MODELLING INFLATION IN CROATIA

TANJA BROZ & MARUŠKA VIZEK

Structure of the presentation

• Motivation for the research• Short overview of past inflation behaviour in

Croatia• Review of related work on modelling inflation • Derivation of the long-run sectoral determinants

of inflation • Derivation of short-run inflation model • Concluding remarks

Motivation for the research

• Constructing a structural model of inflation in Croatia

• Encompassing all relevant sectors of the economy

• Knowing inflation drivers is of the most essential importance for fulfilment of the Maastricht convergence criteria

• Previous research obscure

Past inflation behaviour

• 1990 – 1993;

hyperinflation and recession (1616% yoy)

• 1993;

introduction of the Stabilization program based on nominal exchange rate anchor

• 1994 – today;

stable inflation

-4

-3

-2

-1

0

1

2

3

4

5

6

1994. 1995. 1996. 1997. 1998. 1999. 2000. 2001. 2002. 2003. 2004. 2005. 2006.

Consumer price inflation Consumer price core inflation

In %, y-o-y

Related research• Payne (2002) – VAR - wage increase and

currency depreciation are positively correlated with inflation rates; despite hyperinflation no evidence of inflation inertia.

• Botrić and Cota (2006) – SVAR - terms of trade and balance of payments shocks have the strongest impact on prices; Croatia is a small open economy with high import dependency and uncompetitive economic structure.

• Pufnik and Kunovac (2006) - seasonal ARIMA - aggregating CPI component forecasts outperform both the overall CPI and random walk inflation forecasts.

Long-run determinants – Markup (1)

• Assuming one cointegrating vector and linear homogeneity, the derived long run markup relationship becomes:

• Share of unit labour cost in total unit cost (0.54) seems reasonable since Croatian economy is dominated by service sector.

• Large share of import prices in total unit cost is due to high import dependency of Croatian economy.

,-1)_(*)(* pricesimportulccpi

._*0.46-*0.54- pricesimportulccpimarkup

Long-run determinants – Markup (2)

• High correlation with the business cycle

Long-run determinants - Excess money (1)

• Cointegrating vector includes M1, CPI, GDP, real estate prices, money market rate, interest rate on foreign exchange deposits and a trend

trendforexr

ziborestatepgdpcpimmoneyexcess

*0.04-_*8.55-

*2.67_*0.39*2.97*4.071_

Long-run determinants - Excess money (2)

• Excess money exhibits counter-cyclical properties

Long-run determinants – NEER• By visual inspection only, the presence of B-S effect could not be

confirmed.• Hence for short-run inflation function only the first differences of kuna

nominal effective exchange rate is used.

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

4.40

4.45

4.50

4.55

4.60

4.65

REER Productivity differential between Croatia and its main trading partners

Long-run determinants – Excess demand

Inflation function (1)

• We regressed Dcpit on

– long run solutions – markupt-1, excessmoneyt-1, DNeert-i, and Dgdp_gapt-i, and

– short run dynamics Dppit-i, Dtrading_partners_cpit-i, Dm1t-i, Dimport_pricest-i, Dr_forext-i, where i takes 1 to 3

)57.0(32.0),94.0(18.0)88.0(,26.0),54.0(79.0

),58.0(72.0,89.8,009.0,60.0,2200611995

)06.0()002.0()04.0(

_*25.0_*33.0_*18.0

)001.0()07.0()00.0()00.0(

*41.01*07.0_*05.0*30.0

2

312

2221

RESETChowNormalityARCH

ARcriterionSchwarzsigmaRQQT

forexDLrgapDLgdppricesDLimport

DLNeerDLmmoneyLexcessLmarkupDLcpi

Inflation function (2)

• The model passes various diagnostic tests, encompassed the unrestricted general model, variables are statistically significant and have theoretically plausible signs (except nominal effective exchange rate).

)57.0(32.0),94.0(18.0)88.0(,26.0),54.0(79.0

),58.0(72.0,89.8,009.0,60.0,2200611995

)06.0()002.0()04.0(

_*25.0_*33.0_*18.0

)001.0()07.0()00.0()00.0(

*41.01*07.0_*05.0*30.0

2

312

2221

RESETChowNormalityARCH

ARcriterionSchwarzsigmaRQQT

forexDLrgapDLgdppricesDLimport

DLNeerDLmmoneyLexcessLmarkupDLcpi

Inflation function (3)• Diagnostic features of the inflation model

1995 2000 2005

0.00

0.02

0.04

DLcpi Fitted

1995 2000 2005

-1

0

1

2r:DLcpi (scaled)

-3 -2 -1 0 1 2 3

0.1

0.2

0.3

0.4

0.5 Densityr:DLcpi N(0,1)

2003 2004 2005 2006

0.00

0.02

0.041-step Forecasts DLcpi

Inflation function (4)• Recursive Analysis of Inflation Model

2000 2005

-0.02

0.00

0.02

Res1Step

2000 2005

0.25

0.50

0.75

1.001up CHOWs 1%

2000 2005

0.25

0.50

0.75

1.00Ndn CHOWs 1%

2000 2005

0.25

0.50

0.75

1.00Nup CHOWs 1%

Inflation function (5)• Recursively estimated coefficients of the inflation model

2000 2005

-1.0

-0.5

0.0

Recursive estimation of DLcpi

DLr_forex_3 2*SE

2000 2005

-0.25

0.00

0.25 DLm1_2 2*SE

2000 2005

-2

-1

0

1DLNeer_2 2*SE

2000 2005

0

1DLimport_prices_2 2*SE

2000 2005

-1

0

1 DLgdp_gap_1 2*SE

2000 2005

-0.1

0.0

0.1Lexcess_money_2 2*SE

2000 2005

-0.5

0.0

0.5 Lmarkup_1 2*SE

Concluding remarks

• All long run relationships significantly influence quarter-on-quarter inflation rate.

• Markup and excess money, imposed as error correction terms are the most significant variables for explaining the short run behaviour of inflation.

• Price and monetary variables are found to be important in explaining inflation variance.

• Magnitude of monetary aggregates influence on CPI is marginal.

• Pass-through of Croatian trading partners CPI and Croatian PPI on inflation in Croatia are not significant.

• Presented diagnostic tests suggest that the model could be used for forecasting purposes.