RDP 9807: Inflation Targeting in a Small Open Economy 2. Inflation Targeting in an Open Economy

Standard theoretical analysis of inflation targeting is based on a closed economy model. Aggregate demand depends negatively on the lag of the real interest rate. Aggregate supply is represented by a Phillips curve where inflation equals its expectation plus some adjustment for inflationary pressure associated with the output gap. The output gap is assumed to have a positive impact on inflation, often with a one-period lag. Thus a change in the interest rate is transmitted to inflation with a two-period lag via output.

Given the control lag with which policy affects inflation, the central bank must adopt a procedure known as inflation forecast targeting (Svensson 1996). In the simplest models, under strict inflation targeting, policy is set based on a two-period ahead forecast of inflation so as to move towards the target within the minimum control lag. However, under flexible inflation targeting the inflation forecast horizon is longer than the minimum control lag, depending on the nature of the central bank's preferences. These preferences are reflected in an objective function that penalises deviations of inflation from target, and possibly deviations of output from potential, and may also reflect some preference for interest-rate smoothing.

Recent contributions by Svensson (1998) and Ball (1998) extend models of inflation targeting to an open-economy setting. This requires explicit consideration of both the demand and exchange rate channels of monetary policy.

Ball (1998) includes the indirect effect of exchange rates on inflation via the output gap, as well as the direct effect of the exchange rate on the domestic price of imported intermediate inputs. The direct effect implies that the central bank can target inflation with a one-period lag. Thus movements in the exchange rate have strong implications for the setting of interest rates.^[3] Ball shows that an economy subject to exchange rate fluctuations will be forced to make frequent adjustments to the interest rate in order to achieve a strict inflation target. This will generate volatility in output which may be unacceptable to the monetary authority. To avoid excessive variability in the economy, Ball suggests targeting a refined measure of inflation which abstracts from the direct, but temporary effects of changes in the exchange rate.

Svensson (1998) presents an open-economy model where goods are either imported, or domestically produced. He assumes that the economy is small in the market for imported goods, but not in the world market for its own output. This justifies the distinction between CPI-inflation targeting, and targeting inflation of domestically produced goods. Strict CPI-inflation targeting relies on the direct exchange rate channel to stabilise inflation at a short horizon, and thus introduces substantial fluctuation in other variables. For this reason, Svensson recommends flexible CPI-inflation targeting which stabilises inflation and output at a longer horizon.

The appropriate distinction for a small open economy is that between aggregate inflation and inflation in the non-traded sector.^[4] Domestically produced goods are of two broad classes, traded or non-traded, the prices of which are determined by different factors. Traded goods prices are determined in world markets and converted into domestic currency prices at the prevailing exchange rate; non-traded goods prices are in large part determined by domestic conditions.

In this paper we present a model of a small open economy with a traded and non-traded sector. The exchange rate is forward looking and responds instantly to shocks and unexpected changes in the path of the policy instrument. Inflation in the traded sector is determined by changes in the exchange rate. Given the interaction between the interest rate and the exchange rate, monetary policy can impact upon the traded component of aggregate inflation rapidly due to the direct effect of the exchange rate. Non-traded inflation is affected indirectly via demand, and directly by the effect of policy on inflation expectations, and on the exchange rate.

Overlaying our representation of the economy is the central bank's loss function which embodies preferences over inflation, output stability, and interest-rate smoothing. An aggregate inflation target may induce excessive volatility in the interest rate to offset exchange rate shocks. A non-traded inflation target may induce excessive volatility in the exchange rate when policy is adjusted to offset supply or demand shocks. Under such circumstances it may be appropriate to adopt a more flexible aggregate inflation target over a longer forecast horizon.

Footnotes

This is of relevance to the construction of a Monetary Conditions Index (MCI), and is discussed by Ball (1998). [3]

For an early discussion of the distinction between traded and non-traded goods see Salter (1959) and Swan (1960). Interestingly, Swan considers a small open-economy model in which policy-makers are concerned about internal price stability, making this early work relevant to the discussion of inflation targeting. See also Pitchford (1993). [4]