Stockpiling in the time of COVID-19

Abstract

Objectives.

Consistent with behaviour observed in prior crises, individuals are stockpiling supplies during the novel coronavirus (COVID-19) pandemic. The goal of this study was to describe stockpiling behaviour in response to COVID-19 and investigate individual predictors of stockpiling.

Methods.

Workers (N = 363, 54.72% male, 44.65% female, 0.63% other; M_age = 38.41, SD = 12.48, range = 18-78) were recruited from Amazon’s Mechanical Turk and completed a survey about their stockpiling of 13 items, as well as behaviours and opinions surrounding the COVID-19 pandemic and their political affiliation.

Results.

Participants stockpiled, on average, approximately 6 items, and toilet paper was the item most commonly procured. Approximately 25% of the sample acquired a gun or other weapon in response to the pandemic and approximately 20% of participants stockpiled gold or other precious metals. Stockpiling was more commonly observed among individuals who were more conservative, worried more about the pandemic, and social distanced less.

Conclusions.

Individual, societal, and ideological implications are discussed.

Background

The novel coronavirus (COVID-19) was declared a pandemic on 11 March 2020 (World Health Organization, 2020) and a United States (US) national emergency on 13 March 2020. In response, most state and federal public health officials instituted social distancing, contract tracing, mask-wearing, and hand-washing recommendations in an attempt to slow the spread of the virus, although responses differed by region and implementation timelines varied. Other messaging from state and federal officials about mask-wearing, gathering in groups, and stockpiling food and other essential items has been conflicting. For example, state health departments in Hawaii and Minnesota urged residents to purchase non-perishable food, prescription medication, and sanitary supplies, while the director of Centers of Disease Control and Prevention (CDC) discouraged stockpiling of supplies (Brooks & Hay, 2020 February 28).

Despite contradictory messaging, consistent with prior crises and infectious disease outbreaks, COVID-19 has led to stockpiling (i.e., procuring real or perceived emergency supplies) and panic buying which can result in hoarding (i.e., procuring and/or storing excessive supplies). During natural disasters, gasoline supply disruptions have caused some retail service stations to run out of gasoline, including during the aftermath of Superstorm Sandy on the East Coast of the United States in 2012 (Sterman & Dogan, 2015). There has been similar panic buying among US residents amid winter storm and hurricane seasons (Kulemeka, 2010), and fear of avian (bird) flu triggered hoarding of the antiviral drug oseltamivir (Tamiflu) (Pollack, 2005).

Stockpiling and hoarding stem from a human’s response, either rationally or emotionally, to scarcity. Scarcity may cause stress, anxiety, and fear or panic, leading people to build private stocks or place orders for more than they need (Sterman & Dogan, 2015), potentially in an attempt to regulate their distress (Rajkumar, 2020). Such psychological uncertainties may stimulate behavioural phenomena like hoarding, sales postponement (i.e., a business strategy which maximizes possible benefit and minimizes risk by delaying further investment by the manufacturer into a product or service until the last possible moment), and rife stockpiling in food markets to prepare for future uncertainty (Headey, 2011). In addition to scarcity, the experience of COVID-19 itself is linked to psychological distress (Qiu et al., 2020), which may confer additional motivation to stockpile or panic buy supplies. This is substantiated theoretically by both commodity theory (Brock, 1968) and prospect theory (Tversky & Kanneman, 1992) which link how scarcity and risk aversion, respectively, may explain why consumers stockpile or hoard essential items. Regarding the latter, if perceived future inability to obtain necessary goods is judged as risky, risk aversion could motivate hoarding as a means of protecting oneself from risky events, even if the likelihood of occurrence is low. Therefore, stocking up on supplies is, to some extent, human nature, and should not be regarded as entirely irrational behaviour. However, as a social behaviour, stockpiling could be harmful by disrupting supply chains and causing shortages for others.

The goal of this study was to describe stockpiling in response to COVID-19 and identify individual predictors of stockpiling, including political affiliation, COVID-19 worry, and social distancing. We chose to examine political affiliation given the polarizing landscape of US politics. Indeed, the threat of COVID-19 is viewed differently by conservatives versus liberals (Conway et al., 2020) and the types of COVID-19-related Twitter messages among US congress people varied according to their political party (Green et al., 2020), which suggests that citizens from predominantly Democratic regions heard different messages than those from predominantly Republican regions. Another factor we explored was COVID-19 worry. Concern surrounding the pandemic is ubiquitous, and one study using a UK sample suggested that worry was the most common negative emotion (Kleinberg et al., 2020). Finally, we evaluated the association between social distancing and stockpiling as there is evidence that political beliefs impact compliance to social distancing orders and engage in in-person purchasing (Painter & Qiu, 2020).

Methods

Participants and procedures

Workers (N = 361, 54.72% male, 44.65% female, 0.63% other; M_age = 38.41, SD = 12.48, range = 18-78) were recruited from Amazon’s Mechanical Turk (MTurk). MTurk matches ‘workers’ (i.e., participants) with ‘requestors’ (i.e., researchers) via the Internet. Workers browse titles, descriptions, and compensation associated with researcher-generated tasks and participate in those that are of interest to them (see Paolacci & Chandler, 2014 for more details on the MTurk method). Participants consented and completed a survey about their behaviours and opinions surrounding the COVID-19 pandemic and their political affiliation between 8 April 2020 at 4 P.M. EST and 10 April 2020 at 2 A.M EST. Participants were compensated $1.00 USD.

The sample was primarily White/Caucasian (79.87%), followed by Black or African American (11.64%), Asian (3.77%), more than one race (2.52%), and American Indian, Alaskan Native, Hawaiian Native, or Other Pacific Islander (1.89%). Sixty-seven per cent of the sample held a Bachelor’s degree or higher. Annual income ranged from less than $10,000 USD to more than $260,000 USD with 50% of participant incomes falling between $40,000 USD and $100,000 USD. Forty-nine per cent of the sample voted for Donald Trump in the 2016 election and 58% voted for Barack Obama in 2012. The average number of individuals residing in participant homes was 3.47 (range 1-8+). Regarding infection, the majority of participants had not had COVID-19, while 5.36% of participants ‘definitely’ had COVID-19, 15.14% ‘probably’ did, and 18.61% were unsure. This study was deemed exempt based on category 2 of the Revised Common Rule by the Institutional Review Board at Brown University.

Measures

Demographics

Participants indicated their race, highest education level, annual household income, and number of people in their household. Additionally, participants were asked: ‘Do you think you’ve had Coronavirus (COVID-19)?’ with response options ranging from 1 (definitely not) to 5 (definitely yes).

Stockpiling

A sum score (ranging from 0 to 13) was calculated by summing the number of items (response options were no/yes) that participants indicated they ‘stocked up on’ as a result of the COVID-19 pandemic from the following options: toilet paper, rice, pasta, bread, canned goods, water, medicine, guns and other weapons, gold and other precious metals, cash, wood, gasoline, alcohol. Cronbach’s alpha (α) for the sum score was .88.

Political affiliation

Participants reported on their political affiliation on a scale from 1 (very much democrat) to 9 (very much republican), consistent with Cohen (2003).

COVID-19 worry

Participants were asked: ‘How worried are you about each of the following events as a result of Coronavirus (COVID-19)’ on a scale of 1 (not at all) to 5 (a great deal): I will not have access to food, I will not be able to retire as planned, police will not protect me in the event of major social unrest, I will become very ill or die, someone close to me will become very ill or die, my medical care for non-Coronavirus matters will be negatively impacted, my mental health (e.g., depression, anxiety) will be negatively impacted, I will be not able to pay my rent or mortgage, and I will need to go on, or increase my level of, government assistance. A mean score was calculated across the 9 items (α = .90).

Social distancing

The degree to which participants think they engaged in social distancing was measured with one item: ‘How much social distancing would you say you engaged in as a result of Coronavirus (COVID-19)’ with response options on a 5-point scale.

Data analysis

Data for participants with survey completion times less than or equal to 5 minutes (n = 45) were removed, resulting in an analytic sample of 318. Following an evaluation of the descriptive characteristics of the sample, correlational analyses were conducted to examine the associations among study variables. Finally, a hierarchical multiple regression was conducted, such that block 1 included demographic covariates (income, education, age, gender, and number of individuals in the home) and focal variables were entered in block 2 (political affiliation, COVID-19 worry, social distancing). Change in R² from block 1 to block 2 represents how much additional variance in stockpiling the focal predictors explains beyond the covariates entered in block 1.

Results

Table 1 presents descriptive statistics and correlation coefficients. The item that individuals most commonly stockpiled was toilet paper (63.21%), followed by canned goods (59.18%), rice (57.41%), bottled water (56.96%), pasta (56.19%), bread (53%), medicine (52.7%), cash (45.89%), alcohol (37.7%), gasoline (35.96%), firewood (25.8%), guns or other weapons (24.52%), and gold or other precious metals (20.25%). On average, participants stockpiled approximately 6 items (M = 5.83, SD = 3.97). All variables, with the exception of income, were significantly associated with stockpiling. There was a statistically significant difference in stockpiling between males and females (F(1, 297) = 5.32, p = .022), such that males (M = 6.32 SD = 4.02) stockpiled more than females (M = 5.26 SD = 3.89).

Table 1.

Descriptive statistics and bivariate correlations among study variables

	1	2	3	4	5	6	7
1. Stockpiling sum	5.83 (3.97)
	301
2. Affiliation	.24**	5.19 (3.13)
	301	318
3. COVID-19	.54**	.07	2.83 (1.03)
	301	318	318
4. Social distancing	−.13*	.03	.04	4.15 (0.89)
	301	317	317	317
5. Age	−.16**	.003	−.16**	.07	38.41 (12.48)
	283	300	300	299	300
6. Individuals in home	.45**	.23**	.44**	−.09	−.18**	3.47 (1.85)
	300	315	315	314	297	315
7. Income†	.05	.09	−.05	−.02	.11	.15**	6.29 (3.30)
	295	309	309	308	291	308	309

Note. Means (standard deviations) and ns are on the diagonal. Bivariate correlations and ns are on the off-diagonal.

^**p < .01

^*p < .05

^†Income was coded in $10,000 USD increments; the minimum was 1 = <$10,000 USD, and the maximum was 19 = >$260,000 USD.

Results from the hierarchical multiple regression model are reported in Table 2. In step 1, having obtained a Bachelor’s degree or more education and number of individuals in the home were associated with more stockpiling. With the focal variables added to the model in step 2, the education effect was fully attenuated (step 1 R² = .22; ΔR² = .18 for step 2 p < .001). Increased stockpiling was observed among those who were more conservative, worried more about the pandemic, had more people in the home, and socially distanced less. Although inversely correlated, age did not significantly predict stockpiling.

Table 2.

Hierarchical multiple regression results with stockpiling as the dependent variable (N = 274)

				95% CI
Effect	b	SE	β	LL	UL	p
Step 1
Constant	5.106	1.14		2.86	7/35	<.001
Education	1.170	.549	.119	.089	2.25	.034
Income	−.086	.068	−.071	−.219	.047	.205
Age	−.026	.017	−.083	−.061	.008	.135
People	.894	.123	.410	.653	1.135	<.001
Sex	−.772	.432	−.097	−1.622	.078	.075
Step 2
Constant	1.96	1.43		−.849	4.77	.171
Education	.074	.504	.007	−.919	1.067	.884
Income	.010	.061	.008	−.110	.130	.872
Age	−.011	.016	−.035	−.042	.020	.478
People	.417	.121	.191	.178	.656	.001
Sex	−.543	.384	−.068	−1.30	.214	.159
Political Affiliation	.185	.063	.146	.061	.309	.004
COVID-19 worry	1.78	.216	.457	1.36	2.21	<.001
Social distancing	−.574	.220	−.127	−1.01	−.141	.010

Note. CI = confidence interval; LL = lower limit; People = number of individuals living in the home; UL = upper limit.

R² = .22 for step 1: ΔR² = .18 for step 2 (p < .001).

Discussion

Stockpiling has been widely observed during COVID-19. This study sought to describe stockpiling in response to COVID-19 and to identify individual predictors of this behaviour. Stockpiling was common among this sample and more commonly observed among individuals with more COVID-related worry and who social distanced less. It was also observed more among conservative participants, as measured by a one-item scale from very liberal to very conservative.

Toilet paper was the item that individuals most commonly stockpiled. This is not surprising given the unprecedented media attention to the stockpiling of toilet paper, in particular. Scarce items increase in value when they are newly scarce (Cialdini, 2009). In other words, people value items that have recently become restricted more than those that have been restricted all along, which could help explain the stockpiling of toilet paper and other essentials during COVID-19. Competing with others for scarce resources only increases our perceived need for them, creating a potential treadmill effect, and it may be that the influential power of scarcity that results in stockpiling can arise from the emotionarousing quality that makes rational thinking difficult (Cialdini, 2009).

Nearly a quarter of the sample purchased guns and other weapons. It is possible that stockpiling weapons, as well as gold and other precious metals (20% of the sample reported this), may reflect preparation for anarchy or fear that COVID-19 will impact the ability of police agencies to protect their communities, but this should be explored further. The latter possibility is plausible given that we found a positive association between COVID-19 worry – which included an item about the police’s ability to protect during social unrest – and stockpiling. One concern about this finding is that the extent to which it is comprised of first-time gun buyers who may be prone to more accidents (Lang & Lang, 2020).

Individuals who adhered to social distancing measures engaged in less stockpiling. This pattern may be driven by trait agreeableness, collective consciousness (i.e., a shared understanding of social norms surrounding COVID-19), or other variables not measured here. Results from an Ipsos MORI poll of adults in March 2020 (Ipsos, 2020) indicated that younger individuals were more accepting of public stockpiling and increased their purchasing due to COVID-19 more than older individuals. Our results corroborated more stockpiling among younger people in the correlational analyses, but the effect was washed out in the regression model, indicating that it is better accounted for by the other predictors.

Individuals who are more conservative stockpiled more. Interestingly, in this time of political hyper-polarization in the United States, liberals and conservatives may be experiencing a very different pandemic (Conway, Woodard, Zubrod & Chan, 2020). Conservatives seem to be less concerned about the coronavirus pandemic; only 35% of conservatives (compared to 68% of liberals) are concerned about the virus (Malloy & Schwartz, 2020). In another survey, 42% of Republicans feared that they or someone in their family might be exposed to coronavirus, while 73% of Democrats and 64% Independents agreed (Brownstein, 2020). It is somewhat surprising, then, that conservative participants stockpiled more than liberal participants. It may be that conservative participants are more likely to live in regions with less access to food and other goods. If so, supply chain interruptions would be more likely to impact them, which could have motivated their stockpiling behaviour despite lower levels of COVID-19 worry. Further, activating thinking about COVID-19 elevates Americans’ anxiety and indirectly promotes their social conservatism as well as support for more conservative presidential candidates (Karwowski et al., 2020).

Limitations and future directions

One limitation of this study is that the stockpiling measure was dichotomous and did not allow for participants to indicate the extent to which they stocked up on items (i.e., hoarding) due to COVID-19. The same score would be applied to someone who purchased one of the products listed versus several. Additionally, it is not clear how results from an MTurk sample can be generalized to the broader population. MTurk workers are, for example, more highly educated and more politically liberal compared to Americans as a whole (Levay, Freese, & Druckman, 2016), although this sample primarily voted for Donald Trump in the 2016 election. That said, it is not necessarily less generalizable than other convenience samples (e.g., college students) and was chosen as our method for data collection due to the importance of quickly assessing COVID responses in the brief period of widespread domestic lockdown.

Several of the measures were created for the purpose of this study; their validity has not been fully tested, and future research should consider examining the psychometric properties. Our measure of COVID diagnosis relied on self-report, and we did not obtain information on whether or not participants were tested. Further, the sample largely identified as White or Caucasian. This conversation is particularly relevant to COVID-19, which disproportionately impacts racial minorities in terms of both infection rates and deaths (Tai, Shah, Doubeni, Sia, & Wieland, 2020). Future studies may consider oversampling for racial minorities to better understand the disproportionate impact of COVID-19 on these individuals.

Additionally, results from this sample may not generalize to people in other countries due to differences in population (e.g., demographic groups, collectivist vs. individualistic attitudes) and system (e.g., medical systems, treatment access, gun regulations) features. The same argument may be made for different regions within the United States, as different state and federal approaches may have had a significant effect on public understanding of the pandemic and the subsequent response. Future research should replicate these findings in non-US samples and evaluate regional differences within the United States. Finally, the rapidly changing landscape of our knowledge about COVID and safety recommendations/ requirements may be responsible for heightened anxiety surrounding the pandemic. Our results should be interpreted in light of this global and national confusion.

Conclusions

Stockpiling for COVID-19 has been common. In our sample of MTurk workers in the United States, 84% stockpiled at least one item with an average of 5.83 (SD = 3.97) among the 13 items assessed. While stockpiling is a well-documented response to emergencies, our results also suggest that the behaviour is predicted by conservative political affiliation, COVID-19 worry, social distancing, and number of individuals in the home.

Statement of contribution.

What is already known on this subject?

Consistent with prior crises and infectious disease outbreaks, COVID-19 has led to stockpiling (i.e., procuring real or perceived emergency supplies). Stockpiling stems from an individual’s response, either rationally or emotionally, to scarcity, which may elicit stress, anxiety, fear, or panic, leading people to build private stocks or place orders for more than they need. In addition to scarcity, the experience of COVID-19 itself is linked to psychological distress, which may confer additional motivation to stockpile.

What does this study add?

• Increased COVID-related worry and decreased social distancing predicted stockpiling.

• Stockpiling was predicted by conservative political affiliation.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.