## _wp16166

## Source details

**Canonical URL:** [_wp16166](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16166.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16166.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16166.pdf.json)

---

### I. Introduction — framing the questions
- Headline CPI averaged just 4% between 2000 and 2006, then "averaged more than 9% between 2006 and 2013."
- Headline CPI peaked at 12.1% year-on-year growth in November 2013 and "collapsed to 4.3% in December 2014" — a fall of nearly 800 bps in 13 months — before accelerating to 5% in October 2015.
- Key empirical questions:
  - What was responsible for the sharp decline in inflation over the last two years?
  - Is the decline durable or transitory?
- Candidate contributing factors listed: collapse in global oil and commodity prices; sharp fall in global food prices; a new monetary policy regime aimed at anchoring inflation expectations; a new government addressing food supply bottlenecks; continued restraint on agricultural support prices.
- Method: simple econometric model (augmented Phillips Curve) to quantify contributions of different factors to disinflation.

### Major findings (overview)
- Inflation evolution driven by:
  - business cycle (output gaps),
  - adaptive and forward-looking expectations,
  - institutional mechanisms (MSP determination and backward-looking wage indexation) that amplify persistence,
  - global factors (oil, food prices, exchange rates).
- Quantitative decompositions for disinflation between fiscal years 2013/14 and 2014/15:
  - Close to half of the disinflation attributed to moderation in historical dynamics of inflation (softening of backward-looking expectations and institutional processes).
  - About 20% explained by a sharp decline in the growth of the discretionary component of MSPs (after controlling for wages, input costs, global food prices).
  - About a third attributed to forward-looking expectations (proxying the new monetary policy regime and perceived permanence of global commodity collapses).
- Additional insight: once lagged inflation is included, wages lose explanatory power for current inflation — wages behave more as an "outcome" than a "driver" consistent with backward-looking indexation.

### II. Context: inflation persistence and breadth (2006–14)
- CPI-Industrial Workers (CPI-IW) averaged 9.1% year-on-year between January 2006 and December 2014 (96 months).
- Persistence:
  - CPI > 9% for 51% of those 96 months.
  - CPI > 8% for 63% of those 96 months.
- Breadth:
  - Non-food, non-fuel (core) inflation averaged 7.8% during this period — "just 130 bps below headline" — indicating broad-based inflation.

### III. Followed by the sharp disinflation (aggregate and sectoral facts)
- New all-India CPI:
  - averaged 10.1% in 2012/13 and 9.8% in 2013/14.
  - average inflation fell to 6% in 2014/15 — 400 bps lower than the previous two years.
  - in 2015/16 (April–October) averaged 4.6% — another 140 bps lower.
- Momentum (annualized quarterly growth, seasonally adjusted) fell from 12.9% in last quarter of 2013 to 2.9% q/q, saar in August–October 2015.
- Distribution of disinflation between 2013/14 and 2014/15:
  - 62% attributable to disinflation in food.
  - 38% attributable to non-food (22% core, 16% fuel).
- Food disinflation concentration:
  - Cereals and vegetables (accounting for < 35% of CPI food basket) are responsible for 92% of the food disinflation.
- Definitions:
  - Fuel includes “transport and communication” (large fuel component).
  - Core excludes food, fuel, transport and communication.

### IV. Model, data and identification
- Model: single-equation augmented Phillips Curve; complementary general VAR in Appendix.
- Data/specification:
  - All variables entered as year-on-year growth rates.
  - Quarterly data from 2000Q2 to 2015Q1.
  - Baseline uses 8 lags of CPI inflation and HP-filtered potential output (HP filter with λ = 1600).
  - Output gap, global crude prices, exchange rate lagged by one quarter.
  - Domestic variables: changes in MSP (lagged), rural wages (lagged), dummy for below-normal monsoon (1 when south-west monsoon rainfall < 85%).
  - New Regime Dummy = 1 for 2014Q1–2015Q1 to proxy forward-looking expectations after introduction of flexible inflation targeting by the RBI.
- Endogeneity/addressing simultaneity:
  - MSPs and wages lagged one quarter to mitigate reverse causality.
  - Inclusion of eight lags of domestic headline inflation and global food inflation to address omitted-variable bias related to annual MSP/wage-setting.

### V. Single-equation empirical findings (selected estimates and diagnostics)
- Baseline regression:
  - Observations 60
  - R-squared 0.96
  - D-W-statistic 1.96
  - B-G serial correlation test (LM test) 0.16
  - AIC 2.50
  - SBC 3.33
- Key coefficients (all lags and variable labels as reported):
  - Output gap - HP filter - 1Q lag: 0.52** (one percentage point increase in output gap raises headline inflation by 52 bps; significant at 5%).
  - CPI inflation lags (1Q to 8Q): 0.41**, 0.23, 0.18, -0.30**, 0.47**, 0.01, 0.01, -0.21** (indicating persistence).
  - Dummy for new monetary policy regime#: -1.43** (reduces headline CPI by 143 bps on average; significant at 5%).
  - Annual growth in MSP - 1Q lag: 0.06^ (one percentage point increase in MSP growth associated with 6 bps increase in headline CPI; weakly significant at 15%).
  - Annual growth in wages - 1Q lag: -0.04 (statistically and economically insignificant once lags of inflation included).
  - Dummy for below normal monsoon: 0.67* (increases headline inflation by 67 bps; significant at 10%).
  - Annual growth in $/Re. exchange rate - 1Q lag: -0.08** (1 percentage point appreciation of rupee reduces inflation by 8 bps; significant at 5%).
  - Annual growth in world crude price - 1Q lag: 0.01^ (1 percentage point increase in crude raises CPI by 1 bps; significant at 15%).
  - Annual growth in world food price index - various lags: 0.02, -0.03^, 0.01, 0.00, 0.01, -0.02, 0.04*, 0.00 (some lags significant, e.g., 7Q lag 0.04*).

### VI. Inflation, wages and MSPs — institutional dynamics (Box 1 & Box 3 synthesis)
- Simultaneity and institutional context:
  - Rural wages, MSPs and inflation interact; rural wages often indexed ex post (e.g., MGNREGA wage indexation to previous year’s CPI-RL).
- Wages:
  - Backward-looking due to formal indexation; empirical result: wages significant when MSPs included but cease to be significant once multiple lags of inflation are added.
  - Eight lags of inflation explain about 80% of the variation in wage growth (Figure 5 result referenced).
  - Conclusion: wages operate more as an "outcome" of past inflation than an independent driver.
- MSPs:
  - MSP inflation serves as a floor for key food groups and retains statistical significance even after controlling for lags of inflation and global food prices.
  - Regression of MSP growth on eight lags of CPI inflation and eight lags of world food prices explains 68% of MSP variation — implying part of MSP growth is driven by lagged inflation and world food prices but a discretionary component remains.

### VII. Decomposition of disinflation (2013/14 → 2014/15) — baseline and alternative periods
- Baseline decomposition (Table 3a):
  - Decline in model predicted inflation: 3.3 percentage points (100%).
  - Lagged Inflation: 1.546 percentage points (46%).
  - MSP: 0.721 percentage points (21%).
  - New regime: 1.133 percentage points (33%).
  - Global crude: 0.12 percentage points.
  - Exchange rate: 0.411 percentage points.
  - Output gap: -0.3 percentage points (-8%) [i.e., output gap exerted upward pressure].
  - Global food: -0.1 percentage points (-4%).
- Alternative period (first three quarters of 2015 vs same period 2014):
  - Oil dropped by 66% → oil’s contribution rose to 11% in that period.
  - Global factors (food, exchange rate, oil) accounted for 42% in that alternative period vs 10% in baseline.
  - Even then, lagged inflation accounted for 52% and with MSPs made up 60% together.
- Tabulated key summary claims (from Box 3 conclusions):
  - 20% of disinflation explained by sharp decline in discretionary component of MSPs.
  - 45% attributed to moderation in historical dynamics of inflation (backward-looking/adaptive expectations and institutional persistence).
  - Almost 35% attributed to forward-looking expectations (new monetary policy regime and more benign future commodity expectations).
  - Role of global factors (crude and exchange rates) < 15% in baseline period; larger in alternative horizons.
  - Output gaps closed and were putting upward pressure on inflation (business cycle did not drive recent disinflation).

### VIII. Robustness checks and VAR corroboration
- Robustness:
  - Alternative wage treatment (first difference), alternative monsoon definitions, alternative output gap measures, agriculture production growth used — key variables remain robust.
  - Coefficient on MSP ranges from 0.05 to 0.12 across specifications; a one percentage point MSP increase associated with 5–12 bps increase in headline CPI.
  - Lagged dynamics of inflation remain largest contributor across specifications (37–52% range).
  - New regime dummy accounts for 14–44% of decline across specifications.
  - Using Christiano-Fitzgerald filter for output gap increases output gap effect to 60 bps without affecting other results.
- VAR (lags = 8, Choleski ordering: global factors → output gap → wages → MSP → inflation):
  - Impulse response: 1 s.d. shock to MSP growth raises inflation by 0.3 percentage points; effect not persistent (indistinguishable from zero by second quarter).
  - Wages: smaller, non-persistent effects relative to MSP.
  - Variance decomposition: 50% of inflation variability explained by output gap, 30% by MSP, 10% by inflation dynamics.
  - VAR corroborates single-equation results: lagged inflation, MSP, and regime change dummy account for largest share of disinflation; crude and exchange rate minor in baseline period.

### IX. Key empirical conclusions (concise)
- Lagged inflation/persistence (adaptive expectations and institutional propagation) accounts for roughly 45–52% of the disinflation in baseline and alternative specifications.
- The discretionary component of MSPs explains about 20% of the disinflation in the preferred specification (higher if lagged determinants omitted).
- Forward-looking expectations (proxied by New Regime Dummy) explain roughly 33% (range 14–44% across specifications).
- Global factors (crude, food, exchange rate) explain a small share in baseline (< 15%) but can explain more in other horizons (e.g., oil 11% in early 2015 period).
- Wages are predominantly an outcome of lagged inflation (indexation) and have little independent contemporaneous effect on headline CPI once lagged inflation is controlled for.

*Source: _wp16166 - References .............................................................................................................*

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

### _wp16166 - References .............................................................................................................

### I. Introduction: framing the questions
- Over the last 15 years, inflation has been a symptom of macroeconomic stability in India; headline CPI averaged just 4% between 2000 and 2006, then rose and "averaged more than 9% between 2006 and 2013."
- Headline CPI peaked at 12.1% year-on-year growth in November 2013 and "collapsed to 4.3% in December 2014" — a fall of nearly 800 bps in 13 months — before accelerating to 5% in October 2015.
- Key empirical questions posed:
  - What was responsible for the sharp decline in inflation over the last two years?
  - Is the decline durable or transitory?
- Factors present during the disinflation episode: collapse in global oil and commodity prices, sharp fall in global food prices, a new monetary policy regime aimed at anchoring inflation expectations, a new government addressing food supply bottlenecks, and continued restraint on agricultural support prices.
- Methodological approach: uses a simple econometric model to quantify contributions of different factors to disinflation.

### Major findings summarized
- Evolution of inflation in India is driven by a mix of:
  - the state of the business cycle (output gaps),
  - adaptive and forward-looking expectations,
  - institutional mechanisms (determination of agriculture support prices and backward-looking wage indexation) that amplify persistence,
  - global factors (oil and food prices, exchange rates).
- Quantitative decompositions from model simulations for the disinflation between fiscal years 2013/14 and 2014/15:
  - "Close to half of the disinflation between fiscal years 2013/14 and 2014/15 can be attributed to a moderation in the historical dynamics of inflation which influence contemporaneous inflation." This likely reflects softer backward-looking (adaptive) expectations and changes in institutional processes (wage and MSP setting) that increase persistence.
  - "About 20% of the disinflation can be explained by a sharp decline in the growth of the discretionary component of MSPs," after controlling for wages, other input costs, and global food prices.
  - "About a third of the disinflation can be attributed to forward-looking expectations," likely capturing the effect of the new monetary policy regime announced in January 2014 and the perceived permanence of the collapse in oil and commodity prices.
- Additional noteworthy empirical insight:
  - Lagged inflation is a significant determinant of wages; once lagged inflation is included in the model, the explanatory power of wages for inflation falls sharply — implying wages are more an "outcome" than a "driver" of the inflation process, consistent with backward-looking wage indexation.

### II. Some context: the stubbornness and breadth of retail inflation (2006–14)
- CPI-Industrial Workers (CPI-IW) averaged 9.1% year-on-year between January 2006 and December 2014 (96 months).
- Persistence statistics:
  - CPI was above 9% for 51% of those 96 months.
  - CPI was above 8% for 63% of those 96 months.
- Breadth of inflation:
  - Non-food, non-fuel (core) inflation averaged 7.8% during this period, "just 130 bps below headline," indicating that inflation was broad-based and not solely food-driven.
- The recent disinflation (post-2013) is framed as dramatic relative to this prior persistence and breadth.

### Organization of the paper (as presented)
- Section II: context on the state of play before the disinflation.
- Section III: documentation of the sources of disinflation.
- Section IV: estimation of an augmented Phillips Curve for India and quantification of factor contributions to disinflation.
- Section V: conclusions and policy implications.

*Source: _wp16166 - References .............................................................................................................*

### 2010. The purpose of a core inflation index is to get an accurate measure of the current inflation

### _wp16166 - 2010. The purpose of a core inflation index is to get an accurate measure of the current inflation

### III. Followed by the Sharp Disinflation
- New all-India CPI averaged 10.1% in 2012/13 and 9.8% in 2013/14.
- Average inflation fell to 6% in 2014/15 – 400 bps lower than the previous two years.
- In 2015/16 (April–October) inflation averaged 4.6% – another 140 bps lower.
- Momentum of inflation (annualized quarterly growth in seasonally adjusted CPI) fell from 12.9% in the last quarter of 2013 to 2.9% q/q, saar from August–October 2015.
- Distribution of the disinflation between 2013/14 and 2014/15:
  - 62% of the decline in headline CPI attributable to disinflation in food.
  - 38% attributable to non-food; within non-food, 22% of the decline is attributable to core and 16% to fuel prices.
- Food disinflation concentration:
  - Cereals and vegetables (accounting for less than 35% of the CPI food basket) are responsible for 92% of the disinflation in food between 2013/14 and 2014/15.
- Note on definitions:
  - Fuel includes “transport and communication” which has a large component of fuel.
  - Core excludes food, fuel, transport and communication.

### IV. Why Has Inflation Declined? Quantifying the Contribution of Different Factors
- Approach:
  - Estimate a single-equation augmented Phillips Curve where inflation is a function of the output gap, inflation expectations, global factors (crude and food prices), and structural characteristics unique to India.
  - Complementary analysis using a general VAR in the Appendix for identification and dynamics.
- Data and specification:
  - All variables entered as year-on-year growth rates.
  - Quarterly data from 2000Q2 to 2015Q1.
  - Baseline uses 8 lags of CPI inflation and HP-filtered potential output (HP filter with λ = 1600).
  - Output gap, global crude prices, and exchange rate lagged by one quarter.
  - Additional domestic variables: changes in minimum support prices (MSP), rural wages (lagged), and a dummy for below-normal monsoon (takes value 1 when south-west monsoon rainfall < 85%).
  - A “New Regime Dummy” (binary 1 for 2014Q1–2015Q1) to proxy forward-looking expectations following introduction of flexible inflation targeting by the RBI.
- Identification and endogeneity concerns:
  - MSPs and wages may be contemporaneously determined with inflation; MSP and wages are lagged by one quarter to mitigate reverse causality.
  - Inclusion of eight lags of domestic headline inflation and global food inflation to address omitted-variable bias linked to annual MSP/wage-setting processes.
  - Wage growth series showed a unit root under Clemente, Montañés, Reyes (1998) tests; robustness includes first difference of wage growth.

### A. Single Equation Model (functional form and interpretation)
- Phillips-curve conceptualization:
  - Inflation (πt) depends on expected inflation (πt^e) and deviations of activity from its natural level (xt − x*t).
  - Hybrid expectations modeled via lags of headline CPI (adaptive) and the New Regime Dummy (forward-looking proxy).
- Additional explanatory variables included: lagged rural wage growth, lagged MSP growth, below-normal monsoon dummy, lagged global food prices, lagged global crude prices, Rs./$ exchange rate.

### B. Empirical Findings (summary of main results and magnitudes)
- Baseline regression explains about 96% of the variation in headline CPI (R-squared 0.96).
- Output gap:
  - A one percentage point increase in the output gap (1Q lag) increases headline inflation by 52 bps; statistically significant at the 5% level (coefficient 0.52**).
- Inflation expectations / persistence:
  - Several lags of CPI inflation are economically and statistically significant, indicating persistence; lags capture adaptive expectations and institutional propagation (MSP, wage indexation).
- New Regime Dummy:
  - The New Regime Dummy reduces headline CPI by 143 bps on average; statistically significant at the 5% level (Dummy for new monetary policy regime# -1.43**).
  - Interpretation: proxies forward-looking expectations anchoring and/or captures effects of dramatic collapse in oil prices in 2014.
- Minimum Support Prices (MSPs):
  - A one percentage point increase in MSP growth (1Q lag) is associated with a 6 bps increase in headline CPI; weakly statistically significant at the 15% level (Annual growth in MSP - 1Q lag 0.06^).
  - Coefficient interpreted as the “discretionary” component of MSPs after controlling for input costs and global food prices.
- Wages:
  - Annual growth in wages (1Q lag) statistically and economically insignificant once lags of inflation are included (Annual growth in wages - 1Q lag -0.04).
  - Suggests wage variability largely captures past inflation rather than driving current inflation.
- Monsoon:
  - Below-normal monsoon dummy increases headline inflation by almost 67 bps; significant at the 10% level (Dummy for below normal monsoon 0.67*).
- Global factors:
  - World food prices: some lags show explanatory power (e.g., 7Q lag 0.04*), but primary impact appears via MSPs.
  - World crude price: a one percentage point increase in global crude prices (1Q lag) results in a 1 bps increase in headline CPI; statistically distinguishable from zero at the 15% level (Annual growth in world crude price - 1Q lag 0.01^).
  - Exchange rate: a 1 percentage point appreciation of the rupee against the US$ (1Q lag) is associated with a reduction in inflation by 8 bps; significant at the 5% level (Annual growth in $/Re. exchange rate - 1Q lag -0.08**). (Exchange rate represented as US$ per rupee.)
- Model diagnostics and fit:
  - Observations 60
  - R-squared 0.96
  - D-W-statistic 1.96
  - B-G serial correlation test (LM test) 0.16
  - AIC 2.50
  - SBC 3.33
- Overall interpretation:
  - Disinflation since 2013/14–2014/15 driven largely by food disinflation (62%), concentrated in cereals and vegetables (92% of food disinflation), but with a meaningful non-food contribution (38%).
  - Key drivers include a negative output gap effect reversal, anchored expectations (New Regime Dummy), favorable movements in global commodity prices (including crude), exchange rate appreciation, and discretionary components of MSPs.
  - Wages do not appear to be an independent driver of headline inflation once persistence and other factors are controlled for.

### Key statistics and coefficient table (selected coefficients from Table 2)
- Output gap - HP filter - 1Q lag: 0.52**
- Annual growth in wages - 1Q lag: -0.04
- Annual growth in MSP - 1Q lag: 0.06^
- Dummy for below normal monsoon: 0.67*
- Annual growth in $/Re. exchange rate - 1Q lag: -0.08**
- Annual growth in world food price index - various lags: 0.02, -0.03^, 0.01, 0.00, 0.01, -0.02, 0.04*, 0.00
- Annual growth in world crude price - 1Q lag: 0.01^
- Dummy for new monetary policy regime#: -1.43**
- CPI inflation lags (1Q to 8Q): 0.41**, 0.23, 0.18, -0.30**, 0.47**, 0.01, 0.01, -0.21**

*Table 2: Determinants of Inflation in India — Dependent variable: Year-on-Year CPI Inflation Rate ( ** significant at 5% level, * significant at 10% level, ^ significant at 15% level. #: NEWREGIME is binary 1 for 2014Q1–2015Q1).*

*Source: _wp16166 - 2010. The purpose of a core inflation index is to get an accurate measure of the current inflation.*

### Box 1: Inflation, Wages and MSPs: What’s Driving What?

### Box 1: Inflation, Wages and MSPs: What’s Driving What?

### Simultaneity and institutional context
- There is inherent simultaneity among rural wages, minimum support prices (MSPs) and inflation in India; each variable is potentially a function of the other two.
- Understanding temporal directionality is important for: (i) model specification for inflation, (ii) identifying whether MSPs and rural wages are outcomes of inflation or drivers of it, and (iii) understanding high degrees of inflation persistence in India.
- In India’s rural economy a large fraction of the labor force is in the unorganized sector without centralized wage negotiation; feedback from inflation to rural wages is largely “ex post” through administered increases to MGNREGA wages, which are formally indexed every year to the previous year’s CPI-RL inflation and have become the floor of rural wages.

### Wages: outcome versus driver
- Wages are backward looking because of formal indexation to past CPI-RL inflation; today’s wages are influenced more by yesterday’s inflation than by expected future inflation.
- Empirical findings:
  - Wages have economic and statistical significance in the presence of MSPs.
  - Wages cease to be significant once more than one lag of inflation is added.
- Interpretation:
  - Wage variability, controlling for slack, is more an outcome of previous inflation (consistent with backward-looking indexation) than an independent source of inflationary pressure.

### MSPs: independent explanatory power
- MSP inflation serves as a floor price for key food groups and importantly influences food and headline inflation.
- CACP (Commission for Agricultural Costs and Prices) methodology: input costs (which correlate with past inflation) are only one factor; other factors cited include demand and supply; price trends in the market, both domestic and international; inter-crop price parity; terms of trade between agriculture and non-agriculture; likely implications of MSP on consumers of that product.
- Empirical findings:
  - MSPs retain statistical significance even after controlling for multiple lags of inflation and other factors.
  - The influence of MSPs on inflation is robust to the addition of multiple other controls.
- Reverse causality concerns are mitigated because CACP methodology appears based more on historical inputs than expected inflation.
- MSPs and wages have bidirectional dynamics:
  - Past wages play a role in MSP setting (per CACP).
  - MSPs likely drive demand for agricultural labor and thereby rural wages, independent of the business cycle or supply shocks.

### Net assessment on drivers of inflation and persistence
- Empirical conclusion: wages are much more an outcome of the inflation process rather than a driver; a component of MSPs is more “exogenous” and has independent explanatory power even after controls.
- The backward-looking nature of wage indexation and MSP setting are institutional structures that can accentuate the impact of exogenous shocks and lead to persistence of inflation in India.

### Quantifying contributions to disinflation (2013/14 to 2014/15)
- Model performance:
  - Model explains more than 95% of the quarterly variation in inflation.
  - Note on robustness: dropping several lags for parsimony still yields R^2 = 91%.
- Predicted inflation declined by 3.5 percentage points between 2013/14 and 2014/15; actual decline was 3.3 percentage points.
- Table 3a decomposition (contributors to disinflation between 2013/14 and 2014/15):
  - Decline in model predicted inflation: 3.3 percentage points (100%)
  - Lagged Inflation: 1.546 percentage points (46%)
  - MSP: 0.721 percentage points
  - New regime: 1.133 percentage points (33%)
  - Global crude: 0.12 percentage points
  - Exchange rate: 0.411 percentage points
  - Output gap: -0.3 percentage points (-8%)
  - Global food: -0.1 percentage points (-4%)
- Interpretation:
  - Lags of inflation (proxying adaptive expectations and persistence from institutional settings) account for 46% of the disinflation.
  - The new regime dummy (proxying forward-looking inflation expectations) accounts for 33% of the disinflation.
  - The discretionary component of MSPs accounts for 21% of the disinflation between 2013/14 and 2014/15.
  - Global crude prices and exchange rate stability together account for less than 13% of the disinflation.
  - Global food prices and a closing output gap exerted upward pressure on inflation in 2014/15, contributing negatively to explain the disinflation.

### Box 2: Why not a larger role for oil?
- Time-period dependence:
  - In the baseline period (2013/14 to 2014/15), one-quarter lagged year-on-year oil prices declined by 8% on average; oil accounts for a miniscule percentage (2%) of the disinflation in this period.
  - If the horizon is shifted, global factors can play a larger role: oil prices dropped by 66% during the first three quarters of 2015 relative to same period in 2014; oil’s contribution rose to 11% in that period.
  - For that particular time period, the role of global factors (food, exchange rate, oil) rose to 42% compared to just 10% in the baseline period.
- Even in periods where global factors rise in importance, domestic factors remain predominant:
  - Lagged inflation accounts for 52% of the disinflation in the alternative period; together with MSPs they account for 60%.
- Reasons for limited pass-through of oil declines:
  - A small fraction of the oil price decline flowed through into retail prices because retail gasoline and high speed diesel prices were subsidized before oil prices fell.
  - A large fraction of the oil price decline was captured by the government through successive tax increases.
  - Example: between October 1, 2014 and March 15, 2016 crude prices fell 56%, but retail prices of gasoline and high speed diesel only fell 13% and 21% respectively.

### Robustness checks and additional numerical findings
- Robustness specifications:
  - Column [2]: include first difference in wage growth to address non-stationarity in wages.
  - Column [3]: alternative definition of weak monsoon (deficiency of 10% below normal).
  - Columns [4]-[6]: alternative definitions of output gap.
  - Column [7]: use growth in agriculture production instead of weak monsoon dummy.
- Key robustness outcomes:
  - Key variables of interest reported in Table 2 are robust across specifications.
  - Coefficient on wages is statistically indistinguishable from zero in all specifications.
  - Several lags of inflation and the regime change dummy remain significant across most specifications.
  - Baseline results are robust to measurements of the output gap; using the Christiano-Fitzgerald filter increases the effect of output gap on inflation to 60 bps without affecting other results.
  - Monsoon impact: a deficiency greater than 15% has a much larger impact on headline inflation (67 bps) versus the 10% threshold (27 bps), reinforcing non-linearity.
  - World food prices: first, second and seventh lags remain statistically significant at least at the 20 percent level.
- MSP coefficient and interpretation:
  - The coefficient on MSP ranges from 0.05 to 0.12 in columns [1]-[10] of Table 4.
  - Interpretation: a one-percentage point increase in MSP is associated with a 5-12 bps increase in headline CPI inflation.
  - The estimated MSP coefficient is statistically significant at least at the 20% level in all specifications.
- Decomposition robustness:
  - Repeating the decomposition using coefficients from columns [3]-[10] of Table 4: lagged dynamics of inflation remain the biggest contributor to the disinflation between 2013/14 and 2014/15, with their contribution ranging from 37-52%.
  - The new regime dummy accounts for 14-44% of the decline in inflation.

*Source: Box 1, _wp16166 - Box 1: Inflation, Wages and MSPs: What’s Driving What?*

### Box 3: The Role of MSPs: Separating the Wheat from the Chaff

### Box 3: The Role of MSPs: Separating the Wheat from the Chaff

### Identification strategy and sensitivity checks
- Baseline finding: MSPs explain 20% of the disinflation in the baseline specification.
- Rationale: MSPs themselves are influenced by input factors proxied by historical lags of CPI and global food prices; when those determinants are controlled for, the “discretionary” component of MSPs explains a smaller share of disinflation.
- Sensitivity regressions:
  - Column [8] (basic specification dropping all lags of global food prices and CPI) — estimated MSP coefficient rises from 0.06 in the preferred model to 0.32 in column [8].
  - Columns [9] and [10] (drop only selected determinants; include only one and four lags of world food price respectively) — MSP coefficients rise relative to baseline column [1] but remain lower than column [8].
  - Implication: Omitting lagged global food prices and CPI causes MSP to proxy for those omitted influences; specifications omitting these variables are likely mis-specified and inflate the MSP coefficient.

### Quantifying MSP contribution under alternative specifications
- Using the larger MSP coefficients (e.g., column [8]) increases the estimated role of MSPs:
  - MSP completely explains the disinflation between 2013/14 and 2014/15 if estimates from column [8] are used.
  - Using specifications that include some omitted variables (columns [9] and [10]) reduces MSP’s contribution to between 43-48%.
  - Interpretation: world food prices are a significant determinant of MSP, and accounting for prior years’ world food prices reduces the estimated exogenous contribution of MSPs.

### Explaining MSP growth with lags
- Regression of MSP growth on eight lags of CPI inflation and eight lags of world food prices:
  - The lagged variables explain 68% of the variation in MSP growth.
  - Conclusion: a substantial fraction of MSP variation is explained by lagged inflation and world food prices, consistent with CACP’s use of multiple information sources.

### Wages, lagged inflation, and inflation transmission
- Alternative specifications (Table 5) show:
  - Wages have a positive association with headline CPI only in the absence of lags of inflation (Columns [3] and [4]).
  - Using contemporaneous wages and dropping lags of inflation increases the apparent effect of wages.
  - Specific estimate: a one percentage point increase in wage growth increases headline inflation by 0.26 bps in Column [4], statistically significant at the 5% level.
  - Adding even one lag of CPI (Column [5]) restores Durbin Watson statistic close to 2 but reduces the economic significance of wages by 80%.
  - Figure 5 result: eight lags of inflation explain about 80% of the variation in wage growth.
- Interpretation:
  - Lagged inflation is a significant determinant of wages (reflecting backward-looking wage indexation).
  - Once MSPs, output gaps, and lagged inflation are controlled for, wages have no independent influence on headline CPI.
  - Wages operate more as an “outcome” of inflation dynamics than an independent driver, though they matter through indexation mechanisms.

### Vector autoregression (VAR) corroboration and dynamic effects
- VAR specification: lag length of 8 quarters; Choleski ordering with global factors most exogenous, then output gap, wages, MSP, then inflation.
- Impulse response findings:
  - A one standard deviation shock to growth of MSP increases inflation by 0.3 percentage points; the effect is statistically significant but not persistent and is statistically indistinguishable from zero by the second quarter.
  - Wages have a much smaller and non-persistent effect than MSP.
- Variance decomposition over the sample:
  - 50% of variability in inflation is explained by output gap.
  - 30% by MSP.
  - 10% by dynamics of inflation.
- VAR broadly validates single-equation results: lagged inflation, MSP, and the regime change dummy account for the largest share of the disinflation between 2013/14 and 2014/15; crude and exchange rate played minor roles in this period.

### Key empirical conclusions (from the paper’s conclusions)
- 20% of the disinflation can be explained by a sharp decline in the growth of the “discretionary” component of MSPs.
- 45% of the disinflation can be attributed to a moderation in the historical dynamics of inflation that influence contemporaneous inflation (softening of backward-looking/adaptive expectations and institutional persistence via wage/MSP setting).
- Almost 35% of the disinflation can be attributed to forward-looking expectations (which may capture the new monetary policy regime and more benign future expectations for oil and commodities).
- The role of global factors (global crude prices and exchange rates) in explaining the disinflation between 2013/14 and 2014/15 is less than 15%, though their contribution rises for alternative time periods.
- Output gaps closed during this period and thus were putting upward pressure on inflation (i.e., the business cycle did not drive the recent disinflation).

*Source: Box 3, _wp16166 - Box 3: The Role of MSPs: Separating the Wheat from the Chaff*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp16166.pdf_