Introduction

Research Context

Over the past several years, U.S. households have faced rising food costs, particularly since the COVID-19 pandemic and subsequent supply chain disruptions.

Key Issues:

Food inflation has consistently outpaced general inflation
Families adjust by purchasing cheaper foods or reducing spending on fresh produce
Rising food insecurity alongside higher grocery prices

This Analysis:

Period: 2019-2023 (5 years including pandemic)
Focus: Understanding how food price inflation affects American households
Data: FRED (food prices) + USDA CPS-FSS (food security)

Research Questions

Three Key Questions

How do changes in food prices relate to food insecurity rates?
- Direct relationship between inflation and household food security
Does food inflation correlate with household spending pattern shifts?
- Do people change their eating habits (home vs. restaurant)?
Is there a lagged relationship between inflation and food insecurity?
- Do effects appear immediately or over time?

Data Preparation

Data Sources & Processing

# Download food price data from FRED
food_home <- fredr(series_id = "CUSR0000SAF11", 
                   observation_start = as.Date("2018-01-01"),
                   observation_end = as.Date("2023-12-31"))

food_away <- fredr(series_id = "CUSR0000SEFV", 
                   observation_start = as.Date("2018-01-01"),
                   observation_end = as.Date("2023-12-31"))

# Calculate monthly YoY inflation on full series
monthly_home <- food_home %>%
  arrange(date) %>%
  mutate(
    inflation_home = (value / lag(value, 12) - 1) * 100,
    year = lubridate::year(date)
  )

monthly_away <- food_away %>%
  arrange(date) %>%
  mutate(
    inflation_away = (value / lag(value, 12) - 1) * 100,
    year = lubridate::year(date)
  )

# Compute annual averages for 2019-2023
annual_inflation <- monthly_home %>%
  filter(year >= 2019, year <= 2023) %>%
  group_by(year) %>%
  summarise(avg_inflation_home = mean(inflation_home, na.rm = TRUE), .groups = "drop")

inflation_away <- monthly_away %>%
  filter(year >= 2019, year <= 2023) %>%
  group_by(year) %>%
  summarise(avg_inflation_away = mean(inflation_away, na.rm = TRUE), .groups = "drop")

# Combine inflation data
inflation_data <- annual_inflation %>%
  left_join(inflation_away, by = "year") %>%
  mutate(avg_inflation_total = (avg_inflation_home + avg_inflation_away) / 2)

Data Preparation (cont.)

Food Security Data

# Load food security data (combined from CPS-FSS annual files)
food_security <- read_csv("food_security_annual_summary.csv", show_col_types = FALSE) %>%
  filter(year >= 2019, year <= 2023)

# Merge all data
data <- inflation_data %>%
  left_join(food_security, by = "year")

Combined Dataset: Food Inflation and Food Security (2019-2023)
Year	Avg Inflation (%)	Food Insecure (%)	Food Secure (%)	Very Low Security (%)
2019	1.98	10.78	88.99	3.61
2020	3.42	11.65	88.10	3.76
2021	4.00	10.36	89.51	3.44
2022	9.52	13.38	86.40	4.88
2023	6.12	14.16	85.72	4.92

Note: Food security data collected from ~120,000 households per year via CPS Food Security Supplement

Overview: 2019-2023 Trends

The Story in Numbers

Food Inflation:

2019: 1.98% (baseline)
2020: 3.42% (pandemic onset)
2021: 4.00% (supply chain)
2022: 9.52% PEAK
2023: 6.12% (moderating)

Food Insecurity:

2019: 10.8%
2020: 11.6%
2021: 10.4%
2022: 13.4%
2023: 14.2% HIGHEST

+3.4 percentage points from 2019 to 2023

Question 1: Food Prices & Food Insecurity

How do changes in food prices relate to food insecurity?

# Correlation
cor_result <- cor(data$avg_inflation_total, data$food_insecure_pct, 
                  use = "complete.obs")

# Regression
model1 <- lm(food_insecure_pct ~ avg_inflation_total, data = data)

Key Statistics:

Correlation: r = 0.753 (Strong positive)
R² = 0.567 (57% of variation explained)
Coefficient: β = 0.423

Interpretation: For every 1% increase in food inflation, food insecurity increases by approximately 0.42 percentage points.

Question 1: Time Trends

Question 1: Correlation Plot

Question 1: Summary

Key Findings

Strong Positive Correlation (r = 0.753)

Food inflation and food insecurity move together
Relationship is substantial and immediate
2022 inflation spike (9.52%) → highest insecurity (13.4%)

Regression Results

R² = 0.567: inflation explains 57% of food insecurity variation
Coefficient = 0.42: 1% inflation increase → 0.42pp insecurity increase
2022 case: inflation increased 5.5pp → predicted 2.3pp insecurity increase

Limitations

Small sample size (n=5) limits statistical significance (p=0.14)
COVID-19 pandemic may confound relationship
Cannot establish pure causality

Question 2: Spending Patterns

Does food inflation correlate with spending shifts?

# Download spending data
pce_food <- fredr(series_id = "DFOORC1A027NBEA",
                  observation_start = as.Date("2019-01-01"),
                  observation_end = as.Date("2023-12-31"))

pce_services <- fredr(series_id = "DFSARC1A027NBEA",
                      observation_start = as.Date("2019-01-01"),
                      observation_end = as.Date("2023-12-31"))

# Calculate annual spending
spending <- pce_food %>%
  mutate(year = year(date)) %>%
  filter(year >= 2019, year <= 2023) %>%
  group_by(year) %>%
  summarise(food_home_spending = mean(value), .groups = "drop")

services <- pce_services %>%
  mutate(year = year(date)) %>%
  filter(year >= 2019, year <= 2023) %>%
  group_by(year) %>%
  summarise(food_away_spending = mean(value), .groups = "drop")

# Combine and calculate ratio
spending_data <- spending %>%
  left_join(services, by = "year") %>%
  left_join(inflation_data, by = "year") %>%
  mutate(
    home_away_ratio = (food_home_spending / food_away_spending) * 100,
    pct_change_ratio = (home_away_ratio / lag(home_away_ratio) - 1) * 100
  )

Question 2: Spending Data

Home vs. Away Food Spending Ratio
Year	Inflation (%)	Home/Away Ratio (%)	Change from Prior Year (%)
2019	1.98	2.65	NA
2020	3.42	2.89	9.23
2021	4.00	2.45	-15.28
2022	9.52	2.39	-2.41
2023	6.12	2.38	-0.23

# Correlation
cor_spending <- cor(spending_data$avg_inflation_total, 
                    spending_data$home_away_ratio, 
                    use = "complete.obs")

Key Statistics:

Correlation: r = -0.644 (Moderate negative)
2019-2023 ratio change: -9.9%

Question 2: Time Trends

Question 2: Correlation Plot

Question 2: Summary

Key Findings

Negative Correlation (r = -0.644)

As inflation increased, home/away ratio decreased
Contrary to traditional economic theory
Americans did NOT shift to home cooking during inflation surge

The Pattern:

2020: COVID spike in ratio (2.89%) - forced home cooking
2021-2023: Steady decline to 2.38% despite 9%+ inflation
2022 paradox: Highest inflation, yet lowest ratio

Possible Explanations:

Post-pandemic “revenge spending” on dining experiences
Strong wage growth (2021-2023) enabled sustained restaurant spending
Restaurant dining became a priority/non-negotiable expense
Pent-up demand overwhelmed price sensitivity

Question 3: Lagged Effects

Is there a lagged relationship?

# Create lagged variables
lagged_data <- data %>%
  arrange(year) %>%
  mutate(
    inflation_lag1 = lag(avg_inflation_total, 1),
    inflation_lag2 = lag(avg_inflation_total, 2)
  )

# Calculate correlations for different lags
cors <- tibble(
  Lag = c("Same Year", "1-Year Lag", "2-Year Lag"),
  Correlation = c(
    cor(lagged_data$avg_inflation_total, lagged_data$food_insecure_pct, 
        use = "complete.obs"),
    cor(lagged_data$inflation_lag1, lagged_data$food_insecure_pct, 
        use = "complete.obs"),
    cor(lagged_data$inflation_lag2, lagged_data$food_insecure_pct, 
        use = "complete.obs")
  ),
  N = c(5, 4, 3)
)

Question 3: Lag Comparison

Correlations at Different Time Lags
Time Lag	Correlation	Sample Size
Same Year	0.753	5
1-Year Lag	0.734	4
2-Year Lag	0.996	3

Question 3: Current vs. Lagged

Question 3: Summary

Key Findings - Effects Are Immediate

Contemporaneous Effects Dominate

Same year correlation: r = 0.753 (n=5)
1-year lag: r = 0.734 (n=4) - similar strength
2-year lag: r = 0.996 (n=3) - unreliable due to small sample

What This Tells Us:

Food insecurity responds immediately to price changes
No evidence of delayed effects getting stronger
Households feel financial pressure right away
No “slow burn” effect detected

Important Caveats:

5-year window severely limits lag analysis
Need 10-15 years to study longer-term effects
2-year lag result likely spurious (only 3 data points)
COVID-19 creates structural break in data

Overall Conclusions

Main Takeaways

1. Strong Direct Relationship

Food inflation and food insecurity move together (r = 0.753)
Effect is immediate and substantial
2022 inflation spike demonstrates real-world impact
Each 1% inflation increase → ~0.4pp increase in food insecurity

2. Unexpected Spending Patterns

Negative correlation between inflation and home/away ratio
Post-pandemic behavior trumped price sensitivity
Restaurant spending remained resilient 2021-2023
COVID-19 created anomalous 2020 peak

3. Immediate, Not Delayed Effects

Impacts are contemporaneous, not lagged
Households respond immediately to price changes
No evidence of “slow burn” multi-year effects
Limited data precludes definitive lag conclusions

Policy Implications

Recommendations for Action

1. Immediate Intervention Required

Don’t wait for lagged effects to manifest
Food insecurity spikes happen simultaneously with price increases
Emergency assistance programs must scale rapidly

2. Proactive Program Scaling

Use inflation data as early warning system
Increase SNAP/food assistance benefits when inflation accelerates
Target interventions before households fall into crisis

3. Monitor Real-Time

Track food insecurity alongside price indices
Correlation strong enough to predict need
Monthly monitoring instead of annual surveys

4. Post-Pandemic Context

Consumer behavior may have permanently shifted
Traditional price elasticity models may not apply
Need updated understanding of food spending priorities

Study Limitations

Important Caveats

1. Small Sample Size

Only 5 years of data (2019-2023)
Limited statistical power
P-values not significant at α=0.05 level
Results should be considered exploratory

2. COVID-19 Confounding

2020-2021 not “normal” economic conditions
Pandemic created structural break in spending patterns
Difficult to isolate pure inflation effects
Recovery dynamics overlap with inflation surge

3. Temporal Aggregation

Annual data masks monthly variation
Critical inflation spikes may be smoothed out
Within-year dynamics lost

4. Single Outcome Measure

Food insecurity only; no health outcomes included
Cannot assess longer-term dietary quality impacts
Missing nutrition and health consequences

Technical Appendix

Data Sources & Methods

Data Sources:

CUSR0000SAF11: CPI Food at Home (FRED/BLS)
CUSR0000SEFV: CPI Food Away from Home (FRED/BLS)
DFOORC1A027NBEA: Real PCE Food at Home (FRED/BEA)
DFSARC1A027NBEA: Real PCE Food Services (FRED/BEA)
CPS Food Security Supplement: Annual survey (USDA-ERS/Census)

Sample Sizes:

CPS-FSS: ~120,000 households per year
Aggregated to 5 annual observations due to data availability
Food security rates calculated using household survey weights (HHSUPWGT)

Conclusion

Summary Statement

This analysis reveals a strong, immediate relationship between food price inflation and household food insecurity during the 2019-2023 period.

Key insights:

Clear impact: 9.5% inflation in 2022 → 14.2% food insecurity by 2023
Unexpected resilience: Restaurant spending maintained despite inflation
Immediate effects: No lag time between price rises and food security impacts

food price inflation to dietary behaviour - Presentation

Introduction

Research Context

Research Questions

Three Key Questions

Data Preparation

Data Sources & Processing

Data Preparation (cont.)

Food Security Data

Overview: 2019-2023 Trends

The Story in Numbers

Question 1: Food Prices & Food Insecurity

How do changes in food prices relate to food insecurity?

Key Statistics:

Question 1: Time Trends

Question 1: Correlation Plot

Question 1: Summary

Key Findings

Question 2: Spending Patterns

Does food inflation correlate with spending shifts?

Question 2: Spending Data

Key Statistics:

Question 2: Time Trends

Question 2: Correlation Plot

Question 2: Summary

Key Findings

Question 3: Lagged Effects

Is there a lagged relationship?

Question 3: Lag Comparison

Question 3: Current vs. Lagged

Question 3: Summary

Key Findings - Effects Are Immediate

Overall Conclusions

Main Takeaways

1. Strong Direct Relationship

2. Unexpected Spending Patterns

3. Immediate, Not Delayed Effects

Policy Implications

Recommendations for Action

1. Immediate Intervention Required

2. Proactive Program Scaling

3. Monitor Real-Time

4. Post-Pandemic Context

Study Limitations

Important Caveats

1. Small Sample Size

2. COVID-19 Confounding

3. Temporal Aggregation

4. Single Outcome Measure

Technical Appendix

Data Sources & Methods

Data Sources:

Sample Sizes:

Conclusion

Summary Statement

Thank You