如何为分组样本数据创建符合条件的税率变量

Got it, let's work through this together. Your panel dataset has country-year observations, and we need to create a tax rate variable that hits all three of your requirements. Since you're already using data.table, we'll lean into its fast grouping capabilities to make this clean and efficient.

Step 1: Prep the Data (and Stick with Data.Table) #

First, let's convert your data back to a data.table (it's far more efficient for grouping operations):

DT <- as.data.table(DT)

Step 2: Assign a Base Tax Rate to Each Country #

To ensure the same country's yearly rates never differ by more than 5%, we'll start with a base rate for each country. We'll set this base between 10% and 35%—this way, even if we adjust up or down by 5%, we'll never go outside your 10%-40% range.

# Get unique countries in your dataset
unique_countries <- unique(DT$country)

# Generate base tax rates (10-35%, with 0.1% precision)
set.seed(123) # Keep results reproducible
country_base_tax <- data.table(
  country = unique_countries,
  base_tax = sample(seq(10, 35, by = 0.1), length(unique_countries), replace = TRUE)
)

Step 3: Generate Yearly Tax Rates for Each Country-Year Group #

Now we'll merge the base rates into your dataset, then generate a unique tax rate for each country-year pair—ensuring it's within ±5% of the country's base rate, and all observations in the same country-year get the same rate.

# Merge base rates into the main dataset
DT <- DT[country_base_tax, on = "country"]

# Generate tax rates per country-year group
set.seed(123)
DT[, tax_rate := sample(seq(
  max(10, base_tax - 5),  # Ensure we don't go below 10%
  min(40, base_tax + 5),  # Ensure we don't go above 40%
  by = 0.1
), 1), by = .(country, year)]

# Clean up: remove the base_tax column if you don't need it
DT[, base_tax := NULL]

Verify Your Conditions #

Let's double-check that all requirements are met:

• Same rate per country-year: Run DT[, .N, by = .(country, year, tax_rate)]—each group's N should match the number of observations for that country-year.
• Rates between 10-40%: Run range(DT$tax_rate)—you'll see it's bounded correctly.
• Yearly difference ≤5% per country: Run DT[, diff(range(tax_rate)), by = country]—all values will be ≤5.

Tweaks if Needed #

• If you want integer tax rates, change by = 0.1 to by = 1 in the seq() calls.
• Adjust the precision (e.g., by = 0.01 for two decimal places) if you need more granular rates.

内容的提问来源于stack exchange，提问作者Tom