## _wp08175

## Source details

**Canonical URL:** [_wp08175](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08175.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08175.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08175.pdf.json)

---

### Introduction
- Germany’s 3 percentage point increase in the value-added tax (VAT) rate in January 2007 was one of the largest such hikes in industrial countries.
- Announcement period lasted about 13 months amid a nascent economic recovery, rapid gains in employment, and rising oil and food prices.
- Study focus:
  - Examine inflationary effects of Germany’s VAT hike, with emphasis on core inflation.
  - Investigate whether anticipatory behavior (“inflation smoothing”) led to price increases prior to the formal implementation.
  - Empirical approach: compare inflation dynamics of CPI/HICP items liable to the VAT hike (VAT items) with non-VAT items before and after implementation, controlling for demand shifts, time-specific and seasonal effects, administrative/policy one-offs.

### Key findings
- Pass-through and timing
  - Pass-through of the higher VAT rate to consumers was rather limited overall.
  - Price increases were phased in during 2006 in anticipation of the VAT hike (inflation smoothing), explaining why core inflation after January 2007 did not rise as much as anticipated.
  - Core inflation rose steadily in 2006 and jumped in January 2007; a further rise in April 2007 was due to administrative increases in education costs.
  - After accounting for the education effect, core inflation in the second half of 2007 declined significantly, indicating a base effect related to price increases in the second half of 2006.
- Heterogeneity
  - Producers of items with captive demand (e.g., durable goods) were able to take advantage of consumption shifting and raise prices more.
  - VAT items experienced faster inflation growth than the euro area from 2005 until end-2007, especially in 2006.
- Demand-side evidence
  - Consumption increased 2.2 percent in the last quarter of 2006 over the same period of the previous year; increase strongest in December and more pronounced for cars and other durables.
  - Consumption contracted by 0.5 percent in the first quarter of 2007 (both figures in annualized terms).

### Stylized facts and numerical magnitudes
- VAT coverage and mechanical impact
  - VAT hike affected all items liable to the standard rate of 16 percent (items under the reduced rate such as basic food or books and entertainment were not affected).
  - At a full and immediate pass-through the VAT rate increase would have implied a 2.6 percentage point jump in prices of the affected items.
  - With three-fifths of the items in the CPI basket affected (the VAT items), the full impact on the CPI would have been 1.4 percentage points (Bundesbank, 2006).
- Price-setting frequency around the event
  - Rise in frequency of price increases in goods associated with the VAT hike: about eight percent in the second half of 2006, and ten percent in January 2007, compared to below five percent for the period 2003-2005.
  - Increase of seven percent in the frequency of price decreases was observed in January 2007.

### Theoretical motivations for inflation smoothing
- Sticky information / staggered price setting
  - Firms update information and re-optimize prices infrequently because gathering information and altering prices is costly.
  - Announcement of a VAT hike can lead firms that can adjust prices before implementation to raise prices toward the new target, producing an inflation smoothing effect.
  - If firms can change prices frequently, postponement until implementation may be optimal—so both directions are theoretically possible.
- Imperfect competition and consumption shifting
  - Limited competition and downward-sloping demand imply announcement-induced intertemporal consumption shifting brings purchases forward, raising demand during the announcement phase.
  - Firms can extract higher rents via higher prices; effects larger for goods with larger demand shifts (e.g., durables) and in less competitive markets.

### Empirical strategy, data, and controls
- Data
  - Monthly panel of annual inflation rates for two-digit items of the harmonized consumer price index (HICP) in Germany covering 2005m1 to 2008m1.
  - Two-digit level chosen to allow differentiation while remaining aggregated enough to estimate effects on aggregate core inflation.
  - Final sample: 45 items with a weight of 80 percent of the overall HICP (initial universe 53 items; energy items 19 and 32, seasonal goods and tobacco dropped).
- Empirical specification and identification
  - Model annual inflation rate for commodity i at time t as function of lagged inflation, item-specific time-varying factors, individual and time fixed effects, and group-specific effects to capture VAT announcement/implementation.
  - VAT-specific variables:
    - VAT 06 trend: one for VAT items in January 2006, then increases monthly by one until December 2006.
    - VAT 07 implementation: one for VAT items in January 2007 only.
    - VAT 07 trend: one for VAT items in January 2007, then increases monthly by one until December 2007.
  - Controls: euro area item-level inflation (euro area excluding Germany synthetic aggregates), linear time trend, monthly dummies, administrative price change dummies (education April 2007, medical services January 2006), durable goods dummy for 2006, price-setting power measure based on 2001–2002 price movements.

### Main empirical results (baseline estimation and robustness)
- Baseline estimation setup
  - Dependent variable: HICP-weighted inflation rate of 45 two-digit items; sample January 2005 - January 2008.
  - Estimation methods: FE (fixed effects) and GMM (dynamic panel).
  - Observations: 1665; Number of groups: 45; R-squared: 0.68 (FE).
- Key coefficient estimates (as reported across specifications)
  - Lagged HICP-weighted inflation: 0.76, 0.77 (z-statistics (17.70)** and (17.80)**).
  - VAT 06 trend: 0.001 (z-statistics reported as (2.92)**, (2.92)**, (2.09)*, (2.07)* across specs).
  - VAT 07 implementation: 0.020, 0.020, 0.010, 0.003 (z-statistics (6.73)**, (5.28)**, (1.8), (0.80)).
  - VAT 07 trend: -0.0002 (z = (-0.36)) and 0.0004 (z = (0.7)) in reported cells.
  - VAT Dummy (time-invariant): 0.020, 0.020 (z = (4.93)**, (4.91)** in GMM columns).
- Inference on VAT effects
  - VAT 06 trend is positive and significant throughout specifications, interpreted as evidence for anticipatory inflation smoothing.
  - VAT 07 implementation shows a strong and positive immediate impact in January 2007.
  - No empirical support for a delayed pass-through throughout 2007 (VAT 07 trend not significant).
  - GMM dynamic panel shows high persistence in inflation but main VAT results unchanged: VAT 06 trend remains significant; no delayed price increases post-implementation.

### Exploring alternative explanations and heterogeneity
- Tests for alternative explanatory factors (Table 2)
  - World Cup 06 dummy for May and June 2006: no measurable effect in column (1).
  - World Cup-related restaurants and hotels dummy: picks up significant price increases during the 2006 World Cup but does not account for VAT 06 trend effect.
- Support for demand-shift explanation
  - Durable goods dummy in 2006: positive and statistically significant (specification (3)), indicating price increases concentrated among items with larger demand increases, especially at year-end.
  - Supplier price-setting power: inclusion of a 2002-based measure shows significance in some specifications (full numeric estimates reported in Table 2 of the source).

### Decomposition, pass-through, and aggregate implications
- Model decomposition (based on specification (5) of Table 2)
  - Table I-3. Decomposition of Core Inflation as Predicted by the Model (annual average differences):
    - Actual increase: 2006 = 0.35 ; 2007 = 1.00
    - Total model: 2006 = 0.41 ; 2007 = 0.73
    - VAT total: 2006 = 0.36 ; 2007 = 0.40
      - Durables: 2006 = 0.03 ; 2007 = ...
      - Price setting power: 2006 = 0.03 ; 2007 = ...
      - General increase: 2006 = 0.30 ; 2007 = ...
    - Non–VAT: 2006 = 0.05 ; 2007 = 0.33
    - Residual: 2006 = -0.06 ; 2007 = 0.27
    - Memorandum item: Inflation pass-through: 2006 = 0.24 ; 2007 = 0.49
- Interpretation and magnitudes
  - Model predicts an increase in core inflation due to VAT items of a total of 0.76 percentage point in 2006 and 2007, with 0.36 percentage points in 2006 and 0.40 percentage point in 2007 concentrated upfront at implementation (January 2007).
  - In 2006 core inflation is estimated to have increased by 0.41 percentage point, and by 0.73 percentage points in 2007, totaling 1.15 percentage points; two thirds of the predicted increase was VAT related.
  - Among VAT items, durable goods and items in markets with imperfect competition accounted for 0.06 percentage point of the increase in core inflation in 2006 — one sixth of the estimated VAT-driven increase (0.36 percentage point) in 2006.
  - Estimated cumulative pass-through rate of the VAT hike to core inflation: 73 percent.
    - Timing breakdown: average pass-through 24 percent in 2006 and 49 percent in 2007 (primarily January).

### Methodological notes and calculation procedure
- Sample construction
  - Initial universe: all 53 commodity items with respective HICP-weights.
  - Core inflation sample excludes energy-related items 19 and 32, seasonal goods (fruit, vegetable, and package holidays), and tobacco; leaving 45 items with a total HICP weight of 80 percent.
  - Core index excludes items 1, 7, 8, 13, 19, 30, 32, and 44.
- Procedure to compute model-predicted effects on core inflation
  - a. Multiply actual value for each variable in each month by its estimated coefficient from Table 2, specification (5), and by the number of items in the group (VAT or non-VAT).
  - b. Sum components to obtain predicted inflation rates for each month (annual rates).
  - c. 2006 overall effect: difference between the predicted inflation rate between December and January 2006.
  - d. 2007 implementation effect: difference between the average predicted core inflation rates of the first three months of 2007 and the last three months of 2006.
  - e. Contributions of VAT items obtained by repeating steps c and d for VAT items only.

### Conclusion and policy implications
- Empirical conclusion
  - Evidence of an acceleration of inflation in all VAT items in 2006, a jump in January 2007, but no delayed price increases throughout 2007.
  - Inflation acceleration was stronger among durables and items in less competitive markets, consistent with rent extraction motives.
  - Joint significance test of VAT 06 trend, durable goods, and price setting power: χ2 = 8.5 with p value = 3 percent, rejecting the null of no effect.
- Policy-relevant implications
  - The inflationary profile of a large tax hike is affected by the length of the announcement period: advance price adjustments smooth the inflation profile and can avoid large spikes that risk triggering second-round effects.
  - Incentives for inflation smoothing appear linked to intertemporal consumption shifting; items experiencing larger demand increases are more affected.
  - Understanding these channels is important given several countries’ intentions to increase indirect taxation.

*Source: _wp08175 (excerpts and appendices provided).*

### References..............................................................................................................

### _wp08175 - References

### Introduction
- Germany’s 3 percentage point increase in the value-added tax (VAT) rate in January 2007 was one of the largest such hikes in industrial countries.
- The announcement period lasted about 13 months and occurred amid a nascent economic recovery, rapid gains in employment, and rising oil and food prices.
- Despite concerns, the actual inflation increase was much lower than feared.
- Focus of the study:
  - Examine inflationary effects of Germany’s VAT hike, with emphasis on core inflation.
  - Investigate whether anticipatory behavior (“inflation smoothing”) led to price increases prior to the formal implementation.
  - Empirical approach: compare inflation dynamics of CPI items liable to the VAT hike (VAT items) with non-VAT items before and after implementation, controlling for demand shifts, time-specific and seasonal effects, administrative/policy one-offs.

### Key findings (as presented)
- Pass-through of the higher VAT rate to consumers was rather limited overall.
- Producers of items with captive demand (e.g., durable goods) were able to take advantage of consumption shifting and raise prices more.
- Price increases were phased in during 2006 in anticipation of the VAT hike (inflation smoothing), which explains why core inflation after January 2007 did not rise as much as anticipated.
- Core inflation:
  - Rose steadily in 2006 and jumped in January 2007.
  - Further rise in April 2007 due to administrative increases in education costs.
  - After accounting for the education effect, core inflation in the second half of 2007 declined significantly, indicating a base effect related to price increases in the second half of 2006.
- Cross-country comparison:
  - German prices in 2006 rose more rapidly than prices of a comparable basket in the euro area, especially for VAT items.
- Demand-side evidence:
  - Consumption increased 2.2 percent in the last quarter of 2006 over the same period of the previous year; increase strongest in December and more pronounced for cars and other durables.
  - Consumption contracted by 0.5 percent in the first quarter of 2007 (both figures in annualized terms).

### Stylized facts and numerical magnitudes
- VAT specifics:
  - The VAT hike affected all items liable to the standard rate of 16 percent (items under the reduced rate such as basic food or books and entertainment were not affected).
  - At a full and immediate pass-through the VAT rate increase would have implied a 2.6 percentage point jump in prices of the affected items.
  - With three-fifths of the items in the CPI basket affected (the VAT items), the full impact on the CPI would have been 1.4 percentage points (Bundesbank, 2006).
- Price-setting frequency changes observed around the VAT event:
  - Rise in frequency of price increases in goods associated with the VAT hike: about eight percent in the second half of 2006, and ten percent in January 2007, compared to below five percent for the period 2003-2005.
  - Increase of seven percent in the frequency of price decreases was observed in January 2007.

### Rationale for inflation smoothing (theoretical motivations)
- Sticky information / staggered price setting:
  - Firms update information and re-optimize prices infrequently because gathering information and altering prices is costly (Mankiw and Reis, 2002).
  - Announcement of a VAT hike can lead firms with the opportunity to adjust prices before implementation to raise prices toward the new target, producing an inflation smoothing effect.
  - If firms can change prices frequently, however, postponement until implementation may be optimal—so both directions are theoretically possible.
- Imperfect competition and consumption shifting:
  - If competition is limited and firms face downward-sloping demand curves, announcement-induced intertemporal consumption shifting (bringing purchases forward) raises demand during the announcement phase.
  - Firms can extract higher rents via higher prices; effects larger for goods with larger demand shifts (e.g., durables) and in less competitive markets.
  - Illustrated by monopolistic pricing: announcement shifts demand out, raising the equilibrium price during the announcement period (prices p_i** and p_j** in the text).

### Empirical analysis: data and strategy
- Data:
  - Monthly panel of annual inflation rates for two-digit items of the harmonized consumer price index (HICP) in Germany covering the period 2005m1 to 2008m1.
  - Two-digit level chosen to allow differentiation across items while remaining aggregated enough to estimate effects on aggregate core inflation.
- Empirical specification:
  - Model annual inflation rate for commodity i at time t (p_t,i) as a function of lagged inflation (p_{t-1,i}), item-specific time-varying factors (X_it), individual and time fixed effects (Y_i, Z_t), and group-specific effect G_{Σj,} to capture tax policy announcement effects on VAT-liable items (Σ_j).
  - Estimations include static and dynamic panel specifications.
- Identification logic:
  - Systematic inflationary discrepancies between VAT and non-VAT items prior to 2007 that are not accounted for by controls are attributed to the announcement of the VAT hike.
  - Differences at implementation and post-implementation provide information about pass-through and lagged price increases.

### Empirical hypotheses tested
- Whether the VAT announcement caused earlier-than-expected price increases (announcement effect / inflation smoothing).
- Whether differentiated price effects depend on:
  - Degree of competition (imperfect competition amplifies announcement-period price increases).
  - Size of demand shocks (larger for durables and less frequently purchased items).
  - Macroeconomic context (whether output gap closing and stronger domestic demand in 2006 explain price increases instead of the VAT announcement).

*Source: _wp08175 - References (excerpts from the provided PDF content)._

### appendix lists all 53 commodity items with their respective HICP weights and whether they

### _wp08175 - appendix lists all 53 commodity items with their respective HICP weights and whether they 

### Appendix scope and sample
- Appendix lists all 53 commodity items with their respective HICP weights and whether they are considered durable and affected by the VAT hike. The values for the latter measure vary between 0 and 1 depending on the share of goods within a specific commodity item which are affected by the VAT hike.
- Two energy related items (19 and 32) were dropped from the analysis.
- Seasonal goods (fruit, vegetable, package holidays), and tobacco were dropped because of high volatility.
- Final sample: 45 items with a weight of 80 percent of the overall HICP.
- Sample period for empirical estimation: January 2005 - January 2008.

### Empirical objective and aggregation approach
- Objective: Trace effects of the VAT hike on aggregate core inflation by estimating different inflation trends for VAT items and non-VAT items.
- Aggregation approach: Estimate HICP-weighted inflation rates by weighing each individual item inflation by its weight in the HICP basket to obtain an average VAT-group effect that is not driven by small-weight items.
- Dependent variable definition and rationale discussed in the appendix.

### Control variables and identification
- Euro area wide inflation effects:
  - Item-by-item inflation rates computed for the euro area excluding Germany by aggregating country level 2-digit HICP data to a synthetic aggregate index.
  - Empirical specification includes euro area specific inflation variables for each HICP item separately.
- Time trend:
  - Linear trend included; other specifications (trend squared) explored but not statistically significant.
- Monthly time dummies and idiosyncratic shocks:
  - Dummy variables for each month introduced.
  - Control variables included for administrative price changes for education services in April of 2007 and medical services in January 2006.
- VAT-specific variables:
  - VAT 06 trend: tests announcement/anticipation effect in 2006 (takes value of one for VAT items in January 2006, then increases monthly by one until December 2006).
  - VAT 07 implementation: takes value of one for VAT items in January 2007 only.
  - VAT 07 trend: takes value of one for VAT items in January 2007, then increases monthly by one until December 2007.
  - Fixed effect and GMM models test for time-invariant differences in inflation rates.
- Durable goods:
  - Dummy for durable goods in 2006 to capture demand shifting during the run-up period.
- Price-setting power:
  - Measured using rate of inflation rate increase between December 2001 and January 2002 for price movements above the 75th percentile of price increases (exploiting euro currency introduction price adjustments).

### Baseline estimation results (Table 1 summary)
- Dependent variable: HICP-weighted inflation rate of 45 two-digit items of the index; sample January 2005 - January 2008.
- Estimation methods: FE (fixed effects) and GMM (generalized method of moments).
- Key coefficient estimates and statistics (as reported across specifications (1)-(4)):
  - Constant: 0.001, 0.001, -0.010, -0.010.
  - Lagged HICP-weighted inflation: 0.76, 0.77 (z-statistics shown as (17.70)** and (17.80)**).
  - Trend: 0.001, 0.001, -0.00003, -0.0001 (trend significant in FE, not in GMM).
  - VAT 06 trend: 0.001 (z-statistics reported as (2.92)**, (2.92)**, (2.09)*, (2.07)* across specs).
  - VAT 07 implementation: 0.020, 0.020, 0.010, 0.003 (z-statistics (6.73)**, (5.28)**, (1.8), (0.80)).
  - VAT 07 trend: -0.0002 (z = (-0.36)) and 0.0004 (z = (0.7)) in reported cells.
  - VAT Dummy (time-invariant): 0.020, 0.020 (z = (4.93)**, (4.91)** in reported GMM columns).
- Model diagnostics and sample information:
  - Monthly dummies: Y in all specifications.
  - Price shocks (education fees April 2007 and medical services January 2006): included in all specifications.
  - Test for no residual autocorrelation: Rejected in FE specifications, Not Rejected in GMM specifications.
  - R-squared: 0.68.
  - Observations: 1665.
  - Number of groups: 45.
  - Absolute value of z statistics in parentheses. * significant at 5 percent; ** significant at 1 percent.

### Main findings on VAT-related inflation effects
- VAT announcement effect (VAT 06 trend):
  - Positive and significant throughout specifications, implying inflation rates increased more strongly among VAT items during 2006 after controlling for item-specific, time-specific, and euro area wide effects.
  - Interpreted as evidence for an anticipatory inflation smoothing effect.
- VAT implementation effect (VAT 07 implementation):
  - Strong and positive effect measuring the immediate impact of the VAT hike upon implementation.
- Post-implementation effect (VAT 07 trend):
  - Model rejects hypothesis of a delayed pass-through; no support for a post-implementation inflation trend in 2007.
- Persistence and robustness:
  - Dynamic panel (GMM) estimates show inflation is highly persistent, but main VAT results remain unchanged: VAT 06 trend remains significant and no evidence of delayed price increases after VAT implementation.
- Aggregate interpretation:
  - Evidence that inflation was low in Germany but VAT items experienced faster inflation growth than the Euro area from 2005 until end-2007, driven by the VAT announcement and implementation effects (termed “inflation smoothing”).

### Exploring the inflation smoothing hypothesis (Table 2 summary)
- Tests for alternative explanations for 2006 VAT-item inflation acceleration:
  - World Cup 06 dummy (May and June 2006): column (1) shows no measurable effect.
  - World Cup-related restaurants and hotels dummy: column (2) picks up significant price increases during the 2006 World Cup, but does not explain away the positive VAT 06 trend effect.
- Tests supporting demand-shift explanation:
  - Durable goods dummy in 2006 (specification (3)) shows a positive and statistically significant effect, supporting that 2006 price increases were concentrated among items with larger increases in demand; further tests show price increases concentrated at year-end (not shown).
  - Supplier price-setting power:
    - Separate tests include the 2002-based measure of price-setting power; column (3) of Table 2 reports inclusion of this measure (full numeric estimates for this variable are presented in Table 2 but are not reproduced in the provided text).

*Source: _wp08175 - appendix lists all 53 commodity items with their respective HICP weights and whether they*

### introduction.

### introduction.

### Key empirical findings
- Evidence of an acceleration of the inflation rate in all VAT items in 2006, of a jump increase in January 2007, but no empirical support of delayed price increases throughout 2007.
- Among VAT items, inflation accelerated faster among durables and items in less competitive markets, indicating that rent extraction may have been a motive.
- Dynamic panel estimation: only the VAT trend variable is statistically significant; high standard errors for durability and price setting power indicate multicollinearity. A joint significance test of VAT 06 trend, durable goods, and price setting power produced χ2 = 8.5 with p value = 3 percent, rejecting the null of no effect.

### Determinants of German Core Inflation (summary of Table 2 results)
- Dependent variable: HICP-weighted inflation rate of 45 two-digit items; sample January 2005 - January 2008.
- Specification notes:
  - Estimation: FE (fixed effects) for columns (1)-(5); GMM for column (6).
  - Monthly dummies: Y in all specifications.
  - Price shocks control: Y in all specifications.
  - R-squared: 0.68 in FE specifications.
  - Observations: 1665; Number of groups: 45.
- Selected coefficient estimates and significance (absolute value of z in parentheses; * significant at 5 percent; ** significant at 1 percent):
  - Trend: 0.001 (2.50)** / (2.50)* / (3.21)** / (3.19)** / (2.50)* across specifications.
  - VAT 06 trend: 0.001 (2.92)** ; 0.001 (2.87)** ; 0.001 (2.14)* ; 0.0005 (1.79)* in successive specs.
  - Durable goods: 0.010 (2.34)* ; 0.010 (2.19)* ; 0.002 (1.43).
  - Price setting power: 0.020 (3.05)** ; 0.020 (2.88)** ; 0.003 (1.09).
  - VAT 07 implementation: 0.020 (6.73)** ; 0.020 (6.73)** ; 0.020 (6.41)** ; 0.020 (6.28)** ; 0.020 (6.76)** ; 0.005 (1.80)*.
  - ΔPt-1: 0.76 (17.4)** in the specification where reported.
- Test for no residual autocorrelation: Rej. in FE specs; Not Rej. in GMM.

### Aggregate inflation effects (model decomposition)
- Based on coefficients of specification (5) in Table 2 (best fit specification).
- Table I-3. Decomposition of Core Inflation as Predicted by the Model (annual average differences):
  - Actual increase: 2006 = 0.35 ; 2007 = 1.00
  - Total model: 2006 = 0.41 ; 2007 = 0.73
  - VAT total: 2006 = 0.36 ; 2007 = 0.40
    - Durables: 2006 = 0.03 ; 2007 = ...
    - Price setting power: 2006 = 0.03 ; 2007 = ...
    - General increase: 2006 = 0.30 ; 2007 = ...
  - Non–VAT: 2006 = 0.05 ; 2007 = 0.33
  - Residual: 2006 = -0.06 ; 2007 = 0.27
  - Memorandum item: Inflation pass-through: 2006 = 0.24 ; 2007 = 0.49
- Model implications:
  - The model accounts for a significant share of the actual aggregate increase in core inflation in 2006 and 2007.
  - The model predicts an increase in core inflation due to VAT items of a total of 0.76 percentage point in 2006 and 2007, with 0.36 percentage points in 2006 and 0.40 percentage point in 2007 concentrated upfront at the time of implementation (January 2007).
  - In 2006 core inflation is estimated to have increased by 0.41 percentage point, and by 0.73 percentage points in 2007, for a total of 1.15 percentage points. Two thirds of the predicted increase in core inflation was VAT related.
  - Among VAT items, durable goods and items in markets with imperfect competition accounted for 0.06 percentage point of the increase in core inflation in 2006 — one sixth of the estimated VAT-driven increase (0.36 percentage point) in 2006.
- Estimated cumulative pass-through rate of the VAT hike to core inflation: 73 percent.
  - Broken down by timing: average pass-through 24 percent in 2006 and 49 percent in 2007 (primarily January).

### Data and sample construction
- Initial universe: All 53 commodity items with respective HICP-weights.
- Core inflation sample excludes energy-related items 19 and 32, seasonal goods (fruit, vegetable, and package holidays), and tobacco; leaving a total of 45 items with a total HICP weight of 80 percent.
- The core index excludes items 1, 7, 8, 13, 19, 30, 32, and 44.
- Table A1 provides HICP commodity items, weights, and classifications for durables, VAT exposure, and price setting power; total weight sums to 1000.0 in the provided list.

### Methodology (summary)
- Estimated inflation in VAT items relative to non-VAT items using monthly HICP at the two-digit level.
- Regression framework (described formula): the annual inflation rate for any commodity item is modeled with controls for:
  - α2: general price increases due to common shocks or common cyclical conditions (e.g., euro area).
  - α3: effect of the VAT, elaborated for timing.
  - Additional control variables αj for commodity-specific issues (durability, competitive market structure).
- Models used: random-effect models (stated to fit the data better than fixed-effects in some contexts) and fixed-effects specifications reported in Table 2; dynamic panel (GMM) also used.
- Interpretation of coefficients: coefficients attached to VAT-group dummies measure the weighted average effect of included VAT items on core inflation; to obtain an impact on aggregate HICP core inflation the coefficients need to be multiplied by the number of VAT items.

### Procedure for calculating model-predicted effects on core inflation
- Steps used to compute values in Table I-3:
  - a. Take the actual value for each variable in each month and multiply it by its estimated coefficient from Table 2, specification (5), and by the number of items in the group (VAT or non-VAT).
  - b. Sum these components to obtain predicted inflation rates for each month (annual rates).
  - c. To calculate the 2006 overall effect, take the difference between the predicted inflation rate between December and January 2006.
  - d. To calculate the 2007 implementation effect, take the difference between the average predicted core inflation rates of the first three months of 2007 and the last three months of 2006.
  - e. To calculate contributions of VAT items to the 2006 overall effect or to the 2007 implementation effect, follow steps c and d but only for VAT items in core inflation.
- Notes:
  - Multiplying the group average effect by the total number of items yields the aggregate inflation effect of VAT items as a group.
  - The passthrough computation takes account of the fact that the sample excludes many non-VAT items.
  - If one takes the differences between the January 2007 and December 2006 rates the results are similar.

### Conclusion and policy implications
- An increase in inflation in VAT items contributed 0.36 percentage point to core inflation in anticipation (2006), and a further 0.73 percentage point increase upon implementation in 2007.
- As consumer demand increased in 2006, producers raised prices, more so in durable goods; the January 2007 increase was more muted (0.4).
- Cumulative VAT pass-through to core inflation: 73 percent.
- Policy-relevant implications:
  - The inflationary profile of a large tax hike is likely affected by the length of the announcement period: advance price adjustments smooth the inflation profile and can avoid large spikes that risk triggering second-round effects.
  - Incentives for inflation smoothing appear linked to intertemporal consumption shifting; items experiencing larger demand increases are more affected.
  - Understanding these channels is important given several countries’ intentions to increase indirect taxation.

*Source: _wp08175 - introduction.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08175.pdf_