The impact of the Covid pandemic public policies in Chile on consumption

Using survey data, I simulate the counterfactual impact of the Chilean policies during the pandemic on household consumption. I (cid:133)nd that aggregate consumption would have fallen by 16.7% in the absence of public transfers and a quarantine (cid:135)exibilization policy. Consumption would still fall by 10.2% with a quarantine (cid:135)exibilization policy but without public transfers. Overall, with a quarantine (cid:135)exibilization and all the public transfers combined, household consumption was still 6.2% below its pre-pandemic period. Relative to a scenario with quarantine (cid:135)exibilization but without income transfers, I (cid:133)nd that the income, tax, monetary policy, expenses measures were the most progressive policies and increased total consumption by 2.2%, while the debt deferral and pension withdrawals increased consumption by 0.7% and 1.3%, respectively. The policies(cid:146)impact is highly heterogeneous, with 21.5% of the households increasing their individual consumption relative to its pre-pandemic level.


Introduction
Chile has su¤ered from the global crisis induced by the Covid-19 pandemic, with a National Emergency decreed on March 16 of 2020. The cost of the pandemic are estimated to be around 1.3 to 2 percentage points of annual GDP for each month of strict containment measures, with annual GDP growth in Chile for 2020 being -5.8%, which represents a drop in 7.5 percentage points in annual GDP growth relative to the previously estimated trend (Central Bank of Chile 2021).
This article provides an estimate of the impact of the Covid shock and the public policies implemented to mitigate the crisis on the consumption of the Chilean households. As a developing economy, Chile has a signi…cant amount of socioeconomic inequality, reporting a Gini coe¢ cient of 0.46 in 2017, the second highest among the OECD countries (OECD 2021). Chile also has a large fraction of informal workers relative to the developed economies 1 , with several workers having no access to o¢ cial unemployment insurance. It is also the case that in Chile many households are either entirely formal or entirely informal (OECD/ILO 2019), therefore many families are unable to compensate losses in informal employment by resorting to a family member with more secure income. Furthermore, around 10% of the Chilean households have no access to credit (Madeira 2019), being unable to …nance shortfalls in income with debt. For these reasons, it is important to analyze whether the heterogeneous impact of the policy measures across families of di¤erent backgrounds. Among the policies implemented there was a general tax deferral, a monetary policy rate reduction of 150 basis points, an employment protection scheme for workers with a frozen schedule or reduced hours, income support programs targeted at the poor and middle class, a voluntary deferral of debt payments o¤ered by banks and other …nancial institutions, a government sponsored loan and three pension withdrawal programs from the individual pension accounts.
To estimate the impact of the di¤erent public policies I use the Chilean Household Finance Survey (Encuesta Financiera de Hogares, in Spanish, hence on, EFH). The EFH data has detailed information on the family demographics, the income and labor status of its members, plus the assets (real estate, …nancial portfolio and pension accounts) and debts across di¤erent loan categories. 1 Informal workers represent around 28% of the labor force in Chile, which is high compared to the 18% informal employment rate in developed economies (OECD/ILO 2019). However, Chile's informal employment as a fraction of the labor force is similar or below most Latin American countries and it is substantially below the 70% informal employment rate estimated for the entire developing and emerging countries (OECD/ILO 2019).
This survey has therefore broad information on several categories, which are required to estimate the impact of the di¤erent policies, which may di¤er according to the overall household income level, the number of children, the number of retirees, the income of each member, whether the workers are on a formal labor contract or not, families'public utilities expenditures, the value of their properties, the funds of their pension accounts and the values of their debts across di¤erent loan types (credit cards, mortgages, installment loans). I aggregate the di¤erent policies in 3 groups (public policies (tax deferral, monetary policy, income and expenses support), debt deferral, pension withdrawal policies) and analyze their cumulative e¤ect together. The …rst set of public policies (tax deferral, monetary policy, income and expenses support) comprises a set of many policies implemented by the authorities that involve transfers to the households, while the debt deferral and pension withdrawal policies involve transfers from private agents (banks, other private lenders, the private managers of the workers' individual pension accounts) to the households. Finally, I analyze the counterfactual impact on consumption of the quarantine ‡exibilization policy ("Step by Step", in Spanish, Paso a Paso) which relaxed gradually the mobility restrictions across counties.
The EFH dataset is an essential tool for calibrating the impact of the di¤erent policies, due to: i) its comprehensive information on demographics (income support policies depend on the number of members and income of each household); ii) exhaustive information on the real assets of the household (which is essential for evaluating the real estate tax deferral); iii) information on the debt value of di¤erent types of loans, such as mortgages, installment loans, credit cards and lines (essential to calibrate the e¤ects of the debt deferral programs), and iv) information on the value of the individual pension accounts (essential for evaluating the impact of the pension withdrawals).
The availability of this information at the household level allows to portray the heterogeneity of the policy bene…ts across individual households, rather than an analysis of representative agents.
To calibrate the impact of the di¤erent policies on household consumption, I consider four major components: i) an expenditures model for each of the 12 categories of goods, ii) a simulation of the heterogeneous labor shocks received by workers before and during the pandemic, iii) the transfers received by each household according to the di¤erent policies implemented during the pandemic, and iv) a rule of thumb calibration of the Covid restrictions on the consumption of di¤erent goods combined with a monthly dataset of the quarantine stages in each Chilean county. The empirical model of consumption choice across 12 types of goods obtains the empirical elasticities of consumption according to current income, permanent income, home ownership, number of children, adults and retirees, plus the age and education of the household head, which are estimated from detailed expenditure information available from the Chilean Family Expenditures Survey. The estimated model of expenditures is then applied to the EFH sample to obtain the simulated expenditures before the pandemic and for each month during the pandemic. These for each household in the pandemic period is then obtained as a weighted average between the pre-pandemic period expenditures and the expenditures simulated for each stage of the quarantine ‡exibilization, with more strict quarantine stages implying a smaller weight of the pre-pandemic expenditures. The weight between pre-pandemic and quarantine stage period expenditures is then speci…ed according to the households'county of residence and its "Step by Step" quarantine stage at a monthly frequency. The simulated results are therefore based on several strong assumptions regarding the consumption choice model and the shocks experienced by households. However, the simulated results compare well to aggregate results for the monetary costs of each policy and the estimated variation in consumption observed by national accounts during this period.
The simulated results show that aggregate consumption would have fallen by 16.7% relative to the pre-pandemic level in the absence of transfers to the households and a quarantine ‡exibilization policy. With the quarantine ‡exibilization and in the absence of household transfers, aggregate consumption would fall by 10.2%, but the support policies softened this to a fall of 6.2%. Almost all the families bene…tted from the public support, increasing their consumption relative to a no policy scenario. Individual consumption relative to 2019 for each household changed between -15% to +7.5% once all the policies are accounted for. Around 78.5% of the households still decreased their consumption during the pandemic, while 21.5% increased their consumption.
Our study is related to a growing literature on how surveys can inform about the …nancial Our study adds to this literature by using detailed microeconomic data to calibrate the crisis'heterogeneous impact of the pandemic and its related public policies on consumption in Chile, which complements other studies analyzing the e¤ects on the Chilean household mortgage and consumer loan default during this period (Madeira 2021) or the recent study of Barrero et al. (2020) studying the e¤ect of the policy transfers on income and consumption. Relative to Barrero et al. (2020), our analysis is done at the individual level of each household in the sample and not a representative agent framework by income quintiles 2 . A disadvantage of our framework is that the perceived consumption reported by households in a survey are far below the consumption aggregates in national accounts data.
For instance, the total consumption in our survey dataset for 2017 corresponds to just 34% of the 2 It is worth noting that the analysis of Barrero et al. (2020) is not calibrated at the individual household level.
Their analysis instead uses the fraction of income subsidies (from the CASEN 2017) and the fraction of the consumer and mortgage debt (from the EFH 2017) of each income quintile, then it assumes that each income quintile receives a proportional fraction of the total value of each policy. This analysis is more similar to a representative agent framework with 5 agents, rather than an analysis of heterogeneous households based on the available micro-data. Also, their analysis rests on the strong assumption that the public income bene…ts of the Covid policies were distributed in the same way as the other public subsidies in 2017. In the same way, their analysis does not re ‡ect that consumer installment loans and credit cards or lines are treated di¤erently in terms of the debt deferral programs. GDP, while the fraction of consumption for the households and non-pro…t institutions was 63.4% in national accounts. This under-reporting has several causes, with some being that households are reporting out-of-the-pocket expenditures and do not consider the payments of private insurance or government paid health services or education (Attanasio and Pistaferri 2016), while other causes are due to the under-reporting of some goods such as alcohol or videogames which are subject to social stigma (Crossley and Winter 2014). Furthermore, survey data of consumption tends to under-report expenditures in durables (Attanasio and Pistaferri 2016), which were the type of goods that increased the most during the last quarter of 2020. Another weakness in the methodology of this article is that all the analysis is done in partial equilibrium and there is no consideration of general equilibrium e¤ects between consumption, government spending and revenues and …rms' activity. Finally, this article analyzes the impact on consumption of the quarantine ‡exibilization policy, but, due to a lack of data, the analysis is unable to study how consumers substituted across di¤erent goods and stores as restrictions in mobility changed (see Goolsbee and Syverson 2021).
This work is organized as follows. Section 2 describes the Chilean Household Finance Survey, while section 3 summarizes the quantitative calibration of each policy measure. Section 4 reports the estimates of the empirical consumption model of the individual household, while section 5 summarizes the counterfactual impact of each policy on the consumption of di¤erent goods and across households during the pandemic period. Finally, section 6 summarizes the policy implications.

The Chilean Household Finance Survey
To quantify the potential policy impact I use a sample of 4,549 households from the most recent Household Finance Survey (EFH) wave, implemented in 2017. This survey has detailed measures of the household's demographics, income, assets (…nancial portfolio, real estate) and debts, including mortgage, educational, auto, retail and banking consumer loans. Households also report whether they applied for any loans, any rejected loan applications, and the motives of their consumer loan contracts. Furthermore, I use the survey's information to obtain a measure of each household i's permanent income, given as the sum of its non-labor income (a i ) plus the labor earnings of each labor force member k: P i;t = a i + P k P k(i);t . The permanent income of each household member is given by P k(i);t = (Y k;i (1 u k;i;t ) + Y k;i rr k;i u k;i;t ), where Y k;i is worker k's earnings when in Education and age correspond to the household respondent (the member of highest income).
All values use household weights. employment, u k;i;t = u(x k(i) ; t) is its probability of being in an unemployment spell, and rr k;i is its replacement ratio of income during unemployment relative to the earnings while working, conditional on the mean of workers with similar characteristics of education, sex, age, industry, income quintile and region in the Chilean Employment Survey (Madeira 2015(Madeira , 2018. The EFH survey has an over representation of richer households, since rich households have more complex …nances in terms of assets and debts and also represent a higher portion of the economic activity. To adequately correct for the over representation of wealthier households, all the statistics in this article use expansion factors (or population weights), meaning each observation is weighted with a number f i representing the statistical number of households equivalent to i. Table 1 shows the age, current household income and permanent household income, according to the education of the household head. More than half of the household heads have completed the secondary school or less, with more educated households also reporting on average a younger age.
There is a signi…cant amount of income inequality in Chile, with the percentile 75 showing a current or permanent income that is almost three times the income reported by the percentile 25 of the households. Furthermore, there is a strong education premium in Chile, with postgraduate educated households reporting an average income that is more than twice the average household income.
Income increases signi…cantly with each education level: households with a college degree have more than twice the average income of those with just a secondary education and the households with a postgraduate degree earn more than 50% the income of those with a college degree.
In Table 2 I summarize some characteristics of the households'assets, according to the household size as measured by the number of its members. It shows that 62.7% of the households own their main home. This is relevant, because it shows that almost two thirds of the households should have been able to bene…t from the deferral of the real estate tax payments implemented in Chile. Only 14% of the households are composed of a single person. Home ownership is increasing with the household size, while the pension balance to income ratio decreases signi…cantly with additional members. Finally, the data shows that less than 1% and 8% of the borrowers have defaulted on their mortgage and consumer loans, respectively. However, the fraction of borrowers with arrears in mortgages or consumer loans increases signi…cantly for households with four members or more. Therefore most household borrowers should have been with no arrears and able to take advantage of the debt deferral programs implemented by the Chilean banks and other lenders. ii) the Employment Protection Law, which allows companies to give workers access to income through the public unemployment insurance system while temporarily suspending their activity or retaining the workers on a 50% labor schedule; iii) a deferral of the public utilities'payments; iv) on May 17 the government also announced the distribution of 2. A Middle Class bonus was announced in August with a single payment (not to be repeated) for workers that lost at least 30% of their income relative to the previous year, giving 500, 400, 300, 200 and 100 thousand pesos for workers with a prior monthly income, respectively, between 400 thousand and 1.5 million, 1.5 and 1.6 million, 1.6 and 1.7 million, 1.7 and 1.8 million, and between 1.8 and 2 million pesos.
iii) in August of 2020 the tax administration sponsored a program of zero interest rate loans of up to 650,000 pesos 4 , which was available for workers that had a monthly income above 400,000 pesos during 2019 but that experienced an income fall above 30% after the beginning of the pandemic in 2020. This tax administration sponsored loan had a top amount no higher than 650 thousand pesos, with each worker being able to request up to 3 loans during a period of three months. For the repayment of this zero interest rate loan, the government would make an amortization in the annual tax returns of each worker in 2022 for 10% of the loan amount, and a 30% amortization 2023, 2024 and 2025. The yearly tax collected loan amount would be limited to up to 5% of the yearly taxable income, plus a smaller installment equivalent to 3% of the monthly wage. The remaining debt would be forgiven after 2025 if the loan amount is not yet repaid.
On July 30th of 2020 the Congress implemented an exceptional measure that allowed all workers to withdraw a signi…cant amount of up to 150 UF 5 (around 5,500 USD) from their accumulated individual pension accounts 6 . Each member of the pension system (anyone who has held a formal job in the past) can withdraw up to 100% of its funds for accounts with a value below 35 UF, up to 35 UF for accounts between 35 and 350 UF, up to 10% of the funds for accounts between 350 and 1,500 UF, and 150 UF for accounts above 1,500 UF. Although this measure is not a loan, it can be viewed as a similar measure as a household borrowing from his own future pension income. A second Pension Withdrawal was legislated on December 10th of 2020. A third Pension Withdrawal was implemented on the 28th of April of 2021, but its analysis is not considered in this article, because it is limited to studying the impact of measures during the 13 month period between March of 2020 and March of 2021. This policy measure was possible because Chile has a social security mostly based on compulsory contributions (up to a maximum taxable wage) that workers make to pension funds in private companies. In ordinary times, these pension funds can only be used after age 65, but this law allowed for a withdrawal in cash, check or deposit, without penalties.

Calibration of the di¤erent public policies
To evaluate the public policies I evaluate each month between March of 2020 and March of 2021, updating the monthly income of each household based on three components: i) the unemployment rate of each group of workers at each month t based on their type given by x k = (gender, age, education, industry, residence in capital area or not), ii) the fraction of the labor force that in each period enters a frozen work relationship (F W t ) or a reduced hour schedule (RW t ), iii) the public support policy bene…ts received by the households during the Covid crisis (ps i;t ).
The unemployment risk (u k;t ) of the EFH workers k are based on the mean statistics for 108 worker types (given by a vector x k of their education, age, industry, residence in capital area or not) from the Chilean Employment Survey (ENE) between March of 2020 and March of 2021. The unemployment risk u k;t is de…ned as the probability that the worker is unemployed at a given period (U k;t = 1) conditional on his characteristics x k . Conditional on the workers' characteristics x k = fSantiago Metropolitan area or not, Industry (primary, secondary, tertiary sectors), Gender, Age ( 35, 35 54, 55), Education (secondary school or less, technical degree, college)g, the empirical estimation of the probabilities u k;t is obtained as and the replacement ratio of income during unemployment with rr = 0:60 denoting the replacement ratio of income during unemployment, Y k;i being the EFH survey's reported working income for worker k in household i, a i non-labor income, ) denotes whether the worker k of household i at time t experienced either unemployment or a frozen relationship or reduced hour schedule, and u i;k;t ; F W i;k;t ; RW i;k;t being pseudo-uniform random numbers. Data for the probability of workers entering into a frozen relationship (F W t ) or a reduced schedule (RW t ) in each month was obtained from the Chilean Administrator of the Unemployment Insurance, with series only available at an aggregate level and with no heterogeneity across workers.
The total public bene…ts ps i;t received by household i in period t considers the sum of the total income, expenses and monetary policy support (psY i;t ) with the debt deferral and tax sponsored loan policies (psDs i;t ) and the pension withdrawal policies (psP ension i;t ): The bene…t value of the public support measures received by each household, ps i;t , is calibrated using their income, children, real estate properties, county of residence, loans (mortgages, consumer loans, credit cards, lines of credit, and other debts). To account for the time-variation of the programs I create dummy variables with the name of the month in capital letters denoting a bene…t introduced that month and kept afterwards, ex: M ARCH t 1(t M arch 2020). 15 cubic meters of water per month (roughly 10,000 pesos) and 60 months of a free internet plan provided by the state (roughly, 30,000 pesos): The Employment Protection Law bene…ts are then estimated as: with f e k;i being a dummy denoting whether worker k has a formal employment contract.
The real estate tax deferral for each household i is given as ; 000; 000)), with V i;v denoting the survey reported property appraisal value and v = 0; 1; 2; 3 being the main family home and up to 3 other properties that may be owned by the family. The tax rate 0:025% is applied to properties every quarter, but it is divided by three to be measured monthly.
The deferral of tax debts is taken to be the VAT rate (19%) for the monthly income reported by households from their micro businesses or self-employment: T DD i = 0:19 P k Y k;i se k;i , with se k;i being a dummy variable for whether worker k is a micro-entrepreneur or in formal self-employment.
The bene…t obtained from the lower stamp tax (a reduction from a monthly rate of 0.033% to 0%) and monetary policy rate is given as B_ST _M P R i = (0:00033+0:0125=12) P 3 rt=1 P 3 l=1 L i;rt;l , where rt denotes the debt type (1 bank credit card, 2 retail credit card, 3 bank credit line) and l = 1; 2; 3 denotes up to 3 loans reported by the household in each debt type, assuming that households keep similar amounts of revolving loans as in 2017. The Monetary Policy Rate reduction of 1.25% is divided by 12 to be measured in monthly terms. Other loan categories reported in the EFH, such as banking consumer installment loans, retail installment loans, educational, automobile and credit union debt, typically have maturities of 12 months or more and at a …xed interest rate, therefore these do not apply for lower stamp tax and interest rate. Also, since some households may become more indebted, while other households may lose access to debt during the pandemic, I do not include new loan creation to compute these bene…ts.
The total income, expenses and monetary policy support psY i;t for each month t is therefore: The ‡exible credit card scheme and the debt deferral for non-defaulting customers (Df i = 0) is measured as , being equivalent to one third of the monthly bank and retail credit card bills (rt = 1; 2) plus the debt service of banks and credit unions consumer installment loans (Ds i;rt;l ) and the mortgage debt service (M ds i;v ) for the main home (v = 0) and up to three other properties (v = 1; 2; 3).
The government sponsored zero interest rate loan of up to 650,000 pesos (given in three monthly installments) for each worker with an above 30% income loss corresponds to a total household support of P ubLoan i = P k (650; 000=3)1(LossY i;k;t 0:30), with LossY i;k;t denoting the Income Loss faced by the worker k in the household i at time t in 2020 relative to its permanent income in at the unemployment risk (u k;i;t=2019 = u(x k(i) ; t)) that a worker of his characteristics faced in 2019.
The total policy support that households received in terms of a lower debt service (due to a lower monetary policy rate, lower stamp tax, and the debt deferral scheme) sums up as with the debt deferral amount of six installment payments being spread across the twelve months of the year, while the tax sponsored loan is spread over an eight month period.
I account for the pension withdrawals, with each withdrawal allowing every member of the pension system (anyone who has held a formal job in the past) to withdraw up to 100% of its funds for accounts with a value below 35 UF, up to 35 UF for accounts between 35 and 350 UF, up to 10% of the funds for accounts between 350 and 1,500 UF, and 150 UF for accounts above 1,500 UF. 97% of the workers requested their pension withdrawal within the …rst 2 months (Central Bank of Chile 2020). The value of the pension policy withdrawal for each k member is given by pw n k;i = min(P W I n k;i ; 35U F )1(P W I n with P W I n k;i denoting the pension account balance before the withdrawal is done. Therefore P W I n=1 k;i = P W I k;i for the …rst pension withdrawal and P W I n=2 k;i = P W I k;i pw n=1 k;i for the second pension policy withdrawal, with P W I k;i denoting the self-reported pension account balance in the EFH survey 8 . Note that the second pension withdrawal is by de…nition smaller than the …rst withdrawal, since several pension accounts are either fully withdrawn or substantially reduced after the …rst withdrawal. The impact of the two pension withdrawals on the monthly income is then given by 8 P W I k;i is self-reported for the main household respondent, while for the other members it is imputed from a log-linear regression with their log-work income, gender, education level and a quadratic term of the age.

6) psP ension
with n(LossY i;t ) denoting the parameter for the household to spread its spending of the pension idiosyncratic term that is independent across households (i) and divisions (d).
Note that the consumption in surveys is substantially di¤erent from national accounts. In particular, the total consumption in the EPF survey dataset for 2017 corresponds to just 34% of the GDP, while the fraction of consumption for the households and non-pro…t institutions was 63.4% in national accounts. The di¤erence between national accounts' consumption and survey self-reported consumption does not necessarily imply that one of the data sources is incorrect, but rather that these datasets have a di¤erent de…nition for consumption (   College education, in particular, is strongly associated with expenditures in Education, Health, Communications, Restaurants, and Other products. , , denote 1%, 5% and 10% statistical signi…cance.

Summarizing the policy impact on consumption
I then apply the estimated models ( d ) to the EFH households to obtain the counterfactual impact on consumption of the policies p for each month t between March of 2020 and March of 2021: where CovCS d is an exogenous pandemic shock which decreases spending in some divisions d  Similar consumption ratios were built for a scenario without the "Step by step" program, that is assuming that the quarantine phase was always complete (P P i;t = 0): 12) AC p i;d (P P i;t = 0) =  Table 4 summarizes the size of each set of policies a¤ecting the Chilean households as a fraction of the GDP. The total policy support of 15.1% of GDP is slightly lower than the 17.4% reported in Table 5  The numbers obtained in Table 4 for each policy amount are quite close to the budget numbers.

Size of the public bene…ts received by the households
The Employment Protection Law of 0.6% matches the budget amounts for the policy between  due to the fact that their work considers only the debt deferral amounts obtained by August of 2020. The …rst pension withdrawal is about 6.2% of the GDP, which is slightly below the 6.9% value reported in Barrero et al. (2020). The second pension withdrawal obtained from the EFH calibration is just 3.4% of the GDP, which is signi…cantly below the 6.7% value reported in Barrero et al. (2020), but this result is to be expected due to the under-reporting of …nancial assets and pension amounts that is common in survey datasets (Christelis et al. 2013, Bover et al. 2016). Table 5 shows the fraction of households that received no bene…ts for each group of policies, …nding that only 0.2% of the households did not receive any support. However, there are big di¤erences across groups of policies. Only 6.5% and 0.9% of the households were excluded from, respectively, the pension withdrawals and the Income, taxes, monetary policy reduction (MPR), expenses support. Debts, on the other hand, are concentrated on the richer households. Around 54.6% of the households did not bene…t from the deferral of the loans, with this fraction being as high as 80.9% and 70.3% for the households in the income quintiles 1 and 2. Therefore the large majority of the poorer households did not bene…t from the debt deferral, while only 28.9% of the households in the upper income quintile did not bene…t from the loan deferral programs.
In terms of the distribution of the bene…ts among households with positive amounts (ps p i;t > 0), ) are calculated only for the households that received a positive amount of bene…ts, that is for those with ps p i;t > 0. on the income growth of its bene…ciaries, independently of the income quintile. This shows that the pension withdrawals and debt deferral were not progressive policies at all, since the highest quintiles received similar ratio of bene…ts relative to income and in absolute money amount that is much more due to their higher income levels (as shown in Table 1). Table 6 shows the results of the estimated policy impacts for the aggregate consumption (AC Sum ) and its divisions (AC d ) for the months of July and August of 2020, which mark the peak of the  with the national accounts reports for August of 2020 (Central Bank of Chile 2021).

The impact of the di¤erent public policies on consumption
The Covid shock in July-August had an heterogeneous impact across products, because of the exogenous e¤ect of the lockdown on shopping and grocery plus the income elasticities of each good  Table 7 shows the impact of the di¤erent policies on the household consumption between March of 2020 and March of 2021. It di¤ers substantially from Table 6, because the 13 month period considered in Table 7 includes months with lower quarantine restrictions and lower unemployment rates than the July-August months portrayed in Table 6 (the "peak" of the pandemic). Now the simulations for the scenario of no policies predict a fall in consumption of 10.2% for this period, which is signi…cantly better than the 17.2% drop for the July of 2020 period alone. Similarly, all Communications, Clothing, Furnishings and Recreation also grow by around 5% after all the policies are implemented. This stands in contrast to the strong growth in the consumption of durables found after the pension withdrawals were implemented in Chile, which showed record sales of cars and other durables (Central Bank of Chile 2021). This is due to purchases of durables being  Table 7 to make the comparison easier between the scenarios with and without the quarantine ‡exibilization program. under-estimated by consumption surveys, since such goods are imputed as a ‡ow of use rather than as lumpy purchases that change strongly during the business cycle (Attanasio and Weber 2010). Table 8 shows the counterfactual impact of the di¤erent policies on consumption during March 2020 to March 2021 under a full quarantine policy and no ‡exibilization of activity from the "Step by step" program: P P i;t = 0. The results for the "All policies scenario" show that total consumption was 4.3% higher with the quarantine ‡exibilization. There are also substantial di¤erences across product divisions with the quarantine ‡exibilization program. Without the quarantine ‡exibilization program there would have been more expenses in Food, Health and Communications, but less consumption for the other products especially Clothing, Recreation and Education which would have been 10% lower. Furnishings and Other Goods consumption would also have been around 6% lower without the quarantine ‡exibilization program. Furthermore, the quarantine ‡exibilization program insures a higher consumption between 5.2% to 6% more for the scenarios with just a few individual support policies implemented instead of the All policies scenario. The di¤erence is substantial for the scenario with No Support Policies, since in that case the quarantine ‡exibilization program insures 6.5% more in total consumption. Table 9 shows the fraction of families that either decreased or increased their consumption relative to 2019 for each policy, including the scenarios with and without quarantine ‡exibilization.
No family kept its consumption exactly at the same level as in 2019. The results show that with the quarantine ‡exibilization and all the transfer policies implemented together, around 21.5% of the households increased their consumption during the pandemic period relative to 2019. Now I show how the results are a¤ected by alternative speci…cations of the model calibration. Table 10 shows two robustness checks regarding the exogenous pandemic shocks to consumption, which imply a more negative e¤ect of the restrictions for the purchase of some goods. Alternative 1 considers that the pandemic has no impact on Food and Health goods, instead of the positive impact considered in the baseline calibration. Alternative 2 considers that the pandemic has a negative impact on all products and considers a worse e¤ect for Food, Clothing, Housing, Furnishings, Health, Transport, Recreation, Restaurants and Other goods. The estimates for total consumption in Alternative 1 are slightly below those in the Baseline (see Table 8), showing for the no policy scenario a fall in consumption of 11.6% and 19.1%, respectively, with and without the quarantine ‡exibilization program. Under the all policies scenario, Alternative 1 shows a drop in consumption of 7.7% and 13%, respectively, with and without the quarantine ‡exibilization program. The estimates for total consumption in Alternative 2 are much lower than in the Baseline (Table 8), showing for the no policy scenario a fall in consumption of 15.7% and 25.9%, respectively, with and without the quarantine ‡exibilization program. Under the all policies scenario, Alternative 2 shows a drop in consumption of 12.3% and 20.6%, respectively, with and without the quarantine ‡exibilization program. However, both Alternative 1 and Alternative 2 coincide with the Baseline in terms of the big impact of the quarantine ‡exibilization, which is 7.5% and 10.2%, respectively, for Alternatives 1 and 2 under the no policy scenario and 5.3% and 8.3%, respectively, for Alternatives 1 and 2 under the all policies. The alternatives also coincide with the Baseline in terms of the substantial impact of the all policies, which increase consumption by 3.9% and 3.4% under Alternatives 1 and 2, respectively (for the case of the quarantine ‡exibilization program scenarios). Figure 1 shows the heterogeneous policy impact on the distribution of the households'consumption relative to their individual consumption in 2019 (AC p i;Sum ). Figure 2 shows the individual consumption growth relative to the scenario with no policy support ( ). The probability density function in both …gures is estimated with an Epanechnikov kernel using the Silverman's bandwidth rule. Both …gures are based on the counterfactual simulations with the quarantine ‡exibilization. Figure 1 shows that the impact on consumption of the Covid shock in the No Policy scenario on consumption is centered between -20% to 0%, therefore all households decreased their Debt plus pension policies All policies consumption due to the pandemic. The individual consumption improves signi…cantly with each policy, especially with the Income, taxes, monetary policy reduction (MPR), expenses support, which shows a consumption di¤erence relative to 2019 between -15% to +5%. With all the policies combined, the consumption di¤erence relative to 2019 changes to between -15% to +7.5%. Figure 2 shows that all the policies analyzed increased the individual consumption of all the households relative to the no policies scenario, therefore not a single household reduced its consumption after each policy. The debt deferral had a small impact for most households, with most households increasing their consumption by less than 2.5% and a modal increase around 0.4%.
The Income, tax, monetary policy reduction (MPR), expenses support had a wide impact, with the individual consumption of each household increasing between 0% and 5% and a modal increase of 1.2%. This is a very e¢ cient impact for the Income group of policies, since this group of policies implied just one third of the budget of the pension withdrawals. The pension withdrawals implied an increase in individual consumption between 0% and 3.75%, with a modal increase around 0.8%.
The debt plus pension policies (debt deferral, tax sponsored loans, pension withdrawals) jointly had a substantially higher impact, increasing individual consumption between 0% and 5%, with a modal increase around 1.3%. Finally, the "All policies" scenario increased individual consumption between 0% and 10% relative to the "No Policies" scenario, with a modal increase of 1.8%. In summary, all households increased their consumption relative to the No Policies scenario.

Conclusions
Using micro-data from the Household Finance Survey (EFH), this work estimates the counterfactual impact of the Covid crisis'public policies on the Chilean household consumption. The work provides an analysis of the demand shock in Chile across households and di¤erent products, although other works emphasize that Covid implied an heterogeneous supply and demand shock across sectors Based on the calibration from the EFH sample, the income, tax, monetary policy, expenses support amounted to 3.3% of the GDP in transfers for the households, while the debt policies and the pension withdrawals amounted to 2.7% and 9.6% of the GDP. However, the income group of policies showed the most progressive distribution, with transfers of 0.4% of the GDP for the …rst income quintile and around 0.8% of the GDP for each of the other income quintiles. Although the lowest income quintile receive a lower money amount, the transfer represented a higher relative income increase for those families, and therefore this group of policies had a strongly progressive design.
The debt policies and pension withdrawals were much less progressive, implying, respectively, a transfer of only 0.1% and 0.4% of the GDP to the lowest income quintile, while transferring 1.4% and 4.1% of the GDP to the top income quintile, respectively. Only 0.9% of the households did not bene…t from the income group of policies, while 54.6% and 6.5% of the households did not bene…t from the debt deferral and pension withdrawals, respectively. Furthermore, the income group of policies implied a much stronger increase in income for the lowest income quintiles, while the debt deferral and pension withdrawals implied higher income increases for its bene…ciaries in the top income quintiles. The reason is because the pension withdrawals bene…t more the higher income families and those with formal employment, while the debt deferral bene…ts are concentrated on the top income households which have larger mortgages and consumer loans.
Total consumption could have fallen by 16.7% in a scenario with no policy transfers to the households and no quarantine ‡exibilization. Just with the quarantine ‡exibilization policy, the authorities softened the blow to consumption to a fall of 10.2% relative to a no-pandemic scenario.
Without the quarantine ‡exibilization program there would have been more expenses in Food and Health, but less consumption for the other products such as Clothing, Recreation, Education, Furnishings and Other goods. Relative to a scenario with quarantine ‡exibilization but without income transfers, I …nd that the income, tax, monetary policy, expenses measures increased total consumption by 2.2%, while the debt deferral and pension withdrawals increased consumption by 0.7% and 1.3%, respectively. The debt plus pension policies combined (deb deferral, tax sponsored loans, pension withdrawals) could have increased consumption by 2.5%. However, with all the policies combined there was an increase in consumption of 4% relative to a scenario with no income transfers. With the quarantine ‡exibilization and all the transfer policies, the total consumption between March of 2020 until March of 2021 was still 6.2% below the pre-pandemic level. This drop of 6.2% is just slightly below the fall of 7.4% in national accounts' consumption during the last three quarters of 2020 and the …rst quarter of 2021 relative to the last quarter of 2019.
Finally, there was no household that reduced its consumption after the transfer policies. Relative to a scenario with no transfers, the all policies combination increased individual consumption between 0% and 10% for most households, with a modal increase of 1.8%.