key: cord-0056349-j874405l
authors: Asebedo, Sarah D.; Liu, Yi; Gray, Blake; Quadria, Taufiq Hasan
title: How Americans used their COVID‐19 economic impact payments
date: 2021-01-06
journal: nan
DOI: 10.1002/cfp2.1101
sha: bef2ed147574fa8a926781e27a04407e57aee8c6
doc_id: 56349
cord_uid: j874405l

This study investigates how Americans used their CARES Act Economic Impact Payments (EIP) for their spending needs, spending wants, and financial transactions. The results from a sample of 1,172 Amazon MTurk users collected in July 2020 suggest that EIP use varied across spending and financial transaction categories. Those with job instability, less financial resources, and more people to care for received essential support. A smaller proportion of the population spent at least some of their EIP on their wants. Americans' primarily focused their EIP spending on housing, food, and hobbies. In addition, people were able to improve their financial situation through investing and debt reduction. Decisions to save the EIP were related to economic recovery concerns; a broad policy package and public messaging strategy that offers assurance of economic recovery and stability might enhance policy effectiveness for boosting immediate economic growth through EIP spending.

provides financial planners with insight into how people vary in their financial needs, wants, and goals during an economic crisis; and how professionals might offer advice during times of stability to help their clients prepare for future uncertainty. This study builds upon Slemrod (2003, 2009) and Leigh (2012) by investigating economic stimulus payment use within the context of the COVID-19 health pandemic for spending needs, spending wants, and financial transactions; furthermore, this study incorporates detailed subcategories such as saving, investing, debt repayment, housing, food, hobbies, travel, donations, etc. In addition, this study generates unique data about respondents' view of their spending as a need or a want. While researchers have studied economic stimulus payment use during other economic crises (e.g., 2007-2009) , research for COVID-19 economic stimulus spending is only beginning.

Policymakers use economic stimulus payment programs in response to large-scale economic recessions to address two major challenges: individual economic hardship and stimulation of the broader economy (Kaplan & Violante, 2014b) . For example, the Economic Growth and Tax Relief Reconciliation Act of 2001 provided $300 to $600 Federal stimulus payments for U.S. households and reduced the tax rate during the 2001 economic recession. Similarly, policymakers responded to the Great Recession of 2007-2009 with economic stimulus payments under the Economic Stimulus Act (ESA) of 2008 that were designed to increase aggregate consumption and savings, mitigate the effects of the recession on Americans, and invigorate a stagnant economy. The base ESA payments ranged from $300-$600 for single individuals, $600-$1,200 for couples, and an additional $300 per child for households who qualified for the child tax credit. More recently, policymakers signed the CARES Act into law to mobilize economic relief and to stimulate the economy in response to the economic crises resulting from the 2020 COVID-19 health pandemic. According to the CARES Act, a single U.S. citizen (or U.S. resident alien) with an adjusted gross income under $75,000 (phased out up to $99,000) received the full $1,200 economic impact payment (EIP); married couples filing jointly without children with an adjusted gross income below $150,000 (gradually phased out up to $198,000) received a $2,400 EIP; in addition, households who qualified for the child tax credit received $500 per child.

Whether individuals optimize the relief and economic effects of stimulus payments partially depends on how they use the money, which has generated research interest regarding how people respond to economic stimulus payments and tax rebates (Broda & Parker, 2014; Cole, Thompson, & Tufano, 2008; Coronado, Lupton, & Sheiner, 2005; Kaplan & Violante, 2014a , 2014b Misra & Surico, 2014) . For example, Parker, Souleles, Johnson, and McClelland (2013) found that households spent about 12-30% of their stimulus payments (on average) on nondurables (e.g., food, clothing, and gasoline). Similarly, Leigh (2012) found that 40% of households allocated their 2008 ESA payment towards spending, while 24% allocated it towards saving and 35% towards debt reduction. Broda and Parker (2014) found that households increased their spending by about 10% during the week they received their 2008 ESA payments; this result was more pronounced for low-income and low-wealth households, which is consistent with recent research that found economic stimulus payments increased spending (for mainly nondurables) among low-income households (Baker, Farrokhnia, Meyer, Pagel, & Yannelis, 2020; Chetty, Friedman, Hendren, & Stepner, 2020) .

While research is more limited, studies have found that behavioral, economic, and demographic characteristics are associated with economic stimulus payment spending. Shapiro and Slemrod (2009) found that those with positive expectations of the economy (behavioral) were substantially more likely to spend their 2008 ESA payment rather than save. Lower income households compared to higher-income households (economic) tended to use their 2008 ESA payments in ways that produced stimulative economic effects because they spent more of their payments, whereas higher income households were more likely to save and invest (Elmendorf & Furman, 2008; Orszag, 2008) . Furthermore, Shapiro and Slemrod found that older adults (demographic) were more likely to spend more of their 2008 ESA payment. However, Shapiro and Slemrod did not incorporate spending categories; therefore, it is unclear where and in what ways the 2008 ESA payments were spent and how these trends differed by demographic characteristics.

The economic stimulus payment literature centers upon higher level themes, such as saving and spending; therefore, it is necessary to review the broader literature to determine other behavioral, economic, and demographic factors that might affect people's pre-crisis financial preparation that contribute to their stimulus payment decision making.

Planning, investing, and seeking advice are pre-crisis behaviors. For example, people who consult with a financial planner tend to contribute more towards retirement savings and have higher incomes (Martin Jr & Finke, 2014) ; however, those who primarily consult with family are more likely to have a low income (Chang, 2005) . Risk tolerance is related to long-term planning behavior and a greater likelihood of being a saver (Fisher & Anong, 2012) . Risk tolerance is positively correlated with investing in riskier assets, such as holding a greater proportion of stocks in addition to experiencing greater investment returns (Corter & Chen, 2006; Heo, Nobre, Grable, & Ruiz-Menjivar, 2016) .

Stock ownership and stock market performance also influence consumption decisions. People tend to spend more money when stock prices are increasing, with spending on durable goods increasing more than nondurables (Poterba, Samwick, Shleifer, & Shiller, 1995) . Households also spend more on food when stock prices increase but dining out increases at a much higher rate than dining in (Mankiw & Zeldes, 1991) . A positive economic outlook, or the wealth effect caused by increased statement balances, can explain this higher spending. When expectations are low, individuals increase savings (precautionary savings), spend more on needs, and less on wants (Coibion, Gorodnichenko, & Weber, 2020) . The results from Coibion et al. (2020) align with results from Reed and Crawford (2014) , who found that people reduce spending during recessions with the deepest reductions concentrated in travel, transportation, and clothing expenses, while spending on housing, utilities, and food decline by the least (Reed & Crawford, 2014 ).

Lower income and wealth households consumed their 2001 tax rebate at a rate of 10-40 cents per dollar, whereas those with higher income and liquid wealth varied considerably in their consumption patternsspending either $0 or most (Misra & Surico, 2011) . Those who are younger, have few liquid assets, and lower income are more likely to spend money on nondurable goods (e.g., food; Souleles, 1999 Souleles, , 2002 . Those with higher subjective and objective financial literacy are more likely to have emergency savings (Babiarz & Robb, 2014) , less debt (Gathergood, 2012) , and more investments (Xiao, Chen, & Sun, 2015) . Income is also important to saving behavior: Mills and Amick (2010) found that those with higher incomes were savers who were prepared for future income instability. Furthermore, Maroto (2018) found that individuals with a higher net worth were able to save and invest more than those with a lower net worth, and that households with more children saved less, spent more on needs, and spent less on wants. These pre-crisis financial behaviors may affect the use of stimulus payments during a crisis, as liquidity constrained households rely on stimulus payments to provide basic necessities, while others can rely more on savings.

Under the traditional life cycle hypothesis, individuals contend with present and future spending by smoothing consumption through the use of debt and savings to maximize utility over the life cycle (Ando & Modigliani, 1963) . Money is considered fungible within this approach (i.e., without labels or accounts) with each dollar treated the same as another regardless of its source; however, research has shown that spending and saving habits deviate quite markedly from these assumptions. To explain these deviations, Shefrin and Thaler (1988) developed the behavioral life cycle hypothesis (BLCH) introducing that money is nonfungible (i.e., different sources are not treated in the same manner), and that an individual's proclivity to spend is affected by self-control, mental accounting, and framing. Under the BLCH, money is often assigned to one of three mental accounts depending on its source and intended use with a varying temptation to spend (i.e., marginal propensity to consume; MPC): (a) current income (high MPC), (b) current assets (moderate MPC), and (c) future income (low MPC; Shefrin & Thaler, 1988) . For example, money within a retirement savings account (future income or a current asset) is often viewed and spent differently than money from an individual's paycheck (current income). Additionally, an individual's preference for ambiguity affects how they approach decision making with uncertain outcomes, further complicating the decision-making process for spending income and assets (Kahn & Sarin, 1988) .

The BLCH motivates the main hypothesis of this study: individuals seek to maximize utility in the disposition of their CARES Act economic impact payment (EIP) by choosing goods or services with characteristics that match preferences affected by the current economic environment. Due to the combined psychological, economic, and demographic effects contributing to people's COVID-19 experience, it is expected that behavioral elements under the BLCH (such as self-control, framing, and mental accounting) will affect how Americans choose to use their EIP. Furthermore, the CARES Act EIP is considered an unexpected windfall under the BLCH, where a household had limited or no opportunity to adjust prior saving behavior to account for this future income (Shefrin & Thaler, 1988) . Therefore, Americans are likely to frame their EIPs as a current income windfall that carries a higher MPC than that of non-windfall gains (Arkes Hal et al., 1994; Shefrin & Thaler, 1988) . The BLCH provides insight into how people might frame and use their CARES Act EIP, thereby informing these two hypotheses:

H1: Americans are more likely to frame their EIP as a current income windfall and therefore allocate more of their payment towards spending because of a higher MPC.

H2: Americans are less likely to frame their EIP as a current asset or future income and therefore will be less likely to allocate more of their payment towards financial transactions that result in wealth accumulation outcomes (i.e., save, invest, or reduce debt).

This study employed ordered probit models for each dependent variable given the ordinal measurement described below. Marginal effects are calculated to determine the magnitude of the effects on each observed dependent variable. Survey data were generated through a sample of American adults aged 18 and over from Amazon's MTurk platform. This survey method facilitated a prompt data gathering process during the time of EIP receipt with a more diverse population than a traditional convenience sample (Berinsky, Huber, & Lenz, 2011) ; however, it is important to note that the MTurk population is not considered representative of the U.S. population (Goodman, Cryder, & Cheema, 2013) . The survey was developed with Qualtrics and distributed through MTurk under the title "Covid-19 Economic Stimulus Use Survey." The sample consisted of 1,172 survey respondents who indicated they received an EIP, had used at least a portion of it for spending or financial transactions, had at least a 95% HIT (Human Intelligence Task or survey) approval rating, and had completed at least 500 HITs. Respondents were paid $1.50 for completing the survey. About 50% of the sample came from these states: California, Texas, Florida, New York, Illinois, North Carolina, Pennsylvania, and Georgia; the largest proportion of respondents resided in California, Texas, and Florida (27.58%) . Given the use of human subjects, this study was reviewed and approved by the Texas Tech University Human Research Protection Program (IRB2020-508).

There are 26 dependent variables, estimated through 26 separate ordered probit models. Within the survey, respondents first indicated how they allocated their EIP to these broad categories: spending needs, spending wants, and financial transactions. Responses were analyzed as three separate dependent variables through three distinct models. If respondents indicated they used at least a little of their EIP in one or more of these broad categories, they were then asked to indicate how they used their EIP within subcategories. Spending need and spending want subcategories included (a) housing, (b) food, (c) clothing, (d) transportation, (e) travel, (f) donation, (g) children, (h) education, (i) hobby and recreation, and (j) health care. Each of these subcategories (10 for needs, and 10 for wants) were analyzed as separate dependent variables through 20 different models. If respondents indicated they used at least a little of their EIP for financial transactions, they were then asked to indicate how they used their EIP within three subcategories: saving, investing, and debt repayment. Each of these financial transaction subcategories was analyzed as separate dependent variables through three different models. Each dependent variable is measured on a fivepoint Likert scale, indicating the extent to which respondents used their EIP within a particular category: (a) none at all, (b) a little, (c) some, (d) most, or (e) all.

The dependent variables were designed to capture the potential allocation of the EIP across multiple spending and financial transaction categories. Therefore, it is possible for respondents to be included in more than one model. To control for possible response error across multiple categories, respondents were restricted from signifying they spent most or all on more than one category. Additionally, if most was selected in one category, respondents were not allowed to answer some, most, or all for any other category. For example, a respondent could select most for spending needs, none for spending wants, and a little for financial transactions. In response to the subsequent questions for the subcategories, the respondent could indicate they spent all of their spending needs allocation on food, whereas for financial transactions they could indicate they used most of the proportion allocated to financial transactions for debt repayment and a little for savings. In this case, based on skip logic employed through Qualtrics, the survey would not ask the respondent about subcategories for spending wants because they indicated none for that main category. Figure 1 provides a visual depiction of this survey flow.

To capture individuals' monetary needs, decision-making, spending preferences, and financial behaviors during times of crisis, the survey was constructed with variables that encompass behavioral, economic, and demographic factors, including (a) behavioral: who respondents consulted with about EIP decision making, subjective financial literacy, objective financial literacy, financial-risktaking attitude, stock ownership, changes in stock allocation, COVID-19 economic recovery outlook, recession expectations, stock market expectations, and maskwearing and social distancing preferences; (b) economic: workforce status, change in workforce status, EIP amount, net wealth, and income; and (c) demographic: age, gender, race, education, marital status, and the number of household dependents.

Behavioral. The primary consultant (if any) for EIP decision making was measured by asking respondents: "Who was (or will be) the primary person or professional that was (or will be) most influential in your decisionmaking?" There were seven possible responses: (a) financial planner/ advisor, (b) accountant, (c) other professional, (d) family member(s), (e) friend(s), (f) other, and (g) I did not consult with anyone. Subjective financial literacy was measured from 0 (strongly disagree) to 10 (strongly agree) in response to this statement: "I have a high level of overall financial knowledge." Objective financial literacy was measured by the sum of scores for six questions about interest, inflation, bond prices, compound interest, mortgages, and diversification (Lusardi & Mitchell, 2011 , 2017 . The Cronbach's alpha reliability coefficient for objective financial literacy is low at 0.59 in this study's sample; however, this measure is widely used in the literature and the Cronbach's alpha is within the range of other samples: 0.56-0.84 (Murphy, 2013; Sarigül, 2014; Skagerlund, Lind, Strömbäck, Tinghög, & Västfjäll, 2018) . Perceived willingness to take financial risk (financial-risk-taking attitude) was measured on a 0 (lowest risk level) to 10 (highest risk level) scale with this question from Dohmen et al. (2011) : "How willing are you to take risks in financial matters?" Stock ownership was a binary variable (yes/no), with those indicating stock ownership if they replied "yes' to holding stocks individually or through an investment vehicle. Stock position change was measured on a 1 (greatly reduced) to 7 (greatly increased) scale by using this question: "As of today, how much have you changed your position in stocks compared to the beginning of this year?" Respondents who answered less than four were categorized as reduced; respondents who answered more than four were categorized as increased; respondents who answered exactly four were categorized as no change.

The length of an expected U.S. economic recovery from the COVID-19 pandemic was estimated with this question: "How long do you think it will take the U.S. economy to recover from COVID-19 once restrictions are lifted?" Responses were coded as less than 12 months, 1-3 years, and longer than 3 years. Short-term recession expectations were derived from this question: "How likely do you think that the U.S. economy will experience a recession in the next 12 months?" Possible responses consisted of not likely at all, somewhat likely, very likely, and not sure. Short-term stock market expectations were estimated with this question: "What are your expectations that the stock market (or indices such as the Dow Jones or S&P 500) will be higher four weeks from now?" Possible responses included positive, negative, no change, and not sure. Last, respondents' preference for wearing a mask and social distancing was measured with this question: "Do you follow the Centers for Disease Control and Prevention's guidelines for wearing a mask and social distancing?" Responses were yes, no, and prefer not to say.

Economic. Workforce status was coded as 1 for fulland part-time workers, otherwise 0. Change in work force status was measured by comparing current labor force status and labor force status prior to the COVID-19 pandemic, with a change coded as a 1, otherwise 0. Household income was coded with three income categories: less than or equal to $40,000, $40,001-$100,000, and greater than $100,000. Net worth was estimated with four categories: less than or equal to $0, $1-$99,999, $100,000-$249,000, and greater than $250,000. COVID-19 EIP levels were estimated with four categories: less than $500, $500-$1,200, $1,201-$3,400, and more than $3,400. Demographic. Demographic variables included age, gender, race, education, household size, and marital status. Age was measured as a continuous variable. For gender, females were coded as a 1, with males coded as a 0. Whites were coded as a 1, with non-Whites coded as a 0 for race. Respondents' highest level of education was estimated with three categories: high school, bachelor's degree, and graduate degree (master's or PhD). Household size was estimated by the number of household dependents and measured as a continuous variable. Married couples were coded as a 1; otherwise, 0.

Sample descriptive statistics are summarized in Tables 1  and 2. Table 1 shows descriptive statistics for EIP allocation towards spending needs, spending wants, and financial transaction categories and subcategories. Most individuals used some portion of their EIP on spending needs (83%), with 55% spending some portion on wants, and 65% on financial transactions. Greater distinctions are drawn when comparing who used most or all (categorical level 4 or 5) of their EIP in certain areas, with 48% spending most or all on spending needs, but only 6% spending on wants, and 24% on financial transactions. Of those who allocated most or all of their EIP towards financial transactions, 35% used most or all for savings, 21% for debt, and only 6% for investing.

Amongst subcategories for spending needs, 23% spent most or all on housing and 16% spent most or all on food. For spending wants, housing (8%) and food (11%) were also the most dominant areas of spending, with about 7.5% of the sample also using most or all of their EIP for hobby and recreation wants. All other subcategories for spending needs and wants were around 1-2%, suggesting that Americans concentrated their EIP spending primarily on housing, food, and hobbies and recreation (wants only). In summary, most individuals primarily used their EIP for immediate needs, while fewer were able to save and invest for the future or enjoy spending in ways that contribute to life experience and greater well-being (wants).

Subcategory descriptive statistics are provided in Table 2 , organized by each main dependent variable: spending needs, spending wants, and financial transactions. Those who used most or all of their EIP for financial transactions did not consult with others, had higher objective and subjective financial literacy, were more willing to take financial risks (financial-risk-taking attitude), and had higher incidences of stock ownership than the spending need and want samples. A larger share had also reduced or maintained their stock allocation, believed an economic recovery from COVID-19 would take longer than 1 year, believed a recession was not likely at all or very likely in the next 12 months, exhibited a preference for masks and distancing, received higher EIPs, had higher incomes, were older, White, married, and had a graduate degree.

A larger share of respondents who used most or all of their EIP for spending needs experienced a work force change due to the pandemic, were not sure about the possibility of a recession, held a negative outlook on the stock market, received an EIP between $500-$1,200, had lower income and wealth, no college education, were female, non-White, unmarried, and had more household dependents compared to the financial transaction and spending want samples.

A greater proportion of those that used most or all of their EIP for spending wants had consulted with a financial planner, family, and friends. A greater proportion also believed that an economic recovery would take less than 12 months, a recession was somewhat likely, were averse to social distancing and mask guidelines, were employed, received an EIP less than $500, had a positive or neutral outlook on the stock market, had a wealth level above $100,000, were White, male, had a bachelor's degree, and had fewer household dependents. It is important to note that these profiles give a general sense of differences between the samples for each main category; however, they do not indicate statistical differences between them.

The significant variables for each dependent variable are presented in summary tables along with a description of the main findings. See these tables for full results: (a) (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) Recession ( (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) Income $40k-$100k (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) Behavioral (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) Expectations of stock market after 4 weeks (negative) Positive (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5 (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) Behavioral (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5 (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All ( (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) Cut 4 (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) Behavioral (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All ( (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All ( (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) Behavioral (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5 (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5 (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) Prefer not to say ) financial-risk-taking attitude and higher objective financial literacy allocated more of their EIP to financial transactions. Those with a bearish outlook on a U.S. economic recovery from the pandemic also allocated more of their EIP to financial transactions, which suggests that a stimulus payment program accompanied by greater confidence in an economic recovery is important to stimulate consumer spending instead of financial transactions (i.e., saving, investing, or debt reduction). It is surprising that financial planners did not play a role in EIP decision making for financial transactions given their expertise in this area. Instead, consulting with professionals other than financial planners and accountants (compared to no one) was associated with less allocated to financial transactions. Given that those who consulted with financial planners allocated more to spending wants (see Table 12 ) may signal that pre-crisis preparation reduced the need to use EIP funds for investing, saving, and debt repayment. Those with job instability and less financial resources (income and net worth) allocated less of their EIP to financial transactions, which indicates a potential need to take care of immediate expenses over that of financial transactions. Those with smaller EIPs put less towards financial transactions than those with larger EIPs, suggesting respondents treated larger payments as more "wealth-like" (Shefrin & Thaler, 1988) . Last, larger households allocated less of their EIP towards financial transactions while older adults allocated more.

6.1.1 | Saving Table 8 provides a summary of the saving results. A bearish outlook on a U.S. economic recovery was associated with saving more of the EIP, which provides additional evidence for the notion that confidence in the future economy may be important to stimulate consumer spending with an economic stimulus program. Despite financial planners' expertise, they did not matter for saving decisions. Instead, consulting with other professionals (compared to no one) was related to allocating more to saving.

Having less wealth along with more budget constraints via larger households were related to saving less of the EIP, while those with a higher education saved more. These results suggest that saving decisions were primarily associated with economic recovery perceptions along with having more personal resources to support saving behavior. (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) None at all (1) A little (2) Some (3) Most (4) All (5) Cut 1 having a higher financial-risk-taking attitude and an existing stock position that the respondent increased during the pandemic in addition to holding a bachelor's degree.

Older adults and those with lower wealth invested less of their EIP. It is interesting that there were no other behavioral factors outside of investment-related attributes that mattered; even consulting with financial planners did not play a role in investing decisions, which is surprising. Table 10 provides a summary of the debt repayment results. Allocating more of the EIP to debt repayment was primarily associated with economic and demographic attributes: experiencing a job change but still currently working, having a lower net worth, and larger households. Those with more education along with investors who increased their stock position during the pandemic allocated less of their EIP to debt repayment. These results suggest that EIPs helped those with less financial resources, more job instability, and more financial constraints via larger households improve their financial situation through debt repayment. As Americans increased credit card debt at the onset of the pandemic (Karpman, Zuckerman, Gonzalez, & Kenney, 2020) , it is possible that this debt was repaid with EIP funds; however, this study did not discern whether the repaid debt was accumulated prior to or during the pandemic.

6.2 | Spending 6.2.1 | Spending needs Table 11 provides a summary of the spending needs results. Consulting with professionals (other than financial planners and accountants) was connected to spending more EIP dollars on needs. Investors and those with greater objective financial literacy spent less of their EIP on needs. Those that expected stock market stability or who were unsure about stock market performance (compared to a negative expectation) in the near future, along with those who thought a recession was not likely (compared to very likely) in the next 12 months spent less of their EIP on needs. Experiencing a job change during the pandemic, having a low to moderate net worth and income (compared to high net worth and income), having a bachelor's degree (but not a graduate degree, compared to a high school degree), and a larger household mattered to spending more EIP dollars on needs. These combined results suggest that EIPs provided much needed support to those experiencing job instability with less financial resources and more people to care for. Table 12 provides a summary of the spending wants results. Consulting with financial planners, family, and friends (compared to no one) was connected to spending more EIP dollars on wants. This suggests that financial planners, along with family and friends, might play an important role in helping people make spending decisions in areas that enhance their life experience and wellbeing above and beyond basic needs. Those who were more optimistic (somewhat likely compared to very likely) about avoiding a recession in the next 12 months spent more on wants. Taking spending needs and wants recession results together, it appears that respondents with more recession optimism spent less of their EIP on needs; however, that optimism translated to more spending on wants. Those with higher subjective and objective financial literacy, a job change, a spouse, and a lower net worth put less of their EIP towards wants.

This section summarizes the results for each separate spending category model that was estimated according to how respondents viewed their EIP spending. For example, housing was estimated via two separate models predicting spending for (a) housing needs, and (b) housing wants. Results for housing, food, clothing, transportation, travel, donations, children, education, hobby/recreation, and healthcare are outlined below.

Housing. Table 13 provides a summary of the housing results. Older individuals and investors allocated less of their EIP to housing-related expenses regardless of whether they considered them needs or wants. Having a low to moderate income mattered to spending more EIP dollars on housing needs; consulting with professionals (other than financial planners and accountants), having a larger household, and being married mattered to spending more on housing wants. With more work and school conducted at home during the pandemic, it may be that larger households used more of their EIP funds to enhance their home environment, and consulting with a professional was important to this decision. Financial planners did not matter to housing decisions even though housing comprises a significant proportion of the household budget. The remaining results associated with spending less on housing varied across needs and wants: those with higher objective financial literacy (wants), who increased their stock position (needs), were uncertain about near-term stock market performance (wants), and who are employed were connected to less EIP spending on housing.

Food. Table 14 provides a summary of the food results. Consulting with a financial planner or accountant was related to spending more on food wants, whereas spending more on food needs was associated with consulting with family. Larger households and those with a greater financial-risk-taking attitude allocated more of their EIP to food-related expenses regardless of whether they considered them needs or wants, while White individuals allocated less in general. Views on near-term stock market performance mattered, with both uncertainty (wants) and a positive expectation (needs) resulting in less EIP spending for food (compared to a negative expectation). When it comes to the economy, a slightly more optimistic expectation about an impending recession was associated with more EIP spending on food-related needs. Personal economic circumstances mattered for both food-related needs and wants as those who experienced a job change spent more of their EIP on food needs, while those with a low to moderate income (compared to high income) and a lower EIP spent more on food wants. Overall, it appears that EIPs played an important role in facilitating access to food for larger households and those facing job instability, while contributing to an enhanced life experience through foodrelated wants for those with a low to moderate income. Clothing. Table 15 provides a summary of the clothing results. Greater subjective financial literacy (less to needs and wants) and higher financial-risktaking attitude (more to needs and wants) both played a role in spending EIP dollars on clothing in general.

The objective financial literacy (less to clothing needs) and stock ownership (more to clothing needs) results aligned with the results of their subjective counterparts outlined above. It is interesting that the subjective measures were more robust in predicting both needs and wants than the objective measures for financial literacy and financial-risk-taking attitude. Consulting with family and friends were important to spending more EIP funds for clothing needs. A slightly more optimistic view of an impending recession and having less wealth were connected to more spending on clothing-related needs, while uncertainty about a recession was linked to less. Positive near-term stock market performance expectations were important to spending less for clothing wants. Last, being older (wants), male (wants), and married (needs) were connected to less EIP spending on clothing.

Transportation. Table 16 provides a summary of the transportation results. Consulting with accountants was related to more EIP spending on transportation wants, whereas consulting with friends was connected to spending more for transportation needs. These results provide additional information about how economic and stock market perceptions played a role in decision making. Uncertainty about an impending recession (needs and wants) and uncertainty regarding a near-term change in the stock market (wants) were both connected to less spending for transportation expenses. However, having some sense of certainty about stock market stability (no change compared to negative) along with a little optimism about an impending recession (somewhat likely compared to very likely) were connected to allocating more EIP dollars to transportation expenses. Mask and distancing preferences emerged with those in favor of mask and distancing guidelines using less of their EIP for transportation wants than those not in favor of these guidelines. This result is concerning when combined with the finding that older individuals (an at-risk population for COVID-19) allocated more of their EIP to transportation wants despite the possible increase in health risk inherent in transportation. Last, those with low to moderate wealth levels (compared to higher wealth) used more of their EIP for transportation needs and wants; on the other hand, those with lower EIPs (indicative of those with higher income or smaller households) spent less on transportation needs, consistent with the wealth results.

Travel. Table 17 provides a summary of the travel results. Consulting with financial planners, accountants, and friends were related to more EIP spending on travel needs. Investors who increased their stock position during the pandemic, and those with a greater financial-risktaking attitude allocated more of their EIP to travel needs. The notion that optimism about the economy and an impending recession is connected to increased spending came through for travel needs and wants, similar to other areas: Those believing a recession is somewhat likely (compared to very likely) spent more on travel needs, while those expecting a long economic recovery (compared to <12 months) spent less on travel wants. Similar to transportation expenses, those in favor of mask and distancing guidelines were more likely to spend less of their EIP on travel needs, while those who preferred not to disclose whether they were in favor of masks and distancing guidelines spent more of their EIP on travel wants. Last, those with higher education (wants) along with larger households (needs) spent more on travel; men, older individuals, and those with less wealth spent less on travel wants.

Donations. Table 18 provides a summary of the donation results. Consulting with financial planners (compared to no one) mattered to allocating more EIP dollars to donation needs and wants. Friends were connected to donating more for needs, whereas anyone else in an individual's social network was related to donating more for wants. Additionally, consulting with family was associated only with a decreased likelihood of spending none of the EIP for donation needs (see Table 5 ). These combined results suggest that individuals rely significantly on their social network for donation decisions, and within this network financial planners were connected more broadly to donations considered both needs and wants. Recession and economic perceptions mattered to donation decisions. Those that were more optimistic (not sure and not likely, compared to very likely) about avoiding an impending recession allocated less of their EIP to donations for needs and wants. This result contradicts other spending results where more optimism about recession likelihood generally resulted in more spending. However, similar to spending results, those that thought the economy would take longer to recover from the pandemic (more bearish) allocated less of their EIP to donation wants. Mask and distancing preferences were important to EIP donation decision making with those in favor of mask and distancing guidelines (compared to not in favor) allocating less of their EIP to donation wants, while those that preferred not to say allocated more to donation needs. Last, married households, those with low to moderate wealth, and who experienced a job change donated less of their EIP, while older adults and larger households donated more. Children. Table 19 provides a summary of the children results. The results that mattered to allocating EIPs across child-related spending needs and wants were consulting with family (more), larger households (more), being married (more), greater objective financial literacy (less), making no change to existing stock positions (less), and expecting an extended economic recovery from the pandemic (less). Once again, a bearish outlook on economic recovery was connected to less EIP spending. Last, those with a lower income used more of their EIP for child-related wants, whereas those with low to moderate wealth spent less on childrelated needs than those with greater wealth.

Education. Table 20 provides a summary of the education results. Consulting with others (financial planners, accountants, other professionals, and family) was important to spending more EIP funds for a combination of education needs and wants, with friends important to just needs and other professionals important to just wants. The broad impact of financial planners across education needs and wants might be because education planning is a fundamental component of financial planning. Similar to other results, a belief in a longer economic recovery from the pandemic was linked to less spending for both education needs and wants. Additionally, uncertainty about near-term stock market performance was connected to less EIP spending for education wants, while optimism about avoiding an impending recession was connected to more EIP spending for education needs. Mask and distancing preferences emerged for spending EIP dollars for education, similar to other socially oriented categories noted above (i.e., transportation, travel, and donations): Those in favor of masks and distancing guidelines (compared to those not in favor) spent less of their EIP on education wants; those that preferred not to disclose their preference spent more on education needs. Last, more EIP education spending was associated with larger households (wants and needs), higher education (wants), and job instability (needs), whereas less wealth (wants) and low to moderate income (needs) were related to spending less. Hobby and Recreation. Table 21 provides a summary of the hobby and recreation results. Family (needs) and friends (wants) were important to allocating more EIP funds towards hobbies and recreation expenses but not professionals of any kind. This result could potentially be due to the natural involvement of friends and family in hobby and recreation activities. A bullish nearterm outlook on the stock market was associated with more spending on hobby and recreation wants. This result provides more evidence to a theme that emerged in this study: a sense of optimism and confidence is connected to more spending. However, the recession expectation results were mixed. Consistent with other results, uncertainty about an impending recession (compared to very likely) was connected to less EIP spending for hobby and recreation expense needs. Contrary to other results, a slightly more optimistic outlook (recession somewhat likely compared to very likely) was connected to less spending for wants, not more. It may be that hobby and recreation expenses feel more discretionary, even when considered a need, and therefore a stronger sense of confidence and optimism in the economy is required to spend more in this area. Consistent with other socially oriented spending categories (transportation, travel, donations, and education), mask and distancing preferences mattered with a preference not to disclose connected to more EIP spending for hobby and recreation needs. Last, less EIP spending for hobby and recreation expenses were associated with job instability (needs), low net worth (needs), higher education (wants and needs), and being married (wants and needs), while being White (needs) and having a higher financial-risk-taking attitude (needs) were associated with spending more.

Health Care. Table 22 provides a summary of the health care results. Consulting with others was important to EIP spending for health care, with family associated with spending more on health care needs while consulting with financial planners or accountants were important for spending more on health care wants. These results are similar to other results that signal a trend where professionals appear to matter for spending on wants, whereas family and friends were more prevalent for spending on needs. Results for economic and stock market outlook generally supported the notion that uncertainty and a bearish outlook on a U.S. economic recovery from the pandemic were connected to less spending. Last, there was a consistent relationship across health care needs and wants for wealth, as those with lower wealth spent less of their EIP on health care compared to those with greater wealth, while those with a lower EIP (indicative of higher income or a smaller household) spent more on healthcare wants (consistent with the wealth result). Last, older (wants) and more highly educated individuals (needs) spent more of their EIP on health care.

These results should be interpreted within the context of their limitations. First, the sample was predominantly White, married, and more educated when compared to the U.S. population. Because the findings resulted from a nonprobability crowdsourcing sample (MTurk), the results are not generalizable to the broader U.S. population. Future studies might use nationally representative samples to capture how people use economic stimulus payments for saving, spending, and other financial decisions. Additionally, this study only estimates the short-term use of one economic stimulus payment and there could be a second stimulus check at some point. Therefore, it is likely that the total effect of all stimulus payments will vary from the effects of the first payments alone and additional research will be needed to investigate these combined effects. Furthermore, this study was conducted soon after EIPs were issued. Therefore, this study had to treat the actual and planned usage of the EIP in the same way. How individuals were planning to use the money while completing the survey might have been significantly different from how they actually used the money. Also, this study does not investigate the change in expenditures before and after the COVID-19 pandemic, such as food, housing, clothing, etc. Furthermore, Americans experienced different modes and times of receipt of their EIP. With the dynamic COVID-19 environment, it is possible that this mode and time separation could have affected how people decided to allocate their payment; however, this aspect was not captured within the study. A future study could analyze longitudinal effects of economic stimulus payments on individuals that covers the entire period of economic instability. Moreover, due to concerns of survey fatigue, the survey included a one-item measure for risk attitude from Dohmen et al. (2011) and did not incorporate a multi-item construct for risk tolerance. A future study could use a robust measure of risk tolerance while accounting for the potential change in risk tolerance that can occur during times of significant crises and stress (Heo, Grable, & Rabbani, 2020) . Last, Cronbach's alpha for objective financial literacy was low within this sample. Even though it was constructed according to Lusardi and Mitchell's (2017) financial literacy measure, caution should be exercised in interpreting the results as the reliability is suspect.

The combined descriptive and regression results provide partial support for hypothesis one: due to behavioral factors under the BLCH, people that tend to frame their EIP as a windfall are more likely to spend more of it on needs or wants (H1). Additionally, the results provide partial support for hypothesis two: people allocate smaller portions of their EIP towards financial transactions because of the high marginal propensity to consume windfall income as suggested by the BLCH (H2). The descriptive results partially support these hypotheses because most individuals (83%) allocated some portion of their EIP towards spending needs, whereas less (65%) used some portion for financial transactions. However, an even smaller proportion (55%) used a little to all of their EIP for spending wants. When comparing how participants used the majority of their EIP, 6% spent most or all on spending wants, while about four times as many used most or all on financial transactions (24%), and about eight times as many participants spent most or all on spending needs (48%).

These combined descriptive results suggest that a majority of participants used their EIP for spending at some level; however, almost a quarter of the sample still used a significant portion of their EIP for financial transactions that included saving, investing, and debt reduction. These descriptive results suggest that windfall framing and a larger marginal propensity to consume EIP income was present but varied across respondents, which is consistent with Leigh (2012) , and could be partially due to larger payments feeling more wealth-like with a lower marginal propensity to consume (Shefrin & Thaler, 1988) . This study generated partial evidence for this wealth effect as smaller EIPs were associated with less allocated towards financial transactions than larger ones. The regression analyses facilitate the ability to control for behavioral, economic, and demographic characteristics to determine underlying attributes associated with variation in how people allocated their EIP across different categories. The results generated from the ordered probit models suggest that behavioral, economic, and demographic characteristics offer a partial explanation for this variability. Broader themes that emerged from these results are discussed next.

Consulting with others. Approximately half of the sample did not consult with anyone when making their decision for how to allocate their EIP, which indicates a significant proportion of participants preferred to make the decision on their own. When individuals decided to consult with someone, they leaned on friends, family, accountants, professionals, and financial planners. Consulting with a financial planner was significantly related to spending a larger EIP portion for donations, education, healthcare, and travel; those who consulted with a financial planner also spent less on food needs and more on food wants. This spending pattern is consistent with spending during an economic boom, where dining out and travel increases while spending on eating at home decreases (Reed & Crawford, 2014) . This result suggests financial planners had a broad impact in ways that denote thriving-not just surviving-amidst a significant and extended health and economic crisis. That being said, it is surprising that financial planners did not play a role in saving, investing, and debt reduction decisions, which are traditional financial planning topics. The only professional relationship that mattered for these financial transactions was with other professionals (other than financial planners and accountants) for financial transactions in general and more specifically, for saving decisions. Therefore, financial planners have an opportunity to expand their impact to other spending areas and to increase their impact within traditional financial transactions (saving, investing, and debt reduction) when clients receive an EIP. Additionally, a theme emerged that suggested consulting with a nonformal network of family and friends was related to increased EIP spending for needs, whereas consulting with a more formal network of professionals (accountants, financial planners, etc.) was more frequently associated with increased spending for wants. This result requires additional research as it varied to some extent across spending categories but could reflect that the population of clients using professional services have more discretionary cash flow to support spending wants. This is supported by existing research that suggests those who consult with a financial planner have a higher income, while those who primarily consult with family have a lower income (Chang, 2005; Martin Jr & Finke, 2014) . Therefore, there may be a need to expand professional services and financial advice to those focused more on their needs. to fears about an extended economic recovery along with having more personal resources (e.g., higher education and wealth) to support saving behavior. Mask and Distancing Preferences. Masks and distancing preferences consistently predicted EIP spending in areas that typically involve some level of human interaction or social outreach. Those that were in favor of following mask and distancing guidelines (compared to those not in favor) spent less on transportation, travel, donations, education, and hobby and recreation. These spending areas are also in sectors that have struggled amidst the pandemic. A curious result was that older individuals (a greater health-risk population) allocated more of their EIP to transportation wants despite the possible increase in health risk inherent in transportation-related activities.

Overall, EIPs provided much needed support to those experiencing job instability with less financial resources and more people to care for. The results suggest that EIPs helped those who experienced a job change during the pandemic because these individuals generally allocated a greater proportion of their payment to their needs and less to their wants or financial transactions. However, this finding also suggests that those who experienced a job change during the pandemic may need even more help during and post-pandemic to continue to make ends meet and to eventually recover and rebuild their savings. Additionally, the results suggest that those with lower wealth are less able to save or put money towards goods and services that enhance their quality of life as compared to those with greater financial resources. For example, EIPs played an important role in providing for spending needs for larger and less wealthy households in addition to those facing job instability, while contributing to an enhanced life experience through spending wants for those with greater wealth. Thus, individuals with greater financial resources and a stable job had more flexibility to use their EIP in ways that contributed to their quality of life and well-being above and beyond basic needs. Those that suffered a job loss and/or change with less financial resources (or drained financial resources) may need additional financial and psychological support to recover both during and following periods of economic crises and uncertainty.

The purpose of this study is to investigate how individuals chose to allocate their economic impact payments (EIP) provided through the CARES Act towards their spending needs, spending wants, or their financial situation through financial transactions (i.e., saving, investing, and debt reduction). Additionally, this study aims to discern how individuals allocated their EIP within a variety of sub-categories, such as housing, transportation, travel, education, hobbies, and charitable donations. This study employed the Behavioral Life Cycle Hypothesis (BLCH) to analyze allocation decisions. Overall, behavioral characteristics did play a role in how individuals framed and used their EIPs as expected for windfall income as outlined within the BLCH; however, individuals varied significantly in their decision to put this money towards their needs, wants, and financial situation. The combined results generated useful and interesting themes as it relates to individual decision-making for economic stimulus payment use. Overall, there are two key takeaways that are particularly relevant for policymakers and financial planners.

First, as noted in the discussion section, people's outlook on the economy in addition to the market were consistently related to EIP spending and saving decisions, with optimism and certainty generally related to increased spending. If the goal of an economic stimulus payment program is to stimulate consumer spending in the short run to boost the economy, then it would behoove policymakers to consider a broad policy package and public messaging strategy that offers assurance of economic recovery and stability in order to boost immediate economic growth.

Second, financial planner use within the sample was higher than other professionals; however, the impact of this use was much more limited for EIP decision making for financial transactions-an area in which financial planners have specific expertise. It may be that this lack of a relationship is a natural consequence of adequate pre-crisis planning so those that used financial planners did not feel the need to put their EIP towards financial transactions in the first place. Instead, the results showed that those consulting with financial planners generally spent more on wants in areas such as travel, donations, food, and health care. Yet, it is interesting that financial knowledge and financial planner use did not play a role for financial transaction behaviors; what mattered most from a behavioral perspective was perceived willingness to take financial risk (financial-risk-taking attitude), existing stock ownership, and perspective on the economy and stock market. In fact, behavioral attributes were the most robust in predicting allocation of stimulus money in all areas, which underscores the relevance of client psychology within financial planning (Chaffin, 2018) . These findings suggest that financial planners played a role in clients' EIP spending in areas that helped them thrive through the pandemic, yet an opportunity exists to enhance financial planners' relevance for financial transaction decisions and potentially for pro-bono services to expand the reach of the profession in times of both crises and stability.

Sarah D. Asebedo https://orcid.org/0000-0002-8680-5031 Yi Liu https://orcid.org/0000-0001-5683-0538

The "life cycle" hypothesis of saving: Aggregate implications and tests. The American Economic Review

The psychology of windfall gains

Financial literacy and emergency saving

Income, liquidity, and the consumption response to the 2020 economic stimulus payments (No. w27097)

Using mechanical Turk as a subject recruitment tool for experimental research

The economic stimulus payments of 2008 and the aggregate demand for consumption

What coronavirus could mean for the global economy

Client psychology

With a little help from my friends (and my financial planner)

How did covid-19 and stabilization policies affect spending and employment? A new real-time economic tracker based on private sector data (No. w27431

The cost of the covid-19 crisis: Lockdowns, macroeconomic expectations, and consumer spending (No. w27141

Where does it go? Spending by the financially constrained. Spending by the Financially Constrained

The household spending response to the 2003 tax cut: Evidence from survey data

Do investment risk tolerance attitudes predict portfolio risk?

Individual risk attitudes: Measurement, determinants, and behavioral consequences

If, when, how: A primer on fiscal stimulus, The Hamilton Project, The Brookings Institution

Relationship of saving motives to saving habits

Self-control, financial literacy and consumer over-indebtedness

Data collection in a flat world: The strengths and weaknesses of mechanical Turk samples

A test of the association between the initial surge in COVID-19 cases and subsequent changes in financial risk tolerance

What role does financial risk tolerance play in mediating investing behavior

Modeling ambiguity in decisions under uncertainty

A model of the consumption response to fiscal stimulus payments

A tale of two stimulus payments

The COVID-19 pandemic is straining Families' abilities to afford basic needs

How much did the 2009 Australian fiscal stimulus boost demand? Evidence from household-reported spending effects

Financial literacy and planning: Implications for retirement wellbeing (No. w17078)

How ordinary consumers make complex economic decisions: Financial literacy and retirement readiness

The consumption of stockholders and nonstockholders

Saving, sharing, or spending? The wealth consequences of raising children

A comparison of retirement strategies and financial planner value

Can savings help overcome income instability?

Heterogeneous responses and aggregate impact of the 2001 income tax rebates, discussion paper 8306

Consumption, income changes, and heterogeneity: Evidence from two fiscal stimulus programs

Psychosocial factors and financial literacy

Options for responding to short-term economic weakness. Testimony, Senate Finance Committee

Consumer spending and the economic stimulus payments of

Stock ownership patterns, stock market fluctuations, and consumption

How does consumer spending change during boom, recession, and recovery

A survey of financial literacy among university students

Consumer response to tax rebates

Did the 2008 tax rebates stimulate spending?

The behavioral life-cycle hypothesis

Financial literacy and the role of numeracy-How individuals' attitude and affinity with numbers influence financial literacy

The response of household consumption to income tax refunds

Consumer response to the Reagan tax cuts

Bureau of Labor Statistics, Civilian unemployment rate

Age differences in consumer financial capability

How Americans used their COVID-19 economic impact payments

Optimism and Certainty. Within the results, a theme emerged that suggested a positive outlook regarding a U.S. economic recovery (faster recovery), optimism about avoiding a recession, and more certainty about near-term stock market performance were significantly related to participants spending more of their EIP. These results are consistent with expectations that pandemic induced drops in consumer confidence would lead to a reduction in consumer spending (Carlsson-Szlezak, Reeves, & Swartz, 2020) , in addition to results from Shapiro and Slemrod (2009) who found that positive expectations of the economy were related to increased 2008 ESA spending over that of saving. These combined results suggest that if the goal of an EIP program is to stimulate the economy with enhanced consumer spending in the short run, mere injection of money may not be enough. EIPs accompanied by assurances of an economic recovery and future stability may increase consumer confidence and boost immediate economic growth through increased consumer spending.Wealth Accumulation Behavior Persistence. It appears that attributes associated with wealth accumulation behaviors (such as a greater financial-risk-taking attitude, wealth level, saving, investing, and reducing debt) persist in the midst of crisis and uncertainty. For example, the results suggest that investors used their EIP to continue to improve their financial situation and build additional wealth. EIPs helped those with less resources, more job instability, and more financial constraints via larger households improve their financial situation through debt repayment. Saving behavior persisted, but functioned a little differently, as it appeared to be more closely related