Update: COVID-19

Introduction

COVID-19 has changed behaviors of individuals as consumers, employees, or business owners in many aspects. Due to the concerns of public health from COVID-19, most governments have imposed measures to limit the spread this virus to their citizens, such as state and community-level lock down in the U.S, contact-tracing in S. Korea, national level lown down in Newzealand. These measures and fears of COVID-19 among consumers have changed the ways of consumers spend and consequence on local business owners (e.g., Argente et al., 2020 and Baker et al., 2020). S. Korea is not an execption. The overall economy of S. Korea are negatively affected by COVID-19 (Kirk, 2020), and even small business owners (Shin, 2020).

To alleviate this economic hardship on individuals and local business owners, the Ministry of Interior and Safety of S. Korea, and metroplitian city and state governments have launched stimulus payment programs to provide funds to all households (by the centerl government) or targeted housholds (local governments) under which the payments can only be used to specific stores and sectors.

This paper empirically estimate the effect of these stimulus payments and spillover effects on local business owners.

This paper is unique in three aspects. First, even though there is growning literature that have examined changes of consumer behaviors and the effect of stimulus payments on consumer behaviors during the pandemic or idosycratic shocks in the US, this paper would be one of the firs papers that captures comprehensive payements (covering 99.5% of the households in S. Korea) targeting for local small business (excluding big gerocery marts, online shoppoing malles, and etc.).

Second, using consumer spendings at local markets in Seoul, a metroploitan city, we measure the spillover of the changes in consumer spending on restaurants, at leaset at some aggregated levels.

[I am not sure if the following is possible] Third, with some assumptions on average consumption behaviors, we use a structure model of the demand for eating out.

Stimulus Payments in S. Korea During COVID-19

Korean governement and Seoul Metropolitan Goverment have provided stimulus payments to most citizens in Seoul, Korea.

Unlike the stimulus payments in the U.S. (initiated by the 2020 CARES act), these stimulus payments target on the increase of consumption and its spillover effects on small business owners. There are several restrictions on the usage of these stimulus payments. These restrictiion will be discussed below.

Korean Government Stimulus Payments (KG Stimulus Payments)

The officeical name of this stimulus payment is “Korean Emergency Relief Funds.” This is a primalry focus.

The Korean Government provides emergency relief funds to almost all households in S. Korea (99.5% of households).

Period of Application for KG Stimulus Payments

Application for the stimulus payments by Korean governement were open from May 11 to June 5, 2020. Once applicants receives their payments, they were able to use the funds by the end of August. After then, unused funds would be donated to other subsidy programs for those who are in hardship during the period of COVID-19.

Since we have data unitl the 2nd quarter of 2020, the spending of these payments could not happen during the second quarter. However, a study revels that high increase in consumer spending right after the stimulus measures were implemented.(See the new article.)

Amounts of Korean Gov.Stisdfss

Depending on the number of the members of households, the amount of the payments varies.

Number of Members in Household	1	2	3	4
Amount (KRW)	400,000	600,000	800,000	1,000,000
Amount (USD)	348.74	523.10	697.47	871.84

Forms fo Stimulus Payments

Two forms of the KG stimulus payments are provided

Gift certification: Local stores under the network of Zero Payment, the payment flotform that provides low expensices related to credit card transaction, mostly helping local small business owners. If individiuals choose this options, they will receive addition 10% of the funds that they receive.
Debit/Credit Cards: Most credit card firms provide stimulus payments to applicants’ credit cards or debit cards.
Pre-paid Credit cards: Applicants visits local community service centers and apply for this types of KG stimulus payments.

Restrictions on Stores

Even though the restrictions on the usage of KG stimulus payments vary on the forms of the payments, all the forms of the stimulus payements follows the following restriction:

Sectors	Brands or Examples	Equvalant in the U.S
Supermarket	Homeplus, Emart	Stop & Shop, Walmarket
Department	Lottee, Hyundai	Marcy’s, Nordstorm
Online Transaction	Online shopingmall, Food delivery apps	Amazon, Uber Eat
Electroincs	HiMart, Samsung Digital Plaza	Best Buy
Entertainment	entertainment establishment	Clubs
Face-to-Face Service	Message, face-to-face service	Message, Nail Salon
Leisure	Golf, Golf practice, Screen golf, singing room	Golf, Mini Golf
Gaming	Casino, Lottery, Game Room	Casino, Lottery, Game room
Gift Certificate	Gift certificate store
Jewerly	Jewerly	Jewerly
Tax and Fees	Governement, local tax, fees	governement, local tax, fees
Duty Free	Duty free stores	Duty free stores
Insurance	Life, National Pension, Health, Auto	Life, National Pension, Health, Auto
Auto-Transfer	Pubic Tranfortation, Telecommunication, other auto-transfer	Pubic Tranfortation, Telecommunication, other auto-transfer
Others	Adults, Gun Shop	Adults, Gun Shop

This restirction are commonly used for most government stimulus payments in Korea, such as susidies for child-care and delivery.

Seoul Stimulus Payment

Seoul Metroplitan Government provided stimulus payments for qualifying households in Seoul. The office name of this payment is “Seoul Metropolitan Government Emergency DisasterRelief Fund.”

Eligibiliy for Seould Stimulus Payments

Households whose annual incomse lower 100 % of the standard median incomes, the standard income that the Ministry of Health and Welfare sets up annually as a standard for welfare programs in Korea.

Number of Members in Household	1	2	3	4	5	6
Monthly Income (KRW)	1,757,194	2,991,980	3,870,577	4,749,174	5,627,771	6,506,368
Monthly Income (USD)	1531.99	2608.53	3374.52	4140.52	4906.51	5672.51

For example, a household with 4 members is eligible if the monthly incomes is less than $ 4140.52 (USD).

Period of Seoul Stimulus Payment

The application period for Seoul stimulus payment was from March 11 to May 15, 2020. The experation date is August 30, 2020.

Amounts of Seoul Stimulus Payment

Number of Members in Household	1~2	3~4	over 5
Amount (KRW)	300,000	400,000	500,000
Amount (USD)	261.55	348.74	435.92

Eligible individuals can receive two stimulus payments (KG and Seoul stimulus payments) together.

Forms of Seoul Stimulus Payment

Two forms of this stimulus payments are 1) gift certificates, and 2) pre-paid credit cards. The details of those forms are the same as the above.

Restrictions on Stores

The same restrictions are applied to Seoul stimulus payments. One additional restriction is that individuals can only use their payments at relevant stores in Seoul.

Model

This paper uses discrete choice models to analyze how stimulus payments during COVID-19 changes consumer behavior, related to eating out (either in stores or take-outs). The availability of data sets makes series of assumptions:

Aggregate data per restaurant categories 1: Only aggregated levels of data of restaurants in the market are available. All resturants are sorted into one of the following cateogries:
- Korean
- Chinesses
- Japanness
- Western
- Chichken
- Fast food
- Snack
- Bakery
- Coffee & Beverage
- Pub & Bar.
Aggregate data per restaurant categories 2: Since all information of resturants is aggregated into the above categories, information of each restuarnts is not available. The following information for each restaurant category is given:

Number of Restaurants
Number of Checks (total, based on time period in a days, days of a week, etc)
Estimates of Sales (total, based on time period in a days, days of a week, etc)

Check per one household: Resturant sales are measured by combined all checks related to credit card and cash transaction. It is possible that multiple individuals or households appear on a check. However, given the data sets, I cannot capture this.
Sales per check: Even though sales per check vary with bundles of food and drinks ordered, this measures how much consumer spend aaverage spending per transaction (per visit). In addition, this measure is widely used in the restaurant industry to measure consumer spending along with another measure, sales per cover (total sales dividied by the number of consumers at the resturants). This can be used as average prices for each resturant category.

Demand

We use the logit model to capture consumer decision to buy a restaurant meal. A baseline model is the following (Model 2) - Consumer $i$ choose $j (j=1, \cdots, J)$ at market $m (m=1, \cdot, M)$ at time $t(t=1,\cdots, 14)$:

\[ u_{ij}^{mt} = \alpha p_{j}^mt + X_{j}^{mt} \beta + \beta_y Y^{mt} + \xi_{j}^{mt} + \epsilon_{ij}^{mt} \]

where * $p_j^{mt}$: Average spendings per check in restaurant category $j$ * $X_j$ includes the number of restuants for the restaurant category * $Y^{mt}$ : the median incomes at the market level * $\xi_j$: the unobserved product chacteristics for $j$ * $\epsilon$: the demand shock; it follows the Type I extreme value distribution.

To see how stimulus payments affect consumer behavior, we use a specification of $y$ :

\[ Y^{mt} = y^{mt} + D_{sp}^mt \cdot sp \]

where $y_{mt}$ represents the aveage income in market $m$ at time $t$, $D_s$ is an indicator variables for stimulus payments, and $sp$ is the average stimulus payments.

The average stimulus payments could be differ on consumers’ income(Seould stimulus payments target for those whose income is below the standard median income. We will add additional specification on income later.

Using Model 1, I crete alternative specifications:

Model 2: \[ u_{ij}^{mt} = \alpha \frac{p_{j}^mt}{Y^{mt}} + \beta X_{j}^{mt} + \beta_y Y^{mt} + \xi_{j}^{mt} + \epsilon_{ij}^{mt} \]

Model 3: Nest Logit \[ u_{ij}^{mt} = \alpha \frac{p_{j}^mt}{Y^{mt}} + \beta X_{j}^{mt} + \beta_y Y^{mt} + \xi_{j}^{mt} + (1-\sigma)\epsilon_{ij}^{mt} \]

Possible nests would be time, and type of meal types.

Model 4: Random Coefficients

\[ u_{ij}^{mt} = \alpha_i p_{j}^mt + \beta X_{j}^{mt} + \xi_{j}^{mt} + \epsilon_{ij}^{mt} \]

where $\alpha_i = \alpha + \sigma Y_i^{mt}$ and $Y_i^{mt}$ follows the normal distribution.

$Y_i^{mt}$ varies with different average incomes over time and across markets. In addition, after the stimulus payments are transferred to consumers, this should be changeed based on the stimulus payment distributino rules.

Supply Side: Entry/Exit Decision

Another effect of the stimulus payments is to help local business owners. This is indirect effects resulting from the increase in consumer spending due to the stimulus payments. To estimate the effect on local business owners, we can use either discrete choice models that captures the probility of exit or entry/exit models that consider an agent decision as well as other players’ decision (discrete games)).

To use the discrete choice model, we need to have infomation of individual owners’ decisions of exit from the market. However, this information is not available in our current data sets. Additional data collection would be required.

For entry/exit games(strategic interactions), we would use the early models of entry/exit models: Bresinhan and Ressis (1991, 1994) or Seim (2006). Since all firm information is aggregated to the restaurant category, it would be easier to start with the model of Bresnahan and Ressis (1991). With this model, we will estimate entry threshold.

However, except the Bresinhan and Ressis model, the information of entry/exit at the firm level is necessary. Thus, additional information will be needed.

Data

Varibles information

The following website shows the information of available variables and their definition in Korean and English: Variable Description

Income

The following table shows variables related to estimated income and expeditures.

Variable Names	Description
MT_AVRG_INCOME_AMT	Monthly average income
INCOME_SCTN_CD	Income catergory code
EXPNDTR_TOTAMT	Total expenditure
FDSTFFS_EXPNDTR_TOTAMT	Grocery expenditure
CLTHS_FTWR_EXPNDTR_TOTAMT	Clothing shoes expenditure
LVSPL_EXPNDTR_TOTAMT	Living expenditure
MCP_EXPNDTR_TOTAMT	Medical expenditure
TRNSPORT_EXPNDTR_TOTAMT	Transportation expenditure
LSR_EXPNDTR_TOTAMT	Leisure expenditure
CLTUR_EXPNDTR_TOTAMT	Culture expenditure
EDC_EXPNDTR_TOTAMT	Education expenditure
PLESR_EXPNDTR_TOTAMT	Entertainmalet expenditure

MT_AVRG_INCOME_AMT: Monthly average household incomes collected by the National Health Insurance Service managing a univerisal health insurance serivice for all Koreans. This is funded by contribution of employees and empolyers, Korean governments and tobaco surcharges. For the contributions, employees contributes 6.67% of their monthly salaries: a half paid by employers and another half debuted from employees’ salaries. If one is not employed, their perimum are determined by their values of real estates.
Since Income information is only available at small markets, we analyze these markets and their consumers.

Number of Households

This information of the number of household is important in estimating the amount of the stimulus payments are tranferred to each households. Since the current data does not include this information, we use the average household information of the districts.

load(file="df_all5.Rdata")
# sum(is.na(df_all5$TOT_HH)) # TOT_HH is the number of the household in "Dong", which is bigger that the market defined in the data set. 

df_all5$HHR_1 <- df_all5$HH_1/df_all5$TOT_HH
df_all5$HHR_2 <- df_all5$HH_2/df_all5$TOT_HH
df_all5$HHR_3 <- df_all5$HH_3/df_all5$TOT_HH
df_all5$HHR_4 <- df_all5$HH_4/df_all5$TOT_HH
df_all5$HHR_5 <- with(df_all5, (HH_5+HH_6+HH_7+HH_8+HH_9+HH_10)/TOT_HH)
df_all5$HHR_6 <- with(df_all5, (HH_5+HH_6+HH_7+HH_8+HH_9+HH_10)/TOT_HH)

# Table
table_hhr <- unique(df_all5[, c("TRDAR_CD_NM","HHR_1", "HHR_2", "HHR_3", "HHR_4", "HHR_5")])


table_hhr %>%
  ggplot(aes(x=HHR_1)) + geom_histogram() +
  labs(title="Distribution of Household with One person", 
       x="Percentage of One-Person Household", y="Count") + 
  scale_color_brewer(palette="Dark2")

table_hhr %>%
  ggplot(aes(x=HHR_2)) + geom_histogram() +
  labs(title="Distribution of Household with Two persons", 
       x="Percentage of Two-Person Household", y="Count") + 
  scale_color_brewer(palette="Dark2")

table_hhr %>%
  ggplot(aes(x=HHR_3)) + geom_histogram() +
  labs(title="Distribution of Household with Three persons", 
       x="Percentage of Three-Person Household", y="Count") + 
  scale_color_brewer(palette="Dark2")

table_hhr %>%
  ggplot(aes(x=HHR_4)) + geom_histogram() +
  labs(title="Distribution of Household with Four or More persons", 
       x="Percentage of Four-Person Household", y="Count") + 
  scale_color_brewer(palette="Dark2")

table_hhr %>%
  ggplot(aes(x=HHR_5)) + geom_histogram() +
  labs(title="Distribution of Household with Four or More persons", 
       x="Percentage of Five or More Persons Household", y="Count") + 
  scale_color_brewer(palette="Dark2")

summary(table_hhr[, c("HHR_1", "HHR_2", "HHR_3", "HHR_4", "HHR_5")])

##      HHR_1            HHR_2            HHR_3            HHR_4        
##  Min.   :0.0771   Min.   :0.1000   Min.   :0.0375   Min.   :0.02188  
##  1st Qu.:0.3604   1st Qu.:0.1951   1st Qu.:0.1407   1st Qu.:0.10220  
##  Median :0.4337   Median :0.2166   Median :0.1676   Median :0.13341  
##  Mean   :0.4444   Mean   :0.2105   Mean   :0.1642   Mean   :0.13765  
##  3rd Qu.:0.5110   3rd Qu.:0.2325   3rd Qu.:0.1919   3rd Qu.:0.16553  
##  Max.   :0.8208   Max.   :0.2847   Max.   :0.2807   Max.   :0.42196  
##      HHR_5         
##  Min.   :0.006802  
##  1st Qu.:0.032865  
##  Median :0.041740  
##  Mean   :0.043223  
##  3rd Qu.:0.052215  
##  Max.   :0.114907

avg_hhr <- as.data.frame(colMeans(table_hhr[, c("HHR_1", "HHR_2", "HHR_3", "HHR_4","HHR_5")]))

colnames(avg_hhr) <- c("Average Percent")
avg_hhr$HHR <- c("One", "Two", "Three", "Four", "Five or More")

avg_hhr <- avg_hhr[, c("HHR", "Average Percent")]
row.names(avg_hhr) <- NULL
avg_hhr

##            HHR Average Percent
## 1          One      0.44443378
## 2          Two      0.21046482
## 3        Three      0.16422473
## 4         Four      0.13765393
## 5 Five or More      0.04322274

Since most decisions related to distributing stimulus payments are based on the monthly income in Feburary, 2020. We use the incomes of Feburary, 2020, as reference incomes: the income of the first quarter of 2020.

# summary(df_all5$MT_AVRG_INCOME_AMT)

table_income <- unique(df_all5[, c("STDR_YY_CD", "STDR_QU_CD", "TRDAR_CD_NM", "MT_AVRG_INCOME_AMT")])

table_income$time.id <- table_income %>% group_indices(STDR_YY_CD, STDR_QU_CD)

df.income <- table_income %>% 
  group_by(STDR_YY_CD, STDR_QU_CD, time.id) %>% 
  summarize(mean = mean(MT_AVRG_INCOME_AMT*0.001)) %>%
  mutate(time = paste0(STDR_YY_CD-2000, "Q", STDR_QU_CD))

ggplot(df.income, aes(x=as.factor(time) , y=mean, group=1)) + 
    geom_line() + 
    labs(title="Change in Avereage Montly Income", x="Time", 
       y="Average Montly Income (KWN, 1000s)", 
       caption ="Staring July 3, 2018, employees should join the National Health Insurance if they work more than 8 days(60 hours) per month.") +
    theme( plot.caption = element_text(hjust=0, size=9),
           plot.caption.position = "plot")

table_income %>%
  filter(STDR_YY_CD==2020, STDR_QU_CD==1) %>%
  ggplot(aes(x=MT_AVRG_INCOME_AMT*0.001)) + geom_histogram() +
  geom_vline(xintercept = 3086, linetype="dotted", 
                color = "blue") +
  geom_vline(xintercept = 2992, linetype="dotted", color="red")+
  geom_text(aes(x=3500, y=100, label="Means of Income = 3,086"), colour="blue", 
            vjust= 1.2, angle=0, size=3) +
  geom_text(aes(x=3500, y=110, label="Standard Median Income = 2,992"), colour="red", 
            vjust= 1.2, angle=0, size=3) +
  labs(title="Distribution of Montly Income in 2020 at the Market Level", x="Monthly Income", y="No. of Markets", 
       caption ="Average household size is 2.082; The standard median income (red) is based on two-person households.")

Variation in Stimulus Payments

While the nation stimulus payments are transferred to all Koreans, regardless of their incomes, Seoul stimulus payments are provided to those who are qualified: household incomes are below the standard median income set by the Korean governement.

To estimate appropriate amounts of Seoul stimulus payments provided to qualified individuals, we need to know about the distribution of the montly incomes. The current data only gives the average montly income. To obtain estimated distributions of household incomes, we need addtional information of standard deviation and skewness of this distribution since most income distributions tend to be positively skewed.

According to a report by Shinhan Bank, one of the five biggest banks in South Korea, in 2018 the ratio of standard deviation to average in household income in Seoul (Standard Deviation / Mean) is 1.38.(Standard Deviation to Average in Seoul: 1.38). Using this ratio, we could calculate means and standard deviations for each market and create a normal distribution. Since the standard deviations are great than the means, there are many observations whose incomes are less than zero. addtional information or theories is needed to estimate the distribution of the income.

For example, we might need to create the income distribution as the following graph.

xbar <- 200
xsd  <- 200 *1.38
hist(rnorm(200, mean=xbar, sd=xsd))

# skewness = 3, kutyosis = 20

library(PearsonDS)
moments <- c(mean = 200,variance = xsd^2,skewness = 3, kurtosis = 20)
eg <- rpearson(200, moments = moments)
hist(eg)

## Sales Data The informatio of the sales in our data set is approximated sales based on three major credit card firms: KB, SH, and BC. Using the weights of markets and resturant categories, the estimated sales are obtained.

Even though some parts of cash transaction and Zero Payment, the gift certificate form of the stimulus payment, are considered as in the weights, it is not clear that all transactions of these two payment types are included since the estimated sales are based on the credit card transaction.

We use average sales per ticket, the amount of money used for each transaction. In order to use the average sales per ticket as an alternative of prices, this should be consistent over time. The following figures show that the average sales per ticket are consistent for all restaurant categories, except Pre-Cook Meal.

# Average sales per category
df_all5 %>% 
  group_by(STDR_YY_CD, STDR_QU_CD, SVC_DES) %>%
  summarize(avg.sales = mean(THSMON_SELNG_AMT, na.rm = T)) %>%
  mutate(time = paste0(STDR_YY_CD-2000, "Q", STDR_QU_CD)) %>%
  ggplot(aes(x=as.factor(time), y=avg.sales/10000/3, group=SVC_DES, color=SVC_DES)) +
    geom_line() +
    labs(title="Aggregate Monthly Sales Per Category", y="Sales Per Categories")

# Average sales per store per catergory
df_all5 %>% 
  group_by(STDR_YY_CD, STDR_QU_CD, SVC_DES) %>%
  summarize(avg.sales = mean(THSMON_SELNG_AMT, na.rm = T)) %>%
  mutate(time = paste0(STDR_YY_CD-2000, "Q", STDR_QU_CD)) %>%
  ggplot(aes(x=as.factor(time), y=avg.sales/10000/3, group=SVC_DES, color=SVC_DES)) +
    geom_line() +
    labs(title="Aggregate Monthly Sales Per Category", y="Sales Per Categories")

# Sales per ticket
df_all5$sales.ticket <- with(df_all5, THSMON_SELNG_AMT/THSMON_SELNG_CO)

df_all5 %>% 
  group_by(STDR_YY_CD, STDR_QU_CD, SVC_DES) %>%
  summarize(avg.sales.ticket = mean(sales.ticket, na.rm = T)) %>%
  mutate(time = paste0(STDR_YY_CD-2000, "Q", STDR_QU_CD)) %>%
  ggplot(aes(x=as.factor(time), y=avg.sales.ticket/10000/3, group=SVC_DES, color=SVC_DES)) +
    geom_line() +
    labs(title="Aggregate Monthly Sales Per Ticket", 
         y="Sales Per Tickets", x="Time", color="Restaurant Category")

Sales Amount Per time period

THSMON_SELNG_AMT
TMZON_00_06_SELNG_AMT
TMZON_06_11_SELNG_AMT
TMZON_11_14_SELNG_AMT
TMZON_14_17_SELNG_AMT
TMZON_17_21_SELNG_AMT
TMZON_21_24_SELNG_AMT

Sales counts(Tickets)

THSMON_SELNG_CO
TMZON_00_06_SELNG_CO
TMZON_06_11_SELNG_CO
TMZON_11_14_SELNG_CO
TMZON_14_17_SELNG_CO
TMZON_17_21_SELNG_CO
TMZON_21_24_SELNG_CO

# Sales
summary(df_all5$THSMON_SELNG_AMT)

##      Min.   1st Qu.    Median      Mean   3rd Qu.      Max.      NA's 
## 1.190e+02 4.481e+07 1.303e+08 3.260e+08 3.659e+08 1.815e+10     46186

summary(df_all5$TMZON_00_06_SELNG_AMT)

##       Min.    1st Qu.     Median       Mean    3rd Qu.       Max.       NA's 
##     -21471          0    2049784   27192042   23801559 2584930149      46186

summary(df_all5$TMZON_06_11_SELNG_AMT)

##       Min.    1st Qu.     Median       Mean    3rd Qu.       Max.       NA's 
##  -26775477          0    1473462   19934579   16095443 5317563946      46186

summary(df_all5$TMZON_11_14_SELNG_AMT)

##       Min.    1st Qu.     Median       Mean    3rd Qu.       Max.       NA's 
##   -4834906    3798456   21906780   63555147   67260835 6649092044      46186

summary(df_all5$TMZON_14_17_SELNG_AMT)

##       Min.    1st Qu.     Median       Mean    3rd Qu.       Max.       NA's 
##   -4421861    3769524   14988892   38953489   45487996 7086169302      46186

summary(df_all5$TMZON_17_21_SELNG_AMT)

##       Min.    1st Qu.     Median       Mean    3rd Qu.       Max.       NA's 
##   -7734606   13224585   40394108  107935256  111145610 8566333243      46186

summary(df_all5$TMZON_21_24_SELNG_AMT)

##      Min.   1st Qu.    Median      Mean   3rd Qu.      Max.      NA's 
## 0.000e+00 3.258e+06 1.970e+07 6.843e+07 7.397e+07 4.954e+09     46186

# No. of ticketes
summary(df_all5$THSMON_SELNG_CO)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##       1    2411    7420   24514   24783  773249   46186

summary(df_all5$TMZON_00_06_SELNG_CO)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##      -6       0      66    2069     848  154798   46186

summary(df_all5$TMZON_06_11_SELNG_CO)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##     -25       0      98    2981    1622  152588   46186

summary(df_all5$TMZON_11_14_SELNG_CO)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##      -5     243    1720    5141    5981  242656   46186

summary(df_all5$TMZON_14_17_SELNG_CO)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##     -10     224    1027    3596    3850  197177   46186

summary(df_all5$TMZON_17_21_SELNG_CO)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##       0     731    2375    6768    7413  215846   46186

summary(df_all5$TMZON_21_24_SELNG_CO)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##      -1     165     865    3960    3323  161272   46186

Estimation

Key Issues of Estimation

1. Highly Aggregated at the market level

The information of sales and stores are aggregated to the restaurant category at the market level. For example, the sum of the sales of all Korean restaurants are available, while that of individual restaturents in this category is not available.

2. Limited information of income distribution and households

The informatio of the income distribution at the market level is important so that we can calculate how much stimulus payments are transferred to in

Update: COVID-19

Jung Hwan Koh

2020 11 16

Introduction

Stimulus Payments in S. Korea During COVID-19

Korean Government Stimulus Payments (KG Stimulus Payments)

Period of Application for KG Stimulus Payments

Amounts of Korean Gov.Stisdfss

Forms fo Stimulus Payments

Restrictions on Stores

Seoul Stimulus Payment

Eligibiliy for Seould Stimulus Payments

Period of Seoul Stimulus Payment

Amounts of Seoul Stimulus Payment

Forms of Seoul Stimulus Payment

Restrictions on Stores

Model

Demand

Supply Side: Entry/Exit Decision

Data

Varibles information

Income

Number of Households

Variation in Stimulus Payments

Estimation

Key Issues of Estimation

1. Highly Aggregated at the market level

2. Limited information of income distribution and households

3. Prices as sales per ticket (check)

Demand Side

Supply Side