability, confidentiality, and privacy are substantially underdeveloped among State LMI Offices.  The failure to systematically address these issues is perhaps the largest single system-wide threat to the long-term expanded use of wage records

      D.        Require all Agency employees or agents who will have access to WDE information to sign a Confidentiality Agreement.

E.         Designate an employee representative to act as the contact person for communications concerning the purposes of this MOU.

F.         Provide statistical reports to WDE upon request. Costs for the reports shall be agreed upon by the parties prior to the provision of reports and will be based upon approximate actual costs to the Agency for such services. Copies of all published reports will be provided to WDE.

G.        Produce summary statistics related to labor market dynamics and outcomes in written form for public consumption as requested by the Workforce Development Council.

H.   Destroy all personally identifiable information received from WDE pursuant to this MOU when it is no longer needed for the purposes for which it is requested.

6.         Responsibilities of Wyoming Department of Education. WDE shall:

        A.        Provide the Agency with electronic copies of the current and prior two years of   student data for all available vocational students.

        B.        Provide the Agency with current student data for all vocational students during the first full week of (month to be provided here by WDE).

C.        Designate a representative with whom Agency staff can consult on an as needed basis.

D.        Provide documentation for technical information relating to the vocational student file and notify Agency of substantive changes to the files’ content.

E.   Provide the following variables for each vocational student record: ssn or student identification number identifier, ssn or student number, date of birth, gender,  race, resident status, state of origin, county of origin, gpa, academic program, school or institution code, degree or certificate type, major, cip, hours, standing, start term  date, graduation date, last name, first name, middle name (documented equivalents are acceptable).

F.   Provide spread sheet examples identifying the computations to be applied to each variable, and the variable identification, in statistical reports the Agency is to provide to WDE for outcome measures.

G.        Ensure that Agency statistical reports and analysis reproduced by WDE and its contractors include source citation referencing the AWyoming Department of Employment, Research and Planning (R&P) and the date the statistical report was produced as provided by R&P.

                                    First Name                   Alpha               30 characters

                                    Middle Name               Alpha               30 characters

                                    Year                             Alpha               4 characters

                                    Quarter                        Alpha               1 character

                                    UI account #                Alpha               15 characters

                                    Qtr. Earnings                Numeric           8 numbers(whole dollars)

                                    Industry (NAICS)        Alpha               6 characters

                                    Ownership                    Alpha               1 character

                                    State ID (FIPS)            Alpha               2 characters

                                    County ID (FIPS)         Alpha               3 characters

                                    Resident of state           Yes or No        3 characters

                                    Enrolled in State           Yes or No        3 characters

                                    Type of Enrollment       Alpha               30 characters 

                                                2)         The Reciprocating State will perform the comparison of the Requesting State records to administrative data. 

                                                3)         The Reciprocating State's obligation to provide information is contingent upon the availability of the information within its own computer system.               

                                                4)         The Reciprocating State shall not be liable for any and all damages, including consequential damages, arising from inaccuracies in the information provided.  

Wyoming Legislative Service Office

213 State Capitol Cheyenne, Wyoming 82002 Telephone: (307) 777-7881 Fax: (307) 777-5466 E-mail:

lso@state.wy.us Website: http//legisweb.state.wy.us

Memorandum

Date: November 26, 2001

To: Members, Management Audit Committee From: Program Evaluation Staff

Subject: Preliminary Follow-up Information

Turnover Among Correctional Officer Staff

At the July meeting, members directed LSO staff to provide updated information about turnover among correctional officers at the Wyoming State Penitentiary. The Committee asked the following question: Did the pay raises that went into effect in July 2000 lead to higher correctional officer retention and reduced turnover at the Wyoming State Penitentiary? Members asked for this information by December 1, 2001, as a preliminary to the follow-up scheduled for the Spring of 2002 on the report Turnover and Retention in Four Occupations.

We worked with two agencies to fulfill this request. First, we asked the Department of Corrections to respond to a series of specific points from the original report, and to provide the most recent data possible. They sub-contracted

part of the research and submitted a lengthy response with attachments, all of which are included as Attachment A. 

Second, we contracted with the Department of Employment, Research and Planning Section (R&P), to provide wage and employment data and analysis regarding the individuals who terminated employment as correctional officers since the July 2000 pay raise went into effect. Our summary of their research is included as Attachment B and contains more detail than we include here…

Wage Records Research Index | Labor Market Information