out, health status, and in-migration patterns,werelyonCPSdatafrom2005through2008.TheMarchsupplementtotheCensusBureau’sCPS has been described as “the survey of record” and “the most viable estimate” of theuninsured.
TheBureauoftheCensusadmin-isters the CPS for the Bureau of LaborStatistics, which scientifically selects the sam-ple to represent the civilian noninstitutionalpopulation. The CPS is the official source fornational health insurance estimates like thewidely cited estimate of 46 million uninsuredU.S.residents.TheCPShasaskedabouthealthinsurancesincethe1980s,andthosequestionshave been largely unchanged since 1994.
Theresponse rate for the March supplement isexceptionally high compared to other volun-tary household-based surveys. The nonre-sponse rate for the health insurance questionsin Massachusetts in 2008 was 16 percent.Nonresponse rates for other surveys measur-ingtheeffectsofChapter58havebeenashighas 55 percent
and 68 percent.
Unlike thosesurveys,theCensusBureauincludesresidenceswithout telephones by virtue of conductinginterviews both by telephone and in person.The CPS data are publicly available from theCensus Bureau.
To our knowledge, ours isthefirststudytoemploydatafromtheMarch2009 supplement to the CPS, which covers allofcalendaryear2008,andthefirsttoexamineMassachusetts two years prior to the mandate(2005–2006) and two years after the mandate(2007–2008).Considerabledifficultiesarisewhenwetry to measure the impact of a complex piece of legislation such as Chapter 58. For example,the outcomes of interest may be influencedby other changes occurring at the same time.ThefactthatthevariouselementsofChapter58 took effect at different times may furthercomplicate the picture.SimilartoLonget.al.,
weemployadiffer-ence-in-differences model to control for many factorsthatmightalsoinfluencetheoutcomesof interest. We compare outcomes in Massa-chusettstothoseofotherNewEnglandstates:Maine, New Hampshire, Vermont, RhodeIsland, and Connecticut. We include controlsforpovertythresholds,maritalstatus,sex,edu-cation,race/ethnicity,andfixedeffectsforstateand year. Our “Chapter 58 effect” is thereforeidentified from the interaction of state andyear. We weight all regressions with the CPSweights, stratify by age group, and estimatemodels without imputed values. We attributeanydifferencesbetweenMassachusettsandtheremaining New England states to the Massa-chusetts law. Our overall results on gains ininsurance coverage are very similar to those of Longetal.WeareunawareofanypublishedestimateofthefullcostofChapter58,includingcoststhat do not appear in government budgets—which is significant in itself. For data on thecosttotheCommonwealthofMassachusettsand the federal government, we rely on esti-mates published by the Massachusetts Tax-payers Foundation.
For estimates of thecosts imposed on the private sector, we rely onpersonalcommunicationswithstafffromthe Massachusetts Taxpayers Foundation.
A primary objective of Chapter 58 is toexpand health insurance coverage in Massa-chusetts, with the goal of universal coverage.In this section, we examine how many Massachusetts residents remain uninsured,and how much of the increase in coveragesince2006canbeattributedtoChapter58.
For 2003 through 2006, the CPS reportedthattheuninsuredrateinMassachusettshov-ered around 10 percent. Massachusetts’ unin-sured rate was low compared to the nationalaverageof15to16percentduringthatperiod.It was especially low relative to southwesternstates,wheretheuninsuredrateoftenexceeds20percent. Various estimates exist of how many Massachusetts residents currently lack healthinsurance. The Commonwealth relies on onesurveythatprovidesanestimateof2.6percent
We are unawareof any publishedestimate of thefull cost of Chapter 58.