Chicago Fed National Activity Index: Diffusion Index (CFNAIDIFF) Comparison

• Class: ECON 6635-01

• School: Pompea College of Business

• University: University of New Haven

• Author: Sai Rahul Chirra

Introduction

To understand the dynamics of the U.S. economy, it is essential to consider both national and localized perspectives. This analysis develops a diffusion index by incorporating employment, industrial production, and consumer sentiment data. The resulting index is then compared with the Chicago Fed National Activity Diffusion Index (CFNAIDIFF), offering valuable insights into patterns of economic growth and contraction over time.

Data selection and Diffusion Index Construction.

The diffusion index reflects economic expansion and contraction based on pulling data for three selected variables. Here selected variables are Employment, industrial production, and housing starts.

suppressWarnings({
  suppressPackageStartupMessages({
    library(tidyverse)
    library(quantmod)
    library(tsbox)
    library(zoo)
  })
})

options(digits = 3, scipen = 99999)
graphics.off()

# Load economic variables
getSymbols(c("PAYEMS", "INDPRO", "HOUST"),
           src = "FRED", return.class = 'xts', 
           from = "2010-01-01", to = Sys.Date())

[1] "PAYEMS" "INDPRO" "HOUST" 

Comparison of Custom Diffusion Index and CFNAIDIFF

• Correlation Coefficient:

  • The correlation between your custom Diffusion Index and the Chicago Fed National Activity Diffusion Index (CFNAIDIFF) indicates the extent to which the two series move together.

  • A strong positive correlation (close to 1) would imply that your index is effectively capturing the same economic dynamics as CFNAIDIFF.

  • For example, if the correlation is 0.8 or higher, it suggests that your index is a good proxy for CFNAIDIFF, reflecting similar economic trends over the analyzed period.

# Data preprocessing
employment <- PAYEMS
industrial_production <- INDPRO
housing  <- HOUST

employment_ss <- employment["2010-01-31/2024-09-01"] |> ts_ts()
industrial_ss <- industrial_production["2010-01-31/2024-09-01"] |> ts_ts()
housing_ss <- housing["2010-01-31/2024-09-01"] |> ts_ts()

mydata <- cbind.data.frame(employment_ss, industrial_ss, housing_ss)

mydf <- mydata %>%
  mutate(
    emp_diff = tsibble::difference(employment_ss, differences = 1),
    ind_diff = tsibble::difference(industrial_ss, differences = 1),
    house_diff = tsibble::difference(housing_ss, differences = 1)
  ) %>%
  dplyr::select(emp_diff, ind_diff, house_diff) %>%
  na.omit()

Visual Observations (First Plot):

• General Alignment: There are periods where the trends of the custom Diffusion Index and CFNAIDIFF align closely. This suggests that your methodology for constructing the index is capturing key aspects of economic activity, particularly in times of economic growth or decline.

• Divergence Periods: At certain points, the indices deviate from each other (e.g., the blue line moves significantly above or below the orange line). These divergences could result from differences in the variables included in your index versus those in CFNAIDIFF, such as:

  • CFNAIDIFF uses a broader range of variables beyond employment, industrial production, and consumer sentiment.

  • Your index may have specific sensitivities, such as being more influenced by employment or housing data.

• Post-September Observations:

  • Since CFNAIDIFF data ends in September 2024, your custom index provides valuable insights into recent economic activity.

  • If your index shows a decline after September, it could indicate a weakening economy, while stability or growth would suggest resilience.

# Construct diffusion index
mydf_mat <- apply(mydf, 2, sign)
pos <- apply(mydf_mat, 1, function(row) sum(row > 0))
neg <- apply(mydf_mat, 1, function(row) sum(row < 0))
tot <- pos + neg
index <- (pos / tot - neg / tot) * 100
ma_index <- rollmean(index, 7, align = "right", na.pad = TRUE)

Date <- seq.Date(from = as.Date("2010-05-01"), length.out = length(index), by = "month")
diffusion_df <- cbind.data.frame(Date, index, ma_index)

ggplot(diffusion_df, aes(x = Date, y = index)) +
  # Line for the diffusion index
  geom_line(color = "navyblue", size = 0.8) +
  # Smoothed trend line with transparency
  geom_smooth(color = "red", fill = "lightblue", alpha = 0.3, size = 1.2) +
  # Adding labels to key points
  geom_text(data = diffusion_df[c(1, nrow(diffusion_df)), ],
            aes(label = paste0(round(index, 2), "%")), 
            vjust = -1, size = 3.5, fontface = "bold", color = "darkred") +
  # Adding a horizontal reference line
  geom_hline(yintercept = 0, linetype = "dashed", color = "black", size = 0.8) +
  # Vertical line for a key event (e.g., COVID-19)
  geom_vline(xintercept = as.Date("2020-03-01"), linetype = "dotted", color = "darkgreen", size = 1) +
  annotate("text", x = as.Date("2020-03-01"), y = -90, 
           label = "COVID-19", color = "darkgreen", size = 4, hjust = 0, angle = 90) +
  # Title, subtitle, and axis labels
  labs(
    title = "U.S. Economic Diffusion Index Over Time",
    subtitle = "Analyzing economic trends",
    x = "Year",
    y = "Diffusion Index (%)"
  ) +
  # Adjust y-axis limits to zoom out
  scale_y_continuous(limits = c(-120, 120)) +
  # Enhance visual appearance with themes
  theme_minimal(base_size = 14) +
  theme(
    plot.title = element_text(size = 20, face = "bold", hjust = 0.5, color = "darkblue"),
    plot.subtitle = element_text(size = 14, face = "italic", hjust = 0.5, color = "blue"),
    axis.title.x = element_text(size = 14, face = "bold", color = "black"),
    axis.title.y = element_text(size = 14, face = "bold", color = "black"),
    axis.text = element_text(size = 12, color = "black"),
    panel.grid.major = element_line(color = "gray85", size = 0.5),
    panel.grid.minor = element_blank(),
    legend.position = "none",
    plot.background = element_rect(fill = "white", color = "white")
  )

Long-Term Trend in Diffusion Index

• Visual Observations (Second Plot):

  • Overall Trend: There is a slight long-term decline in the custom Diffusion Index (indicated by the red smoothed line), suggesting a gradual weakening in economic activity over the last decade. This could reflect structural challenges in the economy, such as slowing growth in key sectors.

  • COVID-19 Impact: The sharp drop in the index around March 2020 corresponds to the onset of the COVID-19 pandemic. This emphasizes the index’s sensitivity to major economic disruptions.

  • Recovery and Plateau: Post-2020, the index appears to recover but stabilizes at a lower level compared to the pre-pandemic period. This could indicate a slower recovery or a new “normal” for the economy.

• Short-Term Observations:

  • The index shows periodic fluctuations, which could reflect the cyclical nature of the economy. For example, regular peaks and troughs might align with business cycles or seasonal patterns in employment and industrial production.

  • The shaded light blue area (confidence band for the trend line) suggests some variability but also provides a clear central trajectory for the economy’s performance.

# Load CFNAIDIFF
getSymbols("CFNAIDIFF", src = "FRED", return.class = 'xts', from = "2010-01-01")

[1] "CFNAIDIFF"

cfnaidiff_ss <- CFNAIDIFF["2010-05-01/2024-09-01"] |> ts_ts()
min_length <- min(length(ma_index), length(cfnaidiff_ss))
ma_index <- ma_index[1:min_length]
cfnaidiff_ss <- cfnaidiff_ss[1:min_length]
Date <- Date[1:min_length]

# Prepare the comparison data frame
compare_df <- cbind.data.frame(Date, diffusion_index = ma_index, CFNAIDIFF = cfnaidiff_ss)

correlation <- cor(compare_df$diffusion_index, compare_df$CFNAIDIFF, use = "complete.obs")

# Enhanced comparison plot with interactivity
plotly::ggplotly(
  ggplot() +
    geom_line(data = compare_df, aes(x = Date, y = diffusion_index, color = "Diffusion Index"), size = 1.2) +
    geom_line(data = compare_df, aes(x = Date, y = CFNAIDIFF * 100, color = "CFNAIDIFF"), size = 1.2, linetype = "dashed") +
    scale_color_manual(
      values = c("Diffusion Index" = "navy", "CFNAIDIFF" = "red")
    ) +
    labs(
      title = "Comparison of Diffusion Index and CFNAIDIFF",
      x = "Year",
      y = "Index Value",
      color = "Legend"
    ) +
    theme_minimal(base_size = 8) +
    theme(
      plot.title = element_text(size = 10, face = "bold", hjust = 0.5),
      legend.position = "bottom"
    )
)

NA
NA

Key Economic Insights

• Strengths of the Custom Diffusion Index:

  • It effectively captures major economic events, such as the COVID-19 recession.

  • The index tracks the CFNAIDIFF well in many periods, making it a credible proxy for broader economic trends.

• Weaknesses or Limitations:

  • Differences during divergence periods suggest that the inclusion of additional variables could improve alignment with CFNAIDIFF.

  • Sensitivity to short-term data fluctuations may require smoothing or additional preprocessing for clearer interpretation.

Recent Economic Conditions (Post-September 2024)

• If your custom index is trending downward after September 2024:

  • This may indicate a softening in economic activity, potentially due to weaker industrial production, employment, or consumer sentiment.

  • Such trends could reflect challenges like rising interest rates, inflationary pressures, or other macroeconomic headwinds.

• If the index is stable or rising:

  • This suggests resilience in the economy, with positive contributions from the included variables.

Conclusion

The Diffusion Index reflects notable economic fluctuations over time, with recent declines driven by drops in employment, industrial production, and consumer sentiment. In contrast, CFNAIDIFF shows steadier national activity, with less volatility compared to the Diffusion Index. The weak correlation of 0.157 highlights the differences in their focus: the Diffusion Index is more sensitive to localized economic changes, while CFNAIDIFF captures broader national trends. The outlook suggests that the Diffusion Index indicates regional or sectoral weaknesses, whereas CFNAIDIFF points to a stable national economy without widespread contraction.