Seðlabanki Íslands Is inflation control more difficult in

Seðlabanki Íslands Is inflation control more difficult in very small open economies? The Challenges of Globalisation for Small Open Economies with Independent Currencies 31 May – 1 June 2007 Thórarinn G. Pétursson Central Bank of Iceland

General consensus • Keiko & Hunt (2006) – “… inflation stabilization is more challenging in Iceland than in other industrial countries primarily because of the relative magnitudes of the economic shocks” – “. . . Inflation-output variability trade-off faced by the monetary authorities in Iceland is much less favorable than those faced by the monetary authorities in many other industrial countries, even other small inflationtargeting countries like New Zealand” – “Iceland New Zealand are both subject to to bigger shocks than the other large countries. The key difference between Iceland New Zealand is that the inflation and exchange rate shocks in Iceland are considerably larger than those in New Zealand” • Hunt (2007) – “Macroeconomic volatility in Iceland is already much higher than in all other industrial countries. To a large extend this simply reflects the small size of the economy”

Country sample • Want observations on “very small” countries that are informative for the more developed countries such as ICE… • … without extending the analysis to all the very small Caribbean islands that are very poor and undeveloped • The full sample therefore includes 34 countries – Median population 9. 5 m – Median GDP per capita 29. 8 thousand PPP adjusted US$ – GDP about 50% of world GDP (15% of world population) • Population distribution of the 34 countries – – – – POP size < 1 m: 3 countries POP size 1 -5 m: 7 countries POP size 5 -10 m: 8 countries POP size 10 -20 m: 6 countries POP size 20 -50 m: 3 countries POP size 50 -100 m: 5 countries POP size > 100 m: 2 countries

Inflation more volatile in small countries. . . • Standard deviation of inflation 1985 -2005 • Inflation measured as annualised quarterly changes in seasonally adjusted consumer prices • Countries ranked according to PPP adjusted GDP in 2006

. . . and harder to predict • Standard deviation of inflation forecast errors from 1995 -05 • Estimated from a 40 quarters rolling window VAR model including inflation, import price inflation, output gap and interest rate

What can explain this? • A simple New-Keynesian Phillips curve gives some clues πt = απt-1 + (1 – α)Etπt+1 + φyt + λxt + εt • • πt: inflation and Etπt+1: expected inflation yt: the output gap xt: exchange rate shocks εt: other supply shocks • Key factors therefore seem to be – Persistence of inflation shocks and variability of inflation expectations – Variability of real shocks and exposure to idiosyncratic shocks – Variability of exchange rate shocks and its pass-through to prices – Variability of other shocks – Credibility of monetary policy and its transparency

Smaller economies are more open…

. . . more exposed to idiosyncratic shocks. . . • Measured as the contemporenous correlation of the cyclical components of domestic and world GDP

… and with more volatile real economy

Interest rates seem more volatile. . .

. . . and monetary policy less predictable • Standard deviation of interest rate forecast errors from 1995 -05 • Estimated from a 40 quarters rolling window VAR model including inflation, import price inflation, output gap and interest rate • An alternative measure using a forward-looking Taylor rule gives very similar results

But inflation seems less persistent. . . • Estimate a unilateral inflation equation allowing for a break in mean inflation of unknow timing πt = (α + βDt) + γπt-1 + ΣiφiΔπt-i + εt • Inflation persistence is given by γ

… and exchange rates more stable

But exchange rate noise is greater… • A signal extraction approach suggested by Durlauf and Hall • Gives a lower bound for the variance of model noise in rational expectations models • Closely related to the variance of the exchange rate risk premium

… and the pass-through to prices greater • Estimate a VAR model with domestic and foreign inflation, exchange rate changes, the output gap and short-term interest rate • Generate accumulative impulse responses for a 1% shock to the exchange rate using the generalised approach of Pesaran and Shin Not included

Cross-country inflation volatility • What explains the cross-country variation of inflation volatility (INFVOL)? • Potential explanations – Country size: SIZE – Country openness: OPEN – Real economy fluctuations: YVOL – Importance of idiosyncratic shocks: INTER – Inflation persistence: PERS – Monetary policy predictability: RSE – Exchange rate pass-through: PT – Exchange rate noise: EXNOISE

Cross-country inflation volatility INFVOL = -0. 004 – 0. 025 OPEN + 0. 279 RSE (0. 3) (2. 0) (4. 2) + 0. 182 PT + 0. 230 EXNOISE (6. 3) (3. 3) N = 33, R 2(adj) = 0. 72, SE = 0. 028 • Other variables not significant • Robustness with respect to country sample – PT and EXNOISE robust to excluding any country – OPEN significant at the 10% level, except when excluding ISR (p = 0. 60) and MAL (p = 0. 13) – RSE significant at the 1% level, except when excluding TUR (p = 0. 56)

Interpretation • So the general view seems correct – Inflation control is more difficult in small open economies to the extend that they have noisier exchange rates and a larger pass-through to inflation • But there is hope – Adopt a credible and transparent inflation-fighting monetary policy that can reduce the pass-through, make monetary policy more predictable and contribute to reducing other sources of volatility – Reform other macro policy to give better support to inflation fighting credibility – These policy reforms can also contribute to reducing FOREX risk premia • Structural issues – More open to international trade – Improve functioning of FOREX market – Join a larger currency area?