Date post: | 22-Apr-2018 |
Category: |
Documents |
Upload: | vuongkhanh |
View: | 216 times |
Download: | 2 times |
Forecasting prices in a low inflation environment -
A disaggregated approach over the short term
John Harnett & Javier Papa
1. Introduction & objective
2. Existing evidence
3. The disaggregated approach
4. Estimation strategy & data sources
5. Main results
6. Conclusions, implications & further research
2
Outline
Current Environment
Persistent ultra low inflation or “lowflation” (IMF term)
Low inflation persists even in Ireland, where inflationary
pressures are traditionally larger than in Europe
The consequences of near-zero price growth though are
closely related to those of deflation:
higher real debt stocks and real interest rates
greater unemployment
risk of liquidity trap (Japan)
Monetary policy (i.e. QE) in in the euro area is struggling to
raise inflation rates
3
The main objective is to develop forecasting tools to produce (disaggregated) inflation
projections in the current context of low inflation in Ireland.
1. Introduction & objective
High Inflation
Low Inflation ?
International evidence
Inflation dynamics in the current near-zero price growth environment have been recently
studied for the US (Zaman 2015) and Europe (Conti et al 2015, Riggi et al 2015).
However, there are not many studies/forecasts for the current context of low inflation in
Ireland…
What the existing evidence says on the drivers of Irish inflation?
External factors (e.g. exchange rate, foreign prices) are significant for traded sectors while
domestic factors (e.g. wages, productivity) are significant for non traded sectors (Slevin
2003).
Other domestic factors though (e.g. employment growth, unemployment) were also found to be
important at the aggregate level [Birmingham et al (2006), Gerlach et al (2015)].
Many central banks (including the ECB) look to forecast inflation at a disaggregated level
[Kanutin (2012), Giannone et al (2010), Célérier (2009)].
4
2. Existing evidence
Development StageModel Version1
Completed
Under
Investigation
Judgement Based
Harmonised Index of Consumer Prices (HICP)
Unproc.
Food
Proc.
FoodNEIG Energy
Core
ServicesRents
5
Understand the different dynamics
underlying headline inflation.
Aggregate “Core Inflation”, Goods,
Services (Core + Rents) as needed.
Investigate real economy factors rather
than monetary ones.
6.2%43.8% 10.6%21.1% 12.8% 5.6%
2. The disaggregated approach
Estimation Strategy → Both long run and short run (e.g. error correction) models estimated
Long Run: 𝑃𝑖𝑡 = 𝐶0 + 𝐸𝑖𝑡 . 𝑃𝑖𝑡∗ + 𝐸𝑀𝑃𝑡−1+ 𝜀𝑡
Short Run: ∆𝑃𝑖𝑡= 𝐶0 + 𝛽1𝑃𝑖𝑡−𝑖 + 𝛽2𝐸𝐶𝑀𝑡−1 + 𝛽3𝑗 𝑗=0𝑘 ∆ (𝐸𝑖𝑡 . 𝑃𝑖𝑡
∗ )𝑡−𝑗 + 𝛽4𝑗 𝑗=0𝑘 ∆ 𝐸𝑀𝑃𝑡−𝑖−𝑗 + 𝛽5𝑗 𝑗=1
𝑘 ∆𝑃𝑖 𝑡−1 + 𝑢𝑡
(𝑃𝑖= Irish annual HICP inflation for sub-index I; 𝐸𝑖 = annual rate of change of exchange rate variable; 𝑃𝑖𝑡∗ = foreign price annual inflation in sub-index I; EMP = annual
employment growth; 𝐶0 = constant, 𝜀𝑡& 𝑢𝑡 = residuals )
𝑃∗: when suitable foreign price data was not available – used exchange rate as external influence and time trend
Model Specification → Test for domestic and external factors (following existing literature)
→ Forecasting in quarter (Qt) as explanatory variables need to be available at Qt and Qt+j
Data Sources → quarterly time series constructed using:
HICP inflation – Eurostat: HICP annual rate of change, monthly data
Core Services Inflation – CSO: Core Services price index, monthly data
Annual Employment Growth – CSO: QHNS
Annual Rate of Change in Exchange Rates – Central Bank of Ireland: € to GB£ & NEER
Oil Prices – US$ per barrel, ICE, Brent Crude, FOB North Sea, daily spot and futures
6
2. Estimation strategy & data sources
-2.00
-1.00
0.00
1.00
2004Q2 2005Q4 2007Q2 2008Q4 2010Q2 2011Q4 2013Q2 2014Q4 2016Q2 2017Q4
∆CSI ACTUAL ∆CSI FITTED
-15.00
-10.00
-5.00
0.00
5.00
10.00
15.00
2004Q2 2005Q4 2007Q2 2008Q4 2010Q2 2011Q4 2013Q2 2014Q4 2016Q2 2017Q4
∆ENERGY ACTUAL ∆ENERGY FITTED
7
∆ENERGY = 0.11∆OIL + 0.24*∆ENERGYt-1 + 0.05*ENERGYt-1 - 0.28*ECMt-1
ENERGY = Energy Prices (annual) inflation
Oil = Oil Prices (annual rate of change)
ECM = Error Correction Term (residuals from long run relationship)
∆NEIG = 0.18*∆EMPt-1 + 0.64∆NEIGt-2 - 0.04∆EU_BPt-1 - 0.25*ECMt-1
NEIG = NEIG Prices (annual) inflation
EU_BP = Euro / Sterling Exchange Rate (annual rate of change)
Emp = Employment Growth (annual rate of change
ECM = Error Correction Term (residuals from long run relationship)
∆CSI = 0.3*∆EMPt-5 - 0.05CSIt-1 - 0.48CSIt-4
CSI = Core Services prices (annual) inflation
Emp = Employment growth (annual rate of change)
Energy Inflation
-2.00
-1.00
0.00
1.00
2.00
3.00
2004Q2 2005Q4 2007Q2 2008Q4 2010Q2 2011Q4 2013Q2 2014Q4 2016Q2 2017Q4
∆NEIG ACTUAL ∆NEIG FITTED
5. Main results
Non Energy Industrial Goods Inflation (NEIG)
Core Services Inflation (CSI)
-10.00
-5.00
0.00
5.00
10.00
2004Q2 2005Q4 2007Q2 2008Q4 2010Q2 2011Q4 2013Q2 2014Q4 2016Q2 2017Q4
Contribution to ENERGY
Oil ENERGY (T-I) ∆ENERGY(t-1) ECM (t-1) ∆ENERGY FITTED
-1.80
-1.30
-0.80
-0.30
0.20
0.70
1.20
1.70
2004Q2 2005Q4 2007Q2 2008Q4 2010Q2 2011Q4 2013Q2 2014Q4 2016Q2 2017Q4
Contribution to CSI
CSI(t-1) ∆CSI(t-4) ∆EMP (t-5) ∆CSI FITTED
5. Main results
8
The Energy chart shows the importance of temporary /
external influences (e.g. oil prices) versus domestic
effects on future inflation path.
NEIG inflation is mostly driven by it owns lagged value,
which is probably accounting for the delayed impact of
foreign prices.
The ECM terms also plays a critical role which may be
accounting for domestic prices adjustment to long term
international prices (i.e. ppp effect)
Domestic factors seem to be more important to explain
Core Services inflation, which is a predominantly non
traded sector.
Employment growth in previous periods appears to
account more for Core Services inflation growth, during
the economic downturn.
-1.50
-0.50
0.50
1.50
2.50
2004Q2 2005Q4 2007Q2 2008Q4 2010Q2 2011Q4 2013Q2 2014Q4 2016Q2 2017Q4
Contribution to NEIG
∆Emp(t-1) ∆NEIG(t-2) ∆EU_BP(t-1) ECM(t-i) ∆NEIG FITTED
6. Conclusions, implications & further research
Conclusion: What is the inflation outlook over the short term?
Services inflation will continue to drive general prices upwards,
Energy goods will act as a drag on inflation this year before rebounding in 2017,
NEIG will continue to act as a drag on inflation over the next two years.
Based on this model, inflation is not likely to remain in a low-inflation zone, but rather to increase moderately in 2016
before accelerating in 2017.
Implications for policy-makers
The forecasting tools presented here are expected to assist policy-makers in anticipating whether a general rise (decline)
in inflation rates is due to inflationary (deflationary) processes or due to sector-specific underlying dynamics
Implications for other users
Inflation forecasts are widely used across the civil service as an input (via deflators) to estimate real economy variables
such as real GDP, real wages, productivity, real interest rates, as well as evaluations / assessment purposes etc.
9
6. Conclusions, implications & further research
What are the areas for further research ?
Investigate feasibility of developing food consumer price forecast models
Complement disaggregate approach (bottom up) with aggregate model (top down)
Models testing:
Current ECM estimates are consistent with non-parametric models (i.e. quantile regression)
Further model testing is planned though (e.g. Bayesian Autoregressive models)
Robustness check:
Enhance model specification by including explanatory variables on labour costs and foreign prices
10
11
Thank you for your attention!