Abstract
Republic of Poland: Selected Issues
Forecasting Inflation in Poland1
Inflation in Poland has stayed below the lower bound of the target band for about two years with external shocks adding to downward pressure during 2014. Alongside, inflation developments have been increasingly disjoint from the gradual narrowing of the output gap, while declining inflation expectations suggest that indirect and second-round effects from low inflation may be taking hold. Forecasting inflation in the context of persistent low inflation requires a set of complementary analytical tools. To help inform monetary policy advice, we explore the drivers of inflation and forecast inflation using several different models, allowing also for indirect and second-round effects. Under a no-policy-change scenario, the models point to a protracted period of low inflation, despite continued growth momentum.
A. Stylized Facts
1. Inflation in Poland has declined markedly since mid-2012 despite monetary easing. After running well above the 2.5 percent target,2 year-on-year inflation declined rapidly from 4.3 percent in June 2012 to 0.2 percent one year later as growth weakened. Despite a substantial monetary easing cycle, which entailed a decline in the main policy interest rate from 4.75 percent at end-October 2012 to 2.5 percent in July 2013, inflation failed to return to target and fell into deflationary territory in July 2014. Since then, disinflation has continued, with year-on-year inflation reaching a historic low of -1.6 percent in February 2015 before moderating in March and April. Two additional policy interest rate cuts were implemented (50 basis points in October last year and again in March this year).
2. Low inflation has been prevalent across all major components of inflation. Contributing to the slowdown in headline inflation was not only subdued food and energy price inflation (as well as disappearing base effects3 in July 2014) but also weakening core inflation. While the primary measure of core inflation (headline inflation, excluding food and energy) declined less rapidly than headline inflation, it nonetheless fell to close to zero toward the end of last year (Figure 1).
Recent Inflation Developments, 2004–15
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Recent Inflation Developments, 2004–15
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Recent Inflation Developments, 2004–15
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
3. While economic growth strengthened, inflation remained low. As the monetary easing translated into a pick-up in economic activity,4 real GDP growth strengthened. Quarter-on-quarter (seasonally adjusted) growth picked up in the second quarter of 2013, marking a start of recovery. As a result, the negative output gap started to narrow. In 2014, growth reached 3.4 percent on the back of robust domestic demand growth, while external demand remained subdued. The strong growth, combined with continued disinflation, highlights the challenge in projecting inflation over the monetary policy horizon.
4. In this chapter, we employ a suite of models to determine the main drivers of inflation and provide a range of inflation forecasts to assess the likelihood of protracted low inflation. First, we consider the main factors underlying recent inflation developments and assess the importance of first-round indirect5 (working through input-output linkages) and second-round (working through expectations) effects of external shocks for headline inflation. Then, using a variety of models, we provide possible forecast paths for inflation. This should help inform the likelihood of inflation returning to the target band in the near term in the absence of further monetary policy action.
B. Drivers of Recent Inflation Developments
5. The historically tight link between headline CPI and the output gap has weakened. Allowing for a transmission lag of two quarters, year-on-year headline inflation has until recently closely followed output gap developments (Figure 2).6 This is also evident from a quarterly regression of headline CPI on the 2-quarter lagged output gap, using data since the introduction of the 2.5 percent inflation target in 2004. However, since late-2013, inflation has started to diverge from the path implied by the output gap. This suggests that other factors than the degree of domestic slack are currently important for understanding inflation dynamics. Hence, a traditional model that relies only on the output gap for projecting inflation is becoming less reliable and need to be complemented with more elaborate models.
Potential Output and Inflation, 2004–14
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Potential Output and Inflation, 2004–14
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Potential Output and Inflation, 2004–14
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
6. The output gap-inflation link is also weakening for core inflation. A regression of core inflation on the two-quarter lagged output gap suggests core inflation dynamics have until recently been better anchored to real sector activity than headline inflation. While this simple relationship cannot fully capture the magnitude of all past fluctuations in inflation, it has often fared well in fitting the direction of change. However, core inflation has also failed to pick up recently, despite the narrowing output gap and accelerating domestic demand and wages.
7. Weak core inflation may indicate economic slack or increasing prevalence of indirect and second-round effects. On the one hand, weak core inflation may indicate that the output gap is currently larger than what some measures suggest.7 However, various publicly available output gap estimates point to a narrowing gap, which also appears consistent with the rapid pace of unemployment reduction and robust nominal wage growth. On the other hand, the protracted nature of the weak core inflation implies that the dynamics are not solely explained by one-off factors. Subdued core inflation amid falling international commodity prices and accelerating domestic demand may indeed imply the presence of first-round indirect and second-round effects on inflation.
8. Indirect and second-round effects can occur on the back of a number of factors, exogenous to Polish economic growth (Figure 3).
Weak food price inflation. Weak food price inflation materialized on the back of subdued world food price inflation, an unseasonably mild 2013–14 winter in Poland, and the Russian ban on imports from Poland (and other European Union (EU) countries) for a number of food and vegetable items.
Weak energy price inflation. Subdued energy price inflation has been a drag on headline CPI during 2013–14. In addition, tax changes (including a reduction in electricity tariffs in July 2013 and January 2014) and a decline in communication prices further reduced year-on-year inflation. The sharp drop in world oil prices in late-2014 exerted additional strong downward pressure.
Weak imported inflation. Spillovers from low inflation in the euro area have also been a drag on Polish inflation. Foreign value added in Polish domestic aggregate demand amounts to about 18 percent with 40 percent originating from the euro area. Hence, German and euro area inflation at -0.1 and -0.3 percent, respectively, in February 2015, against the backdrop of a relatively stable euro/zloty exchange rate, have put a drag on Polish inflation.
Drivers of Recent Inflation Developments, 2004–15
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Drivers of Recent Inflation Developments, 2004–15
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Drivers of Recent Inflation Developments, 2004–15
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
9. Weak producer price inflation suggests the prevalence of indirect effects. As imported inflation weakened, producer prices started to decline, while headline CPI inflation was not immediately affected. With producer prices declining sharply during 2012 and remaining subdued since then, low inflation became increasingly prevalent throughout the economy, as low producer prices fed into headline CPI. Hence, headline inflation was affected not only by the direct effect of oil prices and their presence in the consumer basket but also through indirect effects such as declining transport costs resulting from lower oil prices. The decline was also reflected in the GDP deflator, with year-on-year inflation in the GDP deflator falling below zero at the end of last year.
10. In addition, declining inflation expectations indicate that second-round effects may be materializing. Consensus forecasts for the following year have dropped significantly. Similarly, the 8-quarter ahead inflation forecast from the Narodowy Bank Polski (NBP) survey of professional forecasters suggests medium-term inflation expectations are also drifting downward (Figure 4). Hence, low inflation is becoming increasingly entrenched in expectations, highlighting concerns about second-round effects in inflation dynamics. Nonetheless, nominal wage growth continues to hold up well at around 3 percent year-on-year.
Inflation and Interest Rate Expectations and Inflationary Pressure, 2006–15
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Inflation and Interest Rate Expectations and Inflationary Pressure, 2006–15
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Inflation and Interest Rate Expectations and Inflationary Pressure, 2006–15
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Quantifying indirect and second-round effects
11. Indirect and second-round effects translate shocks into core inflation and inflation expectations. While a transitory supply shock affects headline inflation, in the absence of indirect and second-round effects, it may not necessarily alter core inflation, which is mainly driven by domestic demand pressures. However, supply shocks can feed into core inflation through its impact on subcomponents of CPI and on inflation expectations. For example, as transport costs drop following a decline in oil prices, services inflation may eventually decline as well. In turn, this exerts downward pressure on core inflation. And as some other components of the CPI affect each other (through input-output type linkages), expectations can adjust and create second-round effects on inflation.
12. Determining the significance of second-round effects is important for policy decisions. Assessing the magnitude and persistence of second-round effects is instrumental in understanding the extent to which current food and energy supply shocks feed into core inflation and become entrenched in inflation expectations. This is particularly important for policy makers who should aim to minimize unanchoring of inflation expectations and the associated second-round effects to the largest extent possible. If models do not account for second-round effects, the policy response needed to contain these effects may be underestimated. While Section C below explores a number of different models for forecasting inflation, we start by quantifying the combined indirect and second-round effects. As a complement to this analysis, the Cross-Country Selected Issues Paper (IMF, 2015) examines the causes and drivers of low inflation in the Czech Republic, Poland, Sweden, and Switzerland. It disentangles the impact of external shocks into direct first-round, indirect first-round, and second-round effects, highlighting the importance of the latter two and associated policy considerations.
Estimating indirect and second-round effects in a disaggregate model
13. We estimate a disaggregate model that allows to decompose shocks to inflation into an inertial component and a factor capturing indirect and second-round effects (Leon, 2012).8 This method involves solving a system of equations to provide a path for year-on-year inflation, taking into account indirect and second-round effects through the various components of inflation and allowing for exchange rate pass-through. The system can be described as follows:
Producer price index (PPI) equation: Inflation in the PPI at period t is modeled as a function of exogenous macro and external variables:
Here, Zt is a vector of variables capturing macro factors, such as the nominal exchange rate vis-à-vis the euro (to account for the pass-through into producer prices), world oil and coal price inflation, labor market dynamics (proxied by the monthly unemployment rate from the labor force survey), and inflation in the euro area to take into account production linkages in supply chain networks with core European countries such as Germany.
Disaggregate equations: Inflation in each of the 12 components (i) of headline CPI (see Appendix II) in period t are expressed as functions of their own lags, lags of other components of headline CPI (which captures indirect and second-round effects), and the PPI:
To allow for sufficient degrees of freedom, we reduce each equation by assessing which variables and lag lengths are most informative in each equation.
CPI equation: The headline CPI equation aggregates the CPI components using weights as they appear in the consumption basket:
Solving the system: Using data starting in January 2010, we solve the system using Broyden’s Method (Broyden, 1965) (see Appendix II for further details and the full set of results). Estimating the model over this period provided a good fit of the model, while still allowing for substantial in-sample fluctuations in inflation.
14. The disaggregate model has a number of advantages. First, it is particularly suitable to examine the determinants of inflation during periods of low inflation as the granularity provided by disaggregate data allows to understand the multi-sectoral linkages, which explain the low inflation environment. That is, the source of forecast variation does not depend only on aggregate macro-economic data but also on the information contained at the disaggregate level. Second, the model allows to estimate the magnitude of combined indirect and second-round effects as measured through (Granger) causality and interaction amongst components forming the CPI. Third, the framework allows to build various scenarios, including shocks to specific macroeconomic variables or idiosyncratic shocks to the CPI components. Finally, the labor market is captured by including the unemployment rate in the PPI equation. As evidenced by the declining unemployment rate and relatively robust wage growth, labor market dynamics do not currently point to significant second-round effects. Nonetheless, weak inflation expectations could gradually result in lower wages as these get renegotiated.
Assessing second-round effects
15. We identify combined indirect and second-round effects by simulating exogenous shocks in the model. We begin the analysis of indirect and second-round effects by assuming a transitory (and alternatively a more protracted) exogenous shock to PPI inflation. Two modes of interaction across CPI components are then explored:
One that allows for feedback effects between groups of the CPI (i.e., uses the model with the full interaction—see interaction term in the disaggregate equation above); and
One that captures inertia by not allowing for feedbacks between components (i.e., assuming the interaction term is zero).
Indirect and second-round effects are then computed as the difference between the effects of the shock on overall CPI inflation derived from the two models.
16. The results suggest that combined indirect and second-round effects are important in Poland. We estimate the effects of declines in energy commodities prices (oil and coal prices) and the effects of disinflation in the euro area. The energy commodity price shock mimics the recent sharp decline in international oil prices. That is, for the oil and coal price shock, we assume a protracted 9 percentage point drop in year-on-year commodity price inflation every month for 7 months as observed between June 2014 and February 2015. For the euro area inflation shock, we assume a decline in the monthly year-on-year euro area inflation rate by 0.1 percentage point for 18 months as observed between July 2013 and January 2015. The results suggest strong and persistent indirect and second-round effects, which remain sizeable, exhibiting strong persistence, for more than a year following the shock. Hence, while the greatest initial impact is explained by inflation inertia, the effect through other CPI components (the indirect and second-round effect) in the consumer basket could be substantial and persistent and should not be disregarded (Figure 5). The effects are sizeable:
The cumulative 63 percentage points decline in oil and coal price inflation over 7 months lowers headline CPI inflation by 0.3 percentage points compared to the baseline CPI inflation (the path suggested by the model absent a shock) after nine months. About 20 percent of this effect is accounted for by indirect and second-round effects. After a year and a half, indirect and second-round effects continue to affect inflation and represent more than 40 percent of the change in CPI inflation.
The cumulative 2 percentage points decline in euro area CPI inflation over 18 months lowers headline CPI inflation by 0.15 percentage points compared to the baseline CPI inflation after about one year. At this horizon, about 25 percent of the effect is accounted for by indirect and second-round effects. Half a year later, indirect and second-round effects continue to affect inflation and represent more than 25 percent of the change in CPI inflation.
Disaggregate Model with Indirect and Second-Round Effects, 2010–16
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Disaggregate Model with Indirect and Second-Round Effects, 2010–16
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Disaggregate Model with Indirect and Second-Round Effects, 2010–16
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
17. The results of these simulations are important to understand the sources of deflationary pressures. The simulations, aimed at replicating long-lasting shocks to commodity and trading partner inflation, show that indirect and second-round effects reach their peak in about a year. This would imply that some of the current deflationary pressures on CPI inflation are carrying over indirect and second-round effects of shocks that materialized in the past and which continue to have contemporaneous effects. The decline in oil prices observed since late-2014 is therefore likely to continue feeding deflationary pressures as indirect and second-round effects emerge with a lag and then persist.
C. Forecasting Inflation
18. To assess inflation going forward, we estimate a range of different forecasting models. This section provides the details of three underlying models: (i) a Short-Term Forecasting System (STFS); (ii) an augmented Phillips curve; and (iii) the disaggregate model with indirect and second-round effects presented above. We allow for different lengths of the underlying time series for the various methods to obtain the best fit for the regressions. Specific forecast assumptions are listed in Appendix I.
Short-term forecasting system
19. The STFS model provides quarterly forecasts based on six different underlying models:
Autoregressive (AR). Fits a univariate autoregressive process of order p on inflation.
Bridge Equations (BRIDGE). Bridges monthly data on various indicators with quarterly inflation data, using increasing amount of information as provided by the most recent data releases.
Bivariate VARs (BIVAR). Combines, at the monthly frequency, VAR forecasts from a number of bivariate VARs with indicators for inflation together with inflation.
Bayesian VAR (BVAR). Similar to the BIVAR but extends by pooling a set of useful indicators that may exhibit dynamic interactions with inflation into a single VAR equation.
Dynamic Factor Model (DFM). Utilizes the information content of a large set of macroeconomic indicators by assuming that the co-movement of macroeconomic variables is driven by a few (unobserved) common factors.
Dynamic Factor Model with Targeted Predictor (DFTMTP). More parsimonious version of DFM by using a small subset of monthly indicators (“targeted predictor”) to estimate a single common factor (as opposed to r factors in DFM).
20. The STFS model suggests a quarterly inflation path consistent with the desk baseline (Table 1). Based on data dating back to the early 2000s, the STFS model point to seasonally adjusted quarter-on-quarter inflation of around 0.3 percent in the third quarter. While this is broadly in line with the desk baseline of 0.2 percent, the estimate is subject to substantial variation, depending on the underlying model considered. In fact, the STFS forecast for the third quarter varies from 0.1 percent quarter-on-quarter for the DFM model to 1.2 percent for the BRIDGE model.
STFS Model: Inflation Forecast
(Percent, quarter-on-quarter, seasonally adjusted)
STFS Model: Inflation Forecast
(Percent, quarter-on-quarter, seasonally adjusted)
| 2014Q1 | 2014Q2 | 2014Q3 | 2014Q4 | 2015Q1 | 2015Q2 | ||
|---|---|---|---|---|---|---|---|
| Actual | 0.0 | −0.3 | 0.0 | −0.4 | −0.6 | ||
| As of April 1, 2015 | |||||||
| AR | −0.8 | −0.4 | |||||
| BRIDGE | 1.0 | 1.1 | |||||
| BIVAR | −0.8 | −0.6 | |||||
| BVAR | −0.8 | −0.6 | |||||
| DFM | −0.7 | 0.0 | |||||
| DFMTP | −0.7 | 0.0 | |||||
| Current desk baseline | 0.1 | ||||||
STFS Model: Inflation Forecast
(Percent, quarter-on-quarter, seasonally adjusted)
| 2014Q1 | 2014Q2 | 2014Q3 | 2014Q4 | 2015Q1 | 2015Q2 | ||
|---|---|---|---|---|---|---|---|
| Actual | 0.0 | −0.3 | 0.0 | −0.4 | −0.6 | ||
| As of April 1, 2015 | |||||||
| AR | −0.8 | −0.4 | |||||
| BRIDGE | 1.0 | 1.1 | |||||
| BIVAR | −0.8 | −0.6 | |||||
| BVAR | −0.8 | −0.6 | |||||
| DFM | −0.7 | 0.0 | |||||
| DFMTP | −0.7 | 0.0 | |||||
| Current desk baseline | 0.1 | ||||||
21. The STFS model is mainly useful at the short-term horizon. The STFS model may be particularly useful for forecasts at the current-quarter horizon, before the final inflation data are available but after underlying high-frequency indicators have been released. As these models are notoriously backward looking and rely on actual data releases, forecasts at longer horizons are subject to large standard errors. Hence, researchers tend to focus only on near-term forecasts when using the STFS model. In addition, the STFS model can usefully inform about the direction of change in the forecast, hence providing a view on whether the release of new high-frequency data point to upward or downward revisions relative to the previous projection.
Augmented Phillips curve
22. We estimate an open-economy augmented Phillips curve. Our baseline model links inflation to inflation expectations as determined by adaptive expectations, real activity, and the nominal effective exchange rate. Using monthly data from 2004, we explore various permutations of the following equation, including by allowing for forward-looking expectations (Gali and Gertler, 1999) and controlling for imported inflation and supply shocks. This approach is similar to that used in some other recent studies such as Iossifov and Podpiera (2014) and IMF (2014):
Where
πt = headline inflation.
= inflation expectations. We explore results, using both forward-looking consensus forecasts as well as lagged inflation, consistent with adaptive consumer expectations. gapt = real activity gap. Obtained at the monthly frequency by interpolation (cubic spline) of a quarterly output gap series.
= inflation in the euro area. χt = vector of supply-side shocks and other leading indicators. This includes year-on-year world oil price inflation, world food price inflation (based on prices in U.S. dollars), and the nominal effective exchange rate.
23. The various models explain the majority of fluctuations in inflation. Table 2 shows the results from exploring the significance of lags and taking into account various explanatory variables including year-on-year growth in the money supply (M3) (see for example Kim and Molagoda, 2011). Overall, the models can account for around 95 percent of the variation in inflation. Lagged inflation and the output gap are highly important, and an appreciation of the nominal effective exchange rate is, as expected, associated with lower inflation.
Augmented Phillips Curve: Coefficient Estimates
(Dependent variable: CPI inflation, year-on-year in percent)
Augmented Phillips Curve: Coefficient Estimates
(Dependent variable: CPI inflation, year-on-year in percent)
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | |
|---|---|---|---|---|---|---|---|
| Constant | 1.3205 | 0.8131 | 1.2037 | −1.0441 | −1.0422 | 2.6043 | −1.9214 |
| (0.0115) | (0.0683) | (0.0102) | (0.0000) | (0.0000) | (0.0003) | (0.0000) | |
| CPI inflation (-1) | 0.9427 | 1.1281 | 0.9411 | 0.7696 | 0.7536 | 1.1350 | |
| (0.0000) | (0.0000) | (0.0000) | (0.0000) | (0.0000) | (0.0000) | ||
| CPI inflation (-2) | −0.1851 | −0.2006 | |||||
| (0.0466) | (0.0192) | ||||||
| Inflation expectations 1-year ahead | 1.5672 | ||||||
| (0.0000) | |||||||
| Inflation expectations 1-year ahead (-3) | 0.5619 | 0.5893 | |||||
| (0.0000) | (0.0000) | ||||||
| Output gap (-1) | 0.3661 | ||||||
| (0.0105) | |||||||
| Output gap (-2) | 0.9930 | 1.0075 | −0.3140 | 0.6671 | |||
| (0.0025) | (0.0004) | (0.0318) | (0.0056) | ||||
| Output gap (-3) | −1.5498 | −1.6923 | −1.0169 | ||||
| (0.0095) | (0.0015) | (0.0219) | |||||
| Output gap (-4) | 0.6933 | 0.8129 | 0.4264 | ||||
| (0.0254) | (0.0043) | (0.0701) | |||||
| Output gap (-5) | 0.0819 | ||||||
| (0.0809) | |||||||
| Output gap (-6) | 0.7651 | ||||||
| (0.1446) | |||||||
| Output gap (-7) | −2.1668 | ||||||
| (0.0270) | |||||||
| Output gap (-8) | 1.9145 | ||||||
| (0.0003) | |||||||
| Unemployment gap (-3) | −0.1155 | ||||||
| (0.0014) | |||||||
| NEER | −0.0131 | −0.0259 | |||||
| (0.0122) | (0.0003) | ||||||
| NEER (-1) | −0.0072 | −0.0114 | |||||
| (0.1049) | (0.0141) | ||||||
| NEER appreciation | −0.0548 | ||||||
| (0.0000) | |||||||
| NEER appreciation (-1) | −0.0261 | −0.0263 | |||||
| (0.0000) | (0.0000) | ||||||
| World food price inflation | 0.0100 | 0.0081 | |||||
| (0.0009) | (0.0057) | ||||||
| World food price inflation (-3) | 0.0223 | 0.0206 | 0.0232 | ||||
| (0.0006) | (0.0003) | (0.0000) | |||||
| World food price inflation (-4) | −0.0152 | −0.0175 | −0.0209 | ||||
| (0.0127) | (0.0016) | (0.0001) | |||||
| World oil price inflation | 0.0004 | 0.0028 | 0.0141 | ||||
| (0.8031) | (0.0761) | (0.0026) | |||||
| World oil price inflation (-1) | 0.0082 | 0.0075 | 0.0094 | ||||
| (0.0009) | (0.0011) | (0.0000) | |||||
| World oil price inflation (-2) | −0.0095 | −0.0049 | −0.0060 | ||||
| (0.0003) | (0.0359) | (0.0063) | |||||
| Euro area CPI inflation | 0.1969 | ||||||
| (0.0746) | |||||||
| Euro area CPI inflation (-1) | −0.1739 | ||||||
| (0.1025) | |||||||
| M3 growth (-2) | 0.0495 | ||||||
| (0.0032) | |||||||
| M3 growth (-3) | −0.0501 | ||||||
| (0.0030) | |||||||
| R-squared | 0.9503 | 0.9532 | 0.9625 | 0.9743 | 0.9753 | 0.9553 | 0.8647 |
| Adjusted R-squared | 0.9484 | 0.9510 | 0.9597 | 0.9720 | 0.9729 | 0.9532 | 0.8583 |
| Observations | 134 | 134 | 134 | 134 | 134 | 134 | 134 |
Augmented Phillips Curve: Coefficient Estimates
(Dependent variable: CPI inflation, year-on-year in percent)
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | |
|---|---|---|---|---|---|---|---|
| Constant | 1.3205 | 0.8131 | 1.2037 | −1.0441 | −1.0422 | 2.6043 | −1.9214 |
| (0.0115) | (0.0683) | (0.0102) | (0.0000) | (0.0000) | (0.0003) | (0.0000) | |
| CPI inflation (-1) | 0.9427 | 1.1281 | 0.9411 | 0.7696 | 0.7536 | 1.1350 | |
| (0.0000) | (0.0000) | (0.0000) | (0.0000) | (0.0000) | (0.0000) | ||
| CPI inflation (-2) | −0.1851 | −0.2006 | |||||
| (0.0466) | (0.0192) | ||||||
| Inflation expectations 1-year ahead | 1.5672 | ||||||
| (0.0000) | |||||||
| Inflation expectations 1-year ahead (-3) | 0.5619 | 0.5893 | |||||
| (0.0000) | (0.0000) | ||||||
| Output gap (-1) | 0.3661 | ||||||
| (0.0105) | |||||||
| Output gap (-2) | 0.9930 | 1.0075 | −0.3140 | 0.6671 | |||
| (0.0025) | (0.0004) | (0.0318) | (0.0056) | ||||
| Output gap (-3) | −1.5498 | −1.6923 | −1.0169 | ||||
| (0.0095) | (0.0015) | (0.0219) | |||||
| Output gap (-4) | 0.6933 | 0.8129 | 0.4264 | ||||
| (0.0254) | (0.0043) | (0.0701) | |||||
| Output gap (-5) | 0.0819 | ||||||
| (0.0809) | |||||||
| Output gap (-6) | 0.7651 | ||||||
| (0.1446) | |||||||
| Output gap (-7) | −2.1668 | ||||||
| (0.0270) | |||||||
| Output gap (-8) | 1.9145 | ||||||
| (0.0003) | |||||||
| Unemployment gap (-3) | −0.1155 | ||||||
| (0.0014) | |||||||
| NEER | −0.0131 | −0.0259 | |||||
| (0.0122) | (0.0003) | ||||||
| NEER (-1) | −0.0072 | −0.0114 | |||||
| (0.1049) | (0.0141) | ||||||
| NEER appreciation | −0.0548 | ||||||
| (0.0000) | |||||||
| NEER appreciation (-1) | −0.0261 | −0.0263 | |||||
| (0.0000) | (0.0000) | ||||||
| World food price inflation | 0.0100 | 0.0081 | |||||
| (0.0009) | (0.0057) | ||||||
| World food price inflation (-3) | 0.0223 | 0.0206 | 0.0232 | ||||
| (0.0006) | (0.0003) | (0.0000) | |||||
| World food price inflation (-4) | −0.0152 | −0.0175 | −0.0209 | ||||
| (0.0127) | (0.0016) | (0.0001) | |||||
| World oil price inflation | 0.0004 | 0.0028 | 0.0141 | ||||
| (0.8031) | (0.0761) | (0.0026) | |||||
| World oil price inflation (-1) | 0.0082 | 0.0075 | 0.0094 | ||||
| (0.0009) | (0.0011) | (0.0000) | |||||
| World oil price inflation (-2) | −0.0095 | −0.0049 | −0.0060 | ||||
| (0.0003) | (0.0359) | (0.0063) | |||||
| Euro area CPI inflation | 0.1969 | ||||||
| (0.0746) | |||||||
| Euro area CPI inflation (-1) | −0.1739 | ||||||
| (0.1025) | |||||||
| M3 growth (-2) | 0.0495 | ||||||
| (0.0032) | |||||||
| M3 growth (-3) | −0.0501 | ||||||
| (0.0030) | |||||||
| R-squared | 0.9503 | 0.9532 | 0.9625 | 0.9743 | 0.9753 | 0.9553 | 0.8647 |
| Adjusted R-squared | 0.9484 | 0.9510 | 0.9597 | 0.9720 | 0.9729 | 0.9532 | 0.8583 |
| Observations | 134 | 134 | 134 | 134 | 134 | 134 | 134 |
24. The Phillips curve regressions point to inflation remaining below the target band for a protracted period (Figure 6). While the sixth model predicts inflation to reach the target by the end of 2016, the other models point to a more gradual pick-up, with several models remaining below the target band during the projection period. The models suggest that after accounting for monetary policy transmission lags from the recent policy interest rate cuts as factored into staff’s baseline growth projections, inflation is not expected to reach the 2.5 percent target by end-2016. While staff’s baseline projection is set to reach the lower end of the target band by end-2016, the results point to downside risks to this forecast.
Augmented Phillips Curve, 2009–16
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Augmented Phillips Curve, 2009–16
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Augmented Phillips Curve, 2009–16
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Disaggregate model with indirect and second-round effects
25. The disaggregate model also points to inflation well below the target band for a protracted period (Figure 7). As an alternative forecast method, we project inflation using the disaggregate model presented in section B above. While the model is somewhat sensitive to assumptions on lags in the underlying models for CPI components, the baseline forecast from this approach points to inflation of around 1.3 percent by end-2016. This is broadly in line with results from the augmented Phillips curve presented above.
Disaggregate Model with Second-Round Effects, 2010–16
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Disaggregate Model with Second-Round Effects, 2010–16
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Disaggregate Model with Second-Round Effects, 2010–16
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
D. Outlook and Risk
Assessing the inflation forecast
26. Staff’s current baseline inflation forecast incorporates continued temporary deflation followed by protracted low inflation. Year-on-year deflation is expected to continue through 2015. This results from a generalized lack of inflationary pressure. With world oil prices assumed to increase only at a gradual pace, energy price inflation would remain a relatively minor contributor to positive inflation. Food price inflation is also expected to remain subdued at a worldwide level. In turn, despite robust domestic demand growth, indirect and second-round effects as well as imported low inflation will prevent a marked pick-up in core inflation. Overall, headline inflation is therefore projected to return to the lower end of the target band only by late-2016.
Risks to the inflation outlook
27. The models suggest the baseline inflation forecast is subject primarily to downside risk (Table 3 and Figure 8). The current baseline inflation projection falls in the upper range of what the models indicate. Hence, the likelihood of further prolonged low inflation is high. Nonetheless, the 2016 year-average is broadly in line with the NBP’s March projection of just below one percent (Table 4).
Baseline Forecast and Model Results, 2010–16
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Baseline Forecast and Model Results, 2010–16
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Baseline Forecast and Model Results, 2010–16
Citation: IMF Staff Country Reports 2015, 183; 10.5089/9781513518589.002.A001
Inflation Forecasts From Various Models
(Percent, year-on-year)
Inflation Forecasts From Various Models
(Percent, year-on-year)
| Actual | Phillips | Disagg. | Desk | NBP | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | baseline | March 2015 | |||||
| 2014 | July | −0.2 | |||||||||||
| August | −0.3 | ||||||||||||
| September | −0.3 | ||||||||||||
| October | −0.6 | ||||||||||||
| November | −0.6 | ||||||||||||
| December | −1.0 | ||||||||||||
| 2015 | January | −1.4 | |||||||||||
| February | −1.6 | ||||||||||||
| Forecast | March | −1.5 | −1.6 | −1.4 | −1.4 | −1.7 | −1.7 | −1.5 | −1.1 | −1.3 | |||
| April | −1.1 | −1.7 | −1.2 | −1.2 | −2.0 | −2.0 | −1.5 | −0.8 | −1.3 | ||||
| May | −1.8 | −0.9 | −1.1 | −2.2 | −2.2 | −1.4 | −0.6 | −1.3 | −1.1 | ||||
| June | −1.8 | −0.7 | −1.1 | −2.3 | −2.3 | −1.3 | −0.4 | −1.1 | −1.0 | ||||
| July | −1.8 | −0.6 | −1.1 | −2.2 | −2.2 | −1.2 | −0.3 | −0.9 | −0.9 | ||||
| August | −1.8 | −0.4 | −0.9 | −2.1 | −2.1 | −1.1 | −0.3 | −0.8 | −0.9 | ||||
| September | −1.7 | −0.2 | −0.8 | −1.9 | −1.8 | −0.9 | −0.2 | −0.8 | −0.8 | ||||
| October | −1.6 | 0.0 | −0.6 | −1.7 | −1.6 | −0.7 | −0.1 | −0.7 | −0.6 | ||||
| November | −1.6 | 0.1 | −0.4 | −1.4 | −1.4 | −0.5 | 0.2 | −0.6 | −0.5 | ||||
| December | −1.5 | 0.3 | −0.2 | −1.1 | −1.0 | −0.2 | 0.5 | −0.4 | −0.2 | ||||
| 2016 | January | −1.4 | 0.4 | 0.1 | −0.8 | −0.7 | 0.2 | 0.8 | −0.2 | 0.0 | |||
| February | −1.2 | 0.6 | 0.4 | −0.7 | −0.5 | 0.5 | 0.7 | −0.1 | 0.1 | ||||
| March | −1.1 | 0.7 | 0.2 | −0.8 | −0.7 | 0.9 | 0.7 | 0.0 | 0.2 | ||||
| April | −1.0 | 0.8 | 0.3 | −0.7 | −0.6 | 1.2 | 0.7 | 0.2 | 0.4 | ||||
| May | −0.8 | 1.0 | 0.4 | −0.5 | −0.4 | 1.4 | 0.8 | 0.4 | 0.6 | ||||
| June | −0.7 | 1.1 | 0.5 | −0.5 | −0.3 | 1.7 | 0.9 | 0.5 | 0.8 | ||||
| July | −0.6 | 1.2 | 0.6 | −0.4 | −0.3 | 1.9 | 0.9 | 0.7 | 0.9 | ||||
| August | −0.4 | 1.3 | 0.7 | −0.4 | −0.2 | 2.1 | 0.9 | 0.8 | 1.0 | ||||
| September | −0.3 | 1.4 | 0.7 | −0.3 | −0.2 | 2.3 | 0.9 | 1.0 | 1.1 | ||||
| October | −0.2 | 1.5 | 0.8 | −0.3 | −0.2 | 2.4 | 1.0 | 1.1 | 1.3 | ||||
| November | 0.0 | 1.5 | 0.9 | −0.2 | −0.1 | 2.5 | 1.0 | 1.2 | 1.4 | ||||
| December | 0.1 | 1.6 | 1.0 | −0.2 | 0.0 | 2.6 | 1.0 | 1.3 | 1.5 | ||||
| Avg. | 2015 | −1.7 | −0.7 | −1.0 | −1.8 | −1.8 | −1.1 | −0.5 | −1.0 | −0.9 | −0.5 | ||
| 2016 | −0.6 | 1.1 | 0.6 | −0.5 | −0.3 | 1.6 | 0.9 | 0.6 | 0.8 | 0.9 | |||
Inflation Forecasts From Various Models
(Percent, year-on-year)
| Actual | Phillips | Disagg. | Desk | NBP | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | baseline | March 2015 | |||||
| 2014 | July | −0.2 | |||||||||||
| August | −0.3 | ||||||||||||
| September | −0.3 | ||||||||||||
| October | −0.6 | ||||||||||||
| November | −0.6 | ||||||||||||
| December | −1.0 | ||||||||||||
| 2015 | January | −1.4 | |||||||||||
| February | −1.6 | ||||||||||||
| Forecast | March | −1.5 | −1.6 | −1.4 | −1.4 | −1.7 | −1.7 | −1.5 | −1.1 | −1.3 | |||
| April | −1.1 | −1.7 | −1.2 | −1.2 | −2.0 | −2.0 | −1.5 | −0.8 | −1.3 | ||||
| May | −1.8 | −0.9 | −1.1 | −2.2 | −2.2 | −1.4 | −0.6 | −1.3 | −1.1 | ||||
| June | −1.8 | −0.7 | −1.1 | −2.3 | −2.3 | −1.3 | −0.4 | −1.1 | −1.0 | ||||
| July | −1.8 | −0.6 | −1.1 | −2.2 | −2.2 | −1.2 | −0.3 | −0.9 | −0.9 | ||||
| August | −1.8 | −0.4 | −0.9 | −2.1 | −2.1 | −1.1 | −0.3 | −0.8 | −0.9 | ||||
| September | −1.7 | −0.2 | −0.8 | −1.9 | −1.8 | −0.9 | −0.2 | −0.8 | −0.8 | ||||
| October | −1.6 | 0.0 | −0.6 | −1.7 | −1.6 | −0.7 | −0.1 | −0.7 | −0.6 | ||||
| November | −1.6 | 0.1 | −0.4 | −1.4 | −1.4 | −0.5 | 0.2 | −0.6 | −0.5 | ||||
| December | −1.5 | 0.3 | −0.2 | −1.1 | −1.0 | −0.2 | 0.5 | −0.4 | −0.2 | ||||
| 2016 | January | −1.4 | 0.4 | 0.1 | −0.8 | −0.7 | 0.2 | 0.8 | −0.2 | 0.0 | |||
| February | −1.2 | 0.6 | 0.4 | −0.7 | −0.5 | 0.5 | 0.7 | −0.1 | 0.1 | ||||
| March | −1.1 | 0.7 | 0.2 | −0.8 | −0.7 | 0.9 | 0.7 | 0.0 | 0.2 | ||||
| April | −1.0 | 0.8 | 0.3 | −0.7 | −0.6 | 1.2 | 0.7 | 0.2 | 0.4 | ||||
| May | −0.8 | 1.0 | 0.4 | −0.5 | −0.4 | 1.4 | 0.8 | 0.4 | 0.6 | ||||
| June | −0.7 | 1.1 | 0.5 | −0.5 | −0.3 | 1.7 | 0.9 | 0.5 | 0.8 | ||||
| July | −0.6 | 1.2 | 0.6 | −0.4 | −0.3 | 1.9 | 0.9 | 0.7 | 0.9 | ||||
| August | −0.4 | 1.3 | 0.7 | −0.4 | −0.2 | 2.1 | 0.9 | 0.8 | 1.0 | ||||
| September | −0.3 | 1.4 | 0.7 | −0.3 | −0.2 | 2.3 | 0.9 | 1.0 | 1.1 | ||||
| October | −0.2 | 1.5 | 0.8 | −0.3 | −0.2 | 2.4 | 1.0 | 1.1 | 1.3 | ||||
| November | 0.0 | 1.5 | 0.9 | −0.2 | −0.1 | 2.5 | 1.0 | 1.2 | 1.4 | ||||
| December | 0.1 | 1.6 | 1.0 | −0.2 | 0.0 | 2.6 | 1.0 | 1.3 | 1.5 | ||||
| Avg. | 2015 | −1.7 | −0.7 | −1.0 | −1.8 | −1.8 | −1.1 | −0.5 | −1.0 | −0.9 | −0.5 | ||
| 2016 | −0.6 | 1.1 | 0.6 | −0.5 | −0.3 | 1.6 | 0.9 | 0.6 | 0.8 | 0.9 | |||
NBP Inflation Projections as of March 2015
(Percent, year-on-year, unless otherwise indicated)
NBP Inflation Projections as of March 2015
(Percent, year-on-year, unless otherwise indicated)
| 2014 | 2015 | 2016 | 2017 | ||
|---|---|---|---|---|---|
| CPI | 0.0 | −0.5 | 0.9 | 1.2 | |
| Food | −0.9 | −0.7 | 1.4 | 1.6 | |
| Energy | −1.0 | −3.0 | 1.1 | 1.7 | |
| Core (net of food and energy) | 0.6 | 0.4 | 0.6 | 0.9 | |
| Output gap (percent of potential) | −1.0 | −0.6 | −0.6 | −0.3 | |
NBP Inflation Projections as of March 2015
(Percent, year-on-year, unless otherwise indicated)
| 2014 | 2015 | 2016 | 2017 | ||
|---|---|---|---|---|---|
| CPI | 0.0 | −0.5 | 0.9 | 1.2 | |
| Food | −0.9 | −0.7 | 1.4 | 1.6 | |
| Energy | −1.0 | −3.0 | 1.1 | 1.7 | |
| Core (net of food and energy) | 0.6 | 0.4 | 0.6 | 0.9 | |
| Output gap (percent of potential) | −1.0 | −0.6 | −0.6 | −0.3 | |
28. The risk of continued deflation remains high, though some upside risks could materialize.
Sustained decline in energy prices. Energy price inflation has historically been an important contributor to headline inflation. On the downside and considering the presence of sizable indirect and second-round effects on inflation, additional oil price declines could delay the projected pick-up in inflation. On the upside, the potential for a rapid turnaround in oil prices would help counter existing indirect and second-round effects already at work.
Low imported inflation. On the downside, persistently low inflation in the euro area could continue to weigh on imported inflation. Additional downside risks stem from potential zloty appreciation in the context of abundant liquidity associated with European Central Bank (ECB) quantitative easing and sizable positive interest-rate differentials with European emerging market peers. On the upside, the ECB’s monetary easing could have larger-than expected positive effects on euro area inflation, lifting imported inflation.
29. While models may be poor predictors of turning points, a near-term rapid pick-up in inflation is unlikely. While models may fit well historical observations, they often do poorly when forecasting turning points. Hence, results should be interpreted with care, considering the uncertainty related to the forecast of underlying input variables as well as general model uncertainty. In fact, history has shown that Polish inflation can quickly pick up in response to exogenous shocks. Nonetheless, considering the general low-inflation environment in Europe, a substantial near-term pick-up in inflation is considered a low probability event.
E. Conclusion
30. The empirical analysis shows strong evidence of combined indirect and second-round effects. Simulations of shocks to oil price inflation or euro area inflation reveal that a substantial portion of the decline in inflation is owing to feedback effects among the components of the CPI. These feedback effects, generated by the indirect impact of supply shocks on CPI subgroups or by affecting inflation expectations, maintain the downward pressure on prices.
31. The inflation forecasting models point to continued low inflation, with the current baseline in the upper range of the suggested paths. The current baseline projects inflation to enter the target band by end-2016, following continued near-term deflation, and to reach the 2.5 percent target during 2018. While a couple of models predict a return to the target band by end-2016, the majority of the models predict a more gradual pick-up. Hence, there are potential downside risks to the baseline path.
32. Given downside risks to the inflation forecast, monetary policy should remain accommodative. The presence of sizeable and persistent indirect and second-round effects, combined with low imported and commodity price inflation, suggests that the likelihood of a rapid return to the inflation target is low. In this environment, further downward shocks to commodity prices or unwarranted upward exchange-rate pressures could undermine the inflation objective. Policy makers should therefore stand ready to implement additional policy interest rate cuts if inflation expectations continue to decline or if interest-rate differentials continue to widen, attracting unwarranted capital inflows.
References
Broyden, C. G., 1965, “A Class of Methods for Solving Nonlinear Simultaneous Equations,” Mathematics of Computation, Vol. 19, No. 92, pp. 577–593, American Mathematical Society, October.
Demchuk, O., Lyziak, T., Przystupa, J., Sznajderska, A., and Wrobel, E., 2012, “Monetary Policy Transmission Mechanism in Poland. What Do We Know in 2011?” National Bank of Poland, Working Paper No. 116, Warsaw.
Gali, J. and Gertler, M., 1999, “Inflation Dynamics: A Structural Econometric Analysis,” Journal of Monetary Economics, Vol. 44(2), October, pp. 195–222.
IMF, 2014, “Sweden: Staff Report for the 2014 Article IV Consultation,” IMF Country Report No. 14/261, Washington, DC, International Monetary Fund.
IMF, 2015, “Low Inflation in European Inflation Targeters: Causes, Spillovers, and Policy Responses,” Cross-Country Report on Inflation, Selected Issues, Washington, DC, International Monetary Fund.
Iossifov, P. and Podpiera, J., 2014, “Are Non-Euro Area EU Countries Importing Low Inflation from the Euro Area?” IMF Working Paper No. WP/14/191, International Monetary Fund.
Kim, D. and Molagoda N., 2011, “Russian Federation: Selected Issues Paper,” Chapter I, Leading Indicators for Inflation in Russia, IMF Country Report No. 11/295, International Monetary Fund, September.
Leon, J., 2012, “A Disaggregate Model and Second Round Effects for the CPI Inflation in Costa Rica,” Munich Personal RePEc Archive (MPRA) Paper No. 44484, posted online in February 2013.
Appendix I. Data and Forecast Assumptions
Data
The quarterly output gap is computed from the annual estimate of potential real GDP. A cubic spline is fitted to the annual potential real GDP series, inserting the annual value as the second-quarter observation. The level of the series is then adjusted such that the average difference with the original annual series is zero. The output gap is then computed, using data on actual quarterly (seasonally adjusted) real GDP. A monthly output gap series is subsequently obtained from a cubic spline of the quarterly series.
Forecast
Quarterly forecasts for world oil and food prices are from the World Economic Outlook, Global Assumptions. Monthly forecast paths are then computed consistent with the quarterly paths. Other assumptions are consistent with staff’s macroeconomic framework at the time of estimation.
Appendix II. Disaggregate Model
Year-on-year CPI inflation is decomposed into 12 underlying components and their current weights in the consumption basket are recorded (Table AII.1).
CPI Components and Weights in the Consumption Basket
(Percent)
CPI Components and Weights in the Consumption Basket
(Percent)
| Component | Weight |
|---|---|
| Headline CPI | 100 |
| Food & nonalcoholic beverages | 24.4 |
| Alcoholic beverages & tobacco | 6.5 |
| Clothing & footwear | 5.4 |
| Housing, water, electricity, gas, & other fuels | 21.1 |
| Furnishing, house equipment, & routine maintenance | 4.9 |
| Health | 5.2 |
| Transport | 9.0 |
| Communication | 5.3 |
| Recreation & culture | 6.4 |
| Education | 1.0 |
| Restaurants & hotels | 5.2 |
| Miscellaneous goods & services | 5.7 |
CPI Components and Weights in the Consumption Basket
(Percent)
| Component | Weight |
|---|---|
| Headline CPI | 100 |
| Food & nonalcoholic beverages | 24.4 |
| Alcoholic beverages & tobacco | 6.5 |
| Clothing & footwear | 5.4 |
| Housing, water, electricity, gas, & other fuels | 21.1 |
| Furnishing, house equipment, & routine maintenance | 4.9 |
| Health | 5.2 |
| Transport | 9.0 |
| Communication | 5.3 |
| Recreation & culture | 6.4 |
| Education | 1.0 |
| Restaurants & hotels | 5.2 |
| Miscellaneous goods & services | 5.7 |
Forecasting CPI
As discussed by Leon (2012), using data available from the dynamics of the components allows setting up a disaggregate model for inflation. The transmission mechanism from domestic and external macroeconomic variables into the CPI inflation is as follows: macro-economic variables enter the model via the PPI and then affect the components that form the CPI indirectly via the effect of the PPI on each component. The different components then interact with each other and headline CPI is re-constructed using the consumption basket weights for each particular group.
The model is composed of the following equations:
PPI equation:
where Z is a vector of variables capturing macro factors, such as the nominal exchange rate (to account for the pass-through into producer prices), world oil and coal price inflation, labor market dynamics (proxied by the monthly unemployment rate from Eurostat, based on the quarterly labor force survey), and inflation in the euro area to take into account production linkages in supply chain networks with core European countries such as Germany. Results from estimating the PPI equation are shown in Table AII.2.
Coefficient Estimates from PPI and CPI Component Equations
Coefficient Estimates from PPI and CPI Component Equations
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | (11) | (12) | (13) | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Dependent variables → | PPI | G01 | G02 | G03 | G04 | G05 | G06 | G07 | G08 | G09 | G10 | G11 | G12 |
| PPI inflation (-1) | 0.340*** | 0.145*** | 0.0854** | 0.0829** | 0.105*** | 0.0220* | 0.111** | 0.378*** | 0.227** | 0.106*** | 0.246*** | 0.0404*** | 0.0681*** |
| (0.0535) | (0.0394) | (0.0340) | (0.0394) | (0.0335) | (0.0121) | (0.0442) | (0.102) | (0.106) | (0.0392) | (0.0744) | (0.00919) | (0.0133) | |
| Euro area inflation | 1.543*** | ||||||||||||
| (0.201) | |||||||||||||
| Unemployment rate | −0.668*** | ||||||||||||
| (0.177) | |||||||||||||
| Oil price inflation (-1) | 0.00785** | ||||||||||||
| (0.00349) | |||||||||||||
| Coal price inflation (-2) | 0.0413*** | ||||||||||||
| (0.00326) | |||||||||||||
| NEER (percent change) | −0.179*** | ||||||||||||
| (0.0187) | |||||||||||||
| G01(-1) | 0.838*** | 0.342*** | −0.595*** | 0.0292** | |||||||||
| (0.0570) | (0.121) | (0.164) | (0.0110) | ||||||||||
| G02(-1) | 0.891*** | −0.245*** | 0.981*** | 0.0797*** | |||||||||
| (0.0669) | (0.0879) | (0.275) | (0.0227) | ||||||||||
| G03(-1) | 0.383*** | 0.148*** | −0.177** | −0.0201** | |||||||||
| (0.0917) | (0.0521) | (0.0725) | (0.00984) | ||||||||||
| G04(-1) | 0.593*** | 0.0622** | 0.669*** | −1.271*** | 0.0960*** | ||||||||
| (0.0814) | (0.0271) | (0.187) | (0.277) | (0.0246) | |||||||||
| G05(-1) | −0.852*** | 0.750*** | 0.282** | 2.588*** | −0.0724** | ||||||||
| (0.167) | (0.0479) | (0.138) | (0.547) | (0.0313) | |||||||||
| G06(-1) | 0.249* | −0.0876*** | |||||||||||
| (0.127) | (0.0230) | ||||||||||||
| G07(-1) | 0.0514* | 0.0579** | 0.578*** | 0.215** | −0.101*** | ||||||||
| (0.0308) | (0.0246) | (0.106) | (0.0843) | (0.0304) | |||||||||
| G08(-1) | 0.655*** | −0.0645*** | 0.239*** | ||||||||||
| (0.0796) | (0.0239) | (0.0445) | |||||||||||
| G09(-1) | 0.0859* | −0.362* | 0.705*** | ||||||||||
| (0.0470) | (0.200) | (0.0757) | |||||||||||
| G010(-1) | 0.0367* | 0.0655* | 0.611*** | 0.0223*** | |||||||||
| (0.0191) | (0.0355) | (0.0735) | (0.00771) | ||||||||||
| G011(-1) | 0.877*** | 1.229*** | 0.0769* | 1.013*** | 0.821*** | ||||||||
| (0.283) | (0.233) | (0.0410) | (0.0260) | (0.0467) | |||||||||
| G012(-1) | −1.000*** | 0.445** | −1.383*** | ||||||||||
| (0.277) | (0.172) | (0.405) | |||||||||||
| Intercept | 4.972*** | 0.0233 | −0.835* | −3.905*** | 1.509*** | −0.221** | 0.823*** | −4.639*** | −0.725** | 2.060*** | −1.588*** | ||
| (1.531) | (0.148) | (0.479) | (0.606) | (0.332) | (0.0878) | (0.222) | (1.172) | (0.348) | (0.535) | (0.0967) | |||
| Observations | 62 | 62 | 62 | 62 | 62 | 62 | 62 | 62 | 62 | 62 | 62 | 62 | 62 |
| R-squared | 0.986 | 0.918 | 0.885 | 0.865 | 0.964 | 0.977 | 0.831 | 0.943 | 0.894 | 0.837 | 0.913 | 0.998 | 0.985 |
Coefficient Estimates from PPI and CPI Component Equations
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | (11) | (12) | (13) | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Dependent variables → | PPI | G01 | G02 | G03 | G04 | G05 | G06 | G07 | G08 | G09 | G10 | G11 | G12 |
| PPI inflation (-1) | 0.340*** | 0.145*** | 0.0854** | 0.0829** | 0.105*** | 0.0220* | 0.111** | 0.378*** | 0.227** | 0.106*** | 0.246*** | 0.0404*** | 0.0681*** |
| (0.0535) | (0.0394) | (0.0340) | (0.0394) | (0.0335) | (0.0121) | (0.0442) | (0.102) | (0.106) | (0.0392) | (0.0744) | (0.00919) | (0.0133) | |
| Euro area inflation | 1.543*** | ||||||||||||
| (0.201) | |||||||||||||
| Unemployment rate | −0.668*** | ||||||||||||
| (0.177) | |||||||||||||
| Oil price inflation (-1) | 0.00785** | ||||||||||||
| (0.00349) | |||||||||||||
| Coal price inflation (-2) | 0.0413*** | ||||||||||||
| (0.00326) | |||||||||||||
| NEER (percent change) | −0.179*** | ||||||||||||
| (0.0187) | |||||||||||||
| G01(-1) | 0.838*** | 0.342*** | −0.595*** | 0.0292** | |||||||||
| (0.0570) | (0.121) | (0.164) | (0.0110) | ||||||||||
| G02(-1) | 0.891*** | −0.245*** | 0.981*** | 0.0797*** | |||||||||
| (0.0669) | (0.0879) | (0.275) | (0.0227) | ||||||||||
| G03(-1) | 0.383*** | 0.148*** | −0.177** | −0.0201** | |||||||||
| (0.0917) | (0.0521) | (0.0725) | (0.00984) | ||||||||||
| G04(-1) | 0.593*** | 0.0622** | 0.669*** | −1.271*** | 0.0960*** | ||||||||
| (0.0814) | (0.0271) | (0.187) | (0.277) | (0.0246) | |||||||||
| G05(-1) | −0.852*** | 0.750*** | 0.282** | 2.588*** | −0.0724** | ||||||||
| (0.167) | (0.0479) | (0.138) | (0.547) | (0.0313) | |||||||||
| G06(-1) | 0.249* | −0.0876*** | |||||||||||
| (0.127) | (0.0230) | ||||||||||||
| G07(-1) | 0.0514* | 0.0579** | 0.578*** | 0.215** | −0.101*** | ||||||||
| (0.0308) | (0.0246) | (0.106) | (0.0843) | (0.0304) | |||||||||
| G08(-1) | 0.655*** | −0.0645*** | 0.239*** | ||||||||||
| (0.0796) | (0.0239) | (0.0445) | |||||||||||
| G09(-1) | 0.0859* | −0.362* | 0.705*** | ||||||||||
| (0.0470) | (0.200) | (0.0757) | |||||||||||
| G010(-1) | 0.0367* | 0.0655* | 0.611*** | 0.0223*** | |||||||||
| (0.0191) | (0.0355) | (0.0735) | (0.00771) | ||||||||||
| G011(-1) | 0.877*** | 1.229*** | 0.0769* | 1.013*** | 0.821*** | ||||||||
| (0.283) | (0.233) | (0.0410) | (0.0260) | (0.0467) | |||||||||
| G012(-1) | −1.000*** | 0.445** | −1.383*** | ||||||||||
| (0.277) | (0.172) | (0.405) | |||||||||||
| Intercept | 4.972*** | 0.0233 | −0.835* | −3.905*** | 1.509*** | −0.221** | 0.823*** | −4.639*** | −0.725** | 2.060*** | −1.588*** | ||
| (1.531) | (0.148) | (0.479) | (0.606) | (0.332) | (0.0878) | (0.222) | (1.172) | (0.348) | (0.535) | (0.0967) | |||
| Observations | 62 | 62 | 62 | 62 | 62 | 62 | 62 | 62 | 62 | 62 | 62 | 62 | 62 |
| R-squared | 0.986 | 0.918 | 0.885 | 0.865 | 0.964 | 0.977 | 0.831 | 0.943 | 0.894 | 0.837 | 0.913 | 0.998 | 0.985 |
Disaggregate equations:
Each of the components (i) of CPI at period t can be expressed as a function of their own lags, lags of other components of the CPI (the interaction terms capture the combined indirect and second-round effects) and the PPI:
To allow for sufficient degrees of freedom, we reduce each equation by assessing which variables and lag lengths are most informative. Some equations exclude intercepts whenever those are not statistically significant or exhibit unrealistically large values that may adversely affect the forecast of some components of the CPI. The full representations of the CPI group equations are shown in Table AII.2.
Headline CPI equation:
The model then adds the final equation, which aggregates the CPI groups into headline CPI using historical weights in the consumption basket:
Solving the system of equations:
In turn, this creates a system of 14 equations to be solved. To solve the model, we use Broyden’s Method (Broyden, 1965). Broyden’s method is a modification of Newton’s method, which tries to decrease the calculational cost of each iteration. Hence, the method shares many of the properties of Newton’s method, including the fact that it is not dependent on the ordering of the equations in the system and that it will generally converge quickly in the vicinity of a solution.
The period of analysis covers January 2010 and onwards. Data on year-on-year inflation are from the National Institute of Statistics (GUS).
Assessing indirect and second-round effects
We identify combined indirect and second-round effects by simulating exogenous shocks in the model. We begin the analysis of indirect and second-round effects by assuming a transitory (an alternatively a longer-lasting) exogenous shock to PPI inflation. Two effects of the shocks are then derived: (i) one that uses the model with interaction terms between components of CPI; and (ii) one that captures inertia by not allowing for feedbacks between components (i.e., assuming the interaction terms are zero). Combined indirect and second-round effects are then computed as the difference between the effects of the shock on overall CPI inflation derived from the two models.
Prepared by Lone Christiansen and Christian Ebeke. The econometric analysis in this chapter is based on information available as of March 30, 2015.
Since the beginning of 2004, the monetary policy guidelines have consisted of a continuous inflation target of 2.5 percent with a permissible fluctuation band of ±1 percentage point.
Various tariff changes in July 2013 led to a net one-off increase in inflation. Hence, the disappearance of the base effect in July 2014 resulted in a discrete decline in year-on-year inflation.
Demchuk et al. (2012) estimate that the maximum response of annual GDP growth following a rise in short-term interest rates, maintained for four quarters, occurs 4 quarters after the rate change. The maximum response of inflation occurs after six quarters.
We use the terms “first-round indirect effect” and “indirect effect” interchangeably throughout this chapter.
Please see Appendix I for methodological details on the quarterly output gap series.
The revision in potential output across a number of countries following the onset of the 2008–09 global financial crisis points to the inherent uncertainty associated with this measure.
We would like to thank Jorge Leon for his guidance in the implementation of the disaggregate inflation model.
