Health Affairs, 25, no. 2 (2006): w1-w14 (Published online 17 January 2006) doi: 10.1377/hlthaff.25.w1 © 2006 by Project HOPE	New Online   Senate Health Reform Bill   Rewarding Providers   Public Option Policy Brief   Health Reform & Abortion   Delivery System Reform

This Article Abstract Figures Only Reprint (PDF) Technical Appendix Submit a response to this article Alert me when this article is cited Alert me when Comments are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Long, S. K. Articles by Graves, J. A. Search for Related Content PubMed PubMed Citation Articles by Long, S. K. Articles by Graves, J. A.

Web Exclusives

DataWatch

Are Adults Benefiting From State Coverage Expansions?

Abstract

 
This study provides a rigorous evaluation of state efforts to expand insurance coverage in California, Massachusetts, New Jersey, and Wisconsin. Overall, parents in Wisconsin and parents and childless adults in Massachusetts experienced the largest expansions in public coverage, with few, if any, offsetting reductions in private coverage. In contrast, the coverage expansions for parents in California and New Jersey led to increased enrollment, but often at the expense of private coverage. Because constraints on state resources are putting pressure on expansion efforts, we find evidence that cutbacks will place more adults at risk of being uninsured.

Both before and after the implementation of SCHIP, some states were also interested in addressing the coverage problems facing low-income adults. For the most part, states expanded coverage to the parents of children who were already eligible for coverage under Medicaid or SCHIP, but a few states also addressed the needs of childless adults, who are not typically covered by public programs unless they are aged, blind, or disabled.

All of this activity has spawned a growing body of literature evaluating the effects of these policy changes on health insurance coverage for adults.⁴ In this paper we focus on the impacts of recent coverage expansions targeted at adults in California, Massachusetts, New Jersey, and Wisconsin. We provide a rigorous evaluation of efforts to expand insurance coverage to parents in all four states and, in Massachusetts, to childless adults.

Overview Of State Coverage Expansions, 1997–2002

Top Overview Of State Coverage... Methods And Data Study Results Discussion NOTES

 
Prior to the states’ efforts to expand coverage, eligibility for public coverage varied (Exhibit 1). California, New Jersey, and Wisconsin undertook relatively straightforward expansions in Medicaid or SCHIP benefits for parents, while Massachusetts undertook a more complicated effort that included expansions for both parents and childless adults. After the expansions, eligibility ranged from 14 percent of nondisabled parents in Massachusetts to 24 percent of all parents in Wisconsin.

View larger version (27K): [in this window] [in a new window]   EXHIBIT 1 Changes In Eligibility For Public Coverage Among Adults In Four States Following State Coverage Expansions

California. In 2000, California used Section 1931(b) authority to extend eligibility under its Medicaid program (Medi-Cal) to parents with incomes up to 100 percent of poverty (an increase from its prior coverage level of about 74 percent of poverty). The following year, the state extended Medicaid benefits to aged, blind, and disabled people with incomes up to 133 percent of poverty (an increase from 100 percent). Although in 2002 California applied for and received approval of a Section 1115 waiver to expand coverage to the parents of SCHIP children with incomes up to 250 percent of poverty, state budget pressures and a change in administrations halted efforts to implement that expansion. Under California’s coverage expansions, Medicaid eligibility for parents increased from about 15 percent of all parents in the state in 1999 to 21 percent in 2002 (Exhibit 1).

New Jersey. Beginning in 2000, New Jersey expanded Medicaid coverage under FamilyCare to parents with incomes up to 200 percent of poverty (an increase from 41 percent) and to childless adults with incomes up to 100 percent of poverty (they were not previously eligible) under a Section 1115 waiver. Unfortunately, the 2001 economic recession, state budget concerns, and rising health care costs lead New Jersey to pull back from its coverage expansions. By the end of 2001, the state had closed enrollment for childless adults. In July 2002, New Jersey closed new enrollment for parents and scaled back medical benefits for childless adults.

Despite the state’s decision to scaleback its expansions, New Jersey expanded eligibility for parents from 7 percent of all parents in 1999 to 17 percent in 2002 (Exhibit 1). Here, we focus on New Jersey’s expansion of coverage to parents.

Wisconsin. Wisconsin expanded coverage to parents with incomes up to 185 percent of poverty (an increase from 51 percent) under a Section 1115 waiver in the second half of 1999. The expansion of coverage under BadgerCare increased the share of parents eligible for coverage from 6 percent of all parents in early 1999 to 24 percent in 2002 (Exhibit 1).

Massachusetts. Massachusetts began its effort to expand health care coverage to its entire low-income population in 1997 with the creation of MassHealth, a statewide demonstration project under a Section 1115 waiver. Prior to MassHealth, parents with incomes up to about 86 percent of poverty were eligible for Medicaid coverage, while childless adults were not eligible for any coverage unless they were disabled.

Under MassHealth, Medicaid benefits are provided to pregnant women with incomes up to 200 percent of poverty; parents with incomes up to 133 percent of poverty; long-term unemployed adults with incomes up to 133 percent of poverty; unemployed adults receiving unemployment benefits with incomes up to 400 percent of poverty; and all disabled adults, regardless of income.

MassHealth provides premium assistance support to parents with incomes of 133–200 percent of poverty and childless adults with incomes below 200 percent of poverty who have access to employer-based coverage. And, although not insurance coverage per se, Massachusetts also provides health care through publicly subsidized providers to low-income adults who are not eligible under any of the expansions outlined above.

Under MassHealth, the share of parents eligible for insurance coverage (defined as Medicaid benefits or premium assistance support) increased from 9 percent of all nondisabled parents in 1997 to 14 percent in 2002. Over the same period, the share of nondisabled childless adults eligible for coverage rose from zero to 20 percent. In the analysis below, we focus on the effects of Massachusetts’ expansion efforts for nondisabled parents and childless adults.

State coverage changes since 2002. Although this study focuses on the impact of state coverage expansions before 2002, it is worth noting that despite serious state budget shortfalls, California and Wisconsin have maintained the coverage expansions they had implemented by 2002. In Massachusetts, which implemented a much broader set of expansions, state budget difficulties have led the state to curtail eligibility, reduce benefits, and increase cost sharing for some populations since 2002. New Jersey froze new enrollment of childless adults in 2001 and parents in 2002. However, the state received a Health Insurance Flexibility and Accountability (HIFA) waiver in 2003 to enroll up to 12,000 parents with incomes up to 133 percent of poverty who had applied but were not enrolled prior to the state’s 2002 enrollment freeze.

Methods And Data

Top Overview Of State Coverage... Methods And Data Study Results Discussion NOTES

 
We present a brief summary of our study methods here. A more detailed description of our methods is provided in an online technical appendix.⁵

Difference-in-difference approach. We took advantage of the "natural experiments" that occurred in the four study states to compare coverage for adults before and after each state implemented its eligibility expansion. To control for underlying trends in insurance coverage not related to the eligibility expansions, we subtracted changes in health insurance coverage over the same time period for a comparison group of adults who were not affected by the state’s policy changes. We used these comparison groups in a "difference-in-difference" or comparative change framework.⁶ Thus, for example, if public coverage increased by ten percentage points in Wisconsin following the expansion of coverage and, over the same time period, increased by six percentage points for the comparison group, the net effect of the expansion would be a gain in public coverage of four percentage points.

For California, New Jersey, and Wisconsin, we used 1999 as the pre-expansion period and 2002 as the post-expansion period. For Massachusetts, which implemented its expansion earlier, we used 1997 as the pre-expansion period and combined data from 1999 and 2002 for the post-expansion period.

Because there is no perfect comparison group in the absence of random assignment, we drew on several alternate comparison groups to provide the counter-factuals for our analysis. These counterfactuals provide an estimate for what would have happened in each study state if there had been no expansion effort. By using multiple comparison groups, we identified the states where our estimates were sensitive to the particular comparison group chosen (which makes us less confident in the reliability of the results) and the states where our findings were robust across the comparison groups (which increases our confidence in the reliability of the results). The comparison groups for parents in each study state include the following: (1) model 1—parents in the same state with incomes just above the income cutoff for public program eligibility (referred to as near-eligible parents); (2) model 2—childless adults in the same state with income below the income cutoff for eligibility for parents (referred to as eligible childless adults); and (3) model 3—parents in other states who would have been eligible for public coverage under the study state’s post- expansion eligibility rules if they had lived in that state (referred to as eligible adults).⁷

The near-eligible-parent group (model 1) allowed us to compare eligible parents with parents in the same state having somewhat higher incomes (and, thus, subject to the same economic and health care environment), while the eligible-childless-adult group (model 2) allowed us to compare eligible parents with other adults in the same state having similar incomes. However, neither in-state comparison group allowed us to compare the low-income parents who are subject to the coverage expansion with a group of low-income parents who are not subject to it. For that comparison, we turned to the cross-state comparison group of parents in other states (model 3). To estimate the impacts of the expansion in coverage to childless adults in Massachusetts, we used variations of model 1 (based on near-eligible childless adults) and model 3 (based on eligible childless adults in comparison states).

Selecting states for the cross-state comparison groups. For each study state, we selected a group of comparison states (model 3) that (1) did not implement an expansion in coverage for adults during the study period for that study state (1997–2002 for Massachusetts and 1999–2002 for the other three study states); and (2) had levels of Medicaid eligibility and private coverage similar to the study state prior to the study state’s coverage expansion.⁸

Because there is wide variation in population characteristics across the states, we used propensity score matching methods to reweight the sample, to ensure that the characteristics of eligible adults in the cross-state comparison groups are similar to those of eligible adults in the study state.

Multivariate model specification. We isolated the effects of the state expansions on coverage through difference-in-difference models and multivariate regression methods. The regression models included a rich set of variables to control for differences in the treatment and comparison groups (beyond treatment status) and differences within each group over time that could affect our outcome of interest: health insurance coverage.⁹ We estimated linear probability models for ease of computation and to facilitate comparisons across alternative models.

The models of insurance coverage included characteristics of the individual, job characteristics, the premium contribution they would likely face to purchase private health insurance, characteristics of their local health care market, and the economic conditions of the local area (Exhibit 2).¹⁰

Crowding out may occur because newly eligible people drop private coverage to enroll in the public program or because people enrolled in the public program stay enrolled even if they get the option of private coverage through, for example, a new employer. It was not possible with the data used in this study to distinguish between these two paths through which crowding out can occur.

We tested for whether the share of the increase in public coverage attributable to a decline in uninsurance and the share attributable to the crowding out of private coverage are significantly different from zero. One limitation of our estimates of crowd-out, like those generated by others, is that we cannot assess the extent to which the private coverage that is being replaced by public coverage is "adequate." We return to this point later.

Data. The primary source of data for this study was the National Survey of America’s Families (NSAF) for 1997, 1999, and 2002. NSAF, a nationally representative household survey of more than 42,000 families, is uniquely suited for this study because it oversamples low-income households, oversamples families in thirteen focal states (including our study states), and provides data before and after the implementation of the public program expansions. Because of NSAF’s complex design, we relied on replication methods to obtain accurate variance estimates of the coefficients in our model and of our estimates of crowd-out.

We obtained area characteristics from the Area Resource File (ARF); we calculated county premium data from the Medical Expenditure Panel Survey (MEPS) and County Business Patterns (CBP) data. The outcome measure for our analysis was insurance coverage at the time of the survey. People were assigned to the following hierarchical insurance groups in each year: (1) private coverage (includes coverage from a current or former employer or union or under a military program and insurance purchased by the individual), (2) Medicaid/SCHIP, (3) other (includes coverage not captured elsewhere), and (4) uninsured.

Limitations. First, although we used a strong quasi-experimental design and controlled for a wide array of individual and area characteristics in the regression analysis, and although our results were generally consistent across the comparison groups, it is always possible with quasi-experimental methods that unmeasured differences between the treatment and comparison samples may exist that confound the impact estimates. Second, as with all analyses based on survey data, our ability to detect small changes in insurance coverage that result from state coverage expansions was constrained by the sample sizes in the survey.

Study Results

Top Overview Of State Coverage... Methods And Data Study Results Discussion NOTES

 
In all of the study states, public coverage rates increased significantly (Exhibit 3). The gains were largest among parents in New Jersey, Wisconsin, and California. However, there was also a significant reduction in rates of private coverage among eligible parents in each of these states. These reductions were sufficiently large among parents in New Jersey and California that the rates of uninsurance were not statistically different across the years. In contrast, uninsurance fell significantly for parents in Wisconsin and for parents and childless adults in Massachusetts.

Exhibit 4 isolates the impact of the state coverage expansions based on the difference-in-difference models and regression analysis using the alternate within-and across-state comparison groups. The entries in Exhibit 4 are estimates of the net effect of the expansion in coverage, after trends in coverage over the study time period and individual and area characteristics were controlled for.

We also found that private coverage eroded for parents following the expansion in public coverage beyond that observed in the comparison groups, but those estimates are not always statistically significant. In New Jersey there were significant reductions in private coverage for parents relative to all three comparison groups. In California we found lower rates of private coverage for parents following the expansion of public eligibility based on all the comparison groups; however, the estimates are only significantly different from zero for one comparison group. By contrast, in Massachusetts and Wisconsin the estimates of the change in private coverage following the expansion in public eligibility were relatively small in magnitude and not significantly different from zero relative to any of the comparison groups, which suggests little if any net change in private coverage as a result of the expansion efforts in those states.

For childless adults in Massachusetts, the findings are quite different. For that population, both public and private coverage expanded as a result of MassHealth relative to the comparison groups. Given that the key component of MassHealth for nondisabled childless adults is premium assistance for workers with access to employer-sponsored coverage or unemployed workers with access to Consolidated Omnibus Budget Reconciliation Act (COBRA) coverage through a former employer, it is likely that the increase in private coverage we observed under MassHealth reflects the availability of such subsidies.¹³ Since private coverage increased significantly for childless adults in Massachusetts under MassHealth, there is no crowding out for that population group.

Exhibit 5 uses the estimates from Exhibit 4 to construct our best estimates of crowd-out for parents. The exhibit shows the share of the increase in public coverage for parents in each of the study states related to a decline in uninsurance and the share related to a decline in private coverage. The latter provides the estimate of crowd-out for each state.

The evidence in California and Massachusetts is more mixed, with the estimate of crowd-out from one comparison group in each state significantly greater than zero and the remaining estimates not statistically significant. In California, where the estimates of the declines in uninsurance and private coverage were not generally statistically significant (see Exhibit 4), crowd-out could be as low as 0 percent and as high as 100 percent. In contrast, in Massachusetts, where we found a significant reduction in uninsurance for parents (Exhibit 4), crowd-out is significantly less than 100 percent.

The conclusion from Exhibit 5 is that crowding out seems to occur but that its magnitude likely varies across states and is sensitive to the choice of comparison groups. In fact, crowding out might be quite small or nonexistent in some states, particularly for parents in Wisconsin. However, larger sample sizes are needed to obtain precise estimates of the degree of crowding out.

Discussion

Top Overview Of State Coverage... Methods And Data Study Results Discussion NOTES

 
Overall, Wisconsin and Massachusetts have been most successful at reducing uninsurance among adults. In Wisconsin, BadgerCare expanded public coverage rates by double digits above that of the comparison groups, while private coverage held reasonably steady. Similarly, in Massachusetts we observed increases in public coverage for both nondisabled parents and childless adults as a result of MassHealth. However, for childless adults in Massachusetts there was no offsetting reduction in private coverage. In fact, private coverage expanded for this population relative to the comparison groups, likely reflecting the very low incomes of many in that expansion population and the reliance on premium assistance programs for nondisabled childless adults under MassHealth.

The coverage expansions in California and New Jersey also led to an increase in public enrollment among the eligible adults, but there was no evidence of a significant reduction in uninsurance in California and limited evidence in New Jersey relative to their respective comparison groups.

State policymakers can view these findings as strong evidence that adults in the income groups represented by these eligibility expansions are not reluctant to enroll in public insurance programs. We found no evidence to suggest that there is enough stigma associated with Medicaid or SCHIP to cause adults to pass up the opportunity to enroll. In part, this may be a result of the rebranding of these programs in three of the states we studied (BadgerCare in Wisconsin, FamilyCare in New Jersey, and MassHealth in Massachusetts). California was the exception, but its Medicaid program has been called Medi-Cal since its inception.

Our findings also show that policymakers should not assume that they are only enrolling people who would have otherwise been uninsured. Some crowding out of private coverage should be expected, although the degree should not be expected to be uniform across states. While the crowding out of private coverage does increase the direct costs to the public sector, it is likely that at least some share of the private coverage that is replaced by public coverage provides relatively limited benefits or has high premiums, deductibles, and copayments, or both. Crowd-out of inadequate coverage may well yield benefits to families (in terms of better health and financial security) and their communities.

Explaining the variations. The lack of uniformity across states may be attributable to a number of factors, and, although we cannot assess the role that each may play, it is important to keep in mind that they can influence our findings and our ability to generalize to other states and other populations. First, variation within and across states in the private health insurance market, the benefits and costs of private coverage, and the benefits and costs of public coverage likely provide different incentives to eligible adults about enrolling in a public program versus opting for a private plan.

Second, not all states started from the same initial eligibility standards, and the extent of the state expansions varied. For example, New Jersey and Wisconsin had less generous programs for parents prior to the policy change and bigger expansions than the other two states.

Third, the effects of the expansions may be related to specific policies beyond program eligibility. For example, Massachusetts provided premium assistance support rather than direct Medicaid coverage for some expansion populations. New Jersey imposed a waiting period for parents who had employer-sponsored coverage prior to enrolling in FamilyCare, which it waived for parents coming from nongroup or COBRA coverage. This might have allowed some parents to drop employer-sponsored coverage to join the program.

Finally, although we approached the methodological challenge of selecting appropriate comparison groups using similar criteria in each state, the objective might not have been met equally well. In that case, some of the variation in effects we observed might simply be the result of the inherent difficulty in evaluating natural experiments.

Policy implications. Together, this study and earlier work by Richard Kronick and Todd Gilmer show the effects of public coverage expansions for adults in eight states enacted during the decade preceding the recent state budget crises.¹⁴ Insurance coverage expanded in five of the eight states for some or all of the target populations. In the other states, there were either gains in coverage that were not statistically significant or no change that could be measured with the methods used. These findings suggest that when states have the resources to expand public coverage to adults, they have a good chance of being able to reduce their uninsurance rates.

During the past few years, state resources have been constrained as a result of the recession. This led a number of states that were leaders in the expansion of health insurance to cut back on their programs. Some states, such as Massachusetts and Washington, made modest cutbacks in eligibility, while others, such as Tennessee, Missouri, and New Jersey, made more extensive program changes. Despite the budget constraints, some states, such as Arizona and Illinois, have continued to expand their public coverage of adults. The ebb and flow in these programs is likely to be inevitable, given the shifts in resources available to states. However, the results of this study show that if public programs are cut, more adults will be at risk of being uninsured and, as a result, at risk of adverse health consequences. This is likely to be increasingly true in the future, as employer-sponsored coverage continues to erode as a result of rising health insurance premiums, cuts in the benefit package, increased cost-sharing requirements, and fewer firms offering coverage.¹⁵

Editor's Notes

 
Sharon Long (slong{at}ui.urban.org) and Stephen Zuckerman are principal research associates at the Health Policy Center, Urban Institute, in Washington, D.C. John Graves is a research associate there.

This research is part of the Assessing the New Federalism Project and received funding from the Robert Wood Johnson, Annie E. Casey, W.K. Kellogg, Henry J. Kaiser Family, Ford, John D. and Catherine T. MacArthur, Charles Stewart Mott, and David and Lucile Packard Foundations, among others.

NOTES

Top Overview Of State Coverage... Methods And Data Study Results Discussion NOTES

 

1. L. Dubay, J. Haley, and G. Kenney, "Children’s Eligibility for Medicaid and SCHIP: A View from 2000," Assessing the New Federalism Policy Brief no. B-41, March 2002, http://www.urban.org/UploadedPDF/310435.pdf (accessed 3 January 2006).
2. G. Kenney, J. Haley, and A. Tebay, "Children’s Insurance Coverage and Service Use Improve," Snapshots of America’s Families 3, no. 1, July 2003, http://www.urban.org/UploadedPDF/310816_snapshots3_no1.pdf (accessed 3 January 2006).
3. S. Zuckerman, "Gains in Public Health Insurance Offset Reductions in Employer Coverage among Adults," Snapshots of America’s Families 3, no. 8, September 2003, http://www.urban.org/UploadedPDF/310850_snapshots3_no8.pdf (accessed 3 January 2006).
4. See, for example, J.B. Mitchell et al., "Impact of the Oregon Health Plan on Access and Satisfaction of Adults with Low Income," Health Services Research 37, no. 1 (2002): 33–42[Medline]; L. Moreno and S.D. Hoag, "Covering the Uninsured through TennCare: Does It Make a Difference?" Health Affairs 20, no. 1 (2001): 231–239[Abstract/Free Full Text]; R. Kronick and T. Gilmer, "Insuring Low-Income Adults: Does Public Coverage Crowd Out Private?" Health Affairs 21, no. 1 (2002): 225–239[Abstract/Free Full Text]; and S.K. Long and S. Zuckerman, "MassHealth Succeeds in Expanding Coverage for Adults," Inquiry 41, no. 3 (2004): 268–279.[Web of Science][Medline]
5. The technical appendix is available online at http://content.healthaffairs.org/cgi/content/full/hlthaff.25.w1/DC2.
6. J.W. Wooldridge, Econometric Analysis of Cross Section and Panel Data (Cambridge, Mass.: MIT Press, 2002).
7. The lower bound for the income band for the near-eligible group in a state is set at the income level that captures 99 percent of those predicted to be eligible under the rules in place in that state. The near-eligible parents comparison group constitutes a 175-percentage-point band above that income cutoff. Sensitivity analyses using different income bands and income cutoff points yielded results similar to those reported here. For New Jersey, we included both eligible and near-eligible childless adults because the state expanded coverage temporarily to some low-income childless adults. We excluded the childless adults who would have been eligible for that expansion from the comparison group. Because Massachusetts expanded coverage to both parents and childless adults, there is no model 2 comparison group in that state.
8. Since Massachusetts implemented MassHealth in 1997, it is both a study state and, for the period 1999–2002, a comparison state for other study states. The states included in the cross-state comparison groups for each of the study states were as follows. California: FL, KY, LA, MS, NM, NV, OR, TN, TX, and WV; New Jersey: AL, AR, CO, GA, ID, IL, IN, KS, MA, MI, MT, NC, NE, PA, SC, VA, and WY; Wisconsin: AR, DE, HI, IA, IL, KS, MA, MD, MI, NE, NH, PA, VA, and WY; Massachusetts: DE, GA, IA, MI, NC, PA, SC, and WY.
9. Institute of Medicine, Coverage Matters: Insurance and Health Care (Washington: National Academies Press, 2001); and R.M. Andersen, "Revisiting the Behavioral Model and Access to Care: Does It Matter?" Journal of Health and Social Behavior 36, no. 1 (1995): 1–10.[CrossRef][Web of Science][Medline]
10. Data comparable to those in Exhibit 2 for the comparison groups are available in the online technical appendix as Exhibit A.2; see Note 5.
11. G. Davidson, L. Blewett, and K.T. Call, "Public Program Crowd-Out of Private Coverage: What Are the Issues?" Research Synthesis Report no. 5, June 2004, http://www.rwjf.org/publications/synthesis/reports_and_briefs/pdf/no5_researchreport.pdf (accessed 3 January 2006).
12. An exhibit similar to Exhibit 3 that summarizes changes in coverage for the comparison groups during the study period is available in the online technical appendix as Exhibit A.3; see Note 5.
13. S. Silow-Carroll and T. Alteras, "Massachusetts: A Case Study in Childless Adult Coverage" (Washington: Henry J. Kaiser Family Foundation, August 2004).
14. Kronick and Gilmer, "Insuring Low-Income Adults."
15. P. Fronstin, "Uninsured Rose in 2002 as Number of Americans with Employment-based Health Benefits Declined," EBRI Notes 24, no. 11 (2003), http://www.ebri.org/pdf/notespdf/1103notes.pdf (accessed 27 December 2005).

                      What's this?

This article has been cited by other articles:

				S. K. Long On The Road To Universal Coverage: Impacts Of Reform In Massachusetts At One Year Health Aff., July 1, 2008; 27(4): w270 - w284. [Abstract] [Full Text] [PDF]

				S. T. Callahan and W. O. Cooper Continuity of Health Insurance Coverage Among Young Adults With Disabilities Pediatrics, June 1, 2007; 119(6): 1175 - 1180. [Abstract] [Full Text] [PDF]