Healthcare spending through the Pandemic: the impact of high-cost medical events on household finances

Box 1: Our Sample

To assess healthcare spending, we use a data asset based on the personal Chase checking accounts of 1.6 million families, from January 2019 through December 2021. Administrative banking data provides a unique, high-frequency lens into consumer finances, with transaction-level views into income and expenditures. For the purposes of our research, the unit of analysis is the primary account holder, whom we subsequently refer to as a family. We focus on families for whom their Chase checking account is likely to be their primary checking account and provide a good window into their financial life. Specifically, we select account holders who had at least five transactions per month across their checking accounts and at least $12,000 in annual income each year. Our income metric represents all non-transfer inflows into the checking account, which includes payroll income, cash and paper checks, social security income, etc. Note that this reflects take-home income, different from gross or pre-tax measures of income reflected in most public data sources.

We assess a family’s total spending by summing checking account outflows via debit card or electronic channels, excluding debt payments and transfers, as well as credit card transactions for families who have a Chase credit card. Given our goal of assessing healthcare spending, it is important to have as complete an understanding of each family’s overall spending patterns as possible. With that in mind, we cap the amount of unknown spending—cash withdrawals, cashback transactions on debit or credit cards, and payments to non-Chase credit cards—for families in our sample. Because we cannot assess where these funds are ultimately spent, we remove families from our sample if unknown spending comprises 20 percent or more of their total spending.

For the families who meet the above criteria, we classify spending transactions to identify healthcare spending. For debit and credit card transactions, we infer the expense category based on the merchant category code; for electronic transactions, we analyze the text description associated with the transaction to infer category. Since we do not have itemized receipt-level information, everything purchased as part of the same transaction is categorized uniformly. This is an easy task for specialized merchants, such as dentists, but more difficult in cases such as drug stores, which sell many different types of goods, not all of which are related to healthcare. For this reason, we chose not to classify drug store spending as healthcare spending. Ultimately, our healthcare spending metric is comprised of five sub-categories: self-explanatory Hospital, Doctor, Dentist, and Vision categories, along with an Other category, which includes spending on chiropractors, medical devices and nursing and home health aides.

It is important to note that our healthcare spending metric measures out-of-pocket healthcare expenditures only. Dollar amounts in this report therefore represent a lower bound of families’ total healthcare spending. Our metric does not include health insurance premiums, or healthcare spending conducted directly via health savings account (HSA) debit cards, nor does it adjust for later HSA reimbursements. This is particularly important to keep in mind when comparing out-of-pocket healthcare spending across groups. If the groups in question have different levels of insurance coverage—higher deductibles, higher copays, fewer services covered free of charge—that will translate to higher healthcare spending by our measurement, even if the underlying services were the same.