The Marginal Propensity to Consume of 2020 COVID-19 Stimulus Payments in Peru

The COVID-19 pandemic has led to unprecedented economic challenges across the world. To combat these challenges, the government of Peru used fiscal stimulus in the form of direct subsidies paid to vulnerable populations for social protection and to stimulate the economy. Using 514 survey responses collected both in-person and online, the objectives of this study were to calculate the marginal propensity to consume (MPC) for Peruvian subsidy recipients and to evaluate the heterogeneity amongst beneficiaries based on four individual factors: pre-pandemic savings, financial inclusion (bank account ownership), survey response type (online vs in-person), and domicile location (residing in Lima Metro or not). Overall, survey responses showed an average MPC of 0.89, which was greater than subsidy-inspired MPC studies from high-income countries like the United States, United Kingdom, and Japan. There was a statistically significant relationship between MPC and liquidity, which corroborated previous studies on MPC from other countries. Relationships between similar programs in Peru and high-income countries for the impact, effectiveness, and purpose of direct stimulus payments are discussed.


Introduction
The initial impact of COVID-19 in Peru was similar to many other countries: swift, unpredictable, and frightening. Official notification of the first case came on 6 March 2020, and quickly led to closed borders and movement restrictions by 21 March (El Comercio, 2020;Gestión, 2020). The rapid onset of the pandemic and the various resulting restrictions led to an unprecedented economic crisis for the country. Year-on-year gross domestic product (GDP) decreased by approximately 40% in April 2020 and unemployment levels in Lima increased from approximately 7% in February 2020 to over 16% by July 2020 (BCRP, 2021). Generally, low levels of household savings and scarce social protection programs in the country made life difficult for Peruvians once restrictions were implemented. Considering the totality of the problem, the Peruvian government initiated large-scale subsidy programs to support homes and businesses.
A substantial component of the Peruvian stimulus response to combat the economic effects of the pandemic was paid out as subsidies, called bonos in Peru, that went directly to qualifying households as cash transfers. D. U. 027-2020, the first decree made by the Peruvian executive branch announcing these stimulus payments, stated that the payment was for the economic protection of households. This implied the main goal of this stimulus was likely social protection, which is an important distinction (Decreto de Urgencia N° 027-2020. Fiscal policies of this nature can not only be created to stimulate the economy but also to protect vulnerable populations. In many cases, both outcomes are likely occurring at the same time. Zhang, Thelen, and Rao (2009), writing for the United Nations Development Program (UNDP), define social protection as "policy interventions that are intended to reduce poverty and vulnerability (including transitory poverty and vulnerability due to economic or other shocks) and to improve human welfare". In addition to social protection, the elevated MPC of subsidy recipients likely led to a strong positive effect on GDP.
Surveyors reported many people asking for help in getting their subsidies and confusion as to why they did not get one when a neighbor, friend, or relative in a similar financial situation did. A large part of subsidy eligibility was determined by the economic classification of a household, which is made by the SISFOH (Sistema de Focalización de Hogares). The SISFOH falls under The Ministry of Development and Social Inclusion (Ministerio de Desarrollo y Inclusión Social [MIDIS]). Households in Peru are classified as either "extremely poor", "poor", or "not poor". In Lima Metro, for example, a household with private medical insurance, a car less than ten years old, a monthly income over S/ 948 Peruvian soles ($270 USD) per capita, or a monthly electric bill over S/ 25.11 ($7.17 ) would be registered as "not poor" and would generally be ineligible for the subsidy payments (MIDIS, 2019). The subsidy for independent workers (Bono Independiente) was an exception as it was available only to "not poor" households that had independent workers with a total household income less than S/ 1,200 ($345), amongst other requirements (Decreto de Urgencia Nº 033-2020, 2020).
There exists a significant base of information and research on the MPC that arises from stimulus payments. Recently, studies from 2020 and 2021 about the MPC resulting from stimulus payments in the United States (Economic Impact Payments) provide many examples of how MPC can be measured, the challenges of measuring it, and the heterogeneity amongst subsidy recipients' reactions to receiving payments (Armantier, Goldman, Kosar, & van der Klaauw, 2021;Baker, Foarrokhnia, Meyer, Pagel, & Yannelis, 2020;Coibion, Gorodnichenko, & Weber, 2020;Karger & Rajan, 2021). These studies all demonstrated that government stimulus has varying effects on spending that depend on an individual's or household's economic situation.
Ricardian Equivalence theory suggests that debt-financed fiscal policies like stimulus payments or tax breaks will generally be ineffective because people understand that they will eventually have to pay back the government in some future period, leading beneficiaries to tend to save all their stimulus payments (Mankiw, 2014). Using this logic, Bracco, Galeano, Juarros, Riera-Crichton, and Vuletin (2021) divided households into two basic categories, Ricardian and non-Ricardian. Ricardian households can use credit and savings to smooth out consumption and therefore do not respond to stimulus payments. Non-Ricardian households are those that are unable to smooth their consumption with credit or savings and therefore are more likely to respond to stimulus with increased spending. Non-Ricardian households, either by choice or by circumstance, spend most or all their income in the period it is received. This type of spending has also been called "hand-to-mouth" or "rule-of-thumb" behavior in different contexts (Bracco et al., 2021;Galí , Vallé s, & López-Salido, 2007).
Whether it is called non-Ricardian, rule-of-thumb or hand-to-mouth, this consumer behavior equates to the same basic result that certain households respond to stimulus by increasing spending. Bracco et al. reported that before the pandemic struck, approximately 53% of Peruvians were hand-to-mouth consumers (Bracco et al., 2021). Relatively low levels of savings and financial inclusion suggest that Peruvians are more apt to have high MPCs resulting from stimulus payments, as they do not have easy access to credit or savings, both of which would allow them to behave more like Ricardian households. States, US dollars (Baker et al., 2020) Multiple studies about economic stimulus and MPC in the United States describe households that display hand-to-mouth behavior as being liquidity constrained. There is a large spectrum of liquidity-constrained households, ranging from very poor households whose income is spent primarily on necessities, to ijef.ccsenet.org International Journal of Economics and Finance Vol. 14, No.3;2022 higher-income households that are liquidity constrained due to high monthly spending. While the circumstances of these two household types are different, they have similar challenges due to liquidity. There is an inverse relationship between household liquidity and the MPC resulting from a stimulus payment as households with low liquidity tend to have higher MPCs (Armantier et al., 2021;Baker et al., 2020;Coibion et al., 2020;Karger & Rajan, 2021 Despite the relative wealth of research on stimulus-based MPCs in high-income economies, there is much less data on the rest of the world. In the case of Peru, research of this nature is non-existent. This is possibly due to the unprecedented nature of the stimulus payments. Peru, through MIDIS, has a selection of social programs that include conditional cash transfers for the poor (Juntos), cash transfers for the old and poor (Pension 65), and programs for early childhood care and nutrition (Cuna Mas and QaliWarma), amongst others (MIDIS, 2021). However, there are no recent examples of large-scale temporary stimulus payments like the COVID-19 subsidies. New subsidies were announced and distributed in 2021; however, analysis of these 2021 subsidies was not feasible in the current study because the money was still being distributed when the survey was conducted (Bono 600 and Bono Yanapay).
The objectives of this study were to calculate the MPC for Peruvian subsidy recipients and to evaluate the heterogeneity amongst beneficiaries based on four individual factors: pre-pandemic savings, financial inclusion (as determined by bank account ownership), survey response type (online vs in-person), and domicile location (in Lima Metro or not). These data are useful to estimate the effectiveness and spending response of the stimulus programs, which could help guide future stimulus in Peru or other Latin American countries.

Method
A total of 514 surveys were collected between 22 November and 13 December 2021 using in-person and online surveys (Appendix). In-person surveys were conducted in the Juní n and Lima departments of Peru ( Figure 2). The surveys from Juní n, which is located to the east of Lima in the Andes mountains, were collected in the districts of Tarma and Huaricolca. The surveys from Lima were collected in Ate and Villa El Salvador, both of which are in Lima Metro. These locations were selected based on accessibility, surveyor personal safety, and the determination that there would be a relatively high percentage of subsidy recipients in these areas. In-person surveys were solicited in the surveyors' neighborhoods, places of work, streets, and markets. Surveyors reported that it was frequently difficult to get people to respond, and many people who were approached did not receive any form of subsidy and did not know why. Many people also asked for help from the surveyors in getting the subsidies that they felt they were owed. Online surveys were collected using Google Forms and distributed through the network of co-workers, friends, and family members of students in the economics department at The  MPC is a calculation of a change in consumption divided by a change in income. In the case of these subsidies, MPC is calculated for each individual survey respondent using the answers to questions four and seven (Appendix). Question 4 asked the amount the respondent received, which represents their change in income (∆Y), and question 7 asked how much they spent in the first three months, which represents their change in consumption (∆C). Equation 1 was used to determine the MPC for each survey respondent on a scale from 0 to 1, with 0 meaning zero increase in consumption and all the subsidy going to savings, and 1 meaning that the recipient spent all the subsidy within three months of receipt.

MPC= ∆C/∆Y
The five subsidies were paid out by two different government agencies.  Table 2). It was beyond the scope of the current study to evaluate beneficiary selection or subsidy distribution; however, this process had many shortcomings. A report from July 2020 by the Peruvian Ombudsman (Defensorí a del Pueblo) highlighted many challenges with the 2020 subsidies including outdated beneficiary registries, lack of capacity to deliver the funds, inadequate payment formats for many rural and older people, and cases of corruption (Defensorí a del Pueblo, 2020).
The subsidy payments were mostly given out as cash in the state-run national bank (Banco de la Nación) or as electronic transfers for those who signed up online to receive the money electronically. The five subsidy payments in 2020 were all for a total amount of S/ 760, although sometimes recipients received two payments of S/ 380 each (Bono YoMeQuedoEnCasa or Bono Independiente) ( Table 2). Despite fluctuating exchange rates in 2020, the S/ 760 payments were generally worth around $215. In some cases, survey respondents reported receiving more than one subsidy, or more than S/ 760, and in other cases respondents only received one of the two S/ 380 payments. In 2020, worker's monthly income averaged S/ 1711 ($490) in the Lima Metro area, S/ 1414 ($405) across all urban areas, and S/ 678 ($195) in rural areas (Instituto Nacional de Estadistica e Informatica, 2021). However, the threshold for a household to qualify as "poor", and therefore be eligible for a subsidy, was based on per capita household income. In Lima Metro, the threshold was S/ 948 ($270), in other urban areas in Peru it was S/ 718 ($205), and in rural areas it was S/ 380 ($108).  Vol. 14, No.3;2022 An analysis of variance (ANOVA) on mean ranks (i.e., a Kruskal-Wallis Test) (PROC GLM; SAS Institute, Inc., Cary, North Carolina, USA) was conducted to test for differences among nine principal uses of the stimulus money: food, utilities, healthcare, rent, education, debt, goods, transportation, and mortgage. To reduce the potential for Type I errors, the experiment-wise error rate was limited to α ≤ 0.05 by using a Bonferroni adjustment (α') of [α/(k{k-1}/2)] = 0.0014 (α = 0.05; k =9). Mean ranks were considered different according to Fisher's protected LSD at α' = 0.0014 rather than α = 0.05. Similarly, Kruskal-Wallis tests were conducted to evaluate differences among five categories of pre-pandemic savings [no savings; <S/ 1,000 (<$285); S/ 1,000-3,000 ($285-$860); S/ 3,000-10,000 ($860-$2,860); >S/ 10,000 (>$2,860)] for MPC and number of respondents in each category, resulting in α' = 0.0050.
Wilcoxon Rank Sum Tests (PROC NPAR1WAY; SAS Institute, Inc., Cary, North Carolina, USA) were used to test for differences in MPC among each of four sets of pairwise comparisons exhibiting non-parametric distributions related to pre-pandemic savings, domicile location, survey response type, and financial inclusion.
In each case the null hypothesis was that the two samples were derived from populations with equal means and the alternative hypothesis was that the two samples were derived from populations with different means. Pairwise comparisons were considered significant at α = 0.05. The survey results indicated that the recipients of subsidy payments in Peru in 2020 needed government assistance. The subsidies were spent almost entirely on basic needs like food, utilities, healthcare, and rent ( Figure 3). Notably, 77% of respondents said food was one of the main expenses on which they spent their subsidies, followed by utilities (38%), healthcare (36%) and rent (9%). Respondents were allowed to select more than one option for this question (Appendix, question 6). Figure 3 shows the primary expenses and the number of respondents who reported each category as one of their principal uses of the stimulus. These responses show that the stimulus was generally not spent on luxury goods nor unnecessary items, as some news sources and social media accounts have suggested (Lí bero.pe, 2020; Peru21, 2020). Respondents largely reported spending all their stimulus payments within the first three months of receipt. While 400 of the 514 respondents reported an MPC of 1.0, the remaining 114 had a somewhat even distribution between 0 and 1 ( Figure 4). Overall, the average MPC was 0.89. The 514 respondents reported receiving a combined total of S/ 414,270. Of that amount, S/ 21,370 (5.2%) went to the repayment of debts. Figure 5. Pre-Pandemic savings relative to reported MPC of 2020 Peruvian COVID-19 subsidy recipients. Bars and diamonds labeled with different letters were significantly different at α' = 0.0050. Figure 5 shows the inverse relationship between MPC and pre-pandemic savings of the current study, which was similar to what was reported in MPC studies from other countries. Savings levels among subsidy recipients were low, with only 99 respondents (19%) having more than S/ 1,000 ($285) in savings at the beginning of the pandemic. 413 respondents reported having less than S/ 1,000 in savings, and two did not respond to question 9 about pre-pandemic savings. Of the 99 respondents with savings over S/ 1,000, 27 were from outside of Lima Metro and 72 from inside Lima Metro. This indicated a slightly higher savings rate in Lima Metro, with 21.5% of the respondents inside Lima Metro (72/335) having more than S/ 1,000 in savings relative to 15.3% of respondents outside Lima Metro (27/176). While 63% of the respondents in Lima Metro reported having bank accounts, only 36% of those outside of Lima Metro said the same.

Results
The Wilcoxon Rank Sum Tests showed that there were statistically significant differences in the mean MPC between the populations with and without pre-pandemic savings, which corroborated research from the United States on MPCs and liquidity. Both subsets of the sample had high average MPCs, but the 99 respondents who reported having more than S/ 1,000 ($285) in savings before the start of the pandemic had an average MPC of 0.72, while the respondents (n = 413) with less savings had an average MPC of 0.93. Other statistically significant differences in MPC, as shown using the Wilcoxon Rank Sum Tests, existed between respondents outside of Lima Metro vs. those in Lima Metro, those responding online vs. in-person, and those with and without bank accounts ( Figure 6).  Vol. 14, No.3;2022 The objectives of the current study were not to definitively measure the effect of the subsidies on Peruvian GDP; however, it is possible to estimate the potential economic impact. The total amount designated to the five cash transfers in Peru was S/ 11.4 billion, or approximately $3.25 billion (see the Decretos de Urgencia from Table 2). This amounted to total expenditures on subsidies of around 1.61% of GDP, which was nominally $202 billion in 2020 (The World Bank, 2021). Using data from 1995-2018, Bracco et al. (2021) estimated the social transfer multiplier effect in Latin America was approximately 0.9, with a peak effect on GDP coming one quarter after an unexpected social transfer (like a COVID-19 subsidy payment). With the calculated MPC of 0.89, we can estimate that the stimulus payments led to direct spending in the amount of approximately S/ 10.1 billion ($2.9 billion). This suggests that even with a modest multiplier effect (lower than the 0.9 estimated by Bracco et al. [2021]), there would be a significant positive effect on Peruvian GDP.

Discussion
The use of discretionary fiscal policies in the form of direct subsidies to consumers to stimulate an economy is not a revolutionary idea. Many countries gave cash handouts to their citizens during the COVID-19 pandemic, and the United States had previously enacted similar policies through tax rebates during the 2001 and 2008 financial crises (Parker et al., 2013). Stimulus-based MPC in Peru was discovered to be much higher than that in the United States, United Kingdom, and Japan. While there are many factors involved, this implies the beneficiaries of the stimulus in Peru comprised a population in much greater need. The high level of expenditures on food, housing, and healthcare showed that the subsidies helped on all these fronts and reduced the significant economic shock and transitory poverty caused by the pandemic. The policies enacted in Peru, and the resulting high MPC, demonstrate how fiscal policy can simultaneously stimulate the economy and provide social protection (Zhang et al., 2009), as even a modest multiplier effect would have to lead to a positive impact on Peruvian GDP.
The survey results showed mostly expected spending habits from the stimulus beneficiaries. The most notable trend discovered amongst the sub-groups of the sample population was the tendency to have a lower MPC as savings increased. This unsurprising trend has been seen in other studies of stimulus MPC (Baker et al., 2020;Parker et al., 2013). Respondents without a bank account had higher MPCs, which was also expected based on the assumption that people with bank accounts are generally wealthier, save more, and are less likely to exhibit hand-to-mouth consumer behavior (Demirgüç -Kunt, Klapper, Singer, Ansar, & Hess, 2018).
The slightly lower MPC of respondents who live outside of Lima Metro was an unexpected, although significant, result. The authors assumed that people living in Lima Metro would have had lower MPCs due to factors like more access to financial institutions, a higher probability to have bank accounts, and higher savings rates as indicated by the survey responses. Considering possible causes, the authors believe that a lower cost of living outside of Lima Metro could potentially be a reason. The subsidy represented a much smaller portion of average income in Lima Metro than in rural areas and was therefore likely more impactful for rural recipients. The standard subsidy payment of S/ 760 was equal to 112% of average monthly income in rural areas, however it was just 44% of average income in Lima Metro (Instituto Nacional de Estadistica e Informatica, 2021).
The results support the assertion that subsidies were a good policy decision, as they provided much needed aid to the beneficiaries. There is no indication from the surveys that the subsidies were wasted, as most was spent on food. One critique is that the high average MPC (relative to similar studies in high-income countries) could hide the fact that many more Peruvians could have benefitted from stimulus payments. The high average MPC shows that each sol given out was used effectively to help people and stimulate the economy, but it does not show what an additional sol could have done. Additional rounds of stimulus, larger payments, or payments open to larger groups of the population may have lowered overall average MPC, but at the same time provided social protection for many people who were missed by the enacted subsidy programs. Of note, the method the Government of Peru uses to determine whether a home is in poverty or not could potentially be one of the challenges that allows some vulnerable populations to be overlooked.