Contents

• Introduction
• WIA State Formula Grant Program Background
• Dislocated Worker Program Purpose and Eligibility
• Dislocated Worker Formula Funding Structure
• Trends in State Formula Grant Funding
• Dislocated Worker Program Formula Funding
• Formula Funding—General Concepts
• Dislocated Worker Formula Factors
• Examination of the Dislocated Worker Formula
• How Closely Do the Formula Factors Align with the Population on the Basis of Which Funding is Intended to be Provided?
• How Volatile Are Annual State Funding Allocations?
• The Construction of the "Excess Unemployment" Factor Creates Volatility
• Excess Unemployment and the Problem of a Threshold
• Relative Share and Changes in Factor Values
• Excess Unemployment and Funding Intensity
• Examination of Alternative Formula Factors
• Alignment of Alternative Formula Factors
• Civilian Labor Force
• Mass Layoffs (ML)
• Extended Mass Layoffs (EML)
• Unemployment Insurance—First Payments and Exhaustees
• Job Losers and Persons Completing Temporary Jobs
• Volatility of Alternative Factors
• Civilian Labor Force
• Mass Layoffs and Extended Mass Layoffs
• Unemployment Insurance—First Payments and Exhaustions
• U-2—Job Losers
• Current Factors
• Examination of Alternative Formula Features
• Factor Weight
• Hold Harmless and Stop Gain
• Small State Minimum
• Indexation
• Indexation in the Dislocated Worker Formula
• Conclusion

Summary

The Workforce Investment Act (WIA), enacted in 1998, is the federal government's primary employment and job training legislation. Title I of WIA—Workforce Investment Systems—authorizes job training and related services to unemployed or underemployed individuals. Funds authorized under Title I, Subtitle B of WIA are allocated to states by formula and are used for workforce development activities.

This report analyzes the current allocation formula for one of the three Title I formula grant programs—the dislocated worker program, which is the largest of the three Title I grant programs with annual funding of about $1.2 billion. The dislocated worker program is intended to fund employment and training activities for a specific group of unemployed individuals—dislocated workers. Dislocated workers are distinguished from the general category of adult unemployed individuals for purposes of funding allocations and are defined in WIA by specific criteria such as being part of a mass layoff.

Interest in the dislocated worker program's allocation formula has increased recently because some high unemployment states have lost substantial funds under this program in recent years and there has been significant volatility in year-to-year state funding allocations under this particular funding stream. This report examines the funding formula and explores the pros and cons of alternative allocation procedures. The analysis in this report leads to three conclusions:

• First, the current dislocated worker formula does not align fully with the WIA-defined population of dislocated workers. The factors used for allocating funds in the formula may include but do not focus on the targeted population of dislocated workers.
• Second, as currently configured, the dislocated worker formula results in volatile changes in state allocations from one year to the next. While some volatility in allocations may be desirable so that funds go to the areas of greatest temporal need, the mechanics of the current formula tend to exaggerate the underlying volatility of the factors and lead to large annual changes in the intensity of funding (i.e., funding dollars per funding factor).
• Third, there are several alternatives to the current factors in the dislocated worker formula and several alternative formula features that may potentially offer better alignment with the target population and reduced volatility. In addition, there are options available to mitigate the large swings in specific state annual allocations under the dislocated worker formula.

---

An Analysis of State Formula Grants for Dislocated Worker Activities Under Title I of the Workforce Investment Act (WIA)

Introduction

The Workforce Investment Act (WIA) was enacted in 1998 and replaced the Job Training Partnership Act (JTPA) as the federal government's primary employment and job training legislation.¹ Title I of WIA—Workforce Investment Systems—authorizes job training and related services to unemployed or underemployed individuals. Funds authorized under Title I, Subtitle B of WIA ("Statewide and Local Workforce Investment Systems") are allocated to states by formula and are used for workforce development activities. The three formula grant programs in Title I—youth, adults, and dislocated workers—authorize funding for employment and training activities available through the national system of One-Stop centers and are provided by service providers in local communities. Title I of WIA is administered by the Employment and Training Administration (ETA) of the Department of Labor (DOL).

The majority of funding for WIA Title I programs is provided through state formula grants. For example, in program year (PY) 2010 nearly $3 billion of the total Title I funding of $5.5 billion was allocated by state formula grants.² WIA allocation formulas are specified by statute, which defines formula factors, specifies factor weights, and prescribes the distribution of funds to states and localities. The allocation formulas take the form of mathematical equations through which DOL calculates the specific grant amounts to each state.

The dislocated worker program is the largest of the three Title I grant programs and is intended to provide funding for employment and training activities for a specific group of unemployed individuals—dislocated workers. In recent years, the dislocated worker program has provided approximately $1.2 billion annually in formula grants to states, the majority of which was subsequently suballocated to local areas within states.

Interest in the dislocated worker program has intensified recently for several reasons. First, while the dislocated worker formula is, partially by design, more volatile (i.e., the size of an individual state's grant varies from year to year) than the two other state allocation formulas used in WIA (Adult and Youth programs), the recent recession and the large and rapid rise in the number of unemployed it produced has magnified the formula's allocation volatility. In the past two to three years, several states have seen funding declines from the prior year in excess of 40%. In addition, some states with high and persistent unemployment were the hardest hit in terms of funding cuts, even while experiencing rising levels of unemployment. This recent pattern of large, unpredictable, and in many cases countercyclical, swings in funding fails to meet one of the underlying aims of the dislocated worker program—allocation of funds to states in a manner that consistently reflects their absolute and relative need and that consistently corresponds with changes in the state level of need.

Second, the large shifts in state funding from year to year make it difficult to plan, budget for, and operate a program that is designed to be responsive to economic dislocations. When states have rising unemployment, it becomes more difficult to provide employment and training services with level funding. However, when funding declines, in a manner and at levels that cannot be anticipated, in times of rising unemployment, it is unlikely that states can consistently operate cohesive programs and can meet the demand for services.³

Third, in addition to concerns about the dislocated worker program formula leading to significant volatility and concerns about a mismatch between changes in unemployment and changes in funding, a fundamental misalignment between the formula factors and the intended population of the formula has also come under some scrutiny. WIA authorized a formula specifically to allocate resources for dislocated workers—a subpopulation of unemployed workers in general. Yet the formula as currently designed does not incorporate a measure that specifically identifies this subpopulation of workers. Thus, the stated intent in WIA of targeting funds separately for dislocated workers does not match the formula specified in statute to allocate funds for this population.

As reauthorization of WIA is considered, there is substantial interest in exploring ways to improve the formula design so that it better aligns with the statutorily specified goals of the dislocated worker program. This report addresses issues related to the dislocated worker program allocation formula by

• examining the extent to which problems may exist in the current formula regarding alignment of the formula factors with the program aims and volatility in year-to-year allocations;
• examining how aspects of the current formula's design may contribute to misalignment and volatility; and
• analyzing the strengths and weaknesses of alternative design options that may have the potential of improving alignment and reducing year-to-year volatility in state grant levels.

WIA State Formula Grant Program Background

A majority of funding for programs authorized by Title I of WIA is allocated by formulas.⁴ WIA Title I state formula grant funds are allocated from ETA to individual states by formulas (discussed in the sections below). There are three state formula grant programs in Title I: adult, dislocated worker, and youth. From these allocations, a portion of the funds are retained by the states for statewide activities and the remainder of the funds are suballocated to local Workforce Investment Boards (WIB). There are both required and allowable statewide and local employment and training activities.⁵

Statewide required and allowable employment and training activities include dissemination of the state list of eligible providers of training services, evaluations of state workforce investment programs, assistance to local areas in establishing One-Stop delivery systems, operation of a fiscal and management accountability system in order to report on and monitor the use of WIA funds, and various forms of technical assistance.⁶

Local required and allowable employment and training activities are centered on establishing and operating a local One-Stop delivery system and providing core, intensive, and training services.⁷ The programs for adult and dislocated worker participants in WIA are structured around a sequential service strategy that consists of three levels of services: core, intensive, and training. Service at one level is a prerequisite for service at the next level.⁸ That is, any individual may receive "core" services (e.g., job search assistance).⁹ To receive intensive services (e.g., career planning), WIA requires that individuals be unable, after receiving core services, "to obtain or retain employment that allows for self-sufficiency." To receive training (e.g., occupational skills, on-the-job), after receiving intensive services, an individual must have been unable to obtain or retain employment that allows for self sufficiency. Further, to be considered for training, an individual must also have the "skills and qualifications" to participate successfully in training (as determined by a One-Stop case manager), choose a training service linked to an occupation in the local area (or be willing to locate to another area where the occupation is in demand), and be unable to obtain other grant assistance (e.g., Pell grants) for the training services.

The dislocated worker program is the largest of the three WIA state formula grant programs, representing in FY2010 nearly 40% of total state formula grant funding ($1.2 billion of the approximately $3 billion in state formula funding). Unlike the two other formula grant programs in Title I of WIA—Adult and Youth—the dislocated worker program clearly targets a subpopulation within the larger category of unemployed workers. That is, WIA carves out a separate program and separate funding stream for a population that is identified by a particular type of unemployment. In addition, and unlike the WIA programs for youth and adults, the dislocated worker program receives an annual allocation of more than $200 million ($229 million in FY2010) to fund the National Reserve, which funds, among other items, national emergency grants and various types of programs to respond to major economic dislocations. The funds in the National Reserve are allocated at the discretion of the Secretary of Labor but are part of the total funding for dislocated workers.

Dislocated Worker Program Purpose and Eligibility

The dislocated worker (DW) program provides training and related services to individuals ages 18 and older who are classified as "dislocated." WIA defines a "dislocated worker" as an individual who¹⁰

• has been terminated or laid off (or has been notified of a termination or layoff), is sufficiently attached to the workforce (demonstrated either through eligibility for/exhaustion of unemployment compensation or through other means), and is unlikely to return to the previous industry or occupation;
• has been terminated or laid off (or has been notified of a termination or layoff) as a result of any permanent closure of (or substantial layoff at) a facility or is employed at a facility that is scheduled to close within 180 days;¹¹
• was self-employed but is unemployed because of general economic conditions in the individual's community or because of natural disasters; or
• is a displaced homemaker.¹²

There is some breadth in the definition of a "dislocated worker" under WIA. For instance, eligibility is afforded in cases of anticipated facility closings and for self-employed workers. However, the core eligibility requirement is dislocation due to no-fault termination or facility closing. There is no eligibility requirement under WIA related to the cause of the dislocation. From the perspective of the individual, however, the classification of "adult" or "dislocated" will not make a difference in terms of services funded by formula grants and provided through One-Stop centers. That is, dislocated workers, like others accessing services through a One-Stop center, receive core services and may receive intensive services and training depending on the circumstances and characteristics of the individual.

Dislocated Worker Formula Funding Structure

Funding for the dislocated worker program in WIA consists of two parts: the National Reserve and state formula grants. From total funding appropriated for the Dislocated Workers Activities program in a fiscal year, Section 132(a)(2)(A) specifies that 20% is to be used for a National Reserve account, which provides for National Emergency Grants (NEG) and other services for dislocated workers.

Funding for the dislocated worker state formula grant program is allocated from ETA to states according to a three-factor formula based on each state's relative share of each factor. That is, a state's "relative share" of any formula factor is calculated by dividing the factor population (e.g., number of unemployed individuals) in the state by the factor population in the United States as a whole. After the allocations are made to states, within-state allocations are made based on formulas as well (see below for details).

After funds are allocated from ETA to individual states according to the formula, the governor of each state must reserve not more than 15% of the Dislocated Worker Activity state allocation for statewide "employment and training activities." In addition, of the state allocation for dislocated worker activities, the governor of each state must also reserve not more than 25% for rapid response activities. In sum, not more than 40% of dislocated worker state allocations are reserved at the state level for statewide activities.

From the 25% rapid response reservation, states are required to carry out rapid response activities to assist workers who have been dislocated in obtaining reemployment as quickly as possible. A dislocation event is typically defined as a permanent closure or mass layoffs at a facility or a disaster (natural or otherwise) resulting in mass job dislocation. The services funded under this reserve may include¹³

• establishment of onsite contact with employers and employee representatives immediately after the dislocation event,
• provision of information and access to employment and training programs,
• assistance in establishing a labor-management agreement to determine the employment and training needs of the affected workers,
• provision of emergency assistance, and
• provision of assistance to the affected local community to develop a coordinated response in seeking state economic development aid.

The remainder of the dislocated worker activities funding stream, following the state reservation for statewide and rapid response activities, must be suballocated to local areas based on a state-developed formula that takes into account the following information:

• insured unemployment data,
• unemployment concentrations,
• plant closing and mass layoff data,
• declining industries data,
• farmer-rancher economic hardship, and
• long-term unemployment.

Trends in State Formula Grant Funding

Data in Table 1 show the funding trends for the dislocated worker activities from PY2001 through PY2010. Excluding the National Reserve, which is not allocated to states by a formula, funding for the state formula grants has fluctuated between a high of $1.272 billion in PY2001 to a low of $1.150 billion in PY2003, with a PY2010 level of $1.184 billion.¹⁴ From PY2005 through PY2010, funding for the state formula grants has been relatively stable around an average level of approximately $1.182 billion.

Dislocated Worker Program Formula Funding

This section provides a brief discussion of key concepts used in allocating federal funds to states and localities by formula. Following that, the remainder of the section examines the specific design of the dislocated worker formula and some of the issues that have arisen about the performance of the formula.

Formula Funding—General Concepts

The WIA state formula grant program, like many other federal grant programs, provides funds to states via mechanisms that target areas with the greatest need or concentration of eligible individuals.¹⁵ In general, allocation formulas use some combination of population-based factors as a way to direct federal resources toward areas with large concentrations of the program's target group. That is, a factor represents a quantifiable measure of some population or characteristic of the population that reflects the program's target. Programs that allocate funding in this manner usually have a cap on total federal spending, thus the allocation factors determine each jurisdiction's share of the total amount available. The discussion below is not intended to be exhaustive but only to provide a sense of the range of components that make up a formula.

Many formula programs target aid to low-income individuals. While programs for low-income individuals often allocate funds in part on a measure of economic need (e.g., population with income at or below poverty guidelines), not all formula factors are need-based. Some factors are population-based (e.g., rural population), some are characteristic-based (e.g., number of low-income children without health insurance), and some are education-based (e.g., number of individuals ages 16 and over who have not completed high school). Further, some programs base allocations in whole or in part on historic spending patterns, which may reflect a wide variety of factors (e.g., TANF block grants are allocated based on historic spending under predecessor programs). If current population-based formulas were applied to programs using historic spending as a base, the distribution of resources might change significantly. The specific data sources used in formulas are also significant and may be specified in statute. If a formula relies on older data or data subject to large subsequent revisions, funding distributions might not reflect current need.

Moreover, there are various formula features that are often used to mitigate large changes in a particular jurisdiction's formula-based allotment from one year to the next. For example, formulas might include "hold-harmless" provisions to limit annual changes to a certain percentage of the previous year's allocation, "small-state" minimums to guarantee less populous states a minimum level of funding, and "floors" and "ceilings" to create some degree of predictability in annual allocations.

As will be discussed below, the WIA formulas are principally focused on distributing funds in accordance with varied measures of unemployment and states' relative share of unemployment. The WIA dislocated worker formula considers three measures of unemployment—total unemployment, long-term unemployment, and "excess" unemployment. The dislocated worker formula does not employ the use of formula features used in many federal formulas that serve as a hedge against volatility (e.g., ceilings and floors, hold harmless provisions).

Dislocated Worker Formula Factors

Funding for the DW program under WIA is distributed to states by a formula grant. The state grant formula for the DW program is set out in WIA Section 132(b)(2)(B)(ii) and ETA publishes guidance on formula descriptions and data factor definitions. The formula is comprised of the factors summarized in Table 2. Each factor is equally weighted, which means that approximately $400 million (one-third of the roughly $1.2 billion total state formula grant for dislocated workers) is allocated on the basis of each state's relative share of each factor in Table 2.

The DW formula in WIA uses factors similar to those originally developed in the Comprehensive Employment and Training Act (CETA) of 1973 and the Job Training Partnership Act (JTPA) of 1982. Specifically, Section 603(a)(2)(C) of the Emergency Jobs and Unemployment Assistance Act of 1974 (P.L. 93-567) required 50% of funding provided for transitional employment in "jobs providing needed public services" to be allocated among eligible entities on the basis of the relative unemployment compared to total unemployment in all other jurisdictions, and 25% to be allocated on the basis of the excess number of unemployed persons in the relevant jurisdiction compared to the total number of excess unemployed in all other jurisdictions. "Excess" was defined as unemployment above 4.5%. Additionally, JTPA added long-term unemployment as a formula factor in allocating funds for training and employment services.

Unlike the WIA Title I adult activities funding formula, the dislocated workers' formula does not feature small-state minimum provisions, hold-harmless provisions, or stop-gain provisions. The absence of these provisions means that allocations follow the changes in formula factors and are not governed by reference to prior year allocations.

The underlying unemployment data used in the dislocated worker formula are from the Bureau of Labor Statistics (BLS) Local Area Unemployment Statistics (LAUS) series. State labor force data are derived from statistical modeling that incorporates data from the Current Population Survey (CPS).¹⁶ Formula factors are operationalized as follows:

Total Unemployed

One-third of the allocation is based on the state share of the total number of unemployed individuals in the U.S. as a whole. The official concept of unemployment, as reported by the BLS, is used in the formula. That is, to be counted as unemployed, an individual must be without employment but available and looking for work. In the dislocated worker formula, the time period used for each program year has typically been the 12 months ending on September 30 of the year before the WIA program year. In the most recent program year—PY2010 (July 1, 2010–June 30, 2011)—the 12-month average (October 1, 2008–September 30, 2009) level of unemployment was calculated for each state, and was then converted into a relative share for each state. For example, in the state of New Jersey there were an average of 372,712 unemployed individuals each month in the 12-month period from October 1, 2008, through September 30, 2009. Thus, of the comparable U.S. total of 13,350,467 unemployed individuals in that period, New Jersey had a relative share of 2.79%.

Excess Unemployed

One-third of the allocation is based on the state share of the total number of excess unemployed individuals in states with excess unemployment. As with the measure of total unemployed, the excess unemployed factor starts with the official concept of unemployment. For the excess unemployed factor in the dislocated worker formula, the time period used for each program year has typically been the 12 months ending on September 30 of the year before the WIA program year. This factor is calculated by determining the number of unemployed individuals in excess of the number that would equal an unemployment rate of 4.5%. Then the excess number is converted into a relative share for each state. Again using the state of New Jersey as an example, there were an average of 168,689 individuals each month counted as "excess unemployed" in the 12-month period ending September 30, 2009. This was derived as follows. First, a 4.5% unemployment rate in New Jersey would have represented 204,003 individuals (4.5% of the monthly average number of individuals in New Jersey's civilian labor force for the 12-month reference period). Second, 204,003 was subtracted from the average monthly number of total unemployed in New Jersey in the same period (372,712), resulting in 168,709 excess unemployed individuals. Of the comparable excess unemployed total for all states, 6,343,868, New Jersey had a relative share of 2.66%.¹⁷ It is important to note that states with unemployment rates at or below 4.5% receive a relative share of zero for this factor. The implications of this structure are discussed in detail later in the report.

Long-Term Unemployed

One-third of the allocation is based on the state share of the number of long-term unemployed individuals in the U.S. as a whole. As with the measure of total unemployed, the long-term unemployed factor starts with the official concept of unemployment. For the long-term unemployed factor in the dislocated worker formula, the time period used for each program year has typically been the calendar year beginning two and one-half years prior to the WIA program year (e.g., for PY2008, the long-term unemployment in calendar year 2006 was used). In the most recent program year (PY2010), however, the monthly average for the 12-month period ending September 30 was used as the reference period.¹⁸ This factor is calculated by determining the number of individuals unemployed 15 weeks or more. The number is then converted into a relative share for each state. For example, in the state of New Jersey, there were an average of 186,300 individuals unemployed 15 weeks or more each month in the 12-month period ending September 30, 2009. Of the comparable long-term unemployed total in the U.S., 6,201,000, New Jersey had a relative share of 3.0%.

Examination of the Dislocated Worker Formula

As previously noted, there has been significant interest recently, primarily focused on two aspects of the dislocated worker formula—alignment with statutory intent and volatility. This section of the report examines each topic.

First, with regard to alignment emphasis is placed on examining whether the formula factors focus on and measure the need of the target population of dislocated workers. The dislocated worker formula is unlike the general WIA formula for adults because it specifically identifies a subset of the unemployed population. Therefore, to match this statutory specification, it is necessary that the factors used to measure the population are sufficiently narrow in scope.

Second, with regard to volatility emphasis is placed on examining whether the current formula generates types and amounts of volatility that may be undesirable. That is, some degree of volatility may be necessary and desirable if it corresponds well with fluctuation in need. On the other hand, if the volatility is not necessarily the result of funds following need, then the formula may not be functioning as intended. Additionally, as has been suggested, recently large year-to-year fluctuations in funding may undermine program operations.

How Closely Do the Formula Factors Align with the Population on the Basis of Which Funding is Intended to be Provided?

Alignment is a measure of how well a formula factor captures the target population and may be evaluated by its adherence to the WIA definition of a "dislocated worker." To fully align with the WIA definition of dislocated worker, a formula factor would have to be constructed on the basis of an individual's cause and circumstance of dislocation, occupation, and attachment to the labor force. That is, a perfectly aligned formula factor would indicate that the worker was in at least one of the following four classes:

• Terminated, attached to the labor force. This class is described in WIA Section 101(9)(A) and includes individuals who were terminated or laid off (but not through the fault of the individual worker), who are eligible for or have exhausted UI (or had sufficient labor force attachment but did not qualify for UI because of low earnings or work in an uncovered sector), and who are unlikely to return to their previous occupation or industry.
• Terminated as part of a substantial layoff or closure. This class is described in WIA Section 101(9)(B) and includes individuals who were terminated or laid off as part of a permanent closure or "substantial layoff" at a facility;
• Self-employed. This class is described in WIA Section 101(9)(C) and includes individuals who were self-employed but became unemployed because of general economic conditions or natural disasters.
• Displaced homemakers. This class is described in WIA Section 101(9)(D) and includes individuals who provided unpaid services to family members and were dependent on income from another family member but are no longer supported by family income and are unemployed or underemployed.

Understanding the relative alignment of the dislocated worker formula with the population specified in WIA as "dislocated" requires understanding of the source of data for the formula factors. All three factors in the dislocated worker formula—total unemployed, excess unemployed, and long-term unemployed—are drawn from the same data source: the Current Population Survey (CPS), which is a monthly household survey administered by the U.S. Census Bureau. The three factors in the dislocated worker state formula grant are all based on the concept of unemployed individuals as measured in the CPS. To be counted as "unemployed," an individual must have no employment during the reference week, must be available for work during the reference week, and must have actively searched for work within the previous four-week period.¹⁹ In addition, the CPS provides data on the occupation and industry of individuals.

As will be discussed below, the CPS data on unemployed individuals are not perfectly matched with the WIA definitions of dislocated workers. Data limitations pertaining to each class of worker included in the WIA dislocated worker definition are as follows:

• Terminated, attached to the labor force. The CPS data used in the current dislocated worker formula provide the total number of unemployed but do not provide reason for termination.²⁰ In addition, the unemployment count used in the formula is in no way dependent upon labor force attachment (e.g., it does not consider eligibility for or receipt of unemployment insurance benefits). The derivation of the excess and long-term unemployment factors likewise do not specify the reason for unemployment or attachment to the labor market. As a result, the data from the CPS over-represent the number of dislocated workers because they include unemployed individuals who have lost their job for any reason (not just no-fault termination), who may or may not have sufficient attachment to the labor force, and who may or may not be likely to return to their previous industry or occupation.
• Substantial layoff or closure. As with the first class of worker (terminated, attached to the labor force), the CPS data over-represent the number of dislocated workers because there is not information about the nature of the termination (i.e., whether it was part of a larger layoff or closure).
• Self-employed. The CPS data include self-employed workers in the count of unemployed individuals. The data do not, however, provide a separate count of self-employed workers and do not contain information on the cause of unemployment.
• Displaced homemakers. The CPS data include this class of worker but not as a separately identified group. In other words, there are most likely unemployed workers in the CPS data who meet the criteria of a displaced homemaker but it is not possible to determine how many of these individuals meet these criteria.

In sum, while the CPS data provide accurate and timely information on the U.S. labor force, the CPS is not designed to capture the WIA-defined population of "dislocated" workers. As currently used in the dislocated worker formula, the CPS data do not measure attachment to the labor market, do not classify workers as "unlikely to return the their previous occupation or industry," do not provide information on how wage and salary workers lost their jobs, do not identify self-employed individuals who are unemployed because of local economic conditions, and do not separately identify displaced homemakers. Thus the measures used for allocating funds in the current dislocated worker formula are likely to include but do not focus on the population of dislocated workers targeted by WIA. Another way of stating this is that the current factors in the formula are based on unemployed workers in general (with some weight given to long-term unemployed) but not specifically on dislocated workers.

How Volatile Are Annual State Funding Allocations?

State allocations can shift significantly from year to year. One way to summarize the volatility of allocations is to examine volatility data by "state-years." That is, a state-year represents the allocation in one state for a given year. Comparing one state's allocation to the next year's allocation provides a measure of total volatility. For example, from PY2009 to PY2010 Mississippi's dislocated worker funding allocation decreased 27.4%, which represents one state-year. For the entire period that WIA has been in existence, there are 468 state-years (i.e., 50 states, DC, and Puerto Rico over nine years) between PY2002 and PY2010. Thus, of the 468 state-years, annual allocations increased or decreased by at least 30% from the previous year in 109, or 23%, of the state-years.²¹ The number of states with annual allocation changes greater than 30% ranged from two (PY2004 to PY2005) to 25 (PY2008 to PY2009), with an average of 12 states per year experiencing swings of greater than 30% from the previous year. During certain periods, volatility has been particularly acute. For example, from PY2008 to PY2009 13 states experienced increases of at least 30% over prior year allocations and 12 experienced decreases of at least 30% over prior year allocations.²²

The dislocated worker formula is designed to allow for some volatility—as evidenced by the lack of "hold harmless" or "stop gain" provisions—but "excessive" volatility may make it difficult for states to plan employment and training activities from year to year and often results in a disconnect between allocations and state changes in unemployment.

Volatility in a total state allotment in itself is neither positive or negative in terms of formula performance. In fact, the dislocated worker formula is intended to contain some volatility as it is supposed to track a population (i.e., dislocated workers) that itself may be volatile, especially depending on how movements in the "dislocated" population are measured. The magnitude of year-to-year total funding swings in states like Michigan that have experienced sustained high unemployment (see next section) have raised questions about the effects on sustaining well-functioning programs.

The Construction of the "Excess Unemployment" Factor Creates Volatility

Though year-to-year allocations can be quite volatile, the volatility is driven primarily by one of the three formula factors: excess unemployment. In other words, in a three-factor formula it is the one factor of "excess unemployment" that determines most of the variation in year-to-year funding.²³ It is also the case that any formula based on the concept of "relative share" of factor levels (i.e., percentage of national totals) may result in allocation changes that do not match the changes in factor values (i.e., the number of individuals). The use of a threshold, however, exacerbates this feature of relative share formulas. Finally, because one of the three factors in the dislocated worker formula—excess unemployment—is the primary driver of volatility, the formula is also marked by sizeable changes in per capita (or per factor) allocations from year to year.

Excess Unemployment and the Problem of a Threshold

By including a measure of excess unemployment, the dislocated worker formula is designed in part to target areas with relatively worse employment problems. This factor, however, is not a direct measure of economic dislocation that might occur due to a plant closing or mass layoff in a local area. In addition, the choice of 4.5% as the threshold between normal and excess unemployment is arbitrary in that it does not take into account a state's performance relative to the national unemployment rate or the state's recent unemployment rate.²⁴

The logic of creating a factor of excess unemployment is to allocate additional funds to states with the highest unemployment, but the threshold created by this measure creates volatility in the funding allocations to individual states. Volatility in an individual state's allocation is driven by whether the state crosses the 4.5% threshold from year to year and the degree to which the number of unemployed exceeds the threshold. Unlike the Adult and Youth program funding formulas, the dislocated worker allotment formula does not contain "hold harmless" provisions that specify a minimum or maximum allotment that a state must receive each program year. Because the formula is based on a state's relative share of unemployment and because there is no hold harmless provision, allotments may shift substantially between states, creating gains for some and losses for others.

The formula does not contain features that affect the calculation, other than the weights themselves, but the statutory definition of "excess unemployment" effectively serves as a "threshold" feature that is not present in the other two factors. That is, the cutoff of 4.5% means that states with a statewide unemployment rate below 4.5% (the statutorily defined rate) will not receive an allotment based on this factor (i.e., their share from the excess unemployment is 0 x 1/3 = 0). The remaining states—those with unemployment in excess of 4.5%—then split the one-third of the dislocated worker funding that is allocated through the excess unemployment factor. For example, in PY2008 23 states had unemployment in excess of 4.5% and thus split one-third of the dislocated worker funding ($394.6 million, which is one-third of the state formula allotment of $1.18 billion), based on each of the 23 states' relative share of total excess unemployment. In PY2009, 37 states had unemployment in excess of 4.5% and split one-third of the dislocated worker funding ($394.6 million, which is one-third of the state formula allotment of $1.18 billion). Thus, 14 more states shared the same pot of funding in PY2009 than had shared the same amount in PY2008.²⁵

The effect of this threshold feature of the excess unemployment factor on individual states can be illustrated by looking at the experience of two states in PY2008 and PY2009:

• In PY2008, Michigan was one of 23 states with unemployment in excess of 4.5%. Michigan had a total of 128,258 individuals in excess unemployment and its relative share of excess unemployment among the 23 states was 22%. Of Michigan's total dislocated worker funding of $130.8 million in PY2008, $87.4 million was from the excess unemployment factor in the funding formula. In PY2009, Michigan was one of 37 states with unemployment in excess of 4.5%. Michigan had a total of 194,649 individuals in excess unemployment in PY2009 and its relative share of excess unemployment among the 37 states was 8.9%. In PY2009, Michigan's total dislocated worker funding dropped to $75.1 million. Of this $55.8 million decline in funding ($130.8 million - $75.1 million), $52.0 million was a result of Michigan's changing relative share of excess unemployment from PY2008 to PY2009. In sum, despite an increase in excess unemployment of just over 66,000 (52%), Michigan lost $55.8 million (43%) of its dislocated funding.
• In PY2008, Florida was one of 29 states with no excess unemployment. Thus, of Florida's total dislocated worker funding of $31.4 million in PY2008, $0 was from the excess unemployment factor in the funding formula. In PY2009, Florida was one of 37 states with unemployment in excess of 4.5%. Florida had a total of 141,829 individuals in excess unemployment in PY2009 and its relative share of excess unemployment among the 37 states was 6.5%. In PY2009, Florida's total dislocated worker funding increased to $77.1 million. Of this $45.7 million increase in funding ($77.1 million - $31.4 million), $25.8 million was a result of Florida's changing relative share of excess unemployment from PY2008 to PY2009.

To illustrate the threshold effect more generally, Figure 1 shows the relationship in all states between changes in the value of excess unemployed and changes in allocations from PY2009 to PY2010. Specifically, Figure 1 shows the amount by which the total number of excess unemployed (right vertical axis) changed from PY2009 to PY2010 and the amount by which the one-third of the total allocation attributable to that factor changed (left vertical axis). Because the overall allocation for the dislocated worker formula grant was the same in PY2009 and PY2010 ($1.184 billion), the analysis shows exactly how much the excess unemployed factor contributed to each state's overall funding change. That is, each factor is weighted one-third and thus contributes one-third of the total $1.184 billion, or approximately $395 million.

Figure 1. Changes in Excess Unemployed and the Allocation from the Excess Unemployed Factor, PY2009–PY 2010

Source: CRS analysis of Department of Labor (DOL) data.

Data in Figure 1 illustrate how the use of a fixed threshold of a 4.5% unemployment rate to define excess unemployment does not always result in allocations that match the change in actual economic conditions in the states. While the excess unemployment factor is based on levels (not change in levels) and thus reflects any given state's share of all excess unemployed in a given allocation year, it often leads to sharply reduced levels of funding in some states when additional states trigger on to the dislocated worker excess unemployment factor in the formula.²⁶ Specifically:

• The formula often results in divergent allocations among states experiencing similar increases in levels of excess unemployed. For example, from PY2009 to PY2010, Texas, New York, and Michigan saw similar increases in the number of excess unemployed (240,000, 227,000, and 211,000, respectively). Yet the changes in allocations from the dislocated worker formula ranged from $10.1 million more to Texas and $3.6 million more to New York to $10.1 million less to Michigan. Similarly, Kansas and Mississippi each had approximately 25,000 additional excess unemployed from PY2009 to PY2010, but Kansas received an additional $1.4 million in funds while Mississippi received $2.4 million less in funds compared to PY2009 allocations.
• The formula results in transferring funds away from some states (in some cases, states with persistently high unemployment) in times of generally rising unemployment. In Figure 1, 12 states were "new" in PY2010.²⁷ That is, these 12 states received a $0 allocation from the excess unemployed factor in PY2009 but triggered on in PY2010 and received an allocation. Of course, because 12 additional states were "on" for allocation purposes in PY2010 and because there was no additional funding for the dislocated worker grant formula, by necessity funds for the "new" states had to be "paid for" by reallocating from states currently receiving an allocation from the excess unemployment factor.²⁸

Relative Share and Changes in Factor Values

The sort of mismatches between changes in economic conditions and changes in allocations shown in Figure 1 are exacerbated by the threshold in the excess unemployment factor but may also exist, albeit to a lesser degree, in the absence of a such a feature. In a program such as the WIA dislocated worker program, which is characterized by nearly flat annual funding and a formula based on each state's relative share of formula factors, normal changes in the values of the formula factors may cause some shifting of funds away from states with increased need. For example, combining the data in Figure 1 with funding and factor data for the state of California illustrates the performance of the three formula factors:

• From PY2009 to PY2010, excess unemployment increased by 649,431 in California but its allocation from this factor decreased by $19.7 million (Figure 1 shows this portion of the total allocation).
• From PY2009 to PY2010, total unemployment increased by 653,128 in California and its allocation from this factor increased by $779,376.
• From PY2009 to PY2010, long-term unemployment increased by 460,300 in California but its allocation from this factor decreased by $930,710.
• In PY2009, California's total allocation from the dislocated worker formula grant was $212.3 million and its total allocation in PY2010 was $192.4 million—a loss of $19.9 million. Because overall funding for the dislocated worker formula grant was the same in both years, the change in allocation was fully attributable to changes in California's relative share of each factor.

Taken together, California's decrease in dislocated worker funding is almost entirely a function of the excess unemployment factor. That is, despite increases in the absolute level of unemployment for all three formula factors, California's allocation from total unemployment went up slightly and its allocation from long-term unemployment went down slightly; however, these small changes were completely offset by its large allocation drop from excess unemployment. The same phenomenon occurred for allocations from the excess unemployed and the long-term unemployed factors—increases in the number of unemployed but decreases in funding for these factors. The change in funding from the long-term unemployed factor, however, was considerably less than the change from excess unemployed because there is no threshold for long-term unemployed. That is, all states are always "on" the long-term unemployed factor (i.e., there is always some level of long-term unemployed) and thus the fluctuations are not as sharp as they are for excess unemployed.

Excess Unemployment and Funding Intensity

Another way of understanding the distributional consequences of the volatility generated by the current formula for dislocated workers is to look at funding intensity or per capita funding. That is, volatility matters not just in terms of total allocations but also in allocations per capita, which may generate a different kind of inequity from year to year. From an operational standpoint, large changes in resource intensity from year to year may make it difficult to provide consistency and stability in services.

Figure 2 provides data on the funding from each factor (excess, total, and long-term unemployed) divided by the target population. For example, in PY2009 the total state formula allocation for dislocated workers ($1.2 billion) was distributed equally by the three formula factors, or approximately $395 million per factor. Divided by the relevant population for each factor, this yielded funding of $182 for each excess unemployed individual, $43 for each unemployed individual, and $123 for each long-term unemployed individual. The per capita figure measures the intensity of funding—resources per intended recipient of services.

Figure 2. Factor Funding per Capita in the Dislocated Worker Formula

Source: CRS analysis of DOL data.

The data in Figure 2 show three aspects of the distributional effects of the dislocated formula:

• The overall intensity of funding declines as the number of targeted individuals increases. The per capita funding for all three formula factors has ranged from $1,703 in PY2002 to $155 in PY2010. This change simply reflects the changes in the targeted populations as unemployment rises and falls.
• The allocation per capita from the excess unemployed factor has a much larger range in resource intensity than the allocation per capita from the total and long-term unemployed factors. Whereas the funding per excess unemployed person has ranged from $1,351 (PY2002) to $62 (PY2010), the funding for the total and long-term unemployed factors has ranged from $312 to $64 and from $67 to $30, respectively. Even removing PY2002 from the analysis, the ratio of highest per capita funding to lowest per capita funding (e.g., $681 in PY2008 divided by $62 in PY2010) is far higher for excess unemployed (11) than for total unemployed (1.9) or long-term unemployed (3.4).
• The annual changes in funding intensity are almost entirely generated by one factor—excess unemployed. For example, from PY2008 to PY2009, which was a period of rapidly rising unemployment, the total per capita allocation in the overall dislocated worker formula dropped from $909 to $348, or a total of $562. Of this total, $500 was due to the drop in per capita funding for excess unemployed individuals. This is primarily due to the threshold effect of the excess unemployed factor. Factors without "cliffs" (i.e., total and long-term unemployed) do not have large and abrupt changes in the populations for which funding is provided.

Examination of Alternative Formula Factors

Before reviewing alternatives to the current formula, the examination thus far has demonstrated the following:

• The current dislocated worker formula factors do not align well with the general concept and the WIA-specific definition of dislocated workers. The current factors are measures of different facets of unemployment but do not measure the target population.
• The current formula leads to highly volatile, and in many cases countercyclical, state annual allocations. The volatility alone may make it difficult to plan and operate programs and services. In addition, an increase in a factor value (e.g., an increase in the number of excess unemployed) is often accompanied by a decrease in funding for that factor.
• The current formula results in significant variation in the intensity of funding, or the allocation per capita (per capita in the target population). Large annual fluctuations in the allocation per capita (see Figure 2) means significant inequity across years in the resources available per targeted population.
• The excess unemployed factor in the formula generates much of the volatility in the formula. The construction of this factor, which uses a threshold over which states are either on or off of the allocation, exacerbates any natural trends that shift funding among states from year to year.

Given that the dislocated worker formula is likely to receive attention in upcoming WIA reauthorization deliberations, this section of the report considers alternative formula factors and features, which may have the potential of enhancing alignment and reducing volatility. While another approach to minimizing the effects of volatility might be to increase appropriations to create a constant level of funding during changes in target populations, in the current fiscal environment it is not clear that this is an approach likely to be taken.

No existing data source aligns perfectly with the WIA definition of a dislocated worker. Based on the review of the dislocated worker formula above, however, the existing factors are a closer measure of the unemployed population than the population of dislocated workers. This mismatch between the formula factors and the underlying population is important because the WIA state formula grant program makes a distinction between unemployed populations by having two separate funding streams for adults—the adult allocation and the dislocated worker allocation. The adult allocation is based on three factors, two of which—unemployment in areas of substantial unemployment and excess unemployment—are variations of the same core measure used in the dislocated worker formula: total unemployed. There appears to be a mismatch between the intent of the formula construction, as expressed in creating different funding streams for adult and dislocated workers, and the practice of using similar or overlapping measures for the two state grant programs.

Because there have been concerns about this mismatch, this section reviews several alternative factors for possible use in the WIA dislocated worker formula. The factors are evaluated on their alignment with the target population of dislocated workers and the attendant volatility of these factors. Following evaluation of the formula factors, there is an analysis of the ways in which varied formula features might be adjusted to improve the alignment between intent of the formula and its construction.

Alignment of Alternative Formula Factors

Alignment is a measure of how well a formula factor captures the target population. To align with the WIA definition of dislocated worker, a formula factor would have to provide information on an individual's cause and circumstance of dislocation, occupation, and attachment to the labor force. That is, a perfectly aligned formula factor would indicate that the worker is in one of the four classes discussed in the section above on the current formula. To reiterate, WIA provides four sets of characteristics to define a dislocated worker:

• terminated or laid off (or has been notified of a termination or layoff), sufficiently attached to the workforce, and unlikely to return to the previous industry or occupation;
• terminated or laid off (or has been notified of a termination or layoff) as a result of any permanent closure of (or substantial layoff at) a facility or employed at a facility that is scheduled to close within 180 days;
• self-employed but unemployed because of general economic conditions in the individual's community or because of natural disasters; or
• displaced homemaker.

Table 3 provides information on the current WIA dislocated worker formula factors and six possible alternative factors.²⁹ In addition to providing a definition for each factor, Table 3 provides a depiction of the characteristics captured by current and possible alternative formula factors. The four categories of dislocated workers, as defined by WIA, are listed in columns 3 through 6 of the table. In each of the table's rows, there is an indication of the alignment between the current or alternative factor and the characteristics of each category of dislocated worker. This indication is based on an analysis of the underlying data source from which each factor is drawn. For example, individuals in the "terminated, attached to the labor force" category are among those included in the measure of "total unemployment," but they are not identifiable as a distinct subset. On the other hand, individuals who are "self-employed" are not by definition included in the "mass layoffs" measure. The sections immediately following Table 3 provide a broader summary discussion of each alternative factor's relationship to the measurement of dislocated workers as defined by WIA.³⁰

Civilian Labor Force

All of the alternative factors discussed in this report are targeted more toward the dislocated worker population than total unemployment, with the exception of the size of the civilian labor force (CLF). The CLF, which is calculated based on the CPS, is comprised of all employed and unemployed individuals ages 16 and older who are not confined to an institution (e.g., prison) and are not on active duty in the U.S. military. Thus, it is the broadest possible measure of the labor market. As such, it is not aligned with the dislocated worker population but it is a stable measure of each state's share of the workforce. If used as one of multiple factors in a dislocated formula, the state share of the CLF could serve as a base of stability in allocations in contrast to other factors that are more targeted and that may vary widely from year to year.

Mass Layoffs (ML)

The Mass Layoff Statistics (ML) program is administered by BLS and tracks layoff events that result in separations from employment.

• To be counted in the ML data, a layoff event must occur in which at least 50 initial claims for unemployment insurance (UI) are filed against an establishment during a consecutive five-week period.
• The ML data include the total number of mass layoff events and the total number of initial claimants for UI benefits (i.e., person initiating a request either for determination of entitlement or eligibility for compensation).
• State data include DC and Puerto Rico and are available for all industries.
• The ML data series is published monthly and has a lag of about a month between the reference month and the availability of the data (e.g., data for September 2010 were released on October 22, 2010).³¹

Overall the ML data series appears to be aligned more with the dislocated worker population than a broader measure of unemployment. For each class of dislocated worker defined by WIA, the ML data capture the following:

• Terminated, attached to the labor force. The ML data cover this class of dislocated workers, with some exceptions. The ML data capture workers who are terminated from employment through no fault of their own but do not specify the degree of labor force attachment.
• Substantial layoff or closure. As with the first class of worker (terminated, attached to the labor force), the ML data capture the target population. For this class of worker in particular, the ML data are well aligned because there is no reference in WIA to these workers having sufficient attachment to the labor force. In fact, the ML data include workers who are and are not sufficiently attached to the labor force to qualify for UI benefits.
• Self-employed. The ML data by definition do not include self-employed workers.
• Displaced homemakers. The ML data by definition do not include displaced homemakers.

Thus, the number of initial claimants in the ML series captures dislocations that occur as part of a "substantial layoff" event and involve individuals "sufficiently attached" the labor force (i.e., as part of a mass layoff, workers filing initial claims meet the eligibility requirement of being unemployed through no fault of their own). The ML data do not capture layoff events involving fewer than 50 individuals and thus would exclude smaller dislocation events. In addition, while the ML data series provides the number of initial claimants filing for UI following a mass layoff, it does not yield an actual account of those eligible for compensation (i.e., it is not an exact match for "sufficiently attached" to the labor market as defined in WIA).

As a practical consideration, the relative share (i.e., each state's share of the national total) of a factor such as mass layoffs may not be well aligned with the "need" generated by an ML event. The actual impact of a large dislocation event may be measured better by the size of the dislocation as a proportion of each state's civilian labor force. For example, in 2009 there were 5,513 initial claims from mass layoff events in Delaware and 5,206 in West Virginia, which represented 0.2% and 0.18%, respectively, of all ML initial claims in the United States. As a share of each states' civilian labor force, however, the number of initial claims from mass layoff events was 1.3% in Delaware and 0.65% in West Virginia in 2009. In terms of the dislocation effect, therefore, a similar number of layoffs affected a much larger share of the workforce in Delaware than in West Virginia.

Extended Mass Layoffs (EML)

The Extended Mass Layoff Statistics (EML) program is administered by BLS and tracks layoff events that result in separations from employment.

• To be counted in the ML data, a layoff event must occur in which at least 50 initial claims for unemployment insurance (UI) are filed against an establishment during a consecutive five-week period and at least 50 workers have been separated from jobs for more than 30 days.
• The EML data include the total number of mass layoff events, the total number of separations (i.e., number of workers displaced during an EML event regardless of whether they file for UI), and the total number of initial claimants for UI (i.e., person initiating a request either for determination of entitlement or eligibility for compensation).
• State data include DC and Puerto Rico and are available for private, nonfarm industries.
• The EML data series is published quarterly, with approximately a month lag between the reference month and the availability of the data (e.g., data for the second quarter of 2010 were released on August 11, 2010).³²

Overall the EML data series appears to be more aligned with the dislocated worker population than a broader measure of unemployment. For each class of dislocated worker defined by WIA, the EML data capture the following:

• Terminated, attached to the labor force. The EML data cover this class of dislocated workers. The EML data capture workers who are terminated from employment through no fault of their own but do not specify the degree of labor force attachment.
• Substantial layoff or closure. The EML data capture the target population. For this class of worker in particular, the EML data are well aligned because there is no reference in WIA to these workers having sufficient attachment to the labor force. The EML data include both "initial claimants" and "total separations," which means that all individuals terminated through a mass layoff or closure are counted, not only those who request determination of eligibility for UI.
• Self-employed. The EML data by definition do not include self-employed workers.
• Displaced homemakers. The EML data by definition do not include displaced homemakers.

Thus, the number of separations and initial claimants in the EML series captures dislocations that occur as part of a "substantial layoff" event and involve individuals "sufficiently attached" to the labor force (i.e., as part of a mass layoff, workers filing initial claims meet the eligibility requirement of being unemployed through no fault of their own). Like the ML series, the EML data do not capture layoff events involving fewer than 50 individuals and thus would exclude smaller dislocation events. On the other hand, the EML data series provides the total number of separations, in addition to the total number of initial claimants filing for UI, which means that it is perhaps the closest factor for the second class of dislocated worker in WIA.

Unemployment Insurance—First Payments and Exhaustees

Data on the Unemployment Insurance (UI) program are collected and distributed by the Employment and Training Administration (ETA) of DOL. Data include the number of initial claims, first payments, weeks claimed, weeks compensated, and exhaustions.

• An "initial claim" occurs when an individual files a request to determine entitlement to or eligibility for UI or when an individual files a request to begin a subsequent period of eligibility within a benefit year or period of eligibility.
• A "first payment" refers to the initial payment made in a benefit year for a specific program and may be considered a proxy for program beneficiaries.
• An "exhaustion" occurs when an individual draws the final payment of their original entitlement for a given program.
• State data include DC and Puerto Rico.
• The UI data series is published weekly.³³

Overall the UI data appear to be better aligned with the concept of dislocation than a broader measure of unemployment. For each class of dislocated worker defined by WIA, the UI data capture the following:

• Terminated, attached to the labor force. An individual receiving UI benefits is by definition terminated (i.e., without employment through no fault of their own) and sufficiently attached to the labor force. The number of initial claims is therefore very well aligned with the definition of dislocated worker. The number of first payments and the number of exhaustees are even better aligned with the concept of dislocation because they reflect actual evidence of labor force attachment. In the case of exhaustees, in addition to being terminated from employment and being attached to the labor force, individuals in this category also may be less likely to return to a previous occupation or industry. The UI exhaustion data do not capture the likelihood of return to a previous occupation or industry but serve as a proxy for longer-term dislocation.
• Substantial layoff or closure. As with the first class of worker (terminated, attached to the labor force), the UI data capture the target population. For this class of worker, the UI data would pick up individuals whose employment was terminated but would exclude workers who are not sufficiently attached to the labor force to qualify for UI benefits.
• Self-employed. Self-employed workers are not typically eligible for UI benefits.
• Displaced homemakers. The UI data do not include displaced homemakers.

The number of initial UI claimants captures dislocations that involve individuals who are terminated from employment and who may be "sufficiently attached" to the labor force. In other words, some percentage of initial claims will be determined to be ineligible because of insufficient labor force attachment, voluntary quits, inadequate earnings, or similar reasons. The UI first payment and exhaustion data, on the other hand, have the advantage of capturing a population of individuals who had clear attachment to the labor force, lost employment through no fault of their own, and have been unable to return to work before receiving or exhausting UI benefits.

Job Losers and Persons Completing Temporary Jobs

As part of the Local Area Unemployment Statistics (LAUS) program, BLS provides the official unemployment rate and five other measures of labor underutilization. One of these measures, U-2, is the number of job losers and persons who completed temporary jobs. A "job loser" is an individual who becomes unemployed through job loss rather than through a voluntary quit or a recent entrance into the labor market. State U-2 data include DC but not Puerto Rico and the U-2 data series is published quarterly.³⁴

Overall the U-2 data are better aligned with the concept of dislocation than a broader measure of unemployment but not as targeted as other measures. For each class of dislocated worker defined by WIA, the U-2 data capture the following:

• Terminated, attached to the labor force. A job loser or a person completing a temporary job is by definition a terminated worker (i.e., without employment through no fault of their own). However, the U-2 data do not provide information on the degree of attachment to the labor force.
• Substantial layoff or closure. The U-2 data do not indicate the detailed circumstances of job loss. A person who lost a job as part of a substantial layoff or firm closure would be counted as a job loser in the U-2 data, but a job loser who was not a part of such a layoff or closure would also be counted in the U-2 data.
• Self-employed. A job loser could have been self-employed but this would not be identified in the U-2 data.
• Displaced homemakers. The U-2 data do not include displaced homemakers.

The U-2 measure of unemployment is more targeted than total unemployment in that it includes only individuals who have lost employment through no fault of their own. It does not provide information on the degree of labor force attachment, however.

Volatility of Alternative Factors

In addition to alignment, volatility is an important element to consider when examining the performance of a formula. Volatility may be expressed by measuring changes from allocation period to allocation period for a single state or changes in the distribution among states for a single period. In discussions of allocation formulas, the term volatility is typically used to describe the annual changes in a state's funding. There will always be some volatility in annual formula grant allocations, as the intent of many allocation formulas, particularly the dislocated worker formula, is to direct funding to states and localities in which there are the greatest needs at a given point in time. As discussed previously, however, the factors in the dislocated worker formula have generated fairly volatile shifts in year-to-year allocations. It has also been suggested that fluctuations in funding shift in a manner that is not necessarily well aligned with shifts in economic conditions. In addition, the use of the "threshold" formula feature for the excess unemployment factor drives much of the volatility in the dislocated worker allocations and creates "unnecessary" (i.e., above the level of natural fluctuations in the underlying data) volatility.

Volatility is an issue of concern if the allocations do not match the conditions in the location to which funds are distributed. Additionally, high levels of volatility can make it extremely difficult for state and substate entities to budget, plan, and operate programs. In practical terms, it is very difficult to deal with large swings in annual funding levels from a major funding source, particularly large swings in per capita funding. As discussed previously, it is the threshold feature (i.e., 4.5% unemployment) of the excess unemployed factor that generates most of the sizeable swings in per capita funding (see Figure 2). Therefore, this section covers the volatility of alternative formula factors without mechanisms such as a threshold. The addition of any formula features that create a "cliff" or "on-off" effect would be expected to exacerbate any underlying volatility in the data itself.

In general, broader measures of labor market activity (e.g., the size of the civilian labor force tends to be relatively stable and predictable because the demographic characteristics of the population tend to change slowly and be predictable) are relatively stable from year to year, while more targeted measures (e.g., mass layoffs) are more episodic in nature and thus vary a great deal from year to year. Factors that use features, such as a threshold, may exacerbate the inherent volatility of a given measure. This section of the report examines whether alternative formula factors may have the potential of reducing volatility.

One way to measure the contribution of each factor to overall volatility is to compare each state's relative share of each factor over time. A presentation of each state's relative share of each factor for each year would require 4,212 cells of data (9 factors x 52 entities x 9 program years), which would make volatility comparison difficult. Instead, there is a useful summary statistic—the coefficient of variation—to compare the volatility of alternative formula factors for the dislocated worker formula.

The coefficient of variation, which is the result of dividing the standard deviation (i.e., the typical variation from the average) by the mean, provides a way to compare the dispersion or variation of factor values with widely varying means.³⁵

In this approach, the resulting coefficient for each state is a summary of volatility for all years for each factor. A higher coefficient of variation indicates more average dispersion in that state. More importantly, the range of coefficients of variation across states indicates how much volatility is going to be associated with the particular factor. In the case of the excess unemployment factor, for example, coefficient values range from zero to three. On the other hand, coefficient values for the civilian labor force factor range from 0.005 to 0.05. Because state relative share determines state allocations, factors with a wider range of values (e.g., excess unemployment) will yield much greater funding shifts (i.e., changes in relative share) than factors with a smaller range of values (e.g., civilian labor force).

There are at least three broad conclusions that may be drawn from the analysis of volatility in this section (The full results of the analysis are in Appendix B).

• First, the current excess unemployment factor is by far the most volatile of any of the current dislocated worker formula factors. Of all nine factors, three—"excess unemployment," "mass layoffs," and "extended mass layoffs"—show the greatest volatility. In terms of individual states, for the average of the PY2002 through PY2010 period, in 44 of the 52 states, the coefficient of variation for the excess unemployment factor is the highest of all factor coefficients.
• Second, the underlying data for measuring unemployment broadly are not nearly as volatile as the constructed factor of excess unemployment. The threshold feature of the excess unemployment factor generates considerable variation, even on par with the episodic measures, mass and extended mass layoffs. For example, the smallest coefficient of variation for excess unemployment (California) is above the highest coefficient for regular unemployment (Nevada).³⁶ Another way of stating this is that only the mass layoffs and extended mass layoffs tend to have the highest levels of "naturally" occurring volatility, which is intuitive given that they are generated from episodic, irregular events. But even the minimum coefficient of variation values for mass layoffs and extended mass layoffs are well below the minimum value for excess unemployment.
• Third, the coefficient of variation data support the general proposition that broader measures of labor market activity (e.g., civilian labor force, unemployment insurance claims) have much lower volatility than more narrow measures (e.g., mass layoffs, excess unemployment).

The analysis of coefficients of variation demonstrates the effects that each factor exerts on the real driver in changes in annual allocations for any given state in a given year—each state's relative share of the factor or factors in the formula. The ranges in coefficients of variation for each factor indicate how much average volatility there is for that factor. For example, coefficient of variation values for excess unemployment range from 0.2325 to 3.0, which is far higher than the range for total unemployment (0.051–0.228) or for long-term unemployment (0.066–0.471). The excess unemployment factor results in more allocation volatility than any other factor, regardless of which individual state has the highest coefficient of variation, because of the comparatively large fluctuations in relative share it generates.

The remainder of this section summarizes the volatility analysis for each of the alternative factors, followed by a brief comparison to the volatility of the current formula factors.

Civilian Labor Force

The measure of the civilian labor force is the most stable of all the factors considered in this report, with a coefficient of variation range from 0.005 to 0.054. As this is the broadest measure of the labor force, it does not fluctuate a great deal from year to year. Even when states experience large changes in unemployment, the civilian labor force will not necessarily change much since it includes the employed and unemployed. The stability of CLF as a formula factor comes with the tradeoff that it is not targeted at all on dislocated workers. Rather, inclusion of the CLF in a formula factor would provide some measure of stability across states to go along with the other more volatile, but more targeted, factors.

Mass Layoffs and Extended Mass Layoffs

The occurrence of mass layoffs and extended mass layoffs makes these factors volatile by definition. That is, mass layoff and extended mass layoff events are episodic and may not follow a predictable trend based on other indicators of economic activity.³⁷ Indeed, the coefficients of variation for mass layoffs and extended mass layoffs show that these factors are somewhat close to excess unemployment in volatility.³⁸ The mass layoff coefficients range from 0.055 to 1.153 and the extended mass layoffs coefficients range from 0.136 to 1.199. In addition, the range—the difference between the lowest and highest value in the data series—is 1.097 and 1.063, respectively, for mass and extended mass layoffs. These ranges are higher than the range for all other factors with the exception of excess unemployment. This volatility, however, may serve to align the dislocated formula with the target population, as the nature of dislocation varies greatly by geography and time. While fluctuations in this factor itself (without additional volatility created from the features of the formula) may lead to sizeable reallocations from year to year, the volatility would be a result of actual layoff occurrences rather than a function of a formula feature (as in the excess unemployment factor).

Unemployment Insurance—First Payments and Exhaustions

The two measures of UI are used in this analysis are on the lower end of volatility as measured by the coefficient of variation. First payments range from 0.019 to 0.797 and exhaustions range from 0.040 to 0.677. The UI factors are relatively stable because they do not depend on a threshold and tend to track overall movements in total unemployment. In addition, states always have some level of UI first payments and exhaustions so that states will maintain some relative share of the UI payments and exhaustions every year. In addition, measures of UI, whether first payments or exhaustions, would align better with parts of the WIA-defined dislocated worker population as these data capture "attachment to the labor force."

U-2—Job Losers

This alternative measure of labor underutilization from BLS is not highly volatile, with a coefficient of variation ranging from 0.035 to 0.285. This range is comparable to the range for total unemployment but the U-2 series is considerably more targeted than the number of total unemployed by excluding those who voluntarily quit. In addition, the job loser measure may be better aligned with the concept of a dislocated worker because the job loss could represent an occupation to which one is unlikely to return. One drawback to this data series is that it does not include Puerto Rico.

Current Factors

Compared to the alternative formula factors, two of the three current factors are relatively stable and one—excess unemployment—is the most volatile of all factors.

Total Unemployment

The measure of total unemployment is relatively stable, with a coefficient of variation range from 0.05 to 0.23. As this is a broad measure of the labor force, it tends not to move in sudden, large swings. Some states will of course experience large changes in unemployment from year to year but no state will move from zero to some number of unemployed because all states have some unemployment at all times. The permanent presence of some unemployment makes for a smoother series over time and precludes the threshold effect that other factors generate.

Long-Term Unemployed

The long-term unemployed factor is somewhat stable but has a wider coefficient of variation range than regular unemployment: from 0.066 to 0.471. The long-term unemployed measure falls between total unemployment and excess unemployment in terms of its coverage of the population. That is, it captures a more narrow part of the labor force (those unemployed 15 or more weeks) than total unemployment but it does not have a threshold trigger such that states are unlikely to have a zero value for this factor in any given year. As the total number of long-term unemployed has increased to a much higher level in the past two years, many states have seen an increase in their relative share, but each state has had at least some long-term unemployed in each program year.

Excess Unemployed

The excess unemployed factor is not well aligned (see previous discussion) with the dislocated worker population and is the most volatile of the current factors. The coefficients of variation range from 0.233 to 3. In fact, the maximum coefficient of variation for the excess unemployed factor is higher than the maximum coefficient of variation for all other factors. As indicated previously, it is not the nature of the unemployment data itself that drives this degree of volatility. Rather it is the use of a formula feature—a threshold of 4.5%—that creates such a volatile series.

The use of a firm threshold combined with nearly flat funding drives volatility in allocations by adding and removing states, essentially creating an "on-off" switch. In the period of PY2002 through PY2010, the number of "zero" states (i.e., states with no excess unemployment) has ranged from three to 34. This includes single-year swings of 19 (PY2002 to PY2003), 14 (PY2008 to PY2009), and 12 (PY2009 to PY2010) states. So, for example, in PY2008 an allocation of $394.6 million was divided among 23 states (an average of $17.2 million per state), while in PY2009 an allocation of $394.6 million was divided among 37 states (an average of $10.7 million per state).

Examination of Alternative Formula Features

In addition to factors, formula features, which are elements of a formula that adjust the factors in a formula, provide another method to alter the allocation of funding. At one end of the spectrum, a formula with only one factor (e.g., total unemployment) might have no features. That is, each entity subject to the formula would receive an allocation based solely on that entity's relative share of the factor in use. Most formulas, however, use a combination of features that transform the underlying factors. In its current form, the WIA dislocated worker formula uses two features—weighting of factors and a "threshold" feature for the excess unemployment factor. The discussion below covers numerous possible formula features, many of which are used in other federal state or local formula grants, and provides information on the likely effect of each feature. The list is not exhaustive but includes features found in other WIA formulas as well as features in other allocation formulas.

Factor Weight

This is a measure of the relative importance that each factor has in contributing to the overall allocation. Unless a formula has only a single factor, weights must be used to determine the amount that each factor counts in the formula. That is, weighting reflects prioritization associated with each factor. In the current WIA dislocated worker formula, each factor is equally weighted at one-third of the total, which means that all factors are considered to be of equal value in allocating funding.

Weighting could be used in the dislocated worker formula to place higher priority on certain factors. For example, if a higher weight were applied to the most volatile of the three factors—excess unemployment—the overall allocation would become more volatile, but this might be considered a worthwhile tradeoff if excess unemployment aligned well with the target population of dislocated workers. Because the current factors do not align well with the target population, however, weighting might be used to reduce volatility by allocating a greater share of the formula on the basis of a more stable factor, such as total unemployed. Furthermore, weighting could be used to establish a baseline allocation (e.g., high weight on a very stable factor such as civilian labor force), followed by lower weights on more targeted and more volatile factors such as mass layoffs or extended mass layoffs. Weighting offers a straightforward mechanism to change the relative contribution of factors without altering other features of the formula.

Hold Harmless and Stop Gain

Some formulas establish a minimum and/or maximum grant equal to a specified percentage of the amount received in a previous year. Usually, this is the immediately preceding year, although sometimes it is a "base year" that may be several years in the past. The minimum (maximum) percentage may be the full amount received in the previous year (i.e., 100%) or, more often, some lesser (greater) percentage (e.g., 85% for hold harmless, 130% for stop gain). Raising a state to its hold harmless level almost always reduces grants to other states that do not benefit from the hold harmless. Hold harmless amounts are only guaranteed if funds are sufficient to pay for them. If not, hold harmless amounts are ratably reduced to meet the level of the appropriation.³⁹ In the current WIA state formula grants, the Adult and Youth formulas contain a 90% hold harmless and a 130% stop gain provision.⁴⁰ Provisions to limit decreases and cap increases provide some measure of stability in state funding but may also weaken the linkage between changes in the state factors and the funding received by that state (i.e., creating an artificial floor or ceiling may make the formula less responsive than it would otherwise be to changes in the values of formula factors).

Such features could be employed in a number of ways in the dislocated worker formula to mitigate against large swings in year-to-year funding. The introduction of a "hold harmless" and "stop gain" provision in the dislocated worker formula would limit funding losses from year to year but would also cap gains for states experiencing a large increase in relative share of one of the formula factor values. On the other hand, the use of a hold harmless provision in the dislocated worker formula without a corresponding stop gain provision would mitigate losses but allow some additional funding to go to states with increased relative shares of formula factors. Because losses would be mitigated, however, there would be less available funding for reallocation than in the absence of a hold harmless. Finally, the use of hold harmless and stop gain provisions would also allow for different degrees of volatility. For example, a tighter range (e.g., 90% to 110%) would keep allocations relatively stable from year-to-year but would not be as responsive to changes in formula factor values as a formula that allowed greater volatility but guaranteed some minimum allocation from year to year (e.g., 70% to 130%).

Small State Minimum

In addition to hold harmless amounts, which are always expressed in terms of a percentage of a previous year grant (or relative share of a grant), some allocation formulas contain a state minimum grant expressed primarily in terms of a percentage of all allocations to states or as a fixed dollar amount per state. Such minimum grant provisions are aimed at providing what might be considered a minimum "viable" grant to all states. State minimums are set at a percentage of total state grants (typically 0.25%, 0.35%, or 0.5%). Occasionally, they are fixed dollar amounts (e.g., $500,000) or the greater (or lesser) of a fixed amount or a percentage of the total. In some cases, one or more "caps" may be placed on these minimums. When applying the minimum, the money to increase grants to states that would otherwise receive less than the minimum amount comes from all other states, which would see their initial grants ratably reduced.

In the current WIA state formula grants, the Adult and Youth formulas contain a small state minimum grant equal to 0.25% of the total available funding for states. In the dislocated worker formula, establishing a small state minimum grant would likely raise the allocation for a few states that typically do not have high levels of unemployment or large swings in unemployment (e.g., in PY2010 North Dakota would have received about $3 million, instead of the $690,086 it actually received, if the dislocated worker formula had a minimum grant of 0.25% of the total funding for states).

The use of a small-state minimum (or a general minimum) could be used to mitigate some of the allocation volatility, or at least provide some degree of a baseline allocation, by setting a minimum amount guaranteed to each state. Because the dislocated worker formula can, and often does, result in states receiving $0 from the excess unemployment factor, a state minimum could be used to guarantee that no state fully triggers off on this factor but instead receives a minimum grant in years in which it has no excess unemployment.

Indexation

As discussed in a previous section, the use of a threshold for excess unemployment in the current WIA dislocated worker formula creates considerable volatility in funding. The use of a fixed threshold (i.e., at least 4.5% unemployment qualifies as excess unemployment) as an index creates a cliff, which states are either on or off. Because of this cliff effect, allocations become volatile as states trigger on and trigger off the excess unemployment factor. As an alternative, some formulas use indexation of factors to moving averages (e.g., unemployment in one state compared to the average unemployment in that state for the three previous years) or the national average of a factor (e.g., excess state unemployment could be tied to the current national rate of unemployment). Because the effects of indexation depend on the structure and reference point, the remainder of this section demonstrates the potential use of indexation in the WIA dislocated worker formula.

Indexation in the Dislocated Worker Formula

Although the current WIA dislocated worker formula, which uses a formula feature of fixed indexation of 4.5% to define excess unemployment, is intended to target funding on states with high or increasing unemployment (which, as discussed previously, is not strictly aligned with the dislocated worker population as defined by WIA), the use of a fixed point of reference yields results that do not always achieve this goal or results in funding decreases in times of unemployment increases. Despite the problems with using a factor that triggers a state completely on or off of formula funding (i.e., the current fixed threshold of 4.5%), there are at least two alternatives to the use of a fixed index to measure excess unemployment that may mitigate some of the current volatility issues.

Indexing to the National Average Unemployment Rate

Indexing the excess unemployment factor in the current WIA formula to the national unemployment rate would provide a relative measure of unemployment that would reflect changing economic conditions. On the other hand, states with unemployment consistently below the national average would never cross the threshold into "excess" unemployment and thus would not receive an allocation for this portion of the formula. Indexing excess unemployment to the national average unemployment rate would have resulted in the following (conclusions below are based on an analysis of the PY2002 to PY2010 period):

• Of the 468 state-years (i.e., 50 states, DC, and Puerto Rico over nine years) between PY2002 and PY2010, 40% (188), were excess (i.e., unemployment was above the national average unemployment rate), while the remaining 60% (280) were not excess (i.e., unemployment was at or below the national average unemployment rate). This distribution is nearly the converse of the distribution using the 4.5% threshold in current law (i.e., 66% were non-zero and 34% were zero).
• Sixteen states would have never qualified for excess unemployment, while under the current formula, only three states had no excess unemployment in the PY2002 to PY2010 period.
• There would have been minimal variation in the number of states triggering on and off of the excess unemployment threshold. In the PY2002 to PY2010 period, the number of states with excess unemployment would have ranged from 18 to 25, compared to the range of 18 to 49 in the existing formula.

Using the national average would target funds toward states that have unemployment in excess of the national average as opposed to a statutorily specified fixed threshold. In addition, using the national average unemployment rate as the threshold would reduce volatility in annual allocations but would come at the cost of nearly one-third of the states never triggering on to excess unemployment. In essence, the use of a moving national unemployment rate maintains the concept of a threshold but reduces volatility by shrinking the number of states going on or off of the excess unemployment trigger. For states with unemployment consistently higher than the national average, there would be fewer sharp decreases in funding (e.g., Michigan lost 42% from PY2008 to PY2009 under the current formula), while for consistently low-unemployment states the excess unemployment trigger would cease to be a source of funding even if unemployment increased (but stayed below the national average).

Indexing to Recent State Unemployment Levels

Another alternative would be to measure excess unemployment in terms of a state's recent unemployment levels. That is, the index for excess unemployment might be based on each state's recent history of unemployment, such that "excess" would refer only to the individual state's recent unemployment levels. This "look back" factor would tie the concept of state excess unemployment to recent trends in that state and thus would remove the fixed threshold concept of the current formula. Because the index would essentially measure change from a relative base (i.e., states would not move from zero to non-zero or vice versa), this measure should reduce volatility. The "look back" factor would compare unemployment in the most recent year to the trailing average for the past several years.

As an example of the indexation approach, a three-year moving average was used as the baseline for each state's excess unemployment in a given year (e.g., the excess unemployment for a given state in PY2010 was determined by comparing unemployment in PY2010 to the average of PY2006 through PY2009). Thus the excess amount was calculated as the difference between the current year's level of unemployment and the average level of the three prior years. To determine each state's relative share of the change, the number of excess unemployed in a given state was divided by the total number of excess unemployed for that year.

Indexing excess unemployment to the each state's recent unemployment rate would have resulted in the following (conclusions below are based on an analysis of the PY2002 to PY2010 period):

• Of the 468 state-years (i.e., 50 states, DC, and Puerto Rico over nine years) between PY2002 through PY2010, 62% (292), were excess (i.e., unemployment was above the average level in the previous three years), while the remaining 38% (176) were not excess (i.e., unemployment was at or below the average level in the previous three years).
• All states would have had some excess unemployment during the period of PY2002 through PY2010 (under the current formula, three states had zero excess unemployment in the PY2002 to PY2010 period).

Using a moving baseline approach, such as that outlined above, would target funds toward states that have increased levels of unemployment relative to recent past levels but "underfund" states with high but stable levels of unemployment. This would allow states with consistently lower levels of unemployment to receive increased funding if unemployment increased, even if it did not surpass a fixed threshold. A state moving from 2% to 4% unemployment, for example, would receive additional allocations under this approach. Volatility could be reduced by freezing (or setting a percentage of prior year allocations below which funding could not drop) a state's allocations in times of decreasing unemployment. On the other hand, for states with consistently high levels of unemployment but small increases in a downturn, this approach would not provide much additional funding. For example, if a state's unemployment averaged 12% and increased some or remained stable in a downturn, that state would not gain much funding despite having a high and stable level of unemployment.

Conclusion

The analysis in this report supports four conclusions.

First, the current factors used in the WIA state grant formula for dislocated workers are not fully aligned with the target population. The degree of misalignment varies by factor but in general the current factors measure a much broader labor market group—the unemployed—than the subgroup of dislocated workers as defined in WIA.

Second, allocations from the dislocated worker formula are volatile, with much of the volatility driven by the excess unemployment factor. While some volatility is desirable in a formula (to direct allocations to areas with changes in the dislocated worker population), there is often a disconnect between economic conditions and allocations. The scale of the volatility (swings in excess of 30% or more from year to year) threatens the consistency of program operations.

Third, beyond factor-driven volatility, however, the practice of appropriating level amounts of funding for the dislocated worker program irrespective of changes in unemployment and job dislocations also contributes to large swings in state allocations. In the PY2001 through PY2010 period, appropriations for allocation to states through the WIA dislocated worker program have averaged $1.19 billion per year, with a range from $1.15 billion to $1.27 billion. Thus, given a largely constant level of funding, as economic conditions and the number of dislocated workers increases or decreases, the WIA dislocated worker program allocates the same total amount of funding in different ways. The flat funding creates a zero-sum allocation, such that one state's gain in funding must result in another state's loss of funding, even if economic conditions point in the same direction for most or all states.

Fourth, there are several alternative factors available to measure the dislocated worker population. The alternative factors appear to be more closely aligned with the WIA-defined population of dislocated workers and are generally less volatile. There are also options to add formula features that may reduce some of the fluctuations in funding.

Appendix A. Additional Factors for the Dislocated Worker Formula

Other factors for the WIA dislocated worker formula were considered for analysis in this report. Those factors considered but not presented in this report are listed below, along with a brief justification for their exclusion from full analysis.

Unemployment Insurance Initial Claimants. This series is part of the UI program data. An "initial claim" occurs when an individual files a request to determine entitlement to or eligibility for UI or when an individual files a request to begin a subsequent period of eligibility within a benefit year or period of eligibility. This measure of labor market dislocation is timely (reported weekly) and is reasonably well aligned with one of the WIA definitions of a dislocated worker. Because one of the WIA definitions of a dislocated worker includes "attachment to the labor force" as a criteria, the "first payments" series (analyzed in the text of this report) is slightly more aligned as it includes only those who receive a payment (i.e., have clear attachment to the labor market). In sum, because there is no tradeoff in terms of timeliness or data availability and because the category of initial claimants includes some individuals who do not have sufficient labor force attachment to qualify for UI benefits, the UI initial claimants series was not used in this report.

Permanent Job Losers. This series is part of the CPS data and counts permanent job losers and persons who completed temporary jobs. From an alignment perspective, this data series would match part of the definition of a WIA dislocated worker well as it captures workers who have lost jobs through no fault of their own and who may be unlikely to return to a previous occupation or industry. The main drawback to this series, however, is timeliness. Due to sample size considerations, three-year averages would be required to generate reliable estimates at the state level. Given the importance of aligning allocations with need, a lag of this size may not reflect current conditions. In addition, data for Puerto Rico are not available from this series.

Displaced Worker Survey (DWS). This series is a biennial supplementary survey to the CPS. In the DWS, displaced workers are defined as persons at least 20 years of age who lost or left their jobs because of a company move or closure, insufficient work, or abolishment of a position or shift. The survey, which is conducted in January, covers the three previous calendar years (e.g., the January 2010 DWS covered the January 2007 through December 2009 period). The DWS would align very closely with the target populations in the WIA dislocated worker formula. Data are not available at the state level, however, due to the sample size of the DWS.

Appendix B. Coefficient of Variation Analysis for Current and Alternative WIA Dislocated Worker Formula Factors

Footnotes

1.	WIA contains five titles. For a comprehensive overview of the Workforce Investment Act, see CRS Report R41135, The Workforce Investment Act and the One-Stop Delivery System, by [author name scrubbed].
2.	A program year in WIA runs from July 1 through June 30, while the federal fiscal year runs from October 1 through September 30. For example, PY2010 runs from July 1, 2010, through June 30, 2011, while FY2010 runs from October 1, 2009, through September 30, 2010. Appropriations made in a fiscal year fund WIA programs in the corresponding program year. For example, FY2010 (October 1, 2009–September 30, 2010) appropriations provide funding for PY2010 (July 1, 2010–June 30, 2011).
3.	For example, GAO noted that, "some states have reported that this volatility makes program planning difficult. While some degree of change in funding is to be expected due to changing dislocations in the workforce, changes in funding do not necessarily correspond to these changes." It is notable that this GAO finding was reported in 2003, before the period of even greater volatility in 2008 and 2009. U.S. General Accounting Office, Workforce Investment Act: Issues Related to Allocation Formulas for Youth, Adults, and Dislocated Workers, GAO-03-636, April 2003, p. 5, http://www.gao.gov/new.items/d03636.pdf.
4.	Of the $5.5 billion PY2010 Title I funding, approximately $3 billion was allocated by state formula grants, $1.7 billion was allocated to Job Corps, and the remaining $800 million was allocated primarily by competitive grants for national programs.
5.	The focus of this report is on the allocation formula for one Title I program rather than the services provided by activities authorized by Title I. For a more complete discussion of the array of required and allowable activities, see CRS Report R41135, The Workforce Investment Act and the One-Stop Delivery System, by [author name scrubbed]. For in-depth information about youth activities authorized under Title I of WIA, see CRS Report R40929, Vulnerable Youth: Employment and Job Training Programs, by [author name scrubbed]
6.	Required and allowable statewide activities are described in Sections 129(b) and 134(a) of WIA.
7.	Required and allowable local employment and training activities are described in Sections 134(d) and 134(e) of WIA.
8.	Statutory provisions do not specify an amount of time an individual must spend or the number of attempts that must be made to gain employment before moving to the next level in the sequence of services. WIA regulations provide additional guidance on the sequence of services but do not set time or job application requirements. Specifically, the regulations stipulate that an individual must receive at least one core service before receiving intensive services and must receive at least one intensive service before moving to training services. The regulations clearly state that there is "no Federally-required minimum time period for participation" in core or intensive services before receiving the next level of service.
9.	The workforce development system designed by WIA is premised on universal access, such that an adult age 18 or older does not need to meet any qualifying characteristics in order to receive core services. However, Section 134(d)(4)(E) of WIA stipulates that in the event funds allocated for employment and training activities are "limited," priority is to be given to recipients of public assistance and other low-income individuals for intensive and training services. It is left to the discretion of the local WIB, in consultation with the state's governor, to determine this prioritization.
10.	"Dislocated worker" is defined in WIA Section 101(9).
11.	WIA Section 101(9)(B)(iii) also stipulates that for purposes of receiving WIA services other than training, intensive, or supportive, an individual may also be considered "dislocated" if employed at a facility in which a "general" announcement of closure has been made.
12.	WIA defines a "displaced homemaker" as an individual who has been providing unpaid services to family members within the home and who has been supported by the income of another family member but is no longer supported by that income and is experiencing difficulty in obtaining or upgrading employment.
13.	Required rapid response activities are described in Section 134(a)(2)(A) of WIA. The term "rapid response activity" is defined in Section 101(38) of WIA.
14.	Title I state formula grant allocations include the District of Columbia and Puerto Rico. In this report, the term "states" thus refers to the 50 states plus DC and Puerto Rico.
15.	This general discussion of formula funding draws from CRS Report R41625, Federal Benefits and Services for People with Low Income: Programs, Policy, and Spending, FY2008-FY2009, by [author name scrubbed]. Specifically, see pages 32-35.
16.	The CPS is a monthly household survey that includes approximately 60,000 households. Data collected are based on the status reported for the calendar week including the 12th of the month. The 12-month averages used in the WIA dislocated worker formula typically are the averages for the 12-month period ending September 30 of the year prior to the PY (e.g., For PY2010, the average unemployment was calculated for the 12 months ending September 30, 2009).
17.	The total excess unemployed for the U.S. is calculated by summing all of the excess unemployed individuals in states with excess unemployment. That is, if a state does not have an unemployment rate above 4.5%, that state by definition has no excess unemployment and thus does not contribute to the total number of excess unemployed in the United States.
18.	The factor definitions are described in statute but the operationalization of factors (e.g., measurement period) is determined by the Secretary of Labor.
19.	The CPS is the official source of monthly labor force statistics for the United States. The reference week is generally the calendar week that contains the 12th day of the month.
20.	There is a variable in the CPS monthly survey that covers the reason for a respondent's unemployment. A major drawback of using these data is timeliness. Due to sample size issues, pooling of three years is necessary to obtain reliable estimates. This means that the time lag would range from one to three years. There is a more timely alternative from the Bureau of Labor Statistics that will be discussed in the section on alternative formula factors.
21.	Each "year" in this analysis represents the change from one program year to the next (e.g., PY2002 is the change from PY2001 to PY2002). Note that the use of PY2002 as the starting year for analysis of the factors in this report (rather than PY2001, which was the first year of WIA allocations) is due to the availability of allotment formula data factors, which are available on the DOL website starting in PY2002. See http://www.doleta.gov/budget/statfund.cfm.
22.	The American Recovery and Reinvestment Act of 2009 (ARRA; P.L. 111-5) provided about $1.24 billion in additional state formula grant funding in FY2009, which was slightly more than the regular FY2009 appropriation of $1.18 billion. In this report, the additional funding for the dislocated worker program provided by the ARRA is not considered in discussions of funding volatility. The major purpose of this report is to evaluate the WIA dislocated worker formula as it functions under the normal appropriations process. If supplemental funding for the WIA dislocated workers state formula grant program occurred regularly, then the analysis in this report would change to reflect this. However, appropriations for the WIA dislocated workers program has remained relatively flat over the entire period of WIA—an average of $1.19 billion per year—with no regularly occurring supplemental to address spikes in unemployment such as those in 2008. The relatively flat funding actually allows the effects of the formula itself to be seen more clearly since fluctuating funding does not drive much of the change in annual state-by-state allocations.
23.	A more technical comparison of factor volatility may be found below.
24.	To clarify, what is considered excess unemployment varies and depends to some extent on the economic conditions in a given period of time. Prior to the 1990s, for example, many economists thought U.S. unemployment could not go much below 6% without sparking inflation. By the late 1990s, however, the combination of low unemployment (around 4% nationally) and low inflation led some to suggest that the "non-accelerating inflation rate of unemployment" (NAIRU) might be in the 4%-5% range. As noted previously in this report, the 4.5% threshold for defining "excess" unemployment has been around since at least the early 1970s and is near the "full employment" rate of 4% as defined by the Humphrey-Hawkins Full Employment Act of the late 1970s.
25.	To reiterate a point previously raised in this report, the funding discussed here does not include additional funds provided by the ARRA. While the additional ARRA funding ameliorated some of the funding losses in states with allocation losses in PY2009, additional funding is not part of the regular WIA formula. The focus of this report is on the functioning of the regular WIA dislocated worker formula. If supplemental appropriations were a regular congressional response to the funding volatility in the WIA dislocated worker formula, the analysis in this report would change. However, the intent of the analysis here is to understand the formula in the absence of one-time or irregular supplements in funding.
26.	Of the 468 state-years (i.e., 50 states, DC, and Puerto Rico over nine years) between PY2002 and PY2010, 66% (307), were excess (i.e., unemployment was above 4.5%), while the remaining 34% (161) were not excess (i.e., unemployment was at or below 4.5%). Thus, by WIA's definition, a majority of states have excess unemployment a majority of the time. In that sense, having excess unemployment is the norm rather than the exception. It is possible that a minority of states accounted for a majority of the "excess" state-years. However, this is not the case. Though not presented here, a separate analysis shows that in the nine years from PY2002 to PY2010, 38 states had unemployment rates in excess of 4.5% in at least five of those years.
27.	The 12 states are Hawaii, Idaho, Iowa, Kansas, Maryland, Montana, New Hampshire, New Mexico, Oklahoma, Utah, Virginia, and Wyoming. Three states—Nebraska, North Dakota, and South Dakota—did not have excess unemployment in PY2009 or PY2010 and are not included in Figure 1.
28.	All of the states that had excess unemployment in PY2009 (37 including DC and Puerto Rico) also experienced increases in the number of excess unemployed from PY2009 to PY2010. Because the formula is based on the concept of relative share, the 37 states with existing excess unemployment allocations did not receive proportionate changes in funding to provide funds for the 12 new states. In fact, 16 of these 37 states saw decreases in funding from the excess unemployed factor. Whether or not a state received a decrease or increase from PY2009 to PY2010 depended on each state's relative (to other states) increase in excess unemployed.
29.	Three other factors were considered but not used in the analysis in this report. For a brief description of these factors and the reasons for their exclusion, see Appendix A.
30.	The alignment of current factors is covered earlier in the report so there is no additional discussion of those factors in this section. The current factors are included in Table 3 as a means of comparison.
31.	Detailed information on ML program is available at http://www.bls.gov/mls/.
32.	Detailed information on EML program is available at http://www.bls.gov/mls/. The EML program also provides information on the reason for layoff, including "worksite closure." These data, however, are not reported at the state level.
33.	Detailed information on UI program is available at http://workforcesecurity.doleta.gov/unemploy/uifactsheet.asp.
34.	Detailed information on LAUS program for state alternative measures of labor underutilization is available at http://www.bls.gov/lau/stalt.htm.
35.	In this report, the measure of central tendency used is the mean, which is calculated as the arithmetical mean of the data values. That is, the mean is the sum of all data values divided by the number of observations in the data. Technically, the standard deviation is the square root of the variance (the sum of each observation's squared deviation from the mean) for a particular set of data. A larger standard deviation indicates that the data are more dispersed across units, while a smaller standard deviation indicates that the data are less dispersed. Coefficients of variation are computed for each factor by state share as an average across all years. For example, Alabama's coefficient of variation for excess unemployment is computed by dividing Alabama's standard deviation of its relative share of excess unemployment (across all program years from 2002 through 2010) by its mean relative share of excess unemployment (across all program years from 2002 through 2010). In this example, Alabama's mean value was 0.0096 and the standard deviation was 0.0075. Thus, the coefficient of variation was 0.79.
36.	There are three states—Nebraska, North Dakota, and South Dakota—for which it is not possible to calculate a coefficient of variation for the excess unemployment factor. These three states did not have unemployment above 4.5% for any of the program years in this analysis. These three states were given a value of zero and are not counted in the discussion of the range of values for coefficients of variation.
37.	While there is some correlation between the occurrence of mass layoff events and overall economic activity (i.e., gross domestic product), it is not a perfect relationship. For example, while real (inflation-adjusted) GDP grew at a rate of 4.1% from 1999 to 2000, the number of initial claimants from mass layoff events grew by 17%. Similarly, real GDP did not change from 2007 to 2008 but the number of initial claimants from mass layoff events grew by 33%.
38.	To reiterate the definitions from Table 3, the measures for mass layoffs and extended mass layoffs are initial claimants and total separations, respectively. The initial claimants data are the only measure available for the mass layoffs series. While initial claimants and total separations are available for the extended mass layoffs series, total separations are used because they should include individuals who are part of an extended mass layoff event regardless of their eligibility for UI benefits.
39.	Ratable reduction or ratable increase is the process of either reducing or increasing grants as initially calculated in order to adjust for the level of available appropriations or application of certain formula factors, such as a state minimum or stop gain provisions. These reductions or increases are applied in proportion to initial grants (i.e., they are "ratable"). For example, raising certain states to minimum grant amounts requires that funds be redistributed from states with initial grants above the minimums. Ratable reduction reduces funds in proportion to their initial grants for states above minimum levels and redistributes these funds to states with initial grants below minimum levels. When ratable reduction occurs, all states above the minimum have their initial grants reduced by the same percentage, resulting in different dollar amount changes.
40.	The structure of a hold harmless or stop gain provision could vary but the intended result would be to limit year-to-year changes in allocations to a specified level. For example, in the current WIA Adult and Youth state grant formula, the hold harmless provision is 90% of a state's relative share (rather than of the allocation amount itself) of prior year funding and the stop gain provision is 130% of a state's relative share of prior funding (rather than of the allocation amount itself).