## Appendix B.

## Source details

**Canonical URL:** [Appendix B.](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022255-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022255-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022255-print-pdf.pdf.json)

---

### Transmission by maturity of the interest rate shock
- Measurement details:
  - Credit card spending: year-on-year percent change of the 90-day moving average of daily credit card spending.
  - Intensive margin: year-on-year percent change of the 90-day moving average of spending per transaction.
  - Extensive margin: year-on-year percent change of the 90-day moving average of the number of transactions.
  - Confidence intervals reported as 90 percent confidence intervals.
- Empirical findings:
  - A one-standard deviation increase in the 3-month yield tends to generate an immediate decline in credit card spending which falls by about 0.8 percent within a month.
  - The negative effect of a 3-month shock persists for about 6 months.
  - Interest rate shocks to 10-year yields are not associated with statistically significant changes in credit card spending after controlling for 2-year yields.
  - Implication: traditional changes in policy rates (operating on short-term rates) affect household spending rapidly; forward guidance and quantitative easing (operating on medium- and long-term rates) involve considerably longer transmission lags.
- Note on volatility comparison (verbatim from source):
  - "the standard deviation of 2-year shocks is larger than the standard deviation of 3-month shocks, 30 and basis points respectively 21. Thus, an increase in the 3-month yield of a same magnitude to an increase in the 2-year yield would generate a peak contraction in consumer spending by about 1.2 percent."

### Asymmetry between contractionary and expansionary shocks
- Model extension:
  - Interact 2-year yield shocks with dummies D+ and D− to separate positive (tightening) and negative (easing) shocks.
- Empirical findings:
  - Positive interest rate shocks trigger a significant decline in credit card spending; the impact is more pronounced when separating signs than in the unconstrained specification.
  - Negative interest rate shocks are unable to stimulate consumption and generate responses that are not statistically significant.
- Robustness and interpretation:
  - Results align with literature suggesting stronger effects from monetary tightening than from easing.
  - Concerns about information effects (monetary announcements revealing outlook) motivate further identification.

### Identification using stock price responses around policy announcements
- Identification strategy:
  - Include interaction terms between interest rate movements around policy announcements and contemporaneous stock price responses (Euro STOXX 50 index).
  - Logic: a hawkish monetary tightening should increase rates and lower stock prices; an information shock that upgrades the outlook would increase both rates and stock prices.
- Empirical findings:
  - In most cases in the sample, interest rates and stock prices moved in opposite directions, indicating prevalence of monetary policy shocks over information shocks.
  - Positive 2-year interest rate shocks coupled with contemporaneous declines in stock prices (signaling an exogenous tightening) trigger a significant contraction in credit card spending.
  - Positive 2-year shocks coupled with increases in stock prices tend to have expansionary effects on spending, though the expansionary effect is shorter lived.
  - Negative interest rate shocks continue to have no significant effects on credit card spending irrespective of the reaction in stock prices.
- Additional evidence:
  - Appendix D shows the coefficient ψ_h_p in equation (3) is positive and statistically significant between 4 and 9 months after the shock.

### Heterogeneous effects across spending categories and user groups
- Empirical approach:
  - Panel version of the regression aggregates credit card spending by groups G_i (spending categories, spending types, age groups, income level) and estimates conditional responses.
- By spending category:
  - Consumer discretionary spending contracts strongly after monetary tightening.
  - Consumer staples spending does not decline; it modestly increases, possibly reflecting substitution from discretionary goods.
- By spending type:
  - No significant differences in transmission between online and in-person purchases.
  - Note: "only about half of the credit card transactions in our sample are differentiated between online and in-person purchases."
- By demographics:
  - No clear differences across age groups.
  - No evidence of differences across gender.
- By income level:
  - Monetary tightening has stronger effects on spending by high-income users; a proposed interpretation is that high-income consumers are less borrowing-constrained and therefore more responsive to intertemporal substitution when interest rates rise.
  - Caution: income inequality in Germany is less pronounced than in other countries and credit card users may have a narrower income distribution than the overall population.
- Data and future improvements:
  - Access to richer income indicators and wider card usage across income segments would help revisit these findings as privacy protocols and data collection expand.

### Conclusions and implications
- Data advantages and main summary findings:
  - Novel daily transaction-level data from more than 1 million credit accounts in Germany allow precise matching of monetary policy shocks and spending responses.
  - Increases in 2-year interest rates triggered by monetary policy announcements have a significant negative impact on credit card spending that materializes with a lag of approximately 6 months.
  - Short-term rate shocks transmit much faster to consumer spending; long-term rate shocks have no significant effects after controlling for 2-year rates.
  - Monetary policy effects are highly asymmetric: tightening reduces spending markedly, while easing fails to stimulate consumption significantly.
  - Contractionary effects are particularly pronounced when rate hikes are accompanied by stock price declines (capturing exogenous tightening); rate hikes accompanied by stock price increases can have short-lived expansionary effects (capturing positive information shocks).
  - Heterogeneity: strong contraction in discretionary spending, modest increase in staples, stronger effects on high-income users, similar effects across online/in-person purchases, gender, and age.
- Research and policy implications:
  - Findings bear on the relative effectiveness of conventional policy (short-term rates) versus unconventional tools (affecting medium/long-term rates).
  - Results raise questions for policymakers about how to support aggregate demand during downturns given the apparent ineffectiveness of monetary easing on consumption.
  - Future research directions include cross-country replication as Fable Data expands card transaction collection, richer user-level demographic/economic data as privacy protocols improve, and applying the empirical approach to other shocks such as fiscal policy.

*Source: Appendix B, wpiea2022255-print-pdf - Appendix B.*

### Appendix B.

### Appendix B.

### Transmission by maturity of the interest rate shock
- Analysis expands the baseline econometric specification to include interest rate shocks on 3-month and 10-year yields to compare effects by maturity.
- Measurement details:
  - Credit card spending: year-on-year percent change of the 90-day moving average of daily credit card spending.
  - Intensive margin: year-on-year percent change of the 90-day moving average of spending per transaction.
  - Extensive margin: year-on-year percent change of the 90-day moving average of the number of transactions.
  - Confidence intervals reported as 90 percent confidence intervals.
- Empirical findings:
  - A one-standard deviation increase in the 3-month yield tends to generate an immediate decline in credit card spending which falls by about 0.8 percent within a month.
  - The negative effect of a 3-month shock persists for about 6 months.
  - Interest rate shocks to 10-year yields are not associated with statistically significant changes in credit card spending after controlling for 2-year yields.
  - Implication: traditional changes in policy rates (operating on short-term rates) affect household spending rapidly; forward guidance and quantitative easing (operating on medium- and long-term rates) involve considerably longer transmission lags.
- Note on volatility comparison (verbatim from source):
  - "the standard deviation of 2-year shocks is larger than the standard deviation of 3-month shocks, 30 and basis points respectively 21. Thus, an increase in the 3-month yield of a same magnitude to an increase in the 2-year yield would generate a peak contraction in consumer spending by about 1.2 percent."

### Asymmetry between contractionary and expansionary shocks
- Model extension: interact 2-year yield shocks with dummies D+ and D− to separate positive (tightening) and negative (easing) shocks (equation (2) in source).
- Empirical findings:
  - Positive interest rate shocks trigger a significant decline in credit card spending; the impact is more pronounced when separating signs than in the unconstrained specification.
  - Negative interest rate shocks are unable to stimulate consumption and generate responses that are not statistically significant.
- Robustness and economic interpretation:
  - Results align with literature suggesting stronger effects from monetary tightening than from easing (references cited in source).
  - Concerns about information effects (monetary announcements revealing outlook) motivate further identification.

### Identification using stock price responses around policy announcements
- Identification strategy:
  - Expand specification to include interaction terms between interest rate movements around policy announcements and contemporaneous stock price responses (Euro STOXX 50 index), as in equation (3).
  - Logic: a hawkish monetary tightening should increase rates and lower stock prices; an information shock that upgrades the outlook would increase both rates and stock prices.
- Empirical findings:
  - In most cases in the sample, interest rates and stock prices moved in opposite directions, indicating prevalence of monetary policy shocks over information shocks.
  - Positive 2-year interest rate shocks coupled with contemporaneous declines in stock prices (signaling an exogenous tightening) trigger a significant contraction in credit card spending.
  - Positive 2-year shocks coupled with increases in stock prices tend to have expansionary effects on spending, though the expansionary effect is shorter lived.
  - Negative interest rate shocks continue to have no significant effects on credit card spending irrespective of the reaction in stock prices.
- Additional evidence:
  - Appendix D shows the coefficient ψ_h_p in equation (3) is positive and statistically significant between 4 and 9 months after the shock (verbatim statement from source).

### Heterogeneous effects across spending categories and user groups
- Empirical approach:
  - Panel version of the regression (equation (4)) aggregates credit card spending by groups G_i (spending categories, spending types, age groups, income level) and estimates conditional responses.
- Key heterogeneous effects:
  - By spending category:
    - Consumer discretionary spending contracts strongly after monetary tightening.
    - Consumer staples spending does not decline; it modestly increases, possibly reflecting substitution from discretionary goods.
  - By spending type:
    - No significant differences in transmission between online and in-person purchases.
    - Note: "only about half of the credit card transactions in our sample are differentiated between online and in-person purchases." (verbatim from source)
  - By demographics:
    - No clear differences across age groups.
    - No evidence of differences across gender.
  - By income level:
    - Monetary tightening has stronger effects on spending by high-income users; a proposed interpretation is that high-income consumers are less borrowing-constrained and therefore more responsive to intertemporal substitution when interest rates rise.
    - Caution: income inequality in Germany is less pronounced than in other countries and credit card users may have a narrower income distribution than the overall population.
- Data and future improvements:
  - Access to richer income indicators and wider card usage across income segments would help revisit these findings as privacy protocols and data collection expand.

### Conclusions and implications
- Data advantages and main summary findings:
  - Novel daily transaction-level data from more than 1 million credit accounts in Germany allow precise matching of monetary policy shocks and spending responses.
  - Increases in 2-year interest rates triggered by monetary policy announcements have a significant negative impact on credit card spending that materializes with a lag of approximately 6 months.
  - Short-term rate shocks transmit much faster to consumer spending; long-term rate shocks have no significant effects after controlling for 2-year rates.
  - Monetary policy effects are highly asymmetric: tightening reduces spending markedly, while easing fails to stimulate consumption significantly.
  - Contractionary effects are particularly pronounced when rate hikes are accompanied by stock price declines (capturing exogenous tightening); rate hikes accompanied by stock price increases can have short-lived expansionary effects (capturing positive information shocks).
  - Heterogeneity: strong contraction in discretionary spending, modest increase in staples, stronger effects on high-income users, similar effects across online/in-person purchases, gender, and age.
- Research and policy implications:
  - Findings bear on the relative effectiveness of conventional policy (short-term rates) versus unconventional tools (affecting medium/long-term rates).
  - Results raise questions for policymakers about how to support aggregate demand during downturns given the apparent ineffectiveness of monetary easing on consumption.
  - Future research directions include cross-country replication as Fable Data expands card transaction collection, richer user-level demographic/economic data as privacy protocols improve, and applying the empirical approach to other shocks such as fiscal policy.

*Source: Appendix B, wpiea2022255-print-pdf - Appendix B.*

### References

### References

### Bibliographic highlights
- Contains citations to empirical and methodological literature on monetary policy, high-frequency identification, big data for economic statistics, consumption dynamics, and transmission mechanisms. Notable entries include:
  - Abraham, Katharine G., Ron S. Jarmin, Brian Moyer, and Matthew D. Shapiro.2022. Big Data for Twenty-First-Century Economic Statistics.
  - Altavilla, Carlo, Luca Brugnolini, Refet S. Gürkaynak, Roberto Motto, and Giuseppe Ragusa.2019. “Measuring euro area monetary policy.” Journal of Monetary Economics, 108: 162–179.
  - Andersen, Asger Lau, Amalie Sofie Jensen, Niels Johannesen, Claus Thustrup Kreiner, Søren Leth-Petersen, and Adam Sheridan.2021. “How do households respond to job loss? Lessons from multiple high-frequency data sets.” CEPR Discussion Paper 16131.
  - Auclert, Adrien.2019. “Monetary policy and the redistribution channel.” American Economic Review, 109(6): 2333–67.
  - Bernanke, Ben S, and Kenneth N Kuttner.2005. “What explains the stock market’s reaction to Federal Reserve policy?” The Journal of finance, 60(3): 1221–1257.
  - Chetty, Raj, John N. Friedman, Nathaniel Hendren, Michael Stepner, , and the Opportunity Insights Team.2020. “The Economic impacts of COVID-19: Evidence from a new public database built using private sector data.” NBER Working Paper 27431.
  - Christiano, Lawrence J, Martin Eichenbaum, and Charles L Evans.1999. “Monetary policy shocks: What have we learned and to what end?” Handbook of macroeconomics, 1: 65–148.
  - Gagnon, Joseph, Matthew Raskin, Julie Remache, and Brian Sack.2011. “The financial market effects of the Federal Reserve’s large-scale asset purchases.” International Journal of Central Banking, 7(1): 3–43.
  - Kaplan, Greg, Benjamin Moll, and Giovanni L. Violante.2018. “Monetary policy according to HANK.” American Economic Review, 108(3): 697–743.
  - Romer, Christina D., and David H. Romer.2000. “Federal Reserve information and the behavior of interest rates.” American Economic Review, 90(3): 429–457.
- The full references list provides sources on identification strategies (high-frequency identification, local projections), monetary policy surprise decomposition, distributional impacts, and the use of administrative and transaction data in consumption analysis.

### Appendix A — Interest rate shocks
- Data and measurement:
  - Shocks to 2-year German bond yields are retrieved from the Euro Area Monetary Policy Event-Study Database, compiled by Altavilla et al. (2019).
- Findings:
  - The distribution of shocks is "fairly symmetrically distributed around zero."
  - In almost 60 percent of the cases, interest rates and stock prices moved in opposite directions, consistent with effects of exogenous changes in the stance of monetary policy.
- Figures:
  - Figure A.1a: Distribution of interest rate shocks (basis points) around monetary policy announcements between 2017 and 2021.
  - Figure A.1b: Scatter plot of changes in the 2-year yield against concomitant changes in the Euro STOXX 50.

### Appendix B — Real vs nominal spending
- Definition:
  - Real spending is computed by deflating the nominal amount by monthly CPI.
  - Credit card spending is computed as the year-on-year percent change of its 90-day moving average.
- Finding:
  - "The impact of monetary policy on credit card spending is largely unchanged no matter whether spending is deflated by CPI or not."
- Estimation details:
  - Regressions include controls for the stage of the pandemic, the stringency of lockdowns, an indicator for income support and debt relief measures, as well as day-of-the-week and day-of-the-year fixed effects.
  - Figures show point estimates with 90 percent confidence intervals.

### Appendix C — Monetary policy shock types and credit card spending
- Methodology:
  - Altavilla et al. (2019) estimate latent factors using yield changes at different maturities during ECB monetary policy announcements and rotate factors to match specific tools.
- Identified shock types:
  - A factor in the press release window interpreted as a shock to the “target” policy rate.
  - Three factors during the press conference window interpreted as “timing” shock, “forward guidance” shock, and “quantitative easing” shocks.
- Findings on transmission timing:
  - “Target” shocks tend to have an immediate impact on credit card spending.
  - Forward guidance and quantitative easing shocks involve considerably longer transmission lags with peak effects on consumption materializing between 9 and 11 months after the shock.
- Figures:
  - Figure C.1 panels (a) to (d) present responses of credit card spending to target, timing, forward guidance, and quantitative easing shocks, derived from a specification that includes all factors.

### Appendix D — Interactions between interest rate and stock price shocks
- Estimation setup:
  - Regression coefficients on interaction terms between interest rate shocks and stock price shocks are reported for equation (3).
- Key findings:
  - Panel D.1a: The coefficient ψ_h_p on the interaction term when the interest rate shock is positive is "positive and statistically significant between 4 and 9 months after the shock," implying that positive interest rate shocks have less contractionary effects on spending—or potentially expansionary effects—if stock prices increase.
  - Panel D.1b: The coefficient ψ_h_p for negative interest rate shocks "is not statistically different from zero," implying stock price movements do not affect the transmission of negative interest rate shocks to household spending.
- Distributional/conditional responses:
  - Figure D.2 reports the estimated impact of credit card spending to an interest rate shock over the distribution of the stock price response, 6 months after the monetary policy event.
  - Over most of the density of stock price changes (between -15 and 15 standard deviations), the estimated impact of an interest rate increase on credit card spending goes from -9.8 to 10.1 percent and is statistically significant.
  - A decline in interest rates does not produce any significant change in spending regardless of the size of the stock price response.
- Notes on estimation:
  - Credit card spending is the year-on-year percent change of its 90-day moving average.
  - Interest rate shocks capture movements of the 2-year German bond during monetary policy announcements.
  - Stock price responses capture movements of the Euro STOXX 50 index during monetary policy announcements.
  - All regressions include controls for the stage of the pandemic, the stringency of lockdowns, an indicator for income support and debt relief measures, as well as day-of-the-week and day-of-the-year fixed effects.
  - Figures display point estimates with 90 percent confidence intervals.

*Monetary Policy and Credit Card Spending Working Paper No. WP/2022/255*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022255-print-pdf.pdf_