Crowdfunding in Health Care Consumption Appendix

Appendix Appendix A: Robustness to Multiple Search Phrase We browse the Go Fund Me website under five available categories, 2 medical (cancer, surgery) and 3 non-medical (education, creative, and competition). We scrape the text of all one hundred campaigns associated with each browse category. After initial word processing on the scraped text using the Natural Language Tool Kit in python, we create word frequency maps to identify the most commonly used words in each category (Figures A1-A5). We select a total of ten search phrases among those frequently used in the search categories. Appendix Table A1: Browse, Search, and Analytical Category Terms Browse Category Cancer Surgery Education Creative Competition Search Phrase Cancer Surgery treatment School student Music Film Art Team Competition Analytical Category Medical Education Competition Table A2: Summary Statistics of Sample of Crowdfunding Campaigns, Removing 25% Low Probability Medical Total Count Percent Multiples Count After Drop Mean Campaign Rate * Std. Dev Mean Sentiment Score Std. Dev Percent Post- Expansion 28,980 .287 26,692 2.851 3.211 -.010 .016 .566 Education Creative Competition Total 17,291 .303 15,531 2.119 4.185 -.028 .017 .571 5,111 .288 4,942 .958 1.844 -.032 .019 .626 7,189 .235 6,754 1.180 1.949 -.027 .020 .598 11,669 .280 53,919 1.946 3.209 -.023 .020 .585 * Campaign Rate constructed as state-year-month campaign count divided by number of nonelderly adult state residents with incomes below the Federal Poverty Level. Table A3: GoFundMe Campaigns by Fundraiser Category, Dropping Lowest Probability Campaigns with Multiple Category Tags Model I Model II Medical Education Creative Post- Expansion Std. Err P-Value State, Year-month FE State covariates Campaign Sentiment Competition Medical Education Creative Competition -.207 .069 .003 -.037 .088 .677 -.076 .070 .276 -.071 .080 .372 -.256 .071 .000 -.013 .080 .873 -.164 .107 .125 -.105 .059 .076 X X X X X X X X X X X X X X X X Appendix Figure A1: Word frequency map for campaigns tagged “Surgery” Appendix Figure A2: Word frequency map for campaigns tagged “Cancer” Appendix Figure A3: Word frequency map for campaigns tagged “Education” Appendix Figure A4: Word frequency map for campaigns tagged “Creative” Appendix Figure A5: Word frequency map for campaigns tagged “Competition” Table B1: Summary Statistics of Sample of Crowdfunding Campaigns, Dropping 1st and 100th percentiles of sentiment scores Medical Education Creative Competition Total Total Count 28,980 17,291 5,111 7,189 11,669 Count After Drop 28,301 17,003 5,029 7,069 57,402 Mean Campaign Rate * 2.991 2.277 .973 1.214 2.044 Std. Dev 3.364 4.389 1.864 2.084 3.368 Mean Sentiment Score -.011 -.027 -.031 -.026 -.022 Std. Dev .014 .015 .018 .017 .018 Percent PostExpansion .565 .571 .627 .599 .585 * Campaign Rate constructed as state-year-month campaign count divided by number of nonelderly adult state residents with incomes below the Federal Poverty Level. Table B2: GoFundMe Campaigns by Fundraiser Category, Dropping 1st and 100th percentiles of sentiment scores Model I Model II Medical Education Creative Post- Expansion Std. Err P-Value State, Year-month FE State covariates Campaign Sentiment Competition Medical Education Creative Competition -.150 .056 .007 -.095 .100 .343 -.095 .067 .156 -.065 .073 .379 -.156 .072 .031 -.131 .096 .175 -.162 .111 .144 -.088 .059 .136 X X X X X X X X X X X X X X X X Appendix C: Restricting to Years 2013 and Later Appendix Table C1: Summary Statistics of Crowdfunding Campaigns, Restricting to Years 2013 and Later Medical Total Count Count After Drop Mean Campaign Rate * Std. Dev Mean Sentiment Score Std. Dev Percent PostExpansion Education Creative Competition Total 28,980 28,924 17,291 17,240 5,111 5,097 7,189 7,167 11,669 58,428 3.095 3.459 2.340 4.539 .987 1.878 1.228 2.088 2.097 3.461 -.010 .016 -.027 .016 -.032 .019 -.027 .019 -.023 .019 .571 .576 .628 .600 .589 * Campaign Rate constructed as state-year-month campaign count divided by number of nonelderly adult state residents with incomes below the Federal Poverty Level. Table C2: Difference-in-Difference Estimates of Medicaid Expansion Effects on GoFundMe Campaigns by Fundraiser Category, Restricting to Years 2013 and Later Model I Model II Medical Education Creative Post- Expansion Std. Err P-Value State, Year-month FE State covariates Campaign Sentiment Competition Medical Education Creative Competition -.166 .061 .006 -.088 .100 .381 -.049 .081 .548 -.071 .072 .326 -.199 .076 .009 -.107 .093 .251 -.162 .104 .118 -.098 .058 .087 X X X X X X X X X X X X X X X X